Biophysical Studies of the First Binding Domain of Receptor 2A to Assess the Significance of Phosphorylation and Mutations

by

Elvin Dominic de Araujo

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Department of Chemistry University of Toronto

©Copyright by Elvin Dominic de Araujo 2015

Biophysical Studies of the First Nucleotide Binding Domain of 2A to Assess the Significance of Phosphorylation and Mutations

Elvin Dominic de Araujo

Doctor of Philosophy

Department of Chemistry University of Toronto

2015 Abstract The sulfonylurea receptor 2A (SUR2A) are ATP-binding cassette (ABC) transporters that form regulatory subunits in ATP-sensitive potassium channels (K ATP ) channels found in metabolically active tissues. In K ATP channels, four SUR proteins surround four pore-forming

Kir6 subunits. By sensing intracellular [ATP]/[ADP] ratios, KATP channels couple the metabolic state of the cell to the membrane potential and therefore have crucial roles in many biological processes. For example, K ATP channels in the are vital for proper regulation, whereas cardiac K ATP channels contribute to shortening of action potentials which may protect the heart against arrhythmias.

Gating of K ATP channels is a complex process that involves multiple ligands and domains. ATP binding at the Kir subunits closes the pore, whereas MgATP binding and hydrolysis at the SUR nucleotide binding domains (NBDs) results in channel opening. Thus, the

NBDs are critical sites of regulation for the K ATP channel, although the molecular mechanisms for how the NBDs alter channel activity is not well understood. This is mainly a consequence of poor solubility of the isolated SUR NBDs. Here we have developed various strategies that have allowed us to perform detailed NMR experiments and have provided molecular level information

ii on the conformation of SUR2A NBD1. These studies enabled us to investigate the structural, dynamic and functional effects of phosphorylation, disease-causing mutations and drug binding in NBD1. Our data suggest that SUR2A NBD1 contains a number of disordered loops that may function as regulatory regions. Phosphorylation or mutations in these regions, can alter their interactions with the core of the protein, thereby affecting nucleotide binding at NBD1 as well as the equilibrium between different NBD1 conformations that ultimately modulate the activity of the K ATP channel.

iii

Acknowledgments

There are a number of people who have truly enhanced my graduate experience over the past years. My deepest gratitude is to my supervisor Prof. Voula Kanelis who has always been supportive and has helped me grow as a well-rounded researcher. I am grateful to her mentorship and unwavering support and enthusiasm, and she never misses an opportunity to turn something into a teachable moment.

I am grateful to my committee members, Prof. R. Scott Prosser and Prof. Drew Woolley, for their expertise, insightful suggestions and technical support over the years. Prof. Scott Prosser is an avid researcher and his enthusiasm helped instill my interest in biophysics during my undergraduate years. Prof. Drew Woolley is very patient and accommodating and has taught me many valuable lessons in biochemistry and chemical biology. I would also like to thank Prof. Ulrich Krull, Prof. Mark Nitz and Prof. Xiao-an Zhang for helping shape my understanding of biochemistry and NMR. I would like to acknowledge Prof. Brian Shilton, Prof. Jumi Shin and Prof. Deborah Zamble for reviewing my thesis. I would also like to acknowledge Dr. John Rubinstein for all of his suggestions and advice on all of our publications and Prof. Barry Green for his support.

I am also appreciative for the support of my several lab mates over the past years with special thanks to Dr. Jorge P. Lopez-Alonso, Marijana Staglijar, Lynn K. Ikeda, Clarissa R. Sooklal, Serisha Moodley, Sasha Weiditch and Alexandria Albanese. My past and present colleagues have been instrumental to create an enjoyable and engaging working environment. I would also like to acknowledge my colleagues from other labs who have all helped me in various ways and have provided me with a remarkable graduate experience throughout the years. I am grateful to the support from the many past and present members of the Prosser Lab including Dr. Sameer Al-Abdul-Wahid, Dr. Rohan Alvares, Dr. Libin Ye, Tae Hun Kim and Joshua Hoang. I would also like to thank Dr. Sacha Larda from the Prosser Lab for his leadership during our time on the CPS graduate student union. I am grateful to my colleagues from the Shin Lab (Dr. Sam Sathiamoorthy, Dr. Pam Nge, Dr. Ichiro Inamoto, Dr. Inder Sheroan, Dr. Antonia DeJong and Alexandra Strak) for their support. I am also thankful to the support from the MacDonald Lab

iv

(Dr. Quasim Saleem and Angel Lai), Krull Lab (Uvaraj Uddaysankar), Kay lab (Dr. Ranjith Muhandiram) and the Stewart Lab (Dr. Colin De Mill, Marzena Serwin).

I would especially like to thank Prof. Peter Macdonald for providing me multiple opportunities to grow as an educator and teach organic chemistry. I would also like to extend my appreciation to the many members of the Chemical and Physical Sciences Department for their expertise and assistance while I was teaching, including Prof. Jumi Shin, Prof. Patrick Gunning, Prof. Juris Strautmanis, Dr. Sreekumari Nair, Dr. Krish Radhakrishna, Liz Kobluk, Angela Sidoriak, Rubina Lewis, Heidi Moore, and Donna Coulson.

I am grateful to doctoral grants from the Canadian Institutes of Health Research and Queen Elizabeth II Science and Technology.

Finally, I would especially like to thank my parents and my sister for their unwavering support throughout my graduate studies.

v

Table of Contents

Acknowledgments ...... iv Table of Contents ...... vi List of Figures ...... x List of Tables ...... xiii List of Equations...... xiv List of Abbreviations ...... xv Chapter 1 Introduction ...... 1

1 Overview of K ATP channels ...... 1

1.1 Molecular Architecture of K ATP channels...... 1 1.2 Overview of the Kir subunit ...... 2 1.3 Overview of the SUR subunit ...... 4 1.3.1 Overview of the ABC transporters ...... 4 1.3.2 General Structure of ABC transporters ...... 5 1.3.2.1 Arrangement of domains in ABC proteins ...... 6 1.3.2.2 Subdivisions in the ABCC subfamily ...... 8 1.3.2.3 Structure of the Membrane Spanning Domains (MSDs) ...... 8 1.3.2.3.1 The role of MSD0 and the L0 linker ...... 9 1.3.2.4 Structure of the Nucleotide Binding Domains (NBDs) ...... 11 1.3.3 SUR isoforms and splice variations ...... 13

1.4 Nucleotide regulation of K ATP channel activity and kinetics ...... 15 1.4.1 Phosphotransfer Networks ...... 18

1.5 Physiological Role of K ATP channels ...... 20

1.6 Additional Regulation of K ATP channels ...... 22 1.6.1 Pharmaceutical Regulation ...... 22 1.6.1.1 ...... 22 1.6.1.2 Potassium channel openers ...... 24 1.6.2 Phosphatidylinositol ...... 25

1.6.3 Phosphorylation of the K ATP channel ...... 26

1.6.4 Other known regulators of the K ATP channel ...... 27 1.7 Biophysical techniques ...... 27 1.7.1 NMR Spectroscopy ...... 28 1.7.1.1 TROSY-HSQC ...... 29 1.7.2 Fluorescence Spectroscopy ...... 31 1.7.3 Circular Dichroism...... 33 1.8 Thesis Framework ...... 35 Chapter 2 Expression and Purification of Isolated SUR2A NBD1 ...... 36

vi

2 Overview ...... 36 2.1 Introduction ...... 36 2.2 Methods ...... 40 2.2.1 Expression and purification of SUR2A NBD1 ...... 40 2.2.2 Protein concentration determination ...... 42 2.2.3 Sample storage ...... 42 2.2.4 Screening buffer conditions for thermodynamic stability ...... 42 2.2.5 Fluorescence Nucleotide Titrations...... 43 2.2.6 Nucleotide NMR Titrations ...... 44 2.3 Results and Discussion ...... 44 2.3.1 Thermal denaturation studies and buffer screening ...... 44 2.3.2 Purification strategies and long term storage ...... 50 2.3.3 Nucleotide Titrations by NMR spectroscopy ...... 54 2.3.4 Comparison with other screening techniques...... 57 2.3.5 Application to other proteins...... 59 2.4 Conclusions ...... 60 Chapter 3 Resonance Assignments for SUR2A NBD1 ...... 61 3 Overview ...... 61 3.1 Introduction ...... 61 3.1.1 Triple Resonance Assignments ...... 61 3.1.2 Specific Amino Acid Labelling ...... 64 3.2 Methods ...... 66 3.2.1 Protein Expression for NMR Resonance Assignment ...... 66 3.2.2 Specific Amino Acid Labelling ...... 66 3.2.2.1 Specific Amino Acid 15 N Labelling ...... 66 3.2.2.2 Specific Amino Acid 15 N Suppression Labelling ...... 67 3.2.3 Three Dimensional NMR Experiments ...... 68 3.2.4 NMR and Fluorescence Screening of Ligands ...... 68 3.3 Results and Discussion ...... 69 3.3.1 Triple Isotopically Labelled Protein Expression ...... 69 3.3.2 Optimization of NBD1-ΔN samples for NMR spectroscopy ...... 71 3.3.2.1 Temperature ...... 71 3.3.2.2 Nucleotide and Buffer Component Screening ...... 73 3.3.3 Resonance Assignment Experiments ...... 81 3.3.3.1 14 N Unlabelling Strategies ...... 81 3.3.3.2 15 N Labelling Strategies ...... 84 3.3.4 NBD1-ΔN Resonance Assignment Analysis ...... 88 3.3.5 NBD1-ΔN Nucleotide Binding Analysis ...... 90

vii

3.3.6 Lobe II of NBD1-ΔN affects protein stability ...... 94 3.4 Conclusions ...... 95 Chapter 4 Phosphorylation Studies of SUR2A/SUR2B NBD1 ...... 97 4 Overview ...... 97 4.1 Introduction ...... 97 4.2 Methods ...... 99 4.2.1 Protein Purification and Expression ...... 99 4.2.2 NMR Experiments ...... 100 4.2.3 Phosphorylation of NBD1 (S615-L933) & NBD1-ΔC (S615-D914) ...... 101 4.2.4 Fluorescence Nucleotide Binding Experiments ...... 101 4.2.5 Thermal Stability Measurements ...... 102 4.2.6 Dynamic Light Scattering Studies ...... 102 4.3 Results ...... 103 4.3.1 Identification of phosphorylation sites in SUR2A ...... 103 4.3.2 Phosphorylation dependent spectral changes in SUR2A NBD1 ...... 107 4.3.3 Removal of the N-terminal region mimics phosphorylation of NBD1 ...... 112 4.3.4 Phosphorylation increases the nucleotide binding affinity of NBD1 ...... 116 4.4 Discussion ...... 118 Chapter 5 The Disease-Causing Mutation, V734I, Alters the Conformation and Nucleotide Binding of SUR2A NBD1 ...... 121 5 Overview ...... 121 5.1 Introduction ...... 121 5.2 Methods ...... 123 5.2.1 Protein Expression ...... 123 5.2.2 Fluorescence Quenching ...... 124 5.2.3 Fluorescence Nucleotide Binding ...... 125 5.2.4 NMR Spectroscopy ...... 125 5.3 Results ...... 126 5.3.1 Altered conformation of wild type and mutant SUR2A NBD1 ...... 126 5.3.2 Phosphorylation of the N-tail in NBD1-V730I partially masks the molecular effects of the V730I mutation ...... 130 5.3.3 Removal of the N-terminal tail rescues the activity of mutant NBD1 ...... 132 5.3.4 The N-tail (Exon 14) interacts with the disease-causing region (Exon 17) of NBD1 ...... 136 5.4 Discussion ...... 137 Chapter 6 Expression and Purification of SUR2A NBD2 ...... 142 6 Overview ...... 142 6.1 Introduction ...... 142 6.1.1 Asymmetry of NBDs in ABC transporters ...... 143

viii

6.2 Methods ...... 144 6.2.1 Protein Expression and Purification ...... 144 6.2.1.1 Purification of different SUR2A NBD2 constructs under native conditions ...... 144 6.2.1.2 Purification of SUR2A NBD2 under denaturing conditions ...... 144 6.2.2 NMR Experiments ...... 145 6.3 Results ...... 145 6.3.1 SUR2A NBD2 Expression and Purification Profiles ...... 145 6.3.2 NBD2 domain boundary screening ...... 147 6.3.3 Alternative strategies for obtaining SUR2A NBD2 ...... 148 6.3.4 SUR2A NBD1 as a solubility enhancer for SUR2A NBD2 ...... 149 6.4 Discussion ...... 151 Chapter 7 Conclusions & Future Directions ...... 153 7.1 Summary ...... 153 7.2 Discussion ...... 155 7.3 Future Directions ...... 157 7.3.1 Intramolecular interactions within NBD1 ...... 157 7.3.2 Intermolecular interactions of NBD1 with other domains ...... 158 7.3.3 Drug Binding Analysis of NBD1 ...... 159

7.3.4 Full K ATP channel structural studies ...... 160 Chapter 8 Appendix Drug binding to SUR2A NBD1 ...... 162 8 Overview ...... 162 8.1 Introduction ...... 162 8.2 Methods ...... 165 8.2.1 Materials ...... 165 8.2.2 Protein Expression and Purification ...... 165 8.2.3 Homology Models ...... 166 8.2.4 Drug Binding Studies by NMR Spectroscopy ...... 166 8.2.5 Intrinsic Tryptophan Fluorescence Experiments ...... 167 8.2.6 NMR Relaxation Experiments ...... 168 8.2.7 Fluorescence Nucleotide Binding Experiments ...... 168 8.3 Results ...... 170 8.3.1 NMR titration studies indicate that NBD1 of rSUR2A mediates specific interactions with pinacidil ...... 170 8.3.2 Specific Binding of pinacidil by SUR2A NBD1 displays a low affinity ...... 176 8.3.3 Binding of pinacidil changes the affinity of NBD1 for MgATP ...... 177 8.4 Discussion ...... 179 References ...... 182 Copyright Acknowledgments ...... 210

ix

List of Figures

Chapter 1 ...... 1

Figure 1.1: Schematic representations/models of EM images of the K ATP channel ...... 1

Figure 1.2: Models depicting the side/top view of Kir6.2 association in the K ATP channel ...... 2 Figure 1.3: Structure of ABC transporter, Sav1866, showing the canonical domains in different colours ...... 6 Figure 1.4: Schematic representation of domains in the Kir and SUR subunits ...... 7 Figure 1.5: TM arrangement in the MSDs of different ABC transporters ...... 8

Figure 1.6: Role of MSD0 and the L0 linker in the KATP channel ...... 10 Figure 1.7: Homology model and schematic of the CFTR NBD dimer ...... 12 Figure 1.8: Genomic arrangement of sur and kir on 11 and 12 ...... 13

Figure 1.9: Different conformations of the K ATP channel ...... 15

Figure 1.10: Different models of nucleotide regulated gating of the K ATP channel ...... 16 Figure 1.11: Phosphotransfer network in different cell types ...... 18

Figure 1.12: Role of K ATP channels in insulin release and metabolism in the pancreas cell ...... 20 Figure 1.13: Chemical structures of compounds classified as either potassium channel openers and sulfonylurea drugs ...... 23 Figure 1.14: HSQC pulse sequence and corresponding produce operators ...... 30 Figure 1.15: Jablonski diagram depicting the various transitions following absorption of a photon by a molecule .. 32 Figure 1.16: Electron orbital diagram of the amide bond ...... 34 Chapter 2 ...... 36 Figure 2.1: Structure-based sequence alignment of SUR2A NBD1 with ABC transporters ...... 38 Figure 2.2: Structural and biochemical characterization of SUR2A NBD1 S615-L933 ...... 39 Figure 2.3: Denaturation profiles of SUR2A NBD1 through circular dichroism and fluorescence spectroscopy ..... 45 Figure 2.4: Thermal unfolding assays screening different buffer conditions ...... 47 Figure 2.5: SUR2A NBD1 has a higher affinity for MgATP than MgADP ...... 49 Figure 2.6: Overview of purification of SUR2A NBD1 ...... 51 Figure 2.7: MgATP binding causes chemical shift changes for specific NBD1 resonances ...... 55 Figure 2.8: Comparison of chemical shift changes in NBD1 with MgATP ...... 58 Chapter 3 ...... 61 Figure 3.1: Atoms detected in six common triple resonance assignments ...... 63 Figure 3.2: Growth curves of transformed BL21 RIP Codon Plus E. coli cells in different media ...... 70

x

Figure 3.3: Comparison of TROSY-HSQC of NBD1-ΔN ...... 71 Figure 3.4: TROSY-HSQC temperature profile for NBD1-ΔN...... 72 Figure 3.5: ATP titrations with NBD1-ΔN...... 74 Figure 3.6: ADP titrations with NBD1-ΔN ...... 75 Figure 3.7: TROSY-HSQC of NBD1-ΔN with non-hydrolyzable ATP analogues ...... 76 Figure 3.8: pH titrations with NBD1-ΔN...... 78 Figure 3.9: NaCl titrations with NBD1-ΔN...... 79 Figure 3.10: Osmolyte titrations of 100 μM NBD1-ΔN...... 80 Figure 3.11:. Strip plots of sequential correlations from the resonance assignment data ...... 82 Figure 3.12: NMR spectra of NBD1-ΔN expressed with specifically unlabelled amino acids...... 83 Figure 3.13: NMR spectra of NBD1-ΔN expressed with specifically 15 N labelled amino acids...... 85 Figure 3.14: Resonance assignments for SUR2A NBD1-ΔN...... 87 Figure 3.15: Resonance assignments plotted on surface of SUR2A NBD1-ΔN model...... 88

Figure 3.16: Concentration dependence of R 1ρ relaxation rates...... 89 Figure 3.17: Residue specific changes upon ATP binding in NBD1-ΔN...... 90 Figure 3.18: Ribbon diagram of SUR2A NBD1-ΔN upon changes with ATP and ADP...... 91 Figure 3.19: Ribbon diagram of SUR2A NBD1-ΔN with changes upon ATP-analogue binding...... 92 Figure 3.20: Manganese binding to NBD1-ΔN...... 93

Figure 3.21: Changes to NBD1-ΔN with salt, glycerol and D2O...... 95 Chapter 4 ...... 97 Figure 4.1: Models/Schematics of full length SUR2A and SUR2B NBD1...... 98 Figure 4.2: Identification of resonances from the NBD1 core, the N-terminal tail, and the C-terminal tail...... 103 Figure 4.3: Time-course of PKA phosphorylation of SUR2A NBD1 ...... 105 Figure 4.4: Mass spectrometry data identify the T632 and S636 phosphorylation sites...... 106 Figure 4.5: Spectral changes in NBD1 with phosphorylation and removal of the N-terminal tail...... 108 Figure 4.6: Spectral changes in NBD1-ΔC with phosphorylation and removal of the N-terminal tail...... 110 Figure 4.7: Phosphorylation alters the equilibrium of N-terminal tail interactions...... 113 Figure 4.8: Temperature-dependent changes in spectra of non-phospho-NBD1, phospho-NBD1, and NBD1-ΔN. 114 Figure 4.9: Phosphorylation increases the size of NBD1...... 115 Figure 4.10: Nucleotide binding to non-phospho-NBD1, phospho-NBD1, and NBD1-ΔN ...... 117 Figure 4.11: Structural model for effects of phosphorylation on the N-terminal tail with the NBD1 core ...... 119 Chapter 5 ...... 121

xi

Figure 5.1: Quenching studies of wild type and mutant NBD1 ...... 127 Figure 5.2: TNP-ATP binding to wild type and mutant NBD1...... 128 Figure 5.3: TROSY-HSQC spectra of wild type and mutant NBD1 at different concentrations...... 129 Figure 5.4: TROSY-HSQC of phosphorylated NBD1-V730I at different temperatures...... 130 Figure 5.5: Quenching studies of phosphorylated wild type and phosphorylated mutant NBD1 ...... 131 Figure 5.6: TNP-ATP binding to wild type and mutant and phosphorylated mutant NBD1...... 132

Figure 5.7: Quenching studies of wild type and mutant NBD1-ΔN...... 133 Figure 5.8: TROSY-HSQC of phosphorylated NBD1 and phosphorylated NBD1-V730I...... 134 Figure 5.9: Fluorescence emission spectra of NBD1 and various Trp mutants...... 135 Figure 5.10: TROSY-HSQC of 200 μM NBD1-ΔN (black) and NBD1-ΔN,Δ17...... 137 Figure 5.11 Model of the effects of different regulatory regions in NBD1...... 139 Chapter 6 ...... 142 Figure 6.1: Purification of SUR2A NBD2 ...... 146 Figure 6.2: Gel filtration following phosphorylation of NBD2...... 148 Figure 6.3:. Purification of NBD2 under denaturing conditions ...... 149 Figure 6.4: UV trace of the gel filtration of NBD1 and NBD2...... 150 Figure 6.5:. Schematic of the NBD1-NBD2 fusion constructs ...... 151 Chapter 7 ...... 153 Figure 7.1: Model of cooperativity of the SUR NBDs...... 156 Figure 7.2: The ED domain in SUR1 and SUR2...... 158 Figure 7.3: 19 F-NMR of Fluoro-Trp labelled NBD1-ΔN ...... 160 Chapter 8 ...... 162

Figure 8.1 Models/structures of SUR and Kir6. x proteins, and common K ATP channel openers ...... 163 Figure 8.2 Saturation of pinacidil binding to NBD1 in the presence of MgATP ...... 171 Figure 8.3 Diazoxide does not interact with SUR2A NBD1 in the presence of MgATP ...... 172 Figure 8.4 DMSO-mediated changes in NMR spectra of SUR2A NBD1 ...... 173

15 Figure 8.5 N R 2 relaxation data for 250 μM SUR2A NBD1 ...... 175 Figure 8.6 Binding of pinacidil to SUR2A NBD1 monitored by intrinsic Trp fluorescence ...... 176 Figure 8.7 TNP-ATP binding SUR2A NBD1 in the absence and presence of pinacidil ...... 177

xii

List of Tables

Chapter 1 ...... 1 Table 1.1: Subfamilies of human ABC Transporters ...... 5 Table 1.2: Splice variations in the sur2 ...... 14

Table 1.3: Subunit assembly of K ATP channels based on tissue localization ...... 20 Chapter 2 ...... 36 Table 2.1: Summary of NBD1 melting temperatures with various ligands ...... 48 Table 2.2: MgATP concentration affects the final NBD1 purification yield ...... 52 Chapter 3 ...... 61 Table 3.1: Magnetization transfer in 3D resonance assignment experiments ...... 64 Table 3.2: NMR Resonance Assignment Experiment Parameters ...... 68 Table 3.3: Extent of Detectable Metabolic Scrambling with Different Growth Conditions ...... 84 Chapter 5 ...... 121 Table 5.1: Iodide quenching constants for various NBD1 constructs ...... 135 Table 5.2: Acrylamide quenching constants for various NBD1 constructs ...... 136 Chapter 6 ...... 142 Table 6.1:Expression and Purification of Various SUR2A NBD2 Constructs ...... 147 Chapter 8 ...... 162 Table 8.1 Dissociation Constants for Interactions of SUR2A NBD1 with Nucleotide...... 178

xiii

List of Equations

Chapter 1 ...... 1 Equation 1.1: Magnetic moment of a nucleus ...... 28 Equation 1.2: Energy difference in a magnetic field ...... 28 Equation 1.3: Fluorescence quenching ...... 33 Chapter 2 ...... 36 Equation 2.1 Dissociation constant determination for TNP-nucleotide binding to NBD1...... 43 Equation 2.2 Combined chemical shift calculation. in ppm ...... 44 Chapter 4 ...... 97 Equation 4.1 Combined chemical shift calculation in Hz...... 100 Chapter 5 ...... 121 Equation 5.1 Corrected Stern-Volmer quenching ...... 124 Equation 5.2: Inner filter effect correction ...... 125 Chapter 8 ...... 162 Equation 8.1 Drug binding to NBD1 ...... 168

Equation 8.2 Determining R 2 rates from R 1ρ rates ...... 168

xiv

List of Abbreviations

ABC ATP-binding cassette KSV Stern-Volmer ADVANCE Action in Diabetes and Vascular Disease MDR multiple drug resistance AK adenylate kinase MRP multi-drug resistance ALD MSD membrane spanning domain BP binding protein N- amino C- carboxy NBD nucleotide binding domain C42 region the C-terminal 42 residues in SUR Ni 2+ - nickel nitrilotriacetic acid proteins NTA CD circular dichroism NMR nuclear magnetic resonance CFTR transmembrane OABP oligoA binding protein conductance regulator

CK creatine kinase PIP 2 phosphatidylinositol-4,5-bisphosphate DCA deoxycholic acid PK phosphokinases DSS 4,4-dimethyl-4-silapentane-1-sulfonic PKA protein kinase A acid EM electron microscopy PKC protein kinase C ER endoplasmic reticulum P-loop binding loop FDA Food and Drug Administration PMSF phenylmethylsulfonyl fluoride GCN20 general control nonderepressible QANUC Quebec/Eastern Canada High Field NMR facility Gdm-HCl guanidinum hydrochloride SNARE soluble NSF attachment protein receptor HEPES 4-(2-hydroxyethyl)-1- SUMO small ubiquitin like modifier piperazineethanesulfonic acid HSQC heteroquantum single coherence transfer SUR sulfonylurea receptor I spin quantum number TAP transporter associated with antigen processing ICD intracellular domain TM transmembrane INEPT insensitive nuclei enhanced by TMD transmembrane domain polarization transfer IPTG β-D-1-thiogalactopyranoside TNP 2,4,6-trinitrophenyl

KATP ATP-sensitive potassium channels TROSY transverse relaxation optimized channels spectroscopy KCOs potassium channel openers Ulp1 ubiquitin-like-specific protease 1 Kir subunit inwardly rectifying potassium subunit UV ultraviolet

xv 1

Chapter 1 Introduction

1 Overview of K ATP channels

ATP-sensitive potassium (K ATP ) channels were first discovered in cardiac myocytes by Akinori Noma in 1983.202, 218 These channels have since been identified in various biological tissues 203 including the heart, brain, pancreas, kidneys and . KATP channels are involved in the efflux of potassium ions, and their activity is reflective of the metabolic state of the cell.

Thus, K ATP channels couple the membrane potential to the energetic state of the cell leading to several medically and physiologically relevant processes, including cardioprotection as well as insulin and neurotransmitter release. 203 Dysfunction of these channels, as a result of genetic mutation, has been shown to be accountable for diseases such as diabetes mellitus, familial hyperinsulinism, and several cardiovascular disorders. 203

1.1 Molecular Architecture of K ATP channels

KATP channels are hetero-octameric complexes with a combined molecular weight of ~950 kDa.

Molecular cloning and reconstitution experiments have demonstrated that active K ATP channels are comprised of four copies of an inwardly rectifying potassium (Kir) subunit and four copies of a sulfonylurea receptor (SUR) subunit. 51, 282 The four Kir subunits form a central pore that allows for potassium ion efflux. Each of the four Kir subunits are associated with a SUR subunit that serves to regulate channel activity. Low resolution electron microscopy (EM) images of the K ATP channel have provided dimensions of 18 nm in diameter and 13 nm in height. 202 A schematic diagram of the subunit organization based on 18 Å EM images 202 is depicted in Figure 1.1. Top View Side View SUR subunit

Kir subunit

K+ pore

Figure 1.1: Schematic representations of EM images of the K ATP channel. The red and blue show the SUR and Kir subunits respectively. Modeling of canonical ABC domains was performed with the structure of P-glycoprotein. The SUR proteins contain an additional membrane spanning domain 0 (MSD0, shown in yellow) that participates in channel regulation, but is not part of the canonical ABC structure (see section 1.3.2). The MSD0 domain was modeled against alpha helix bundles. Adapted with authorization from EMBO J. Mikhailov et al. 2005. 24 (23): p. 4166-75 (Reference 202 ). Copyright © 2005, European Molecular Biology Organization.

2

1.2 Overview of the Kir subunit

Kir subunits are potassium selective ion channels that conduct current more favourably in the inward direction (into the cell) than the outward direction. 28 The Kir channels are classified into 82 seven subfamilies on the basis of phylogenetics. The pore-forming subunit in KATP channels are members of the Kir6 subfamily (Kir6.1 or Kir6.2, depending on tissue localization). Kir6.1 is generally restricted to vascular smooth muscle, whereas Kir6.2 has a more ubiquitous distribution and is found in the heart, liver and pancreas. Remarkably, Kir6.1 displays only half of the activity of Kir6.2 despite having ~71 % sequence identity. 240 Note that contrary to their name, members of the Kir6 subfamily are only weakly inwardly rectifying and allow for efflux of potassium ions from the cell (see below).2, 131

High resolution structures of prokaryotic KirBac1.1 161 and mammalian Kir3.1 216 have been obtained, as well as other related potassium ion channels such as bacterial KcsA 309 and MthK channels. 29, 333 The general architecture for the Kir channel family is shown in Figure 1.2. A single monomer consists of two membrane spanning domains that are flanked by cytosolic amino (N)- and carboxy (C)-terminal domains. 58 The membrane spanning domains (outer M1 helix and inner M2 helix) are linked by a highly conserved extracellular P-loop. The P-loop

Side view of Kir dimer Top view of Kir tetramer

Figure 1.2: Models depicting the side/top view of Kir6.2 association in the K ATP channel. The membrane spanning domains from two different subunits are depicted in the side view. For clarity, the cytoplasmic domains of the other subunits are also depicted. The M1 and M2 domains are embedded in the membrane while the slide helix interacts at the bilayer-cytosol interface. Four Kir subunits associate to form a central pore for potassium ion permeation as shown in the top view. Adapted by permission from Macmillan Publishers Ltd: Nat. Rev. Neurosci. Bichet et al Merging functional studies with structures of inward-rectifier K + channels, Copyright © 2003 (Reference 29 ).

3 forms part of the selectively filter for potassium ions with the conserved motif, TXGY/FG. An active potassium channel is generated as a tetramer of Kir subunits.262 The P-loop, M2 helix and intracellular C-terminus from each of the four subunits contribute to the formation of the central pore (Figure 1.2). The selectivity for potassium ion permeation results because residues in the selectivity filter interact with the K + ions to mimic the hydration shell. 16 Modeling and structural analysis have illustrated that the bond lengths for potassium ion coordination to the carbonyl- oxygen atoms of the amide backbone closely match the size of the hydration shell for K + ions. 115, 219 Coordination of ions of different atomic radii, such as smaller sodium ions, is thermodynamically disfavoured as the selectivity filter from each subunit would have to substantially contract to accommodate the smaller solvation envelope of Na +.219

In addition to the M1 and M2 helices, there is a short helix known as the slide helix between the N-terminus and the M1 domain. This helix is amphipathic and sits at the interface of the cytoplasm and lipid bilayer. 195 The slide helix is thought to be responsible for ligand transduction. Studies of patients with mutations in the slide helix (eg. Andersen syndrome) have shown that there are diminished interactions of the slide helix with the C-terminal domain. 74 These mutations reduce the extent of channel activation and inward rectification.

Inward rectification by Kir channels is achieved by the presence of endogenous, small, positively charged molecules, most significantly intracellular Mg 2+ and polyamines. 138 These molecules can bind to the cytosolic N- and C- terminal domains of the Kir subunits and prevent re-binding and reverse passage of the potassium ions. 153 Electrophysiology and mutation studies have indicated that different residues in the cytosolic domains of Kir subunits can influence the degree of inward rectification for the channel. 138 As such, Kir channels can be classified as strong, weak or intermediate depending on the extent of inward rectification. 153 Furthermore, the

Kir6 subfamily that is employed in K ATP channels, is weakly inwardly rectifying and will allow current to flow at positive membrane potentials. One residue in particular, Asp172 in the M2 helix of a strong rectifier, Kir2.1, is decisive in determining the level of inward rectification for several Kir channels. 290 Mutation of this residue to an Asn (as found in the weakly rectifying Kir1.1 channel) can reverse the strong activity. 112 This site is referred to as the D/N site and modifies the affinity of the Kir subunit for Mg 2+ and the subsequent potassium ion rectification. 112

4

Several Kir channels are also regulated by specific protein-protein interactions. For instance, cytosolic domains of the Kir3 channel, which are involved in signaling pathways, are activated via interactions with the Gβγ subunits of G-proteins. 339 Kir2.3 channels, which are strong rectifiers that are important for membrane excitability in the nervous system, are regulated by a number of protein-protein interactions including CASK 217 (scaffolding protein with homology to calmodulin-dependent protein kinase), LIM protein 172 (actin binding protein) and SAP-97 173 (synapse associated protein). Kir4.1 channel activity is enhanced in the presence of 122 the anchoring protein complex PSD-90/SAP90. In K ATP channels, which are the focus of this study, the Kir6.1 and Kir6.2 subunits are regulated by the sulfonylurea receptor (SUR) proteins.

1.3 Overview of the SUR subunit

The sulfonylurea receptors (SURs) are members of the ATP-binding cassette (ABC) superfamily. ABC transporters couple the energy derived from ATP hydrolysis to the vectorial transport of substrate across a membrane. 250 The SURs are atypical ABC proteins in that they lack any known intrinsic transport capacity, and function only as regulators of the Kir subunits. 209 However, they are still classified as ABC transporters based on structural and sequence similarity to members of the ABC family. Consequently, knowledge of the structure and regulation of other ABC transporters provides a basis for studies of the SUR proteins.

1.3.1 Overview of the ABC transporters

Transport of solutes across a membrane is a highly important and tightly regulated process. In E. coli 10 % of the genome is dedicated to transporter proteins with roughly 5 % encoding ATP- binding cassette (ABC) transporters. 250 ABC transporters are found in all kingdoms of life and have various roles such as osmoregulation, nutrient transport and DNA repair. In , all ABC proteins with transport capacity function as effluxers. There are approximately 6000 known ABC transporters, which make them one of the largest and most ancient protein superfamilies. 59 In the , up to 48 ABC transporters have been identified to date. 313

Proteins are identified as ABC transporters through sequence similarity of the gene cassette and organization. At minimum, ABC proteins are comprised of two membrane spanning domains (MSDs) and two nucleotide binding domains (NBDs). In humans, ABC transporters are subdivided into seven subfamilies (A-G) based on phylogenetic analysis of

5 the NBDs which, as their name implies, bind and hydrolyze ATP.313 Table 1.1 summarizes key properties of the human ABC subfamilies. The sulfonylurea receptor proteins (SUR1 and SUR2 isoforms) are members of the C-subfamily, which also include the cystic fibrosis transmembrane conductance regulator (CFTR), and nine multidrug resistance proteins (MRPs). CFTR is an ion channel that allows for chloride efflux and is well known because specific mutations in the CFTR gene cause the autosomal recessive disease, cystic fibrosis. 95 The MRPs are involved in secretion of toxins, small molecules and ions and are known for imparting drug resistance to cancer cells. Generally, members of the same subfamily, have similar biochemical properties. Studies have shown that substrates of the MRP family, such as verapamil, can also target the 215 KATP channel to modify activity. Conversely, drugs that target the K ATP channel such as glibenclamide, are also active on the ABCC subfamily, MRPs and CFTR. Notably, CFTR has also been elucidated to have an analogous role to the SURs in regulating ion channels, and has also been speculated to form active channels with Kir6 subunits as well.277 Table 1.1: Subfamilies of Human ABC Transporters Subfamily Colloquial Subfamily Names Number of Generalized function ABCA ABC1 12 Cholesterol/fatty acid transport ABCB MDR (Multiple drug resistance) / 11 Bile/peptide transport TAP (Transporter associated with antigen processing) ABCC MRP (Multidrug resistance protein) 12 Ions/toxin efflux ABCD ALD (Adrenoleukodystrophy) 4 Peroxisomal transport ABCE OABP (OligoA binding protein) 1 ABCF GCN20 (General control 3 Gene expression nonderepressible) ABCG White 5 Anion/peptide/sterol efflux

1.3.2 General Structure of ABC transporters

ABC proteins are responsible for the translocation of a remarkable diversity of substrates. Despite the vast assortment of substrates transported, ABC proteins possess a universally common architecture. The first high resolution (3.2 Å) ABC transporter structure was solved by 186 Locher et al in 2002 of the E. coli vitamin B 12 importer. Additional structures of both prokaryotic and eukaryotic full length ABC transporters have since been solved and contributed to the wealth of biochemical and biophysical data of ABC proteins. Crystallographic data have provided further evidence that the canonical ABC structure is comprised of two membrane spanning domains (MSD1 and MSD2) and two nucleotide binding domains (NBD1 and NBD2) which are illustrated in Figure 1.3.61 The MSDs serve to recruit and bind the specific substrates for transport, whereas the NBDs bind and hydrolyze ATP which is thought to provide the power-

6 stroke to facilitate substrate translocation. Each eukaryotic MSD consists of six α-helices, although sequence similarity between different ABC transporter MSDs is typically low. This is likely to accommodate the diversity of target substrates for each transporter. 61, 187 Crystal structures of ABC proteins demonstrate that the helices of the MSDs extend beyond the lipid bilayer. These long transmembrane helical extensions are connected to each other by short helices, termed coupling helices (Figure 1.3). Together the cytoplasmic portion of the helices in the MSDs and the coupling helices form intracellular domains (ICDs). The coupling helices associate with the NBDs thereby providing a tangible link between the MSDs and NBDs. This physical connection may serve as a possible pathway through which conformational changes from ATP binding and hydrolysis at the NBDs are transmitted to the MSDs.61, 187

1.3.2.1 Arrangement of domains in ABC proteins

The arrangement of MSDs and NBDs varies between different ABC transporters. 187 Some prokaryotic ABC transporters encode and express each of the four domains as independent

Extracellular

Membrane Spanning Domains

Intracellular Intracellular Domains

Coupling Helices

Nucleotide Binding Domains

Figure 1.3: Structure of ABC transporter, Sav1866, showing the canonical domains in different colours. The MSDs are depicted in red and green, and the NBDs are shown in blue and yellow bound to ATP molecules (orange). Prokaryotic importers also possess a binding protein. In eukaryotic transporters, the MSDs extend into the cytoplasm to form the ICDs, and contact the NBDs by the short coupling helices. Adapted with authorization from Microbiol Mol Biol Rev. Davidson et al . Structure Function and Evolution of Bacterial ATP-binding Cassette Systems Copyright © 2008. (Reference 61).

7 polypeptides that co-assemble into an active transporter. Prokaryotic ABC transporters have been identified as both importers and exporters. Importer proteins possess an additional soluble domain, known as a binding protein (BP). 116 In Gram negative bacteria, the binding protein exists in the periplasm between the inner and outer cell membranes, where it functions to recruit substrate and shuttle it to the ABC protein for transport. In Gram positive bacteria, the binding protein is often anchored to the lipid membrane, although in some cases it may be tethered directly to the ABC transporter. 116

Eukaryotic ABC transporters solely exist as exporters and the four domains are typically encoded onto a single polypeptide chain (full transporter), and in some cases, two polypeptides (half transporter). 116 The organization of domains in a full ABC transporter typically consists of

(NH 2-MSD1-NBD1-MSD2-NBD2-COOH) which forms the template for the majority of eukaryotic ABC transporters. A half-transporter consists of a dimer of two polypeptide chains with the genomic format of (NH 2-MSD-NBD-COOH). These half transporter systems require proper assembly of the dimer in order to achieve transport function, highlighting the relevance of each domain and their interconnected activity. A functional ramification of this variation in domain organization is that transporter assembly may occur to create homodimers (identical MSDs and NBDs) or heterodimers (different MSD1/MSD2 and NBD1/NBD2). The SUR proteins encode all MSD and NBD domains onto a single polypeptide chain. In humans, half transporters are generally found in the ABCG subfamily. 313

Kir SUR

ABC Transporter

Figure 1.4. Schematic representation of domains in the Kir and SUR subunits. As a member of the long ABCC subfamily, the SUR proteins have an additional TMD0 (also referred to as MSD0) which is shown in yellow. Adapted by permission from Physiol Rev. Flagg et al (2010) Jul 90(3):799-829 (Reference 90 ).

8

1.3.2.2 Subdivisions in the ABCC subfamily

The C subfamily of human ABC transporters is further distinguished as 'long' or 'short' transporter units. This is because in addition to the canonical ABC structure of two MSDs and two NBDs, long members possess an additional membrane spanning domain (MSD0) which is N-terminal to MSD1. 209 MSD0 is coupled to the rest of the ABC protein through a cytoplasmic linker known as the L0 linker. The SUR proteins (both SUR1 and SUR2 isoforms) are classified as long members, as are multiple multidrug resistance proteins (MRP1, MRP2, MRP3 MRP6 and MRP7). A schematic of the SUR protein is illustrated in Figure 1.4. Short members of the ABC proteins include the remaining proteins (CFTR, MRPs) and possess the canonical ABC structure of two MSDs, two NBDs and a long N-terminal cytoplasmic extension. 209

1.3.2.3 Structure of the Membrane Spanning Domains (MSDs)

The hydrophobic membrane spanning domains contain bundles of alpha helices and are generally classified into three separate topologies (type I importer, type II importer or exporter). The arrangement of helices in each of these distinct topologies is depicted in Figure 1.5. Type I Importer Type II Importer Exporter

Figure 1.5: TM arrangement in the MSDs of different ABC transporters. There are three types of transmembrane arrangements in ABC transporters. The top panel highlights the side view of the helices embedded in the membrane. The bottom panel provides a bird`s eye view of the numbered helices in the same membrane spanning domain. Adapted by permission from Macmillan Publishers Ltd: Nat. Rev. Mol. Cell. Bio. 10,218-227, Copyright © (2009) (Reference 250 ).

9

In the Type I importer fold, there are generally five helices in each MSD. The helices are arranged with TM2-5 (transmembrane helices 2-5) forming the circumference of the central pore and TM1 encircling the four helices. 250 The number of helices is also variable with Type I ABC importers, such as ModB and MalG, having six helices in each domain. In the Type II importer fold, there are 10 helices per MSD, as seen with Haemophilus influenzae , HI1471, with a tight packing of the helices. 250 TM2 is positioned towards the center of the pore and provides contacts with the other helices. In the eukaryotic exporter fold, there are six helices for each MSD. Based on the Sav1866 structure, the helices are partitioned into two subdomains composed of helices from each subunit. Each subdomain consists of TM1-2 from one monomer, and TM3-6 from the second monomer. 250

The exporter fold is further distinguished by the connections between the helices. The helices are linked to each other on the extracellular side by short loops. These loops are available for various purposes such as cell signaling. For example, eight glycosylation sites have been identified on the extracellular loops in ABCA4. 34 ABC proteins can also vary in the length of the extracellular loops as ABCA13 possesses an approximately 2500 amino acid extracellular loop between TM1 and TM2, and is the largest known ABC protein. 238, 304 On the intracellular side of eukaryotic proteins, the TM extend approximately 25 Å into the cytosol to form the ICDs. In both exporter and importer folds, the MSDs and NBDs communicate through the coupling helices. In importers, coupling helices contain a Glu-Ala-Ala motif with structurally analogous regions in exporters contacting the MSD and NBD interface. 304

1.3.2.3.1 The role of MSD0 and the L0 linker

The SUR proteins contain an additional N-terminal membrane spanning domain (MSD0) that is connected to MSD1 by the short peptide, the L0 linker.209 The function of MSD0 and the L0 linker in the SUR channel has been studied extensively. Although, the role is not fully understood, it has been found that MSD0 is important for proper trafficking and regulation of the 209 full KATP channel as well as interacting with the Kir6 pore protein.

Functional assembly of the K ATP channel into a complete octameric complex is controlled by protein trafficking from the endoplasmic reticulum. The cytosolic domains of the Kir6 subunits and the SUR subunits possess ER retention signals (Arg-Lys-Arg). 42 Association of the

Kir6 and SUR subunits in the K ATP channels physically mask the ER retention signals and thus

10

prevents these signals from being detected. This allows for release of the K ATP channel from the ER and subsequent trafficking to the plasma membrane. All ER retention signals are blocked only when an octameric complex is fully assembled, and partially formed oligomers will not traffic to the cell membrane. 42 The Arg-Lys-Arg signals are distinct from other known ER retention signals and are also thought to have a similar function for the related ABC transporter, CFTR. Deletion of similar arginine rich regions in CFTR restores trafficking of CFTR misfolded variants caused by the cystic fibrosis-causing mutation ΔF508.42, 44

In addition to trafficking, the MSD0 and the L0 linker have been found to have a significant role in controlling K ATP channel activity. Electrophysiological studies using a variant of Kir6. x that lacks the 26 C-terminal residues, allows for trafficking to the cell surface without SUR subunit association. These experiments have shed light on the relevance of the individual SUR subunits.20 Intrinsic gating of the resulting isolated Kir subunits is shown to be quite low. However, in the presence of the MSD0 domain, this activity is increased 5-fold 20 (Figure 1.6). This remarkable increase is reduced when the MSD0 region is extended to include successive regions of the L0 linker as well. This reduction in activity, suggests that the MSD0 has stimulatory effects and L0 linker has inhibitory effects on Kir channel activity.20 Furthermore,

Kir6.2

P0max = 0.086 ± 0.016 ms

Kir6.2 + MSD0

P0max = 0.44 ± 0.068 ms

Kir6.2 + MSD0 + L0 linker

P0max = 0.23 ± 0.012 ms

Figure 1.6: Role of MSD0 and the L0 linker in the KATP channel. The homology model (left) suggests a critical role of MSD0 and the L0 linker in communication between the SUR and Kir subunits. Electrophysiology experiments (right) of the isolated Kir subunit show low probability of channel opening. In association with the MSD0 domain this open probability increases substantially, but is reduced by extending the MSD0 to include the L0 linker. Adapted with permission from Babenko & Bryan. J. Biol. Chem. (2003) Oct; 278(43); 41577-80 (Reference 20 ).

11

SUR proteins lacking MSD0 cannot regulate the K ATP channel. Considering that ATP binding and hydrolysis at the NBDs results in channel opening, it is likely that conformational changes in the nucleotide binding domains of the SUR proteins are likely transmitted to MSD1 and MSD2 and these changes influence the interactions of the L0 linker with the Kir subunits. Furthermore, in ABCC1, the L0 linker is found to be membrane associated. 22 When the L0 linker is truncated by 10 residues, it loses its affinity for the bilayer and transport ability of the full transporter was also eliminated. 22 This further suggests a critical role for the L0 linker in relaying conformational changes from the NBDs to the Kir subunits.

1.3.2.4 Structure of the Nucleotide Binding Domains (NBDs)

The nucleotide binding domains possess several conserved elements, and as such, are often employed for classification and characterization of ABC transporters. The conventional NBD structure has been determined from high resolution crystal structures of prokaryotic and eukaryotic full-length transporters as well as isolated NBDs. All NBDs possess two domains, lobe I and lobe II which function together in ATP-binding and hydrolysis. 251

The overall structure of lobe I harbours two well conserved subdomains. The first subdomain is a structure similar to other F 1-ATPase enzymes in which a central core of β-sheets are surrounded by α-helices and is often referred to as the α/β subdomain. There is a second subdomain in lobe I that is known as the antiparallel β-sheet subdomain and is involved in interactions with the ribose sugar and nitrogenous base during nucleotide binding. 130 Lobe I is the core domain and is typified by the characteristic Walker A and Walker B motifs. 6 The Walker A and Walker B motifs are hallmark features of ATP binding proteins and are not solely exclusive to ABC transporters. The Walker A (or P-loop, for phosphate binding) motif has a consensus sequence GXXXXGK(T/S), in which the invariant Lys coordinates to the γ-phosphate of the nucleotide. 60 The preceding XXGK residues have main-chain atoms that form a cavity that approximates the size of a phosphate ion with the amide protons facing inwards. This backbone (and not side chain) bonding to the phosphate may explain the observed variability in Walker A sequences. The Walker B motif is C-terminal to the P-loop, although conformationally close to the Walker A site, with a consensus sequence ФФФФD/E, where Ф is a hydrophobic residue. The conserved acidic residue (E/D) is thought to coordinate to the Mg 2+ ion.

12

Lobe I

Lobe II

Figure 1.7: Homology model and schematic of the CFTR NBD dimer. NBD1 is shown in green and NBD2 is shown in blue. The NBDs associate in a head-to-tail manner such that both NBDs complete the binding site around each ATP molecule. Relative positions of the important motifs, including the Walker A Lys, D-loop, Walker B glutamate and H-loop are shown around the bound ATP molecules, as well as the Walker C motif from the opposing NBD. The relative positions of lobe I and lobe II on NBD1 are also indicated. Adapted by permission from Macmillan Publishers Ltd: Nature, Gadsby et al. 440(7083); 477-483 Copyright © 2006 (Reference 95 ).

Lobe II is comprised of an ABC-specific α helical subdomain that contains the ABC signature sequence LSGGQ (also referred to as the Walker C motif). 251 This sequence is specific and often diagnostic to ABC transporters. The signature sequence assists in ATP binding and hydrolysis. Lobe I and lobe II are joined through a molecular hinge region that allows them to move during nucleotide hydrolysis. Lobe II is thought to be mainly responsible for relaying conformational changes in the NBDs to the coupling helices and thereby the MSDs.251

There are also a number of other well conserved motifs within the NBDs. The A-loop is present in lobe I and contains an aromatic residue that is approximately 25 residues N-terminal of Walker A, where it is positioned in the loops connecting the first two β-strands in the β-sheet subdomain. 7 This residue serves to create a stabilizing π-π stacking interaction with the nitrogenous base of the nucleotide. The D-loop is also located in lobe I and contains the consensus sequence, SALD, and assists in coordinating the nucleophilic water molecule.140, 237 An H-loop and Q-loop have also been identified in lobe II, which refer to conserved His and Gln residues respectively. 250 The His residue is thought to act in unison with the D-loop and/or glutamate of Walker B to hydrogen bond to the nucleophilic water molecule. 140, 341 Recently, computer simulations predicted the H-loop to be responsible for the transfer of protons between reactants in an ABC transporter, HlyB, which assists in nucleotide hydrolysis. 341 The Gln residue

13 of the Q-loop has been shown to associate with the magnesium ion, γ-phosphate and water molecule. Biochemical evidence has also implicated the Q-loop in a signaling role between nucleotide binding and transporter ability. 327

Each ABC transporter contains two NBDs which associate with each other to form a labile dimer. It is well established that NBD1 and NBD2 associate in a head-to-tail manner, so that lobe I of NBD1 interacts with lobe II of NBD2 and vice versa 250 (Figure 1.7). The NBD dimer contains two separate sites for nucleotide binding, with each site formed by both NBDs. The Walker A and Walker B motif from one NBD form half of the binding site, with the Walker C from the opposing NBD completing the binding site. 250 The spatial arrangement of the dimeric interface was observed in isolated NBDs, including Rad50 120 and MJ0796335 , but also in crystal structures of full ABC proteins, including BtuCD 186 and PgP.5

Although the NBDs possess several conserved elements across all ABC proteins, outside of these regions can be quite variable. Changes in the NBD sequence and structure can possibly affect the activity of the entire ABC protein. Therefore, studies into the SUR NBDs can provide insights into regulation of the entire K ATP channel since limited information is currently available.

1.3.3 SUR isoforms and splice variations

The SUR proteins, as previously mentioned, are members of the C-subfamily of ABC proteins and serve to regulate the pore-forming Kir6 subunit in K ATP channels. There are two SUR proteins, SUR1 (ABCC8 gene) and SUR2 (ABCC9 gene) and various splice isoforms of each of 209 these proteins. Although both SUR proteins form regulatory subunits in KATP channels, SUR1 is predominantly found in KATP channels of pancreatic cells and SUR2 is found in KATP channels

Figure 1.8: Genomic arrangement of sur and kir on and 12. The close proximity of these two genes suggests an additional level of transcriptional regulation. SUR2 exists as two splice isoforms with alternate usage of exon 38A or 38B. Adapted by permission from Physiol Rev. Flagg et al (2010) Jul 90(3):799-829 (Reference 90 ).

14 of the heart, skeletal, vascular and non-vascular muscle. SUR1 is comprised of 39 exons and is found on the short arm of chromosome 11, at 11p15.1. SUR2 is slightly shorter (38 exons) and is found on the short arm of , at 12p11.12. The shorter length of SUR2 compared to SUR1, is the result of a pseudo-exon deletion (exon 18) which corresponds to a region in NBD1 following the Walker A motif.90

Sequencing of SUR2 mRNA transcripts from various tissues suggests alternative splicing of the SUR2 gene. The two most common splice isoforms are SUR2A and SUR2B which result from alternate usage of the final exon. In the SUR2 gene, the last two exons each correspond to 42 residues and are referred to as either 38A or 38B. SUR2A uses exons 1-38A, whereas SUR2B employs exons 1-37 and 38B 174 (Figure 1.8). Thus, SUR2A and SUR2B differ only in the final

42 residues of the protein. This single difference in exon usage has significant influence on K ATP channel activity and pharmaceutical sensitivity. For example, K ATP channels with SUR2A are not sensitive to the potassium channel opener, diazoxide. However, the drug is active on SUR2B containing channels suggesting a crucial role for the final exon in the SUR2 protein. 332

Additional splice variants of the SUR2 gene have also been observed. Minor levels of mRNA transcripts containing deletions of exon 14 from SUR2A have also been recovered from cardiac tissue (named SUR2C). RNA transcripts of SUR2A lacking exon 17 (referred to as SUR2D) have been identified in cardiac tissue and in the kidneys and have shown to have lower MgATP-induced gating compared with SUR2A.50 The functional significance of each splice variant has not been fully characterized as well as the possibility of the corresponding variant in the SUR2B background. Several splice variants have also been observed in SUR1. 174 Table 2 summarizes various splice variations identified for the sur2 gene.

The genes for both SUR and Kir subunits are located on the same chromosome. It has been speculated that these genes were once linked since recombinantly-combined SUR and Kir proteins have been shown to be active. 174 The spatial proximity of both sur and kir genes may

Table 1.2: Splice variations in the sur2 gene 174, 340 Splice Variant Exon Usage Tissue Localization SUR2A 1-38A Cardiac Tissue/Smooth Muscle SUR2B 1-37, 38B Vascular Muscle SUR2C 1-13, 15-38A Cardiac Tissue SUR2D 1-16, 18-38A Cardiac/Kidneys SUR2A (Δ17,18) 1-16,19-38A Cardiac Tissue

15

also suggest one possible method of regulation for K ATP channels at the transcriptional level 132 although conclusive evidence has not been obtained. Nevertheless, regulation of the K ATP channel by nucleotide and pharmaceutical compounds has been thoroughly investigated.

1.4 Nucleotide regulation of K ATP channel activity and kinetics

Regulation of K ATP channels is a complex and intricate process managed by multiple protein domains and small molecules. Overall, K ATP channels operate under an intrinsic gating mechanism,84 referring to the observation that these channels are active (open) in the absence of any nucleotide or ligand. This activity can be measured via electrophysiology and is viewed as several short bursts of membrane depolarization and recovery, spaced between relatively longer inactive periods.

KATP channel activity is elaborately modulated by intracellular nucleotide 84 concentrations. KATP channels possess nucleotide binding sites on each Kir and SUR subunit, with binding at each site resulting in contrasting processes. ATP binds to the Kir subunit in a Mg 2+ -independent manner, resulting in channel closure, and consequent membrane depolarization (Figure 1.9). Other , such as ADP, GTP, GDP are also shown to elicit channel closure, albeit with a reduced response, suggesting that electrostatic interactions of the triphosphate predominate at the affinity pocket of the Kir subunit. 84, 303

Binding of Mg-ADP at the SUR protein leads to loss of affinity for nucleotide at the Kir subunit. 209 The loss of binding at the Kir subunit, results in channel re-opening and re-stimulates

Closed (ATP at Kir) Open (MgADP at SUR) K+

ADP

ATP ATP ADP

ATP ATP

Figure 1.9: Different conformations of the K ATP channel. ATP can bind to the Kir subunit which results in channel closure. Alternatively, MgADP can bind at the SUR NBDs which results in ATP dissociating from the inhibitory sites on the Kir subunit and the channel re-opening. Adapted by permission from Physiol Rev. Flagg et al (2010) Jul 90(3):799-829. Modified from reference 90 .

16

KATP channel activity. The presence of MgADP is achieved either through direct binding of the nucleotide or as a consequence of SUR NBD hydrolysis of bound MgATP (Figure 1.9). Thus,

KATP channels act as acute sensors of cellular energetics by responding to relative intracellular concentrations of ATP and ADP. Generally, studies have shown that high concentrations of ATP result in channel inhibition (channel closure) whereas low concentrations result in stimulatory effects (potassium ion efflux). 209

The octameric organization of K ATP channels introduces a further level of channel regulation as nucleotide can potentially bind in different permutations to each of the four different Kir and SUR subunits in a K ATP channel. As such, single K ATP channel kinetics are highly complex and less understood, although three different models have been proposed to explain observed channel activity (Figure 1.10). In each model, the K ATP channel is considered to function as a single ‘gating unit’ or as a composition of various ‘gating units’. 53 A gating unit refers to the minimum complex that can be switched to the open or closed conformation.

In the Hodgkin and Huxley Model, the individual Kir subunits act as separate gating 53 units, and thus a K ATP channel is comprised of 4 independent gating units. If any of the gating

Monod -Wyman -Changeux Model

Open Closed

Hodgkin and Huxley Model Dimer of Dimers Model

Open Closed

Open Closed

1 Gating Units × 4 Subunits

2 Gating Units × 2 Dimers

1 Gating Unit

Figure 1.10: Different models of nucleotide regulated gating of the K ATP channel. Each circle/square represents a single gating unit. The circles represent a gating unit in the open conformation, whereas the square depict a closed conformation. The models differ in the number of gating units in the K ATP channel Adapted by permission from: Craig et al J. Gen. Physiol. Jul 2008 132(1); 131-44 (Modified from reference 53 ).

17 units exist in the closed conformation it will disrupt the integrity of the central potassium pore, and therefore the entire channel will be impermeable to potassium ions. Consequently, a single ATP binding event at any of the Kir subunits will result in complete channel closure. As a result, MgADP binding will be required at each SUR subunit in order to stimulate activity. 53 This model was elaborately tested by Enkvetchakul et al by examining channel kinetics of Kir6.2 subunits, 83 with and without the SUR1 isoform. These studies suggested that the KATP channel samples various states during activity, which include transitions through an open state, to a short lived closed conformation and then to a longer closed conformation. The transition from the open state to the short lived closed state is thought to be ATP-independent. However, the initial short closed state can be converted into the longer lived 'closed' state by the presence of ATP.178 The presence of the SUR protein favours the open state as well as presence of a known K ATP regulator and phospholipid, phosphatidylinositol phosphate (PIP).179

In the Monod-Wyman-Changeux Model, the full K ATP channel functions as one gating unit. Each ATP binding/dissociation event is cumulative towards a total free energy of the channel. 53 Both the open and closed states have a minimum energy threshold that would lead to

KATP channel activity or inhibition. Studies by Craig et al employed K ATP channels with the individual subunits mutated to yield defective ATP binding sites. 53 The channels were assembled with a known number of mutated or wild type subunits and the resulting channel kinetics suggested that ATP binding supports the Monod-Wyman-Changeux mathematical model.

The data from these studies as well as others,1, 270 also suggest that at least two ATP binding events are required to allow for channel closure. These data as well as computer modeling studies have suggested that the channels function as a series of two gating units (dimer of dimers models).53 Under this model, a gating unit is comprised of two subunits (either adjacent or opposing Kir proteins, as this is currently unknown) which is referred to one of two possible dimers. Both dimers (gating units) operate independently, and each gating unit must be in the open conformation in order for channel activity to be observed. The different conformations of each dimer are achieved by ATP binding to each monomer. Recent evidence for this model has also been accumulated by Hosy et al where the stoichiometry for K ATP activation by ADP was recorded in K ATP channels expressed in Xenopus oocytes and supports the dimer of dimers model.123 Although a definitive model has not be determined to explain all

18

the observed data, in each model, KATP channel activation is dependent on relaying changes in ATP / ADP concentrations to the direct microenvironment of the channel.

1.4.1 Phosphotransfer Networks

KATP channels are responsive to changes in nucleotide concentrations. However, the physiological presence of a true apo-state is disputed as millimolar intracellular concentrations of nucleotide are present in the cell, even during periods of metabolic stress.80, 81 Furthermore, studies have shown that intracellular ATP concentrations under all metabolic states, are 10-100 times greater than the expected values for inhibitory nucleotide concentrations. Therefore the

KATP channel should remain closed under all physiological conditions. Even increases in intracellular ADP are not shown to be of enough significance to elicit a change in channel activity. It has been postulated, that the presence of phosphotransfer networks in different cell types may adjust the microenvironment of the K ATP channel and allow for proper transduction of 80 81 the metabolic state of the cell to the KATP channel (Figure 1.11).

Figure 1.11: Phosphotransfer networks. Phosphotransfer networks allow for rapid dissemination of metabolism signals throughout the cell. Different enzymes, spatially distributed throughout the cell, cycle through reaction substrates to rapidly transfer information regarding changes in substrate concentration. In certain tissues, adenylate kinase (AK), creatine kinase (CK) or other phosphokinases (PK) are thought to transfer the metabolic signals to the KATP channels, This cycling allows for changes to the microenvironment of the K ATP channel to be rapidly sensed. Adapted by permission from Physiol Rev. Flagg et al (2010) Jul 90(3):799-829. Modified from reference 90 .

19

Phosphotransfer pathways are distinct from phosphorylation signaling cascades and provide strict spatial control with temporal regulation.81 In phosphotransfer pathways, a cellular distributed network of enzymes transfer phosphoryl groups between different substrate molecules. 163 A commonly identified enzyme in the phosphotransfer pathway is adenylate kinase which catalyzes the reaction 2 ADP → ATP + AMP. A series of adenylate kinase enzymes are located throughout the cell, referred to as phosphoryl wires. These enzymes transfer phosphoryl groups between the substrates successively along the phosphotransfer wires.81 The phosphotransfer reactions propagate throughout the cell in a manner analogous to a wave. In this way, ligands (ATP/ADP/AMP) are not required to passively diffuse throughout the entire cell to their final destination in order to exert their effects. 81 The phosphotransfer pathway has been identified as a highly efficient method of transferring metabolic signals throughout the entire cell as it does not rely on diffusion rates which would require large concentration gradients of ATP and ADP in order to be kinetically feasible in the cell. 163 Other organisms, such as cyanobacteria also show evidence for ATP and NADPH cycling. 65 In addition to adenylate kinase (AK), creatine kinase (CK) and other phosphokinase enzymes have also been implicated in phosphotransfer networks. Furthermore, these enzymes are found at or near the K ATP channel and have also been shown to associate with the channel through co-precipitation studies. Moreover, the phosphotransfer rates of these enzymes have been found to be comparable to the rates of 80 KATP channel activity.

In vivo studies into the phosphotransfer network connectivity with K ATP channels have also implicated a role for adenylate kinase and creatine kinase in channel activity. 80 Transgenic mice that are lacking adenylate kinase have shown to have compromised K ATP channel regulation.37 Moreover, these mutations could be rescued by upregulation of creatine kinase, another phosphotransfer enzyme. NMR studies with 18 O and 31 P NMR exchange have demonstrated that under hypoxia, there is an increased rate in adenylate kinase activity and 79 phosphoryl group cycling which is coupled to increases in KATP channel activity. Co- immunoprecipitation studies have also shown that creatine kinase associates with the K ATP channel.54 Creatine kinase mediated phosphotransfer pathways are thought to allow for the precise metabolic state of the cell to be signaled to the K ATP channel to induce the proper protein conformation and thereby, biological response.

20

1.5 Physiological Role of K ATP channels

KATP channels are present in metabolically active cells, and their activity is influenced both by the cell type and metabolic requirements. Table 1.3 shows the presence of various K ATP channels and subunit composition. Gene knockout studies of Kir6.1 and Kir6.2 in mice have phenotypes of angina as well as sudden death syndrome. Similar phenotypes are also observed with gene knockouts of SUR. 204, 293

Table 1.3: Subunit assembly of K ATP channels based on tissue localization Tissue SUR Subunit Kir Subunit Pancreas SUR1 Kir6.2 Brain SUR1 Kir6.2 Heart SUR2A Kir6.2 Smooth Muscle SUR2A Kir6.2 Vascular Smooth Muscle SUR2B Kir6.1

The role of K ATP channels is well understood in the islets of the pancreas where they 15 serve to regulate insulin release (Figure 1.12). Under resting conditions, the K ATP channel remains open, allowing for release of potassium ions and maintenance of a negative membrane potential. An increase in blood glucose levels stimulates ATP production and increases the relative concentration of ATP to ADP. The ATP binds to the Kir subunit which leads to channel closure and subsequent membrane depolarization. This depolarization event triggers the opening of voltage-gated Ca 2+ channels which leads to calcium ion influx. The resulting increase in intracellular calcium ions prompts a cascade of metabolic signals that lead to insulin release. 15

Mutations within the K ATP channel can lead to hyperglycemia. For example, transgenic mice

Figure 1.12: Role of K ATP channels in insulin release and metabolism in a pancreatic cell. In the resting state of pancreatic cells, the channel is open and the membrane is hyperpolarized. Under high glucose conditions, ATP binds to the Kir subunit and closes the channel resulting in membrane depolarization. This depolarization triggers calcium gated channels to open which stimulates insulin release. Adapted with permission from Ashcroft. Am J Physiol Endrocrinol Metab. 2007 Oct. 293 (4) (Reference 15 ).

21 with mutations in Kir6 close to the inhibitory ATP-binding site, show the same phenotype as mice with glycolytic enzyme mutations. In both cases, insulin is produced in the pancreas, however the K ATP channel remains open resulting in nominal insulin release and hyperinsulinism.252

A similar role for the KATP channel has been defined in cardiac tissue. Under normal conditions, the high ATP levels maintained by the heart allows for the K ATP channel to be inhibited (in contrast to the resting state of an open channel in pancreatic cells).91 Depletion of energetic resources results in the presence of MgADP at the SUR NBDs and subsequent channel opening and membrane polarization. This polarization event limits the influx of calcium ions and 91 prevents muscle motility especially during periods of stress. The opening of K ATP channels also shortens action potentials and is therefore cardioprotective during periods of ischemic insult.

Cardiac K ATP channels have also been recognized to have a major role in ischemic preconditioning. For instance, dog myocytes lacking K ATP channels cannot withstand long 334 ischemic periods without cardiac arrest compared to dogs expressing normal KATP channels. 141

The opening of one percent of all membrane KATP channels has been shown to be adequate to elicit a spatial and temporal response in membrane potential.343 This therefore suggests an overcapacity in potential K ATP channel activity in cardiac tissue. It has been suggested the evolutionary response for a large number of cardiac K ATP channels is to maintain homeostasis during periods of elevated stress and increased metabolic activity. This method requires higher than usual cardiac activity and it is thought that K ATP channels can provide steady cellular energetics during stress periods.343 Furthermore, under periods of high exercise, transgenic mice lacking K ATP channels cannot perform as well, coupled to sudden death 142 syndrome. It should also be noted that in the reverse case, significant overexpression of K ATP channels in the transgenic mice also results in heart failure and (100 %) embryonic lethality.301

As such, the hypothesis detailing the involvement of KATP channels in elevated homeostasis is still under investigation.

KATP channels are also found in the kidneys, brain and smooth muscle. Their regulation is less studied but is thought to be similar to the pancreatic and cardiac cells. The resting state of the channel mainly depends on the membrane potential of the corresponding cells and available

22

cellular energy metabolites. Energy metabolites can also influence KATP channels indirectly through affecting gene expression. 47 Studies in mice have shown that during periods of starvation and low glucose, K ATP channel mRNA transcripts are less apparent. The transcriptional regulation of K ATP channels is less understood and adds another layer of complexity to the various methods of channel regulation. 233

1.6 Additional Regulation of KATP channels

1.6.1 Pharmaceutical Regulation

The pharmacology of K ATP channels and the SUR subunit is directly tied to their history and their initial discovery is based on serendipitous events. 191 In 1942, Marcel Jambon in France was studying the treatment of typhoid disease and observed that one compound of interest, sulfonamide 2254RP, led to hypoglycemia (excess insulin production).190 This outcome is even more remarkable considering that excess insulin production was also observed in undernourished patients. Further studies delving into the effects of this compound, led the researchers to localize the major effects to the β-cells of the pancreas and to suggest sulfonylureas/sulfonamides as possible therapeutic agents for patients with insulin deficiencies. Currently, there are seven commercialized sulfonylureas for treatment of hypoinsulinism, such as that found in diabetes. 209 Several have also been in use prior to their understanding of K ATP channels as targets.

The mechanism of action for various drugs that target K ATP channels is diverse and not well understood. Generally, KATP channel active compounds are classified into two categories, sulfonylureas and potassium channel openers (KCOs). Sulfonylureas exert their action by closing the K ATP channel and have a parent sulfonylurea structure. KCOs are not chemically related to each other and as their name suggests, lead to membrane depolarization.209 Structures of common sulfonylureas and KCOs are shown in Figure 1.13.

1.6.1.1 Sulfonylureas

Sulfonylureas are commonly prescribed for control of type II diabetes where they act to stimulate insulin production. 209 Due to rapid clearance from the body, many of these drugs need to be administered immediately prior to food intake. Second generation sulfonylureas are shown to be more potent in their activity on K ATP channels than their predecessors, although there is no formally recognized structural basis between the different drugs.

23

Sulfonylureas bind to two distinct sites on K ATP channels, a low affinity site on the Kir subunit, and a high-affinity site on the SUR protein. The low affinity site is not thought to have 239 any medical relevance. The stoichiometry of sulfonylurea interaction with the K ATP channel is not definitively described as tritium-labelled glibenclamide (sulfonamide) studies from different groups, have indicated both quantitative (4 site) binding and minimal (1 site) binding is sufficient to induce channel activity. This apparent difference may be a result of relative concentrations of nucleotide and drug solutions between the different experiments. 239

Generally, sulfonylureas have a 100-1000 fold greater affinity for SUR1 compared to SUR2 isoforms.16 This difference is thought to be localized to the substitution of Ser1237 in

Potassium Channel Openers Sulfonylureas and Other Potassium Channel Closers

Diazoxide Glibenclamide

Cromakalim Meglitinide

Pinacidil Rapaglinide

P1075 Tolbutamide Figure 1.13: Chemical structures of compounds classified as either potassium channel openers or sulfonylurea drugs.

24

SUR1 to a Tyr residue in SUR2. Ser1237 is located between TM15 and TM16 of MSD2.16 Mutation of Ser1237 eliminates tolbutamide binding although binding of other sulfonamides is not impacted to same degree. Recombinant studies have also shown that TM5 and TM6 may also have a role in defining the sulfonylurea binding pocket as these regions have shown to be critical for glibenclamide binding as well. 239

The mechanism of interaction of sulfonylureas with K ATP channels has been well studied in the absence of and presence of nucleotide. Sulfonylureas, such as gliclazide, inhibit activation of the channel. However, it is unclear whether this inhibition is the result of displacement of nucleotide binding to the SUR NBDs by the sulfonylurea or because of an allosteric effect caused by sulfonylurea binding that blocks the effects of MgADP at the SUR protein from being conveyed to the channel. Radiolabelled nucleotide studies have suggested displacement of MgATP from the SUR NBD1 protein upon drug binding, suggesting competitive inhibition.239 This is also supported by the fact that there is a reduction in drug potency when MgADP is present in drug binding studies. It should also be noted that despite saturating conditions for sulfonylurea binding, the channel does not reach greater than 80 % of full inhibition. This suggests that drug binding only reduces the probability for channel opening and without blocking 239 KATP channel activity completely.

Although sulfonylureas can be prescribed for diabetes or similar conditions, they also have certain side effects including the potential to be teratogenic. Additionally, the FDA advises that they may increase risk to cardiovascular disease due to their interaction with K ATP channels in the heart. However, studies by ADVANCE (Action in Diabetes and Vascular Disease) have not been able to identify a strong correlation in these side effects.207, 228

1.6.1.2 Potassium channel openers

Various potassium channel openers (KCOs) have been identified to date and they exert different effects on K ATP channels based on SUR subunit composition. The net result of KCO stimulation of K ATP channels is a reduction in calcium ion influx via hyperpolarization of the cell and subsequent blockage of the Ca 2+ gated channels. Since investigations of KCOs were initially performed on vascular smooth muscle tissues, they were originally identified as Ca 2+ ion antagonists. 236 KCOs have been identified to interact with the SUR subunit of the protein. Common cardiovascular drugs include pinacidil, diazoxide and nicorandil. Pinacidil, as well as

25 its derivatives such as P1075, is a highly used and commercially available KCO. Studies with the tritium labelled P1075 demonstrate that it binds to the SUR subunit but not the Kir6. x subunit. Furthermore, electrophysiology studies indicate that isolated Kir6 channels are not responsive to the drug. However, the response to KCOs is restored when Kir6. x subunits are co-expressed with the SUR subunit. 194

Pancreatic K ATP channels composed of SUR1 are only weakly activated by pinacidil but are strongly opened in the presence of diazoxide. By contrast, K ATP channels in the heart, comprised of SUR2A subunits, are strongly activated by pinacidil and not by diazoxide. Interestingly, channels containing the SUR2B isoform are activated by both pinacidil and diazoxide. This differential regulation between SUR2A and SUR2B suggests a critical role for the C-terminal 42 residues of the SUR protein which is the only location where the SUR2A and SUR2B splice isoforms differ. The C-terminal region of SUR2B is more similar to SUR1 (~74 %) than SUR2A (~33 %), explaining the sensitivity of both SUR1 and SUR2B to diazoxide compared to SUR2A.280

Similar to sulfonylurea receptors, the stoichiometry for KCO binding is not yet fully elucidated. Hill coefficients from binding studies of different KCOs with excised K ATP channels suggested more than 1 KCO molecule binds per K ATP channel. However, experiments performed with re-assembled Kir and SUR subunits have implicated only 1 KCO for the entire channel. As with sulfonylurea studies, some of these discrepancies may be associated due to relative concentrations of drug and nucleotide and/or hydrolysis activity of the channel. 169

1.6.2 Phosphatidylinositol

In addition to protein domains and ligands, the lipid content of the bilayer dramatically alters the 283 KATP channel open probability. Anionic phospholipids, most significantly phosphatidylinositol-4,5-bisphosphate (PIP 2), have been shown to activate the channel. Phosphatidylinositol is the most abundant phospholipid with the additional phosphorylated forms, phosphatidylinositol-4,5-bisphoshate, inositiol-1,4,5-trisphosphate and diacylglycerol comprising 1 % of the total bilayer composition. PIP 2 regulates a number of ion channels such as Ca 2+ -ATPases and Na +/Ca 2+ exchangers in the brain.127 Other Kir channels are also regulated by 283 PIP 2, such as Kir1.1 and Kir 3.4.

26

KATP channels in the presence of PIP 2 are less sensitive (at the Kir6. x site) to ATP 86 inhibition. Rb efflux electrophysiological studies on mutant K ATP channels suggested the PIP 2 binding interface is located to two regions (rich in basic residues) on the Kir subunit at the C- terminus. Competition experiments involving Kir6 fusions with maltose binding protein have shown that phosphatidylinositol competes with the binding of TNP-ATP (a strongly fluorescent

ATP analogue often used in nucleotide binding studies), with an order of PIP 3>PIP 2≈PIP>PI which likely suggests the importance of the electrostatic interaction of the lipid head group for 193 Kir subunit binding. Although the physiological role for PIP 2 activation of channels is still being investigated a general role for PIP 2 activation of a number of membrane bound proteins is to ensure their activity is only present in the plasma membrane since PIP 2 has a low abundance in other organelle membranes. This reduces the likelihood for promiscuous activity of a 127 /channel prior to reaching the cell surface. PIP 2 is also plays a role in a number of signal transduction pathways. Therefore PIP 2 is well connected into sensing the metabolic state of the cell to provide another layer of metabolic control over K ATP channels.

1.6.3 Phosphorylation of the K ATP channel

Post-translational modifications have been identified in the K ATP channel. These include phosphorylation sites on both the Kir6. x and SUR subunits. Phosphorylation by protein kinase A has been identified on Ser385 in Kir6.1. 43 Three phosphorylation sites have also been identified in the SUR2B subunits, Thr633 (in NBD1) and Ser1387 and Ser1467 (in NBD2). 243, 279 Phosphorylation of the SUR subunit on the NBDs produces stimulatory effects and increases channel activity. Mutation of these residues removes PKA dependent activity.279 Engineered disulfide bridges between different domains have suggested that phosphorylation allows for more efficient communication between the NBDs and MSDs allowing for the increased channel activity. 279 A novel protein kinase C phosphorylation site has also been identified on Ser382 on the Kir6 subunit. 18 Phosphorylation at this site is suggested to down regulate channel activity by disrupting interactions between the SUR and Kir subunits. PKC may also affect the distribution 18 of K ATP channels at the lipid bilayer surface.

27

1.6.4 Other known regulators of the K ATP channel

There are a number of accessory proteins/molecules that are demonstrated to associate and mediate K ATP channel. Although an exhaustive list is not provided here, an overview of some important known regulators is described.

The SNARE protein, syntaxin 1A, which regulates exocytosis interacts with the SUR 227 proteins at the nucleotide binding domains. The syntaxin 1A/SUR interaction inhibits KATP channel activity and this inhibition is reversed by the presence of MgADP or the potassium 227 channel opener diazoxide. Recent studies have also depicted that changes in PIP 2 can block syntaxin binding in a concentration dependent manner. 179

The intracellular pH can also effect channel activity, with basic pH favouring strong inward rectification and lower pH reducing inward rectification. The pH dependence was localized to a single residue, His216, in the Kir6.2 subunit. 24 Although the pH dependence was only observed in this study with the Kir6.2 subunit, studies by Kang et al have shown that syntaxin 1A will bind to the SUR NBDs in a pH dependent manner, specifically at low pH values. 147, 323 This pH dependence provides another mechanism of regulation in vascular cells 323 since the O2/CO 2 equilibrium is also pH dependent.

Acetyl-CoA esters are also known to activate the KATP channel as they are structurally analogous to PIP 2. They have been identified as important components for signal transduction pathways occurring in the pancreas. 52 Acetyl-CoA esters are found to reduce sensitivity of the channel to ATP. Even compounds with lower physiological abundances, such as hydrogen sulfide, a gastrotransmitter, have been identified as a potential regulator of K ATP channels in vascular smooth muscle. Endogenous hydrogen sulfide can stimulate K ATP channels to hyperpolarize the membrane. 297

1.7 Biophysical techniques

Significant understanding of the functional implications of the structure in the SUR NBD proteins involved probing different protein conformations. Although various techniques were employed at different stages in each project to address specific aims, three main techniques were consistently utilized to analyze the various proteins. Studies by NMR spectroscopy elucidated residue level information of the SUR proteins and also allowed information regarding dynamics

28 to be obtained. Fluorescence spectroscopy and circular dichroism require significantly lower concentrations of protein but probe more global changes to the molecule.

1.7.1 NMR Spectroscopy Nuclear magnetic resonance (NMR) spectroscopy is a technique that can be applied to the study of proteins. NMR takes advantage of the magnetically active nuclei (nuclei for which I ≠ 0, where I is the spin quantum number). The value of the spin quantum number is based on the number of protons and neutrons in the nucleus. 39 A nucleus will be NMR active as long as both the number of protons and the number of neutrons are not even numbers. Nuclei with a non-zero spin quantum number will possess a magnetic moment, μ. The value of the magnetic moment is given by the equation 39 : μ = (γ Ih)/2π , (Eq 1.1) where γ is the gyromagnetic ratio of a specific nucleus, h is Plank`s constant. The magnetic moment is a vector quantity, and as such, possesses a magnitude and direction. However, in the absence of a magnetic field, each nucleus orients in a random manner, resulting in no observable difference between various nuclei. Upon the application of a strong magnetic field, the nuclei are no longer degenerate and split into a thermodynamic distribution of various energetic states. 39 The number of states available depends on the spin quantum number, I. For commonly used biological nuclei, 1H, 15 N and 13 C (I = ½), there are two available states. The energy difference between these two states is related through the equation: ΔE = μB, (Eq 1.2) where B is the strength of the external magnetic field. Furthermore, the energy difference between the two states is small and on the order of kT . Thus, there is only a small number of excess nuclei in the lower energy state than in the high energy state. As a result, NMR is an intrinsically low sensitivity technique and requires relatively larger sample concentrations (milligram quantities).151

As any absorption spectroscopy, nuclei can be promoted to the higher energetic state by supplying energy (pulse) of the appropriate wavelength. In NMR, the energy of this transition is in the radio-frequency region of the electromagnetic spectrum. Furthermore, the local environment of the nuclei remarkably adjusts the energy of the transition. This is because the electron cloud of the nuclei is influenced by the electron clouds of nearby atoms. These changes

29 substantially influence the magnetic field experienced by the nuclei and will influence their resonant frequency.151

Since the energy difference between the two states is quite low, the excited state remains significantly populated at thermal equilibrium. Since a tiny fraction of each NMR sample (1 in 10 5) is interrogated following a pulse sequence, the pulse sequence is generally repeated several times in order to increase signal-to-noise. As a result of this repetition and the length of the pulse sequence, other processes such as relaxation of the nuclei to the original ground state also need to be considered to ensure reproducibility in the experiment. 39

There are two types of relaxation processes that significantly influence NMR experiments. The first is referred to as T 1 relaxation and refers to the thermal relaxation. Following an appropriate pulse on an NMR sample, the net magnetization vector will be perturbed from equilibrium (away from the z-axis). The gradual return to equilibrium is referred to as T 1 relaxation. Generally T 1 relaxation rates are quite long (~1.5 s), and practically not achieved. Instead, a constant waiting period is employed in between each repetition of the pulse sequence.39

A second type of relaxation is T 2 relaxation. T2 relaxation rates depend on the sample and refer to how quickly the nuclear spins in the sample dephase from each other. Following the applied radiofrequency the excited molecules, as a consequence of rotational motion in solution, experience different local environments. This in turn influences their local magnetic field and

NMR absorption peak. This T 2 relaxation depends on a number of factors, for example a higher viscosity solvent will allow for slower tumbling times and more solvent-molecule interactions that increase T 2 rates. Larger proteins (>30 kDa) have slower tumbling times. Elevated T 2 relaxation rates result in lower resolution and broadened line widths. 39

1.7.1.1 TROSY-HSQC

One dimensional 1H NMR is often insufficient for NMR studies of proteins, as spectra are too overlapped to be of use. In order to avoid this, 2D NMR is employed using samples that are isotopically enriched with the spin ½ nuclei, 15 N or 13 C (Note that naturally abundant 12 C nuclei are not NMR active, whereas 14 N has a spin of 1 and thus has more complex transitions than spin ½ nuclei). One common experiment is the heteroquantum single coherence transfer (HSQC)

30 experiment. The pulse sequence and product operators are shown in Figure 1.14. The overall experiment consists of excitation of 1H nuclei, followed by an INEPT period where the magnetization is transferred to the directly attached secondary nucleus (eg. 15 N). The chemical shifts of the 15 N nuclei are allowed to evolve and the magnetization is then transferred back to the 1H nuclei through a second INEPT period. The pulse sequence ends with an acquisition period in which the 1H nuclei chemical shifts are evolved. 39

Upon closer examination of the pulse sequence in Figure 1.14, the initial 90°x on proton creates transverse magnetization. The following pulse sequence is a spin-echo and refocuses the magnetization from chemical shift evolution on 1H. The simultaneous 180° pulses on 1H and 15 N during this period allow for evolution of magnetization due to J HN coupling, which results in 1 15 transfer of magnetization from H to N to create antiphase magnetization of the form (-2I xSz). The 90° pulses following the spin echo, allow the antiphase magnetization to be transferred to 15 the N (-2I zSy). This overall sequence (including the spin-echo) is known as an INEPT period and transfers polarization from one nucleus to another. 39

15 In the second spin echo, the chemical shift is allowed to evolve on N only. The 180°x pulse on 1H ensures that chemical shift evolution on 1H and that J HN coupling evolution is refocused. Following the second spin echo sequence, 90° pulses are applied on both nuclei to create antiphase magnetization of the form 2I yS2zcos w Nt1, where the term (cos w Nt1) describes modulation of the magnetization by the chemical shift of the 15 N nucleus. Double quantum

o o o o o o 90 x 180 x 90 y 180 x 90 x 180 x

1 I ( H) τ1 τ1 t1/2 t1/2 τ1 τ1

o o o o 180 x 90 x 90 x 180 x

Decoupling S (15 N) coherence

I1y cos(π JIS τ1) -2I 1zS2y w +2I S cos( w t ) w -2I 1z S2y cos( 2t1) 1y 2z 2 1 -I1x cos( N 2t1) -2I 1x S2z sin(π JIS τ1) +2I 1z S2x sin( w 2t1) -2I 1y S2x sin( w 2t1) S2z If τ1 = 1/4J IS , simplifies to Final term is multiple -2I 1x S2z quantum and remains unobservable (therefore omitted)

Figure 1.14: HSQC pulse sequence and corresponding produce operators.

31

coherence magnetization of the form -2I ySxsin w Nt1is also created, but not be observed and is therefore ignored. The second INEPT sequence results in creation of in-phase magnetization of 1 15 the form -Ixcos w Nt1, which will then evolve due to chemical shift on H. Decoupling on N during the acquisition period ensures that there is no J HN coupling.

The pulse sequence that we employ is modified from the HSQC in various manners. Most significantly, large proteins suffer from loss in resolution due to faster relaxation rates

(shorter T 2). This faster T 2 relaxation manifests as broader line widths and lower resolution data. In the absence of 15 N decoupling during acquisition, four peaks are observed since each crosspeak is a doublet in both dimensions.39 In addition, each component peak possesses different linewidths. The differential broadening is caused by interference between dipolar and chemical shift anisotropy relaxation mechanisms. In a standard HSQC experiment, decoupling allows for rapid sampling between each of these different states and the four peaks collapse into a single peak with average relaxation properties of the individual peaks. In a transverse relaxation optimized spectroscopy (TROSY) HSQC, only the slowest relaxing component (narrowest peak) is selected through phase cycling. The result is increased spectral quality for large proteins. The TROSY pulse sequence increases the size limit from 30 kDa to 50 kDa. Proteins as large as 100 kDa have also been analyzed with the TROSY-HSQC, although deuteration of the protein is required. 260

The TROSY-HSQC is also gradient enhanced, where a gradient induces a heterogeneous magnetic field throughout the sample that varies in a linear manner. Applying a gradient will cause magnetization at one position in the sample to precess at a different rate than at another position in the sample. Applying the reverse gradient will undo this process. Although this process may have an apparent null effect, any molecules that spatially move within the interval between the two gradients will not have their magnetization fully restored. Thus, gradients can also be applied at various stages in the pulse sequence to dephase solvent signals and suppress the strong water peak. In the TROSY-HSQC, the gradients are also used for coherence transfer selection of the slowest relaxing term and suppression of the anti-TROSY terms. 39

1.7.2 Fluorescence Spectroscopy

NMR spectroscopy requires significant protein concentrations (0.1 - 1 mM, depending on the NMR experiment to be run), whereas fluorescence spectroscopy can provide information of

32 protein structure and conformation with lower amounts of protein (1 μM). When light is directed towards a sample, it can interact with a solution in a number of different manners (Figure 1.15). In the UV-visible range, a large portion of light is transmitted through dilute protein-buffer samples. However, if the wavelength of light matches the energy between the ground and excited state of a molecule in the solution, it can be absorbed. The subsequent relaxation pathway for this molecule is determined by a number of factors, most predominantly the electronic nature of the molecule. Following excitation, the molecule will relax to the lowest vibration level of the excited state. If this level in the excited state (S 1) overlaps with vibrational levels in the ground state (S 0), relaxation will occur to the ground state by non-radiative transitions. However, if there is no overlap between energy levels, a photon may be emitted which is referred to as fluorescence. The energy of the emitted photon is of lower energy than the initial wavelength, except in rare cases. 263

The fluorescence observed for a molecule depends on several parameters such as the extinction coefficient (likelihood for photon absorption), quantum yield (likelihood for photon emission), sample concentration and the conditions/local environment. The extinction coefficient and quantum yield are intrinsic properties of the fluorophore and cannot be easily manipulated. For example, Trp has a quantum yield of 0.2 (20 % of absorbed photons are relegated to fluorescence relaxation) and most significantly contributes to intrinsic fluorescence of proteins. However, all Trp-containing proteins do not exhibit the same fluorescence because the local environment of each Trp is not identical. When exchange occurs for a fluorescent residue (or group) from a solvent-exposed environment to solvent-buried environment, changes in two

Figure 1.15: Jablonski diagram depicting the various transitions following absorption of a photon by a molecule. The ground vibrational states are shown in red and the excited vibrational states are shown in green. Adapted from Kochuveedu & Kim. Nanoscale, 2014, 6:4966-4984 with permission of The Royal Society of Chemistry. (Reference 154 ).

33 parameters are often observed. The wavelength for maximum emission is generally red shifted which is the result of the polar solvent molecules re-aligning their dipole to match the excited state of the fluorophore and thereby lowering the energy of the excited state. Further, this red- shift is often accompanied by a reduction in fluorescence intensity and is observed due to additional solvent collisions that non-radiatively dissipate the energy of the excited state, and reduce the probability for a photon emission.164

This property of solvent quenching can be extended by introducing a secondary molecule and observing the changes in fluorescence. A fluorophore (eg. labelled ATP) that is introduced into a solution will theoretically have low fluorescence due to solvent quenching. Upon burial into the hydrophobic core of a protein (eg. binding to a specific site), the solvent has a reduced probability of collisional quenching due to sterics, and a corresponding increase in fluorescence is observed. These fluorescence properties can be exploited to obtain binding constants for nucleotide to different proteins. 164

Alternatively, introducing a molecule that is not fluorescent, can also provide information on solvent accessibility. A large molecule can collide with a fluorophore in the excited state and dissipate the energy non-radiatively. This is known as collisional quenching and can be related to the exposure of the fluorophore through the equation 164 : = 1 + (Eq. 1.3)

In equation 1.3, Fo represents the fluorescence intensity in the absence of the quencher, F represents the fluorescence intensity at concentration of quencher, Q. The parameter kq represents the quencher rate coefficient and τ 0 is the excited state life time. The product of kq and

τ0 is often provided as the Stern-Volmer constant, K SV . Higher K SV constants are indicative of solvent exposure of the fluorophore. Thus, fluorescence is a useful tool for probing various conformations of a protein.

1.7.3 Circular Dichroism

Circular dichroism allows for rapid analysis of the overall secondary structure of a protein. Similar to fluorescence spectroscopy low concentrations of protein can be employed which enhance its utility in rapidly screening various solution conditions for proteins. CD spectroscopy is a variation of UV absorption spectroscopy in which a sample is sequentially interrogated with

34 right and left-handed circularly polarized light over a range of wavelengths. 275 A chiral molecule will differentially absorb right and left-handed polarized light and the difference in absorption will result in the emission of elliptically polarized light. The degree of ellipticity as a function of wavelength is plotted in a CD spectrum. For protein CD spectroscopy, typical absorption spectra are expected for specific secondary structures. 275 The archetypal spectra for predominantly alpha helical and beta sheet proteins are shown in Figure 1.16. The CD absorption band for proteins arises from the amide bond. Because different secondary structures have different orientations of the amide bonds with respect to one another, there is a characteristic CD spectrum for each secondary structure. As depicted in the simplified orbital scheme, promotion from the n oxygen lone pair to the carbonyl antibonding orbital (π*) is not electrically allowed. However, the transition can still occur magnetically. Therefore this absorption emerges with weak intensity in * the 220 nm range. A much stronger band is observed for the π o-π transition which gives rise to a peak in the 180-200 nm region. 254 These transitions allow for rapid analysis of the secondary structure of a protein.

* * π* πo π n π ~190-210 nm ~220 nm n 55000

25000 πo MRE

-5000

π -35000 nꞌ 160 180 200 220 240 260 Wavelength (nm) Figure 1.16: Electron orbital diagram of the amide bond. In circular dichroism, the most common transitions for proteins occur from π o → π* and n → π* level. Ideal CD profiles for alpha helices (red), beta sheet (blue) and intrinsically disordered proteins (yellow) are shown. CD spectra adapted from Miles & Wallace. Chem. Soc. Rev. 2006 Jan 35(1):39-51 with permission of The Royal Society of Chemistry. (Reference 205 ).

35

1.8 Thesis Framework

KATP channels are of clinical and medical relevance due to their ability to couple the metabolic state of the cell to the membrane potential. The SUR nucleotide binding domains are critical for ensuring proper regulation of K ATP channels, although biochemical and biophysical data of the NBDs is quite limited. As such, we sought to investigate molecular mechanisms through which the SUR NBDs regulate the K ATP channel. Part of the reason for limited structural and biophysical information is that the isolated SUR NBDs are difficult to express as monomeric proteins and are prone to precipitation or aggregation. As such, our primary goal was to develop strategies for soluble expression and achieve high-yield purification of SUR2A NBD1 from heterologous expression in E. coli (Chapter 2).

Once we were able to generate quantities of protein amenable for detailed NMR studies, we sought to assign the correlations in the NMR spectra of NBD1 to the individual amino acids, in order to obtain residue specific information. Due to low quality of triple resonance assignment data, we have developed labelling techniques to augment the resonance assignment data (Chapter 3). This has allowed us to study conformational changes in NBD1, such as those caused by nucleotide binding, in a site specific manner.

These studies have provided us with a platform to examine various mechanisms of regulation of SUR2A NBD1. These studies include examining the conformational and biochemical effects of NBD1 phosphorylation (Chapter 4), disease-causing mutations in NBD1 (Chapter 5), and drug-binding (Chapter 8/Appendix). Our results have offered insights into different modes of regulation in SUR2A NBD1 by the interaction of various disordered regions.

We have also attempted to characterize interactions of SUR2A NBD1 with its binding partner, SUR2A NBD2. Strategies into purifying SUR2A NBD2 as a soluble and monomeric protein are explored in Chapter 6. Finally, Chapter 7 summarizes progress in understanding regulation by SUR2A NBD1 and offers perspectives into the different mechanisms of molecular- level regulation in NBD1 and how this influences K ATP channel activity.

36

Chapter 2 Expression and Purification of Isolated SUR2A NBD1 2 Overview

This chapter demonstrates the expression and purification of the isolated NBD1 from SUR2A, the regulatory subunit of cardiac KATP channels. The aim of this study was to develop conditions that enable the purification of large quantities of soluble and monomeric SUR2A NBD1. This would also allow for concentrated samples of NBD1 to be stable at 30 oC for long periods of time so that detailed NMR experiments and other structural studies could be performed. Initially, a domain boundary screen, testing the expression and solubility of SUR2A NBD1, was performed as part of M.Sc. work and published in Biochemistry. 1 This work is summarized in the introduction section of this chapter, including some figures, with permission from the publisher. This chapter focuses on strategies employed for optimizing purification yields and has allowed for a ten-fold increase in protein yield. Significant portions of this chapter were published in Protein Purification and Expression 2 and reproduced with permission. Author contributions: E.D.A. generated all SUR2A NBD1 samples, conducted the thermodynamic stability screens, nucleotide binding experiments, and analyzed all fluorescent data. V.K. analyzed the NMR data.

2.1 Introduction

An understanding of the molecular basis for both regulation of the K ATP channels and disease causing mutations is precluded by the lack of high-resolution structural information for K ATP channels. Structural studies of the NBDs would benefit from generating these samples as isolated proteins. As previously described in Chapter 1, although the NBDs are part of a much larger complex, the modularity of the SUR proteins, (the distinct MSD and NBD domains) provides a foundation for studying isolated NBDs. Variability of the arrangement and connectivity between different ABC transporters 61, 72 provides further evidence of the modularity of the ABC protein

1 de Araujo, E.D. , Ikeda, L.K., Tzvetkova, S., and Kanelis, V. (2011) The First Nucleotide Binding Domain of the Sulfonylurea Receptor 2A Contains Regulatory Elements and Is Folded and Functions as an Independent Module. Biochemistry . 50:6655-66. (Reprinted with permission from American Chemical Society). 2 de Araujo, E.D and Kanelis V. (2014) Successful development and use of thermodynamic screen for purification of SUR2A NBD1. Protein Purification and Expression . 103:38-47. (Reprinted with permission from Elsevier).

37 superfamily. For example, in several mammalian ABC proteins, including the SURs, the minimum structure is encoded within a single polypeptide chain with the arrangement of MSD1- NBD1-MSD2-NBD2.72 By contrast, other ABC proteins, such as human TAP transporters, are complexes of two polypeptide chains, with each chain comprised of one MSD and one NBD.72 As a further variation, many prokaryotic ABC proteins frequently consist of MSD and NBD domains that are encoded and expressed as independent proteins and co-assemble into the full transporter complex.61, 72 This variation of how the MSDs and NBDs are arranged on different polypeptide chains in different proteins suggests that the MSDs and NBDs function as independent interacting modules and can be studied in isolation.

The identification of the correct domain boundaries is essential for heterologous expression of soluble NBDs. Most significantly, domain boundaries should be selected such that all structured regions are included and the disordered regions are minimized. Truncation of structured regions can destabilize the overall fold of the domain and promote protein precipitation. Extending the boundaries of the domain to include unstructured regions is also detrimental as these flexible regions may reduce solubility and also lead to protein aggregation. Determination of the correct domain boundaries in the NBDs of ABC transporters is further complicated by the low sequence identity between the domains beyond the ATP-binding sequences involved in ATP-binding and hydrolysis. 152

Initial structure-based sequence alignments of the SUR NBDs (Figure 2.1) suggested different possibilities for domain boundaries of the SUR NBDs. Expression and purification profiles have suggested domain boundaries of S615-L933 as the regions encompassing SUR2A NBD1, as this construct gave the greatest amount of soluble protein per L of E. coli culture and could be concentrated (Figure 2.2A). 2D 15 N-1H TROSY-HSQC NMR data demonstrate dispersion in the 1H dimension that is characteristic of a folded protein (Figure 2.2B). The NMR spectra also display resonances with markedly different intensities. The most intense resonances have 1H chemical shifts between 7.8 ppm and 8.2 ppm (Figure 2.2Ci) and are likely from disordered regions in the protein, such as loops or disordered residues at the N- and C- termini. Other resonances have broad, weak signals (Figure 2.2Cii and Ciii), which are likely due to protein motions on the μs-ms timescale. We can also rule out dimerization or protein aggregation as being the cause for differential broadening, as our additional studies demonstrated that SUR2A NBD1 is monomeric at NMR concentrations. 189

38

Figure 2.1: Structure-based sequence alignment of SUR2A NBD1 with ABC transporters. The grey shapes above the sequence alignment represent known structural elements of CFTR NBD1, with α-helices represented as cylinders, β-sheets as arrows, and 3 10 -helices as open circles. Secondary structure elements in known structures are highlighted in purple for β-strands, blue for α-helices, and green for 3 10 -helices. Conserved residues in the Walker A and Walker B motifs, signature sequence, and Q, D, and H loops are in bold, with the motif labelled below the alignment. An asterisk (*) highlights a conserved aromatic C-terminal to the β1 strand, which is involved in binding the adenine ring of ATP. Residues comprising a predicted disordered region in SUR2A NBD1, thought to be analogous to the regulatory insert of CFTR, are highlighted in light orange. A previously known PKA phosphorylation site, T632, in rSUR2A and mSUR2C is indicated by a red open circle above the alignment.

39

A

B C

Figure 2.2: Structural and biochemical characterization of SUR2A NBD1 S615-L933. (A) Expression, purification and characterization of different SUR2A NBD1 constructs. (B) 2D 15 N 1H TROSY-HSQC spectrum of SUR2A NBD1 (0.32 mM) with 5 mM Mg 2+ and 5 mM ATP in 20 mM Na + phosphate, pH 7, 150 mM NaCl, 2 % (v/v) o glycerol, 5 mM DTT, 10 % (v/v) D 2O at 30 C at 600 MHz. Chemical shifts were referenced to 4,4-dimethyl-4- silapentane-1-sulfonic acid (DSS). Resonances of backbone nuclei, as well as those from side chain nuclei from Trp, Asn, and Gln residues, are shown in black. The black resonances enclosed in the dashed box are from the side chain Trp indole HN. There are six Trp residues in the protein and six resolved resonances in the dashed box. The gray resonances are of opposite sign, caused by spectral aliasing, and are possibly from Arg NεHε side chain correlations (marked with an “*”) or backbone NH correlations from Gly residues (marked with a “+”). (C) Selected regions of the spectra showing peaks of markedly different intensities. A trace through the approximate center of the peak is shown at the bottom of each spectrum, illustrating the line shape. Three different resonances were chosen, identified as (i), (ii), and (iii) in panel B. Very intense resonances in the center of the spectra, such as the peak represented by (i), may arise from disordered loops in NBD1. Very weak resonances, such as the peak represented by (iii) are due to conformational exchange in the protein.

2D 15 N-1H correlation spectra of proteins at concentrations of 250 m M or less can be used to monitor global structural changes with mutations or changes in conditions, or to assess binding of the protein to specific ligands.145, 189 However, in order to obtain information at the level of individual residues, each resonance in the 2D 15 N-1H correlation spectrum must be

40 assigned to a specific residue. Although 15 N-1H TROSY-HSQC spectra of NBD1 at 250 m M can be recorded in approximately 40 min, NMR experiments required for resonance assignment of the protein require more concentrated samples of NBD1 (≥0.5 mM or >18 mg/ml) to be stable for many days (>14 days) at 30 °C. Although lower temperature prevents precipitation of NBD1, spectra of NBD1 at lower temperatures suffer from resonance broadening, and hence are not of sufficient quality for detailed NMR studies. Thus, we sought to determine additional conditions that would allow greater yields of soluble NBD1 to be obtained, which would decrease the amount of cell culture used and thus the cost of isotopic enrichment for NMR studies, and would also allow concentrated samples of NBD1 to be stable at 30 °C for many days.

In order to determine conditions to allow for more concentrated samples of SUR2A NBD1 to be stable for longer times, we developed a cost-effective and convenient method for screening buffer conditions and additives that enhance protein stability. Application of the assay identified positive screening conditions to the purification protocol of NBD1 resulted in a 10- fold increase in the yield of the purified protein (75 mg NBD1 per liter E. coli culture versus 8 mg per L culture reported previously). 64 The thermodynamic screen also provided insights into strategies for long-term storage of NBD1. We subjected NBD1 in these new buffer conditions to binding studies with different nucleotides using fluorescence and NMR spectroscopy. Our NMR data suggest that MgATP, MgADP, and MgAMP-PNP binding to NBD1 cause similar chemical shift changes, which result from the interaction of specific NBD1 residues with nucleotides and protein conformational changes associated with nucleotide binding. Further, NMR spectra of NBD1 with MgAMP-PNP are of higher quality compared to using MgATP, indicating that AMP-PNP could be used as the ligand in future NMR studies. The methods and strategies presented here can be employed to enhance purification yields, sample life times, and storage of low stability nucleotide binding domains, such as other ATPases and GTPases.

2.2 Methods

2.2.1 Expression and purification of SUR2A NBD1

Rat SUR2A NBD1 (S615-L933) was expressed as previously described 64 in E. coli BL21 (DE3) CodonPlus® (RIL) cells (Stratagene) with an N-terminal 6×His-SUMO tag. Cell cultures were 15 grown in minimal M9 media, which contained NH 4Cl (Sigma Aldrich or Cambridge Isotope Laboratories) for isotopic labelling, with constant agitation at 37 °C. As the cell cultures reached

41

OD 600 values of 0.4, 0.6, and 0.8, the incubation temperature was progressively decreased to 30 °C, 25 °C, and 18 °C, respectively. Cell cultures were then allowed to incubate at 18 °C for 30 min prior to induction of protein expression by addition of 0.75 mM IPTG. After ~20 h the cells were harvested by centrifugation and the pellets were stored at -20 °C.

Here, we describe the original purification procedure that was previously performed, 64 and include all modifications to the protocol that allow for larger yields of purified soluble protein. Purification of SUR2A NBD1 was carried out at 4 oC for all steps. Each 1 g of cell pellet was resuspended in 8 ml of lysis buffer (20 mM Tris-HCl pH 8.0, 100 mM Arg, 2 mM β- mercaptoethanol, 10 mM MgATP, 5 mM imidazole, 0.2 % [v/v] Triton X-100, 10 % [v/v] glycerol, 2 mg/ml deoxycholic acid, 1 mg/ml lysozyme, 5 mM 6-aminocaproic acid, 5 mM benzamidine and 1 mM PMSF). The cell lysate was loaded onto a 5 mL Ni 2+ -NTA affinity column (GE Healthcare) equilibrated with 20 mM Tris-HCl, pH 7.6, 150 mM NaCl, 10 mM MgATP, 2 % glycerol, and 5 mM imidazole. The Ni 2+ affinity column was then washed with 10 column volumes of equilibration buffer and the 6×HisSUMO-NBD1 fusion protein was eluted using 20 mM Tris-HCl pH 7.3, 150 mM NaCl, 10 mM MgATP, 2 % (v/v) glycerol, and 400 mM imidazole. The elution fractions were immediately diluted 3-fold into a buffer containing 20 mM Tris-HCl pH 7.3, 10 mM MgATP, 5 mM β-mercaptoethanol, and 2 % (v/v) glycerol. The 6×-His-SUMO tag was cleaved from NBD1 with a His-tagged Ulp1 protease.

The resultant mixture, containing isolated SUR2A NBD1, 6×-His-SUMO, and His-Ulp1, was loaded onto a size exclusion column (Superdex 75, GE Healthcare) in 20 mM Tris-HCl pH 7.3, 150 mM NaCl, 2 mM MgATP, 5 mM β-mercaptoethanol, and 2 % (v/v) glycerol. SUR2A NBD1 was then purified to homogeneity by a reverse Ni 2+ -NTA affinity column in 20 mM Tris-HCl, pH 7.3, 150 mM NaCl, 10 mM MgATP, 5 mM β-mercaptoethanol, 2 % (v/v) glycerol, and 5 mM imidazole, which removed small amounts of the 6×His-SUMO and 6×His- Ulp1 proteins that co-elute with NBD1 from the size exclusion column. The purified SUR2A NBD1 was dialyzed into NBD1 buffer (20 mM sodium phosphate, pH 7.3, 2 % [v/v] glycerol, 2 mM DTT) with and without ATP and MgCl 2, as required.

In summary, the original purification protocol 64 was altered in the following manner: (1) The MgATP concentration in the lysis and Ni 2+ column buffers was increased from 2 mM to 10 mM; (2) NaCl was removed from the lysis buffer; (3) Ni 2+ column elution fractions were

42 immediately diluted using a buffer containing 10 mM MgATP, but lacking NaCl; (4) The pH of the Ni 2+ column elution buffer, the dilution buffer, and the size exclusion column buffer were decreased to 7.3 from 7.6.

2.2.2 Protein concentration determination

Because samples of NBD1 contain MgATP, some of which is bound to NBD1 and some of which is free in solution, A 280 readings cannot be used to determine the protein concentration. Thus, concentrations of NBD1 samples were determined using the Bradford protein assay and confirmed by amino acid analysis (Advanced Protein Technology Centre, Hospital for Sick Children).

2.2.3 Sample storage

Protein samples were stored in solution in 10 mM sodium phosphate, pH 7.0, 5 mM MgATP and 2 % (v/v) glycerol at 4 °C. Samples of NBD1 at concentrations of 250 m M are stable in solution up to three weeks at 4 °C.

Long-term storage of NBD1 is achieved by lyophilization. Samples of NBD1 in the storage buffer at 18 mg/ml (0.5 mM) and 500 m L were frozen rapidly in liquid N 2 and lyophilized overnight using a Modulyo-115 Thermo Electron Corporation lyophilizer. Lyophilized samples were redissolved in 500 m L of water and the buffer exchanged as required.

2.2.4 Screening buffer conditions for thermodynamic stability

The stability of NBD1 in solutions with different components was assayed by performing intrinsic Trp fluorescence thermal denaturation experiments. Fluorescence experiments were conducted on a Fluoromax-4 spectrofluorimeter (Horiba Scientific) equipped with a Peltier unit for precise temperature control. Thermal denaturation of NBD1 was monitored by following the change in the emission spectrum at 350 nm, the wavelength at which the difference in the fluorescence spectra of folded and denatured NBD1 is at a maximum. The excitation wavelength was 295 nm, and the excitation and emission slit widths were 2 nm and 3 nm, respectively. The NBD1 samples were heated from 20 °C to 60 °C in 1 °C increments with a 30 s equilibration time at each temperature. Samples contained 2 m M NBD1 in 10 mM Na + phosphate, pH 7.0, with and without additives, in a total volume of 0.5 ml.

43

2.2.5 Fluorescence Nucleotide Titrations

The Kd value for binding of the fluorescent ADP analogue 2',3'-O-(2,4,6-trinitrophenyl)- adenosine-5'-diphosphate (TNP-ADP, Molecular Probes) to NBD1 was determined and 2+ compared to the K d value for the NBD1/TNP-ATP interaction. Mg and ATP were removed from NBD1 by size exclusion chromatography (Superdex 75 column, GE Healthcare) and 189 replaced with 2.5 m M MgCl 2 and 2.5 m M TNP-ATP or 2.5 m M TNP-ADP. Alternatively, substitution of MgATP with other nucleotides of interest can be effectively accomplished by first removing MgATP via a desalting column followed by addition of the nucleotide of interest. Because of the limited solubility of nucleotide-free NBD1,64 binding experiments were conducted using 10 % (v/v) glycerol at 15 °C. Binding experiments were performed by serial dilutions of the protein from 50-70 m M, depending on the concentration eluted from the size exclusion column, to 0.8-2.0 m M while keeping the concentrations of the MgCl 2 and fluorescent

189 nucleotide constant at 2.5 m M each. A separate sample was generated containing MgCl 2, nucleotide and buffer only for the 0 m M NBD1 sample. Fluorescence spectra of TNP-ADP or TNP-ATP were recorded after each sample was generated using an excitation wavelength of 465 nm and a slit width of 5 nm. Emission spectra were recorded from 485 – 600 nm with a slit width of 7 nm. The K d value for the NBD1-nucleotide complex was determined by monitoring the ratio between the fluorescence intensity at 533 nm, which corresponds to the wavelength where the fluorescence difference of free and bound TNP-nucleotide is at a maximum, and 600 nm to account for any non-specific fluorescence from the protein.223 The titration data were fit to Equation 2.1:

(I¥ - I )  2  I =I + o ()[TNP ]+[NBD1 ]+K - ()[TNP ]+[NBD1 ]+K - 4() [TNP ] ()[NBD1 ] (Eq. 2.1) o  total total d total total d total total  2[TNPtotal]

where I is the fluorescence intensity ratio at a given total concentration of TNP-ATP, [TNP total ],

I∞ is the fluorescence intensity ratio at saturation, I o is the fluorescence intensity ratio in the absence of ligand, K d is the dissociation constant, and [NBD1 total ] is the total concentration of SUR2A NBD1 in the reaction. This equation assumes a 1:1 complex of NBD1 with TNP-ATP or (TNP-ADP).110, 317

44

2.2.6 Nucleotide NMR Titrations Interactions of SUR2A NBD1 with nucleotide and nucleotide analogues were monitored by recording 15 N-1H TROSY-HSQC spectra of NBD1 at 30 °C on a 600 MHz Varian spectrometer equipped with a H(F)CN triple resonance cyroprobe with actively shielded z-gradients. NMR samples of NBD1 were prepared with fresh nucleotides. First, MgATP was removed from samples of NBD1 by size exclusion chromatography. The appropriate nucleotide ( i.e. MgATP, MgADP, MgAMP-PNP) was then introduced immediately into the sample. Solutions of nucleotides were adjusted to pH 7.3 using NaOH so that the addition of the analogue would not change the pH of the samples. Because NBD1 samples lacking nucleotide are sensitive to aggregation and precipitation,64 two samples of 200 m M NBD1 with either a 1 mM Mg- nucleotide or with 50 mM Mg-nucleotide were prepared. NMR spectra of the NBD1 samples with the high and low nucleotide concentrations were initially recorded. Subsequently, these NBD1 samples were combined in different ratios and additional NMR spectra were recorded to obtain intermediate points in the titration. This method allowed us to maintain the protein concentration at a constant level throughout the titration while minimizing the use of expensive nucleotides. NMR spectra were processed with NMRPipe/NMRDraw 75 and analyzed with NMRView.139 The combined chemical shift change was calculated using Equation 2.2: 2 2 0.5 Dd total = ( d HN + Dd N ) (Eq. 2.2)

1 15 The terms, Dd HN and Dd N, are the chemical shift changes in the HN and N dimensions 129 respectively. The NMR samples also contained 10 % D 2O (v/v) and 0.5 mM 4,4-dimethyl-4- silapentane-1-sulfonic acid (DSS) for 1H and 15 N chemical shift referencing.330

2.3 Results and Discussion

2.3.1 Thermal denaturation studies and buffer screening

Our previous work showed that 250 m M samples of SUR2A NBD1 are monomeric, give high quality NMR spectra, and are suitable for NMR titration studies.64, 189 However, NMR samples of NBD1 at various concentrations (250 m M – 500 m M) and NBD1 samples during the purification precipitate over time. We hypothesized that precipitation of NBD1 is caused by individual local protein unfolding events, which expose hydrophobic residues that can interact in concentrated solutions and lead to protein aggregation or precipitation. Thus, conditions that increase the thermodynamic stability of the protein may result in increased sample lifetime.

45

A B

C

Figure 2.3: Determining denaturation profiles of SUR2A NBD1 through circular dichroism and fluorescence spectroscopy. All experiments contain 2 μM protein in 10 mM sodium phosphate pH 7.0 Experiments in the absence of any nucleotide are shown in black, with the presence of 2 mM ATP, AMP or ADP shown in red, green or blue respectively. (A) shows the melting profile by monitoring the CD signal at 222 nm as a function of temperature. (B) depicts the change in intrinsic Trp fluorescence as the concentration of guanidinium hydrochloride is increased. 8M of guanidinium hydrochloride (optical grade) was titrated into the protein solution, thereby increasing the volume of the sample and artificially decreasing the fluorescence intensity. Therefore all guandinium hydrochloride unfolding experiments were normalized to a control titration with the addition of buffer. (C) illustrates the change in intrinsic Trp fluorescence with increasing temperature. Each method demonstrates the same relative increase in stability upon the addition of nucleotide. However, unfolding in each case was irreversible.

These conditions could include the pH of the sample buffer, use of natural ligands for the protein, or the addition of osmolytes that aid in protein stabilization and solubility. In order to determine factors that allow for greater protein stability we assayed the melting temperature (T m) of NBD1 under different buffer conditions and with various stabilizing additives.

46

We have monitored protein stability of SUR2A NBD1 by both circular dichroism and fluorescence melting studies. Fluorescence spectroscopy is more versatile as several solutes can interfere with the CD signal. Changes to protein stability were observed by varying the temperature or concentration of denaturant, guanidinium hydrochloride. As shown in Figure 2.3 various signals were monitored including the CD signal (Fig 2.3A), and the intrinsic Trp fluorescence (Fig 2.3 B,C). Although the denaturation profiles display similar trends, we opted to carry out the thermodynamic buffer screen by monitoring intrinsic Trp fluorescence as a function of melting temperature. Measurements through CD spectroscopy require interrogation of both right and left circularly polarized light which significantly increases the length of the experiment. We have also attempted fluorescence melts performed using SyproOrange, which required screening of protein:dye ratios in order to obtain an observable transition (data not shown). However, the obtained denaturation curves deviated from expected melting profiles using SyproOrange. 213 Also, the use of SyproOrange may interact with different compounds in the buffer screen and influence the results. Therefore, based on the ease of experiment replication, we decided to carry out the thermodynamic buffer screen by monitoring intrinsic Trp fluorescence as function of melting temperature. SUR2A NBD1 contains 6 Trp residues (W616, W677, W727, W756, W877, W906). Homology models of SUR2A NBD1 189 based on the crystal structure of NBD1 from the ABC transporter MRP1 248 indicate that most of the Trp residues are at least partly buried. Thus, the intrinsic Trp fluorescence of SUR2A NBD1 will change upon protein unfolding and can be used to probe protein stability. Trp fluorescence melting curves for

NBD1 in the absence and presence of nucleotides are shown in Fig. 2.3C. The T m was determined by calculating the maximum of the first derivative of the sigmoidal fluorescence melting curves.201 Fitting thermal denaturation curves was not possible because of the steep slopes of the folded and unfolded baselines, likely due to the fact that there are six Trp residues in SUR2A NBD1. The presence of 2 mM MgATP leads to a marked increase in the T m of NBD1, from 27.7 ± 1.2 °C in the apo state to 43.0 ± 0.1 °C in the presence of 2 mM MgATP

(Fig. 2.4). The presence of 2 mM MgADP and MgAMP also results in increases in the T m of NBD1 to 40.6 ± 0.7 °C and 39.3 ± 0.3 °C, respectively.

47

Figure 2.4: Thermal unfolding assays screening different buffer conditions. In each sample, 2 μM NBD1 was incubated with 10 mM NaPhos pH 7.0 and 2 mM ATP (excepting screens where buffer, pH and nucleotide were varied). Each point represents the average of three individual melting temperature experiments of NBD1. The error bars represent the standard deviation of the averaged experiments.

48

Table 2.1: Summary of NBD1 melting temperatures with various ligands Melting Temperature Melting Temperature ATP (mM) ADP (mM) 0.00 27.7 ± 1.2 0.00 27.7 ± 1.2 0.25 34.0 ± 1.0 0.25 34.3 ± 0.3 0.50 38.3 ± 0.6 0.50 36.6 ± 0.0 0.75 39.3 ± 0.3 0.75 38.3 ± 0.3 1.00 41.7 ± 0.6 1.00 40.3 ± 0.3 2.00 43.0 ± 0 2.00 40.6 ± 0.7 5.00 45.3 ± 0.3 5.00 42.3 ± 0.3 10.00 45.7 ± 0.6 10.00 42.3 ± 0.0 AMP (mM) Glycerol (% v/v) 0.00 27.7 ± 1.2 0 43.0 ± 0.0 0.25 33.3 ± 0.3 2 43.3 ± 0.3 0.50 35.7 ± 0.3 4 46.3 ± 0.6 0.75 37.0 ± 0.0 6 48.3 ± 0.3 1.00 37.7 ± 0.3 8 50.3 ± 0.3 2.00 39.3 ± 0.3 10 51.7 ± 0.6 5.00 40.3 ± 0.3 12.5 51.0 ± 0.0 10.00 40.7 ± 0.6 15 52.0 ± 0.3 NaCl (mM) DCA (mg/ml) 0 43.0 ± 0.0 0 43.0 ± 0.0 50 42.3 ± 0.6 1 44.0 ± 1.0 150 41.3 ± 0.6 2 44.3 ± 0.6 200 39.6 ± 1.1 4 45.7 ± 0.6 300 38.3 ± 1.1 6 47.3 ± 0.6 L-Glu (mM) L-Arg (mM) 10 43.0 ±0.0 10 42.7 ±0.6 50 43.3 ± 0.6 50 43.3 ± 0.6 100 44.3 ± 0.6 100 44.6 ± 0.6 Phosphate (mM) TrisHCl (mM) 10 43.0 ±0.6 10 43.0 ±0.0 50 44.3 ± 0.3 50 43.3 ± 0.6 100 43.7 ± 0.6 100 43.3 ± 0.6 Buffer pH Triton-X (% v/v) 5.0 34.3 ± 0.3 0 43.0 ± 0.0 5.5 37.7 ± 0.3 0.20 44.0 ± 0.0 6.0 40.3 ± 0.3 0.50 44.3 ± 0.6 6.5 41.3 ± 0.3 0.75 46.7 ± 0.6 7.0 43.0 ± 0.0 1.0 48.3 ± 0.6 7.3 43.3 ± 0.3 HEPES (mM) 7.6 40.3 ± 0.3 10 43.0 ±0.0 8.0 36.7 ± 1.2 50 43.0 ± 0.0 8.5 34.7 ± 0.6 100 43.6 ± 0.6 DTT (mM) DTT (mM) 0 43.0 ± 0.0 3 43.3 ± 0.6 1 43.3 ± 0.6 5 43.7 ± 0.6 2 44.4 ± 0.0 10 42.3 ± 1.1 The melting temperatures are given with standard deviations of three independent experiments.

49

Addition of increasing amounts of Mg- nucleotides results in concomitant increases in the T m value (Fig. 2.4), reflecting the fact that nucleotides interact with NBD1. Further, the T m value of apo NBD1 of ~28 °C explains the poor quality NMR spectra of the protein in absence of nucleotide.64 NMR spectra of SUR2A NBD1 with MgATP and other Mg-nucleotides are of high quality (see below).189

The lower T m values for NBD1 with MgADP and MgAMP compared with MgATP is consistent with the lower binding affinity of SUR2A NBD1 for TNP-ADP (20.6 ± 2.4 m M; Fig. 2.5) compared to TNP-ATP (8.5 ± 1.1 m M; Fig. 2.5) and also the expected lower affinity of NBDs for AMP based on structures of NBDs with MgATP100, 130, 175, 248, 288, 315, 337 and 149, 175, 315, 335 MgADP. Tm values of 28 °C - 43 °C seen for isolated SUR2A NBD1 are similar to those obtained for NBD1 from CFTR,244 which is in the same subfamily of ABC transporters as 72 the SUR proteins. The relatively low T m values for these NBD1 proteins may reflect the fact that in single-chain ABC proteins, such as SUR2A and CFTR, NBDs make extensive contacts with the coupling helices and the opposite NBD,5, 62, 325 which add to the overall stability of the

NBD. Further, the relatively low T m values of isolated NBDs from mammalian ABC transporters underlie the importance of screening for stabilizing conditions.

Figure 2.5: SUR2A NBD1 has a higher affinity for MgATP than MgADP. The fluorescence titration of TNP-ATP and TNP-ADP are shown as filled circles and open circles, respectively. The titration data were fit assuming a 1:1 complex for the NBD1/nucleotide interaction, as described in the Materials and methods. The fit of the TNP-ATP titration data is shown as a solid line, while the fit of the TNP-ADP titration data is shown as a dashed line. K d values are reported as averages ± standard deviations of three separate experiments.

50

Using the Trp fluorescence denaturation method allows for rapid scanning of additional buffer conditions and stabilizing additives (Fig. 2.4 and Table 2.1). The T m values at varying pH indicate that NBD1 is most stable at pH 7.3. Increasing concentrations of glycerol resulted in increasing protein stability, with no further significant increase in protein stability observed at glycerol concentrations higher than 12.5 % (v/v). In contrast, NaCl concentrations of greater than

50 mM caused decreased T m values of NBD1. Small increases in protein stability were achieved with L-Arg concentrations greater than 50 mM or L-Glu concentrations greater than 100 mM. L- Arg and L-Glu have been shown to increase protein solubility alone 9, 10, 103, 166, 249, 324 and in combination.103 Concentrations of DTT up to 5 mM had no effect on NBD1 stability, while increasing concentrations of deoxycholic acid and Triton-X100 increased NBD1 stability slightly. In terms of buffer salts, phosphate buffer provides modest stabilization of NBD1, but is not compatible with many downstream assays, such as ATP hydrolysis experiments. Low concentrations of HEPES and Tris-HCl did not affect NBD1 stability. Both HEPES and Tris-HCl have a buffering capacity in the pH range (7.3 – 8.0) required in our purification. We have opted to use Tris-HCl in our purifications because of the lower cost of Tris-HCl.

2.3.2 Purification strategies and long term storage

The overall purification scheme is provided in Figure 2.6. Data from the thermodynamic screens were used to alter the composition of specific buffers in the original purification procedure in order to allow for maximum thermodynamic stability of SUR2A NBD1. There were two steps in the original purification protocol where the protein is most vulnerable to aggregation and precipitation: during lysis of the cells and immediately after elution of the 6×His-SUMO-NBD1 fusion protein from the Ni 2+ -affinity column. At both of these purification steps, the concentration of the NBD1 fusion protein is high and dilution does not alleviate protein precipitation. Further, resuspending cell pellets in a large volume of lysis buffer reduces the efficiency of the sonication step, resulting in further decreases in yield of the purified protein.

The most substantial increase in NBD1 stability was observed in the presence of nucleotide (Fig. 2.4). As such, we tested the effect of increasing concentrations of ATP in the cell lysis and Ni 2+ affinity purification buffers on purification of NBD1 (Table 2.2). There is a clear increase in the yield of purified NBD1 with increasing concentration of MgATP throughout the purification. Increasing the concentration of MgATP to 10 mM from 2 mM resulted in an

51

Figure 2.6: Overview of purification of SUR2A NBD1. In Step 1, BL21(DE3) CP-RIL E. coli are transformed with the pET-6×HisSUMO-NBD1 vector. The cultures are grown in 95% M9, 5% LB and the growth curve is monitored by optical density at 600 nm (OD 600 ). At an OD 600 of 0.8 the cultures are induced with 1 mM IPTG and harvested after 18-20 h. The E coli proteome is shown on a 15 % SDS-PAGE at pre- and post induction. In Step 2, cell pellets are lysed by sonication and the gel displaying purification is shown. Gel samples for the insoluble fraction (after cell lysis), the soluble fraction (after cell lysis), and the Ni 2+ column flow through correspond to 1/1000 of the total volume of the fraction. The amount of sample loaded for Ni 2+ column elution fraction corresponds to 1/20000 of the total volume (5 ml) for that fraction. In step 3, the soluble 6×His-SUMO-NBD1 is concomitantly concentrated and digested with 6×HisUlp1. The amount loaded for the 6×HisUlp1 digestion sample is 1/5000 the total volume (15 ml). In Step 4, size exclusion chromatography using a Superdex 75 (S75) column yields a sample containing NBD1 (36.6 kDa calculated molecular mass) and 6×His-SUMO (13.9 kDa calculated molecular mass). The UV-trace (at 280 nm) shows the elution profile of NBD1and 6×His-SUMO from the gel filtration column. The gel samples for the S75 purification are also shown and correspond to 1/5000 the total volume of the elution fractions on a 15% SDS PAGE gel. The reverse Ni 2+ column in step 5 removes residual 6×His-SUMO and any 6×His-Ulp1 to yield purified NBD1. The gel sample for the reverse Ni 2+ column (1/5000 the total volume) was taken after the protein was concentrated to 10 ml. The large amount of protein for the reverse Ni 2+ column demonstrates the high yield of purified NBD1 achieved and that samples of NBD1 at high concentration are attainable.

52 almost 10-fold increase in purified NBD1 obtained (75 mg NBD1 per liter E. coli culture versus 8 mg per L culture; Table 2.2 and reference 64 ). Furthermore, low concentrations of MgATP result in protein precipitation over time, which considerably reduces the final yield. The fact that high concentrations of ATP (>10 mM) are required for optimal purification likely reflects the low binding affinity of SUR2A NBD1 for nucleotide (see below), as well as that many E. coli proteins also bind ATP. Thus, during the sonication and Ni 2+-affinity purification, when many contaminating proteins are present, additional ATP is needed to saturate NBD1. The only stage where the MgATP concentration is at 2 mM, as in the original protocol, is during the size exclusion column step used to separate the isolated NBD1 from the 6xHis-SUMO tag. Because the size exclusion column results in dilution of NBD1 by 10-fold, high concentrations of MgATP (i.e. 10 mM) are not required to prevent aggregation of the sample. Further, because NMR studies require milligram quantities of NBD1, the size exclusion column is run repeatedly and uses a total of ~500 ml of buffer. Thus, a lower MgATP concentration minimizes the amount of ATP used during this step. Table 2.2: MgATP concentration affects the final NBD1 purification yield Concentration of MgATP used in the Final yield of purified NBD1 purification (mM) (mg / L of M9 minimal media) 2 8 5 44 10 75 15 75

We also adjusted the purification buffers on the basis of the other screens. The Ni 2+ column elution buffer and size exclusion column buffer were at the optimal pH of 7.3 for NBD1 rather than at pH 7.6 as in the original protocol. Decreasing the pH from 7.6 to 7.3 stabilizes NBD1 by ~3 °C (Fig. 2.4 and Table 2.1). The lysis and equilibration buffers were kept at pH 8.0 and 7.6, respectively, to allow for efficient binding of the 6×His-SUMO-NBD1 to the Ni 2+ affinity column. NaCl, which was shown to be destabilizing to NBD1 (Fig. 2.4 and Table 2.1), was minimized during lysis and at the Ni 2+ column elution, the two steps where the protein is most concentrated. NaCl was not added to the lysis buffer and the Ni 2+ elution fractions were diluted with a buffer lacking NaCl. The dilution reduced the concentration of NaCl and also lowered the imidazole concentration necessary for the activity of Ulp1 protease, which is used to remove the 6×His-SUMO tag. Note that all sources of salt cannot be removed at these two steps. MgATP, which was also included in the lysis buffer and dilution buffer, significantly alters the ionic strength of the solution, as nucleotides are acidic and require titration with NaOH to pH 7.

53

Thus, it is imperative to remove added NaCl at these two steps in order to prevent aggregation of NBD1. The NaCl concentration was kept at 150 mM in the Ni 2+ column equilibration buffer to minimize non-specific protein binding to the Ni 2+ resin and in the size exclusion column buffer to prevent adsorption of the protein to the Superdex resin. Knowledge of the destabilizing effect of NaCl on NBD1 allowed for determining conditions for successful lyophilization of NBD1. As expected from the thermal stability screen, 150 mM NaCl in the storage buffer prior to lyophilization resulted in aggregation of the re-dissolved NBD1 samples. Eliminating NaCl from the storage buffer allowed for effective lyophilization of NBD1. Lyophilized NBD1 could be re- dissolved and NMR spectra of the re-dissolved NBD1 were identical to spectra of NBD1 samples that were not lyophilized. In addition, removal of NaCl from the storage buffer allowed NBD1 to be concentrated to ~500 m M prior to lyophilization. Achieving NBD1 concentrations of ~500 m M was not possible when the storage buffer contained NaCl.

Other additives that increase NBD1 stability were included in the lysis buffer and then either eliminated or reduced during subsequent purification steps. The lysis buffer included 2 mg/ml deoxycholic acid, 0.2 % (v/v) Triton-X100, 100 mM L-Arg, and 10 % glycerol. Deoxycholic acid is a bile acid that has a non-polar face and polar face, located on opposite sides of the fused ring structure, that may preferentially interact with exposed protein hydrophobic and hydrophilic surfaces, respectively. Similarly, monomers of the amphipathic Triton X-100 can bind proteins. Deoxycholic acid and Triton-X100 cause only small increases in NBD1 stability (Figure 2.4 and Table 2.1) and were only used in the lysis buffer. Glycerol is a natural organic protective osmolyte that stabilizes the folded state of a protein.101, 105, 160, 289 Glycerol has a large effect on NBD1 stability (Figure 2.4), and thus was maintained throughout the purification and in the final sample, albeit at varying concentrations depending on the purification step. Glycerol was kept at 10 % (v/v) during the lysis, but then reduced to 2 % (v/v) during the Ni 2+ column purification and size exclusion purification step. Reducing the levels of glycerol decreased the overall time of the purification. Column flow rates could be significantly increased, which greatly accelerated the purification times, especially the time required for the gel filtration step. Further, centrifugal concentration times of NBD1 either for loading onto gel filtration or final sample preparation were reduced. We wished to decrease the time for concentrating SUR2A NBD1 samples, as extensive centrifugal concentration times have been shown to result in aggregation/precipitation for other NBDs.247 NMR samples of NBD1 contained 2 % (v/v)

54 glycerol, which balances the need for glycerol for protein stability with the requirement of small amounts of glycerol for NMR spectroscopy, as glycerol increases the viscosity of solutions, which results in broadening of NMR signals. Although L-Arg prevents protein aggregation,9, 10, 103, 166, 249, 324 it was not included in NMR samples because at concentrations of > 50 mM required to stabilize NBD1, L-Arg resonances from natural abundance 15 N are seen in NMR spectra. In summary, changes to our original purification protocol allow for a 10-fold increase in the yield of NBD1 samples that are highly pure (Figure 2.6).

2.3.3 Nucleotide Titrations by NMR spectroscopy

We used NMR spectroscopy to probe binding of nucleotides and ATP analogues to NBD1. Because we cannot acquire high quality NMR spectra of NBD1 in the absence of nucleotide,64 we cannot use the apo NBD1 state as the initial point in the titration. The nucleotide concentration was varied from 1 mM to 50 mM, while keeping the concentration of NBD1 constant at 200 μM. Titration of NBD1 with MgATP results in chemical shift changes for ATP binding for 18 backbone resonances (Fig. 2.7A, B) and 2 resonances derived from side chain Trp indoles (Fig. 2.7A, B). One of the Trp indole resonances is from Trp 616, which we assigned by making a deletion mutant that removes residues S615-R617 from our original construct.64 The other Trp indole resonance is likely from Trp 756, which is located in the Q loop that in other NBDs coordinates the g -phosphate group of ATP.100, 130, 175, 248, 288, 315, 337 We do not have assignments for other resonances in SUR2A NBD1 and thus we cannot map other chemical shift changes to specific residues. However, there is a strong correlation between the magnitude of the combined chemical shift change as a function of MgATP concentration ( Dd total ) for multiple resonances. Figure 2.7Ci displays the correlation between the Dd total values for the resonance marked with an asterisk ( “ * ” ), which is highlighted in Fig. 2.7A and B ii , with the Dd total values for four additional backbone resonances (Fig. 2.8A,Bi) and the two Trp indole resonances (Fig.

2.7A, 2.7Biii ). The Dd total values for two additional backbone resonances from the disordered region of the protein (Fig. 2.8A, labelled with a grey diamond and “ + ”) are also correlated (Fig.

2.7Cii ). Correlations for the Dd total values for the remaining 11 backbone resonances are not observed, likely because those resonances suffer from peak overlap. Chemical shift changes are expected for residues that directly bind nucleotide. Consequently, many of the chemical shifts observed are likely from residues in the Walker A and/or Walker B motif. However, chemical

55

Figure 2.7 MgATP binding causes chemical shift changes for specific NBD1 resonances. (A and B) Comparison of 2D 15 N–1H TROSY-HSQC spectra of SUR2A NBD1 (200 μM) with varying concentrations of MgATP and in 20 + 15 1 mM Na phosphate, pH 7.3, 5 mM DTT, 2 % (v/v) glycerol, 10 % D 2O at 30 °C. (A) The full 2D N– H TROSY- HSQC spectrum of SUR2A NBD1 with 1 mM MgATP is shown in the background with resonances from backbone and Trp, Asn, and Gln side chain HN groups colored black. The blue resonances, caused by spectral aliasing, are likely from Arg side chain NeHe and Gly backbone HN correlations. The spectrum of SUR2A NBD1 with 50 mM ATP is shown in the foreground. Resonances colored red and green in this spectrum correspond to the black and blue resonances, respectively in the spectrum of SUR2A NBD1 with 1 mM MgATP. (continued on next page)

56

Continued from Figure 2.7 (B) Selected regions of the 2D 15 N–1H TROSY-HSQC spectrum showing the full titration. Spectra were recorded on 200 μM NBD1 in the buffer conditions described above with the following MgATP concentrations: 1 mM (black), 5 mM (purple), 10 mM (green), 20 mM (blue), 50 mM (red). Arrows in the panels A and B indicate the direction of the chemical shift changes during the titration. The symbols in panels A and B highlight specific resonances which display correlations in their chemical shift changes during the titration, with the correlations displayed in panel C. (C) Comparison of chemical shift changes with MgATP for specific SUR2A NBD1 resonances. The graph on the left displays the combined chemical shift changes ( Dd total ) for four backbone (grey X, grey triangle, black star, light grey oval) and two Trp indole (open diamond, dark grey rectangle) resonances plotted as a function of the Dd tota for the backbone HN resonance marked with an ‘‘*’’, which is highlighted in panels A and B (middle). The backbone resonance labelled with a grey X is highlighted in the top panel of part B. The two Trp indole HN resonances are highlighted in the bottom panel of part B. The graph on the right displays the correlation between the Dd total values for two additional backbone resonances, labelled with a grey diamond and a ‘‘+’’. The concentration of MgATP for each set of points is displayed at the top of the graphs. shift changes may also reflect conformational changes transmitted through the domain by MgATP binding. Because Trp 616 is located in a disordered N-terminal segment of the protein and not in the structured NBD1 core 64 , we attribute the chemical shift changes of the Trp 616 indole resonance as resulting from conformational changes in the protein upon ATP binding. In addition, multiple backbone resonances that exhibit chemical shift changes during the MgATP titration are also likely from disordered regions in the protein (Fig. 2.7A,B,C grey oval, open diamond, grey triangle, black “ + ”, black ). Further, as the concentration of MgATP increases, these resonances exhibit chemical shift changes reflective of more disorder in that they titrate toward the centre of the 1H dimension of 8.2 ppm (Fig. 2.7A). Thus, these chemical shift changes may be due to conformational changes in NBD1 rather than directly from ATP binding. Conformational changes in NBD1, such as movements of lobe I and lobe II are expected upon MgATP binding and dimerization with NBD2, and have been observed in multiple NBDs.149, 315, 335

Note that we observe changes in the NMR spectra of NBD1 throughout the MgATP titration, up to and including a MgATP concentration of 50 mM. Thus, the NMR titration, which uses the natural ligand (Fig. 2.7) rather than the fluorescent analogue (Fig. 2.5), indicates that the affinity of SUR2A NBD1 for MgATP is low (>100 m M), explaining the necessity of high concentrations of MgATP in the purification and sample buffers. Differences in the K d values of NBD1 with ATP and TNP-ATP is also observed for other NBDs, including CFTR NBD1.299 Nonetheless, fluorescent nucleotide analogues have been used successfully to compare binding of NBDs to different nucleotides (eg. ATP vs. ADP) as we have done here, to examine the effect of mutations in NBDs on ATP binding,109, 188, 244 and to assess the effect of drugs on ATP binding.189

57

We have also recorded spectra of NBD1 in the presence of MgADP as well as MgAMP- PNP (Fig. 2.8). Titration of NBD1 with MgADP, MgAMP-PNP, or MgATP results in similar chemical shift changes for many NBD1 resonances. Thus, the spectrum of NBD1 with MgATP (Fig. 2.8, black and blue peaks) is very similar to spectra of NBD1 in the presence of MgADP (Fig. 2.8, red and green peaks) and MgAMP-PNP (Fig. 2.8, red and green peaks). However, there are some chemical shift differences between NBD1 resonances in the presence of MgATP versus MgADP or MgAMP-PNP (Fig. 2.8, arrows). Notably, some of the resonances that exhibit chemical shift changes with MgAMP-PNP also shift during the MgATP titration (Fig. 2.7, arrows highlighted by an asterisk, “*”). In some cases linear chemical shift changes 241 are seen for resonances that shift during the MgATP titration and also shift in the presence of MgAMP- PNP, implying that NBD1 binds MgAMP-PNP with a higher affinity compared to MgATP. Further, there are resonances in the spectrum of NBD1 with MgADP and MgAMP-PNP (Fig. 2.8, purple and cyan circles) that are not observed in spectra of NBD1 with MgATP. Some of the additional resonances are common to spectra of NBD1 with either MgADP or MgAMP-PNP spectrum (Fig. 2.8, purple circles), albeit at slightly different chemical shifts, while others are unique to the specific nucleotide (Fig. 2.8, cyan circles). The additional resonances may either be overlapped in the spectrum of NBD1 with MgATP or may be broadened due to protein motions on the m s-ms timescale. Motions on the m s-ms timescale at multiple positions in the NBD from the Methanococcus jannaschii protein MJ1267 are affected by nucleotide binding.321 Thus, motions in the SUR NBDs are also expected. Notably mutation of residues near the ATP binding site in SUR2A NBD2 does not affect nucleotide binding but only affects the rate of different steps in the ATP hydrolysis cycle, 31 suggesting that these disease-causing mutations alter dynamics in the protein. Resonance assignments and NMR relaxation studies of NBD1 with different nucleotides will highlight dynamic regions of the protein. The higher quality of the spectrum of NBD1 with MgAMP-PNP indicates that this non-hydrolyzable ATP analogue should be used as the NBD1 ligand for the NMR resonance assignment experiments.

2.3.4 Comparison with other screening techniques

This paper presents a thermodynamic screening protocol, which allows for rapid evaluation of potential buffer conditions and additives that promote protein stability, and its application to NBD1 from the mammalian ABC protein SUR2A. The thermodynamic screen presented here is

58

A B

Figure 2.8: Comparison of chemical shift changes in NBD1 with MgATP, MgADP and MgAMP-PNP. The TROSY-HSQC spectrum of NBD1 with 2 mM MgATP is overlaid with the spectrum of NBD1 in the presence of (A) 2 mM ADP and (B) 2 mM MgAMP-PNP. The concentration of NBD1 is 200 μM in each experiment. The spectrum of NBD1 with 2 mM MgATP is in the foreground in both panels with MgADP (A) and MgAMP-PNP (B) in the background in red and green. Purple circles highlight new resonances that are common to spectra of NBD1 with both MgADP and MgAMP-PNP. Cyan circles highlight new resonances specific to the spectrum of NBD1 with either MgADP (A) or MgAMP-PNP (B). Arrows indicate the direction of chemical shift changes for additional NBD1 resonances with MgADP or MgAMP-PNP versus MgATP. A subset of arrows are highlighted by the symbol `*`to indicate resonances that shift with MgAMP-PNP that also change during the MgATP titration.

a variation of the thermal shift assay.213, 314 Also known as the differential scanning fluorimetry assay, this method is used largely in the drug discovery process to monitor binding of low molecular weight compounds to proteins.185, 224 Thermal shift assays involve the use of a fluorescent dye, such as SYPRO orange that is fluorescent only when bound to exposed hydrophobic sites in molten globules or denaturation intermediates, to construct the thermal denaturation curve. Thermal shift assays using SYPRO orange and other dyes can be performed on any instrument that provides precise temperature control and detection of fluorescent dyes, such as real-time PCR machines. Real-time PCR machines also offer the advantage of high- throughput screening of small molecule ligands and stabilizing additives for the target protein. Although we have conducted our screen with a conventional fluorimeter, high throughput screening in our assay is possible with fluorimeters equipped with plate readers.

59

While the thermal shift assay has been successfully used for a number of different proteins, there may be specific advantages to eliminating the dye and conducting the assay in a non-high throughput format. Firstly, the absence of dyes may simplify the screen. Initially, the dye to protein ratio must be determined experimentally in order to optimize the signal and obtain the denaturation curve in which an increase in temperature results in increased fluorescence.158 In the case of SUR2A NBD1 with SYPRO orange, we tried a number of conditions, varying the amount of protein and dye, and only had moderate success in obtaining the expected denaturation curves largely due to high background of a SYPRO orange signal. This is possibly because SYPRO orange binds to multiple sites on denatured intermediates of NBD1. In addition, the advantage of conducting these screens in a non-high throughput format is that proteins with short sample lifetimes are quickly prepared in specific trial conditions and assayed immediately. This approach minimizes the time the protein incubates in different buffer conditions during preparation of separate samples in a 96-well plate. Proteins that have low kinetic stability will show a melting temperature that depends on both the equilibration time and the scan rate.150 For this reason, sample preparation time, as well as the experimental equilibration time, scan rate, and emission collection time should be consistent between different samples. Our approach can also be adapted to using other reporters of protein structure, such as CD spectroscopy. Thermal denaturation studies using CD spectroscopy are also useful if there are components in the buffer that interfere with the fluorescence emission spectra of the protein. Solution conditions that promote protein stability, and thereby minimize aggregation and precipitation, can be further tested using large molecular weight cutoff concentrators to confirm the presence of monomeric proteins.145 Promising solution conditions can be used to prepare NMR samples to record 15 N- HSQC spectra to assess feasibility and/or applied to the purification process to increase protein yield as we have done here.

2.3.5 Application to other proteins

The strategies that we have employed to stabilize SUR2A NBD1 and improve final protein yields can also be applied to purification of other nucleotide binding domains, such as GTPases, as well as other low-stability proteins. Because the screen involves measuring the T m of unfolding using intrinsic Trp fluorescence, low concentrations of protein (~1-2 m M) are used. Thus, this method can be used effectively to screen a large number of conditions for systems with poor protein recovery. In addition to the additives used here, additional osmolytes can be

60 tested, such as glycine, trimethylglycine (i.e. betaine), trimethylamine N-oxide (TMAO), and sarcosine to name a few. Many organic osmolytes are present in nature and serve to protect proteins against denaturating conditions as in human renal medulla cells 212 or the high salt or high temperature environments for extremophiles.63, 171 Knowledge of osmolytes present in cells that also express the protein of interest can provide a starting set of compounds for screening.

2.4 Conclusions

Results from our screening method have enabled a 10-fold increase in the yield of purified SUR2A NBD1 and long-term storage of NBD1 by lyophilization without compromising the stability or NMR spectra of the protein. The increased yield of NBD1 and its increased stability allow us to extend the series of experiments which are now feasible, including NMR resonance assignments and additional NMR relaxation experiments. Conditions that promote NBD1 solubility can be used as a starting point for crystal screens in order to obtain a high resolution X- ray structure of the protein.

61

Chapter 3 Resonance Assignments for SUR2A NBD1 3 Overview

In this chapter, we have expanded our NMR studies of SUR2A NBD1 to acquire residue-level information through resonance assignments of the amide backbone. These studies initially involved screening various conditions through a combination of NMR and fluorescence methodology. The optimal conditions were selected on the basis of protein stability and NMR spectral quality. Samples of the strip plots used in resonance assignment data are also provided. 13 Due to low spectral quality of NMR resonance assignment experiments that provide Cβ correlations, a number of amino-acid specific samples were also generated. As part of the selective amino acid labelling, we have developed a strategy for specific labelling with minimal metabolic scrambling of the 15 N isotope without the use of auxotrophs. This strategy is currently being reproduced as proof of principle with the model proteins human ubiquitin and yeast Smt3 for submission .3 These strategies have allowed for ~65 % complete assignment of the amide backbone of NBD1. We have employed these resonance assignments to examine conformational dynamics of NBD1, such as through relaxation studies and hydrogen-deuterium exchange experiments.

3.1 Introduction

3.1.1 Triple Resonance Assignments

In order to obtain additional residue-level information from NMR spectroscopy the resonance frequency (also known as the chemical shift) of each NMR active nucleus in the molecule must be obtained. In the case of 2D 15 N-1H HSQC experiments, obtaining the 15 N and 1H chemical shift assignments would enable mapping of each resonance in the HSQC experiment to backbone and side chain H-N pairs in the macromolecule. Thus, changes in the HSQC spectrum with ligand binding, mutations, and post-translational modifications can be correlated to specific residues.

3In Preparation

62

NMR resonance assignment of biological molecules, such as proteins, requires the use of three-dimensional triple resonance experiments. As the name suggests, these experiments correlate three different nuclei, 1H, 15 N, and 13 C. The backbone resonance assignment, which will ultimately give the 15 N and 1H resonances observed in a 2D 15 N-1H HSQC, of a 30 kDa protein such as NBD1, generally requires the following six triple resonance experiments: HNCO, HN(CA)CO, HN(CO)CA, HNCA, HN(CO)CACB and HNCACB. The nomenclature for the experiments is based on the pathway for magnetization transfers and the parenthesis indicate the atoms for which magnetization is not detected. Each experiment is initially started as well as detected on the amide proton in order to maximize sensitivity. 38

Magnetization is transferred from the amide 1H nucleus to the 15 N nucleus and then to the 13 C nuclei through INEPT sequences. All of the atoms detected in each of the experiments are illustrated in Figure 3.1. The experiments are generally performed in pairs in which one experiment in the pair gives inter-residue information and one experiment in the pair gives intra- and inter-residue information. For example, in the HNCO, the magnetization transfer pathway is 1 15 13 15 1 described as Hi→ Ni→ Cꞌi-1(t 1)→ Ni(t 2) → Hi(t 3), where t 1, t 2, and t 3 denote acquisition times 13 13 in which magnetization is detected and Cꞌi-1 indicates the C carbonyl carbon on the previous residue (inter-residue). The compliment HN(CA)CO experiment has a magnetization transfer 1 15 13 13 15 1 45 pathway described as Hi→ Ni→ Cꞌi(t 1), Cꞌi-1(t 1)→ Ni(t 2)→ Hi(t 3). Thus a comparison of 1 15 the HNCO and HN(CA)CO at the Hi and Ni frequencies allows for identification of the chemical shift of the 13 C carbonyl chemical shift of the i and i-1 residues. The presence of two peaks in the inter-residue experiments are a result of the magnetization transfer from the amide 15 13 15 N to the Cα atoms. This is because the coupling of N to the inter- and intra-residue Cα atoms is very similar (11 and 7 Hz respectively) which results in magnetization transfer to both atoms. The transfer is more efficient for the intra- correlation and it generally appears with a stronger intensity than the inter-peak. 45

The scalar coupling between the carbonyl 13 C atom and the covalently linked 15 N atom is relatively large (~15 Hz) which allows for efficient magnetization transfer. Although each 13 residue contains two C atoms along the peptide backbone (C α and CO), they possess different chemical shifts (C α ≈ 40-65 ppm and CO ≈ 165-180 ppm). This difference in chemical shift 15 allows magnetization to be transferred selectively from the N to either the C α or CO in each of the resonance assignment experiments. 38

63

Figure 3.1 Atoms detected in six common triple resonance assignments. In each case, the spin system is excited and detected on the amide proton in order to maximize sensitivity of the experiments.

Each resonance assignment experiment varies in sensitivity. Table 3.1 provides a list of the expected correlations for each of the resonance assignment experiments as well as the magnetization transfer pathways. Certain experiments are modified to enhance sensitivity. For example, the HNCACB is often run as the HN(CA)CB for larger, deuterated proteins. The nominal difference between the two experiments occurs with the length of the τ periods during the C α-Cβ INEPT transfers. In the HNCACB, the τ delay (~3.4 ms) allows for both C α and C β cross-peaks to be detected. 38 However, in the HN(CA)CB the delay is 2τ (~6.8 ms) which transfers all the magnetization from the C α to the C β. Although this increases sensitivity for the 38 Cβ atoms, relaxation on the C α during the extended delay period also reduces signal.

Although a full assignment profile could theoretically be generated from only one pair of triple resonance assignment experiments, in practice, generally all six experiments are required.

64

This is mainly due to the lower dispersion of C α and CO chemical shifts and reduced sensitivity for Cβ experiments due to relaxation during multiple magnetization transfers. Furthermore, resonance assignments only map the sequential correlations in the protein but do not indicate which correlation corresponds to which amino acid. In order to assign the resonances, certain inferences can be made based on chemical shift of the C β atoms. The chemical shifts for Cβ atoms are most distinctive in distinguishing amino acids. For example, the chemical shift of the

Cβ atoms for Ala residues generally show correlations at less than 20 ppm, which is far removed from the chemical shifts of C β atoms (30 – 40 ppm) of most amino acids. The C β chemical shifts for Ser (~60 ppm) and Thr (~70 ppm), are also far removed from the chemical shifts of most C β 45 resonances. These distinct C β chemical shifts and the fact that Gly lacks a C β nuclei allows one to use the C α and C β chemical shifts to identify the amino acid type for a few residues, and map the sequential NMR correlations to the sequence. Table 3.1: Magnetization transfer in 3D resonance assignment experiments 3D Experiment Magnetization Transfer Pathway Expected Correlations i - 1 i HNCO H→N→CO( F1 )→N( F2 ) → H( F3 ) X HN(CA)CO H→N→CA→CO( F1 )→N( F2 ) → H( F3 ) X X HN(CO)CA H→N→CO→CA( F1 )→CO→N( F2 ) → H( F3 ) X HNCA H→N→CA( F1 )→N( F2 ) → H( F3 ) X X HN(COCA)CB H→N→CO→CA→CA/CB( F1 )→CA→CO→N( F2 ) → H( F3 ) X HN(CA)CB H→N→CA→CB( F1 )→CA→N( F2 ) → H( F3 ) X X

Because these triple resonance experiments are costly in terms of the spectrometer time required and the 15 N, 13 C, and 2H labelling, we carried out screens to determine conditions for NMR resonance assignment of partially deuterated 13 C/ 15 N labelled SUR2A NBD1. Although we collected the six experiments described above, the poor quality of the C β experiments, in particular the HN(COCA)CB, required NBD1 samples labelled at specific residues to assist in identification of amino acid type.

3.1.2 Specific Amino Acid Labelling

The low sensitivity of the C β experiments is particularly apparent in larger proteins, such as NBD1, and often necessitates additional information to complete backbone resonance assignment for a protein. Often times this involves specific labelling of one type of amino acid within the protein. Several convenient and cost-effective methods have been employed for recombinant proteins expressed in E. coli , each with their own merits. For instance, incorporating specific 14 N atoms in an otherwise 15 N background, termed as 14 N unlabelling

65 strategies, offer a ubiquitous and rapid approach to identifying amino-acid specific resonances in an NMR spectra. 27, 159 This approach has been cleverly optimized to allow for multi-unlabelling profiles to be obtained from a single E. coli culture, although it also requires the use of C β experiments. 174 However, 14 N unlabelling strategies suffer from strong metabolic scrambling for various amino acids including Leu, Val and Ile. Other approaches involve cell-free expression systems that eliminate any scrambling, but require adaptation to a new expression system, which may not always be feasible for the protein of interest. 210, 292 Supplementing metabolic precursors has been used to introduce labelled spin-systems with minimal scrambling, but only as 13 C- labelled side-chains (as opposed to backbone 15 N) labels. Part of the reason for the strong 15 N scrambling, is that during amino acid synthesis, the amide nitrogen atom is incorporated during the final biosynthesis step. For instance, in the case of α-ketoglutarate amino acid derivatives (such as Leu and Val), the enzyme that carries out the final reaction is a specific branch chain transaminase. Scrambling due to the transaminases can be avoided by using auxotrophs. However, this approach may not be optimal for heterologous protein expression in several systems, due to reduced cell growth and protein yields. 326 220

Here we have developed a fast and cost-effective method for 15 N labelling of specific amino acids including, Leu, Ile, Val, Gly and Ser. These residues are often inaccessible to specific labelling/unlabelling strategies due to strong metabolic scrambling. This is unfortunate as these amino acids are well represented in proteins with an average abundance of 7.6 %, 6.8 % and 3.8 % for Leu, Val and Ile, respectively. 99 In order to specifically label these amino acids, we have supplemented E. coli with a minimal amount of the desired 15 N label as well as saturating levels of unlabelled amino acids to allow for reversible feedback inhibition of transaminases. This approach of specific labelling is both efficient and robust as it requires minimal changes to the expression and purification of E. coli recombinant proteins. Further, the inclusion of off- pathway amino acids avoids the use of auxotrophs. Employing these specific labelling samples in combination with triple resonance data allowed assignment of 65 % HN, 62 % N, 73 % CO, 70

% Cα, and 38 % Cβ atoms in SUR2A NBD1. Resonance assignments of NBD1 have allowed residue-specific analysis of the effects of phosphorylation and mutation on the conformation of NBD1, as discussed in Chapters 4 and 5, respectively. Additionally, resonance assignments have allowed for residue-specific analysis of ligand titrations and hydrogen-deuterium exchange studies.

66

3.2 Methods

3.2.1 Protein Expression for NMR Resonance Assignment

Rat SUR2A NBD1-ΔN (D665-L933) fused to an N-terminal 6×His-SUMO tag was transformed in E. coli BL21 (DE3) CodonPlus® (RIL) cells (Stratagene). NBD1-ΔN is lacking an N-terminal disordered region which eliminates a large amount of NMR spectral overlap and therefore makes it more amenable to NMR studies (see Chapter 4). For three-dimensional NMR experiments, the 15 13 transformed E. coli were grown in minimal M9 media supplemented with NH 4Cl and C glucose (Cambridge Isotopes Labs). Since SUR2A NBD1-ΔN is a relatively large protein for 2 NMR studies (~30 kDa), minimal media growths were performed in 70 % H2O as opposed to 2 2 H2O to fractionally incorporate deuterium ( H) atoms in the protein. Incorporation of H reduces relaxation from 1H, which otherwise reduces signal intensity. For expression in deuterated media, the E. coli cultures were grown in several stages to acclimatize their ability to grow in 2 heavy water. Additionally, all solutions (eg. salts, glucose, etc.) were generated in H2O as well.

The E. coli cultures were first transformed with plasmid containing SUR2A NBD1-ΔN. A single colony was grown in 3 mL of unlabelled LB broth with the appropriate antibiotics to maintain selective pressure on the plasmid. The day-culture was used to inoculate a 200 mL overnight culture. The overnight culture contained M9 media, including antibiotics, and was supplemented with 2.5 % of 15 N/ 13 C OD2 media (Silantes). OD2 is a rich growth media used to improve cell culture growth. The overnight cultures were used to inoculate 1 L (70 % D2O) of fresh M9 media containing the same components as the 200 mL overnight culture. As the cell cultures reached OD 600 values of 0.4, 0.6, and 0.8, the incubation temperature was progressively decreased from 37 °C to 30 °C, 25 °C, and 18 °C, respectively. Cell cultures were then allowed to incubate at 18 °C for 30 min prior to induction of protein expression by addition of 0.75 mM IPTG. After ~20 h the cells were harvested by centrifugation and were stored at -20 °C.

3.2.2 Specific Amino Acid Labelling

3.2.2.1 Specific Amino Acid 15 N Labelling

In order to augment the triple resonance assignment data, we generated samples of SUR2A NBD1-ΔN that were 15 N labelled at specific amino acids. The overall protocol involves growing

67

E. coli in 14 N (unlabelled) media and supplementing a labelled amino acid prior to induction. However, E. coli metabolic pathways can lead to scrambling of the 15 N label resulting in labelling of additional residues than the one originally containing the 15 N label. An additional drawback is the cost of the specific label. We have generated samples that are specifically labelled on Gly, Ser, Leu and Val. In order to avoid metabolic scrambling a series of precautions needed to be employed. Leu and Val are members of the pyruvate family of amino acids. The final step of in the biosynthesis pathway involves a reversible transamination with Glu as the nitrogen donor. 99 To inhibit the reverse reaction (which otherwise leads to 15 N scrambling), significant amounts of Glu are supplemented to the media at the initial growth as well as at induction. In addition, equal concentrations of unlabelled pyruvate amino acid derivatives are supplemented. For instance, in the case of 15 N Val labelling, the following 14 N amino acids were supplemented at concentration of 0.5 g/L at the initial inoculation: Leu, Ile, Ala and Glu. Approximately 30 min prior to induction, 25 mg/L of 15 N Val are supplemented to the culture. For labelling of other pyruvate amino acid derivatives, the 15 N amino acid was also introduced to the media prior to induction, and the corresponding unlabelled amino acids were included at the initial growth. In the case of Gly-specific 15 N labelling, 14 N Ser (0.5 g/L) was added at the initial growth stages, with 10 mg/L of 15 N Gly added just prior to IPTG induction. This prevented scrambling of the 15 N label from Gly to Ser. Alternatively, the same growth was performed without the addition of any Ser, to allow for both Gly and Ser to be concomitantly 15 N labelled in NBD1.

3.2.2.2 Specific Amino Acid 15N Suppression Labelling

We also have employed a secondary approach to obtaining specifically labelled samples to augment our NMR resonance data. In reverse labelling or 15 N labelling suppression 159 , E. coli cultures are grown in 15 N enriched M9 media. However, the target ( 14 N) amino acid is provided in large excess to the cultures (0.5 g/L). This discourages E. coli to synthesize the specific amino acid, and bypasses incorporation of the 15 N into the target amino acid. 159 Although scrambling of the unlabelled 14 N amino acid will occur, the correlations for the other non-target amino acids, will still appear in the HSQC spectra (likely with a reduced intensity). The difference spectra of a fully labelled sample and a specifically labelled sample provides the residue-specific peaks. This provides an alternative cost-effective method for amino acid labelling. However, since this

68 method is based on examining the disappearance of a signal, as opposed to a gain of signal (as with 15 N labelling), it is intrinsically less sensitive.

3.2.3 Three Dimensional NMR Experiments

The 3D NMR experiments were set-up by Dr. Tara Sprules at the Quebec/Eastern Canada High Field NMR (QANUC) facility. These experiments were performed on a Varian INOVA 800 MHz equipped with a HCN cyroprobe and xyz -field gradients. Three dimensional HNCO, HNCACO, HN(CO)CA, HNCA, HN(CA)CB, and HN(COCA)CB experiments were performed in order to obtain assignments for the 15 N, 13 C, 13 Cα, 13 Cβ, and 1HN nuclei of each amino acid. The spectra were recorded with the parameters outlined in Table 3.2. Table 3.2 NMR Resonance Assignment Experiment Parameters HNCO HN(CA)CO HN(CO)CA HNCA HN(CA)CB HN(COCA)CB HN(CA)CB t1 (ms) 9.2 12.7 6.0 6.0 3.6 3.6 7.1 t2 (ms) 14.1 28.2 14.1 14.1 14.1 14.1 13.7 1H Spectral 13008.1 13008.1 13008.1 13008.1 13008.1 13008.1 13008.1 Width (Hz) 13 C Spectral 2614.193 3770.100 6032.930 6032.930 14077.007 14077.007 4926.713 Width (Hz) 15 N Spectral 2268.989 2268.989 2268.989 2268.989 2268.989 2268.989 2268.989 Width (Hz) Number of 8 32 16 16 24 24 64 Scans

The HN(CA)CB was run twice, with two different sweep widths in the 13 C dimension. A smaller sweep width improved the signal-to-noise intensity for some correlations, but also leads to spectral-folding which complicates the analysis. NMR spectra were processed with NMRPipe/NMRDraw 75 and analyzed with NMRView.139

3.2.4 NMR and Fluorescence Screening of Ligands

Our resonance assignment was performed on the construct NBD1-ΔN, comprised of residues D665-L933. As previously mentioned, this protein eliminates an N-terminal disordered region (comprising of residues S615-E664) of NBD1 (S615-L933) and therefore reduces significant NMR spectral overlap and also increases the thermodynamic stability of the protein in the presence of MgATP (see Chapter 4). NMR samples for triple resonance assignment experiments need to be concentrated (>500 m M) and must be stable for ~2 weeks at 20 °C -30 °C, prompting us to screen several buffer conditions. In addition, labelling with 2H may affect the stability of the protein requiring additional changes to the buffer conditions. As such, thermodynamic

69 screenings of NBD1-ΔN were performed in a similar manner as described in Chapter 2. In addition to fluorescence thermodynamic screening, a number of samples were also tested by NMR to ensure appropriate spectral quality for given conditions. This involved NMR spectral screening of nucleotide titrations, pH titrations and buffer component conditions. For nucleotide titrations, MgATP was first removed from NBD1 samples through gel filtration. The appropriate nucleotide (previously titrated to pH 7.0 with NaOH) was added immediately to the sample. The pH titrations were conducted in a universal buffer mixture of 100 mM sodium phosphate and 200 mM sodium citrate that has a buffer capacity in the range of pH 3.3-8.2 rather than in phosphate buffer which has a buffering capacity in the pH range of 6.2-8.2. Separate samples were generated by preparing a stock of NBD1-ΔN protein (500 μM) in 20 mM NaPhos pH 7.0. The protein was diluted 5-fold with universal buffer of (100 mM Na 2Phos/200 mM citric acid prepared at various pH values) and the final pH of the resulting solution was recorded. This method of generating separate samples was effective for performing the pH titrations as it bypasses the direction addition of acid or base to the sample. The direct addition of acid or base results in a high local concentration of the acid or base at the point of addition, which may denature protein molecules in the vicinity of the added acid or base before mixing can occur. The denaturation of a few protein molecules can lead to aggregation through the sample. Because this method ensures that the concentration of the protein is identical between all samples, it was also employed to test the effect of NaCl, glycerol, and guanidinium hydrochloride.

3.3 Results and Discussion

3.3.1 Triple Isotopically Labelled Protein Expression

In order to express SUR2A NBD1-ΔN (D665-L933) as a 15 N/ 13 C/70 % 2H labelled protein, the growth media was altered to incorporate the different isotopes. Isotopic enrichment requires growing E. coli cells in minimal media, which results in slower growth rates compared to cells grown in rich LB media (Figure 3.2, red curve). Expression and growth for only 15 N isotopically enriched protein (as previously done) occurs in 95 % M9 minimal media with 5 % LB to generate a larger cell mass, albeit with a slightly reduced isotopic enrichment of the protein. The slight reduction in 15 N incorporation does not significantly affect 2D 15 N-1H HSQC spectra as magnetization is only transferred between two atoms. However, for resonance assignment experiments, the magnetization is transferred multiple times between 15 N and 13 C atoms, and

70

Figure 3.2 Growth curves of transformed BL21 (DE3) RIL Codon Plus E. coli cells in different media. The cells were induced with 1 mM IPTG at an OD 600 of ~0.8. Cell harvesting was performed at ~16-20 h after induction. between sequential 13 C atoms. As such, any unlabelled atoms would disrupt the pathway. Therefore 100 % labelling is an essential prerequisite for resonance assignment and 5 % LB cannot be used. Fortunately, there are a number of isotopically-labelled rich medias that are commercially available. We have employed OD2 media as a substitute for LB. The addition of 2.5 % (v/v) OD2 media in the minimal M9 media increased the growth rate of E. coli similarly to the minimal media alone when 70 % D 2O is used (Figure 3.2, blue vs cyan curve). The growth rate is slower in 100 % D 2O (Figure 3.2, green curve) because the heavy water slows the kinetics of metabolic reactions in E. coli and also reduces the stability of some E. coli proteins. 318

Although the E. coli cultures were grown in 100 % deuterated media, for NMR studies the amide protons need to be exchanged back to 1H. The aliphatic and aromatic protons remain as deuterium. This usually occurs spontaneously during the purification process, as the 1 purification buffers are prepared in H2O which will exchange with the amide backbone. However, as shown in Figure 3.3, following purification of NBD1-ΔN expressed in 100 % deuterated media, not all resonances are present in the 15 N-1H HSQC. This implies that not all deuterium atoms are exchanged to 1H. The resonances which are absent likely correspond to residues within the core of the protein which are inaccessible to solvent. To rectify this problem the protein was incubated at 30 °C in order to increase amide-solvent exchange rates. Furthermore, 100 mM guanidinum hydrochloride was added to the protein solution, to assist in

71

Figure 3.3 Comparison of TROSY-HSQC of NBD1-ΔN purified from E. coli grown in 100 % deuterated media (red) and 0 % deuterated (normal) media (black). There is an offset between the two spectra due to the deuterium isotope effect. Not all resonances are observed when the protein is expressed in fully deuterated media. unfolding and allow access to protein interior. Unfortunately, the resonances could not be recovered after these incubation periods. Further increases in the incubation temperature or denaturant concentration were not attempted due to irreversibly unfolding of the sample. To overcome this problem, the growth media was reduced to 70 % D 2O / 30 % H 2O. This compromises the level of deuterium that will be incorporated into the sample. However, following purification of the partially deuterated protein, all resonances were observed in the 15 1 N- H HSQC. Therefore, the 70 % D 2O mixture was employed for the growth media.

3.3.2 Optimization of NBD1-ΔN samples for NMR spectroscopy

3.3.2.1 Temperature

Since SUR2A NBD1-ΔN samples need to be stable at concentrations of 500 μM for several days in order to acquire the 3D NMR data sets, we screened for conditions that promote long term stability of the protein. In order to determine the temperature to carry out the resonance assignment experiments we examined a series of HSQC spectra of NBD1-ΔN with increasing temperature from 10 °C - 50 °C in steps of 5 oC. As shown in Figure 3.4B (red circles), a small subset of the peaks, most notably the arginine side chains, are of reduced intensity with increasing temperature. However, the correlations of the HSQC for the vast majority of peaks

72

30oC

25 oC

Figure 3.4: TROSY-HSQC temperature profile for NBD1-ΔN (70 μM). The temperature was varied from 10-50 °C in steps of 5 oC. Only NMR spectra from 10, 20, 30 and 40 °C are shown in (B). The Arg NεHε side chain correlations are circled in red. The peak intensity for one representative peak is plotted as a function of temperature (A). To assess sample life time, two 500 μM NBD1-ΔN samples were incubated at 30 °C and 25 °C. Aliquots were removed from the samples each day, centrifuged to remove any precipitate and run on an SDS-PAGE gel. were most intense in the range of 30-35 °C. In addition to peak intensity, the life time of the sample must also be considered when screening for the optimal temperature. To investigate the long-term stability of SUR2A NBD1-ΔN, two 500 μM NBD1-ΔN samples were independently incubated at 30 °C and 25 °C for 7 days. Each day, an aliquot was removed and centrifuged to clear any precipitated protein. The supernatant was run on an SDS-PAGE gel to evaluate the relative concentration of protein remaining in solution (Figure 3.4B). The protein sample

73 incubated at 30 °C, showed signs of more significant precipitation following the 48 h time point, whereas the sample incubated at 25 °C was stable for 6 days. Thus we decided to run our resonance assignment experiments at this temperature.

3.3.2.2 Nucleotide and Buffer Component Screening

Using the thermodynamic stability assays described in Chapter 2 for NBD1, we screened optimal buffer conditions for NBD1-ΔN. As described in Chapter 2, the presence of MgATP had the largest influence on thermodynamic stability. Based on these results, we have probed the effect of increasing MgATP concentration on the thermodynamic stability and secondary structure of NBD1-ΔN. The fluorescence melting studies (Figure 3.5), suggested that 2-5 mM MgATP would be sufficient to improve the thermodynamic stability of NBD1-ΔN. We also performed nucleotide titrations with MgATP and MgADP by NMR and circular dichroism to examine the effect of the nucleotide on the protein (Figure 3.5-3.6). At high concentrations of MgATP or MgADP (>5 mM) sharp ridges appear in the NMR spectra due to natural 15 N abundance from nucleotide. These noise peaks can theoretically be removed with proper gradients but this can also reduce protein signal and complicates the pulse sequences further. Based on the NMR titration studies, we opted to maintain 5 mM MgATP in NBD1-ΔN samples.

We have also screened NBD1 in the presence of different nucleotide ligands (AMP- PNP/AMP-PCP; Figure 3.7). The use of non-hydrolyzable ATP analogues was explored as there is precedent for hydrolysis of nucleotide over time with other nucleotide-binding proteins. This hydrolysis may result in chemical shifts in NMR spectra over time, undermining the utility of the triple resonance assignment experiments. In comparison to the natural ligand, the non- hydrolyzable analogues are modified at the γ-phosphate to make them less susceptible to hydrolytic cleavage. However for SUR2A NBD1-ΔN, ATP hydrolysis was found to occur at an extremely low rate, likely due to the requirement for dimer formation with complementary SUR2A NBD2. Nonetheless, we recorded NMR spectra of NBD1-ΔN with alternate nucleotides. Spectra of NBD1-ΔN with different ATP analogues were comparable in spectral quality and intensity (Figure 3.7). Notably, the same peaks that shift with nucleotide binding also display chemical shifts in the presence of ATP-analogues, which suggests these residues may be involved in nucleotide binding. We have also screened these analogues to examine their effect on

74

A

B C

Figure 3.5 ATP titrations with NBD1-ΔN (A) 40 min TROSY HSQC of 100 μM NBD1-ΔN with only the beginning and end titration points of 1 and 20 mM MgATP shown in black and red respectively. The insets show the full titration range of 1, 2, 5, 10 and 20 mM MgATP in black, yellow, green, blue, and red respectively. (B) Thermodynamic profile of 2 μM NBD1-ΔN with varying concentrations of MgATP as determined by intrinsic Trp fluorescence. The melting temperature was determined as described in Chapter 2. (C) CD spectra of 10 μM NBD1- ΔN at 15 °C with increasing concentrations of MgATP. Spectra are shown for 0 (black), 10 (red), 20 (blue), 30 (green), 40 (cyan), 50 (orange) and 150 (yellow) μM MgATP concentrations. The spectra are the average of 3 separate scans and were independently blanked against buffer containing the corresponding concentration of MgATP.

75

A

B

Figure 3.6 MgADP titrations with NBD1-ΔN (A) 40 min TROSY HSQC of 100 μM NBD1-ΔN with only the beginning and end titration points of 1 and 20 mM MgADP shown in black and red respectively. The insets show the full titration range of 1, 2, 5, 10 and 20 mM MgADP in black, yellow, green, blue, and red respectively. (B) Thermodynamic profile of 2 μM NBD1-ΔN with varying concentrations of MgADP (left), MgAMP (centre), or MgAMP-PNP (right) determined by intrinsic Trp fluorescence. The melting temperature was determined as described in Chapter 2. MgADP and MgAMP were screened over the concentration range of 0 - 10 mM. MgAMP- PCP was screened over a more reduced range of 0 - 1 mM nucleotide due to the cost of the ligand.

76

Figure 3.7 TROSY-HSQC of NBD1-ΔN with non-hydrolyzable ATP analogues. NBD1-ΔN with 5 mM MgATP is shown in the background in black in all spectra. The spectra of NBD1-ΔN with (A) 5 mM MgAMP-PCP or (B) 5 mM MgAMP-PNP is shown in cyan and red respectively. The insets show the same regions as in Figure 3.5 (MgATP) and 3.6 (MgADP) with to illustrate the effects of the nucleotide on NBD1.

77 thermodynamic stability of NBD1-ΔN. Since the greatest enhancement in melting temperature was achieved with the ATP (Figure 3.5-3.6), we opted to employ the natural ligand for resonance assignment experiments.

We also examined the effect of other ligands and buffer conditions on NMR spectra of SUR2A NBD1-ΔN, including buffer pH, salts and osmolytes (Figure 3.8-3.10). The pH of the solution is an important parameter to optimize as it affects the amide exchange rate with bulk water and influences the stability of the protein (Chapter 2). A pH screen of NBD1 is shown in Figure 3.8. NMR spectra were recorded on samples at varying pH values (4.4 - 8.0). NMR peak intensities increase for lower pH values due to slower amide proton off-rates. At a pH of 4.4, the protein still appears to retain significant levels of secondary and tertiary structure, although certain peaks show chemical shifts that indicate the protein is unfolding (Figure 3.8A). Thermodynamic screens by fluorescence, at various pH values, suggest maximal protein stability in the range of 7.0-7.3 which corresponds well to physiological conditions (Figure 3.8Bi) and what we found for NBD1 (S615-L933). In this case, there is a compromise between increased NMR signal for lower pH values, and increased protein stability. Consequently, we have decided to use pH 7.0 for our NMR studies.

Titration of the protein with increasing (buffered-) acid and following the changes in intrinsic Trp fluorescence results in the pH profile shown in Figure 3.8Biii. The pKa is shown to be 4.92 whereas the theoretically calculated pKa is predicted to be 5.21. However, this discrepancy may be a result of measuring the local changes in the pKa around the Trp residues and can also be influenced by conformational changes. Fitting the data to modified Henderson- Hasselbalch equations that account for multiple pKa values (due to multiple Trp residues) may also provide additional accuracy. Plotting the chemical shift of each NMR peak as a function of the buffer pH, shows the pKa for each amide proton. A sample pH titration curve is displayed for one peak in Figure 3.8Bii. The curve is fit to the Henderson-Hasselbalch equation. Not all the peaks demonstrate a standard fit to the equation, likely because a specific resonance is affected by more than one titratable group or the protein is unfolding.

We have also screened the effects of other buffer components such as NaCl (Figure 3.9), glycerol (Figure 3.10A) and guanidinium hydrochloride (Figure 3.10B) on the basis of thermodynamic stability and NMR spectral quality for NBD1-ΔN.

78

i ii iii

Figure 3.8 pH titrations with NBD1-ΔN. (A) 40 min TROSY HSQC of 90 μM NBD1-ΔN with sample titration points of pH 4.4, 6.5 and 8.0 shown in black, blue and red respectively. The insets show the full titration range of pH 4.4 (black) , 5.3 (violet), 6.1 (yellow), 6.5 (blue) 7.0 (green), 7.5 (cyan) and 8.0 (red). (B) i)Thermodynamic profile of 2 μM NBD1-ΔN in buffers of varying pH values. The melting temperatures were determined as described in Chapter 2. (ii) Proton chemical shift of a sample peak from the NMR titration series as a function of pH. (iii) pH titration monitored by intrinsic Trp fluorescence. 2 μM NBD1-ΔN was titrated with aliquots of 1N HCl in phosphate buffer. The pH and fluorescence were recorded after each addition. Since the fluorescence will decrease due to increasing volume, the trials were normalized to a blank run of titrating only phosphate buffer (n = 3). The data in (Bii) and (Biii) were both fit to the Henderson-Hasselbalch equation.

79

A

B

Figure 3.9 NaCl titrations with NBD1-ΔN (A) 40 min TROSY HSQC of 100 μM NBD1-ΔN with only the beginning and end titration points of 0 and 300 mM NaCl shown in black and red respectively. The insets show the full titration range of 0, 50, 150, 200 and 300 mM NaCl in black, yellow, green, blue, and red respectively. (B) Thermodynamic profile of 2 μM NBD1-ΔN with varying concentrations of glycerol, NaCl, DTT, DCA, Arg and, Glu.

80

A

B

Figure 3.10 Osmolyte titrations of 100 μM NBD1-ΔN with (A) GdmHCl and (B) Glycerol with only the beginning / end titration points of (A) 0, 300 mM GdmHCl or (B) 2, 15% glycerol shown in black, red respectively. The insets show the full titration range of 0, 100, 200, and 300 mM GdmHCl or 2, 5, 10 and 15% glycerol in black, blue, green and red respectively.

81

Overall, the additives and buffer conditions are important for both protein stability and sample lifetime but cannot be used in an excessive amount or they detract from the quality of the NMR spectra. These screens allowed for us to determine useful conditions for NMR resonance assignment spectra.

3.3.3 Resonance Assignment Experiments

Initially, using the HNCO, HNCACO, HN(CO)CA, HNCA and HN(CA)CB only allowed for assignment of ≈10 % of the backbone resonances. A strip plot from the corresponding experiments taken at the amide 1H and 15 N chemical shifts for residues A880 to S885 is shown in Figure 3.11. The main reason for the low sequence coverage was due to low quality C β experiments. Specifically, the HN(COCA)CB experiment could not be successfully run at QANUC on an 800 MHz spectrometer and the majority of visible correlations in the HN(CA)CB were barely above the noise threshold. In the absence of viable C β correlations and with such low sequence coverage, other experiments were required. The most accessible experiments involved amino acid specific labelling.

3.3.3.1 14 N Unlabelling Strategies

The most accessible experiments involved amino acid specific unlabelling. SUR2A NBD1-ΔN 15 14 was expressed in minimal M9 media containing NH 4Cl and an excess of a specific ( N) amino acid. Using this method, SUR2A NBD1-ΔN samples were made that contained the following 14 N-labelled amino acids in an otherwise 15 N labelled background: Ala, Asn, Lys, His, Met, Thr, Val, Glu, Trp, Tyr, Phe, or Cys (Figure 3.12). For instance, there are 14 Arg residues in NBD1- 15 ΔN. NBD1-ΔN was expressed in minimal media that was supplemented with both NH 4Cl and 14 N Arg (0.5 g/L) as the only nitrogen sources. NMR spectra of the resulting protein are missing 15 specific correlations when compared to NBD1-ΔN that is grown in media with only NH 4Cl (Figure 3.12, top left panel). The absent correlations likely correspond to Arg residues. Although 14 peaks are expected to disappear, only 9 peaks were completely absent from the spectra. This may be due to a number of reasons including resonance overlap. From these 9 peaks, only 6 were assigned to the sequence based on triple resonance data. Although several growths were performed to unlabel specific amino acids, certain NMR spectra were inconclusive in the number of peaks which disappeared. This is most likely a result of metabolic scrambling as observed for

82

Figure 3.11 Strip plots of sequential correlations from the resonance assignment data. In the top profile, the initial strip shows a correlation from the HN(CO)CA. An adjacent strip is taken at the same 1H frequency in the HNCA. One correlation in the HNCA strip matches in 13 C frequency to the HN(CA)CO strip. The 13 C frequency of the remaining correlation identifies the next strip in the HN(CO)CA. Similar strips are shown for the HNCO, HN(CA)CO and HN(CA)CB. The HN(COCA)CB was of poor quality and did not produce any viable correlations. Based on the C β frequencies, these strips unambiguously correspond to A880-S885.

83

Figure 3.12. NMR spectra of NBD1-ΔN expressed with specifically unlabelled amino acids. In each case, the appropriate 14 N amino acid was supplemented to 15 N enriched media. Spectra of SUR2A NBD1-ΔN containing the unlabelled amino acid are shown in the foreground in red while spectra of the fully labelled protein are shown in the background in black. Peaks that are absent in the unlabelled spectra correspond to the specific 14 N-amino acid type. Spectra are shown for Arg, Ala, Thr, Lys, His, Met, Val, Gln. The number of expected resonances that should disappear (based on the amino acid abundance in the protein), the number of absent resonances that are actually observed, and the number of resonances that have been assigned are also provided below each spectra.

84

Gln, Val, Ile, Trp and Phe unlabelled samples. Other residues which were known to have strong metabolic scrambling, such as Leu, Asp, Asn and Glu were not attempted.

3.3.3.2 15 N Labelling Strategies

Since 14 N unlabelling is not conclusive for some amino acids, we turned to using 15 N specifically labelled amino acids. There are two important problems that arise with the use of 15 N labels. First, large amounts of 15 N labelled amino acids are difficult to synthetically prepare and therefore are expensive. Second, any metabolic scrambling of the 15 N label to other amino acids is quite prominent and will result in additional peaks in the spectra, undermining the utility of this approach. Depending on where these amino acids are located in the protein, their intensity can vary. For example, resonances from unstructured loops or termini are very intense and can be observed, even if the sites are labelled to a low extent. Initial attempts to solely label with only 15 N-Leu or 15 N-Val were unsuccessful due to strong metabolic scrambling (data not shown). In order to limit metabolic scrambling in E. coli , the media was supplemented with different components along the biosynthesis pathway. For example, Table 3.3 shows the extent of scrambling of 15 N-labelled Val when additional amino acids are supplemented. The metabolic scrambling can be prevented by reversibly inhibiting the catabolic pathways of the amino acid of interest. Scrambling of the 15 N label from 15 N-labelled Val can be inhibited by including 14 N- labelled Leu, Ile, Ala and Glu. Addition of the 14 N-labelled Val in addition to 15 N-labelled Val reduces the amount of labelling. The main advantage is that the growth media uses remarkably low quantities (~ 5 mg / 100 mL culture) of 15 N labelled amino acids, which eliminates additional resonances observed from scrambling and also reduces the cost of labelling. We have used this approach to selectively label proteins with 15 N-Leu, 15 N-Gly, and 15 N-Ser, in addition to 15 N-Val. The corresponding specifically labelled spectra are shown in Figure 3.13. This labelling system can be exploited for several purposes. Primarily, it can assist with triple resonance assignment analysis by confirming the identity of specific amino acid types, as Table 3.3 Extent of Detectable Metabolic Scrambling with Different Growth Conditions Number of Peaks Observed Detectable Metabolic Growth Condition (16 Val in NBD1-ΔN) Scrambling? 15 N Val 36 Yes 15 N Val + 14 N Leu + 14 N Ile 23 Yes 15 N Val + 14 N Leu + 14 N Ile + 14 N Glu + 14 N Ala 16 No 15 N Val + 14 N Val + 14 N Leu + 14 N Ile + 14 N Glu+ 14 N Ala 8 No

85

Figure 3.13 NMR spectra of NBD1-ΔN expressed with specifically 15 N labelled amino acids. In each case, the appropriate amino acid was supplemented to the growth media as a 15 N label shortly before induction. Additional additives were included in the media to prevent metabolic scrambling. In each case, the spectra of the fully labelled protein is shown in black. The spectra that correspond to Gly/Ser are shown in green, Gly in cyan, Leu in yellow and Val in purple. The number of expected resonances that should be present (based on the amino acid abundance in the protein), the number of resonances that are actually observed, and the number of resonances that have been assigned are also provided below each spectra.

86 was the case with SUR2A NBD1-ΔN. Since the pyruvate amino acid derivatives have the highest abundance in protein composition, they provide excellent sequence coverage in order confirm/assist in sequential backbone assignments. These strategies can also be useful in cases where the quality of triple resonance assignment is poor to augment the number of assignment obtained. To improve the quality of these experiments, shorter sweep widths can be employed in the 13 C dimension (for eg. from 10 - 80 ppm to 25 - 50 ppm). This significantly improves signal quality and shortens the time of the experiment. However, it leads to spectral folding of peaks, which results in resonances of negative intensity that are ambiguous to assign. For instance, a resonance that appears with negative intensity at 45 ppm in an HN(CA)CB experiment (performed with a sweep width of 25 ppm) could potentially correspond to a Gly, Ser, Thr or Ala. Identifying the type of the residue of the peak would therefore remove the ambiguity of the spectral folding.

Alternatively, these labelling strategies can be employed to simplify HSQC spectra, especially for larger proteins where there is significant peak overlap and it is difficult to correlate changes to individual resonances. In such cases, labelling of Leu, Val or Ile would provide site specific information. We are currently reproducing these 15 N labelling strategies with the model protein, ubiquitin, as proof of principle for the utility of this method.

Based on the additional information for 10 specifically labelled/unlabelled amino acids, for which conclusive evidence could be obtained, the resonance assignment sequence coverage was extended from 10 % to ~65 %. This is a significant improvement and increases the confidence in the resonance assignment. The specific resonance assignments for NBD1-ΔN are depicted on the TROSY-HSQC in Figure 3.14.

87

Figure 3.14 Resonance assignments for SUR2A NBD1-ΔN. The residue number is provided for all assigned peaks in the TROSY-HSQC of D665-L933 SUR2A NBD1-ΔN. For clarity certain resonance assignments are omitted for peaks close to the centre of the spectra. (B) shows close-up of the resonance assignments in the overlapped regions and includes the resonance assignments omitted from (A). Approximately 65 % of all amide peaks were assigned.

88

3.3.4 NBD1-ΔN Resonance Assignment Analysis

Based on the 3D NMR and labelling/unlabelling data, we were able to obtain assignments for 65

% HN, 62 % N, 73 % CO, 70 % C α, and 38 % C β atoms of SUR2A NBD1-ΔN. The assigned residues for are plotted on the surface of a model for SUR2A NBD1-ΔN in Figure 3.15. There are certain regions for which resonance assignments could not be obtained. These include a 15 residue stretch between β4 and β5 strands in lobe I of NBD1-ΔN, and a patch of 19 residues in between α4 and α5 helices in lobe II. Lobe I is formed by the ATP-binding α/β domain and β- sheet subdomain, while lobe II is comprised of the α-helical subdomain (Chapter 1). The 15 residues between β4 and β5 strands include residues of exon 17 in NBD1-ΔN, which is absent in the naturally occurring splice isoform, SUR2D. 50 We have also analyzed SUR2D NBD1-ΔN which has allowed us to identify the resonances in the 1H-15 N HSQC that likely correspond to this region, based on the peaks that are absent in spectra of SUR2D NBD1- ΔN (see Chapter 5). However, specific assignments were not conclusively established for this region. The 19 residues in between α4 and α5 helices in lobe II include the signature C motif. The correlations for residues N- and C-terminal to these regions have low signal intensity suggesting low quality correlations in these regions. Both sequences also have a repeat sequence of SRSR and QRQR respectively, which also hinders their resonance assignment due to overlap. The internal motions of the protein can dramatically affect the quality of resonance assignment data. To characterize the fast-time scales of motion (ns-ps) of the protein, we have

Rotate180 o

Signature Sequence

Exon 17

Figure 3.15 Resonance assignments plotted on surface of SUR2A NBD1-ΔN model. Surface coloured light blue represented regions for which resonance assignments have been obtained, and regions in white have not been assigned. Dark grey regions represent areas where resonance assignments were obtained for the HSQC at 800 MHz at 25 °C could not be transferred to HSQC spectra at 600 MHz 25 °C. Certain regions lacking resonance assignments are indicated on the model.

89

15 15 examined N longitudal relaxation rates, R 1, and the N transverse relaxation rates, R 1ρ (in order to calculate R2). R2 relaxation rates are often elevated around regions of the protein that are challenging to assign. The R 1 rate fluctuates as a result of chemical shift anistropy and dipolar coupling interactions. Differences in the R 1 rate throughout the backbone also arise from various 258 conformational motions such as the flexibility of a disordered region or loop. The R 2 rate also depends on chemical shift anistropy and dipolar interactions, similar to R 1 relaxation, but it is more sensitive to magnetic fluctuations as with large biomolecules. 257

We have recorded R 1 and R1ρ relaxation rates for SUR2A NBD1 at different protein concentrations as shown in Figure 3.16 (R 1 rates not shown). The R 1ρ rates for NBD1-ΔN samples at 200 μM and 500 μM are ~30 s -1, similar to what we have observed with monomeric NBD1.189 Resonances that were identified to belong to exon 17, through comparison of the

SUR2A NBD1-ΔN and SUR2D NBD1-ΔN spectra, were shown to have very low R 1ρ rates (<15 s-1) suggesting these residues are present in a loop. Resonances around the signature sequence do not possess unusually large relaxation rates, which suggests the poor signal quality may arise from other factors. As expected from their chemical shifts and intensity in the HSQC spectra, the residues of the C-terminal tail (Q915-L933) show very low R 2 rates indicating the residues have limited secondary structure and are present in a disordered loop. Note that R1 and R 2 rates are elevated at high protein concentrations (1 mM) which suggest that the protein aggregates or oligomerizes. Dilution of concentrated NBD1-ΔN samples restores the R 1 and R 1ρ rates to their expected levels (data not shown).

Figure 3.16 Concentration dependence of R 1ρ relaxation rates. The relaxation rates are plotted as a function of residue for different concentrations of NBD1-ΔN (200 μM, 500 μM and 1000 μM). At 1 mM NBD1-ΔN, the R 1ρ rates are significantly elevated suggesting higher order oligomerization.

90

3.3.5 NBD1-ΔN Nucleotide Binding Analysis

With the aid of resonance assignments, the effects of nucleotide binding on NBD1-ΔN could be analyzed in a site-specific manner. NMR titrations of NBD1-ΔN with MgATP (1-20 mM, Figure 3.5) result in a changes at various sites across NBD1-ΔN. The combined chemical shift change from 1 mM to 20 mM MgATP was calculated based on Eq. 2.2 and the most significant changes were plotted on the surface of a model for NBD1-ΔN (Figure 3.17). As expected, the most significant changes are localized to a single face of the protein. In the context of the full SUR protein, this localization of conformational changes is consistent with nucleotide binding. ATP binding at the NBDs triggers dimerization of the domains. Therefore several changes are expected along the dimerization interface of NBD1 to promote interaction with NBD2. Most of the residues that are predicted to be involved in nucleotide binding in NBD1-ΔN exhibit chemical shifts changes upon the ATP titration. For example, residues from the Walker A motif that is involved in γ-phosphate binding, including G704 and C705, show the greatest change upon nucleotide binding. Residues around the Q-loop (including W756, L758, N759), D- loop (including L836) and the Walker B also show significant changes upon ATP addition. We also evaluated these changes in the context of the ADP-bound state. All the residues that shift with ATP binding also shift with ADP binding, with the exception of residues in the Walker A. Only Q702 of the Walker A displayed significant chemical shift changes which is consistent with ADP-binding, since it does not possess a γ-phosphate group. The residues that have significant

Rotate180 o

Figure 3.17 Residue specific changes upon ATP binding in NBD1-ΔN. Combined chemical shift changes were calculated based on NMR spectra changes with ATP. Residues that have combined chemical shift greater than 2 standard deviations are shown in magenta, and between 1-2 standard deviations are in yellow. Light blue, white and grey regions represent regions for which assignments have been determined, not been determined and are not transferred respectively.

91

ATP ADP

Figure 3.18 Ribbon diagram of SUR2A NBD1-ΔN upon changes with ATP (left) and ADP (right). Combined chemical shift changes were calculated based on NMR spectra changes with ATP. Residues that have combined chemical shift greater than 2 standard deviations are shown in magenta, and between 1-2 standard deviations are in yellow. Light blue, white and grey regions represent regions for which assignments have determined, not determined and not transferred respectively. chemical shifts upon ATP and ADP binding are shown in Figure 3.18. For reference, the conserved and consensus sequences of nucleotide binding in NBD1-ΔN are shown in Figure 3.19. Interestingly, the magnitude of changes that are observed with ADP binding are greater than with ATP binding, especially for residues in the vicinity of the Walker B motif. This difference likely arises from the weaker affinity of NBD1-ΔN for ADP compared to ATP. Since both nucleotide titrations of NBD1-ΔN were set up with identical ratios of protein:nucleotide (1:5) at the first point, the protein starts at a lower saturation point in the case of ADP. Therefore, the difference by the final point (1:100) is more exaggerated in the case of ADP than ATP. In addition to the conserved and consensus nucleotide binding sites, there are also significant chemical shift changes in residues close to W727, V730 and N731. These residues are not predicted to interact with nucleotide based on canonical NBD crystal structures, and may represent an addition site important to SUR2A NBD1 activity. The residues are adjacent to an unassigned loop (corresponding to exon 17). This loop is significantly longer (13 residues) than the loops (~3 residues) present in other ABC transporters, based on crystal structures. A disease causing mutation V730I is also present in this region and is explored in further detail in Chapter 5 for its influence in altered nucleotide binding. In addition to changes as a result of ATP or ADP-binding, we have also examined the specific effects of ATP-analogue interactions with NBD1-ΔN (Figure 3.19). Most of these analogues affect residues in a predictable manner. For instance, AMP-PCP most significantly

92

H Loop (H867) D Loop (D839) Walker A (G701-S708) Signature Sequence A Loop (W677) (L808-Q812)

Walker B (I828-D832) Q Loop (Q753)

AMP-PCP AMP-PNP TNP-ATP

Figure 3.19 Ribbon diagram of SUR2A NBD1-ΔN with changes upon ATP-analogue binding. (Top) Consensus and conserved sites in nucleotide binding are illustrated (Bottom) Chemical shift changes upon AMP-PCP (left), AMP- PNP (center) and TNP-ATP (right) binding. Combined chemical shift changes were calculated based on NMR spectra changes with ATP. Residues that have combined chemical shift greater than 2 standard deviations are shown in magenta, and between 1-2 standard deviations are in yellow. Light blue, white and grey regions represent regions for which assignments have determined, not determined and not transferred respectively. alters residues in the Walker A, Walker B and D-loop. These are all regions in close proximity to the γ-phosphate. Studies with AMP-PNP also produce similar changes in the Walker A and Walker B. However, there are additional changes close to the A-loop of NBD1-ΔN, which interacts with the ribose sugar. Crystal structures of NBDs with AMP-PNP indicate that the coordination geometry of the terminal phosphate group to the nitrogen atom introduces a kink in

93 the molecule.32, 302 This bending of the molecule likely alters the mode of binding and the position at which AMP-PNP is held in the binding pocket. An additional analogue was tested by NMR, TNP-ATP, which is covalently modified on ribose sugar with a trinitrophenyl group. As expected, the region on the opposite side of the A-loop is most altered upon TNP-ATP binding. This is likely a result of steric interference of the trinitrophenyl group with the residues opposite to the adenine binding pocket. Changes in the Walker A were not observed, and only minor chemical shifts at residues corresponding to the Walker B and Q-loop. Notably, there is a secondary interaction of TNP-ATP with residues in the final α helix of NBD1-ΔN. This is predicted to be far in space from the ATP-binding site. This helix also shows moderate changes upon ATP and ADP binding and suggests it is important for nucleotide binding. However, crystal structures of proteins with TNP-ATP only suggest conformational changes in a hydrophobic pocket close to the ATP-binding site. 32, 302 Together, the crystal structures and our results of TNP-ATP binding suggest an allosteric effect at the C-terminal helix in SUR2A NBD1, although the origin of this effect is currently unknown. In order to further examine ATP binding with NBD1-ΔN, we have also studied the effects of Mn 2+ binding, in place of Mg 2+ . Mn 2+ is a paramagnetic species that can effectively be used as a substitute for Mg 2+ binding. 245 Residues in close proximity to the paramagnetic ion will

A B

Figure 3.20 Manganese binding to NBD1-ΔN with 5 mM ATP. (A) TROSY-HSQC of NBD1-ΔN with 0 (black), 500 (green), 625 (blue), 750 (yellow), 825 (magenta) and 1000 (red) μM Mn 2+ . The plot (B) shows the effect of Mn 2+ concentration on peak intensity for C705 of the Walker A. Residues with the sharpest decay curves are plotted as magenta coloured on a model of NBD1-ΔN (B, inset). The radius of each sphere is proportional to the observed drop in peak intensity.

94 have their NMR signals broadened significantly due to increased relaxation. Mn2+ was titrated into the NBD1-ΔN samples (containing 5 mM MgATP) and the decrease in signal intensity was recorded (Figure 3.20A). The decay in signal is related to the distance from the paramagnetic Mn 2+ center. Therefore, residues that are in close proximity will decay more rapidly. Concentrations above 1 mM Mn 2+ (with 200 μM NBD1-ΔN) resulted in complete broadening of all signals. A plot of the Mn 2+ decay curve for C705 is shown in Figure 3.20B. The Walker A Lys (K707) that is thought to bind to the metal center is not assigned in spectra at 600 MHz. However, closely positioned residues of the Walker A motif show the most significant broadening, with C705 of the Walker A displaying the greatest loss in intensity. In addition to Walker A residues, residues close to the Q-loop are also broadened, as expected considering their involvement in binding the Mg 2+ ion in the MgATP ligand.336 Additionally, residues 727 and 729 are also significantly broadened. This is unusual as these are not particularly close in space to the Mn2+ based on the model. However, relaxation data, and the location of the resonances in the HSQC, suggest that these residues are also present in a disordered loop, and therefore these residues may adopt a conformation that is close in space to the paramagnetic centre. These residues are also predicted to be altered upon ATP-binding and suggests this region may be important for nucleotide binding. Finally, D876 also shows significant broadening. This residue is predicted to be present in a β-strand far from the ATP binding site, and is unlikely involved in nucleotide binding. This broadening may be the result of an electrostatic interaction between the residue and excess Mn 2+ in the sample.

3.3.6 Lobe II of NBD1-ΔN affects protein stability

As we have seen, changes in ligands can influence the stability of NBD1-ΔN. Most predominantly, nucleotide binding (at lobe I) is responsible for increases in stability of NBD1- ΔN. Therefore, ligands or additives that alter the binding site likely affect the stability of the protein. From our stability screens, both salt and glycerol concentration alter NBD1-ΔN stability. Although they both have global effects on protein conformation, the most significant changes were localized to lobe II (away from the ATP binding site) as shown in Figure 3.21. Furthermore, lobe I is only minimally perturbed by increases in glycerol or NaCl. The fact that thermodynamic stability is altered in both cases, suggests an allosteric effect between lobe I and lobe II. This is not unreasonable as the lobe II of NBD1 interacts with lobe I NBD2 in a cooperative manner.

95

A B C

NaCl Glycerol D2O

Figure 3.21. Changes to NBD1-ΔN with (A) salt, (B) glycerol and (C) D 2O. In (A) and (B), residues that change more than 2 standard deviations are shown in magenta spheres. In (C), residues with a hydrogen/deuterium exchange rate greater on the order of minutes-hours are shown in yellow, and residues with an exchange rate on the order of hours-days are shown in magenta. For all models, light blue, white and grey regions represent regions for which assignments have determined, not determined and not transferred respectively.

Hydrogen-deuterium exchange NMR experiments provide insights into the slow time scale motions of NBD1-ΔN. Lyophilized NBD1-ΔN was re-solubilized in 100 % D 2O and several sequential HSQC experiments were recorded with scan times of approximately 40 minutes each. This disappearance of the peaks over time indicates exchange with deuterium ( 2H), which renders those nuclei invisible to 1H-15 N HSQC spectra. H/D rates are affected by multiple factors including hydrogen bonding and solvent accessibility. Residues with slow exchange rates (half-life on the order of hours/days) are shown in Figure 3.21. These sites comprise the β strand C-terminal to the Walker A motif, the α3 and α5 helices and the final β-strand. Only the Walker B site is partly protected, whereas all ATP-binding sites exchange rapidly with solvent (order of minutes to hours), indicating high solvent accessibility of these regions. The rapid exchange also suggests a fast off rate for nucleotide in the isolated protein.

3.4 Conclusions

Our strategies from specific 15 N amino acid labelling, as well as 14 N unlabelling approaches have allowed us to expand the utility of poor quality triple resonance assignment data, and determine resonance assignments for 65 % of the HN peaks in NBD1-ΔN. Our resonance assignments can be used to gain insights into nucleotide binding and mobility of the protein at the level of individual residues. Additionally, these experiments have provided us with a platform to examine

96 site specific changes in NBD1 as a result of other important regulatory mechanisms such as phosphorylation, disease-causing mutations and drug binding.

97

Chapter 4 Phosphorylation Studies of SUR2A/SUR2B NBD1 4 Overview

In this chapter we have examined the biophysical effects of phosphorylation on SUR2A NBD1. Our results suggest that phosphorylation of the N-tail in NBD1 disrupts its interactions with the core of the protein, which results in increased nucleotide binding. A secondary non-canonical PKA phosphorylation site, that was previously proposed in NBD1, has also been identified. Significant portions of the work presented in this chapter were published in the Journal of Biological Chemistry and reproduced with permission. 4 Author Contributions: E.D.A performed all of the purification and phosphorylation studies on NBD1 as well as the NMR, fluorescence, nucleotide binding and dynamic light scattering studies. C.P.A., C.R.S and M.S. repeated the NMR phosphorylation experiments using NBD1-ΔC and the N-tail. J.P.L-A carried out a subset of these NMR experiments. M.S. and E.D.A performed the molecular cloning. V.K. processed the mass spectrometry data and V.K and E.D.A analyzed the NMR data. V.K. and E.D.A wrote the manuscript and V.K., E.D.A, and C.P.A edited the manuscript.

4.1 Introduction Although nucleotide binding/hydrolysis at the NBDs is sufficient to induce channel currents, 8, 26, 180- KATP channel activity is also affected by phosphorylation of SUR and/or Kir6. x subunits. 182, 243, 253, 278, 279 In the case of vascular K ATP channels, mono-phosphorylation of murine SUR2B 243 279 at three sites (T633 and S1465 , or S1387 ) by protein kinase A (PKA) results in K ATP channel activation, with di-phosphorylation of T633 and S1465 further stimulating channel activity.243 Notably, the phosphorylation sites are conserved amongst species (T635, S1390, and T1547, respectively, in human SUR2A), highlighting their importance in regulating the activity of the SUR NBDs and their subsequent control of K ATP channel function. A molecular-level understanding of how phosphorylation activates K ATP channels is not available, but may involve

4 de Araujo, E.D. , Alvarez, C.P., Lopez-Alonso, J.P., Sooklal, C.R., Stagljar M., Kanelis, V. Phosphorylation- dependent changes in nucleotide binding, conformation and dynamics of SUR2B NBD1. Accepted to Journal of Biological Chemistry (2015). (Reprinted with permission from American Society for Biochemistry and Molecular Biology).

98

Figure 4.1.Schematic diagrams of full length SUR2A and SUR2A NBD1 (A) Schematic diagram of a SUR protein. The transmembrane helices in each membrane spanning domain (MSD0, MSD1, and MSD2) are shown in gray. The L0 linker, shown in orange, connects MSD0 to the minimum ABC transporter structure. Cytoplasmic extensions of the transmembrane helices and coupling helices (CH1, CH2, CH3, CH4) in MSD1 and MSD2 are shown in light purple. Together, these regions form the intracellular domains (ICDs). NBD1 and NBD2 are shown in blue and green, respectively. The N-terminal tail connecting MSD1 to NBD1 is shown in red, while the C-terminal tail connecting NBD1 to MSD2 is shown in yellow. Phosphorylation sites in the N-terminal tail are indicated as red circles. Scale bars on the right indicate the lengths throughout the protein as determined from homology models of SUR2B based on crystal structures of other ABC transporters. The length of the lines representing the L0 linker, N-terminal tail, and C-terminal tail were determined assuming that these linkers are completely disordered. Notice that because the ICDs are only 30 Å in length, the N-terminal tail is drawn with multiple bends in order to fit into the space allocated in the 2D representation. (B) Schematic diagram of NBD1 (residues S615-L933), NBD1-ΔC (residues S615-D914), NBD1-ΔN (residues D665-L933), NBD1-ΔNΔC (residues D665-D914), and N-tail (residues S615-E664). As in panel A, the NBD1 core is in blue, the N-terminal extension that is missing in NBD1-ΔN is in red, and the C-terminal extension that is missing in NBD1-ΔC is in yellow. (C) Schematic ribbon diagram of the homology model of SUR2A NBD1 based on the crystal structure of MRP1 NBD1 (PDB code 2CBZ 248 ). The colouring is as in panels A and B, with the exception that the different subdomains in the NBD1 core are coloured in different shades of blue. The α/β subdomain, which contains the MgATP binding site, is in dark blue, the β-sheet subdomain is in cyan, and the α-helical subdomain is in light blue. The side chain of W616 is shown in cyan, whereas the Q loop (also known as the γ-phosphate loop) is in magenta. The N-terminal tail is shown in an extended conformation because there were no restraints placed on these residues in the modeling. All structure figures were created using MOLMOL. 156

99 altered MgATP binding and hydrolysis and/or intra- and intermolecular interactions of the NBDs, as seen for CFTR.21, 33, 144, 177 Here, we present experiments detailing the effects of phosphorylation on the conformation and nucleotide binding of rat SUR2A NBD1. The sequence of rat SUR2A NBD1 is ~96 % identical to human SUR2A/SUR2B NBD1, including the phosphorylation site T632 (T635 in human SUR2B), with most amino acid changes occurring in unstructured loops.64 The studies presented here employed four different SUR2A NBD1-containing constructs (Figure 4.1B): (1) The complete SUR2A NBD1 (NBD1) consists of residues S615-L933 and is a folded and functional protein domain that possesses a predicted disordered N-terminal tail, which contains the T632 phosphorylation site, and a predicted disordered C-terminal tail;64 (2) SUR2A NBD1 from S615-D914 (NBD1-ΔC) lacks the C-terminal tail, while (3) SUR2A NBD1 from D665-L933 (NBD1-ΔN) lacks the N-terminal tail; (4) SUR2A NBD1 from D665-D914 (NBD1- ΔNΔC) lacks the N- and C-terminal tails and comprises only the canonical NBD1 fold. We also used a SUR2A construct comprised only of the N-terminal tail (S615-E664), herein referred to as N-tail. Data from NMR spectroscopy and dynamic light scattering experiments suggest that phosphorylation results in disrupted interactions between the disordered N-terminal tail and the core of NBD1. Because the N-terminal tail is ~50 residues long (Figure 4.1), it has the potential to adopt multiple conformations, including conformations that are bound to NBD1. Fluorescence studies demonstrate increased nucleotide binding affinity of phosphorylated NBD1 (phospho- NBD1) and NBD1-ΔN. These findings provide insights into the molecular basis by which phosphorylation activates K ATP channels.

4.2 Methods

4.2.1 Protein Purification and Expression

Wild type rat SUR2A NBD1 proteins comprising residues S615-L933 (NBD1), S615-D914 (NBD1-ΔC), D665-L933 (NBD1-ΔN), D665-D914 (NBD1-ΔNΔC), and S615-E664 (N-tail) were prepared as described previously.64, 66 Briefly, the SUR2A NBD1, NBD1-D C, NBD1-D N, NBD1-D ND C, and N-tail proteins were expressed as fusions with a cleavable N-terminal 6×His- SUMO tag in Escherichia coli BL21 (DE3) CodonPlus-RIL (Stratagene) cells. Uniformly 15 N- labelled SUR2A NBD1 proteins were expressed in cells grown 97.5 % 15 N-labelled M9 minimal media and 2.5 % 15 N-labelled E. coli -OD2 media (Silantes). All protein purification steps were

100 conducted at 4 °C. The 6×His-SUMO fusion proteins were isolated using a Ni 2+ -NTA affinity column (GE Healthcare) and the 6×His-SUMO tag was cleaved with 6×His-Ulp1 protease. The SUR2A NBD1, NBD1-D C, NBD1-D N, NBD1-D ND C, and N-tail proteins were next purified to homogeneity by size exclusion chromatography (Superdex 75, GE Healthcare), followed by a reverse Ni 2+ -NTA affinity column to remove small amounts of the 6×His-SUMO and 6xHis- Ulp1 proteins that co-elute with the NBD proteins from the size exclusion column. For the NMR and fluorescence studies, the SUR2A NBD1, NBD1-D C, NBD1-D N, NBD1-D ND C proteins were exchanged into NBD1 buffer (20 mM sodium phosphate pH 7.3, 150 mM NaCl, 2 % [v/v] glycerol, 2 mM DTT) with and without ATP and MgCl 2, as required. The N-tail was exchanged into a buffer identical to the NBD1 buffer, but lacked MgATP and contained 2 mM TCEP-HCl rather than DTT.

4.2.2 NMR Experiments

TROSY-HSQC spectra 232 of SUR2A NBD1, NBD1-D C, NBD1-D N, NBD1-D ND C, and N-tail were recorded at 20 °C, 25 °C, and/or 30 °C on a 600 MHz Varian Inova spectrometer equipped with a H(F)CN triple resonance cryoprobe and actively-shielded z-gradients. Chemical shifts for each spectrum were referenced to 4,4-dimethyl-4-silapentane-1-sulfonic acid (DSS).330 The different samples were exchanged into fresh NBD1 buffer with 5 mM MgCl 2 and 5 mM ATP by dialysis against the same stock of buffer to ensure the solution conditions of the samples were identical. Spectra were processed using NMRPipe/NMRDraw 75 and analyzed with NMRView.139 Chemical shift changes with phosphorylation or removal of the N-terminal tail were determined 66, 129 by calculating the combined chemical shift difference, Dd tot , from the Eq. (4.1).

2 2 0.5 Dd total = ( d HN + Dd N ) (Eq. 4.1)

Only resonances exhibiting a significant combined chemical shift difference in Hz (Dd tot ), which is greater than the average if all Dd tot values plus one standard deviation are considered. For

NMR data presented in this paper, only Dd tot ≥ 8 Hz ppm are considered. As mentioned in Chapter 3, backbone 1H, 15 N, 13 C, 13 Cα and sidechain 13 Cβ resonance assignments for NBD1-ΔN were obtained from standard triple resonance TROSY-based experiments recorded on samples of 0.5 mM NBD1-ΔN that were uniformly 15 N and 13 C-labelled and fractionally 2H labelled to ~50 %. The triple resonance assignment data were run on an 800 MHz spectrometer equipped with a cyroprobe (QANUC NMR Facility, McGill University) or on

101 a 600 MHz spectrometer equipped with an H(F)CN room temperature probe at 25 °C. These data was supplemented with 15 N-1H TROSY HSQC spectra recorded on samples that were either 15 N labelled only on specific amino acids (Gly, Leu, Ser, Val) or 14 N-labelled on specific amino acids (Ala, Arg, His, Lys, Met, Thr) and 15 N-labelled at other positions. Resonance assignments of NBD1-ΔN spectra at 25 °C were transferred to NBD1-ΔN and NBD1-ΔNΔC spectra at 30 °C. The similarity of the NBD1-ΔN spectrum to spectra of other NBD1-containing proteins allowed for straightforward transfer of resonance assignments for most residues. We have obtained assignments for 58-63 % of backbone HN resonances for NBD1-containing proteins. The combination of TROSY-based triple resonance data and specifically labelled samples has allowed for this level of resonance assignment for other NBDs. 143 Assignments of Trp indole HN resonances were obtained by site-directed mutagenesis of NBD1 or NBD1-ΔN.

4.2.3 Phosphorylation of NBD1 (S615-L933) & NBD1-ΔC (S615-D914)

Phosphorylation reactions were carried at 30 °C on purified samples of NBD1, NBD1-ΔC, and N-tail (~200 μM) in the NBD1 buffer with 15 mM MgATP rather than 5 mM MgATP. The phosphorylation reaction was initiated by the addition of 750 units (or 0.6 μM) of the catalytic subunit of protein kinase A (PKA, Promega) and was monitored by recording a series of 2D 1H- 15 N TROSY-HSQC spectra at 30 °C. The NBD1 phosphorylation reaction was allowed to proceed for 16 h, at which point the sample was almost fully phosphorylated (> 97%), as determined by the peak intensities of a resonance corresponding to the indole HN group of W616 in the non-phosphorylated and phosphorylated states. Phosphorylation of N-tail was allowed to proceed for 6 hours. Phosphorylation sites were identified by mass spectrometry following in-gel tryptic digestion (Advanced Protein Technology Centre, Hospital for Sick Children). The phosphorylated NBD1 and NBD1-ΔC were exchanged into the NBD1 buffer, which contained 5 mM MgATP, by size exclusion chromatography. The phosphorylated N-tail was exchanged into the N-tail buffer.

4.2.4 Fluorescence Nucleotide Binding Experiments

The K d value for binding of the fluorescent ATP analogue 2',3'-O-(2,4,6-trinitrophenyl)- adenosine-5'-triphosphate (TNP-ATP, Molecular Probes) to phosphorylated NBD1 and NBD1- D N was determined and compared to binding to NBD1, as done previously.66, 189 The fluorescence nucleotide binding studies were performed on a Fluoromax-4 spectrofluorimeter

102

(Horiba-Jovin, Inc.) equipped with a Peltier unit for precise temperature control. Mg 2+ and ATP were removed from the NBD1 samples using size exclusion chromatography and replaced with

2.5 m M MgCl 2 and 2.5 m M TNP-ATP. Because of the limited solubility of nucleotide-free NBD1 samples, binding experiments were conducted using 10 % (v/v) glycerol at 15 °C. Binding experiments were performed by serial dilutions of the protein from 50-70 m M, depending on the concentration eluted from the size exclusion column, to 0.8-2.0 m M while keeping the concentrations of the MgCl 2 and TNP-ATP constant at 2.5 m M each. A separate sample was generated containing MgCl 2, TNP-ATP and buffer only for the 0 mM NBD1 sample. Fluorescence emission spectra (from 485 nm – 600 nm) of TNP-ATP were recorded immediately after each sample was generated using an excitation wavelength of 465 nm, an excitation slit width of 5 nm, and an emission slit width of 7 nm. The K d value for the NBD1/nucleotide complex was determined by monitoring the ratio between the fluorescence intensity at 533 nm, which corresponds to the wavelength where the fluorescence difference of free and bound TNP- ATP is at a maximum, and 600 nm to account for any non-specific fluorescence from the protein.223 The titration data were fit to the Eq. (2.1). This equation assumes a 1:1 complex of NBD1 with TNP-ATP.110, 317

4.2.5 Thermal Stability Measurements

The thermal stabilities of non-phospho-NBD1, phospho-NBD1, and NBD1-D N were measured using intrinsic Trp fluorescence spectroscopy, as described previously.66 Briefly, thermal denaturation was monitored by following the change in the emission spectrum at 350 nm, the wavelength at which the difference in the fluorescence spectra of the folded and denatured proteins is at a maximum. The excitation wavelength was 295 nm, and the excitation and emission slit widths were 1 nm and 3 nm, respectively. The NBD1 proteins were heated from 10 to 60 °C in 1 °C increments, with a 1 min equilibration time at each temperature. The fluorescence emission at 350 nm was recorded at each temperature in a 0.5 ml cuvette. Thermal denaturation studies were done with 2 μM NBD1 in absence and presence of 2 mM MgATP.

4.2.6 Dynamic Light Scattering Studies

Dynamic light scattering experiments were performed on a Malvern Zetasizer NanoZS instrument with 100 m M samples of non-phospho-NBD1, phospho-NBD1, and NBD1-D N in the

103

NBD1 buffer containing 2 mM MgATP. Samples were centrifuged for 10 minutes at 15,000 rpm prior to each experiment. The hydrodynamic radius was determined from the average of the frequency distribution of particle sizes in three separate experiments for each sample.

4.3 Results

4.3.1 Identification of phosphorylation sites in SUR2A

Phosphorylation of SUR2A NBD1 and NBD1-D C in vitro was achieved by incubating purified proteins with the catalytic subunit of protein kinase A (PKA), the in vivo kinase for the SUR proteins.243, 279 Deletion of residues Q915-L933 from NBD1 to form NBD1-D C results in disappearance of 15 intense resonances centered about 8.2 ppm in the 1H dimension (Figure 4.2A, highlighted with a “+”), confirming previous predictions that the C-terminal tail is disordered.64 Removal of the C-terminal tail thereby results in reduced resonance overlap in the

Figure 4.2 Identification of resonances from the NBD1 core, the N-terminal tail, and the C-terminal tail. Comparison of 2D 15 N-1H-TROSY HSQC spectra of (A) NBD1 (250 μM) and NBD1-ΔC (220 μM) with 5 mM Mg 2+ and 5 mM ATP in 20 mM Na + phosphate, pH 7.3, 150 mM NaCl, 2 mM DTT, 2 % (v/v) glycerol, 10 % (v/v) D 2O at 30 °C at 600 MHz. The spectrum of nonphospho-NBD1 is in the background with resonances of backbone nuclei, as well as those from side chain nuclei from Trp, Asn, and Gln residues, shown in black. The blue resonances are of opposite sign, caused by spectral aliasing, and are possibly from Arg NεHε side chain correlations. The spectrum of NBD1-ΔC is in the foreground. The cyan and magenta peaks in the NBD1-ΔC spectrum correspond to the black and blue resonances, respectively, in the NBD1 spectrum. (B) Comparison of 2D 15 N-1H-TROSY HSQC spectra of (A) NBD1 and N-tail. The NBD1 spectrum is coloured as in panel A. The N-tail spectrum is coloured in orange. The solution conditions of N-tail are identical to that of NBD1, except that the N-tail buffer lacks MgATP and contains 2 mM TCEP-HCl in place of DTT.

104

NMR spectrum, making it possible to see additional chemical shift changes upon phosphorylation with NBD1-D C compared to NBD1. Although disordered, the C-terminal tail residues Q915-L933 significantly enhance the solubility of NBD1.64 Further, residues outside the canonical NBD fold are often involved in regulation of the NBD function.55, 148, 177

NMR spectra of isolated N-tail indicate that, as predicted, residues S615-E664 are disordered (Figure 4.2B). The N-tail used here is part of the intracellular linker (Q600-E664) connecting transmembrane helix 11 to NBD1. Transmembrane helix predictions, sequence alignments, and homology models 189 of SUR2A based on structures of other ABC proteins indicate that helix 11 ends at V599, residues Q600-L607 extend transmembrane helix 11 into the cytoplasm, and that NBD1 begins at D665. Further, the overlay of resonances in the N-tail spectrum with resonances in the NBD1 spectrum from S615-E664 (Figure 4.2 and see below) indicates that the presence of the folded NBD1 core does not induce folding of the N-tail. Notably, residues L607-E664 in SUR2A are primarily predicted as being disordered by PONDR analysis.64 We have also recorded spectra that demonstrate that a longer N-tail protein (S608- E664) is also disordered. Additionally, we have recorded spectra on longer N-tail proteins (eg. Q600-E664). Resonances in spectra of the N-tail from S615-E664 overlay perfectly with resonances in the longer N-tail proteins, indicating that extension of the N-tail does not induce folding of these residues. However, NBD1-containing proteins starting at residues Q600 or S608 aggregate at the high concentrations required for NMR studies, 64 as is often seen when unstructured residues are included in constructs that contain folded domains. Thus, for comparison with NBD1 proteins, we are using an N-tail protein beginning at S615.

Phosphorylation of NBD1, NBD1-D C, and N-tail is monitored by changes in the chemical shift of the indole HN resonance of W616 (Figure 4.3, W616 indole vs. W616 indole - phospho). The identity of the W616 indole HN resonance in NBD1 and NBD1-D C is known from 2D 15 N-1H NMR spectra of a deletion mutant of NBD1 in which residues S615-R617 were removed.64 The W616 indole HN resonance in N-tail is assigned based on resonance overlap between NBD1 and N-tail resonances and the fact that N-tail has only one Trp residue. The combined chemical shift change ( Dd tot of 37 Hz) for the W616 indole HN resonance and its location in a sparsely populated region of the spectrum allows for monitoring of the phosphorylation reaction. Chemical shift changes in the W616 indole HN resonance indicates

105

Figure 4.3. Time-course of PKA phosphorylation of SUR2A NBD1. Only the region of the 2D 15 N-1H-TROSY HSQC showing the Trp indole resonances is displayed. The most left-hand panel displays (0 h) the spectrum before addition of PKA. Spectra were recorded every two hours up to 16 hours. An additional spectrum was also recorded after two weeks. Selected spectra from the reaction are shown. The Trp indole resonance from residue W616 is used to monitor phosphorylation. The W616 Trp indole resonance is labelled W616 indole in the non-phosphorylated state, and is labelled as W616 indole-phospho in the phosphorylated state. that it takes ~16 hours to completely phosphorylate an NMR sample containing 200 m M NBD1 at 30 °C (Figure 4.3) and ~6 hours to completely phosphorylate 200 m M N-tail at 30 °C. Samples of NBD1-D C were also phosphorylated using the same protocol, but to a lower extent. According to peak intensities of the W616 indole HN resonance from the non-phosphorylated and phosphorylated states, NBD1-D C is ~85 % phosphorylated. We could not achieve uniform phosphorylation of NBD1-D C with higher temperatures due to compromised sample stability. Nonetheless, chemical shift changes observed with phosphorylation of NBD1 are also seen with phosphorylation of NBD1-D C. Thus, information gleaned from phosphorylation of NBD1-D C can be applied to NBD1.

The identity of the phosphorylation sites was determined using mass spectrometry following in-gel trypsin digestion (Figure 4.4A). Previous studies determined that NBD1 is phosphorylated at the PKA consensus site T632, as well as at another unidentified site, as a peptide comprising residues C628-I639 with a T632A mutation can still be phosphorylated by PKA.243 Our mass spectrometry data comparing non-phosphorylated and phosphorylated forms of NBD1, NBD1-D C, and N-tail confirm phosphorylation of T632 and identified the second site as S636. Residues T632 and S636 are located in the N-tail of NBD1 (Figure 4.1). No phosphorylation sites were detected in the core of NBD1. These data are consistent with control NMR phosphorylation experiments that show no chemical shift changes in the spectra of NBD1- D N and NBD1-D ND C proteins, which lack T632 and S636 (data not shown).

106

Figure 4.4 Mass spectrometry data identify the T632 and S636 phosphorylation sites. (A) Mass spectrometry data on the non-phospho- and phospho- NBD1. Fragmentation tables for the T632- and S636-containing peptides in non-phospho NBD1 are shown in the left panel. Fragmentation tables for the phosphorylated peptides, the pT632- and pS636-containing peptides, in phospho NBD1 are shown in the right panel. Values highlighted in pink correspond to singly-charged B and doubly-charged B ions, while values highlighted in blue correspond to singly-charged Y and doubly charged Y ions. The values in green correspond to B or Y ions with a loss of ammonia or water. Values corresponding to predicted fragments that are not observed are not highlighted. Fragmentation tables were created with the software Scaffold (Proteome Software). (B) Phosphorylation sites in NBD1. The canonical PKA phosphorylation consensus sequence, which contains T632, is bolded in red, and the non-canonical phosphorylation site S636 is shown in purple. Note that only a subset of the NBD1 sequence is shown.

The phosphorylation site T632 conforms to the PKA consensus site (R/K-R/K-x- pS/pT),231, 256 while S636 comprises a non-consensus PKA phosphorylation site (Figure 4.4B). Phosphorylation of non-canonical sites by PKA has been demonstrated to modulate the function of other proteins. 108, 118, 305 Further, work with model peptides indicates that PKA can phosphorylate a Ser or Thr residue even if the dibasic motif is located up to 8 residues N- terminal of the phosphorylation site, albeit with varying efficiency.319 Thus, the proximity of the two phosphorylation sites in SUR2A NBD1 may allow for the dibasic residues comprising the consensus site at T632 to serve in PKA recognition of the non-consensus site S636.

107

4.3.2 Phosphorylation dependent spectral changes in SUR2A NBD1

In addition to the W616 indole HN group, phosphorylation-dependent chemical shift changes are observed for two additional Trp indole HN and 40 backbone HN resonances in NBD1 (Figure 4.5A) and for one additional Trp indole resonance and 48 backbone resonances in NBD1-D C (Figure 4.6A). Chemical shift changes are expected for residues close in space to the phosphorylation sites. Phosphorylation of T632 and S636 introduces two negative charges each into otherwise polar, but uncharged side chains, and thus significantly changes the chemical environment for residues close to T632 and S636. The cyan circles in Figures 4.5A and 4.6A highlight phosphorylation-dependent chemical shift changes common to NBD1 and NBD1-D C, while the magenta squares highlight phosphorylation-dependent chemical shift changes that are specific to each protein.

Analysis of the NMR spectra indicates that phosphorylation of the N-terminal tail affects disordered and structured residues in NBD1 and NBD1-D C. Of the backbone HN resonances that exhibit chemical shift changes, 6 in NBD1 and 10 in NBD1-D C have intense signals and are centered about 8.2 ppm in the 1H dimension (Figure 4.5A and 4.6A, respectively), suggesting they are from disordered residues. Comparison of spectra of NBD1, NBD1-D N, NBD1-D C, NBD1-D ND C, and N-tail indicates that these intense HN resonances are from the residues in the N-terminal tail (Figure 4.2, highlighted with a “ * ” sign ), as is expected considering the location of the phosphorylation sites, and not the C-terminal tail. Further, the quality of the NBD1-ΔNΔC spectrum indicates that the structural boundaries of the NBD1 core in SUR2A are from residues D665-D914 (Figure 4.6B). Some of the disordered resonances that exhibit chemical shift changes are only seen with phosphorylation of the NBD1-ΔC (Figure 4.6A, magenta squares). However, these resonances may also be present in spectra of NBD1 and phospho-NBD1 but may be overlapped by the additional disordered residues from the C-terminal tail in NBD1. We can not assign the remaining disordered resonances that change with phosphorylation to the N- terminal tail due to resonance overlap in this region of the spectrum. The chemical shift change of the indole HN from W616, which is also located in the disordered N-terminal tail (Figure 4.1C), demonstrates that phosphorylation affects residues far from T632 and S636 in the primary sequence. Chemical shift changes of the W616 indole would not be expected if the N-terminal

108

Figure 4.5 Spectral changes in NBD1 with phosphorylation and removal of the N-terminal tail. (A, i) Comparison of 2D 15 N-1H-TROSY HSQC spectra of non-phospho-NBD1 (250 μM) and phospho-NBD1 (200 μM). The solution conditions for each sample are identical to that described in the legend to Figure 4.2. The spectrum of non-phospho-NBD1 is in the foreground and is coloured as in Figure 4.2. The spectrum of phospho-NBD1 is shown in the background. The red and green peaks in the phospho-NBD1 spectrum correspond to the black and blue resonances, respectively, in the non-phospho-NBD1 spectrum (continued on next page).

109

Figure 4.5 (continued) (A, ii) Selected region of the overlaid spectra shown in panel A. Resonances highlighted by cyan circles highlight phosphorylation-dependent chemical shift changes that also occur in NBD1-ΔC. Magenta squares represent chemical shift changes seen with phosphorylation of NBD1 only. Assignment of selected resonances that exhibit phosphorylation-dependent chemical shift changes is shown. (A, iii) Chemical shift mapping of phosphorylation-dependent changes to specific residues in NBD1. The ribbon diagram is coloured cyan for residues with resonance assignments, white for residues with no resonance assignments, and grey for residues which have assignments in NBD1-ΔN but not phospho-NBD1. The location of Pro residues is also indicated in grey, as these residues do not give signals in the NH-based NMR experiments, such as the 2D 15 N-1H-TROSY HSQC. Cα atoms of residues that exhibit phosphorylation-dependent chemical shift changes are shown as spheres coloured from light pink, to highlight the smallest changes (Δδ tot = 8-10 Hz), to magenta for the largest changes (Δδ tot ≥ 25 Hz). The Cα atoms of T632 and S636 are coloured yellow. (B, i) Comparison of 2D 15 N-1H-TROSY HSQC spectra of non-phospho- NBD1 and NBD1-ΔN (250 μM). The spectrum of non-phospho- NBD1 is shown in the background in black and blue, as in panel A. The spectrum of NBD1-ΔN is in the foreground. The magenta and cyan peaks in the NBD1-ΔN spectrum correspond to the black and blue resonances, respectively, in the non-phospho-NBD1 spectrum. (B, ii) Selected region of the overlaid spectra shown in panel A. Green circles highlight resonances that also display chemical shift changes with phosphorylation. Assignment of selected resonances that exhibit chemical shift changes with deletion of the N-terminal tail only are shown. (B, iii) Chemical shift mapping of N-terminal tail deletion to specific residues in NBD1. The ribbon diagram is coloured as in panel A, except that residues coloured in gray represent residues that have been assigned in spectra of NBD1-ΔN recorded at 800 MHz at 25 °C that can not be transferred to NBD1-ΔN spectra recorded at 600 MHz at 30 °C. Selected residues that exhibit changes with removal of the N-terminal tail, only, are highlighted. tail is completely disordered and residues W616, T632 and S636 were independent of one another.

In keeping with the long-range effects seen with the W616 indole HN resonance change, phosphorylation of the disordered N-terminal tail also affects multiple structured residues in NBD1 and the NBD1-ΔC. Phosphorylation-dependent chemical shift changes in NBD1 are observed for residues throughout the NBD core (Figure 4.5A and 4.6A). These include residues in the β-sheet subdomain (I691, L721, Y726, W727), the α-helical subdomain (T767-G769, R776, Y777, C784, G811), and 27 residues in the α/β subdomain, which contains the MgATP binding site. Notably, many of the α/β and α-helical subdomain residues that exhibit chemical shift changes cluster together at or near the interface between the α/β and α helical subdomains (V748-A751, Q824-T826, F830, L831). Residues in the Q loop (Q753, W756-N759) also display phosphorylation dependent chemical shift changes. MgATP binding results in conformational changes in the Q-loop and, along with NBD1/NBD2 dimerization, also alters the relative orientation of α-helical subdomain with the rest of the protein. Notably, chemical shift changes are also observed for several residues that are located at the predicted NBD1/NBD2 dimerization interface (Q753, W756, G811, H841, E904, H905). Thus, the phosphorylation dependent changes of the functional NBD1 residues may be coupled to altered NBD1 activity.

110

Figure 4.6 Spectral changes in NBD1-ΔC with phosphorylation and removal of the N-terminal tail. (A) 2D 15 N- 1H-TROSY spectra showing phosphorylation-dependent changes in NBD1-ΔC. (i) The spectrum of non- phosphorylated NBD1-ΔC (250 μM) is in the foreground while the spectrum of phosphorylated NBD1-ΔC (80 μM) is in the background, with colouring as described for Figure 4.5A. The solution conditions for each sample are identical to those described in Figure 4.5. (continued on next page).

111

Figure 4.6 (continued) (ii) Selected region of the spectrum shown in panel A. Cyan circles highlight resonances that also display chemical shift changes with phosphorylation of NBD1. Magenta squares highlight resonances that display phosphorylation-dependent chemical shift changes in NBD1-ΔC only. Phosphorylation-dependent chemical shift changes are seen for backbone resonances (black peaks) and resonances from NεHε groups in Arg side chains (green peaks). (A, iii) Chemical shift mapping of phosphorylation-dependent changes to specific residues in NBD1-ΔC. The ribbon diagram is coloured with the resonance assignments for phospho-NBD1-ΔC an d chemical shift changes. The Cα atoms for selected residues that show phosphorylation-dependent chemical shift changes in NBD1-ΔC, but not NBD1, are highlighted. (B, i) Comparison of 2D 15 N-1H-TROSY HSQC spectra of non-phospho-NBD1-ΔC and NBD1-ΔNΔC (320 μM) The spectrum of non-phospho-NBD1-ΔC is shown in the background in black and blue, as in panel A. The spectrum of NBD1-ΔNΔC is in the foreground. The magenta and cyan peaks in the NBD1-ΔNΔC spectrum correspond to the black and blue resonances, respectively, in the non-phospho-NBD1-ΔC spectrum. (B, ii) Selected region of the overlaid spectra shown in panel A. Green circles highlight resonances that also display chemical shift changes with phosphorylation. (B, iii) Chemical shift mapping of N-terminal tail deletion to specific residues in NBD1-ΔC. Residues are coloured based on available resonance assignments for NBD1-ΔNΔC and chemical shift changes. The Cα atoms for selected residues that exhibit chemical shift changes with deletion of the N-terminal tail in NBD1-ΔC, but not NBD1, are highlighted.

Similar changes are also seen upon phosphorylation of the NBD1-ΔC. However, the decreased overlap afforded by the non-phospho and phospho- NBD1-ΔC spectra allow identification of additional residues that exhibit phosphorylation-dependent changes. These include N729 and V730, which are located in a loop connecting the β-sheet and α/β subdomains, and K907 located at the NBD1/NBD2 dimer interface. Phosphorylation also causes chemical shift changes in the indole HN resonances of W727 and W756 in NBD1 (Figure 4.5A) and of W756 and W906 in the NBD1-ΔC (Figure 4.6).

Analysis of the homology model of SUR2A NBD1 indicates that phosphorylation affects residues that are up to ~40 Å away from the end ( i.e. residue E664) of the N-terminal tail (Figure 4.5A, 4.6A). One hypothesis for the long-range chemical shift changes is that the N- terminal tail possesses residual structure and interacts with multiple regions in the core of NBD1, and that phosphorylation alters the structure of the N-terminal tail and/or its interactions with the NBD1 core. The interaction of disordered proteins or disordered residues of a folded protein with multiple binding sites has been observed for many intrinsically disordered proteins.

Chemical shift mapping of phosphorylation dependent changes in NBD1 and the NBD1- ΔC is possible because most backbone and Trp indole HN chemical shift assignments of the NBD1-ΔN can be transferred to spectra of phospho-NBD1 and phospho-NBD1-ΔC. The remaining backbone resonances (2 in NBD1 and 11 in NBD1-ΔC) that display chemical shift changes are not mapped to specific residues either because they have not been assigned in

112 spectra of NBD1-ΔN or because assignments could not be transferred from spectra of NBD1-ΔN to spectra of phospho-NBD1 and/or phospho-NBD1-ΔC.

4.3.3 Removal of the N-terminal region mimics phosphorylation of NBD1

In order to determine how phosphorylation alters the conformation of NBD1, we compared spectra of NBD1, phospho-NBD1 and NBD1-ΔN (Figure 4.5), as well as NBD1-ΔC, phospho- NBD1-ΔC and NBD1-ΔNΔC (Figure 4.6). Almost every residue in NBD1 and NBD1-ΔC that displays phosphorylation-dependent chemical shift changes, also changes in chemical shift upon removal of the N-terminal tail. Chemical shift changes common to phosphorylation and N- terminal tail removal are highlighted by green circles in spectra in Figures 4.5B and 4.6B, respectively, and also mapped onto the NBD1 model. Additional chemical shift changes in NBD1 and NBD1-ΔC with removal of the N-terminal tail (eg. A667-T671, D688-R690 resonances in Figure 4.5B) arise from changing the N-terminal residue in the protein from S615 to D665. This change brings the charged N terminus closer to the NBD1 core, and as expected, causes multiple chemical shift changes for residues in the β-sheet subdomain. In some cases, chemical shift changes observed with removal of the N-terminal tail were also seen with phosphorylation, but the Dd tot value calculated for these residues was not statistically significant. Additional differences observed between phosphorylation and removal of the N-terminal may result from overlap in spectra of NBD1 and phospho-NBD1 versus NBD1 and NBD1-ΔN (and their NBD1-ΔC counterparts), which limits our ability to detect those chemical shift changes. For example, more chemical shift changes are seen with removal of the N-terminal tail compared with phosphorylation for residues in the loop comprised by residues N729-S747.

Notably, a few resonances show linear chemical shift changes 241 when comparing non- phospho-NBD1, phospho-NBD1, and NBD1-ΔN (Figure 4.7). Such linear chemical shift changes reflect fast exchange on the NMR timescale between two distinct states, one state in which the N-terminal tail is bound to the NBD1 core, as in non-phospho-NBD1, and a second state in which the N-terminal tail fully displaced from the NBD1 core, as in NBD1-ΔN. The intermediate position of phospho-NBD1 resonances indicates that phosphorylation only partly displaces the N-terminal tail from the NBD1 core.

113

Figure 4.7 Phosphorylation alters the equilibrium of N-terminal tail interactions. Selected regions of the spectra of non-phospho-NBD1 (black), phospho-NBD1 (red), and NBD1-ΔN (cyan) are overlaid. Arrows highlight the direction of the linear chemical shift changes observed upon phosphorylation of NBD1 and removal of the N- terminal tail. Further differences between non-phospho-NBD1 versus phospho-NBD1 and NBD1-D N are observed when comparing spectra of the proteins at different temperatures (Figure 4.8). While spectra of non-phospho-NBD1 show evidence of broadening in backbone and Trp side chain resonances as the temperature is decreased from 30 °C to 20 °C, spectra of phospho-NBD1 and NBD1-D N change very little with temperature. A similar trend is seen for non-phospho- NBD1-D C, phospho-NBD1-D C, and NBD1-D ND C. For clarity, only a subset of resonances affected by temperature in spectra of non-phospho-NBD1 are highlighted. For example, a number of backbone and Trp indole HN resonances that are observable in spectra of non- phospho-NBD1 at 30 °C (Figure 4.8, solid cyan circles) are missing in spectra of non-phospho- NBD1 at 25 °C and/or 20 °C (Figure 4.8, dashed cyan circles highlight the position of these missing resonances). Notably, some of the non-phospho-NBD1 resonances that completely disappear at 25 °C and/or 20 °C have strong intensities at 30 °C. In contrast, most of these resonances are observable in spectra of phospho-NBD1 at lower temperatures and all are observable in NBD1-D N in spectra recorded at 25 °C and 20 °C. Further, phosphorylation of NBD1 and removal of the N-terminal tail from NBD1 results in the appearance of additional peaks (Figure 4.8, magenta circles). These resonances are likely also present in spectra of non- phospho-NBD1, but their low intensities do not allow them to be unambiguously distinguished from background noise.

114

Figure 4.8 Temperature-dependent changes in spectra of non-phospho-NBD1, phospho-NBD1, and NBD1-ΔN indicate that displacement of the N-terminal tail reduces broadening in the NBD1 core. 2D 15 N-1H-TROSY HSQC spectra for each sample were recorded at temperatures of 30 °C (top panels), 25 °C (middle panels), and 20 °C (bottom panels). Backbone HN resonances, as well as side chain HN resonances from Trp, Asn, and Gln residues are in black. Resonances from Arg NεHε groups are in blue. Solid cyan circles in all panels highlight backbone and Trp indole HN resonances that are seen in spectra of non-phospho-NBD1 at 30 °C, but which experience broadening in spectra of the non-phospho-NBD1 at 25 °C and 20 °C. Dashed cyan circles indicate the positions of the resonances that are broadened in spectra of non-phospho-NBD1 at the lower temperatures. A grey dotted box in the spectrum of non-phospho-NBD1 at 30 °C (upper left) highlights the Trp indole HN resonances. Solid magenta circles highlight backbone resonances that appear in spectra of phospho-NBD1 and NBD1-ΔN. The dashed magenta circles highlight the positions of these resonances in spectra of phospho-NBD1 at 20 °C.

115

The temperature-induced broadening of non-phospho-NBD1 is reversible and we have already determined that NMR samples of non-phospho-NBD1 at 250 μM concentration are monomeric.189 Thus, we do not attribute the broadening to NBD1 dimerization or aggregation. Because broadening in NMR spectra can arise from μs-ms timescale protein motions, these temperature-dependent spectral changes imply differences in NBD1 dynamics, possibly caused by differential interactions of the N-terminal tail with the NBD1 core in the different proteins. In fact, many of the nonphospho-NBD1 resonances that experience temperature-induced broadening also exhibit chemical shift changes with phosphorylation and removal of the N- terminal tail. These include residues in the β-sheet subdomain (I691, Y726, W727), the ATP binding α/β subdomain (G694, L696, I714, A751, W756, L758, F830, L831, K859, D876) and α-helical subdomain (T767, G769, Y777, and C784). Interactions of the N-terminal tail in nonphospho-NBD1 result in multiple NBD1 core residues experiencing μs-ms motions, leading to overall broadening of the spectra. These μs-ms motions are quenched by the limited interactions of the phosphorylated N-terminal tail and NBD1 core, resulting in reduced broadening in spectra of phospho-NBD1. The absence of N-terminal tail binding to the NBD1 core in NBD1-ΔN eliminates broadening in spectra of NBD1-ΔN.

Disruption of N-terminal tail interactions with the NBD1 core by phosphorylation of T632 and S636 would lead to a less compact protein. In order to test this model, we measured the

Figure 4.9 Phosphorylation increases the size of NBD1. Dynamic light scattering profiles of nonphospho-NBD1 (light grey dotted line), phospho-NBD1 (grey dashed line) and NBD1-ΔN (solid black line) are shown. The frequency distribution of particles at the given size (hydrodynamic radius) is shown for each protein. The fit of the profiles gives a hydrodynamic radius of 3.07 ± 0.06 nm for non-phospho-NBD1, 3.66 ± 0.07 nm for phospho-NBD1, and 2.35 ± 0.07 nm for NBD1-ΔN. Values are reported as averages of three measurements ± one standard deviation.

116 hydrodynamic radius of non-phospho- and phospho-NBD1 using dynamic light scattering (Figure 4.9). Phosphorylation of T632 and S636 increased the hydrodynamic radius of NBD1 by ~20 % from 3.07 ± 0.06 nm to 3.66 ± 0.07 nm, consistent with our hypothesis. For comparison, NBD1-D N has a hydrodynamic radius of 2.35 ± 0.07 nm. The hydrodynamic radius of NBD1- D N (30 kDa) is slightly smaller than the hydrodynamic radius of ovalbumin (45 kDa) of 2.81 ± 0.01 nm. Together, the NMR and the light scattering data suggest that phosphorylation of T632 and S636 disrupts interactions of the N-terminal segment with the core of NBD1.

4.3.4 Phosphorylation increases the nucleotide binding affinity of NBD1

PKA phosphorylation of SUR2A at NBD1 and NBD2 results in increased K ATP channel 243, 279 activity. Because K ATP channel gating is also activated by MgATP binding and hydrolysis at the NBDs, we sought to determine whether phosphorylation alters the nucleotide binding properties of NBD1. We used a fluorescently labelled ATP analogue, TNP-ATP, to probe binding of nucleotide to NBD1. 66, 189 TNP-ATP exhibits very low fluorescence in absence of NBD1. Addition of increasing amounts of non-phosphorylated NBD1, phosphorylated NBD1, and NBD1-ΔN resulted in a concentration-dependent and saturable increase in TNP-ATP fluorescence (Figure 4.10A). Fitting of the binding curves reveals a K d value of 8.4 ± 1.1 μM for the interaction of non-phospho-NBD1 with TNP-ATP, as measured previously. 66, 189 In comparison, there is a modest increase in the nucleotide binding affinity for phospho-NBD1 (K d

6.3 ± 0.2 μM) and a 2-fold increase in the nucleotide binding affinity of NBD1-ΔN (K d 4.4 ± 0.7 μM). Mg 2+ was required for these experiments, as no nucleotide binding could be detected without Mg 2+ , as assessed by TNP-ATP fluorescence binding experiments and NMR titrations with ATP. Note that NBD1 samples for these Mg 2+ -free studies were dialyzed against a buffer lacking Mg 2+ and containing Chelex-100 resin to remove any residual metals. The intermediate increase in nucleotide binding affinity of phospho-NBD1 is consistent with a model in which phosphorylation of T632 and S636 disrupt interactions of the N-terminal region with the NBD1 core to partly expose the ATP binding site, with removal of these residues in NBD1-ΔN fully exposing the ATP binding site. The increased nucleotide binding observed for the isolated and monomeric phospho-NBD1 and NBD1 in which the N-terminal tail is completely displaced from binding the protein core (NBD1-ΔN) likely has a greater effect in the intact channel, which contains four SUR2A proteins. Increased MgATP binding at NBD1 will

117

A B

Figure 4.10 Nucleotide binding to non-phospho-NBD1, phospho-NBD1, and NBD1-ΔN. (A) The increasing fluorescence of TNP-ATP is shown as solid circles in black, blue, or red for nonphospho- NBD1, phospho-NBD1, and NBD1-ΔN, respectively. Solid lines displays the fit of the TNP-ATP titration data, assuming a 1:1 complex for the NBD1/nucleotide interaction, as described in Experimental Procedures. The K d values for binding to TNP-ATP are 8.4 ± 1.1 μM for non-phospho-NBD1, 6.3 ± 0.2 μM for phospho-NBD1, and 4.4 ± 0.7 μM for NBD1-ΔN. (B) MgATP binding increases in the thermodynamic stability of non-phospho-NBD1 (black), phospho-NBD1 (blue), and NBD1-ΔN (red). Temperature denaturation studies were performed on 2 μM protein samples in a 1 cm cuvette. Data from temperature denaturation of the NBD proteins in absence and presence of 2 mM MgATP are shown as open circles and filled circles, respectively. In absence of MgATP, the T m value was similar for all proteins (nonphospho-NBD1, 27.7 ± 1.2 °C; phospho-NBD1, 29.7 ± 1.5 °C; NBD1-ΔN, 30.0 ± 1.0 °C). In the presence of 2 mM MgATP, the T m values are 42.3 ± 0.6 °C for non-phospho-NBD1, 44.7 ± 0.6 °C for phospho-NBD1, and 48.3 ± 0.6 °C for NBD1-ΔN. The K d values and T m values are reported as averages of three measurements ± one standard deviation. likely also increase binding of NBD1 and NBD2, which ultimately increases K ATP channel gating. We used intrinsic Trp fluorescence to monitor thermal unfolding of the NBD1 proteins (Figure 4.10B). In the absence of MgATP, phospho-NBD1 and NBD1-D N have melting profiles 66 and melting temperatures (T m ~ 30 °C) that are similar to that previously seen for NBD1. However, in the presence of MgATP, the proteins have different thermal stabilities. MgATP leads to a substantial stabilization of all NBD1 samples at all nucleotide concentrations tested, with NBD1-D N exhibiting the greatest increase in T m values (T m = 48 °C), followed by phospho-

NBD1 (T m = 45 °C) and then non-phospho-NBD1 (T m = 42 °C). This trend in T m values for the different samples is likely a reflection of the difference in nucleotide binding affinity for the different proteins. ATPase activity of the different NBD1 proteins was measured as done previously, and was found to be very low (< 4.0 nmol Pi/ mg protein/ min), likely because NBD2 was not present.

118

4.4 Discussion

Our NMR studies demonstrate that phosphorylation of T632 and S636 affects residues in the disordered N-terminal tail and structured NBD1 core. Residue T632 corresponds to a canonical PKA phosphorylation site that is conserved amongst species, while residue S636 is a variable non-canonical PKA phosphorylation site. Residue S636 in rat and murine SUR2A corresponds to a Pro residue (P639) in human SUR2A, which would promote alternate conformations of the N- terminal tail.73, 192 Further, P639 is part of exon 14 (residues Q638-K673) that is missing in a naturally-occurring splice isoform, known as human SUR2C,49 which further illustrates the regulatory nature of the N-terminal tail. Our NMR data showing that removal of the N-terminal tail mimics phosphorylation of NBD1 suggest that phosphorylation of T632 and S636 disrupts some of the transient interactions of the N-terminal tail with NBD1, with complete disruption achieved by removal of the N-terminal tail in NBD1-D N (Figure 4.11). Chemical shift mapping indicates that these transient interactions occur with residues that exhibit conformational changes with MgATP binding, such as residues in the Q loop and at the interface between the α/β and α- helical subdomains. The more “open” state of NBD1 achieved by phosphorylation and N- terminal tail deletion would result in increased exposure of the MgATP binding site likely, thereby explaining the increased nucleotide binding affinity of phospho-NBD1 and NBD1-D N. Chemical shift mapping also indicates that transient interactions occur with NBD1 residues that bind NBD2. Phosphorylation promotes stabilization of NBD1 into a conformation that allows it to make productive interactions, such as with MgATP and/or NBD2, leading to increased K ATP channel opening. Thus, the model (Figure 4.11) is consistent with the demonstrated 222 243, 278 phosphorylation-mediated activation of K ATP channels.

Because SUR2A NBD2 is also phosphorylated, additional studies using NBD2 are necessary. Further, phosphorylation of the NBDs may affect their interactions with other regions of SUR2A, such as the coupling helices, and/or the Kir6.2 channel pore. Therefore, a full understanding of the molecular basis of the mechanism by multisite phosphorylation regulates

KATP channels will require additional studies to address these questions. Notably, 26, 180 phosphorylation of SUR1 NBD2 also regulates K ATP channel gating, suggesting that SUR subunit phosphorylation is a common regulatory mechanism to control K ATP channel gating.

119

Figure 4.11 Structural model for effects of phosphorylation on the N-terminal tail with the NBD1 core. The NBD1 core is shown as a blue sphere, while the N-terminal tail that contains T632 and S636 is shown as a red line with phosphorylation indicated by open red circles labelled with a P as applicable. Different interactions of the N-terminal tail with NBD1 are displayed schematically on a gradient from a “closed state”, in which the N- terminal tail in bound to the NBD1 core, to an “open state” where the N-terminal tail is removed and can not bind the NBD1 core. Phosphorylation of the N-terminal tail results in an intermediate state in which the N-terminal tail makes limited interactions with the NBD core.

Phosphorylation has been observed for many ABC transporters across all subfamilies, which implies that phosphorylation is a general mechanism for regulating ABC transporter function. ABC transporters are phosphorylated at specific sites in the NBDs and/or in the NBD1- MSD2 linker by various kinases, with the effects of phosphorylation depending on the specific site phosphorylated.291 For example, phosphorylation of ABCA1 at NBD2 is required for full cholesterol transport activity, whereas phosphorylation of the NBD1-MSD2 linker affects protein stability.255, 269, 291 ABCB1, which is also known as P-glycoprotein and is associated with multidrug resistance in cancers, 276 is also phosphorylated at a number of sites in the NBD1- MSD2 linker by PKA, protein kinase C, and casein kinase 2 (reviewed in 291 ). PKC phosphorylation increases the ATPase activity of ABCB1 and regulates efflux of anions.

The best-characterized ABC protein in terms of phosphorylation is the cystic fibrosis transmembrane conductance regulator (CFTR). Like the SUR proteins, CFTR is a member of the C-subfamily of ABC transporters.72 CFTR is phosphorylated by PKA and PKC at multiple sites in the disordered R region, which links NBD1 to MSD2, and by PKA at one site in the regulatory insert (RI) located within NBD1.48, 175, 234, 296 As with SUR2A, phosphorylation of CFTR disrupts transient interactions of the RI and R region with the NBD1 core 21, 144 to promote a more open CFTR NBD1 conformation,114, 144, 338 which then permits binding of NBD1 with coupling helix 1.144 Further, as with SUR2A, phosphorylation of CFTR leads to increased MgATP binding and hydrolysis, and subsequent channel activation.55, 177

120

Studies of CFTR highlight additional functional consequences of phosphorylation, which may also be relevant for SUR2A. There are differences in the conformation of phosphorylated CFTR with the severe CF-causing mutation D F508,3, 144 that partly explain the molecular defects of mutant CFTR compared to the wild type protein.3, 77, 267, 322 Further, deletion of the RI, which mimics phosphorylation,144 increases the response of mutant CFTR to small molecule correctors.3 Phosphorylation also disrupts interactions of the R region with NBD2, but enhances interactions of the R region with the C terminus of CFTR and the accessory proteins STAS and 14-3-3.33 As observed for CFTR, phosphorylated SUR NBDs bearing disease-causing mutations may also possess a different conformation from phosphorylated wild type NBDs. In addition, phosphorylation may affect interactions of the SUR2A NBDs with other proteins, such as the Kir6.2 pore, or with therapeutics, such as pinacidil which binds NBD1.189 Thus, data presented here on the conformation of wild type SUR2A NBD1 provide a platform to address these additional questions.

121

Chapter 5 The Disease-Causing Mutation, V734I, Alters the Conformation and Nucleotide Binding of SUR2A NBD1 5 Overview

In this chapter we have examined the molecular effects of a genetic mutation, V734I, in SUR2A NBD1. We have employed fluorescence and NMR spectroscopy to examine conformational changes in NBD1 that result from the disease-causing mutation. These studies have suggested that the V734I substitution alters the nucleotide affinity and conformation of SUR2A NBD1. We have also investigated the effects of phosphorylation of the N-terminal tail on the conformation of the mutant NBD1 compared to wild type NBD1. Notably, phosphorylation partly corrects the defects caused by the V734I mutation. Full correction is observed when the N-terminal tail is deleted. Previous work on NBD1 showed that phosphorylation results in decreased interactions of the N-terminal tail with the core of NBD1, including at V734. Thus, these studies have provided insights into different regulatory regions within NBD1, the N-terminal tail and the region containing V734. These regulatory regions correspond to exons that are spliced out in naturally occurring isoforms of SUR2A. Removal of exon 14, which forms part of the N- terminal tail forms SUR2C while removal of exon 17, which contains V734, forms SUR2D. This chapter is currently in preparation for submission.5 Author contributions: E.D.A performed all the protein purification, NMR spectroscopy, fluorescence quenching experiments and nucleotide binding experiments. M.S. and J.P.L.-A. performed the molecular cloning. E.D.A. and V.K. wrote and edited the manuscripts.

5.1 Introduction

The enzymatic activity and interactions of the SUR NBDs are critical in regulating K ATP channel gating. As previously discussed, mutations in the NBDs of SUR1 cause neonatal type II diabetes or hyperinsulinism,92 while mutations in the SUR2A NBDs result in various cardiovascular disorders.31, 206, 221 One such mutation is the substitution of Val734 to an Ile in NBD1 of the cardiac-specific human SUR2A isoform. 206 This Val residue is highly conserved across species,

5 In Preparation

122 including human, mouse, rat and rabbit. 206 Homology modeling has illustrated that residue V734 is found in a loop located C-terminal to the Walker A motif between β-strands, β4 and β6, in NBD1.189 In some NBDs this region consists of only 3 residues that forms a tight turn, while this region in other NBDs consists of ≈15 residues that forms a small β-strand, followed by a 3 10 helix and a 2-turn α-helix. In SUR2A NBD1, this loop is longer than the homologous loop found in most NBDs (≈20 residues) and is predicted to be disordered. 64 A naturally occurring splice isoform lacking exon 17 has also been identified in various tissues. This splice isoform is referred to as SUR2D. In addition, a naturally occurring splice isoform also exists that lacks exon 14 (residues S627-F675 of the N-terminal tail) and is known as SUR2D. The significance of these isoforms is discussed below.

Removal of exon 17 from SUR2A to form SUR2D in the context of the full K ATP channel has been shown to significantly reduce gating of the channel in response to nucleotide. 50 Moreover, individuals carrying the V734I mutation have a 6.40-fold increased susceptibility for experiencing myocardial infarctions. 206 Recent electrophysiology studies have suggested that the 126, 226, 287 V734I mutation alters the response of vascular K ATP channels to nucleotide. Modeling of NBD1 based on SAXS data has also predicted the site of this mutation to be involved in several interdomain interactions of the NBDs. 226 Thus, the V734I mutation may compromise these interdomain interactions resulting in the disease state. However, the SAXS data were acquired on an NBD1 protein that lacks the N-terminal tail and also ends at residue E890 and thereby lacks the last three secondary structural elements of the canonical domain. In our hands, the corresponding NBD1 construct (D665-E889) resulted in little soluble protein expressed,64 and thus we have done all of studies on NBD1 proteins that extend to L933. Here we employ NMR and steady-state fluorescence methodology to examine the effects of the V734I mutation on the conformation and nucleotide binding of NBD1. We are using the NBD1 from rat SUR2A. Residue V734 in human SUR2A is V730 in rat SUR2A. We will be using the residue numbering for rat SUR2A from here on. In accordance with previous studies/nomenclature we have employed residues S615-L933 as the full length nucleotide binding domain 1 (referred to as NBD1). This protein contains an N-terminal tail that has been shown to be intrinsically disordered (Chapter 4, & Sooklal C.R., de Araujo E.D. and Kanelis V, manuscript in preparation) and also contains two phosphorylation sites (Chapter 4). The NBD1 protein from D665-L933 that is lacking this N-terminal region and is referred to as NBD1-ΔN. Fluorescence quenching studies of wild type NBD1 (NBD1-WT) and mutant NBD1 (NBD1-

123

V730I) revealed conformational differences between the MgATP-bound states of the proteins. These differences are highlighted by fluorescence TNP-ATP binding and NMR MgATP binding studies which suggest a reduction in nucleotide affinity in mutant NBD1. Notably, these differences in nucleotide affinity and fluorescence quenching between the wild-type and mutant proteins are partially reversed by phosphorylation of the N-terminal region. Complete removal of the N-terminal disordered region (NBD1-ΔN) reveals minimal differences between wild type and mutant proteins. Furthermore, removal of residues encoded by exon 17, results in multiple chemical shift changes in NMR spectra of SUR2A NBD1-ΔN, some of which are similar to those observed to change in response to phosphorylation. This implies that interactions of the disease-causing region with the N-terminal region of NBD1 are important for regulating K ATP channel gating. These findings will enhance our understanding of the molecular consequences of the V734I mutation and the potential importance of intramolecular SUR2A interactions in KATP channel gating.

5.2 Methods

5.2.1 Protein Expression

The expression and purification of soluble and monomeric wild type rat SUR2A NBD1 is described in Chapter 2.64, 66 Constructs with the disease-causing mutation V734I (or V730I with rat SUR2A numbering) were generated with QuikChange Mutagenesis (Stratagene) in both NBD1 and NBD1-ΔN and are referred to as NBD1-V730I and NBD1-ΔN-V730I respectively. Additionally, constructs were designed to eliminate residues V730 to S742 which are encoded by exon 17, to generate the naturally occurring splice variant of SUR2A NBD1 and NBD1-ΔN, which are referred to as NBD1-Δ17 and NBD1-ΔN-Δ17. All Trp mutants were generated through QuikChange Mutagenesis (Stratagene).

All SUR2A NBD1 variants were expressed and purified as previously described. 64, 66 Briefly, each construct was transformed into E.coli BL21 (DE3) CodonPlus-RIL (Stratagene) cells. Cells were grown and induced in 95 % 15 N-labelled M9 minimal media with 5 % LB media at 18 °C for 16 h. Following cell lysis, the soluble 6×HisSUMO-NBD1 was purified using a 5 mL Ni 2+ -NTA affinity column (GE Healthcare) at 4 °C. The 6×His-SUMO tag was cleaved using a 6×His-Ulp1 protease and the isolated NBD1 was separated and purified to homogeneity by size

124 exclusion chromatography (Superdex 75, GE Healthcare) followed by a reverse Ni 2+ -NTA affinity column. Buffers used in the purification are described in Chapter 2.

Purification of NBD1-V730I required higher concentrations of MgATP throughout the purification process (15 mM MgATP versus 10 mM MgATP) in order to avoid precipitation of the isolated domain. In order to ensure that all protein preparations had identical concentrations of MgATP, the proteins were subjected to gel filtration to remove excess MgATP. Following gel filtration, the proteins were supplemented with the appropriate concentration of fresh nucleotide as required for each biophysical study. For the assays requiring the apo state, MgATP was also removed through gel filtration, and samples were supplemented with 5 mM EDTA.

5.2.2 Fluorescence Quenching

Fluorescence experiments were conducted on a Fluoromax-4 spectrofluorimeter (Horiba-Jovin, Inc.) equipped with a Peltier unit for precise temperature control and an automatic titrator. Trp fluorescence quenching studies were performed using solutions containing 0.3 M KI or 0.6 M acrylamide. All quenching solutions were prepared fresh and 0.1 M Na 2S2O3 was added to the KI - 170 stock solutions to prevent formation of I 2 and I 3 species. Trp fluorescence data were recorded at 15 °C, with excitation and emission wavelengths of 295 and 350 nm, respectively. The excitation and emission slit widths of 1.0 and 4.0 nm, respectively. The excitation wavelength of 295 nm was chosen to selectively excite Trp residues in the protein. Fluorescence emission was monitored at 350 nm to probe fluorescence quenching of exposed Trp residues. Samples contained 2 μM NBD1 either with 2 mM MgATP or 5 mM EDTA as needed. In order to account for dilution of NBD1 and changes in ionic strength during quenching experiments with KI, control experiments were repeated with KCl instead of KI. The net change in fluorescence was determined by the equation, F (F F) o = o KI 183, 214 (Eq. 4.1) F (F F) o KCl where Fo is the initial fluorescence and F is the fluorescence at each point in the titration, and

(F o/F) KI and (F o/F) KCl represent the fluorescence ratios in the presence of KI or KCl, respectively. In order to account for the dilution effect during acrylamide quenching experiments, parallel titrations were performed with buffer alone. Quenching data with acrylamide were also corrected for the inner filter effect.165, 183, 266 The absorbance at 295 nm

125

(A 295 ) and 350 nm (A 350 ) of the solution at each titration point were recorded and the observed fluorescence was corrected according to the equation,165, 266

+ A295 A350 F = F 10 2 (Eq. 4.2) corrected observed

The values of A 295 and A 350 at the highest acrylamide concentration used were 0.023 and 0.005, respectively. The corrected fluorescence quenching data for I - and acrylamide were analyzed 170 using the Stern-Volmer equation (Eq 1.2) to obtain the Stern-Volmer quenching constant, K SV .

5.2.3 Fluorescence Nucleotide Binding

The K d value for binding of the fluorescent ATP analogue 2',3'-O-(2,4,6-trinitrophenyl)- adenosine-5'-triphosphate (TNP-ATP, Molecular Probes) for the different NBD1 proteins was determined as previously described.188 Both Mg 2+ and ATP were removed from NBD1 samples using size exclusion chromatography as described in section 5.2.2 and replaced with 2.5 µM

MgCl 2 and 2.5 µM TNP-ATP as described in section 5.2.2. Because NBD1 samples without nucleotide have limited solubility binding experiments were conducted at 15 °C with 10 % (v/v) glycerol in the buffers.

Binding experiments were performed by serial dilutions of protein, starting from 50 -70 µM (depending on the concentration eluted from the size exclusion column) to 0.8 -2.0 µM, while maintaining a constant concentration of MgCl 2 and TNP-ATP at 2.5 µM each. Fluorescence spectra of TNP-ATP were recorded using an excitation wavelength of 465 nm and a slit width of 5 nm. Emission spectra were collected from 485 – 600 nm with a slit width of 7 nm. The K d value for the NBD1-nucleotide complex was determined by monitoring the ratio between the fluorescence intensity at 533 nm, which corresponds to the wavelength where the fluorescence difference of free and bound TNP-ATP is at a maximum, and 600 nm to account for any non-specific fluorescence from the protein.223 The titration data were fit to the equation (2.1). This equation assumes a 1:1 complex of NBD1 with TNP-ATP.110, 317

5.2.4 NMR Spectroscopy

TROSY-HSQC 232 spectra of SUR2A NBD1 proteins were recorded at various temperatures on a 600 MHz Varian Inova spectrometer equipped with a H(F)CN triple resonance cryoprobe and actively-shielded z-gradients. Chemical shifts for each spectrum were referenced to 4,4-

126 dimethyl-4-silapentane-1-sulfonic acid (DSS).330 The samples were prepared with the same dialysis buffer stock to ensure the solution conditions of all samples were identical. Spectra were processed using NMRPipe/NMRDraw 75 and analyzed with NMRView. 139

5.3 Results

5.3.1 Altered conformation of wild type and mutant SUR2A NBD1

Our previous high quality NMR spectra and fluorescence studies of NBD1 have demonstrated that the isolated first nucleotide binding domain of SUR2A is folded, monomeric and binds MgATP. 64, 66, 189 However, resonances in the NMR spectra of the nucleotide-free NBD1 are significantly broadened and hence assessing structural changes from MgATP binding, in NBD1 in the wild type state or with the V730I mutation was not feasible by NMR spectroscopy. To resolve these challenges, we used intrinsic Trp fluorescence to probe conformational changes of NBD1 upon MgATP binding. Conformational changes in proteins have been studied using Trp residues as reporters in several different proteins,68, 167 including ABC transporters. 184 SUR2A NBD1 contains six Trp residues located throughout the protein. Homology models of SUR2A NBD1 based on the NBD1 from the ABCC protein, multidrug resistance protein 1 (MRP1), revealed that two Trp residues are buried in the core and the remaining four Trp residues are partially surface exposed.

Fluorescence quenching of NBD1 by iodide and acrylamide displays conformational changes between the apo- and ATP-loaded states (Figure 5.1). In the case of iodide quenching, binding of MgATP resulted in a substantial increase (~60 %) in the K SV constant from 0.94 ± 0.10 M -1 to 1.51 ± 0.11 M -1. In contrast, the acrylamide quenching constant moderately decreased (~14 %) upon MgATP binding from 3.79 ± 0.06 M -1 to 3.27 ± 0.12 M -1. Decreasing

KSV constants are generally indicative of burial of a Trp residue, while an increase in the K SV value suggests greater exposure of a Trp residue. The opposing changes in iodide and acrylamide

KSV values following the addition of MgATP is reflective of the different Trp populations probed by the different quenchers. Potassium iodide can only penetrate the hydrophobic core to a limited extent and is excluded from anionic and electronegative regions of the protein. Acrylamide generally partitions to hydrophobic patches and can penetrate into the protein interior. The changes in potassium iodide and acrylamide K SV constants of NBD1 upon ATP binding, suggests

127 nucleotide binding induces conformational changes throughout NBD1 , as is the case with other NBDs upon ATP binding. 184

We have also used fluorescence quenching to probe the conformational changes in mutant NBD1-V730I from nucleotide binding (Figure 5.1). The apo state of NBD1-V730I displays K SV values from acrylamide that are ~12 % larger than those observed for apo NBD1-

WT. An even greater increase (~25 %) is obtained in the K SV values from acrylamide for the MgATP-loaded state of NBD1-V730I compared with the MgATP-loaded state of NBD1-WT.

The larger K SV values for apo and MgATP-bound NBD1-V730I compared to the wild type protein indicate an increase in the accessibility of Trp residues in non-polar environments in the mutant. The increased accessibility of Trp residues to acrylamide in NBD1-V730I may result

Figure 5.1: Quenching studies of wild type (black) and mutant (red) NBD1. The fluorescence emission of 2 μM NBD1 either in the presence of 5 mM EDTA (A/C) or 2 mM ATP (B/D) was recorded following the addition of increasing amounts of either potassium iodide or acrylamide quenchers. The values were fit to the Stern- Volmer equation and the K SV values are provided.

128 from greater exposure of a surface Trp surrounded by hydrophobic residues caused by altered structure of the mutant or altered dynamics of the mutant that allow acrylamide easier access to buried Trp residues.

In contrast to that observed with acrylamide, K SV values for iodide quenching of the apoNBD1-V730I show no significant difference from that observed for the wild type. Furthermore, the addition of MgATP to NBD1-V730I does not result in any observable conformational changes via iodide quenching. The similar K SV values for apo and MgATP- bound NBD1-V730I indicate that any conformational changes that occur upon MgATP binding do not result in an overall increase in Trp exposure of the specific Trp residues that are sensitive to I - quenching. Together, the iodide and acrylamide quenching data demonstrate conformational differences in NBD1 and mutant NBD1-V730I.

Previous electrophysiology studies have suggested that the V730I mutation alters the 126, 286 MgATP affinity of the full K ATP channel. As our fluorescence quenching data displayed a differential response of wild type and mutant NBD1 to MgATP, we also evaluated the ability of NBD1-V730I to bind nucleotide using fluorescently labelled TNP-ATP. A higher concentration of mutant NBD1-V730I, compared to wild type protein, was necessary to saturate TNP-ATP fluorescence changes. Fitting the TNP-ATP fluorescence saturation curves resulted in a K d value of 8.4 μM ± 1.1 μM for TNP-ATP binding by NBD1, compared with a K d value of 20.3 μM ± 3.1 μM for NBD1-V730I, indicating the mutation reduces the nucleotide affinity of the isolated NBD1 (Figure 5.2).

Figure 5.2: TNP-ATP binding to wild type (black) and mutant (red) NBD1. The curves are fit to binding equation (2.1) which assumes a 1:1 interaction. The dissociation constants for both proteins are provided.

129

A B i ii

iii iv

Figure 5.3: TROSY-HSQC spectra of wild type and mutant NBD1 at different concentrations. A) Overlay of 250 μM NBD1-WT (black) and 250 μM NBD1-V730I (red). Changes in chemical shift/intensity are circled in cyan. In B) i/ii, the protein is present at 200 μM with 2 mM MgATP. In iii/iv, the protein is present at 400 μM with 2 mM MgATP. In iv, the mutant spectrum resembles the apo-state, while the wild type is essentially unchanged.

The lower affinity for TNP-ATP observed with NBD1-V730I is also consistent with our studies from NMR spectroscopy. NMR spectra of NBD1-WT and NBD1-V730I are shown in Figure 5.3A. As seen for NBD1-WT, the spectra of NBD1-V730I show dispersion in the 1H dimension characteristic of a folded protein,162 indicating that the V730I mutation does not disrupt the overall fold. There are several backbone resonances that show different chemical shifts in the spectra of NBD1-WT and NBD1-V730I. These chemical shift changes may indicate altered interactions in the wild type and mutant proteins of the region that contains residue V730 (Figure 5.3A). Other resonances show different intensities between wild type and mutant proteins, suggesting that the mutation affects some dynamic properties in the protein. Consistent with the different nucleotide affinities of the two proteins, samples of wild-type and mutant protein containing a different ratio of MgATP:protein concentrations, yield remarkably different spectra (Figure 5.3B). For wild type NBD1, spectra remain identical even at lower MgATP concentrations (comparing ranges from 5:1 to 10:1, MgATP:protein concentrations). In contrast, resonances in the NMR spectra of highly concentrated NBD1-V730I are significantly broadened at lower MgATP concentrations (5:1). Increasing the relative concentration of ATP (to yield a

130 molar ratio of 10 MgATP:1 protein) yields higher quality mutant spectra. The broadening of resonances in the spectra of the NBD1-V730I due to lower molar equivalents of MgATP yields NMR spectra similar to that for apo NBD1-WT. This phenomenon supports the finding that isolated NBD1 containing the disease-causing mutation is compromised in binding MgATP.

5.3.2 Phosphorylation of the N-tail in NBD1-V730I partially masks the molecular effects of the V730I mutation

Previous studies of isolated NBD1 have shown the importance of the N-terminal tail for regulating the activity of SUR2A (Chapter 4). The N-terminal tail has been demonstrated to be intrinsically disordered (Chapter 4 & Sooklal, de Araujo and Kanelis, unpublished) and harbours two phosphorylation sites. Phosphorylation of the N-tail disrupts its interactions with the core of NBD1, yielding a hydrodynamically larger and more open conformation. We have examined how phosphorylation of the N-tail affects NBD1-V730I. Phosphorylation of NBD1-V730I was performed in a similar method to phosphorylation of the wild type protein (Chapter 4). NMR spectra of the phosphorylated NBD1-V730I (phospho-NBD1-V730I) exhibit differences in chemical shifts and peak intensities compared to phosphorylated NBD1-WT (phospho-NBD1- WT), suggesting that the mutation affects some dynamic properties in the protein. Additional chemical shift changes from the V730I mutation may also be masked by resonance overlap.

Figure 5.4. TROSY-HSQC of phospho-NBD1-V730I (200 μM) at different temperatures. NMR spectra are shown in black with aliased peaks in red.

131

-1 -1 Ksv = 2.03±0.10 M Ksv = 1.56±0.17 M -1 -1 Ksv = 1.88±0.11 M Ksv = 1.52±0.13 M

Figure 5.5 Quenching studies of phospho-wild type (black) and phospho-mutant (red) NBD1. The fluorescence emission of 2 μM NBD1 either in the presence of 5 mM EDTA (A/C) or 2 mM ATP (B/D) was recorded following the addition of increasing amounts of either potassium iodide or acrylamide quenchers. The values were fit to the Stern-Volmer equation and the K SV values are provided.

Quenching studies of the phosphorylated samples reveal insights into the mechanism of action of the disease causing mutation. Quenching constants (using potassium iodide) are slightly reduced when both NBD1-WT and NBD1-V730I are phosphorylated (Figure 5.5, Table 5.1). The changes in the quenching constants may be also related to the changes in electrostatic charge of the protein due to phosphorylation. However, the differences in quenching constants between the phospho-NBD1-WT, and phospho-NBD1-V730I are much smaller (24 %) compared to differences in K SV values between the non-phosphorylated proteins (46 %). The reduced difference in K SV values may imply that the conformation of phospho-NBD1-V730I is similar to phospho-NBD1-WT.

132

A B

Figure 5.6: (A) TNP-ATP binding to NBD1-WT (black) and NBD1-V730I (red) and phospho-NBD1-V730I (blue) NBD1. (B) TNP-ATP binding to NBD1-ΔN (black) and NBD1-ΔN-V730I (red). The curves are fit to binding equation (2.1) which assumes a 1:1 interaction. The dissociation constants for both proteins are provided.

To examine this hypothesis, we measured the nucleotide binding affinity for phospho- NBD1-V730I (Figure 5.6A). Phospho-NBD1-WT has previously been shown to have an increased affinity for nucleotide as the displacement of the N-tail allows for increased access to the ATP-binding site. Phosphorylation of NBD1-V730I also showed an increase in nucleotide binding affinity (from 20.3 ± 3.1 μM to 6.6 ± 0.3 μM ). This recovery in nucleotide affinity for phospho-NBD1-V730I and the reduced response of NBD1-V730I to fluorescence quenching suggests an increased interaction between the N-tail and the site of the V730I mutation in NBD1 that is disrupted by phosphorylation.

5.3.3 Removal of the N-terminal tail rescues the activity of mutant NBD1

Previous studies have indicated that removal of the N-terminal tail to form NBD1-ΔN, mimics phosphorylation. We have investigated the consequences of the disease-causing V730I mutation in the context of NBD1-ΔN. Consistent with the reduced differences in KSV values observed for the phosphorylated proteins, no significant changes were detected in the K SV constants (with either iodide or acrylamide quenchers) between NBD1-ΔN-WT and NBD1-ΔN-V730I (Figure 5.7). The lack of observed differences suggests that the mutation does not substantially alter the conformation of the NBD1-ΔN protein or at the very least, the local environment of the Trp residues. This is consistent with previous predictions based on SAXS homology modeling for SUR2A NBD1 that contains residues D666-E890.226 These models predicted very limited differences between the wild type and mutant protein. However, the predictions were based on

133

Figure 5.7 Quenching studies of wild type (black) and mutant (red) NBD1-ΔN. The fluorescence emission of 2 μM NBD1 either in the presence of 5 mM EDTA (A/C) or 2 mM ATP (B/D) was recorded following the addition of increasing amounts of either potassium iodide or acrylamide quenchers. The values were fit to the Stern-Volmer equation and the constants are provided. proteins with domains boundaries that are most analogous to NBD1-ΔN and are lacking the N- terminal regulatory region. Our experiments with NBD1-ΔN support these predictions. However, as we have previously shown, extending the domain boundaries to begin at residue S615 reveals intramolecular interactions between the N-terminal tail and the core of NBD1 that are enhanced in the non-phosphorylated mutant.

NMR spectra of the wild type and mutant NBD1-ΔN proteins (in the ATP-bound state) show both proteins are properly folded and that the V730I mutation does not disrupt the overall structure (Figure 5.8). There are a few (<15) resonances that show significant chemical shifts, suggesting the mutation may only alter the immediate environment of the mutated residue. Additionally, there were no significant changes in ATP-binding affinity observed between the wild type (4.4 ± 0.7 μM) and mutant NBD1-ΔN (4.8 ± 0.9 μM, Figure 5.6B). The combined lack of changes in NMR spectra, fluorescence quenching constants, and nucleotide binding constants

134

A B

Figure 5.8. (A) TROSY-HSQC of phosphorylated NBD1 (black) and phosphorylated NBD1-V730I (red) (B) TROSY-HSQC of NBD1-ΔN (black) and NBD1-ΔN-V730I (red). Both spectra are shown at 30 oC with 200 μM of protein. indicates that the V730I mutation does not affect the conformation and activity of NBD1-ΔN. Thus, these data suggest that the V730I disease-causing mutation elicits its detrimental effects by altering interactions of the NBD1 core with the N-terminal tail and possibly with other regions of

SUR2A in the full K ATP channel. These data also indicate that studies of the SUR2A NBDs requires proteins that include regulatory residues outside the canonical domain.

The removal of the N-terminal tail improves the thermodynamic stability of the protein, which allows for a more detailed investigation into fluorescence quenching. For instance, to examine if the observed changes in Stern-Volmer constants were specific for MgATP binding, we performed quenching studies on a Walker A mutant (K707A) which is shown to be insensitive to nucleotide binding. The K SV values for NBD1-ΔN-K707A are identical in the presence or absence of MgATP with the K707A mutant (Table 5.1). Studies of the K707A mutant in the context of the NBD1 background were not possible due to poor solubility of the protein. This implies that the changes observed between wild type and mutant NBD1 are specific to MgATP binding.

135

A B

Figure 5.9. Fluorescence emission spectra of NBD1 and various Trp mutants in the absence (A) or presence (B) of ATP. Only NBD1 W677F shows noticeable changes upon the addition of ATP.

In order to localize the effects of intrinsic Trp fluorescence quenching studies we have employed individual residue substitutions. The emission profile of the various Trp substitutions is depicted in Figure 5.9. The removal of W677 (residue in vicinity of the A-loop) appears to be sensitive to ATP binding with slight changes also observed due to mutation of W906 (residue in the C-terminus). Substitutions of W877 (877A or 877F) destabilized the protein and could not be expressed in E. coli and assayed for the contribution to total protein fluorescence. ATP-binding and thermodynamic stability of all other Trp mutants was assessed and only appeared to be compromised in W677F (data not shown). The quenching constants for each Trp mutant in NBD1-ΔN are provided in Table 5.3, in the presence and absence of nucleotide. These studies support our surface topology predictions by homology modeling.

Table 5.1 Iodide Quenching Constants for various NBD1 constructs

-1 † Iodide Quenching Constant, K SV (M ) No ATP 2 mM ATP S615 - L933 0.94 ± 0.10 1.51 ± 0.11 S615 - L933 V730I 0.90 ± 0.01 0.81 ± 0.04 S615 - L933 Phosphorylated 0.60 ± 0.01 1.01 ± 0.08 S615 - L933 V730I Phosphorylated 0.41 ± 0.05 0.77 ± 0.02 D665 - L933 1.33 ± 0.06 1.02 ± 0.11 D665 - L933 V730I 1.22 ± 0.06 1.08 ± 0.02 D665 - L933 W677F 1.65 ± 0.43 0.64 ± 0.06 D665 - L933 W756F 2.16 ± 0.42 1.19 ± 0.20 D665 - L933 W906F 1.10 ± 0.06 0.72 ± 0.07 D665 - L933 W727F 1.03 ± 0.15 1.51 ± 0.26 D665 - L933 Δexon17 0.96 ± 0.11 0.89 ± 0.18 D665 - L933 K707A 1.22 ± 0.09 1.34 ± 0.03 †Values are reported as an average of n = 3 independent measurements

136

Table 5.2 Acrylamide Quenching Constants for various NBD1 constructs

-1 † Acrylamide Quenching Constant, K SV (M ) No ATP 2 mM ATP S615 - L933 3.79 ± 0.06 3.27 ± 0.12 S615 - L933 V730I 4.26 ± 0.18 4.10 ± 0.06 S615 - L933 Phosphorylated 2.03 ± 0.10 1.56 ± 0.17 S615 - L933 V730I Phosphorylated 1.88 ± 0.11 1.52 ± 0.13 D665 - L933 8.67 ± 0.35 10.46 ± 0.58 D665 - L933 V730I 8.35 ± 0.28 10.55 ± 0.29 †Values are reported as an average of n = 3 independent measurements

5.3.4 The N-tail (Exon 14) interacts with the disease-causing region (Exon 17) of NBD1

Physiological data acquired on full KATP channels containing SUR2D suggest a decreased response to MgATP. 50 We have studied the effects of deleting exon 17 from both NBD1 and NBD1-ΔN. Deletion of exon 17 in NBD1 (referred to as NBD1-Δ17) results in primarily insoluble E. coli protein expression. Any recoverable protein is easily prone to aggregation, even at low protein concentrations and thus our studies on NBD1-Δ17 are limited. However, removing exon 17 in the background of NBD1-ΔN (referred to as NBD1-ΔNΔ17) allowed for increased solubility and/or stability. This increase in stability may implicate an interaction between the N- terminal region and exon 17. NMR spectra of NBD1-ΔNΔ17 show several chemical shift changes that suggest alterations to protein conformation (Figure 5.10A). Many of the residues that are affected by the exon removal are thought to be close in space to exon 17, as predicted by homology modeling. These changes occur in regions close to the region containing exon 17 (Figure 5.10C). Additionally, a number of peaks are also broadened beyond detection suggesting changes in protein dynamics. Based on the resonance assignments, this occurs in regions far in space from the exon encoded region (Figure 5.10D). Moreover, a number of chemical shift changes that occur are also specific to changes that occur with phosphorylation of the N-tail. These changes suggest an interaction between the regions encoded by exon 14, i.e. most of the N-terminal tail, and exon 17 which contains V730. Fluorescence quenching of SUR2D also shows significant decreases in the quenching constants, which may be related to a contraction of the protein conformation due to removal of the loop (Table 5.1). The removal of exon 17 significantly reduces ATP affinity as well (Figure 5.10B). Overall these data suggest both exon 14 and exon 17 play important roles in modulating the conformation and nucleotide binding affinity of NBD1 that alters SUR2A-mediated regulation of KATP channel gating.

137 A B

C D

Figure 5.10.(A) TROSY-HSQC of 200 μM NBD1-ΔN (black) and NBD1-ΔN,Δ17 (red). (B) TNP-ATP binding to wild type (black) and mutant (red) and phosphorylated mutant (blue) NBD1 and NBD1-ΔNΔ17 (green). The curves are fit to binding equation (2.1) which assumes a 1:1 interaction. (C) A model of SUR2A NBD1-ΔN showing residues in magenta that exhibit significant chemical shift changes (> 2 standard deviations) (D) shows residues in NBD1-ΔN (magenta) that have been broadened beyond detection in NBD1-ΔN,Δ17.

5.4 Discussion

2+ Cardiac K ATP channels are essential to prevent intracellular Ca overloading and therefore have a protective role during ischemic events. 31, 206, 221 The genetic, yet conservative, V730I mutation in SUR2A NBD1 has been linked to a 6.4-fold increased risk of myocardial infarctions. 206 Here we have investigated the molecular effects of this mutation on SUR2A NBD1 and NBD1-ΔN

138

that is lacking the N-terminal tail. Full K ATP channel studies into the pathophysiological mechanism of this mutation have implied an altered sensitivity to nucleotides. 126, 287 Furthermore, the effect of the V730I mutation on an NBD1 protein similar, but not identical to NBD1-ΔN, has previously been predicted by SAXS-supported homology modeling of refolded and octameric NBDs. 225 These models suggest V730I is localized to a disordered loop at a protein interface and 225 likely disrupts interactions with other parts of the K ATP channel.

In order to investigate this mutation further, we have examined the biophysical properties of wild type and mutant NBD1. In comparison to the previous SAXS studies, 226 the domain boundaries of NBD1 used in this investigation include C-terminal residues that were shown to be critical for protein expression and stability. 64 The inclusion of these residues have allowed us to express NBD1 as a folded and monomeric protein in high yields.64, 189 Moreover, our domain boundaries also include an additional N-terminal region that is important for regulation of NBD1 (Chapter 4). Models of the truncated NBDs based on SAXS data indicated that V730I is likely not responsible for gross structural changes but likely alters the quaternary structure of the K ATP channel. 226 When we considered NBD1-ΔN, a protein with N-terminal domain boundaries similar to those used in SAXS modeling, our NMR and fluorescence data are consistent with the previous models. NBD1-ΔN-V730I does not show significant conformational changes or altered nucleotide binding compared to the wild type.

Recent data have depicted that NBD1 contains a number of regulatory regions, including an N-terminal tail that is disordered (Chapter 4 & Sooklal, de Araujo and Kanelis, unpublished) and contains two phosphorylation sites. Notably, part of this region also corresponds to an exon, (exon 14), that was found to be absent in the cardiac splice isoform SUR2C. 13, 107 We have examined the effects of the V730I mutation in the context of NBD1 (which includes the regulatory N-terminal tail). Notably, the presence of this regulatory region results in the different fluorescence quenching constants between the wild type and mutant protein. These differences are also supported by our TNP-ATP binding assays that demonstrate NBD1-V730I has a decreased affinity for nucleotide.

The observation that differences in wild type and mutant NBD1 only manifest with the presence of the N-terminal region, suggests this region may mediate allosteric control of the protein. Furthermore, this N-terminal region contains two phosphorylation sites and

139 phosphorylation has been shown to disrupt interactions between the N-terminal region and the core of the protein. We sought to identify the effects of phosphorylation on mutant NBD1. Consistent with our data, phosphorylation of the protein to some extent masks differences between mutant and wild type NBD1. ATP-binding was partially recovered when the mutant was phosphorylated. These results suggest that the N-tail interacts with a region localized to V730I. Removal of this N-tail, such as by phosphorylation, may reduce the interactions and therefore biochemical effects of the mutation.

Alternative splicing of the SUR2A gene has been suspected of modulating nucleotide affinity to address temporal specific needs of the cell.50 This is depicted by changes in burst 13, 50 patterns of K ATP channels upon different splice variant usage of SUR2A in K ATP channels. Our NMR data from NBD1-ΔNΔ17 suggest a number of residues that are affected by exon 17 removal also change with phosphorylation or removal of the N-tail. These results suggest a transient interaction between the two regions. The presence of the region exon 14 decreases nucleotide affinity of NBD1 whereas the presence of residues encoded by exon 17 promotes

ATP affinity. In this way, the specific splicing of the K ATP cell can regulate the activity channel, thereby resetting the membrane potential. Modifying the exon 17 encoded region, as with the presence of the V730I mutation, alters the conformation or interactions of exon 17 with the rest

Figure 5.11. Model of the effects of different regulatory regions in NBD1. A schematic of the primary sequence of NBD1 with the core of the protein shown in blue, the N-tail (encompassing exon 14) shown in red, and exon 17 shown in green. The V730I mutation is shown in yellow. The splicing of different regions results in different effects on the NBD1. The extent of these changes is depicted by the crosses. The positioning of the crosses is based on the K d values of each of the proteins for nucleotide. In the case of NBD1-Δ17, the protein was not soluble, likely due to low stability, and the K d value could not be measured.

140 of NBD1, which results in compromised nucleotide binding leading to disease. A model of the effects of these interactions on NBD1 is shown in Figure 5.11. Previous biochemical data have suggested that NBD1 is required in an "activated" state in order to stimulate channel opening. 23, 24 137 This state is stabilized by different events, such as nucleotide binding at NBD1, MgADP binding at NBD2, or interactions with phospholipids. 23, 24 137 Our data suggest that the presence or removal of different exons in NBD1 may stabilize or destabilize the "activated" state of

NBD1. This, in turn, alters the activity of the K ATP channel.

Although our data strongly suggest an interaction between the N-tail and V730I containing region, this is likely not the only effect of the mutation. For instance, previous studies 286 have shown that mutant K ATP channels are susceptible to the antianginal drug, nicorandil. We have tested nicorandil binding to both wild-type and mutant (NBD1 and NBD1-ΔN) and have not detected any measurable binding even at 10-fold drug concentrations (data not shown). These results suggest that the mutation likely has other effects on K ATP channels including the NBD1/NBD2 interface as well as coupling helices.

Primarily, MgATP forms part of the NBD1/NBD2 heterodimer interface and stimulates NBD dimerization. 17, 288, 316, 337 Our data for all proteins is obtained from monomeric NBD1. As such we can extrapolate that a compromised conformation of monomeric NBD1 in the full channel, coupled to a reduced nucleotide affinity, would result in decreased NBD1/NBD2 dimerization of NBD1-NBD2. Additional consequences on NBD dimerization may be due to structural changes in NBD1-V730, also as suggested by Park et al .226 The loop containing the residue V730 (V730-loop) is located at the base of the ATP-binding α/β subdomain next to the hinge between lobe I and lobe II.64 Lobe I is formed by the ATP-binding α/β subdomain and the β-sheet subdomain, whereas lobe II is formed by the α-helical subdomain. The V730-loop is structured in some NBDs,5, 62, 100, 130, 149, 175, 176, 248, 325, 335 and may be disordered in others including SUR2A NBD1.64 Mutation of V730 to the bulkier Ile may alter the structure of the loop or its interactions with other parts of NBD1. Because of the proximity of the V730-loop to the lobe I-lobe II hinge, structural changes in the V730I-loop may compromise conformational changes or motion of the hinge, which would ultimately affect the orientation of lobe I and lobe II with respect to NBD2.

141

Structural evidence from Sav1866 has suggested another possible mechanism of action through interactions of the coupling helices. 62, 240 Coupling helices provide the mechanical link between the NBDs and MSDs in ABC transporters.119 Again, the V730I mutation may sterically compromise coupling helix binding by NBD1. A lack of structural complimentary between NBD1 and the coupling helices may also be observed for other mutations at the V730 position or of other residues in the V730-loop, but such mutations have not yet been reported. Altered NBD/coupling helix interactions would compromise the communication of MgATP binding and hydrolysis at the NBDs to the MSDs, and ultimately to the channel pore. In addition, altered interactions of NBD1-V730I with the coupling helices may also destabilize full length SUR2A, and thus the K ATP channel. Notably, there are a variety of mutations in the coupling helices of SUR1 that cause hyperinsulinism or diabetes,2, 92 highlighting the importance of the coupling helices and their interactions with the NBDs in the SUR proteins. Thus, a full examination of the effect of the V730I mutation, as well as other mutants, requires additional studies with other regions of SUR2A such as the coupling helices and NBD2, and with the cytoplasmic domains of the Kir6.2 channel pore. However, our samples of SUR2A NBD1 allow for studies of the effect of the V730I mutation of nucleotide binding and structural changes in the isolated protein. These studies demonstrate that both the N-terminal region and the region encoded by exon 17 collectively modulate NBD1 conformation and nucleotide affinity. These findings will shed light on regulation in the K ATP channel complex. The compromised nucleotide binding and altered conformations in NBD1-V730I compared with NBD1-WT likely have many implications for

SUR2A-mediated regulation of K ATP channels.

142

Chapter 6 Expression and Purification of SUR2A NBD2 6 Overview

NBD1 in ABC proteins has been shown to have a number of productive interactions with several other protein domains, most importantly, NBD2. This chapter describes our studies with the isolated protein, SUR2A NBD2. Our goal was to assess the nature of the interactions between SUR2A NBD1 and NBD2 through NMR spectroscopy. As such we have attempted various strategies to purify SUR2A NBD2 as a monomeric protein and analyze its interactions with NBD1. The studies described in this chapter can help provide a framework for future experiments with SUR2A NBD2.

6.1 Introduction

The activity of SUR2A NBD1 is known to be modulated by several protein domains and ligands 35, 41, 285 which is important for regulation of the K ATP channel. These protein domains can contact NBD1 by both stable or transient interactions and alter its nucleotide binding affinity. SUR2A NBD2 is of particular importance as the formation of the NBD1/NBD2 heterodimer complex and/or ATP hydrolysis are important for proper channel regulation. 35 Furthermore, the final 42 residues of NBD2 are the only difference between SUR2A and SUR2B isoforms and result in different pharmaceutical sensitivities of K ATP channels, suggesting critical a role for NBD2 in the full channel. 71 Moreover, several disease-causing mutations have been identified in SUR2A NBD2 which are responsible for cardiac disease, 31 Cantú syndrome,310 and Brugada syndrome. 126 Two of the cardiac-disease causing mutations include A1513T and a frameshift mutation at the codon encoding L1524. Residues A1513 and L1524 are found in the final exon in SUR2A, exon 38A, that is comprised of the 42 C-terminal residues of the protein. These mutations affect the ATP hydrolysis cycle, implying that they may interact with NBD1 differently or have altered protein dynamics. Unfortunately, purification of NBD2 from eukaryotic ABC transporters has been historically challenging. Previous studies of SUR2A NBD2 were performed through refolding of the NusA tagged NBD2 and resulted in mixtures of higher order oligomers and octameric complexes. 226 Dynamic light scattering of these proteins showed particles over 100 nm in diameter confirming a molecular weight complex that is much

143 larger than monomeric NBD2.226 We expect that the hydrodynamic radius of NBD2 will be similar to the core of NBD1 (~3-4 nm), as seen in Chapter 4. Regardless, SAXS and homology modeling experiments performed with the oligomers, suggested dimerization interfaces and contact points between different NBD1 and NBD2 subunits in the full K ATP channel.

6.1.1 Asymmetry of NBDs in ABC transporters

In some prokaryotic ABC transporters, both NBDs are equivalent and therefore there is no real distinction in the individual function of the domains. 59 Even in the case of specific eukaryotic ABC transporters, NBD1 and NBD2 are strongly similar in structure and thought to be functionally equivalent. A well studied example involves the ABCB1 transporter, P- glycoprotein, where NBD1 and NBD2 are thought to function in identical ways. 272, 308 For instance, studies have suggested that ATP-hydrolysis alternates between NBD1 and NBD2 in P- glycoprotein with each transport cycle.36, 271 Cross-talk between the NBDs has also been shown by binding of ATP-transition state analogues to one NBD which prevents binding of a second analogue at the opposite NBD. Further, blocking either NBD site such as through reaction of nitrobenzo-2-oxa-1,3-diazole at the catalytic residues, destroys activity for the full transporter. 273, 274 Additional studies have suggested that there are two hydrolysis events in a single transport cycle, where the first hydrolysis cycle serves to facilitate transport activity and the second event allows for a reset of the transporter to the original state. 25, 311 However in these studies there is no preference for whether NBD1 or NBD2 carries out the initial hydrolysis event and the selection appears to occur in a random order. 264, 265 From these studies, both NBD1 and NBD2 in P- glycoprotein appear to be functionally equivalent.

However, this is not the case for NBD1 and NBD2 of all ABC transporters. Some transporters are thought to possess NBDs that have significantly different nucleotide binding affinities and hydrolysis rates. 125, 294 MRP1, which carries out a similar function to P- glycoprotein, appears to possess two non-equivalent NBDs. 96 ATP hydrolysis appears to predominantly occur at MRP1 NBD2. NBD2 also has higher binding affinity for MgADP than MRP NBD1. The binding of MgADP at MRP2 NBD2 also promotes increased binding of MgATP to NBD1. 96, 124 In CFTR NBD1, the NBDs are also thought to be non-equivalent with slower (if any) hydrolysis occurring at NBD1. 4, 295 This is also the case in the SUR proteins, where the NBDs are thought to behave asymmetrically, with hydrolysis activity occurring at

144

NBD2 and limited (if any) hydrolysis occurring at NBD1. 198, 199 As with other ABC transporters, both NBDs are required for proper activity which allows for cooperativity between the domains. Furthermore, MgADP binding at NBD2 stabilizes the binding of MgATP (and not MgADP) at NBD1. Notably, it was shown that MgATP (and not MgADP) is favoured at the NBD1 site in order to allow for MgADP to open the channel by interacting at NBD2.306

Cooperativity and communication between the NBDs is important for proper functioning of ABC transporters as mutations that that disrupt catalytic activity or nucleotide binding at one NBD negatively impact the opposing NBD. In this chapter, we endeavored to characterize the different protein-protein interactions between SUR2A NBD1 and NBD2. The focus of these studies involved developing strategies for expressing NBD2 as a monomeric protein. Our preliminary studies with SUR2A NBD1/NBD2 suggest we can probe interactions between these two domains.

6.2 Methods

6.2.1 Protein Expression and Purification

6.2.1.1 Purification of different SUR2A NBD2 constructs under native conditions

All SUR2A NBD2 variants were expressed and purified using strategies similar to SUR2A NBD1 purification. 64, 66 Briefly, SUR2A NBD2 was cloned into a pET-15b-6×HisSUMO vector. Each construct was transformed into E. coli BL21 (DE3) CodonPlus-RIL cells. Cells were grown in 95 % M9 minimal media with 5 % LB media at 37 °C and induced with 0.75 mM IPTG at 18 °C for ~16 h. The cells were lysed by sonication and the different fusion proteins were purified using a 3 mL Ni 2+ -NTA affinity column. The equilibration and elution buffers used here were identical to those used in Chapter 2. The elution fractions from the Ni 2+ column were loaded on a size exclusion column to assess the extent of any aggregation (Superdex 75, GE Healthcare). The 6×His-SUMO tag was not removed from the protein to prevent precipitation of NBD2.

6.2.1.2 Purification of SUR2A NBD2 under denaturing conditions

Denaturing purification strategies were also attempted with SUR2A NBD2, followed by refolding of NBD2. In this case, the E. coli cell cultures were grown, harvested and lysed as

145 previously described. However, following the initial sonication the cell pellet was suspended in equilibration buffer containing denaturant. The equilibration buffer contained 6.0 M urea, 20 mM Tris-HCl pH 7.3, 150 mM NaCl, 5 mM β-mercaptoethanol, 5 mM imidazole and 5 % glycerol. The inclusion bodies were solubilized with the aid of sonication. The suspension was nutated for 15 minutes and then cleared by centrifugation at 10000 g. The supernatant was filtered and passed through a 3 mL Ni 2+ -NTA affinity column, washed with 5 column volumes of equilibration buffer and eluted with elution buffer. The elution buffer was identical to the equilibration buffer with 500 mM imidazole in place of 5 mM imidazole. The elution fractions were assessed for protein content through SDS-PAGE.

Refolding of SUR2A NBD2 was performed through two methods. A step-wise dialysis was used to incrementally remove the denaturant from the protein sample. Dialysis was performed systematically against buffers with concentrations of 4.0 M urea, 2.0 M urea and 0 M urea. Alternatively, refolding was attempted through a slow drop-wise dilution into equilibration buffer containing 0 M urea and 0 mM imidazole. In both cases, the protein was subsequently concentrated and loaded onto a S75 gel filtration column to determine the extent of NBD2 oligomerization.

6.2.2 NMR Experiments

TROSY-HSQC of the SUR2A NBDs were recorded at 30 ºC on a 600 MHz Varian Inova spectrometer equipped with a H(F)CN triple resonance cryoprobe and actively-shielded z- gradients. NMR spectra were processed with NMRPipe/NMRDraw 75 and analyzed with NMRView. 139 Chemical shifts for each spectrum were referenced to 4,4-dimethyl-4-silapentane- 1-sulfonic acid (DSS).

6.3 Results

6.3.1 SUR2A NBD2 Expression and Purification Profiles

In order to determine the domain boundaries for SUR2A NBD2, we have employed a similar approach as used previously for SUR2A NBD1. The N-terminal domain boundary was selected on the basis of the first predicted β-strand in NBD2 using our structure-based sequence alignment. 64 Conventionally, the C-terminal domain boundary for NBD2 is thought to occur at

146

the end of exon 37, with the remaining 42 residues from the last exon, 38A, (or exon 38B in SUR2B NBD2) forming an independent regulatory region. 64 However, our alignment indicates that exon 38A (or 38B) contains secondary structural elements that are part of the canonical NBD structure in eukaryotic ABC proteins. As such we have included the residues corresponding to exon 38A as part of SUR2A NBD2. Expression and purification of SUR2A NBD2 was carried out in a similar manner as for SUR2A NBD1. Figure 6.1 provides an overview for the purification of a protein comprising SUR2A NBD2 that contains residues E1300-K1545. SUR2A NBD2 E1300-K1545 was expressed in E. coli and the cells were lysed by sonication. As shown in Lane 3 of Figure 6.1A, following sonication of the cell-lysate, the insoluble fraction includes a large band corresponding to ~45 kDa, likely the His-SUMO- SUR2A-NBD2 fusion protein. This suggests that the fusion protein is expressed within inclusion bodies in E. coli and only a small amount appears to be in the soluble fraction. Nonetheless, the cell lysate was cleared by the centrifugation and loaded onto a Ni 2+ affinity column. The beads were washed and then eluted with buffer containing 500 mM imidazole. Although protein corresponding to the expected size of 6×His-SUMO-NBD2 was recovered following purification, NBD2 also needs to be properly folded and monomeric (no aggregation). To ascertain the extent of any NBD2 oligomerization, the protein was loaded onto a gel filtration

kDa 250 148

98

64

50

36

Figure 6.1: Purification of SUR2A NBD2. (A) SDS-PAGE of NBD2 cell lysis and Ni 2+ -NTA purification. The soluble and insoluble fraction are represented by 1/1000 the volume of the total fraction. The flowthrough and wash were loaded on to the gel at 1/5000 the total volume. The elutions represent 1/1000 the final volume. (B) UV absorption profile of NBD2 and standards loaded onto an S75 gel filtration column. NBD2 is shown in the solid line and the standards are distinguished by the dashed lines. The molecular weights of the standards are given over their corresponding peaks, with blue dextran representing the void volume.

147 column. However, as shown in Figure 6.1B, the protein elutes close to the void volume of the column, indicating some higher order NBD2 oligomer is formed. We have experimented with a number of strategies in order to overcome the challenges of insoluble expression and aggregation of 6×HisSUMO-NBD2.

6.3.2 NBD2 domain boundary screening

In order to minimize the extent of NBD2 aggregation, we have attempted screening domain boundaries for SUR2A NBD2 based on refining the structure-based sequence alignments. This has included extending the N terminus of the NBD2 construct to the predicted helices of the membrane spanning domains and truncating the C terminus to remove exon 38A. Table 6.1 shows a summary of the purification profiles for each of the NBD2 constructs tested.

Table 6.1: Expression and purification of various SUR2A NBD2 constructs SUR2A NBD2 Construct Overexpression in E. coli Purified as monomer? E1300 - K1545 Yes No E1300 - V1503 † Yes No E1307 - V1503 † Yes No R1263 - R1503 † Yes No † Construct subcloned by Marijana Stagljar

For each construct tested in Table 6.1, the fusion protein was primarily expressed in the inclusion bodies with only low levels of recovery following the first Ni 2+ -NTA column. In all cases, any protein that was present in the soluble fraction was loaded onto the gel filtration column and eluted at the void volume of the gel filtration column indicating a higher order oligomer as seen for NBD2 (E1300-K1545, data not shown for all NBD2 constructs). In the case of NBD2 (E1300-K1545), the oligomer peak was collected and dialyzed into the NMR buffer. The resulting sample was concentrated to 100 m M and a 15 N-1H TROSY-HSQC was recorded. The corresponding NMR spectra was of poor quality likely due to the large size of the NBD2 oligomer (data not shown).

We have attempted to separate the individual monomers from the NBD2 oligomer using denaturants such urea. However, denaturant treatment leads to irreversible precipitation of the samples and therefore was not a viable approach. Alternative strategies to overcome NBD2 aggregation involved phosphorylation of the protein. There are two known phosphorylation sites in NBD2. 279 We have used PKA to carry out the phosphorylation at 30 °C using a similar set of

148

kDa 250 148

98

64 50

36

16

Figure 6.2: Gel filtration following phosphorylation of NBD2 (A) SDS-PAGE of fractions eluted from the gel filtration column. (B) The UV profile of phosphorylated NBD2 is shown in a solid line. The molecular weights of the standards are given over their corresponding peaks, with blue dextran representing the void volume in dashed lines. conditions that were employed for NBD1 phosphorylation. 67 As can be seen in Figure 6.2, gel filtration of the phosphorylated NBD2 did not indicate a change in the oligomeric state. However, it should also be noted that the percent phosphorylation achieved in NBD2 was not examined by mass spectrometry, and it is possible that the aggregation may have also prevented phosphorylation.

6.3.3 Alternative strategies for obtaining SUR2A NBD2

In light of the challenges of expressing soluble and monomeric NBD2, we have attempted to purify the inclusion bodies and refold the protein. Protein refolding is less desirable since it can often lead to mixtures of different oligomers or disulfides, or partially unfolded aggregates. 284

Conventionally, partially unfolded protein species can be separated using size exclusion chromatography, although complete removal is unlikely. Refolding was attempted for 6×HisSUMO SUR2A NBD2 (E1300-K1545). The cellular pellet remaining after cell lysis (by sonication) was resuspended in lysis buffer containing (20 mM Tris, pH 8.0, 150 mM NaCl, 5 mM imidazole, 5 % glycerol and 6 M urea) and further sonicated to aid in solubilization. The resulting suspension was cleared by centrifugation, filtered and loaded on to a Ni 2+ -NTA column. The Ni 2+ -NTA column was washed with 5 volumes of equilibration buffer and the 6×HisSUMO- NBD2 protein was eluted with elution buffer. A SDS-PAGE gel of the purification is shown in

149

kDa 250 148

98

64 50

36

16

6

Figure 6.3: Purification of NBD2 under denaturing conditions. (A) SDS-PAGE of fractions eluted from the gel filtration column. (B) Normalized UV profile of refolded NBD is shown in a solid line. Additional peaks in the UV trace may also be due to impurities in the samples. The molecular weights of the standards are given over their corresponding peaks, with blue dextran representing the void volume in dashed lines.

Figure 6.3A. The removal of urea and refolding of NBD2 was attempted by two different methods. A slow drop wise dilution of NBD2 into refolding buffer (20 mM Tris pH 8.0, 150 mM NaCl, 2 % glycerol, 5 mM MgATP) was initially performed. Additionally, a separate sample was subjected to a step-wise dialysis of 4 M, 2 M and 0 M urea which allowed for a controlled reduction in denaturant. However, in each case the gel filtration of the refolded proteins indicated a higher order oligomer (Figure 6.3B).

6.3.4 SUR2A NBD1 as a solubility enhancer for SUR2A NBD2

The data presented above indicate that 6×HisSUMO SUR2A NBD2 is expressed in the insoluble fraction in E. coli and forms high order oligomers in solution. However, SUR2A NBD1 is expressed as a soluble 6×His-SUMO fusion protein in E. coli and isolated NBD1 is a monomer in solution at 250-500 μM concentrations, depending on the construct used. Thus, we attempted to use the complementary binding partner, SUR2A NBD1, to enhance the solubility of the protein using different strategies. We tested using NBD1 at several stages in the purification to enhance NBD2 solubility. For example, we have tested to determine if NBD1 can elicit a change the oligomerization state of NBD2 by mixing equimolar concentrations (50 μM) of NBD1 with

150

Figure 6.4: UV trace of the gel filtration of NBD1 and NBD2. 50 μM of NBD1 and 50 μM NBD2 (bottom line) were mixed and loaded onto the gel filtration column. As a control (top line) 50 μM of NBD2 was independently loaded onto to the column. The presence of NBD1 did not change the elution volume of NBD2.

NBD2 and monitoring any changes using gel filtration. However, as can be seen from Figure 6.4, the presence of NBD1 was not sufficient to promote soluble expression of NBD2. Part of the reason for the lack of response may be because of the low affinity between the isolated NBD1 and isolated NBD2, which would result in little NBD1/NBD2 complex formation at low micromolar concentrations. Because the ABC protein SUR2A is a single polypeptide protein, NBD1 and NBD2 are held in close proximity in the full channel. Thus, the local high concentration of NBD1 and NBD2 negates the requirement for a high affinity interaction between the two domains. In order to compensate for a low affinity, we have designed a construct that contains NBD1 coupled to NBD2 by a linker comprised of Gly and Ser residues (Figure 6.5). This strategy was successful for linking and expressing a fusion for MRP1 NBD1 and NBD2. 320 As such, NBD2 was subcloned into the pET-6×HisSUMO NBD1-ΔN vector along with a Gly/Ser linker and a TEV-protease cleavage site in between NBD1-ΔN and NBD2 (molecular cloning performed by Marijana Stagljar). The length of the linker region between NBD1-ΔN and NBD2 was determined on the basis of distance constraints from a homology model of SUR2A in which NBD1 and NBD2 form a productive dimer. 189 The model indicated that 8 Gly-Ser repeats, for a total of 16 residues, would provide the minimum distance that would allow for productive NBD1/NBD2 heterodimer formation. The His-SUMO-NBD1-NBD2 construct was transformed and expressed in E. coli . However, following cell lysis, the fusion

151

Figure 6.5: Schematic of the NBD1-NBD2 fusion constructs. In (A) the Gly-Ser linker consist of 16 residues, whereas in (B) the linker has been extended to a total of 45 residues. protein was not found in the soluble fraction. The solubility of both SUMO and NBD1 were not sufficient to allow for soluble expression of NBD2 and the fusion protein could not be successfully isolated. There is precedent for a longer linker between the NBD1 and NBD2 protein for improving solubility of the NBD1-NBD2 protein. 320 To test this idea, we have extended the linker to 45 residues of repeating Gly-Ser subunits. However, the increase in linker length did not aid in soluble expression of the NBD1-NBD2 complex, as the purification did not yield significant quantities of protein.

6.4 Discussion

The formation of the SUR NBD1/NBD2 heterodimer is important for K ATP channel gating. We have attempted to investigate the biophysical properties of NBD2 and its interactions with NBD1. In all cases the protein was predominantly expressed in inclusion bodies in E. coli . Gel filtration of purified NBD2 that remained in the soluble fraction eluted as a higher order oligomer complex. The generation of 6×HisSUMO-NBD2 inclusion bodies suggests protein aggregation in the cell during induction, possibly as a result of protein misfolding. The recovery of protein from inclusion bodies is generally performed with the use of chaotropic agents, such as guanidinium hydrochloride. We have also attempted to purify NBD2 under denaturing conditions as well, followed by protein refolding, however the monomeric state could not be recovered in high yields. There are several strategies that may be employed to obtain soluble protein. For example, inclusion bodies are often the result of overexpression of the protein, which leads to high intracellular concentration which creates conditions favourable for aggregation. One strategy to minimize inclusion body formation involves re-cloning NBD2 into a lower expression system that may result in lower concentrations of NBD2 in vivo . This may minimize the likelihood for protein aggregation, although it decreases overall yields. Other simpler approaches may also involve screening the concentration of IPTG and the timing of induction as these can reduce the level of protein expression, thereby increasing the amount of soluble protein recovered. 155

152

Should high levels of expression be required, the protein may be expressed in cells strains that are modified to include vectors to overexpress specific chaperones that aid in molecular folding. 70, 97 69 These cell strains have been used to maximize the level of soluble protein. The use of NBD1 to stabilize NBD2 in vivo is an attractive strategy to obtain soluble protein. However, our inability to recover soluble protein may be related to the activity of the dimer. It is likely that the NBD1/NBD2 complex is active inside the cell, leading to ATP hydrolysis and dissociation of the heterodimer into linked NBD1 and NBD2. The aggregation- prone NBD2 is thus no longer bound to NBD1, leading to insoluble protein expression of 6×HisSUMO-NBD1-NBD2. The activity of the NBD1/NBD2 heterodimer also leads to futile ATP hydrolysis and consumption of E. coli energy reserves. This would also trigger expression of 6×HisSUMO-NBD1-NBD2 into inclusion bodies to prevent waste of any resources. Future attempts to obtain soluble NBD1/NBD2 heterodimer may be assisted by the generation of catalytically deficient mutants, such NBD1-K707A. These mutants have disrupted nucleotide binding/hydrolysis due to mutations of key residues in the Walker A (NBD1-K707A) or Walker B motifs. Other strategies can also be employed to obtain the isolated NBD2. For example, in CFTR NBD1-ΔF508 the mutation is known to de-stabilize the entire domain, which leads to low protein recovery. Certain mutations have been used to compensate for the stability and allow for CFTR-NBD1-ΔF508 to be studied. 235 Corresponding mutations can be introduced into SUR2A NBD2, based on sequence alignments, to optimize the level of soluble protein expression. Alternatively, 6×HisSUMO-SUR2A NBD2 can be characterized to determine its oligomerization state and used to study its interactions with NBD1.

153

Chapter 7 Conclusions & Future Directions 7

7.1 Summary

The NBDs of the sulfonylurea receptor proteins are critical for proper regulation of K ATP channels (Chapter 1). These domains are important in order to couple the metabolic state of the cell (which establishes ATP and ADP concentrations), to the membrane potential for several 35, 83, 209 metabolically active tissues. In cardiac tissue, the K ATP channels open under stress and protect the heart against ischemia. 209 Mutations in the NBDs result in various disorders that reduce the sensitivity of the potassium channels. 35, 286 In order to explore the molecular basis through which the NBDs regulate the potassium channels we have employed several different biophysical approaches.

Initially, in order to investigate the biophysical and biochemical properties of the first nucleotide binding domain of SUR2A, we required monomeric and stable samples of the protein. Generation of these samples involved screening of the expression and solubility profiles for constructs of SUR2A NBD1 with different N- and C-terminal domain boundaries. The domain boundaries tested were determined from structure-based sequence alignments of the SUR NBDs as well as homology modeling (Chapter 2). To optimize both the quantity and quality of protein samples generated, we assessed the influence of various ligands and buffer components on the thermodynamic stability of SUR2A NBD1. This was mainly performed through intrinsic Trp fluorescence-based screens which identified conditions that allowed for a 10-fold increase in protein yield as well as longer sample life times (Chapter 2). These techniques developed for SUR2A NBD1 are also useful for the isolation and purification of high quantities of other low- stability domains, such as NBDs from other ABC proteins including SUR1 NBDs and Ycf1p NBDs.

We have also employed these techniques to screen for conditions to obtain NBD1 samples that would be stable at high concentrations (> 500 μM) for several days. This allowed us to carry triple NMR resonance assignment experiments in order to obtain resonance assignments

154 of NBD1-ΔN. Although the quality of the NMR spectra recorded was low, specifically 13 experiments that correlated the chemical shifts of the side chain Cβ nuclei to those of the backbone 1H and 15 N, we have developed specific 15 N labelling strategies to enhance the usability of the data (Chapter 3). Residues such as Leu, Val, Ile, Ser and Gly which are known to interconvert during amino acid biosynthesis, can be supplemented in a concentration-dependent manner with corresponding unlabelled amino acids to prevent scrambling in E. coli . This approach is both efficient and robust as it requires only minimal changes to the E. coli growth and expression of the recombinant protein. These strategies allowed us to improve the number of resonance assignments determined. From our three-dimensional NMR data and specifically labelled samples, we have obtained ~65 % of the amide backbone assignment for SUR2A NBD1. These studies also allowed us to re-evaluate our NMR data from our screens in a residue- specific manner and analyze the effects of nucleotide binding on SUR2A NBD1 conformation.

The NBD1-ΔN resonance assignments provided a platform to study different regulatory mechanisms of NBD1 at the level of individual residues. Phosphorylation of the NBDs in the full 243, 261 channel is known to enhance K ATP channel activity. We have investigated the regulatory effects of phosphorylation on the biochemical properties of NBD1 (Chapter 4). Our NMR data suggest that phosphorylation alters the protein dynamics of NBD1. Our data suggest that the N- terminal tail of NBD1, which contains the phosphorylation sites, makes transient interactions with the core of the domain, and that these interactions are disrupted upon phosphorylation. The displacement of the N-tail allows for improved access to the active site of the protein. This increases the affinity of NBD1 for ATP, which would promote NBD1/NBD2 dimerization. Thus, these data may explain the increase in K ATP channel activity that is observed upon phosphorylation. Furthermore, a significant portion of the N-tail corresponds to exon 14. Naturally occurring splice isoforms have been identified in cardiac tissue which are lacking exon 14 further highlighting the importance of this regulatory region.

We have also examined the molecular effects of the disease-causing mutation, V734I, in SUR2A NBD1 that leads to increased risk for cardiomyopathy.286 Our studies (Chapter 5) of the mutant protein demonstrate a reduction in ATP affinity, which is also observed in studies of full 286 KATP channels bearing the V734I substitution. Fluorescence quenching indicates the mutation results in protein conformational changes compared to wild type. These changes in conformation and nucleotide affinity are partially reversed by phosphorylation and fully reversed with removal

155 of the N-tail, suggesting an interaction between the loop encoding the V734I mutation and the N- tail of NBD1. Notably, both of these regions correspond to different exons (14 and 17) that are spliced from the mRNA in specific tissues. 50, 113, 174 Likely, these results suggest an important role of exon usage and alternative splicing in regulating the activity of the KATP channels.

125, The interplay between NBD1 and NBD2 is a key step in regulation of K ATP channels. 133 We have attempted to generate NBD2 in order assess the nature of the interactions between the two domains (Chapter 6). Although NBD2 can be overexpressed in significant quantities in E. coli , we were not able to obtain the protein in a monomeric state, even when expressed as a fusion protein with solubility tags and its binding partner, NBD1. Although this is not unusual for eukaryotic NBDs, additional trials to obtain soluble NBD2 protein would be informative for insights into the dimerization interface and the mechanisms of disease-causing mutations in NBD2.

As an additional regulatory mechanism, we have investigated the interactions of NBD1 with different potassium channel openers (Chapter 8/Appendix). Using NMR and fluorescence spectroscopy, we have shown a direct and specific interaction of the drug, pinacidil, with NBD1. This interaction is coupled to an increase in nucleotide binding affinity suggesting a possible mechanism of action for the drug. Other potassium channel openers, such diazoxide, do not show any changes to NBD1 protein conformation or nucleotide affinity. Based on the number of chemical shift changes upon pinacidil addition, as well as the measured dissociation constant, 208, 268, 307 and binding data on full length K ATP channels, it is likely that NBD1 only forms part of 189 the binding site on the K ATP channel.

7.2 Discussion

The NBDs are critical sites for regulation of the KATP channel. Currently, the role of nucleotide hydrolysis in KATP channel activity remains unclear. Some studies have suggested that MgADP binding is sufficient to stimulate activity,106, 281, 344 while other studies support the role of hydrolysis in channel opening. 30, 197, 202, 342 Furthermore, there is also controversy over the occurrence of MgATP hydrolysis at both of the NBDs. Although all experiments support cooperativity between the NBDs, various studies have depicted hydrolysis at solely NBD2 211, 331 while other experiments have detected ATPase activity at the NBD1 site as well.30, 31, 342 However in all cases, catalytic activity proceeds with a faster rate in NBD2 in comparison to

156

NBD1. The asymmetry in the SUR NBDs, coupled to limited hydrolysis at NBD1 suggests an intricate role for NBD1 in K ATP channel regulation.

An overall model of K ATP channel regulation suggests a dual role of ATP binding. ATP binding at the inhibitory Kir6. x sites prevents gating of the channel.12 The presence MgADP at the NBD2 site allows for opening of the channel. 199 This response to MgADP binding at NBD2, is only possible if MgATP is present at the NBD1 site, as shown by photolabelling studies. 306 For instance, the sulfonylurea, glibenclamide, competitively displaces MgATP from NBD1 resulting in channel closure, despite the presence of MgADP at NBD2. It was suggested that the conformation of SUR NBD1 in the MgATP state is required to overcome the inhibitory ATP- bound site on the Kir6. x subunit. 199, 306 Additional studies have expanded upon this model, identifying specific residues that may functionally couple NBD1 in the ATP bound state with the Kir6. x subunit and reverse its inhibitory effects. 137 Therefore, despite having limited (if any) catalytic ability, NBD1 is required in a specific ATP-bound conformation in order to participate in channel activation. A model described by Ueda et al depicting the cooperativity between NBD1 and NBD2 is illustrated in Figure 7.1. 306

Figure 7.1: Model of cooperativity of the SUR NBDs. The red squares represent the NBDs in the apo state which result in channel closure. MgATP binding at NBD1 results in the "signalling competent state" (blue circles) and stabilizes MgADP binding at NBD2 (green circles). The magnesium ions are not shown. Sulfonylurea binding, such as glibenclamide (black circle), can displace MgATP binding at NBD1 resulting in channel closure, despite MgADP binding to NBD2. Our data indicate that phosphorylation, mutations and drug binding affect NBD1 conformation and alter equilibrium ( 1) This scheme is based on the model of reference 306 .

157

In support of this model, our data suggest NBD1 likely samples an equilibrium of different conformations with nucleotide binding, phosphorylation and mutations all affecting the population of the active NBD1 conformer (Equilibrium 1 in Figure 7.1). The binding of nucleotide (MgATP > MgADP > MgAMP) likely promotes stabilization of the NBD1 conformation that promotes opening of the KATP channel. This active state for NBD1 has previously been referred to as the "signaling-competent state". 121, 344 This state likely allows NBD1 to make productive interactions with the rest of the channel to communicate its role in channel opening. Our data indicate that the presence of different regions from NBD1 can alter the equilibrium of states. For example, exon 14 which is found in a disordered N-terminal tail of NBD1 sterically impedes binding of MgATP to NBD1. This reduces the population of the signaling-competent state. Phosphorylation of NBD1 displaces the N-tail from the core of the protein and thereby increases the population of the NBD1 active state. The increase in the NBD1-active state population cooperatively allows for NBD1/NBD2 dimerization which promotes NBD2 to adopt the necessary structural conformation for either binding MgADP or hydrolysis of nucleotide. Removal of exon 17 (which also contains the disease causing mutation, V734I) reduces sensitivity of NBD1 to nucleotide. This likely results in a lower population of NBD1 active states. Our data also suggest the potential for cross-talk between both exon 14 and exon 17 in SUR2A NBD1.

The biological relevance of the effects of each of these exon removals is displayed by the tissue specific splicing of exon 14 or exon 17. Based on the requirements of the cell, different splice isoforms can be generated which will have different activities. This essentially allows the membrane potential of the cell to be altered according to the specific needs of the cell. Thus, alternative splicing of the K ATP channel layers another level of regulation by carefully modulating the equilibrium of SUR2A NBD1 "signaling competent states" and therefore the activity of KATP channels.

7.3 Future Directions

7.3.1 Intramolecular interactions within NBD1

We have provided evidence that SUR2A contains a number of regulatory regions within NBD1, including residues from exon 14 and exon 17. The significance of these regions is likely to modulate the affinity of NBD1 for nucleotide and therefore fine tune K ATP channel gating and

158

Figure 7.2: The ED domain in SUR1 and SUR2. A long stretch of acidic residues are observed following the C- terminus of NBD1. These residues are thought to function in an analogous manner to the R region of CFTR. Adapted by permission from J. Gen. Physiol. Karger et al Mar; 131 (3);185-96, copyright 2008. (Reference 148 ). ultimately the membrane potential. There are other regions within NBD1 that are likely to be involved in regulation of the protein. One such region involves a highly charged and acidic stretch of residues, termed the ED domain 148 (Figure 7.2). This region is analogous the R domain in the ABCC subfamily member, CFTR. 148 The R region is a disordered region in CFTR that is extensively phosphorylated which alters its interactions with CFTR NBD1.329 Although the ED domain has not been demonstrated to contain any phosphorylation sites, the negative charges likely mimic the function of the R domain. This is highly plausible since the N-tail in SUR2A NBD1 (S615 - E664) is analogous to the RI region of CFTR (Chapter 4). When the ED domain is mutated to neutral residues in the full channel, sensitivity to both MgADP and sulfonylureas is removed. 148 Our preliminary studies with NBD1 (that has been extended to include the ED domain, S615-K972) suggest a change in protein dynamics as well as an increase in thermodynamic stability. NBD1 (S615-K972) also displays an increase in thermodynamic stability of the isolated domain, but there is also significant broadening of NMR spectra (de Araujo, Alvarez & Kanelis, unpublished). This suggests the ED domain also alters protein dynamics of NBD1. Additional experiments to assess changes in nucleotide affinity and drug binding capacity of NBD1 with the ED domain would be beneficial to understanding the biochemical effects of the increase in the length of the C-terminal tail.

7.3.2 Intermolecular interactions of NBD1 with other domains

As previously described in Chapter 1, there are several interdomain interactions between NBD1 and several other components of the K ATP channel. The NBDs are important for regulation and need to communicate conformational changes to other parts of the channel including the L0 linker and MSD0 domain. Gating studies have suggested a critical role for these domains in channel activity and modeling has suggested important interactions between NBD1 and the L0 linker. 20, 85 To date, there is very limited structural information available on the additional MSD0 subunit as well as the cytoplasmic L0 linker. Studies into the L0 linker would be useful to

159 identify any residual secondary structure. Interactions of the L0 linker with NBD1 can be probed through binding experiments via NMR spectroscopy similar to our drug binding studies. As noted in Chapter 6, all of these experiments will be enhanced by interaction studies with NBD2, although this requires obtaining soluble and monomeric NBD2. Various strategies have been discussed in Chapter 6 that may allow fusion proteins of NBD2 to be tested with monomeric NBD1 by NMR.

In addition to domains of the K ATP channel, NBD1 is also thought to perform a number of interactions with other accessory proteins. Recently, an interaction between Syntaxin 1A and the SUR NBDs was observed through pull down experiments. 146, 147 The interaction sites of these proteins can be probed by NMR. Preliminary interaction studies have been performed with GST- fusion proteins of different domains of syntaxin1A proteins and SUR2A NBD1. Although we have observed some changes in NBD1, these experiments require control studies of NBD1 with the GST-tag alone (Albanese, de Araujo, Staglijar & Kanelis, unpublished). Further, this interaction was described to be pH dependent. 35 pH titrations with NBD1 and need to be repeated in the presence of syntaxin 1A. These experiments can provide insights into molecular interactions of syntaxin 1A with SUR2A NBD1.

7.3.3 Drug Binding Analysis of NBD1

We have previously investigated the interactions of NBD1 with potassium channel openers (Chapter 8). Likely, other such drugs also possess binding sites that are partially or entirely at the

Figure 7.3: 19 F-NMR of Fluoro-Trp labelled NBD1-ΔN (D665 - L933). (A) The 19 F spectra of fluoro-Trp labelled NBD1 (250 μM ) in black. The predicted deconvolution of the individual peaks in the spectra are shown in cyan with the assignment of each peak. (B) shows how the assignment was determined based on individual mutagenesis of the five Trp residues. JPL-A performed all the experiments to obtain the assignment of the fluorine spectra. ED purified fluoro-Trp labelled wild type NBD1. J.P.L.-A. and E.D. set up the 19 F NMR experiments with the initial assistance of Dr. Sacha Larda and Tae Hun Kim.

160

NBD1 interface. Some of these drugs are also thought to compete with the ATP binding site of NBD1. 35, 94 These pharmaceutical compounds can be screened through both NMR and fluorescence as previously accomplished for pinacidil 189 (Lopez-Alonso, de Araujo & Kanelis, unpublished). We have also employed an alternative technique, 19 F-NMR, for probing the interactions with NBD1. For these experiments, NBD1-ΔN has five Trp residues that are 19 F labelled at one position. Fluorine NMR is useful since the probe is a single atom which generally results in minimal perturbation to the sites of interest (for example, in comparison to labelling the protein with a fluorophore). 46 Fluorine is also not biologically present, which essentially removes any background signals. 196 This strategy can be useful in localizing the binding sites for drug interactions by simplifying the spectra to only five peaks/sites in the protein. Furthermore, fluorine is significantly more electronegative than proton making the probes more sensitive to changes in the chemical environment. 196 The 19 F-NMR spectrum of SUR2A NBD1-ΔN (labelled with 19 F-Trp) is shown in Figure 7.3 with the assignments to each of the individual fluorine Trps. Various drugs can also be screened with fluorine NMR and monitoring their effect on peak position.

7.3.4 Full K ATP channel structural studies

Electrophysiology and biochemical data of the K ATP channel would be greatly enhanced by additional structural information. Currently, there are low resolution (18 Å) 202 EM images for the 93 full K ATP channel as well as 21 Å resolution EM images of tetrameric SUR2B. Cryo-EM is an ideal tool for determining the structure of the full channel, as new developments in detectors have allowed protein structures to be determined with extremely high levels of resolution (~3 Å). 23, 76 Further, low concentrations of protein are required, unlike the sample constraints for X- ray crystallography or NMR studies. The full channel is approximately 950 kDa which is in the range of size limits for cryo-EM studies. 40 The full channel can be expressed and purified using different methods, such as the cookie-cutter approach using membrane encapsulating polymers. 111 These polymers consist of styrene maleic acid molecules that insert into the membrane and arrange into small discs. Some of these discs will encapsulate the tagged-protein of interest, which can then be purified using affinity chromatography. 111 The utility of this technique is that it does not involve the use of detergents. Membrane proteins are purified in their native lipid environment which is likely the reason that the purified proteins appear to have a greater stability and purity. 111, 136, 230 These experiments will provide important structural

161 information that will assist in determining the complex mechanisms involved in regulation of the

KATP channel.

Overall, this work has provided insights into the molecular basis of regulation of SUR2A NBD1 as well as strategies for purifying and studying isolated nucleotide binding domains. The experiments which are proposed here, build on these studies. Structural data on the K ATP channel will also provide a macroscopic view of regulation for the full channel.

162

Chapter 8 Appendix Drug binding to SUR2A NBD1 8 Overview

This chapter describes evidence for a direct and specific interaction of the KATP channel opener, pinacidil, with SUR2A NBD1. This chapter provides NMR studies of SUR2A NBD1 that demonstrate, for the first time, a direct and specific interaction of the drug pinacidil with the protein. Interactions of SUR2A NBD1 with diazoxide, a functionally related but structurally distinct drug, were not observed under the same conditions, demonstrating a specificity of SUR2A NBD1 for drugs. Further, fluorescent ATP binding data show that pinacidil promotes a higher affinity interaction between SUR2A NBD1 and nucleotide, yielding insights into how the 6 drug may activate cardiac K ATP channels. This chapter was published in Biochemistry and is reproduced with permission from the publisher. The text and supplementary information and figures have been slightly modified. Author Contributions: EDA expressed and purified all proteins. EDA and JPL-A performed the fluorescence and NMR experiments. JPL-A performed all NMR drug titrations. JPL-A and VK analyzed the NMR data. VK performed all homology modeling. VK wrote the paper, and VK, JPL-A and EDA edited the manuscript.

8.1 Introduction 268 The SUR proteins are known sites of binding of K ATP channel openers (KCOs). These structurally diverse drugs are used in the treatment of disorders such as hypertension, angina, and ischemic heart disease, and are also used for cardioprotection in surgery.104, 135, 236 Figure 8.1 shows the structures of two commonly studied KCOs, pinacidil and diazoxide. K ATP channels from different tissues display varying sensitivities to these KCOs and the differential responses 14, 209 of K ATP channels to the drugs are attributed to the specific SUR protein. For example, pancreatic K ATP channels, which contain four copies of SUR1, are activated by diazoxide and to 11, 89, 102 a small extent by pinacidil. In contrast, the SUR2A-containing cardiac K ATP channels are activated by pinacidil and generally not by diazoxide,89, 102, 134, 298, 328 except under specific

6 Lopez-Alonso, J.P., de Araujo E.D. , and Kanelis V. (2012) NMR and fluorescence studies of drug binding to the first nucleotide binding domain of SUR2A. Biochemistry . 51:9211-9222. (Reprinted with permission from Elsevier).

163

conditions (see below). Finally, K ATP channels in smooth muscle, which contain SUR2B, are activated by both pinacidil and diazoxide.134, 242 In addition to differences in activation by different KCOs, specific SUR proteins in K ATP channels differ in the degree to which their response to the drugs is modulated by nucleotides. For example, diazoxide activates SUR1- and

SUR2B-containing K ATP channels in the presence of both MgATP and MgADP, while only

MgADP and not MgATP supports diazoxide activation of SUR2A-containing cardiac K ATP channels.57

The differential activation of K ATP channels containing either SUR1 or SUR2 isoforms by pinacidil and diazoxide has been exploited to identify regions of the SUR proteins involved in KCO binding and/or mediation of the effects of the drugs. Binding studies using 3H-P1075, an analogue of pinacidil, and electrophysiological experiments on SUR1-SUR2A chimeras originally localized the pinacidil binding site to a segment of SUR2A comprised of transmembrane (TM) helices 12-17.19 Additional 3H-P1075 binding and channel activation studies narrowed the site of action to residues T1059-L1087 (rat SUR2A numbering) and R1218-

A B

C

Figure 8.1 Models/structures of SUR and Kir6. x proteins, and common K ATP channel openers. (A) Ribbon diagram of the homology model of SUR2A with NBD1 and NBD2 colored in blue and green, respectively. The MSDs are primarily colored in gray, with the exception of part of the cytoplasmic loop between TM helices 13 and 14, that includes coupling helix 3 and is important for pinacidil binding and activation of K ATP channels, which is magenta. The MSD2-NBD2 linker and residues L1249 and T1253 in TM helix 17, also important for K ATP channel activation, are in cyan. The C42 region in NBD2 is indicated. A solid black circle shows the location of a possible drug binding interface. (B and C) Chemical structures of the K ATP channel openers (B) pinacidil and (C) diazoxide.

164

N1320.307 Residues T1059-L1087 encompass the cytoplasmic loop between TM helices 13 and 14 (Figure 8.1A, magenta) and include coupling helix 3 that contacts both NBD1 and NBD2. Residues R1218-N1320 form TM helix 17 and the MSD2-NBD2 linker (Figure 8.1A, cyan), plus the first b -strand of NBD2 (Figure 8.1A, green). Channel activation studies using mutations in which L1249 and T1253 in TM helix 17 in SUR2A were changed to Thr and Met, the corresponding residues in SUR1, respectively, identified these residues as critical for KCO- 208 mediated activation of K ATP channels. Thus from these data, two different regions in the SUR protein appear to be involved in KCO binding and channel activation. The C-terminal 42 residues in the SUR protein, known as the C42 region, define a third region as being involved in KCO binding and KCO-mediated activation. K ATP channels containing either SUR2A or SUR2B, which differ only in the composition of the C42 region but are otherwise identical, display different affinities for various KCOs, with SUR2B-containing channels having a higher affinity for the drugs than SUR2A-containing channels.268 Further, a chimera comprising the SUR2 core structure and the C42 region of SUR1, which is similar in sequence to that of SUR2B, binds P1075, pinacidil, and diazoxide with an affinity similar to SUR2B, consistent with the involvement of the C42 region in drug binding.268 The C42 region is part of the NBD2 domain 64 and homology models of NBD2 of SUR2A and SUR2B NBD2,200 based on the structure of the histidine permease NBD,130 suggest interactions of the C42 region with the ATP-binding site in NBD2. Homology models of the minimum ABC structure of SUR2A (Figure 8.1A), as well as structures of ABC transporters Sav1866,62 MsbA,325 and murine P-glycoprotein,5 which were not available at the time the C42 experiments were done, demonstrate that the C42 region is not located near the TM helices or coupling helix 3. However, the C42 region is at the NBD1/NBD2 dimer interface and hence may affect drug binding because of its influence on NBD1/NBD2 dimerization. The apparent discrepancy resulting from identification of three distant regions in SUR2A in pinacidil binding and activation may be due to pinacidil inducing structural changes in the protein, which are in turn necessary for channel activation. Some of the regions identified by mutation studies may be involved in drug binding while other regions are necessary for transmitting conformational changes through the protein. The KCO binding and channel activation data, as well as structural models of SUR2A, led us to test a model 14 that, along with the previously identified regions that include the coupling helices and the MSD2-NBD2 linker, the NBDs also form part of the KCO binding sites. Thus, we have probed binding of two KCOs, pinacidil and diazoxide, to SUR2A NBD1, which along

165 with coupling helix 3 and NBD2, may be involved in creating a drug binding site. These studies were enabled by the finding that SUR2A NBD1 could be expressed as a soluble and folded domain,64 while SUR2A NBD2 could not. In this paper, we demonstrate direct and specific binding of pinacidil to SUR2A NBD1 in the presence of MgATP using nuclear magnetic resonance (NMR) spectroscopy and use intrinsic Trp fluorescence to determine the affinity of the drug-NBD1 interaction. The pinacidil-NBD1 interaction is weak, having a K d value of 455 ± 37 m M, which is consistent with the hypothesis that NBD1 forms part of the drug binding site and the demonstration of the involvement of other regions of the SUR proteins in previous studies.19, 208, 268, 307 In contrast, interactions between SUR2A NBD1 and diazoxide were not observed under the same conditions. Using fluorescence nucleotide binding studies, we demonstrate that pinacidil increases the nucleotide binding affinity of SUR2A NBD1, providing further evidence for a direct interaction between pinacidil and NBD1. NMR relaxation experiments and size exclusion chromatography demonstrate that SUR2A NBD1 is monomeric under the conditions used in the drug binding studies. Together, these experiments identify an additional component of the binding site for pinacidil on SUR2A, which is likely at an interface between coupling helix

3 and NBD1 that forms part of the drug binding site in the intact K ATP channel. The studies also provide a methodological approach for testing binding of drugs to the NBDs of the SUR proteins.

8.2 Methods

8.2.1 Materials

The K ATP channel openers (KCOs) pinacidil and diazoxide were purchased from Sigma. The KCOs purchased were ≥ 98 % pure according the manufacturer.

8.2.2 Protein Expression and Purification

NBD1 from rat SUR2A (residues S615-L933) was prepared as previously described.64 The sequence of rat SUR2A is ~96 % identical to human SUR2A in this region, with most amino acid changes occurring in unstructured loops.64 Thus, data acquired on rat SUR2A NBD1 are applicable to the human protein.

166

8.2.3 Homology Models

Homology models of rat SUR2A (Figure 8.1A) were generated with the program Modeller.259 A structure-based sequence alignment of the C-family of ABC transporters,72 including isoforms of SUR2,49, 50 based on available structural data of full 5, 62, 325 and half transporters (PDB code 4AYT) available was generated using Clustal W.168, 300 Structural information from the crystal structure of CFTR NBD1, which is similar to SUR2A NBD1 64 was also used in the alignment. The homology models were generated using the crystal structure of Sav1866 (PDB code 2HYD) 62 and the human CFTR F508A NBD1-RE (PDB code 1XMI),176 excluding residues C647– Asp673, as templates. A total of 50 models were generated, from which the 10 lowest energy models were selected for analysis.

8.2.4 Drug Binding Studies by NMR Spectroscopy

Interaction of SUR2A NBD1 with K ATP channel openers (KCOs) was monitored with NMR spectra of SUR2A NBD1. The purified SUR2A NBD1 was dialyzed into NBD1 buffer with 5 mM ATP and 5 mM MgCl 2. The NMR samples also contained 0.5 mM 4,4-dimethyl-4- silapentane-1-sulfonic acid (DSS) for 1H and 15 N chemical shift referencing.330 NMR drug binding studies were done with SUR2A NBD1 at a concentration of 250 m M. TROSY-HSQC spectra 232 of SUR2A NBD1 were recorded at 30°C on a 600 MHz Varian Inova spectrometer equipped with a H(F)CN triple resonance cryoprobe and actively shielded z-gradients. The KCO drugs were solubilized in DMSO. Sequential additions of the drugs were made to the SUR2A NBD1 sample and NMR spectra were recorded. Three separate titrations were done with pinacidil and two with diazoxide, using different preparations of the protein. To account for changes in spectra of SUR2A NBD1 from addition of increasing amounts of DMSO, blank titrations of SUR2A NBD1 in which identical volumes of DMSO without the drug were recorded. Thus, the spectra of SUR2A NBD1 with and without drug shown have identical amounts of DMSO. All NMR data were processed using NMRPipe/NMRDraw 75 and analyzed using NMRView.139 Spectra with and without the drug were compared by calculating the combined chemical shift difference, Dd total , according to the Equation 2.2.

We also recorded 1D-1H spectra for each titration point in order to experimentally determine the drug and DMSO concentrations in solution at each point in the titration.229

167

Addition of drug to the protein sample results in a dilution of the DMSO concentration, which compromises the solubility of the drug. Thus, the amount of drug in solution during the titration may not be equal to the amount of drug added due to precipitation. For pinacidil, we used resonances from the tertbutyl (0.93 ppm), methyl (1.18 ppm), and aromatic (7.19 and 8.35 ppm) protons. These signals were integrated and divided by the total number of protons (signal:proton ratio) that give rise to the signals. For diazoxide, we used resonances from the methyl (2.39 ppm), and aromatic (7.36, 7.73, 7.96 ppm) protons. For DMSO, we used the resonance from the methyl protons (2.72 ppm). The signal:proton ratio for each drug was compared to that for glycerol, for which we determine the concentration based on our starting concentration and the dilution in the titration, to obtain values for the concentration of drug and DMSO in our NMR titration samples. This method allows for determination of the free drug in solution, as resonances for drug bound to the protein would be broadened and/or overlapped with the protein resonances, and hence not observable in the 1D-1H spectrum. The free drug concentrations at the end of the titration were determined to be in 10- to 15-fold molar excess compared with the protein concentration. Thus, the total concentration of drug is underestimated by < 10 %, even if we were to consider saturation of the protein with drug. We also prepared standard samples of pinacidil and diazoxide, in which there was no precipitation of the drug, and verified the drug and DMSO concentrations by the same method. These calibration standards also verified the drug and DMSO concentrations in our titration samples. The 1D spectra were processed and analyzed with the ACD/NMR Processor (Academic Edition, version 12.1) software package (www.acdlabs.com).

8.2.5 Intrinsic Tryptophan Fluorescence Experiments

Kd values for the interaction of pinacidil and NBD1 in the presence of MgATP were determined using intrinsic Trp fluorescence of the NBD1. Fluorescence experiments were conducted on a Fluoromax-4 spectrofluorimeter equipped with an automatic titrator and a Peltier unit for precise temperature control. Experiments were conducted at 15°C, with excitation and emission wavelengths of 298 and 348 nm, respectively, and slit widths of 1.5 and 6 nm, respectively. The excitation wavelength of 298 nm was chosen to selectively excite Trp residues in the protein and the emission wavelength of 348 nm corresponds to the wavelength where the fluorescence difference of NBD1 in absence and presence of pinacidil NBD1 is at a maximum. The affinity

168 between NBD1 and pinacidil was measured using 1 m M protein in NBD1 buffer containing 10 % DMSO to ensure solubility of the drug at the highest concentrations used in the experiment (~ 1 mM). The titration data were fit to Equation 8.1 317 , assuming a 1:1 complex between NBD1 and the drug.

(I¥ - I ) 2  I =I - o {()[KCO ]+[NBD1 ]+K - ()[KCO ]+[NBD1 ]+K - 4() [KCO ] ()[NBD1 ]  o total total d total total d total total  2[NBD1total] (Eq 8.1)

where I is the fluorescence intensity ratio at a given total concentration of pinacidil, [KCO total ], I∞ is the fluorescence intensity ratio at saturation, I o is the fluorescence intensity ratio in absence of the drug, K d is the dissociation constant, and [NBD1 total ] is the total concentration of SUR2A NBD1 in the reaction.

8.2.6 NMR Relaxation Experiments

15 N R 1ρ relaxation experiments were performed at 30 °C on a Varian Inova 600 MHz spectrometer equipped with an HCN triple-resonance cryoprobe with actively shielded z- 86, 157 15 gradients using previously published pulse schemes. N R 1ρ values were measured from six different spectra recorded with delays of 2, 4, 8.5, 14, 21, and 30 ms for a sample of 250 m M NBD1 in the presence of saturating concentrations of MgATP. Relaxation experiments were not recorded in the presence of pinacidil due to limited solubility of the drug over the time required 15 for the acquiring the relaxation data (~2 days). N R 2 values for each residue were obtained by correction of the observed relaxation rate R 1ρ for the offset Δν of the applied spin-lock rf field

(ν 1) to the resonance using the relation 2 R1ρ = R2 sin θ, (Eq 8.2) −1 where θ = tan (ν 1/Δν) and ν 1 was 1824 Hz. All data sets were processed using NMRPipe. Peak intensities were obtained using the Rate Analysis tool in NMRView and used to fit a two e−t*R2 88 parameter function of the form I(t)=I 0 using a Matlab script. Errors in relaxation rates were estimated by Monte Carlo analysis. In total, 24 resolved peaks were analyzed.

8.2.7 Fluorescence Nucleotide Binding Experiments

Kd values for the interaction between SUR2A NBD1 (S615-L933) and the fluorescent ATP analogue 2',3'-O-(2,4,6-trinitrophenyl)- adenosine-5'-triphosphate (TNP-ATP, Molecular Probes) were determined using fluorescence spectroscopy. These studies required removal of ATP

169 present in the NBD1 samples from the purification. Mg 2+ and ATP were removed from NBD1 samples by size exclusion chromatography and replaced with known concentrations (2.5 m M) of

MgCl 2 and TNP-ATP. SUR2A NBD1 elutes from the size exclusion column (Superdex 75, GE Healthcare) at 10.7 ml, which is consistent with momomeric NBD1.64 Binding experiments were performed by serial dilutions of the protein at a constant concentration of the fluorescent TNP- ATP. An initial sample containing 50-70 m M NBD1 (depending on the concentration of apo

NBD1 that eluted from the size exclusion column), 2.5 m M TNP-ATP and 2.5 m M MgCl 2 in the NBD1 buffer with 10 % (v/v) glycerol was generated. Samples with lower NBD1 concentrations were made by serial dilutions into buffer lacking the protein, while maintaining constant MgCl 2 and TNP-ATP concentrations. Thus samples ranged in protein concentration from 0.1 – 1 m M for the lowest NBD1 concentration to 50-70 m M NBD1, with constant Mg 2+ and TNP-ATP concentrations of 2.5 m M. A separate sample was generated containing MgCl 2, TNP-ATP and buffer only for the 0 m M NBD1 sample. Due to lack of stability and solubility of apo NBD1, binding experiments were performed at 15 °C. Fluorescence spectra of TNP-ATP were recorded immediately after each sample was generated using an excitation wavelength of 465 nm and a slit width of 5 nm. Emission spectra were recorded from 485 nm – 600 nm with slit widths of 7 nm. The K d value for the NBD1-nucleotide complex was determined by monitoring the ratio between the fluorescence intensity at 533 nm, which corresponds to the wavelength where fluorescence difference of free and bound TNP-ATP is at a maximum, and 600 nm to account for any non-specific fluorescence from the protein.223 The titration data were fit to the Equation 2.2. This equation assumes a 1:1 complex of NBD1 with TNP-ATP.110, 317

We kept the TNP-ATP concentrations constant,110, 312 rather than the protein concentration as done in other studies 183, 244 because of the lack of stability of apo SUR2A NBD1 over time. By employing serial dilutions, we measured the fluorescence of samples with excess NBD1 before any precipitation occurred. To determine the effects of pinacidil and diazoxide on nucleotide binding, the TNP-ATP binding experiments were performed in the presence of 12.5 m M pinacidil or diazoxide. A concentration of 12.5 m M of drugs was chosen, rather than 50 m M, which is equal to the highest concentration of NBD1 in the titration, because pinacidil has a weak fluorescent signal when excited at 465 nm. The fluorescence of pinacidil at 12.5 m M is negligible.

170

8.3 Results

8.3.1 NMR titration studies indicate that NBD1 of rSUR2A mediates specific interactions with pinacidil

15 N TROSY-HSQC spectra of SUR2A NBD1 were recorded in the absence and presence of the drugs pinacidil (Figure 8.2) and diazoxide (Figure 8.3). As explained in the methods section 8.2.4, and as seen with other drugs used in NMR binding studies,128 pinacidil and diazoxide have poor solubility in the aqueous NBD1 buffer and hence are dissolved in DMSO. Thus, blank titrations of SUR2A NBD1 using only DMSO were also performed to account for changes in the SUR2A NBD1 spectra with increasing concentrations of DMSO (Figure 8.4). Figures 8.2 and 8.3 comparing spectra in absence and presence of drugs were recorded with identical concentrations

Addition of pinacidil to SUR2A NBD1 results in thirteen distinct chemical shift changes throughout spectra of SUR2A NBD1 (Figure 8.2), indicating that the KCO pinacidil interacts with SUR2A NBD1. The number and magnitude of the chemical shift changes are consistent with spectral changes seen upon binding of a small molecule drug to NBD1 of CFTR using similar concentrations of protein and drug.128 Only resonances exhibiting a significant combined chemical shift difference, which is greater than the average of all Dd total values plus one standard

143 deviation, Dd total ‡ 0.04, are considered. Resonances displaying changes in intensity are not considered when analyzing spectral changes because decreases in intensity may be due to small amounts of protein precipitation during the titration. Further, overlapping resonances were also not considered in the analysis. The small number of chemical shift changes observed in NBD1 upon addition of pinacidil is expected due to the small size of the drug molecule and indicates that the drug interacts with specific residues in SUR2A NBD1.of DMSO (v/v, 7.5 %).

171

A

B C D

E i ii

iii

Figure 8.2: Saturation of pinacidil binding to NBD1 in the presence of MgATP. (A) Comparison of 2D 15 N- 1H TROSY-HSQC spectra of SUR2A NBD1 (250 µM) in the presence of increasing concentrations of pinacidil. The spectra were recorded at 30 oC at 600 MHz with 5 mM Mg 2+ and 5 mM ATP in 20 mM Na + phosphate pH 7.3, 150 mM NaCl, 5 mM DTT, 2% (v/v) glycerol, 7.5% (v/v) DMSO, 10% (v/v) D 2O. Chemical shifts for each spectrum were referenced to DSS. The spectrum of SUR2A NBD1 in the absence of pinacidil is in the foreground with resonances coloured in black. The spectrum of NBD1 with 2.4 mM pinacidil is in the middle with resonances coloured blue, whereas that of NBD1 with 3.9 mM pinacidil is in the background in red. Green circles highlight specific chemical shift changes in NBD1 upon addition of pinacidil. (B-D) Selected regions of the spectrum shown in (A). As with panel A, green circles highlight chemical shift changes observed with increasing concentrations of pinacidil. Spectra of NBD1 with 2.4 mM and 3.9 mM pinacidil added to the sample are very similar, with none of the resonances displaying significant chemical shift changes (Δδ total < 0.03 for all resonances), indicating that the saturation point in the titration has been reached. The asterisks (“*”) identify resonances from disordered regions in the protein. (E) The full 1D-1H NMR spectrum of 1.4 mM pinacidil in the NBD1 buffer with 1.4 % (v/v) DMSO is shown in (i). The structure of pinacidil is shown in the top left hand corner of the spectrum, with selected protons numbered in the structure and the corresponding resonances numbered in the spectrum. The chemical shift for the methyne proton indicated by number 5 is at ~4 ppm, and hence this resonance is overlapped with the glycerol resonances. Resonances corresponding to water, glycerol, DMSO, and DSS are also labelled. (ii) and (iii) Selected regions of the 1D-1H NMR spectrum of pinacidil are displayed with peak integration used to determine the concentration of the drug in solution. Because the intensity of DSS and water resonances are similar in both datasets, the two datasets can be compared in terms of signal:noise. The glycerol and DMSO resonances are very intense and much higher than the water and DSS resonances. Chemical shifts for each spectrum were referenced to DSS.

172

A

B C D

E i ii

iii

Figure 8.3 Diazoxide does not interact with SUR2A NBD1 in the presence of MgATP. (A) Comparison of 2D 15 N−1H TROSY-HSQC48 spectra of SUR2A NBD1 (250 μM) in the absence and presence of 3.8 mM diazoxide at 30 °C at 600 MHz. The solution conditions for each sample are identical to those described in the legend to Figure 8.2. The spectrum of SUR2A NBD1 in the absence of diazoxide is in the foreground with resonances colored in black, while that of NBD1 with diazoxide is in the background with resonances colored red. Spectra of SUR2A NBD1 with and without diazoxide are virtually identical. Minor chemical shift changes, which are seen for only two resonances and only with high (3.8 mM) concentrations of diazoxide, are circled (blue, solid). Dashed blue circles highlight resonances for which significant chemical shift changes are seen with pinacidil addition (Figure 8.2), but which are absent in the diazoxide titration. (B−D) Selected regions of the spectrum shown in panel A. The resonance circled in panel D by a solid blue circle shows small chemical shift changes (Δδtotal < 0.03). The asterisks (“*”) identify resonances from disordered regions in the protein. (E) The full 1D-1 H NMR spectrum of 0.7 mM diazoxide in the NBD1 buffer with 0.7 % (v/v) DMSO is shown in (i). The structure of diazoxide is shown in the top left hand corner of the spectrum, with selected protons numbered in the structure and the corresponding resonances numbered in the spectrum. (ii) and (iii) Selected regions of the 1D- 1H NMR spectrum of diazoxide are displayed with peak integration used to determine the concentration of diazoxide and DMSO in solution.

Notably, all of the observed chemical shift changes occurred for resonances from structured residues in the protein. We do not observe chemical shift changes in the intense resonances centered about 8.2 ppm in the 1H dimension (Figure 8.2C, D; marked with an “*”) that result from disordered regions in SUR2A NBD1, such as the disordered region of the b -

173 sheet subdomain insert,64 indicating that these regions in SUR2A NBD1 do not interact with pinacidil. Of the thirteen resonances that exhibit chemical shift changes upon pinacidil addition, nine resonances have 1H chemical shifts of less than or equal to 7.7 ppm (i.e. 7.7, 7.5, and 7.3 ppm) and greater than 8.5 ppm (i.e. 8.6, 9.0, 9.3, 9.5, and two peaks at 8.7 ppm). Additional chemical shifts are seen for residues with 1H chemical shifts of 7.8, 7.9, 8.0 and 8.3 ppm. The lower intensity of these resonances (Figures 8.2C, D; highlighted by cyan circles) compared to the intense resonances described above (Figures 8.2C, D; marked with an “*”) suggest that they are also from structured residues in the protein.

Although Figure 8.2 shows HSQC spectra taken with 0, 2.4 and 3.9 mM pinacidil (with DMSO concentrations (v/v) upto 7.5 %), chemical shift changes were observed with as little as 1.5 mM pinacidil added. These drug concentrations are of the same order of magnitude as the concentrations used to observe pinacidil-mediated activation of K ATP channels in some

A

B C D

Figure 8.4: DMSO-mediated changes in NMR spectra of SUR2A NBD1. Comparison of 2D 15N-1 H TROSY- HSQC (6) spectra of SUR2A NBD1 (250 µM) in the absence and presence of 7.5 % (v/v) DMSO with 5 mM Mg2+ and 5 mM ATP in 20 mM Na phosphate, pH 7.3, 150 mM NaCl, 5 mM DTT, 2 % (v/v) glycerol, 10 % (v/v) D2O at 30 °C at 600 MHz. Chemical shifts for each spectrum were referenced to DSS (5). The spectrum of SUR2A NBD1 in absence of DMSO is in the background with resonances coloured in magenta, 2 whereas the spectrum of NBD1 with 7.5% (v/v) DMSO is in the foreground with resonances coloured in black. (A) DMSO causes chemical shift changes throughout the spectrum, but does not disrupt the structure of NBD1. Spectra of SUR2A NBD1 in the presence of DMSO show dispersion of backbone resonances from ~6.5 to ~9.5 in 1 H dimension, which is characteristic for a folded protein. Further, spectra of SUR2A NBD1 in the absence and presence of 7.5 % (v/v) DMSO are similar. Minimal changes in chemical shifts and peak intensities are observed in spectra of SUR2A NBD1 with concentrations of DMSO as high as 11 % (v/v) compared to 7.5 % DMSO. Dashed blue circles highlight resonances for which significant chemical shift changes are seen with pinacidil addition (Figure 2). (B-D) Selected regions of the spectrum shown in panel A. Interestingly, resonances arising from disordered residues do change in position with DMSO (panels C and D, highlighted by an asterisk, ” * ”). Changes in solution conditions can have significant effects on resonances from disordered proteins or from regions of folded proteins with significant disorder (7). Thus, chemical shift changes for these residues in NBD1 are consistent with their disordered character and exposure to the solvent.

174 electrophysiological studies 57, 78, 208 and were necessary here due to the high protein concentration needed for NMR spectroscopy. Spectra of NBD1 are similar with 2.4 mM and 3.9 mM pinacidil added to the sample Figure 8.2). None of the resonances display significant chemical shift changes ( Dd total < 0.03 ppm for all resonances) between the two spectra, indicating that 2.4 mM pinacidil is sufficient to reach saturation in the titration. Identification of the specific residues in SUR2A NBD1 that interact with pinacidil would require resonance assignment of the protein, which in turn requires concentrated samples of NBD1 (≥ 0.5 mM) to be stable for many days (> 14 days) at 30 °C. We are currently developing strategies to allow for this series of extensive and involved experiments.

In contrast to pinacidil, diazoxide does not bind SUR2A NBD1, or at least binds too weakly to be detected by these experiments. Spectra of 250 m M SUR2A NBD1 in the absence and presence of 3.8 mM diazoxide, a 15-fold excess of the compound, are virtually identical (Figure 8.3). Only two resonances, out of 260 total, displayed a significant, but small difference in chemical shift ( Dd total = 0.04) (Figure 8.3A, circled). There is also another resonance that may change with diazoxide addition (Figure 8.3D, circled), but the Dd total value calculated for this resonance is not statistically significant. In comparison, addition of pinacidil results in thirteen resonances with significant Dd total values of 0.04 – 0.10. Many of the resonances that display chemical shift changes with pinacidil addition do not change with addition of diazoxide (Figure 8.2, solid circles compared to Figure 8.3, dashed circles). Thus, NMR data indicate that pinacidil makes specific interactions with NBD1 while diazoxide does not interact with NBD1 or interacts with a much lower affinity than pinacidil. Spectra with higher concentrations of diazoxide to detect lower affinity interactions could not be recorded due to the solubility of the compound.

In the intact SUR2A protein, as with other ABC transporters, NBD1 and NBD2 form a heterodimer upon binding of MgATP.117 We have recorded 15 N TROSY-HSQC spectra of SUR2A NBD1 at various concentrations (from 50 m M > 300 m M). SUR2A NBD1 elutes from the 24 mL Superdex 75 size exclusion chromatograph column at a volume of 10.7 ml, consistent with it being monomeric. Spectra of SUR2A NBD1 samples at 50 m M concentration, which are taken directly from the eluent (at 10.7 ml) of the size exclusion column without concentration of the sample further, are identical in peak positions and relative intensities, also indicate that the samples of NBD1 up to 300 m M are monomeric.64

175

The absence of dimerization of SUR2A NBD1 in the presence of MgATP is consistent with observations of NBDs from other eukaryotic ABC transporters, in which NBD1 and NBD2 are not identical, that demonstrate monomeric behavior of NBD1 even at the high concentrations used in structural studies.143, 175, 176, 248

Further confirmation of monomeric SUR2A NBD1 in our samples comes from NMR 15 N 15 R2 relaxation studies. N R 2 rates were calculated for 24 resonances, with resonances selected from regions of the spectrum that are derived from structured regions of the protein, such as those with 1H chemical shifts of > 8.5ppm. Resonances with 1H chemical shifts of 7.7-8.5 ppm were excluded because of potential overlap in this region of the spectrum. A subset of decay 15 15 curves from the N R1ρ experiment, which are used to calculate the N R 2 rates, is shown in A B

15 15 1 Figure 8.5: N R 2 relaxation data for 250 μM SUR2A NBD1. (A) Selected regions from the N- H correlation 15 spectrum recorded with the pulse sequence used to determine N R 1ρ rates with a delay time of 2 ms. (B) Examples 15 15 15 of N R 1ρ decay curves for peaks shown in (A). The N R 1ρ rates were used to determine N R 2 rates, as described in the Experimental Procedures section.

176

15 Figure 8.5. The calculated N R 2 rates for these resonances were used to calculate a correlation time ( t m) for SUR2A NBD1 using only the J(0) term for the spectral density function and assuming an order parameter (S 2) of 0.85, which is the average S 2 value expected for a folded 86 15 protein. Only the J(0) term was considered because it dominates the N R 2 relaxation of large molecules, such as proteins.87 We calculate a correlation time of 25.3 – 4.5 ns. This value compares favourably with an expected correlation time of 22 ns calculated for a homology model of SUR2A NBD1 from HYDRONMR,98 indicating that our sample of SUR2A NBD1 is predominantly monomeric under these conditions.

8.3.2 Specific Binding of pinacidil by SUR2A NBD1 displays a low affinity

SUR2A NBD1 contains a total of six Trp residues that are located throughout the protein, that produce a large fluorescence signal (Figure 8.6A). Changes in the environment of one or more Trp residues from direct binding of pinacidil or conformational changes associated with binding have the potential to produce changes in Trp emission spectra in proteins.223 SUR2A NBD1 displays a large fluorescence signal, owing to the six Trp residues in the protein. Indeed, addition of pinacidil results in a large quenching of Trp fluorescence (Figure 8.6A), indicating direct binding of pinacidil to NBD1. Quenching of fluorescence often results from exposure of Trp residues to the solvent, which in this case would suggest conformational changes upon pinacidil binding. However, quenching of fluorescence may also result in other situations. For example,

A B

Figure 8.6: Binding of pinacidil to SUR2A NBD1 monitored by intrinsic Trp fluorescence. (A) Emission spectra of 1 μM SUR2A NBD1 in NBD1 buffer with 5 mM MgATP and 10% (v/v) DMSO in the absence (black) and presence of 1 mM pinacidil (gray). (B) Binding of pinacidil to SUR2A NBD1 as monitored by quenching of Trp fluorescence. The data (open squares) were fit assuming a 1:1 complex (solid line), as described in Experimental Procedures.

177 direct binding of pinacidil to a site involving a Trp may cause quenching depending on the orientation of the aromatic group in pinacidil and a Trp residue in the protein. Quenching may occur because pinacidil binding causes conformational changes in NBD1 that bring acidic residues in close proximity to one or more Trp residues. Regardless of the quenching mechanism, the large changes in the fluorescence spectrum of NBD1 indicate an interaction with pinacidil. Although the spectra shown in Figure 8.6A are for 1 m M SUR2A NBD1 in absence and presence of 1 mM pinacidil, a significant amount of quenching (~50 %) is seen with much lower pinacidil concentrations (i.e. 100 m M). A K d value of 455 – 37 m M was obtained from the pinacidil titration (Figure 8.6B). The relatively low affinity interaction is likely because we are probing pinacidil binding to NBD1 alone and not in the presence of coupling helices, and therefore reflective of having only part of the pinacidil binding site present in the experiment.

8.3.3 Binding of pinacidil changes the affinity of NBD1 for MgATP

After establishing that pinacidil binds SUR2A NBD1 by NMR and fluorescence spectroscopy, we used fluorescence spectroscopy to probe the effect of pinacidil on the nucleotide affinity of NBD1. Although previous work demonstrated the interaction of SUR2A NBD1 (S615-L933) with MgATP using NMR spectroscopy,64 we could not use NMR titration experiments to A B

Figure 8.7. TNP-ATP binding SUR2A NBD1 in the absence and presence of pinacidil. (A) Emission spectra of 2.5 μM TNP-ATP in the NBD1 buffer in the absence (gray line) and presence of 50 μM SUR2A NBD1 (black line). (B) Binding of SUR2A NBD1 to TNP-ATP as monitored by increasing fluorescence of TNP-ATP. TNP-ATP fluorescence titration data acquired in the absence of drug are shown as solid circles and were fit assuming a 1:1 complex, shown by the solid line, as described in the Experimental Procedures. Data from NBD1 titration of TNP- ATP acquired in the presence of pinacidil are shown as solid diamonds with the fit of the TNP-ATP titration data shown as a dashed line. The data and the fit from the TNP-ATP titration in the presence of diazoxide are shown as open circles and a dotted line, respectively. TNP-ATP titration data in the presence of pinacidil and diazoxide were fit assuming a 1:1 complex for the NBD1/nucleotide interaction.

178 determine the affinity of the interaction due to the tendency of the nucleotide-free NBD1 to precipitate at concentrations greater than 100 m M at 30 °C, which are the conditions necessary for the NMR experiments.

The interaction between SUR2A NBD1 and nucleotide was probed using the fluorescent ATP analogue TNP-ATP. TNP-ATP has a tri-nitrophenyl fluorophore on the ribose of the nucleotide and has been employed to study nucleotide binding to ATP binding proteins including kinases 110, 312 and other ABC transporters, such as P-glycoprotein 183 and CFTR.244 The affinities for ATP and ADP may 299 or may not 110 be different from that of their fluorescent analogues. Nonetheless, these probes have been used successfully to compare changes in nucleotide binding in other NBDs under different experimental conditions, such as in wild type and mutant states.109, 188, 244

Upon addition of SUR2A NBD1, the fluorescence emission spectra of TNP-ATP changes so that the intensity increases and the maximum is blue-shifted (Figure 8.7A), corresponding to the direct binding of TNP-ATP by NBD1. Addition of increasing amounts of NBD1 resulted in a concentration-dependent and saturable increase in TNP-ATP fluorescence (Figure 8.7B; closed circles, solid lines). The K d value obtained from these measurements shows tight binding of

NBD1 to TNP-ATP (8.0 m M ± 1.0 m M). A similar K d value was obtained for the interaction of TNP-ATP with NBD1 from CFTR,244 which is in the same subfamily of ABC proteins as SUR2A.72 Table 8.1: Dissociation Constants for Interactions of SUR2A NBD1 with Nucleotide

Nucleotide Kd (μM) TNP-ATP 8.0 ± 1.0 (3) TNP-ATP (in the presence of pinacidil) 3.1 ± 0.5 (3) TNP-ATP (in the presence of diazoxide) 11.9 ± 3.0 (2) The equilibrium dissociation constants (Kd, μM) were determined using fluorescence of the trinitrophenyl moiety of TNP-ATP and are reported as averages ± standard deviations. The number in parentheses indicates number of experiments performed for each binding analysis.

In the presence of 12.5 m M pinacidil, NBD1 titration of TNP-ATP also displays a saturable fluorescence increase (Figure 8.7B; solid diamonds, dashed line). However, the K d value for the interaction between TNP-ATP and pinacidil-bound NBD1 is decreased to 3.1 ± 0.5 m M. The increase in the affinity of NBD1 for TNP-ATP in the presence of 12.5 m M of pinacidil, suggests that pinacidil stabilizes the interaction between the nucleotide and NBD1. Similar titration experiments were performed in the presence of diazoxide (Figure 8.7B; open

179

circles, dotted line). In this case, no change in the affinity of NBD1 and TNP-ATP (K d value = 11.8 ± 3.0 m M) was observed, consistent with the NMR data that did not demonstrate an interaction between diazoxide and SUR2A NBD1. The Kd values measured are shown in Table 8.1.

8.4 Discussion

The data in this paper demonstrate, for the first time, a direct interaction of the KCO pinacidil 15 64 with SUR2A NBD1. N R 2 relaxation experiments and other lines of evidence presented suggest that the NBD1 is not an artificial homodimer, even at the high concentrations of protein and nucleotide used in these experiments. We currently do not have resonance assignments for the protein, and therefore can not specifically identify NBD1 residues that interact with pinacidil. However, the NMR titration data show chemical shift changes for a number of resonances derived primarily from structured regions in the protein upon addition of pinacidil. The intense resonances, which are derived from disordered regions in NBD1 such as the b -sheet subdomain insert and C-terminal extension,64 do not exhibit chemical shift changes upon pinacidil addition. The number and magnitude of the spectral shift changes are consistent with an interaction of pinacidil with specific residues in NBD1.128 The observation that many of the resonances in the NBD1 spectrum do not change upon addition of pinacidil demonstrate that binding of pinacidil does not cause gross structural changes to the protein structure. Further, quenching of Trp fluorescence in NBD1 upon addition of pinacidil and the finding that binding of pinacidil to NBD1 increases the affinity of NBD1 for nucleotide also demonstrate a direct interaction between pinacidil and NBD1. Previous work has demonstrated the involvement of residues in coupling helix 3,307 the MSD2-NBD2 linker,307 TM helix 17,208 and the C42 region 268 in binding of pinacidil and/or mediation of its effects in the intact K ATP channel. Structures of ABC proteins 5, 62, 325 and our homology model of SUR2A revealed that coupling helix 3 forms an interface with NBD1 and NBD2, and the flexible linker can also interact at this site. Therefore, the data we present here on the interaction of pinacidil with NBD1 is consistent with earlier studies that sought to localize the binding site of pinacidil in the intact SUR2A. Of note, there is a Trp residue, Trp 756, at this interface. Although changes in fluorescence can be due to multiple mechanisms, quenching of the fluorescence of Trp756 may occur due to direct binding of pinacidil at this site. Together with residues in coupling helix 3,307 the MSD2-NBD1 linker,307

180 and NBD2, residues around Trp 756 in NBD1 may be involved in pinacidil binding. The relatively low affinity that we measured for the binding of isolated NBD1 with pinacidil is consistent with the hypothesis that NBD1 contributes only part of the pinacidil binding site in the intact channel, with the full binding site comprised of residues from the coupling helices and the MSD2-NBD1 linker,14, 307 and residues of NBD1 and NBD2. In contrast to pinacidil, the NMR data indicate that diazoxide does not bind SUR2A NBD1, as indicated by other studies.19, 56 Earlier studies demonstrated that, in general, diazoxide 14, 89, 102, 209 can activate SUR1-containing K ATP channels, but not SUR2A-containing channels.

Notably, diazoxide sensitivity can be conferred to SUR2A-containing K ATP channels by transfer of MSD1-NBD1 from SUR1 to SUR2A,19 indicating that SUR2A NBD1 is not involved in the diazoxide binding interface in SUR2A, consistent with the lack of interaction between diazoxide and SUR2A NBD1 observed here. Further, while residues L1249 and T1253 in SUR2A are involved in pinacidil activation, mutation of the corresponding residues in SUR1 has no affect on diazoxide-mediated activation of SUR1-containing channels.208 These studies indicate that separate regions mediate the effect of diazoxide and other openers, such as pinacidil.56

Residues involved in KCO binding are not necessarily involved in K ATP channel activation. Chimeras between SUR2A and MRP1 indicated that residues E1305, I1319, and L1313 are important for channel activation by, but not binding 78 of, the pinacidil analogue P1075. Residue E1305 is located in the MSD2-NBD2 linker and I1310 and L1313 are located in NBD2. Recent structures of ABC proteins 5, 62, 325 and our SUR2A homology model show that the MSD2-NBD2 linker contacts coupling helix 3, which in turn contacts NBD1. Thus, interaction of coupling helix 3 with pinacidil could affect the conformation of residues in the MSD2-NBD2 linker, such as E1305, I1319, and L1313, leading to changes in channel activation, possibly by coupling to Kir6.2.78, 246 Conformational changes and dynamics are important to link opener binding to channel activation. Conformational changes are expected in the NBDs from ATP binding and hydrolysis during the gating cycle. Drug binding may enhance these functionally relevant conformational changes. For example, increases in nucleotide binding to NBD1 in intact K ATP channels, as we observed for the isolated SUR2A NBD1, in the presence of pinacidil may enhance structural and dynamic changes required for channel gating. A similar effect has been demonstrated in a recent study showing enhanced conformational changes in CFTR NBD1 upon drug binding that are comparable to the structural changes involved in gating of the CFTR channel.128 Further drug

181

binding studies encompassing different regions of the K ATP channel, such as SUR2A NBD2, as well as different KCOs, are needed to elucidate some of these mechanisms.

182

References

1. Abraham, M.R., V.A. Selivanov, D.M. Hodgson, D. Pucar, L.V. Zingman, B. Wieringa, P.P. Dzeja, A.E. Alekseev, and A. Terzic, Coupling of cell energetics with membrane metabolic sensing. Integrative signaling through creatine kinase phosphotransfer disrupted by M-CK gene knock-out. J Biol Chem, 2002. 277 (27): p. 24427-34.

2. Aittoniemi, J., C. Fotinou, T.J. Craig, H. de Wet, P. Proks, and F.M. Ashcroft, Review. SUR1: a unique ATP-binding cassette protein that functions as an ion channel regulator. Philos Trans R Soc Lond B Biol Sci, 2009. 364 (1514): p. 257-67.

3. Aleksandrov, A.A., P. Kota, L.A. Aleksandrov, L. He, T. Jensen, L. Cui, M. Gentzsch, N.V. Dokholyan, and J.R. Riordan, Regulatory insertion removal restores maturation, stability and function of DeltaF508 CFTR. J Mol Biol, 2010. 401 (2): p. 194-210.

4. Aleksandrov, L., A. Mengos, X. Chang, A. Aleksandrov, and J.R. Riordan, Differential interactions of nucleotides at the two nucleotide binding domains of the cystic fibrosis transmembrane conductance regulator. J Biol Chem, 2001. 276 (16): p. 12918-23.

5. Aller, S.G., J. Yu, A. Ward, Y. Weng, S. Chittaboina, R. Zhuo, P.M. Harrell, Y.T. Trinh, Q. Zhang, I.L. Urbatsch, and G. Chang, Structure of P-glycoprotein reveals a molecular basis for poly-specific drug binding. Science, 2009. 323 (5922): p. 1718-22.

6. Altenberg, G.A., The engine of ABC proteins. News Physiol Sci, 2003. 18 : p. 191-5.

7. Ambudkar, S.V., I.W. Kim, D. Xia, and Z.E. Sauna, The A-loop, a novel conserved aromatic acid subdomain upstream of the Walker A motif in ABC transporters, is critical for ATP binding. FEBS Lett, 2006. 580 (4): p. 1049-55.

8. Arakel, E.C., S. Brandenburg, K. Uchida, H. Zhang, Y.W. Lin, T. Kohl, B. Schrul, M.S. Sulkin, I.R. Efimov, C.G. Nichols, S.E. Lehnart, and B. Schwappach, Tuning the electrical properties of the heart by differential trafficking of KATP ion channel complexes. J Cell Sci, 2014. 127 (Pt 9): p. 2106-19.

9. Armstrong, N., A. de Lencastre, and E. Gouaux, A new protein folding screen: application to the ligand binding domains of a glutamate and kainate receptor and to lysozyme and carbonic anhydrase. Protein Sci, 1999. 8(7): p. 1475-83.

10. Arora, D. and N. Khanna, Method for increasing the yield of properly folded recombinant human gamma interferon from inclusion bodies. J Biotechnol, 1996. 52 (2): p. 127-33.

11. Ashcroft, F.M. and P. Rorsman, Electrophysiology of the pancreatic beta-cell. Prog Biophys Mol Biol, 1989. 54 (2): p. 87-143.

12. Ashcroft, F.M. and F.M. Gribble, Correlating structure and function in ATP-sensitive K+ channels. Trends Neurosci, 1998. 21 (7): p. 288-94.

183

13. Ashcroft, F.M. and F.M. Gribble, Tissue-specific effects of sulfonylureas: lessons from studies of cloned K(ATP) channels. J Diabetes Complications, 2000. 14 (4): p. 192-6.

14. Ashcroft, F.M. and F.M. Gribble, New windows on the mechanism of action of K(ATP) channel openers. Trends Pharmacol Sci, 2000. 21 (11): p. 439-45.

15. Ashcroft, F.M., The Walter B. Cannon Physiology in Perspective Lecture, 2007. ATP- sensitive K+ channels and disease: from molecule to malady. Am J Physiol Endocrinol Metab, 2007. 293 (4): p. E880-9.

16. Ashfield, R., F.M. Gribble, S.J. Ashcroft, and F.M. Ashcroft, Identification of the high- affinity tolbutamide site on the SUR1 subunit of the K(ATP) channel. Diabetes, 1999. 48 (6): p. 1341-7.

17. Atwell, S., C.G. Brouillette, K. Conners, S. Emtage, T. Gheyi, W.B. Guggino, J. Hendle, J.F. Hunt, H.A. Lewis, F. Lu, Protasevich, II, L.A. Rodgers, R. Romero, S.R. Wasserman, P.C. Weber, D. Wetmore, F.F. Zhang, and X. Zhao, Structures of a minimal human CFTR first nucleotide-binding domain as a monomer, head-to-tail homodimer, and pathogenic mutant. Protein Eng Des Sel, 2010. 23 (5): p. 375-84.

18. Aziz, Q., A.M. Thomas, T. Khambra, and A. Tinker, Regulation of the ATP-sensitive potassium channel subunit, Kir6.2, by a Ca2+-dependent protein kinase C. J Biol Chem, 2012. 287 (9): p. 6196-207.

19. Babenko, A.P., G. Gonzalez, and J. Bryan, Pharmaco-topology of sulfonylurea receptors. Separate domains of the regulatory subunits of K(ATP) channel isoforms are required for selective interaction with K(+) channel openers. J Biol Chem, 2000. 275 (2): p. 717-20.

20. Babenko, A.P. and J. Bryan, Sur domains that associate with and gate KATP pores define a novel gatekeeper. J Biol Chem, 2003. 278 (43): p. 41577-80.

21. Baker, J.M., R.P. Hudson, V. Kanelis, W.Y. Choy, P.H. Thibodeau, P.J. Thomas, and J.D. Forman-Kay, CFTR regulatory region interacts with NBD1 predominantly via multiple transient helices. Nat Struct Mol Biol, 2007. 14 (8): p. 738-45.

22. Bakos, E., R. Evers, G. Calenda, G.E. Tusnady, G. Szakacs, A. Varadi, and B. Sarkadi, Characterization of the amino-terminal regions in the human multidrug resistance protein (MRP1). J Cell Sci, 2000. 113 Pt 24 : p. 4451-61.

23. Bartesaghi, A., D. Matthies, S. Banerjee, A. Merk, and S. Subramaniam, Structure of beta-galactosidase at 3.2-A resolution obtained by cryo-electron microscopy. Proc Natl Acad Sci U S A, 2014. 111 (32): p. 11709-14.

24. Baukrowitz, T., S.J. Tucker, U. Schulte, K. Benndorf, J.P. Ruppersberg, and B. Fakler, Inward rectification in KATP channels: a pH switch in the pore. EMBO J, 1999. 18 (4): p. 847-53.

184

25. Becker, J.P., G. Depret, F. Van Bambeke, P.M. Tulkens, and M. Prevost, Molecular models of human P-glycoprotein in two different catalytic states. BMC Struct Biol, 2009. 9: p. 3.

26. Beguin, P., K. Nagashima, M. Nishimura, T. Gonoi, and S. Seino, PKA-mediated phosphorylation of the human K(ATP) channel: separate roles of Kir6.2 and SUR1 subunit phosphorylation. EMBO J, 1999. 18 (17): p. 4722-32.

27. Bellstedt, P., T. Seiboth, S. Hafner, H. Kutscha, R. Ramachandran, and M. Gorlach, Resonance assignment for a particularly challenging protein based on systematic unlabeling of amino acids to complement incomplete NMR data sets. J Biomol NMR, 2013. 57 (1): p. 65-72.

28. Benarroch, E.E., Potassium channels: brief overview and implications in epilepsy. Neurology, 2009. 72 (7): p. 664-9.

29. Bichet, D., F.A. Haass, and L.Y. Jan, Merging functional studies with structures of inward-rectifier K(+) channels. Nat Rev Neurosci, 2003. 4(12): p. 957-67.

30. Bienengraeber, M., A.E. Alekseev, M.R. Abraham, A.J. Carrasco, C. Moreau, M. Vivaudou, P.P. Dzeja, and A. Terzic, ATPase activity of the sulfonylurea receptor: a catalytic function for the KATP channel complex. FASEB J, 2000. 14 (13): p. 1943-52.

31. Bienengraeber, M., T.M. Olson, V.A. Selivanov, E.C. Kathmann, F. O'Cochlain, F. Gao, A.B. Karger, J.D. Ballew, D.M. Hodgson, L.V. Zingman, Y.P. Pang, A.E. Alekseev, and A. Terzic, ABCC9 mutations identified in human disrupt catalytic KATP channel gating. Nat Genet, 2004. 36 (4): p. 382-7.

32. Bilwes, A.M., C.M. Quezada, L.R. Croal, B.R. Crane, and M.I. Simon, Nucleotide binding by the histidine kinase CheA. Nat Struct Biol, 2001. 8(4): p. 353-60.

33. Bozoky, Z., M. Krzeminski, R. Muhandiram, J.R. Birtley, A. Al-Zahrani, P.J. Thomas, R.A. Frizzell, R.C. Ford, and J.D. Forman-Kay, Regulatory R region of the CFTR chloride channel is a dynamic integrator of phospho-dependent intra- and intermolecular interactions. Proc Natl Acad Sci U S A, 2013. 110 (47): p. E4427-36.

34. Bungert, S., L.L. Molday, and R.S. Molday, Membrane topology of the ATP binding cassette transporter ABCR and its relationship to ABC1 and related ABCA transporters: identification of N-linked glycosylation sites. J Biol Chem, 2001. 276 (26): p. 23539-46.

35. Burke, M.A., R.K. Mutharasan, and H. Ardehali, The sulfonylurea receptor, an atypical ATP-binding cassette protein, and its regulation of the KATP channel. Circ Res, 2008. 102 (2): p. 164-76.

36. Callaghan, R., R.C. Ford, and I.D. Kerr, The translocation mechanism of P-glycoprotein. FEBS Lett, 2006. 580 (4): p. 1056-63.

37. Carrasco, A.J., P.P. Dzeja, A.E. Alekseev, D. Pucar, L.V. Zingman, M.R. Abraham, D. Hodgson, M. Bienengraeber, M. Puceat, E. Janssen, B. Wieringa, and A. Terzic,

185

Adenylate kinase phosphotransfer communicates cellular energetic signals to ATP- sensitive potassium channels. Proc Natl Acad Sci U S A, 2001. 98 (13): p. 7623-8.

38. Cavanagh, J., Protein NMR spectroscopy : principles and practice . 2nd ed. 2007, Amsterdam ; Boston: Academic Press. xxv, 885 p.

39. Cavanagh, J., Protein NMR spectroscopy : principles and practice . 2007, Amsterdam ; Boston: Academic Press. xxiii, 885 p.

40. Chacon, P., F. Tama, and W. Wriggers, Mega-Dalton biomolecular motion captured from electron microscopy reconstructions. J Mol Biol, 2003. 326(2): p. 485-92.

41. Challinor-Rogers, J.L. and G.A. McPherson, Potassium channel openers and other regulators of KATP channels. Clin Exp Pharmacol Physiol, 1994. 21 (8): p. 583-97.

42. Chan, K.W., H. Zhang, and D.E. Logothetis, N-terminal transmembrane domain of the SUR controls trafficking and gating of Kir6 channel subunits. EMBO J, 2003. 22 (15): p. 3833-43.

43. Chang, T.J., W.P. Chen, C. Yang, P.H. Lu, Y.C. Liang, M.J. Su, S.C. Lee, and L.M. Chuang, Serine-385 phosphorylation of inwardly rectifying K+ channel subunit (Kir6.2) by AMP-dependent protein kinase plays a key role in rosiglitazone-induced closure of the K(ATP) channel and insulin secretion in rats. Diabetologia, 2009. 52 (6): p. 1112-21.

44. Chang, X.B., L. Cui, Y.X. Hou, T.J. Jensen, A.A. Aleksandrov, A. Mengos, and J.R. Riordan, Removal of multiple arginine-framed trafficking signals overcomes misprocessing of delta F508 CFTR present in most patients with cystic fibrosis. Mol Cell, 1999. 4(1): p. 137-42.

45. Chary, K.V.R., G. Govil, and SpringerLink (Online service), NMR in Biological Systems From Molecules to Humans , in Focus on Structural Biology, . 2008, Springer Science+Business Media B.V.: Dordrecht. p. 530 p.

46. Chen, H., S. Viel, F. Ziarelli, and L. Peng, 19F NMR: a valuable tool for studying biological events. Chem Soc Rev, 2013. 42 (20): p. 7971-82.

47. Chen, Z.C., Y.Z. Cheng, L.J. Chen, K.C. Cheng, Y. Li, and J. Cheng, Increase of ATP- sensitive potassium (K(ATP)) channels in the heart of type-1 diabetic rats. Cardiovasc Diabetol, 2012. 11 : p. 8.

48. Cheng, S.H., D.P. Rich, J. Marshall, R.J. Gregory, M.J. Welsh, and A.E. Smith, Phosphorylation of the R domain by cAMP-dependent protein kinase regulates the CFTR chloride channel. Cell, 1991. 66 (5): p. 1027-36.

49. Chutkow, W.A., M.C. Simon, M.M. Le Beau, and C.F. Burant, Cloning, tissue expression, and chromosomal localization of SUR2, the putative drug-binding subunit of cardiac, , and vascular KATP channels. Diabetes, 1996. 45 (10): p. 1439- 45.

186

50. Chutkow, W.A., J.C. Makielski, D.J. Nelson, C.F. Burant, and Z. Fan, Alternative splicing of sur2 Exon 17 regulates nucleotide sensitivity of the ATP-sensitive potassium channel. J Biol Chem, 1999. 274 (19): p. 13656-65.

51. Clement, J.P.t., K. Kunjilwar, G. Gonzalez, M. Schwanstecher, U. Panten, L. Aguilar- Bryan, and J. Bryan, Association and stoichiometry of K(ATP) channel subunits. Neuron, 1997. 18 (5): p. 827-38.

52. Corkey, B.E., J.T. Deeney, G.C. Yaney, K. Tornheim, and M. Prentki, The role of long- chain fatty acyl-CoA esters in beta-cell signal transduction. J Nutr, 2000. 130 (2S Suppl): p. 299S-304S.

53. Craig, T.J., F.M. Ashcroft, and P. Proks, How ATP inhibits the open K(ATP) channel. J Gen Physiol, 2008. 132 (1): p. 131-44.

54. Crawford, R.M., H.J. Ranki, C.H. Botting, G.R. Budas, and A. Jovanovic, Creatine kinase is physically associated with the cardiac ATP-sensitive K+ channel in vivo. FASEB J, 2002. 16 (1): p. 102-4.

55. Csanady, L., K.W. Chan, D. Seto-Young, D.C. Kopsco, A.C. Nairn, and D.C. Gadsby, Severed channels probe regulation of gating of cystic fibrosis transmembrane conductance regulator by its cytoplasmic domains. J Gen Physiol, 2000. 116 (3): p. 477- 500.

56. D'Hahan, N., H. Jacquet, C. Moreau, P. Catty, and M. Vivaudou, A transmembrane domain of the sulfonylurea receptor mediates activation of ATP-sensitive K(+) channels by K(+) channel openers. Mol Pharmacol, 1999. 56 (2): p. 308-15.

57. D'Hahan, N., C. Moreau, A.L. Prost, H. Jacquet, A.E. Alekseev, A. Terzic, and M. Vivaudou, Pharmacological plasticity of cardiac ATP-sensitive potassium channels toward diazoxide revealed by ADP. Proc Natl Acad Sci U S A, 1999. 96 (21): p. 12162-7.

58. Dart, C., M.L. Leyland, P.J. Spencer, P.R. Stanfield, and M.J. Sutcliffe, The selectivity filter of a potassium channel, murine kir2.1, investigated using scanning cysteine mutagenesis. J Physiol, 1998. 511 ( Pt 1) : p. 25-32.

59. Dassa, E. and P. Bouige, The ABC of ABCS: a phylogenetic and functional classification of ABC systems in living organisms. Res Microbiol, 2001. 152 (3-4): p. 211-29.

60. Davidson, A.L. and J. Chen, ATP-binding cassette transporters in bacteria. Annu Rev Biochem, 2004. 73 : p. 241-68.

61. Davidson, A.L., E. Dassa, C. Orelle, and J. Chen, Structure, function, and evolution of bacterial ATP-binding cassette systems. Microbiol Mol Biol Rev, 2008. 72 (2): p. 317-64, table of contents.

62. Dawson, R.J. and K.P. Locher, Structure of a bacterial multidrug ABC transporter. Nature, 2006. 443 (7108): p. 180-5.

187

63. de Araujo, C., D. Arefeen, Y. Tadesse, B.M. Long, G.D. Price, R.S. Rowlett, M.S. Kimber, and G.S. Espie, Identification and characterization of a carboxysomal gamma- carbonic anhydrase from the cyanobacterium Nostoc sp. PCC 7120. Photosynth Res, 2014. 121 (2-3): p. 135-50.

64. de Araujo, E.D., L.K. Ikeda, S. Tzvetkova, and V. Kanelis, The first nucleotide binding domain of the sulfonylurea receptor 2A contains regulatory elements and is folded and functions as an independent module. Biochemistry, 2011. 50 (31): p. 6655-66.

65. de Araujo, E.D., J. Patel, C. de Araujo, S.P. Rogers, S.M. Short, D.A. Campbell, and G.S. Espie, Physiological characterization and light response of the CO2-concentrating mechanism in the filamentous cyanobacterium Leptolyngbya sp. CPCC 696. Photosynth Res, 2011. 109 (1-3): p. 85-101.

66. de Araujo, E.D. and V. Kanelis, Successful development and use of a thermodynamic stability screen for optimizing the yield of nucleotide binding domains. Protein Expr Purif, 2014. 103C : p. 38-47.

67. de Araujo, E.D., C.P. Alvarez, J.P. Lopez-Alonso, C.R. Sooklal, M. Stagljar, and V. Kanelis, Phosphorylation-dependent changes in nucleotide binding, conformation, and dynamics of the first nucleotide binding domain (NBD1) of the sulphonylurea receptor 2B (SUR2B). J Biol Chem, 2015.

68. de Foresta, B., M. Vincent, J. Gallay, and M. Garrigos, Interaction with membrane mimics of transmembrane fragments 16 and 17 from the human multidrug resistance ABC transporter 1 (hMRP1/ABCC1) and two of their tryptophan variants. Biochim Biophys Acta, 2010. 1798 (3): p. 401-14.

69. de Marco, A., Protocol for preparing proteins with improved solubility by co-expressing with molecular chaperones in Escherichia coli. Nat Protoc, 2007. 2(10): p. 2632-9.

70. de Marco, A., E. Deuerling, A. Mogk, T. Tomoyasu, and B. Bukau, Chaperone-based procedure to increase yields of soluble recombinant proteins produced in E. coli. BMC Biotechnol, 2007. 7: p. 32.

71. de Wet, H., C. Fotinou, N. Amad, M. Dreger, and F.M. Ashcroft, The ATPase activities of sulfonylurea receptor 2A and sulfonylurea receptor 2B are influenced by the C- terminal 42 amino acids. FEBS J, 2010. 277 (12): p. 2654-62.

72. Dean, M., A. Rzhetsky, and R. Allikmets, The human ATP-binding cassette (ABC) transporter superfamily. Genome Res, 2001. 11 (7): p. 1156-66.

73. Deber, C.M., B. Brodsky, and A. Rath, Proline residues in proteins . Encyclopedia of Life Sciences. 2010, Chichester: John Wiley and Sons. 9.

74. Decher, N., V. Renigunta, M. Zuzarte, M. Soom, S.H. Heinemann, K.W. Timothy, M.T. Keating, J. Daut, M.C. Sanguinetti, and I. Splawski, Impaired interaction between the slide helix and the C-terminus of Kir2.1: a novel mechanism of Andersen syndrome. Cardiovasc Res, 2007. 75 (4): p. 748-57.

188

75. Delaglio, F., S. Grzesiek, G.W. Vuister, G. Zhu, J. Pfeifer, and A. Bax, NMRPipe: a multidimensional spectral processing system based on UNIX pipes. J Biomol NMR, 1995. 6(3): p. 277-93.

76. DiMaio, F., Y. Song, X. Li, M.J. Brunner, C. Xu, V. Conticello, E. Egelman, T.C. Marlovits, Y. Cheng, and D. Baker, Atomic-accuracy models from 4.5-A cryo-electron microscopy data with density-guided iterative local refinement. Nat Methods, 2015. 12 (4): p. 361-5.

77. Drumm, M.L., D.J. Wilkinson, L.S. Smit, R.T. Worrell, T.V. Strong, R.A. Frizzell, D.C. Dawson, and F.S. Collins, Chloride conductance expressed by delta F508 and other mutant CFTRs in Xenopus oocytes. Science, 1991. 254 (5039): p. 1797-9.

78. Dupuis, J.P., J. Revilloud, C.J. Moreau, and M. Vivaudou, Three C-terminal residues from the sulphonylurea receptor contribute to the functional coupling between the K(ATP) channel subunits SUR2A and Kir6.2. J Physiol, 2008. 586 (13): p. 3075-85.

79. Dzeja, P. and A. Terzic, Adenylate kinase and AMP signaling networks: metabolic monitoring, signal communication and body energy sensing. Int J Mol Sci, 2009. 10 (4): p. 1729-72.

80. Dzeja, P.P. and A. Terzic, Phosphotransfer reactions in the regulation of ATP-sensitive K+ channels. FASEB J, 1998. 12 (7): p. 523-9.

81. Dzeja, P.P. and A. Terzic, Phosphotransfer networks and cellular energetics. J Exp Biol, 2003. 206 (Pt 12): p. 2039-47.

82. Ehrlich, J.R., Inward rectifier potassium currents as a target for atrial fibrillation therapy. J Cardiovasc Pharmacol, 2008. 52 (2): p. 129-35.

83. Enkvetchakul, D., G. Loussouarn, E. Makhina, S.L. Shyng, and C.G. Nichols, The kinetic and physical basis of K(ATP) channel gating: toward a unified molecular understanding. Biophys J, 2000. 78 (5): p. 2334-48.

84. Enkvetchakul, D. and C.G. Nichols, Gating mechanism of KATP channels: function fits form. J Gen Physiol, 2003. 122 (5): p. 471-80.

85. Fang, K., L. Csanady, and K.W. Chan, The N-terminal transmembrane domain (TMD0) and a cytosolic linker (L0) of sulphonylurea receptor define the unique intrinsic gating of KATP channels. J Physiol, 2006. 576 (Pt 2): p. 379-89.

86. Farrow, N.A., R. Muhandiram, A.U. Singer, S.M. Pascal, C.M. Kay, G. Gish, S.E. Shoelson, T. Pawson, J.D. Forman-Kay, and L.E. Kay, Backbone dynamics of a free and phosphopeptide-complexed Src homology 2 domain studied by 15N NMR relaxation. Biochemistry, 1994. 33 (19): p. 5984-6003.

87. Farrow, N.A., O. Zhang, J.D. Forman-Kay, and L.E. Kay, Comparison of the backbone dynamics of a folded and an unfolded SH3 domain existing in equilibrium in aqueous buffer. Biochemistry, 1995. 34 (3): p. 868-78.

189

88. Finerty, P.J., Jr., A.K. Mittermaier, R. Muhandiram, L.E. Kay, and J.D. Forman-Kay, NMR dynamics-derived insights into the binding properties of a peptide interacting with an SH2 domain. Biochemistry, 2005. 44 (2): p. 694-703.

89. Flagg, T.P., H.T. Kurata, R. Masia, G. Caputa, M.A. Magnuson, D.J. Lefer, W.A. Coetzee, and C.G. Nichols, Differential structure of atrial and ventricular KATP: atrial KATP channels require SUR1. Circ Res, 2008. 103 (12): p. 1458-65.

90. Flagg, T.P., D. Enkvetchakul, J.C. Koster, and C.G. Nichols, Muscle KATP channels: recent insights to energy sensing and myoprotection. Physiol Rev, 2010. 90 (3): p. 799- 829.

91. Flagg, T.P. and C.G. Nichols, "Cardiac KATP": a family of ion channels. Circ Arrhythm Electrophysiol, 2011. 4(6): p. 796-8.

92. Flanagan, S.E., S. Clauin, C. Bellanne-Chantelot, P. de Lonlay, L.W. Harries, A.L. Gloyn, and S. Ellard, Update of mutations in the genes encoding the pancreatic beta-cell K(ATP) channel subunits Kir6.2 (KCNJ11) and sulfonylurea receptor 1 (ABCC8) in diabetes mellitus and hyperinsulinism. Hum Mutat, 2009. 30 (2): p. 170-80.

93. Fotinou, C., J. Aittoniemi, H. de Wet, A. Polidori, B. Pucci, M.S. Sansom, C. Venien- Bryan, and F.M. Ashcroft, Tetrameric structure of SUR2B revealed by electron microscopy of oriented single particles. FEBS J, 2013. 280 (4): p. 1051-63.

94. Frelet, A. and M. Klein, Insight in eukaryotic ABC transporter function by mutation analysis. FEBS Lett, 2006. 580 (4): p. 1064-84.

95. Gadsby, D.C., P. Vergani, and L. Csanady, The ABC protein turned chloride channel whose failure causes cystic fibrosis. Nature, 2006. 440 (7083): p. 477-83.

96. Gao, M., H.R. Cui, D.W. Loe, C.E. Grant, K.C. Almquist, S.P. Cole, and R.G. Deeley, Comparison of the functional characteristics of the nucleotide binding domains of multidrug resistance protein 1. J Biol Chem, 2000. 275 (17): p. 13098-108.

97. Garcia-Fruitos, E., M. Martinez-Alonso, N. Gonzalez-Montalban, M. Valli, D. Mattanovich, and A. Villaverde, Divergent genetic control of protein solubility and conformational quality in Escherichia coli. J Mol Biol, 2007. 374 (1): p. 195-205.

98. Garcia de la Torre, J., M.L. Huertas, and B. Carrasco, HYDRONMR: prediction of NMR relaxation of globular proteins from atomic-level structures and hydrodynamic calculations. J Magn Reson, 2000. 147 (1): p. 138-46.

99. Garrett, R. and C.M. Grisham, Biochemistry . 2nd ed. Saunders golden sunburst series. 1999, Fort Worth: Saunders College Pub.

100. Gaudet, R. and D.C. Wiley, Structure of the ABC ATPase domain of human TAP1, the transporter associated with antigen processing. EMBO J, 2001. 20 (17): p. 4964-72.

190

101. Gekko, K. and S.N. Timasheff, Thermodynamic and kinetic examination of protein stabilization by glycerol. Biochemistry, 1981. 20 (16): p. 4677-86.

102. Glukhov, A.V., T.P. Flagg, V.V. Fedorov, I.R. Efimov, and C.G. Nichols, Differential K(ATP) channel pharmacology in intact mouse heart. J Mol Cell Cardiol, 2010. 48 (1): p. 152-60.

103. Golovanov, A.P., G.M. Hautbergue, S.A. Wilson, and L.Y. Lian, A simple method for improving protein solubility and long-term stability. J Am Chem Soc, 2004. 126 (29): p. 8933-9.

104. Gomma, A.H., H.J. Purcell, and K.M. Fox, Potassium channel openers in myocardial ischaemia: therapeutic potential of nicorandil. Drugs, 2001. 61 (12): p. 1705-10.

105. Graslund, S., P. Nordlund, J. Weigelt, B.M. Hallberg, J. Bray, O. Gileadi, S. Knapp, U. Oppermann, C. Arrowsmith, R. Hui, J. Ming, S. dhe-Paganon, H.W. Park, A. Savchenko, A. Yee, A. Edwards, R. Vincentelli, C. Cambillau, R. Kim, S.H. Kim, Z. Rao, Y. Shi, T.C. Terwilliger, C.Y. Kim, L.W. Hung, G.S. Waldo, Y. Peleg, S. Albeck, T. Unger, O. Dym, J. Prilusky, J.L. Sussman, R.C. Stevens, S.A. Lesley, I.A. Wilson, A. Joachimiak, F. Collart, I. Dementieva, M.I. Donnelly, W.H. Eschenfeldt, Y. Kim, L. Stols, R. Wu, M. Zhou, S.K. Burley, J.S. Emtage, J.M. Sauder, D. Thompson, K. Bain, J. Luz, T. Gheyi, F. Zhang, S. Atwell, S.C. Almo, J.B. Bonanno, A. Fiser, S. Swaminathan, F.W. Studier, M.R. Chance, A. Sali, T.B. Acton, R. Xiao, L. Zhao, L.C. Ma, J.F. Hunt, L. Tong, K. Cunningham, M. Inouye, S. Anderson, H. Janjua, R. Shastry, C.K. Ho, D. Wang, H. Wang, M. Jiang, G.T. Montelione, D.I. Stuart, R.J. Owens, S. Daenke, A. Schutz, U. Heinemann, S. Yokoyama, K. Bussow, and K.C. Gunsalus, Protein production and purification. Nat Methods, 2008. 5(2): p. 135-46.

106. Gribble, F.M., S.J. Tucker, and F.M. Ashcroft, The essential role of the Walker A motifs of SUR1 in K-ATP channel activation by Mg-ADP and diazoxide. EMBO J, 1997. 16 (6): p. 1145-52.

107. Gribble, F.M. and F.M. Ashcroft, Differential sensitivity of beta-cell and extrapancreatic K(ATP) channels to gliclazide. Diabetologia, 1999. 42 (7): p. 845-8.

108. Grifman, M., A. Arbel, D. Ginzberg, D. Glick, S. Elgavish, B. Shaanan, and H. Soreq, In vitro phosphorylation of acetylcholinesterase at non-consensus protein kinase A sites enhances the rate of acetylcholine hydrolysis. Brain Res Mol Brain Res, 1997. 51 (1-2): p. 179-87.

109. Grycova, L., Z. Lansky, E. Friedlova, V. Vlachova, M. Kubala, V. Obsilova, T. Obsil, and J. Teisinger, ATP binding site on the C-terminus of the vanilloid receptor. Arch Biochem Biophys, 2007. 465 (2): p. 389-98.

110. Guarnieri, M.T., B.S. Blagg, and R. Zhao, A high-throughput TNP-ATP displacement assay for screening inhibitors of ATP-binding in bacterial histidine kinases. Assay Drug Dev Technol, 2011. 9(2): p. 174-83.

191

111. Gulati, S., M. Jamshad, T.J. Knowles, K.A. Morrison, R. Downing, N. Cant, R. Collins, J.B. Koenderink, R.C. Ford, M. Overduin, I.D. Kerr, T.R. Dafforn, and A.J. Rothnie, Detergent-free purification of ABC (ATP-binding-cassette) transporters. Biochem J, 2014. 461 (2): p. 269-78.

112. Guo, D., Y. Ramu, A.M. Klem, and Z. Lu, Mechanism of rectification in inward-rectifier K+ channels. J Gen Physiol, 2003. 121 (4): p. 261-75.

113. Hambrock, A., R. Preisig-Muller, U. Russ, A. Piehl, P.J. Hanley, J. Ray, J. Daut, U. Quast, and C. Derst, Four novel splice variants of sulfonylurea receptor 1. Am J Physiol Cell Physiol, 2002. 283 (2): p. C587-98.

114. Hegedus, T., A.W. Serohijos, N.V. Dokholyan, L. He, and J.R. Riordan, Computational studies reveal phosphorylation-dependent changes in the unstructured R domain of CFTR. J Mol Biol, 2008. 378 (5): p. 1052-63.

115. Hellgren, M., L. Sandberg, and O. Edholm, A comparison between two prokaryotic potassium channels (KirBac1.1 and KcsA) in a molecular dynamics (MD) simulation study. Biophys Chem, 2006. 120 (1): p. 1-9.

116. Higgins, C.F., ABC transporters: physiology, structure and mechanism--an overview. Res Microbiol, 2001. 152 (3-4): p. 205-10.

117. Higgins, C.F. and K.J. Linton, The ATP switch model for ABC transporters. Nat Struct Mol Biol, 2004. 11 (10): p. 918-26.

118. Hirling, H. and R.H. Scheller, Phosphorylation of synaptic vesicle proteins: modulation of the alpha SNAP interaction with the core complex. Proc Natl Acad Sci U S A, 1996. 93 (21): p. 11945-9.

119. Hollenstein, K., R.J. Dawson, and K.P. Locher, Structure and mechanism of ABC transporter proteins. Curr Opin Struct Biol, 2007. 17 (4): p. 412-8.

120. Hopfner, K.P., A. Karcher, D.S. Shin, L. Craig, L.M. Arthur, J.P. Carney, and J.A. Tainer, Structural biology of Rad50 ATPase: ATP-driven conformational control in DNA double-strand break repair and the ABC-ATPase superfamily. Cell, 2000. 101 (7): p. 789- 800.

121. Hopfner, K.P., A. Karcher, L. Craig, T.T. Woo, J.P. Carney, and J.A. Tainer, Structural biochemistry and interaction architecture of the DNA double-strand break repair Mre11 nuclease and Rad50-ATPase. Cell, 2001. 105 (4): p. 473-85.

122. Horio, Y., H. Hibino, A. Inanobe, M. Yamada, M. Ishii, Y. Tada, E. Satoh, Y. Hata, Y. Takai, and Y. Kurachi, Clustering and enhanced activity of an inwardly rectifying potassium channel, Kir4.1, by an anchoring protein, PSD-95/SAP90. J Biol Chem, 1997. 272 (20): p. 12885-8.

123. Hosy, E. and M. Vivaudou, The unusual stoichiometry of ADP activation of the KATP channel. Front Physiol, 2014. 5: p. 11.

192

124. Hou, Y.X., L. Cui, J.R. Riordan, and X.B. Chang, ATP binding to the first nucleotide- binding domain of multidrug resistance protein MRP1 increases binding and hydrolysis of ATP and trapping of ADP at the second domain. J Biol Chem, 2002. 277 (7): p. 5110- 9.

125. Hrycyna, C.A. and M.M. Gottesman, Multidrug ABC transporters from bacteria to man: an emerging hypothesis for the universality of molecular mechanism and function. Drug Resist Updat, 1998. 1(2): p. 81-3.

126. Hu, D., H. Barajas-Martinez, A. Terzic, S. Park, R. Pfeiffer, E. Burashnikov, Y. Wu, M. Borggrefe, C. Veltmann, R. Schimpf, J.J. Cai, G.B. Nam, P. Deshmukh, M. Scheinman, M. Preminger, J. Steinberg, A. Lopez-Izquierdo, D. Ponce-Balbuena, C. Wolpert, M. Haissaguerre, J.A. Sanchez-Chapula, and C. Antzelevitch, ABCC9 is a novel Brugada and early repolarization syndrome susceptibility gene. Int J Cardiol, 2014. 171 (3): p. 431-42.

127. Huang, C.L., Complex roles of PIP2 in the regulation of ion channels and transporters. Am J Physiol Renal Physiol, 2007. 293 (6): p. F1761-5.

128. Hudson, R., A.A. Chong, Protasevich, II, R. Vernon, E. Noy, H. Bihler, J. Li An, O. Kalid, I. Sela-Culang, M. Mense, H. Senderowitz, C.G. Brouillette, and J.D. Forman- Kay, Conformational changes relevant to channel activity and folding within the first nucleotide binding domain of CFTR. J Biol Chem, 2012.

129. Hudson, R.P., P.A. Chong, I.I. Prostasevich, R. Vernon, E. Noy, H. Bihler, J.L. An, O. Kaild, I. Sela-Culang, M. Mense, H. Senderowitz, C.G. Brouillette, and J.D. Forman- Kay, Conformational Changes Relevant to Channel Activity and Folding within the first Nucleotide Binding Domain of CFTR. Journal of Biological Chemistry, 2012. in press .

130. Hung, L.W., I.X. Wang, K. Nikaido, P.Q. Liu, G.F. Ames, and S.H. Kim, Crystal structure of the ATP-binding subunit of an ABC transporter. Nature, 1998. 396 (6712): p. 703-7.

131. Inagaki, N., T. Gonoi, J.P.t. Clement, N. Namba, J. Inazawa, G. Gonzalez, L. Aguilar- Bryan, S. Seino, and J. Bryan, Reconstitution of IKATP: an inward rectifier subunit plus the sulfonylurea receptor. Science, 1995. 270 (5239): p. 1166-70.

132. Inagaki, N., J. Inazawa, and S. Seino, cDNA sequence, gene structure, and chromosomal localization of the human ATP-sensitive potassium channel, uKATP-1, gene (KCNJ8). Genomics, 1995. 30 (1): p. 102-4.

133. Islam, M.S., The islets of Langerhans . Advances in experimental medicine and biology,. 2010, Dordrecht ; New York: Springer. xix, 798 p.

134. Isomoto, S. and Y. Kurachi, Function, regulation, pharmacology, and molecular structure of ATP-sensitive K+ channels in the cardiovascular system. J Cardiovasc Electrophysiol, 1997. 8(12): p. 1431-46.

193

135. Jahangir, A. and A. Terzic, K(ATP) channel therapeutics at the bedside. J Mol Cell Cardiol, 2005. 39 (1): p. 99-112.

136. Jamshad, M., J. Charlton, Y.P. Lin, S.J. Routledge, Z. Bawa, T.J. Knowles, M. Overduin, N. Dekker, T.R. Dafforn, R.M. Bill, D.R. Poyner, and M. Wheatley, G-protein coupled receptor solubilization and purification for biophysical analysis and functional studies, in the total absence of detergent. Biosci Rep, 2015. 35 (2).

137. John, S.A., J.N. Weiss, and B. Ribalet, Regulation of cloned ATP-sensitive K channels by adenine nucleotides and sulfonylureas: interactions between SUR1 and positively charged domains on Kir6.2. J Gen Physiol, 2001. 118 (4): p. 391-405.

138. John, S.A., L.H. Xie, and J.N. Weiss, Mechanism of inward rectification in Kir channels. J Gen Physiol, 2004. 123 (5): p. 623-5.

139. Johnson, B.A. and R.A. Blevins, NMRView: a computer program for the visualization and analysis of NMR data. Journal of Biomolecular NMR, 1994. 4: p. 603-614.

140. Jones, P.M. and A.M. George, Role of the D-loops in allosteric control of ATP hydrolysis in an ABC transporter. J Phys Chem A, 2012. 116 (11): p. 3004-13.

141. Jovanovic, A., S. Jovanovic, E. Lorenz, and A. Terzic, Recombinant cardiac ATP- sensitive K+ channel subunits confer resistance to chemical hypoxia-reoxygenation injury. Circulation, 1998. 98 (15): p. 1548-55.

142. Kane, G.C., X.K. Liu, S. Yamada, T.M. Olson, and A. Terzic, Cardiac KATP channels in health and disease. J Mol Cell Cardiol, 2005. 38 (6): p. 937-43.

143. Kanelis, V., R.P. Hudson, P.H. Thibodeau, P.J. Thomas, and J.D. Forman-Kay, NMR evidence for differential phosphorylation-dependent interactions in WT and DeltaF508 CFTR. EMBO J, 2010. 29 (1): p. 263-77.

144. Kanelis, V., R.P. Hudson, P.H. Thibodeau, P.J. Thomas, and J.D. Forman-Kay, NMR evidence for differential phosphorylation-dependent interactions in WT and ΔF508 CFTR. EMBO J, 2010. 29 (1): p. 263-77.

145. Kanelis, V., P.A. Chong, and J.D. Forman-Kay, NMR spectroscopy to study the dynamics and interactions of CFTR. Methods Mol Biol, 2011. 741 : p. 377-403.

146. Kang, Y., Y.M. Leung, J.E. Manning-Fox, F. Xia, H. Xie, L. Sheu, R.G. Tsushima, P.E. Light, and H.Y. Gaisano, Syntaxin-1A inhibits cardiac KATP channels by its actions on nucleotide binding folds 1 and 2 of sulfonylurea receptor 2A. J Biol Chem, 2004. 279 (45): p. 47125-31.

147. Kang, Y., Y. Zhang, T. Liang, Y.M. Leung, B. Ng, H. Xie, N. Chang, J. Chan, S.L. Shyng, R.G. Tsushima, and H.Y. Gaisano, ATP modulates interaction of syntaxin-1A with sulfonylurea receptor 1 to regulate pancreatic beta-cell KATP channels. J Biol Chem, 2011. 286 (7): p. 5876-83.

194

148. Karger, A.B., S. Park, S. Reyes, M. Bienengraeber, R.B. Dyer, A. Terzic, and A.E. Alekseev, Role for SUR2A ED domain in allosteric coupling within the K(ATP) channel complex. J Gen Physiol, 2008. 131 (3): p. 185-96.

149. Karpowich, N., O. Martsinkevich, L. Millen, Y.R. Yuan, P.L. Dai, K. MacVey, P.J. Thomas, and J.F. Hunt, Crystal structures of the MJ1267 ATP binding cassette reveal an induced-fit effect at the ATPase active site of an ABC transporter. Structure, 2001. 9(7): p. 571-86.

150. Kedracka-Krok, S. and Z. Wasylewski, A differential scanning calorimetry study of tetracycline repressor. Eur J Biochem, 2003. 270 (22): p. 4564-73.

151. Keeler, J., Understanding NMR spectroscopy . 2nd ed. 2010, Chichester, U.K.: John Wiley and Sons. xiii, 511 p.

152. Kerr, I.D., G. Berridge, K.J. Linton, C.F. Higgins, and R. Callaghan, Definition of the domain boundaries is critical to the expression of the nucleotide-binding domains of P- glycoprotein. Eur Biophys J, 2003. 32 (7): p. 644-54.

153. Kew, J.N.C. and C.H. Davies, Ion channels : from structure to function . 2nd ed. 2010, Oxford ; New York: Oxford University Press. xiii, 562 p., 8 p. of plates.

154. Kochuveedu, S.T. and D.H. Kim, Surface plasmon resonance mediated photoluminescence properties of nanostructured multicomponent fluorophore systems. Nanoscale, 2014. 6(10): p. 4966-84.

155. Kopetzki, E., G. Schumacher, and P. Buckel, Control of formation of active soluble or inactive insoluble baker's yeast alpha-glucosidase PI in Escherichia coli by induction and growth conditions. Mol Gen Genet, 1989. 216 (1): p. 149-55.

156. Koradi, R., M. Billeter, and K. Wuthrich, MOLMOL: a program for display and analysis of macromolecular structures. J Mol Graph, 1996. 14 (1): p. 51-5, 29-32.

157. Korzhnev, D.M., N.R. Skrynnikov, O. Millet, D.A. Torchia, and L.E. Kay, An NMR experiment for the accurate measurement of heteronuclear spin-lock relaxation rates. J Am Chem Soc, 2002. 124 (36): p. 10743-53.

158. Kranz, J.K. and C. Schalk-Hihi, Protein thermal shifts to identify low molecular weight fragments. Methods Enzymol, 2011. 493 : p. 277-98.

159. Krishnarjuna, B., G. Jaipuria, A. Thakur, P. D'Silva, and H.S. Atreya, Amino acid selective unlabeling for sequence specific resonance assignments in proteins. J Biomol NMR, 2011. 49 (1): p. 39-51.

160. Kumar, R., Role of naturally occurring osmolytes in protein folding and stability. Arch Biochem Biophys, 2009. 491 (1-2): p. 1-6.

195

161. Kuo, A., J.M. Gulbis, J.F. Antcliff, T. Rahman, E.D. Lowe, J. Zimmer, J. Cuthbertson, F.M. Ashcroft, T. Ezaki, and D.A. Doyle, Crystal structure of the potassium channel KirBac1.1 in the closed state. Science, 2003. 300 (5627): p. 1922-6.

162. Kwan, A.H., M. Mobli, P.R. Gooley, G.F. King, and J.P. Mackay, Macromolecular NMR spectroscopy for the non-spectroscopist. FEBS J, 2011. 278 (5): p. 687-703.

163. Lajtha, A., G. Goracci, G. Tettamanti, and SpringerLink (Online service), Handbook of neurochemistry and molecular neurobiology . 2009, Springer US: Boston, MA. p. xv, 757 p.

164. Lakowicz, J.R., Principles of fluorescence spectroscopy . 3rd ed. 2006, New York: Springer. xxvi, 954 p.

165. Lakowicz, J.R., ed. Principles of Fluorescence Spectroscopy, 3rd Edition . 3 ed. 2006, Springer: New York.

166. Lange, C. and R. Rudolph, Suppression of protein aggregation by L-arginine. Curr Pharm Biotechnol, 2009. 10 (4): p. 408-14.

167. Lange, R., P. Anzenbacher, S. Muller, L. Maurin, and C. Balny, Interaction of tryptophan residues of cytochrome P450scc with a highly specific fluorescence quencher, a substrate analogue, compared to acrylamide and iodide. Eur J Biochem, 1994. 226 (3): p. 963-70.

168. Larkin, M.A., G. Blackshields, N.P. Brown, R. Chenna, P.A. McGettigan, H. McWilliam, F. Valentin, I.M. Wallace, A. Wilm, R. Lopez, J.D. Thompson, T.J. Gibson, and D.G. Higgins, Clustal W and Clustal X version 2.0. Bioinformatics, 2007. 23 (21): p. 2947-8.

169. Lawson, K., Potassium channel openers as potential therapeutic weapons in ion channel disease. Kidney Int, 2000. 57 (3): p. 838-45.

170. Lehrer, S.S., Solute perturbation of protein fluorescence. The quenching of the tryptophyl fluorescence of model compounds and of lysozyme by iodide ion. Biochemistry, 1971. 10 (17): p. 3254-63.

171. Lentzen, G. and T. Schwarz, Extremolytes: Natural compounds from extremophiles for versatile applications. Appl Microbiol Biotechnol, 2006. 72 (4): p. 623-34.

172. Leonoudakis, D., L.R. Conti, S. Anderson, C.M. Radeke, L.M. McGuire, M.E. Adams, S.C. Froehner, J.R. Yates, 3rd, and C.A. Vandenberg, Protein trafficking and anchoring complexes revealed by proteomic analysis of inward rectifier potassium channel (Kir2.x)- associated proteins. J Biol Chem, 2004. 279 (21): p. 22331-46.

173. Leonoudakis, D., L.R. Conti, C.M. Radeke, L.M. McGuire, and C.A. Vandenberg, A multiprotein trafficking complex composed of SAP97, CASK, Veli, and Mint1 is associated with inward rectifier Kir2 potassium channels. J Biol Chem, 2004. 279 (18): p. 19051-63.

196

174. LeRoith, D., S.I. Taylor, J.M. Olefsky, and Ovid Technologies Inc., Diabetes mellitus a fundamental and clinical text . 2004, Lippincott Williams & Wilkins: Philadelphia. p. xvi, 1540 p.

175. Lewis, H.A., S.G. Buchanan, S.K. Burley, K. Conners, M. Dickey, M. Dorwart, R. Fowler, X. Gao, W.B. Guggino, W.A. Hendrickson, J.F. Hunt, M.C. Kearins, D. Lorimer, P.C. Maloney, K.W. Post, K.R. Rajashankar, M.E. Rutter, J.M. Sauder, S. Shriver, P.H. Thibodeau, P.J. Thomas, M. Zhang, X. Zhao, and S. Emtage, Structure of nucleotide- binding domain 1 of the cystic fibrosis transmembrane conductance regulator. Embo J, 2004. 23 (2): p. 282-93.

176. Lewis, H.A., X. Zhao, C. Wang, J.M. Sauder, I. Rooney, B.W. Noland, D. Lorimer, M.C. Kearins, K. Conners, B. Condon, P.C. Maloney, W.B. Guggino, J.F. Hunt, and S. Emtage, Impact of the deltaF508 mutation in first nucleotide-binding domain of human cystic fibrosis transmembrane conductance regulator on domain folding and structure. J Biol Chem, 2005. 280 (2): p. 1346-53.

177. Li, C., M. Ramjeesingh, W. Wang, E. Garami, M. Hewryk, D. Lee, J.M. Rommens, K. Galley, and C.E. Bear, ATPase activity of the cystic fibrosis transmembrane conductance regulator. J Biol Chem, 1996. 271 (45): p. 28463-8.

178. Li, L., X. Geng, and P. Drain, Open state destabilization by ATP occupancy is mechanism speeding burst exit underlying KATP channel inhibition by ATP. J Gen Physiol, 2002. 119 (1): p. 105-16.

179. Liang, T., L. Xie, C. Chao, Y. Kang, X. Lin, T. Qin, H. Xie, Z.P. Feng, and H.Y. Gaisano, Phosphatidylinositol 4,5-biphosphate (PIP2) modulates interaction of syntaxin- 1A with sulfonylurea receptor 1 to regulate pancreatic beta-cell ATP-sensitive potassium channels. J Biol Chem, 2014. 289 (9): p. 6028-40.

180. Light, P.E., J.E. Manning Fox, M.J. Riedel, and M.B. Wheeler, Glucagon-like peptide-1 inhibits pancreatic ATP-sensitive potassium channels via a protein kinase A- and ADP- dependent mechanism. Mol Endocrinol, 2002. 16 (9): p. 2135-44.

181. Lin, Y.F., Y.N. Jan, and L.Y. Jan, Regulation of ATP-sensitive potassium channel function by protein kinase A-mediated phosphorylation in transfected HEK293 cells. EMBO J, 2000. 19 (5): p. 942-55.

182. Lin, Y.F. and Y. Chai, Functional modulation of the ATP-sensitive potassium channel by extracellular signal-regulated kinase-mediated phosphorylation. Neuroscience, 2008. 152 (2): p. 371-80.

183. Liu, R. and F.J. Sharom, Fluorescence studies on the nucleotide binding domains of the P-glycoprotein multidrug transporter. Biochemistry, 1997. 36 (10): p. 2836-43.

184. Liu, R., A. Siemiarczuk, and F.J. Sharom, Intrinsic fluorescence of the P-glycoprotein multidrug transporter: sensitivity of tryptophan residues to binding of drugs and nucleotides. Biochemistry, 2000. 39 (48): p. 14927-38.

197

185. Lo, M.C., A. Aulabaugh, G. Jin, R. Cowling, J. Bard, M. Malamas, and G. Ellestad, Evaluation of fluorescence-based thermal shift assays for hit identification in drug discovery. Anal Biochem, 2004. 332 (1): p. 153-9.

186. Locher, K.P., A.T. Lee, and D.C. Rees, The E. coli BtuCD structure: a framework for ABC transporter architecture and mechanism. Science, 2002. 296 (5570): p. 1091-8.

187. Locher, K.P., Review. Structure and mechanism of ATP-binding cassette transporters. Philos Trans R Soc Lond B Biol Sci, 2009. 364 (1514): p. 239-45.

188. Logan, J., D. Hiestand, P. Daram, Z. Huang, D.D. Muccio, J. Hartman, B. Haley, W.J. Cook, and E.J. Sorscher, Cystic fibrosis transmembrane conductance regulator mutations that disrupt nucleotide binding. J Clin Invest, 1994. 94 (1): p. 228-36.

189. Lopez-Alonso, J.P., E.D. de Araujo, and V. Kanelis, NMR and fluorescence studies of drug binding to the first nucleotide binding domain of SUR2A. Biochemistry, 2012. 51 (45): p. 9211-22.

190. Loubatieres-Mariani, M.M., [The discovery of hypoglycemic sulfonamides]. J Soc Biol, 2007. 201 (2): p. 121-5.

191. Loubatieres, A., The hypoglycemic sulfonamides: history and development of the problem from 1942 to 1955. Ann N Y Acad Sci, 1957. 71 (1): p. 4-11.

192. MacArthur, M.W. and J.M. Thornton, Influence of proline residues on protein conformation. J Mol Biol, 1991. 218 (2): p. 397-412.

193. MacGregor, G.G., K. Dong, C.G. Vanoye, L. Tang, G. Giebisch, and S.C. Hebert, Nucleotides and phospholipids compete for binding to the C terminus of KATP channels. Proc Natl Acad Sci U S A, 2002. 99 (5): p. 2726-31.

194. Manley, P.W., C. Loffler-Walz, U. Russ, A. Hambrock, T. Moenius, and U. Quast, Synthesis and characterization of a novel tritiated KATP channel opener with a benzopyran structure. Br J Pharmacol, 2001. 133 (2): p. 275-85.

195. Mannikko, R., C. Jefferies, S.E. Flanagan, A. Hattersley, S. Ellard, and F.M. Ashcroft, Interaction between mutations in the slide helix of Kir6.2 associated with neonatal diabetes and neurological symptoms. Hum Mol Genet, 2010. 19 (6): p. 963-72.

196. Marsh, E.N. and Y. Suzuki, Using (19)F NMR to probe biological interactions of proteins and peptides. ACS Chem Biol, 2014. 9(6): p. 1242-50.

197. Masia, R., D. Enkvetchakul, and C.G. Nichols, Differential nucleotide regulation of KATP channels by SUR1 and SUR2A. J Mol Cell Cardiol, 2005. 39 (3): p. 491-501.

198. Matsuo, M., N. Kioka, T. Amachi, and K. Ueda, ATP binding properties of the nucleotide-binding folds of SUR1. J Biol Chem, 1999. 274 (52): p. 37479-82.

198

199. Matsuo, M., K. Tanabe, N. Kioka, T. Amachi, and K. Ueda, Different binding properties and affinities for ATP and ADP among sulfonylurea receptor subtypes, SUR1, SUR2A, and SUR2B. J Biol Chem, 2000. 275 (37): p. 28757-63.

200. Matsushita, K., K. Kinoshita, T. Matsuoka, A. Fujita, T. Fujikado, Y. Tano, H. Nakamura, and Y. Kurachi, Intramolecular interaction of SUR2 subtypes for intracellular ADP-Induced differential control of K(ATP) channels. Circ Res, 2002. 90 (5): p. 554-61.

201. Mergny, J.L. and L. Lacroix, Analysis of thermal melting curves. Oligonucleotides, 2003. 13 (6): p. 515-37.

202. Mikhailov, M.V., J.D. Campbell, H. de Wet, K. Shimomura, B. Zadek, R.F. Collins, M.S. Sansom, R.C. Ford, and F.M. Ashcroft, 3-D structural and functional characterization of the purified KATP channel complex Kir6.2-SUR1. EMBO J, 2005. 24 (23): p. 4166-75.

203. Miki, T., K. Nagashima, and S. Seino, The structure and function of the ATP-sensitive K+ channel in insulin-secreting pancreatic beta-cells. J Mol Endocrinol, 1999. 22 (2): p. 113-23.

204. Miki, T., M. Suzuki, T. Shibasaki, H. Uemura, T. Sato, K. Yamaguchi, H. Koseki, T. Iwanaga, H. Nakaya, and S. Seino, Mouse model of Prinzmetal angina by disruption of the inward rectifier Kir6.1. Nat Med, 2002. 8(5): p. 466-72.

205. Miles, A.J. and B.A. Wallace, Synchrotron radiation circular dichroism spectroscopy of proteins and applications in structural and functional genomics. Chem Soc Rev, 2006. 35 (1): p. 39-51.

206. Minoretti, P., C. Falcone, A. Aldeghi, V. Olivieri, F. Mori, E. Emanuele, M. Calcagnino, and D. Geroldi, A novel Val734Ile variant in the ABCC9 gene associated with myocardial infarction. Clin Chim Acta, 2006. 370 (1-2): p. 124-8.

207. Monami, M., S. Genovese, and E. Mannucci, Cardiovascular safety of sulfonylureas: a meta-analysis of randomized clinical trials. Diabetes Obes Metab, 2013. 15 (10): p. 938- 53.

208. Moreau, C., H. Jacquet, A.L. Prost, N. D'Hahan, and M. Vivaudou, The molecular basis of the specificity of action of K(ATP) channel openers. EMBO J, 2000. 19 (24): p. 6644- 51.

209. Moreau, C., A.L. Prost, R. Derand, and M. Vivaudou, SUR, ABC proteins targeted by KATP channel openers. J Mol Cell Cardiol, 2005. 38 (6): p. 951-63.

210. Morita, E.H., M. Shimizu, T. Ogasawara, Y. Endo, R. Tanaka, and T. Kohno, A novel way of amino acid-specific assignment in (1)H-(15)N HSQC spectra with a wheat germ cell-free protein synthesis system. J Biomol NMR, 2004. 30 (1): p. 37-45.

211. Nagata, K., M. Nishitani, M. Matsuo, N. Kioka, T. Amachi, and K. Ueda, Nonequivalent nucleotide trapping in the two nucleotide binding folds of the human multidrug resistance protein MRP1. J Biol Chem, 2000. 275 (23): p. 17626-30.

199

212. Neuhofer, W. and F.X. Beck, Survival in hostile environments: strategies of renal medullary cells. Physiology (Bethesda), 2006. 21 : p. 171-80.

213. Niesen, F.H., H. Berglund, and M. Vedadi, The use of differential scanning fluorimetry to detect ligand interactions that promote protein stability. Nat Protoc, 2007. 2(9): p. 2212- 21.

214. Nikolaeva, O., G. Moiseyev, K.K. Rodgers, and J.X. Ma, Binding to lipid membrane induces conformational changes in RPE65: implications for its isomerohydrolase activity. Biochem J. 436 (3): p. 591-7.

215. Ninomiya, T., M. Takano, T. Haruna, Y. Kono, and M. Horie, Verapamil, a Ca2+ entry blocker, targets the pore-forming subunit of cardiac type KATP channel (Kir6.2). J Cardiovasc Pharmacol, 2003. 42 (2): p. 161-8.

216. Nishida, M., M. Cadene, B.T. Chait, and R. MacKinnon, Crystal structure of a Kir3.1- prokaryotic Kir channel chimera. EMBO J, 2007. 26 (17): p. 4005-15.

217. Nix, S.L., A.H. Chishti, J.M. Anderson, and Z. Walther, hCASK and hDlg associate in epithelia, and their src homology 3 and guanylate kinase domains participate in both intramolecular and intermolecular interactions. J Biol Chem, 2000. 275 (52): p. 41192- 200.

218. Noma, A., ATP-regulated K+ channels in . Nature, 1983. 305 (5930): p. 147-8.

219. Noskov, S.Y., S. Berneche, and B. Roux, Control of ion selectivity in potassium channels by electrostatic and dynamic properties of carbonyl ligands. Nature, 2004. 431 (7010): p. 830-4.

220. O'Grady, C., B.L. Rempel, A. Sokaribo, S. Nokhrin, and O.Y. Dmitriev, One-step amino acid selective isotope labeling of proteins in prototrophic Escherichia coli strains. Anal Biochem, 2012. 426 (2): p. 126-8.

221. Olson, T.M., A.E. Alekseev, C. Moreau, X.K. Liu, L.V. Zingman, T. Miki, S. Seino, S.J. Asirvatham, A. Jahangir, and A. Terzic, KATP channel mutation confers risk for vein of Marshall adrenergic atrial fibrillation. Nat Clin Pract Cardiovasc Med, 2007. 4(2): p. 110-6.

222. Ortiz, D., L. Gossack, U. Quast, and J. Bryan, Reinterpreting the action of ATP analogs on K(ATP) channels. J Biol Chem, 2013. 288 (26): p. 18894-902.

223. Pain, R.H., Determining the fluorescence spectrum of a protein. Curr Protoc Protein Sci, 2005. Chapter 7 : p. Unit 7 7.

224. Pantoliano, M.W., E.C. Petrella, J.D. Kwasnoski, V.S. Lobanov, J. Myslik, E. Graf, T. Carver, E. Asel, B.A. Springer, P. Lane, and F.R. Salemme, High-density miniaturized thermal shift assays as a general strategy for drug discovery. J Biomol Screen, 2001. 6(6): p. 429-40.

200

225. Park, S., B.B. Lim, C. Perez-Terzic, G. Mer, and A. Terzic, Interaction of asymmetric ABCC9-encoded nucleotide binding domains determines KATP channel SUR2A catalytic activity. J Proteome Res, 2008. 7(4): p. 1721-8.

226. Park, S. and A. Terzic, Quaternary structure of KATP channel SUR2A nucleotide binding domains resolved by synchrotron radiation X-ray scattering. J Struct Biol, 2010. 169 (2): p. 243-51.

227. Pasyk, E.A., Y. Kang, X. Huang, N. Cui, L. Sheu, and H.Y. Gaisano, Syntaxin-1A binds the nucleotide-binding folds of sulphonylurea receptor 1 to regulate the KATP channel. J Biol Chem, 2004. 279 (6): p. 4234-40.

228. Patel, A., S. MacMahon, J. Chalmers, B. Neal, L. Billot, M. Woodward, M. Marre, M. Cooper, P. Glasziou, D. Grobbee, P. Hamet, S. Harrap, S. Heller, L. Liu, G. Mancia, C.E. Mogensen, C. Pan, N. Poulter, A. Rodgers, B. Williams, S. Bompoint, B.E. de Galan, R. Joshi, and F. Travert, Intensive blood glucose control and vascular outcomes in patients with . N Engl J Med, 2008. 358 (24): p. 2560-72.

229. Pauli, G.F., T. Godecke, B.U. Jaki, and D.C. Lankin, Quantitative (1)h NMR. Development and potential of an analytical method: an update. J Nat Prod, 2012. 75 (4): p. 834-51.

230. Paulin, S., M. Jamshad, T.R. Dafforn, J. Garcia-Lara, S.J. Foster, N.F. Galley, D.I. Roper, H. Rosado, and P.W. Taylor, Surfactant-free purification of membrane protein complexes from bacteria: application to the staphylococcal penicillin-binding protein complex PBP2/PBP2a. Nanotechnology, 2014. 25 (28): p. 285101.

231. Pearson, R.B. and B.E. Kemp, Protein kinase phosphorylation site sequences and consensus specificity motifs: tabulations. Methods Enzymol, 1991. 200 : p. 62-81.

232. Pervushin, K., R. Riek, G. Wider, and K. Wuthrich, Attenuated T2 relaxation by mutual cancellation of dipole-dipole coupling and chemical shift anisotropy indicates an avenue to NMR structures of very large biological macromolecules in solution. Proc Natl Acad Sci U S A, 1997. 94 (23): p. 12366-71.

233. Philip-Couderc, P., N.I. Tavares, A. Roatti, R. Lerch, C. Montessuit, and A.J. Baertschi, Forkhead transcription factors coordinate expression of myocardial KATP channel subunits and energy metabolism. Circ Res, 2008. 102 (2): p. e20-35.

234. Picciotto, M.R., J.A. Cohn, G. Bertuzzi, P. Greengard, and A.C. Nairn, Phosphorylation of the cystic fibrosis transmembrane conductance regulator. J Biol Chem, 1992. 267 (18): p. 12742-52.

235. Pissarra, L.S., C.M. Farinha, Z. Xu, A. Schmidt, P.H. Thibodeau, Z. Cai, P.J. Thomas, D.N. Sheppard, and M.D. Amaral, Solubilizing mutations used to crystallize one CFTR domain attenuate the trafficking and channel defects caused by the major cystic fibrosis mutation. Chem Biol, 2008. 15 (1): p. 62-9.

201

236. Pollesello, P. and A. Mebazaa, ATP-dependent potassium channels as a key target for the treatment of myocardial and vascular dysfunction. Curr Opin Crit Care, 2004. 10 (6): p. 436-41.

237. Ponte-Sucre, A., ABC transporters in microorganisms : research, innovation and value as targets against drug resistance . 2009, Norfolk, UK: Caister Academic. xii, 260 p., 10 p. of plates.

238. Prades, C., I. Arnould, T. Annilo, S. Shulenin, Z.Q. Chen, L. Orosco, M. Triunfol, C. Devaud, C. Maintoux-Larois, C. Lafargue, C. Lemoine, P. Denefle, M. Rosier, and M. Dean, The human ATP binding cassette gene ABCA13, located on chromosome 7p12.3, encodes a 5058 amino acid protein with an extracellular domain encoded in part by a 4.8-kb conserved exon. Cytogenet Genome Res, 2002. 98 (2-3): p. 160-8.

239. Proks, P., F. Reimann, N. Green, F. Gribble, and F. Ashcroft, Sulfonylurea stimulation of insulin secretion. Diabetes, 2002. 51 Suppl 3 : p. S368-76.

240. Proks, P. and F.M. Ashcroft, Modeling K(ATP) channel gating and its regulation. Prog Biophys Mol Biol, 2009. 99 (1): p. 7-19.

241. Pufall, M.A., G.M. Lee, M.L. Nelson, H.S. Kang, A. Velyvis, L.E. Kay, L.P. McIntosh, and B.J. Graves, Variable control of Ets-1 DNA binding by multiple in an unstructured region. Science, 2005. 309 (5731): p. 142-5.

242. Quayle, J.M., M.T. Nelson, and N.B. Standen, ATP-sensitive and inwardly rectifying potassium channels in smooth muscle. Physiol Rev, 1997. 77 (4): p. 1165-232.

243. Quinn, K.V., J.P. Giblin, and A. Tinker, Multisite phosphorylation mechanism for protein kinase A activation of the smooth muscle ATP-sensitive K+ channel. Circ Res, 2004. 94 (10): p. 1359-66.

244. Rabeh, W.M., F. Bossard, H. Xu, T. Okiyoneda, M. Bagdany, C.M. Mulvihill, K. Du, S. di Bernardo, Y. Liu, L. Konermann, A. Roldan, and G.L. Lukacs, Correction of both NBD1 energetics and domain interface is required to restore DeltaF508 CFTR folding and function. Cell, 2012. 148 (1-2): p. 150-63.

245. Rai, S.S., K. Kuchroo, and S.R. Kasturi, Nucleotide binding to tubulin-investigations by nuclear magnetic resonance spectroscopy. Biochim Biophys Acta, 1996. 1292 (1): p. 77- 88.

246. Rainbow, R.D., M. James, D. Hudman, M. Al Johi, H. Singh, P.J. Watson, I. Ashmole, N.W. Davies, D. Lodwick, and R.I. Norman, Proximal C-terminal domain of sulphonylurea receptor 2A interacts with pore-forming Kir6 subunits in KATP channels. Biochem J, 2004. 379 (Pt 1): p. 173-81.

247. Ramaen, O., C. Sizun, O. Pamlard, E. Jacquet, and J.Y. Lallemand, Attempts to characterize the NBD heterodimer of MRP1: transient complex formation involves Gly771 of the ABC signature sequence but does not enhance the intrinsic ATPase activity. Biochem J, 2005. 391 (Pt 3): p. 481-90.

202

248. Ramaen, O., N. Leulliot, C. Sizun, N. Ulryck, O. Pamlard, J.Y. Lallemand, H. Tilbeurgh, and E. Jacquet, Structure of the human multidrug resistance protein 1 nucleotide binding domain 1 bound to Mg2+/ATP reveals a non-productive catalytic site. J Mol Biol, 2006. 359 (4): p. 940-9.

249. Reddy, K.R., H. Lilie, R. Rudolph, and C. Lange, L-Arginine increases the solubility of unfolded species of hen egg white lysozyme. Protein Sci, 2005. 14 (4): p. 929-35.

250. Rees, D.C., E. Johnson, and O. Lewinson, ABC transporters: the power to change. Nat Rev Mol Cell Biol, 2009. 10 (3): p. 218-27.

251. Remaut, H. and R.m. Fronzes, Bacterial membranes : structural and molecular biology . 2014, Norfolk: Caister Academic Press. xii, 500, 14 p.

252. Remedi, M.S., J.C. Koster, B.L. Patton, and C.G. Nichols, ATP-sensitive K+ channel signaling in glucokinase-deficient diabetes. Diabetes, 2005. 54 (10): p. 2925-31.

253. Ribalet, B., S.A. John, and J.N. Weiss, Regulation of cloned ATP-sensitive K channels by phosphorylation, MgADP, and phosphatidylinositol bisphosphate (PIP(2)): a study of channel rundown and reactivation. J Gen Physiol, 2000. 116 (3): p. 391-410.

254. Rodgers, D.S., Circular dichroism : theory and spectroscopy . 2012, New York: Nova Science Publishers. xi, 500 p.

255. Roosbeek, S., F. Peelman, A. Verhee, C. Labeur, H. Caster, M.F. Lensink, C. Cirulli, J. Grooten, C. Cochet, J. Vandekerckhove, A. Amoresano, G. Chimini, J. Tavernier, and M. Rosseneu, Phosphorylation by protein kinase CK2 modulates the activity of the ATP binding cassette A1 transporter. J Biol Chem, 2004. 279 (36): p. 37779-88.

256. Rust, H.L. and P.R. Thompson, Kinase consensus sequences: a breeding ground for crosstalk. ACS Chem Biol, 2011. 6(9): p. 881-92.

257. Ryabov, Y., C.D. Schwieters, and G.M. Clore, Impact of 15N R2/R1 relaxation restraints on molecular size, shape, and bond vector orientation for NMR protein structure determination with sparse distance restraints. J Am Chem Soc, 2011. 133 (16): p. 6154-7.

258. Sahu, S.C., A.K. Bhuyan, A. Majumdar, and J.B. Udgaonkar, Backbone dynamics of barstar: a (15)N NMR relaxation study. Proteins, 2000. 41 (4): p. 460-74.

259. Sali, A. and T.L. Blundell, Comparative protein modelling by satisfaction of spatial restraints. J Mol Biol, 1993. 234 (3): p. 779-815.

260. Salzmann, M., K. Pervushin, G. Wider, H. Senn, and K. Wuthrich, TROSY in triple- resonance experiments: new perspectives for sequential NMR assignment of large proteins. Proc Natl Acad Sci U S A, 1998. 95 (23): p. 13585-90.

261. Sampson, L.J., Y. Hayabuchi, N.B. Standen, and C. Dart, Caveolae localize protein kinase A signaling to arterial ATP-sensitive potassium channels. Circ Res, 2004. 95 (10): p. 1012-8.

203

262. Sansom, M.S., I.H. Shrivastava, J.N. Bright, J. Tate, C.E. Capener, and P.C. Biggin, Potassium channels: structures, models, simulations. Biochim Biophys Acta, 2002. 1565 (2): p. 294-307.

263. Sauer, M., J. Hofkens, and J. Enderlein, Handbook of fluorescence spectroscopy and imaging : from single molecules to ensembles . 2011, Weinheim: Wiley-VCH. ix, 281 p.

264. Sauna, Z.E. and S.V. Ambudkar, Evidence for a requirement for ATP hydrolysis at two distinct steps during a single turnover of the catalytic cycle of human P-glycoprotein. Proc Natl Acad Sci U S A, 2000. 97 (6): p. 2515-20.

265. Sauna, Z.E., M.M. Smith, M. Muller, K.M. Kerr, and S.V. Ambudkar, The mechanism of action of multidrug-resistance-linked P-glycoprotein. J Bioenerg Biomembr, 2001. 33 (6): p. 481-91.

266. Schlamadinger, D.E., D.I. Kats, and J.E. Kim, Quenching of Tryptophan Fluorescence in Unfolded Cytochrome c: A Biophysics Experiment for Physical Chemistry Students. Journal of Chemical Education, 2010. 87 (9): p. 961-964.

267. Schultz, B.D., R.A. Frizzell, and R.J. Bridges, Rescue of dysfunctional deltaF508-CFTR chloride channel activity by IBMX. J Membr Biol, 1999. 170 (1): p. 51-66.

268. Schwanstecher, M., C. Sieverding, H. Dorschner, I. Gross, L. Aguilar-Bryan, C. Schwanstecher, and J. Bryan, Potassium channel openers require ATP to bind to and act through sulfonylurea receptors. EMBO J, 1998. 17 (19): p. 5529-35.

269. See, R.H., R.A. Caday-Malcolm, R.R. Singaraja, S. Zhou, A. Silverston, M.T. Huber, J. Moran, E.R. James, R. Janoo, J.M. Savill, V. Rigot, L.H. Zhang, M. Wang, G. Chimini, C.L. Wellington, S.R. Tafuri, and M.R. Hayden, Protein kinase A site-specific phosphorylation regulates ATP-binding cassette A1 (ABCA1)-mediated phospholipid efflux. J Biol Chem, 2002. 277 (44): p. 41835-42.

270. Selivanov, V.A., A.E. Alekseev, D.M. Hodgson, P.P. Dzeja, and A. Terzic, Nucleotide- gated KATP channels integrated with creatine and adenylate kinases: amplification, tuning and sensing of energetic signals in the compartmentalized cellular environment. Mol Cell Biochem, 2004. 256-257 (1-2): p. 243-56.

271. Senior, A.E. and D.C. Gadsby, ATP hydrolysis cycles and mechanism in P-glycoprotein and CFTR. Semin Cancer Biol, 1997. 8(3): p. 143-50.

272. Senior, A.E., Catalytic mechanism of P-glycoprotein. Acta Physiol Scand Suppl, 1998. 643 : p. 213-8.

273. Senior, A.E. and S. Bhagat, P-glycoprotein shows strong catalytic cooperativity between the two nucleotide sites. Biochemistry, 1998. 37 (3): p. 831-6.

274. Senior, A.E., P. Gros, and I.L. Urbatsch, Residues in P-glycoprotein catalytic sites that react with the inhibitor 7-chloro-4-nitrobenzo-2-oxa-1,3-diazole. Arch Biochem Biophys, 1998. 357 (1): p. 121-5.

204

275. Serdyuk, I.N., N.R. Zaccai, and G. Zaccai, Methods in molecular biophysics : structure, dynamics, function . 2007, Cambridge ; New York: Cambridge University Press. xvi, 1120 p.

276. Sharom, F.J., The P-glycoprotein multidrug transporter. Essays Biochem, 2011. 50 (1): p. 161-78.

277. Sheppard, D.N. and M.J. Welsh, Effect of ATP-sensitive K+ channel regulators on cystic fibrosis transmembrane conductance regulator chloride currents. J Gen Physiol, 1992. 100 (4): p. 573-91.

278. Shi, Y., Z. Wu, N. Cui, W. Shi, Y. Yang, X. Zhang, A. Rojas, B.T. Ha, and C. Jiang, PKA phosphorylation of SUR2B subunit underscores vascular KATP channel activation by beta-adrenergic receptors. Am J Physiol Regul Integr Comp Physiol, 2007. 293 (3): p. R1205-14.

279. Shi, Y., X. Chen, Z. Wu, W. Shi, Y. Yang, N. Cui, C. Jiang, and R.W. Harrison, cAMP- dependent protein kinase phosphorylation produces interdomain movement in SUR2B leading to activation of the vascular KATP channel. J Biol Chem, 2008. 283 (12): p. 7523-30.

280. Shindo, T., M. Yamada, S. Isomoto, Y. Horio, and Y. Kurachi, SUR2 subtype (A and B)- dependent differential activation of the cloned ATP-sensitive K+ channels by pinacidil and nicorandil. Br J Pharmacol, 1998. 124 (5): p. 985-91.

281. Shyng, S., T. Ferrigni, and C.G. Nichols, Regulation of KATP channel activity by diazoxide and MgADP. Distinct functions of the two nucleotide binding folds of the sulfonylurea receptor. J Gen Physiol, 1997. 110 (6): p. 643-54.

282. Shyng, S. and C.G. Nichols, Octameric stoichiometry of the KATP channel complex. J Gen Physiol, 1997. 110 (6): p. 655-64.

283. Shyng, S.L., C.A. Cukras, J. Harwood, and C.G. Nichols, Structural determinants of PIP(2) regulation of inward rectifier K(ATP) channels. J Gen Physiol, 2000. 116 (5): p. 599-608.

284. Singh, S.M. and A.K. Panda, Solubilization and refolding of bacterial inclusion body proteins. J Biosci Bioeng, 2005. 99 (4): p. 303-10.

285. Smith, A.J., T.K. Taneja, J. Mankouri, and A. Sivaprasadarao, Molecular cell biology of KATP channels: implications for neonatal diabetes. Expert Rev Mol Med, 2007. 9(21): p. 1-17.

286. Smith, K.J., A.J. Chadburn, A. Adomaviciene, P. Minoretti, L. Vignali, E. Emanuele, and P. Tammaro, Coronary spasm and acute myocardial infarction due to a mutation (V734I) in the nucleotide binding domain 1 of ABCC9. Int J Cardiol, 2013. 168 (4): p. 3506-13.

205

287. Smith, K.J., A.J. Chadburn, A. Adomaviciene, P. Minoretti, L. Vignali, E. Emanuele, and P. Tammaro, Coronary spasm and acute myocardial infarction due to a mutation (V734I) in the nucleotide binding domain 1 of ABCC9. Int J Cardiol, 2013.

288. Smith, P.C., N. Karpowich, L. Millen, J.E. Moody, J. Rosen, P.J. Thomas, and J.F. Hunt, ATP binding to the motor domain from an ABC transporter drives formation of a nucleotide sandwich dimer. Mol Cell, 2002. 10 (1): p. 139-49.

289. Sousa, R., Use of glycerol, polyols and other protein structure stabilizing agents in protein crystallization. Acta Crystallogr D Biol Crystallogr, 1995. 51 (Pt 3): p. 271-7.

290. Stanfield, P.R., N.W. Davies, P.A. Shelton, M.J. Sutcliffe, I.A. Khan, W.J. Brammar, and E.C. Conley, A single aspartate residue is involved in both intrinsic gating and blockage by Mg2+ of the inward rectifier, IRK1. J Physiol, 1994. 478 ( Pt 1) : p. 1-6.

291. Stolarczyk, E.I., C.J. Reiling, and C.M. Paumi, Regulation of ABC transporter function via phosphorylation by protein kinases. Curr Pharm Biotechnol, 2011. 12 (4): p. 621-35.

292. Su, X.C., C.T. Loh, R. Qi, and G. Otting, Suppression of isotope scrambling in cell-free protein synthesis by broadband inhibition of PLP enymes for selective 15N-labelling and production of perdeuterated proteins in H2O. J Biomol NMR, 2011. 50 (1): p. 35-42.

293. Suzuki, M., R.A. Li, T. Miki, H. Uemura, N. Sakamoto, Y. Ohmoto-Sekine, M. Tamagawa, T. Ogura, S. Seino, E. Marban, and H. Nakaya, Functional roles of cardiac and vascular ATP-sensitive potassium channels clarified by Kir6.2-knockout mice. Circ Res, 2001. 88 (6): p. 570-7.

294. Sydow, S., J. Radulovic, F.M. Dautzenberg, and J. Spiess, Structure-function relationship of different domains of the rat corticotropin-releasing factor receptor. Brain Res Mol Brain Res, 1997. 52 (2): p. 182-93.

295. Szabo, K., G. Szakacs, T. Hegeds, and B. Sarkadi, Nucleotide occlusion in the human cystic fibrosis transmembrane conductance regulator. Different patterns in the two nucleotide binding domains. J Biol Chem, 1999. 274 (18): p. 12209-12.

296. Tabcharani, J.A., X.B. Chang, J.R. Riordan, and J.W. Hanrahan, Phosphorylation- regulated Cl- channel in CHO cells stably expressing the cystic fibrosis gene. Nature, 1991. 352 (6336): p. 628-31.

297. Tang, G., L. Wu, W. Liang, and R. Wang, Direct stimulation of K(ATP) channels by exogenous and endogenous hydrogen sulfide in vascular smooth muscle cells. Mol Pharmacol, 2005. 68 (6): p. 1757-64.

298. Terzic, A., A. Jahangir, and Y. Kurachi, Cardiac ATP-sensitive K+ channels: regulation by intracellular nucleotides and K+ channel-opening drugs. Am J Physiol, 1995. 269 (3 Pt 1): p. C525-45.

206

299. Thomas, P.J., P. Shenbagamurthi, X. Ysern, and P.L. Pedersen, Cystic fibrosis transmembrane conductance regulator: nucleotide binding to a synthetic peptide. Science, 1991. 251 (4993): p. 555-7.

300. Thompson, J.D., D.G. Higgins, and T.J. Gibson, CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res, 1994. 22 (22): p. 4673-80.

301. Toib, A., H.X. Zhang, T.J. Broekelmann, K.L. Hyrc, Q. Guo, F. Chen, M.S. Remedi, and C.G. Nichols, Cardiac specific ATP-sensitive K+ channel (KATP) overexpression results in embryonic lethality. J Mol Cell Cardiol, 2012. 53 (3): p. 437-45.

302. Toyoshima, C., S. Yonekura, J. Tsueda, and S. Iwasawa, Trinitrophenyl derivatives bind differently from parent adenine nucleotides to Ca2+-ATPase in the absence of Ca2+. Proc Natl Acad Sci U S A, 2011. 108 (5): p. 1833-8.

303. Tucker, S.J., F.M. Gribble, P. Proks, S. Trapp, T.J. Ryder, T. Haug, F. Reimann, and F.M. Ashcroft, Molecular determinants of KATP channel inhibition by ATP. EMBO J, 1998. 17 (12): p. 3290-6.

304. Tusnady, G.E., B. Sarkadi, I. Simon, and A. Varadi, Membrane topology of human ABC proteins. FEBS Lett, 2006. 580 (4): p. 1017-22.

305. Ubersax, J.A., E.L. Woodbury, P.N. Quang, M. Paraz, J.D. Blethrow, K. Shah, K.M. Shokat, and D.O. Morgan, Targets of the cyclin-dependent kinase Cdk1. Nature, 2003. 425 (6960): p. 859-64.

306. Ueda, K., J. Komine, M. Matsuo, S. Seino, and T. Amachi, Cooperative binding of ATP and MgADP in the sulfonylurea receptor is modulated by glibenclamide. Proc Natl Acad Sci U S A, 1999. 96 (4): p. 1268-72.

307. Uhde, I., A. Toman, I. Gross, C. Schwanstecher, and M. Schwanstecher, Identification of the potassium channel opener site on sulfonylurea receptors. J Biol Chem, 1999. 274 (40): p. 28079-82.

308. Urbatsch, I.L., G.A. Tyndall, G. Tombline, and A.E. Senior, P-glycoprotein catalytic mechanism: studies of the ADP-vanadate inhibited state. J Biol Chem, 2003. 278 (25): p. 23171-9.

309. Uysal, S., V. Vasquez, V. Tereshko, K. Esaki, F.A. Fellouse, S.S. Sidhu, S. Koide, E. Perozo, and A. Kossiakoff, Crystal structure of full-length KcsA in its closed conformation. Proc Natl Acad Sci U S A, 2009. 106 (16): p. 6644-9.

310. van Bon, B.W., C. Gilissen, D.K. Grange, R.C. Hennekam, H. Kayserili, H. Engels, H. Reutter, J.R. Ostergaard, E. Morava, K. Tsiakas, B. Isidor, M. Le Merrer, M. Eser, N. Wieskamp, P. de Vries, M. Steehouwer, J.A. Veltman, S.P. Robertson, H.G. Brunner, B.B. de Vries, and A. Hoischen, Cantu syndrome is caused by mutations in ABCC9. Am J Hum Genet, 2012. 90 (6): p. 1094-101.

207

311. Varadi, A., G. Szakacs, E. Bakos, and B. Sarkadi, P glycoprotein and the mechanism of multidrug resistance. Novartis Found Symp, 2002. 243 : p. 54-65; discussion 65-8, 180-5.

312. Vas, M., A. Merli, and G.L. Rossi, Antagonistic binding of substrates to 3- phosphoglycerate kinase monitored by the fluorescent analogue 2'(3')-O-(2,4,6- trinitrophenyl)adenosine 5'-triphosphate. Biochem J, 1994. 301 ( Pt 3) : p. 885-91.

313. Vasiliou, V., K. Vasiliou, and D.W. Nebert, Human ATP-binding cassette (ABC) transporter family. Hum Genomics, 2009. 3(3): p. 281-90.

314. Vedadi, M., F.H. Niesen, A. Allali-Hassani, O.Y. Fedorov, P.J. Finerty, Jr., G.A. Wasney, R. Yeung, C. Arrowsmith, L.J. Ball, H. Berglund, R. Hui, B.D. Marsden, P. Nordlund, M. Sundstrom, J. Weigelt, and A.M. Edwards, Chemical screening methods to identify ligands that promote protein stability, protein crystallization, and structure determination. Proc Natl Acad Sci U S A, 2006. 103 (43): p. 15835-40.

315. Verdon, G., S.V. Albers, B.W. Dijkstra, A.J. Driessen, and A.M. Thunnissen, Crystal structures of the ATPase subunit of the glucose ABC transporter from Sulfolobus solfataricus: nucleotide-free and nucleotide-bound conformations. J Mol Biol, 2003. 330 (2): p. 343-58.

316. Verdon, G., S.V. Albers, N. van Oosterwijk, B.W. Dijkstra, A.J. Driessen, and A.M. Thunnissen, Formation of the productive ATP-Mg2+-bound dimer of GlcV, an ABC- ATPase from Sulfolobus solfataricus. J Mol Biol, 2003. 334 (2): p. 255-67.

317. Viguera, A.R., J.L. Arrondo, A. Musacchio, M. Saraste, and L. Serrano, Characterization of the interaction of natural proline-rich peptides with five different SH3 domains. Biochemistry, 1994. 33 (36): p. 10925-33.

318. Vogel, H.J., Calcium-binding protein protocols . Methods in molecular biology. 2002, Totowa, N.J.: Humana Press.

319. Walsh, D.A. and S.M. Van Patten, Multiple pathway signal transduction by the cAMP- dependent protein kinase. FASEB J, 1994. 8(15): p. 1227-36.

320. Wan, L., X. Liang, and Y. Huang, Characterization of the ATPase activity of a novel chimeric fusion protein consisting of the two nucleotide binding domains of MRP1. Arch Biochem Biophys, 2009. 485 (2): p. 102-8.

321. Wang, C., N. Karpowich, J.F. Hunt, M. Rance, and A.G. Palmer, Dynamics of ATP- binding Cassette Contribute to Allosteric Control, Nucleotide Binding and Energy Transduction in ABC Transporters. J Mol Biol, 2004. 342 (2): p. 525-37.

322. Wang, F., S. Zeltwanger, S. Hu, and T.C. Hwang, Deletion of phenylalanine 508 causes attenuated phosphorylation-dependent activation of CFTR chloride channels. J Physiol, 2000. 524 Pt 3 : p. 637-48.

323. Wang, X., J. Wu, L. Li, F. Chen, R. Wang, and C. Jiang, Hypercapnic acidosis activates KATP channels in vascular smooth muscles. Circ Res, 2003. 92 (11): p. 1225-32.

208

324. Wang, X.T. and P.C. Engel, An optimised system for refolding of human glucose 6- phosphate dehydrogenase. BMC Biotechnol, 2009. 9: p. 19.

325. Ward, A., C.L. Reyes, J. Yu, C.B. Roth, and G. Chang, Flexibility in the ABC transporter MsbA: Alternating access with a twist. Proc Natl Acad Sci U S A, 2007. 104 (48): p. 19005-10.

326. Waugh, D.S., Genetic tools for selective labeling of proteins with alpha-15N-amino acids. J Biomol NMR, 1996. 8(2): p. 184-92.

327. Wen, P.C. and E. Tajkhorshid, Dimer opening of the nucleotide binding domains of ABC transporters after ATP hydrolysis. Biophys J, 2008. 95 (11): p. 5100-10.

328. Wheeler, A., C. Wang, K. Yang, K. Fang, K. Davis, A.M. Styer, U. Mirshahi, C. Moreau, J. Revilloud, M. Vivaudou, S. Liu, T. Mirshahi, and K.W. Chan, Coassembly of different sulfonylurea receptor subtypes extends the phenotypic diversity of ATP-sensitive potassium (KATP) channels. Mol Pharmacol, 2008. 74 (5): p. 1333-44.

329. Winter, M.C. and M.J. Welsh, Stimulation of CFTR activity by its phosphorylated R domain. Nature, 1997. 389 (6648): p. 294-6.

330. Wishart, D.S., C.G. Bigam, J. Yao, F. Abildgaard, H.J. Dyson, E. Oldfield, J.L. Markley, and B.D. Sykes, 1H, 13C and 15N chemical shift referencing in biomolecular NMR. J Biomol NMR, 1995. 6(2): p. 135-40.

331. Yamada, M., M. Ishii, H. Hibino, and Y. Kurachi, Mutation in nucleotide-binding domains of sulfonylurea receptor 2 evokes Na-ATP-dependent activation of ATP-sensitive K+ channels: implication for dimerization of nucleotide-binding domains to induce channel opening. Mol Pharmacol, 2004. 66 (4): p. 807-16.

332. Ye, B., S.L. Kroboth, J.L. Pu, J.J. Sims, N.T. Aggarwal, E.M. McNally, J.C. Makielski, and N.Q. Shi, Molecular identification and functional characterization of a mitochondrial sulfonylurea receptor 2 splice variant generated by intraexonic splicing. Circ Res, 2009. 105 (11): p. 1083-93.

333. Ye, S., Y. Li, L. Chen, and Y. Jiang, Crystal structures of a ligand-free MthK gating ring: insights into the ligand gating mechanism of K+ channels. Cell, 2006. 126 (6): p. 1161-73.

334. Yellon, D.M. and J.M. Downey, Preconditioning the myocardium: from cellular physiology to clinical cardiology. Physiol Rev, 2003. 83 (4): p. 1113-51.

335. Yuan, Y.R., S. Blecker, O. Martsinkevich, L. Millen, P.J. Thomas, and J.F. Hunt, The crystal structure of the MJ0796 ATP-binding cassette. Implications for the structural consequences of ATP hydrolysis in the active site of an ABC transporter. J Biol Chem, 2001. 276 (34): p. 32313-21.

209

336. Zaitseva, J., S. Jenewein, T. Jumpertz, I.B. Holland, and L. Schmitt, H662 is the linchpin of ATP hydrolysis in the nucleotide-binding domain of the ABC transporter HlyB. EMBO J, 2005. 24 (11): p. 1901-10.

337. Zaitseva, J., C. Oswald, T. Jumpertz, S. Jenewein, A. Wiedenmann, I.B. Holland, and L. Schmitt, A structural analysis of asymmetry required for catalytic activity of an ABC- ATPase domain dimer. EMBO J, 2006. 25 (14): p. 3432-43.

338. Zhang, L., L.A. Aleksandrov, Z. Zhao, J.R. Birtley, J.R. Riordan, and R.C. Ford, Architecture of the cystic fibrosis transmembrane conductance regulator protein and structural changes associated with phosphorylation and nucleotide binding. J Struct Biol, 2009. 167 (3): p. 242-51.

339. Zhao, Q., T. Kawano, H. Nakata, Y. Nakajima, S. Nakajima, and T. Kozasa, Interaction of G protein beta subunit with inward rectifier K(+) channel Kir3. Mol Pharmacol, 2003. 64 (5): p. 1085-91.

340. Zhou, M., H.J. He, O. Tanaka, R. Suzuki, M. Sekiguchi, Y. Yasuoka, K. Kawahara, H. Itoh, and H. Abe, Localization of the sulphonylurea receptor subunits, SUR2A and SUR2B, in rat renal tubular epithelium. Tohoku J Exp Med, 2008. 214 (3): p. 247-56.

341. Zhou, Y., P. Ojeda-May, and J. Pu, H-loop histidine catalyzes ATP hydrolysis in the E. coli ABC-transporter HlyB. Phys Chem Chem Phys, 2013. 15 (38): p. 15811-5.

342. Zingman, L.V., A.E. Alekseev, M. Bienengraeber, D. Hodgson, A.B. Karger, P.P. Dzeja, and A. Terzic, Signaling in channel/enzyme multimers: ATPase transitions in SUR module gate ATP-sensitive K+ conductance. Neuron, 2001. 31 (2): p. 233-45.

343. Zingman, L.V., D.M. Hodgson, P.H. Bast, G.C. Kane, C. Perez-Terzic, R.J. Gumina, D. Pucar, M. Bienengraeber, P.P. Dzeja, T. Miki, S. Seino, A.E. Alekseev, and A. Terzic, Kir6.2 is required for adaptation to stress. Proc Natl Acad Sci U S A, 2002. 99 (20): p. 13278-83.

344. Zingman, L.V., D.M. Hodgson, M. Bienengraeber, A.B. Karger, E.C. Kathmann, A.E. Alekseev, and A. Terzic, Tandem function of nucleotide binding domains confers competence to sulfonylurea receptor in gating ATP-sensitive K+ channels. J Biol Chem, 2002. 277 (16): p. 14206-10.

210

Copyright Acknowledgments

All figures adapted or reproduced from previously published work have been provided with the appropriate copyright clearance approval and are accordingly acknowledged/cited in the corresponding caption of each adapted figure as per the specified guidelines of each publisher. The full citation is also provided in the References section. Additionally, an online doi/link of the original journal article from where each figure was adapted/reproduced has also been provided below.

Chapter 1: Figure 1.1: doi:10.1038/sj.emboj.7600877 Figure 1.2: doi:10.1038/nrn1244 Figure 1.3: doi:10.1128/MMBR.00031-07 Figure 1.4: doi:10.1152/physrev.00027.2009 Figure 1.5: doi:10.1038/nrm2646 Figure 1.6: doi:10.1074/jbc.C300363200 Figure 1.7: doi:10.1038/nature04712 Figure 1.8: doi:10.1152/physrev.00027.2009 Figure 1.9: doi:10.1152/physrev.00027.2009 Figure 1.10: doi:10.1085/jgp.200709874 Figure 1.11: doi:10.1152/physrev.00027.2009 Figure 1.12: doi:10.1152/ajpendo.00348.2007 Figure 1.15 doi:10.1039/C4NR00241E Figure 1.16 doi:10.1039/B316168B

Chapter 2 Chapter 2 is a combination of two publications. The permission from the publisher was obtained in each case. Reprinted with permission from Biochemistry, de Araujo, E.D., Ikeda, L.K., Tzvetkova, S., and Kanelis, V. The First Nucleotide Binding Domain of the Sulfonylurea Receptor 2A Contains Regulatory Elements and Is Folded and Functions as an Independent Module. 50:6655-66. Copyright (2011) American Chemical Society. Available from: http://pubs.acs.org/doi/abs/10.1021/bi200434d

211

Reprinted from Protein Purification and Expression, 103/November, de Araujo, E.D., Kanelis V., Successful development and use of a thermodynamic stability screen for optimizing the yield of nucleotide binding domains, 38-47, Copyright (2014), with permission from Elsevier. Available from: http://www.sciencedirect.com/science/article/pii/S1046592814001818

Chapter 4: Chapter 4 has been accepted to the Journal of Biological Chemistry at the time of thesis submission. A link to the current online version from JBC is provided. Available from: http://www.jbc.org/content/early/2015/07/21/jbc.M114.636233.abstract

Chapter 7: Figure 7.2 doi: 10.1085/jgp.200709852.

Chapter 8/Appendix: Chapter 8/Appendix was published in Biochemistry. The permission from the publisher was appropriately obtained. Reprinted from Biochemistry, Lopez-Alonso, J.P., de Araujo E.D., and Kanelis V. (2012) NMR and fluorescence studies of drug binding to the first nucleotide binding domain of SUR2A. 51: 9211-9222. Copyright (2012) American Chemical Society. Available from: http://pubs.acs.org/doi/abs/10.1021/bi301019e