Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Development of methods for study of membrane proteins in the absence of detergents. Carlson, Michael 2018

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2018_november_carlson_michael.pdf [ 5.66MB ]
Metadata
JSON: 24-1.0371907.json
JSON-LD: 24-1.0371907-ld.json
RDF/XML (Pretty): 24-1.0371907-rdf.xml
RDF/JSON: 24-1.0371907-rdf.json
Turtle: 24-1.0371907-turtle.txt
N-Triples: 24-1.0371907-rdf-ntriples.txt
Original Record: 24-1.0371907-source.json
Full Text
24-1.0371907-fulltext.txt
Citation
24-1.0371907.ris

Full Text

i   Development of methods for study of membrane proteins in the absence of detergents.   by  Michael Carlson   B.Sc., University of Victoria, 2012      A DISSERTATION SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY  in  THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES   (Biochemistry and Molecular Biology)    THE UNIVERSITY OF BRITISH COLUMBIA  (Vancouver)    August 2018        © Michael Carlson 2018 ii   Committee Page  The following individuals certify that they have read, and recommend to the Faculty of Graduate and Postdoctoral Studies for acceptance, the dissertation entitled:  Development of methods for study of membrane proteins in the absence of detergents.  submitted by Michael Carlson  in partial fulfillment of the requirements for  the degree of  Doctor of Philosophy  in     Biochemistry and Molecular Biology  Examining Committee:  Franck Duong, Biochemistry and Molecular Biology Supervisor  Nobu Tokuriki, Biochemistry and Molecular Biology Supervisory Committee Member  J. Thomas Beatty, Microbiology and Immunology Supervisory Committee Member  Natalie Strynadka, Biochemistry and Molecular Biology University Examiner  Dr. Suzanna Straus, Chemistry University Examiner         iii   Abstract  Membrane proteins are sequestered in a hydrophobic lipid environment, and are therefore resistant to characterization by traditional biochemical techniques which occur in aqueous solution. Traditionally, detergents have been used to solubilize membrane proteins, but these surfactants have detrimental effects on protein form and function.  In this thesis, the form, function, and potential mechanism of membrane protein complexes are investigated in detergent-free buffer.  Our first study is motivated by the lack of flexible reconstitution scaffolds that maintain a lipid-protein environment for membrane protein stabilization. To this end, we design the peptidisc, a simple  method for the universal stabilization of membrane protein complexes in detergent-free solution. Analysis of 5 different membrane protein complexes reconstituted in peptidiscs demonstrate that the method maintains function and increases stability of incorporated membrane proteins.  We extend the method to reconstitute the entire membrane proteome of the bacterium Escherichia coli, demonstrating the peptidisc as a “one-size fits all” scaffold.  We measure the co-fractionation of reconstituted proteins to identify proteins complexes captured in the peptidisc. The method provides a high-throughput, detergent free approach for identifying the protein-protein interactions of a biological membrane.  In the final study, we discover that conformational flexibility of the maltose importer MalFGK2 chloride channel activity in the maltose importer MalFGK2 is linked directly to the transient chloride channel activity. This finding provides support for the expanded alternating access model, which include discrete substeps in the transport model. I discuss the ramifications of the expanded alternating access model on membrane transporter and channel evolution and function. In addition, the development of the peptidisc method is discussed in the broader context of current, detergent free methods for analysis of membrane protein structure, function and interactions.           iv   Lay Summary      Membrane proteins are medically important: they represent over 50% of current drug targets.  However, researchers need special tools to manipulate membrane proteins so that effective drugs may be designed.  Here, I describe a simple, scalable method for the rapid stabilization of membrane proteins called the peptidisc.  We apply the peptidisc to mass spectrometry based protein identification to measures all the different membrane protein interactions (membrane protein interactome) of captured in the peptidisc at once.  As the peptidisc method is very gentle, we can isolate very fragile interactions that are missed by other methods.  Finally, we investigate how cells can selectively move molecules in and out of the cell, and the resulting consequences of this action on membrane permeability. We do this by making a dysfunctional version of a membrane protein that allows accidental leakage of salt into the cell.  We discuss how our findings can provide insight into the possible evolutionary link between two ubiquitous classes of proteins, transporters and channels.                           v   Preface  The following describes publications and manuscripts in preparation or revision that were used as a basis for the text and figures in this thesis. The contributions of supporting authors in each article are described.   Chapter 1: Permission was obtained from the respective journals to reuse information, and is indicated where applicable.   Chapter 2: This study is taken from a manuscript written jointly by myself and my supervisor Dr. Franck Duong.  The manuscript is currently under revision with the title:  “The Peptidisc, a Simple Method for Stabilizing Membrane Proteins in Detergent-free Solution”.  Michael L. Carlson1, John W. Young1, Zhiyu Zhao1, Lucien Fabre2†, Daniel Jun3, Jianqing Li4, Jun Li4, Harveer S. Dhupar1, Irvin Wason1., Allan T. Mills1††, J. Thomas Beatty3, John S. Klassen4, Isabelle Rouiller2, Franck Duong1*.       In this manuscript I conceived and designed all experiments.  I performed the experiments described Figure 2-1(panels a, b and d), Figure 2-3, Figure 2-4, Supplementary Figure 2-1, Supplementary Figure 2-3, Supplementary Figure 2-5, Supplementary Figure 2-6, and Supplementary Figure 2-8. In Figure 2-1c, I purified and reconstituted the MalFGK2 peptidisc, which was imaged by Dr. Lucien Fabre in the group of Dr. Isabelle Rouiller at the University of McGill.  In Figure 2-2, I analyzed and presented the data, while Zhiyu Zhao (Duong lab, University of British Columbia) performed the in-gel reconstitutions. I purified the MalFGK2 protein, while the other proteins were initially purified by other members of the Duong laboratory at the University of British Columbia; John Young (YEG), Allan Mills (FhuA, OmpF), or Harveer Dhupar (BRC).  In Supplementary Figure 2-4, I performed all reconstitutions and delivered the protein sample to Jianqing Li, and Jun Li who performed the native mass spectrometry experiments in the lab of Dr. John Klassen at the University of Alberta.  In Supplementary Figure 2-2, Irvin Wason (Duong lab, University of British Columbia) performed all gels and blots.  I analyzed the gels and presented the data.  vi   Chapter 3:  This study was originally conceived with the help of Dr. Nichollas Scott, Dr. Leonard Foster and Dr. Franck Duong.  The study is from a manuscript in preparation, written jointly with Dr. Greg Stacey.I performed the experiments described in Figure 3-1. 3-2, 3-3, 3-6, 3-7 and Supplementary Figure 3-1 and 3-2.  I performed all Maxquant data analyses of the peptidisc libraries under the guidance of Kyun Lee (Jenny) Moon and Dr. Nicholas Scott (Foster lab, University of British Columbia).  Dr. Greg Stacey (Foster lab, University of British Columbia) performed all co-fractionation data analysis and computational validation described in Figure 3-4 and 3-5.  Supplementary Figure 3-2 was supplied by John Young (Duong lab, University of British Columbia).  Dr. Nik Stoynov and Dr. Jason Rogalski (Foster lab, University of British Columbia) operated the mass spectrometers used to identify the tryptic peptides.               Chapter 4:  This study was conceived by myself, my supervisor Dr. Franck Duong, and Dr. Huan Bao, and is modified from a manuscript I wrote jointly with my supervisor, Dr. Franck Duong.  The manuscript is published under the title: “Formation of a chloride-conducting state in the maltose ATP-binding cassette (ABC) transporter” Michael L. Carlson, Huan Bao and Franck Duong.  The article was published in volume 291, issue 23, corresponding to pages 1119-25, in the Journal of Biological Chemistry on April 7, 2016.   I purified all proteins and performed all experiments produced as figures in this study.                   vii   Table of Contents Abstract ................................................................................................................................................... iii Lay Summary ...................................................................................................................................... iv Preface ...................................................................................................................................................... v Table of Contents .............................................................................................................................. vii List of Tables ....................................................................................................................................... xii List of Figures .................................................................................................................................... xiii List of Symbols, Abbreviations and Terms ......................................................................... xv Acknowledgements ........................................................................................................................ xvii Dedication .......................................................................................................................................... xviii Chapter 1:  Introduction ................................................................................................................. 1 1.1   Challenges of understanding membrane protein and contributions of this work thereof ................. 1 1.2 Membrane mimetics ............................................................................................................................ 2 1.2.1 Detergents .................................................................................................................................... 3     1.2.1.1 Mechanism of detergent solubilization  ................................................................................ 4     1.2.1.2 Disadvantages of detergents for membrane protein solubilization  ...................................... 6 1.2.2 Protein based reconstitution systems ........................................................................................... 8     1.2.2.1 Nanodiscs .............................................................................................................................. 8     1.2.2.2 Saposin lipoparticles ........................................................................................................... 10 1.2.3 Synthetic polymer based reconstitution systems ........................................................................ 13     1.2.3.1 Styrene maleic acid lipoparticles ........................................................................................ 14     1.2.3.2 Amphipols ........................................................................................................................... 15 1.2.4 Peptide based reconstitution systems ......................................................................................... 16         1.2.4.1 Peptergents .......................................................................................................................... 16         1.2.4.2 Lipopeptides ........................................................................................................................ 18         1.2.4.3 Nanostructured beta-sheet peptides..................................................................................... 20         1.2.4.4 ApoA1 mimetic peptides .................................................................................................... 21 1.3 Methods to examine protein-protein interaction networks ............................................................... 23 1.3.1 Genetic approaches .................................................................................................................... 24 1.3.2 Proteomic approaches ................................................................................................................ 24     1.4 Stable amino acid isotopologue labelling in cell culture (SILAC) ................................................... 27 viii       1.5 Prediction of interactomes from co-elution data (PrInCE) ............................................................... 29 1.6 On the structure and interactome of the Escherichia coli (E. coli) cell envelope ............................. 32 1.7 Channels and Transporters ................................................................................................................ 33 1.7.1 Canonical alternating access model ........................................................................................... 33 1.7.2 Ion and water leakage in alternating access transporters ........................................................... 36 1.7.3 The expanded alternating access model ..................................................................................... 36     1.8 Coupling of conformational changes with enzyme activity of MalFGK2 ......................................... 39 1.9 Overview of Objectives .................................................................................................................... 40 Chapter 2:  Design and validation of the Peptidisc method, a simple yet efficient procedure for stabilizing membrane proteins in detergent-free buffer .................................................................................................................................................................... 42 2.1 Introduction ..................................................................................................................................... 42 2.2 Materials and Methods ................................................................................................................... 44 2.2.1 Plasmids and biological reagents ............................................................................................... 44 2.2.2 Preparation of the Nanodisc Scaffold Peptides (NSPs) ............................................................. 44 2.2.3  On-column reconstitution.......................................................................................................... 45 2.2.4  In-gel reconstitution .................................................................................................................. 45 2.2.5  On-bead reconstitution .............................................................................................................. 45 2.2.6  Reconstitution of MalFGK2 and BRC into proteoliposomes .................................................... 46 2.2.7  Reconstitution of BRC in peptidiscs, low-lipid nanodiscs, styrene maleic acid nanoparticles,                DDM and SDS. ......................................................................................................................... 46 2.2.8   FhuA binding assay .................................................................................................................. 46 2.2.9   Absorbance spectroscopy ......................................................................................................... 47 2.2.10 NSPr quantification .................................................................................................................. 47 2.2.11 Lipid extraction and quantification .......................................................................................... 48 2.3 Results .............................................................................................................................................. 49 2.3.1 Solubility comparison of NSP and NSPr ................................................................................... 49 2.3.2 On column reconstitution of MalFGK2 ...................................................................................... 49 2.3.3 Structure, composition and functional activity of MalFGK2 peptidisc ...................................... 50 2.3.4 Composition and functional activity of FhuA peptidisc ............................................................ 52 2.3.5 In gel reconstitution screening of optimal reconstitution ratios (RR50) for MalFGK2, OmpF3,           Sec(YEG)n, and FhuA ................................................................................................................ 52 2.3.6 On bead reconstitution of MalFGK2 into peptidisc .................................................................... 55 2.3.7 The peptidisc increases thermostability of BRC ........................................................................ 57 ix   2.3.8 Comparison of peptidisc to other membrane mimetics for BRC stabilization .......................... 58 2.4 Discussion......................................................................................................................................... 58 2.4.1 The peptidisc methodology is simple and seamlessly integrates into existing membrane protein           purification schemes .................................................................................................................. 59 2.4.2 The peptidisc stabilizes membrane proteins in their functional states ....................................... 59 2.4.3 The peptidisc fills a unique role in comparison to other membrane mimetic systems............... 60 2.4.4 Association of peptide in the peptidisc ...................................................................................... 61 2.4.5 Perspective on the peptidisc method .......................................................................................... 61 Chapter 3:  A high-throughput approach to identify protein complexes of the E. coli cell envelope in detergent-free buffer. ..................................................................... 63 3.1 Introduction ..................................................................................................................................... 63 3.2 Materials and Methods ................................................................................................................... 66 3.2.1 Plasmids and biological reagents ............................................................................................... 66 3.2.2 Preparation of SILAC labelled, E. coli crude membranes. ........................................................ 67 3.2.3 Protein expression and purification ............................................................................................ 67 3.2.4 Preparation of E. coli membrane solubilized in peptidisc libraries............................................ 69 3.2.5 Preparation of E. coli cell membrane solubilized in styrene maleic acid (SMA). ..................... 69 3.2.6 Fractionation of cell envelope libraries in detergent free buffer ................................................ 70 3.2.7 In solution digestion of fractionated samples............................................................................. 70 3.2.8 Liquid chromatography and mass spectrometry analysis (LC-MS/MS) .................................... 71 3.2.9 Binary interaction identification by PrInCE and complex assignment ...................................... 72 3.3 Results .............................................................................................................................................. 73 3.3.1 Capture of the E. coli membrane proteome in peptidisc ............................................................ 73 3.3.2 Fractionation of the SILAC-labeled peptidisc library ................................................................ 77 3.3.3 Large membrane protein complexes are captured in the peptidisc library................................. 78 3.3.4 Prediction of binary interactions from co-elution data (PrInCE) ............................................... 79 3.3.5 Computational validation of binary interactions ........................................................................ 81 3.3.6 Assignment and validation of protein complexes ...................................................................... 82 3.3.7  Validation of binary interactions by affinity purification mass spectrometry (AP-MS) ........... 83 3.3.8  Identification of a unique Type I ATP Binding Cassette (ABC) transporter complex ............. 87 3.4 Discussion......................................................................................................................................... 88 3.4.1 Comparison to other detergent-free membrane solubilization methods. ................................... 89 3.4.2 Quality of the peptidisc reconstituted E. coli cell envelope interactome ................................... 89 x   3.4.3  Methods to increase number of predicted interactions by peptidisc PCP-SILAC .................... 90 3.4.4 Advantages of the peptidisc PCP-SILAC method ..................................................................... 91 3.4.5 Peptidisc PCP-SILAC can detect interactions not available in detergent AP-MS experiments 91 3.4.6 Comparison between ABC transporters in E. coli reveals a potential new mechanism for stable           complex formation. .................................................................................................................... 92 3.4.7 SEC-PCP-SILAC of peptidisc solubilized membranes is a high-throughput approach for           generating high quality membrane interactomes. ...................................................................... 93 Chapter 4:  Identification of chloride channel activity in the maltose transporter MalFGK2 ..................................................................................................................... 94 4.1 Introduction ..................................................................................................................................... 94 4.2 Materials and Methods ................................................................................................................... 95 4.2.1 Plasmids and biological reagents ............................................................................................... 95 4.2.2 Cysteine cross-linking ................................................................................................................ 96 4.2.3 Incorporation of proteins in proteoliposomes ............................................................................ 96 4.2.4 Spheroplast membrane permeability assays............................................................................... 97 4.2.5 Planar lipid bilayer experiments ................................................................................................ 97 4.3 Results .............................................................................................................................................. 98 4.3.1 MalF500 can readily access the transition state ......................................................................... 98 4.3.2 MalF500 is deleterious to the cell .............................................................................................. 99 4.3.3 MalF500 is permeable to chloride ........................................................................................... 102 4.3.4 Ion channel activity of MalF500 .............................................................................................. 103 4.3.5 The periplasmic gate seals MalF500 ........................................................................................ 105 4.3.6 MalF500GGG forms a quasi-permanently open channel ........................................................... 107 4.4 Discussion....................................................................................................................................... 109 4.4.1 Insights into the expanded alternating access model from chloride leakage in MalF500 ........ 109 4.4.2 Role of the periplasmic gate of MalFGK2 in prevention of chloride conductance .................. 110         4.4.3 Insights into the evolution of the ABC transporter turned chloride channel CFTR. ................ 110 Chapter 5:  General Conclusions and Future Work .................................................... 112 References ........................................................................................................................................... 122 Appendices ......................................................................................................................................... 134 Appendix 2A:  Additional Methods - Chapter 2 ................................................................................... 134 2A.1 Protein expression and purification .......................................................................................... 134 2A.2 Native gel electrophoresis ........................................................................................................ 136 xi   2A.3 Dynamic and static light scattering analysis ............................................................................ 136 2A.4 Sample preparation and EM image acquisition........................................................................ 136 2A.5 EM data processing and image analysis ................................................................................... 137 2A.6 Mass spectrometry ................................................................................................................... 137 2A.7 Other methods .......................................................................................................................... 138 Appendix 2B:  Supplementary Figures and Tables - Chapter 2 ............................................................ 138 Appendix 3A: Supplementary Figures and Tables – Chapter 3 ............................................................ 150                                xii   List of Tables Table 1-1 Comparison of membrane mimetics for membrane protein solubilization. .................. 6      Supplemental Table 2-1 Calculated and measured molecular weights of peptidiscs. .............. 147 Supplemental Table 2-2 Molecular weight, diameter, and scaffold stoichiometry of MalFGK2  reconstituted in peptidisc (NSPr) as determined by SEC-MALs and negative stain electron  microscopy………………….. .................................................................................................... 147 Supplementary Table 2-3 Amphipathic scaffolds used in this study. ...................................... 148 Supplementary Table 2-4 Native gel buffer recipes. ................................................................ 149 Supplementary Table 3-1 List of deposited co-fractionation data, interaction lists and complex  assignment from Chapter 3..... .................................................................................................... 153 Supplementary Table 3-2 List of gene ontology terms used to predict protein association with the E. coli cell envelope. ............................................................................................................. 154                               xiii   List of Figures Figure 1-1    Mechanism of micelle formation by amphiphilic detergents .................................... 4 Figure 1-2    The three stages of detergent solubilization of a lipid bilayer with increasing titration of detergent ........................................................................................................................ 6 Figure 1-3    Methods for nanodisc formation ............................................................................... 9 Figure 1-4    Modelled examples of saposin A reconstituted nanoparticles. ............................... 12 Figure 1-5    Chemical structure of synthetic polymer scaffolds for detergent free stabilization of membrane proteins. ....................................................................................................................... 13 Figure 1-6    Molecular models of peptide detergents at neutral pH ........................................... 17 Figure 1-7    Structural comparison of traditional detergent micelle, lipid bilayer, and lipopeptide micelle ........................................................................................................................................... 19 Figure 1-8    Structures of β-strand peptides designed to stabilize integral membrane proteins . 20 Figure 1-9    Protein purification strategies for MS protein-protein interactions (PPI). .............. 25 Figure 1-10  Overview of SILAC. ............................................................................................... 29   Figure 1-11  PrInCE overview ..................................................................................................... 31 Figure 1-12  The one vs two gate model distinction between channels and transporters ............ 35   Figure 1-13  Models for the conformational movements of an ABC transporter due to nucleotide binding as described by the alternating access model. ................................................................. 37 Figure 1-14  Inward (apo) and outward (ATP-bound) structures of the E. coli maltose importer MalFGK2. ...................................................................................................................................... 38   Figure 2-1    The “on-column” reconstitution of MalFGK2. ........................................................ 51 Figure 2-2    Express “in-gel” method for determining optimal reconstitution ratio. .................. 54 Figure 2-3    Direct “on-beads” reconstitution during membrane protein purification. ............... 56 Figure 2-4    Thermostability of the BRC complex in peptidiscs. ............................................... 57 Figure 3-1    Overview of the peptidisc-based SEC-PCP-SILAC workflow. .............................. 74 Figure 3-2    The peptidisc captures detergent solubilized membrane proteins with high efficiency....................................................................................................................................... 76   Figure 3-3    Proteomic analysis of soluble, SILAC labelled E. coli membrane proteins in SMALPs or peptidisc libraries. ..................................................................................................... 79  Figure 3-4    Interaction properties of peptidisc PCP-SILAC. ..................................................... 80 xiv   Figure 3-5    Comparison between the peptidisc interactome and the Babu et al. 2017 cell envelope interactome. ................................................................................................................... 82 Figure 3-6    Validation of predicted interactors by AP-MS. ....................................................... 86  Figure 3-7    Co-fractionation profiles of ABC transporter-SBP complexes in peptidiscs. ......... 88 Figure 4-1    Conformation of MalF500. ..................................................................................... 99 Figure 4-2    Activity of MalF500. ............................................................................................. 101 Figure 4-3    Permeability of MalF500 in spheroplast assays. ................................................... 103 Figure 4-4    Ion channel activity of MalF500 in planar lipid bilayers. ..................................... 104 Figure 4-5    The periplasmic gate seals MalF500. .................................................................... 106 Figure 4-6    Ion channel activity of MalF500GGG. ................................................................. 108 Supplemental Figure 2-1  Solubility tests of NSP and NSPr. .................................................. 139 Supplemental Figure 2-2  Quantification of NSPr content in peptidiscs. ................................ 140 Supplemental Figure 2-3  Quantification of phospholipids trapped in peptidiscs .................. .141 Supplemental Figure 2-4  Native MS of intact peptidiscs. ....................................................... 142 Supplemental Figure 2-5  Structural stability of MalFGK2 peptidisc. ..................................... 143 Supplemental Figure 2-6  Binding activity of FhuA in nanodiscs and peptidiscs. .................. 144 Supplemental Figure 2-7  Capture of SecYEG monomer and dimer in peptidisc and nanodisc...................................................................................................................................................... 145  Supplemental Figure 2-8  Effect of peptidisc on BRC stability. .............................................. 146 Supplemental Figure 3-1  Fractionation profiles for select E. coli membrane protein complexes solubilized in SMA and peptidisc. .............................................................................................. 150 Supplemental Figure 3-2  Purification of YfgM, PpiD, and MipA. ......................................... 151 Supplemental Figure 3-3  Co-elution of SecYEG. ................................................................... 152        xv   List of Symbols, Abbreviations and Terms Å: Angstrom 10-10 metres ATP: Adenosine tri-phosphate ApoA1: Apolipoprotein sub-type A1 BAM:  β-barrel assembly machinery BL21: E. coli strain suitable for protein over-expression BME: β-mercaptoethanol, a reducing agent BRC:  Bacterial Reaction Centre, a membrane protein complex composed of the PufM, PufL, and PuhA proteins from Rhodobacter Sphaeroides. CL: Cardiolipin, an acidic phospholipid made from two covalently attached phosphatidyl glycerol molecules CP3: Copper3(Phenanthroline), an oxidizing agent Cryo-EM: Cryo-electron microscopy DDM: n-Dodecyl β-D-maltoside, a non-ionic detergent DTT: Dithiothreitol, a reducing agent E. coli: Escherichia coli EDTA: Ethylene-di-amine-tetra-acetic acid EM: Electron microscopy FhuA: Outer membrane protein encoded by gene fhuA, devoid of leader peptide IPTG: Isopropyl-1-thio-β-D-galactopyranoside IMV: Inverted (inside out) inner membrane vesicle JW2806:  An auxotrophic strain of E. coli, cannot produce the lysine synthase LysA, meaning it requires addition of lysine in the media to grow. kDa: kilodalton = 1000 gram/mole KM9: Strain of E. coli that lack the F0F1 ATP synthase gene MalFGK2: ABC transporter responsible for active uptake of maltose, heterotrimeric complex composed of the membrane proteins MalF, MalG, and two copies of MalK. MD: Molecular dynamics simulation MSP: Membrane scaffold protein MP:  Membrane protein xvi   NBD: Nucleotide binding domain, a domain in ABC transporters that binds and hydrolyzes ATP NEM: N-ethylmaleimide, blocks disulphide formation NMR: Nuclear magnetic resonance NSP: Nanodisc scaffold protein NSPr: Nanodisc scaffold protein reversed OmpF3: Trimeric complex composed of outer membrane protein F without leader peptide. PAGE: poly-acrylamide gel electrophoresis PCF:  Protein fragment complementation assay PrInCE:  Prediction of Interactomes from Co-Elution data R. Sphaeroides:  Rhodobacter Sphaeroides SecYEG:  The bacterial translocon, responsible for translocation and insertion of proteins into or across the bacterial membrane.  Heterotrimeric complex composed of one copy each of SecY, SecE, and SecG. SDS: sodium dodecyl sulfate SMA: Styrene maleic acid block co-polymer, with each block containing approximately (n) styrene groups and one maleic acid group.  With n being 1, 2 or 3 for the brand names SMA 1000, 2000, and 3000, respectively.  SMALPs: styrene maleic acid lipoparticle, polymer lipid particle formed by solubilization of a lipid bilayer by addition of excess SMA.  These particles contain both polymer and lipids, and may also have membrane proteins incorporated into them. Tris: Trisaminomethane Y2H: Yeast two-hybrid          xvii   Acknowledgements:       I would like to thank my supervisor, Dr. Franck Duong, for allowing me to freedom to explore and develop my passions for problem solving and method development, while teaching me the basics of membrane protein biochemistry. I would also like to thank several laboratory members who made much of this work possible through discussion, collaboration, and support over the years including: J.W. Young, A. Mills, H. Li, H. Bao, H. Dhupar, Z. Zhao, I. Wason, X. Zhang, S. Macdonald, and H. Won.  I would also like to acknowledge the wonderful scientists of the Foster lab, including N. Stoynov, G. Stacey, N. Scott, and K.Y. Moon.  My supervisory committee, including Dr. J. Thomas Beatty and Dr. Nobu Tokuriki, also provided direction over the course of my degree. Most importantly, I would like to thank my wife and best friend, Elena, for her constant smile in tough times and insight when the path was not clear.  Lastly, I would like to thank my wonderful family and friends for supporting me over the years with kindness and patience.  The work presented in this thesis was generously funded by scholarships from the Natural Sciences and Engineering Council of Canada, and the University of British Columbia.                      xviii   Dedication:   -To my best friend and partner, Elena.  -To Mom, Dad, and Rachael. -To you, the reader.  1  Chapter 1:  Introduction 1.1   Challenges of understanding membrane protein and contributions of this work thereof    Membrane proteins play a vital role in the cell. They help to maintain the structural integrity and facilitate the flow of material through biological membranes.  As a result of their easy accessibility at the surface of the cell, and crucial roles in processing environmental information and regulation of cell homeostasis, membrane proteins are prime pharmaceutical targets and often implicated in disease (Yin and Flynn. 2016).  However, our understanding of membrane structure and function has lagged far behind that of soluble proteins.  This is because membrane proteins present several challenges to study. First, because the membrane accounts for a relatively small fraction of the cell volume, expression is generally low and can present a critical bottleneck. Second, biological membranes are insoluble in aqueous solution and generally very heterogeneous. As most biophysical techniques are tailored for purified soluble proteins, membrane proteins often need to be first solubilized in amphiphilic detergents and purified from the lipid bilayer before characterization. However, due to the denaturing effects of detergents, the structure and stability of membrane proteins is often compromised once removed from the membrane.  These denaturing effects can destroy membrane interactions and alter functional activity.  Here, I add new tools available to researchers for analyzing membrane proteins.  I utilize well characterized membrane proteins from the model organism E. coli to validate use of these new tools for the study of membrane proteins. In chapter 2, I present the peptidisc, a simplified method for working with membrane proteins in detergent-free buffer.  In chapter 3, I expand use of the peptidisc to stabilize membrane proteins of the E. coli cell envelope detergent-free buffer, forming a “peptidisc library”.  Analysis of the peptidisc library by protein-correlation profiling allows for investigation of the many different protein-protein interactions in the E. coli cell envelope in a single experiment.    To further understanding of membrane protein mechanism, I also include a separate study in chapter 4 wherein I identify chloride conducting states in a functional mutant (malF500) of the model ABC transporter, the maltose importer MalFGK2.  This later work provide missing evidence for the expanded alternating access model  2  1.2 Membrane mimetics   Approximately 60% of drug targets are located at the cell surface, despite only encompassing ~1/5th of the human proteome (Overington, Al-lazikani, and Hopkins. 2006). However, only 2% of all determined protein structures were membrane proteins (Zhou, Zheng, and Zhou. 2004, Blanco. 2016). One reason is that membrane proteins rely on a hydrophobic environment provided by a lipid bilayer to be stable, functional, and in the proper conformation.  However, to purify membrane proteins for structural studies, they must be solubilized from the heterogenous membrane.  This has motivated many studies on different reconstitution systems that mimic the hydrophobic environment of a lipid bilayer while maintaining membrane protein solubility and monodispersity.  Membrane mimetics include detergent micelles (Chaptal et al. 2017), amphipols (Popot. 2010), peptide detergents, styrene maleic acid lipoparticles (Broecker, Eger, and Ernst 2017; Dörr et al. 2014), and nanodiscs (Ritchie et al. 2009).  Each technique has specific strengths and caveats, which will be reviewed in the following sections, but are also summarized in Table 1.   Mimetic Method of Production Unnatural functional groups? Requires prior detergent solubilization Stable in aqueous buffer without excess scaffold Requires optimization (exogenous  lipid) Buffer instability at low pH or divalent cations Nanodisc Recombinant No Yes Yes Yes No Peptidisc Synthetic No Yes Yes No No saposin A lipoparticle Recombinant No Yes Yes Yes No Amphipols Synthetic Yes No Yes/No No Yes SMALPs Synthetic Yes No Yes No Yes Peptergent Synthetic Yes No No No Unknown Lipopeptide Synthetic Yes Yes No No Unknown [beta]-sheet peptides Synthetic Yes Yes Yes No Unknown Beltide Nanodisc Synthetic Yes Yes No Yes Unknown 3   Table 1-1: Comparison of membrane mimetics for membrane protein solubilization.  1.2.1 Detergents   Detergents are the most widely used membrane mimetic for analysis of membrane proteins.  These are amphipathic molecules, in general consisting of a hydrophilic head group and hydrophobic alkyl tail, much like the lipids that make up the cell membrane.  Detergents are sparingly soluble in their monomeric form, and as concentration increases they associate into multi-unit assemblies called micelles (Figure 1-1).  The outside of a micelle is hydrophilic and formed by the head groups of detergent, while the tails form the hydrophobic micellar core. The critical micelle concentration (CMC) of a detergent is the point where a solution has become saturated by free detergent monomers, and thus becomes thermodynamically stable for micelles to form (Figure 1-1 A, point 2). Because of the mismatch between the width of the detergent head group and the alkyl chain, micelles are generally spherical (Figure 1-1B).  In contrast, the two alkyl chains of a lipid prevent this mismatch, so lipids self-assemble into lipid bilayers.   4   Figure 1-1:  Mechanism of micelle formation by amphiphilic detergents.  A) Light scattering (left axis), and apparent hydrodynamic diameter (right axis), of detergent at increasing concentrations.  B)  Representative schematic of nanoscale detergent structures formed in solution at the indicated titration points 1, 2 and 3 in panel A. 1.2.1.1  Mechanism of Detergent Solubilization    The size of a micelle is determined by a detergent’s aggregation number (the number of individual detergent molecules that are necessary to form a stable assembly), while the charge of the head group denotes whether the detergent is classified as either ionic (+ or -), zwitterionic (+ and -), or non-ionic (no charge).  Because they are structurally very similar to lipids, detergents 5  are able to partition into the lipid bilayer when added in concentrations above their CMC, forming a mixed lipid-detergent bilayer (Figure 1-2, titration point 2). However, mismatch between the detergent and lipids result in bilayers that are unstable. As detergent concentration continues to increase the bilayer eventually breaks (solubilizes) into mixed lipid-detergent micelles (Figure 1-2, titration point 3).  This partitioning, and then eventual disruption of the lipid bilayer can be demonstrated in free liposomes by light scattering measurements.  Titration of detergent therefore results in liposome swelling and increased light scattering, followed by decrease in light scattering as the bilayer eventually collapses and becomes solubilized in excess detergent (Figure 1-2).  Most importantly, when solubilizing a biological membrane, membrane proteins and lipids become incorporated into mixed micelles.    6   Figure 1-2:  The three stages of detergent solubilization of a lipid bilayer with increasing titration of detergent.  A) Light scattering of a solution of liposomes with increasing concentrations of detergent.  B) Schematic detailing the structure of the lipid bilayer at the indicated points in the detergent titration.  Detergent molecules are indicated with red triangles, Lipid molecules are black rectangles.   7  1.2.1.2  Disadvantages of detergents for membrane protein characterization     The ability to release membrane proteins from the insoluble lipid bilayer into soluble particles is the most powerful feature of detergents, as membrane proteins can then be purified using traditional affinity methods.  However, detergent solubilization has several important caveats that complicate characterization of membrane proteins after their solubilization.   First, all detergents are slightly denaturing (Yang et al. 2014; le Maire, Champeil, and Moller 2000).  The denaturing effect of detergents generally decreases with charge of the head group; non-ionic detergents, such as n-Dodecyl β-D-maltoside (DDM), are therefore most commonly used for preparation of membrane proteins (le Maire, Champeil, and Moller 2000; Orwick-Rydmark, Arnold, and Linke 2016).  Second, to maintain micelle stability and membrane protein solubility, detergents must be constantly present in concentrations above their CMC (le Maire, Champeil, and Moller 2000).  This excess in free micelles interferes with many of the analytical techniques used to characterize protein structure and function (Singh and Sigworth 2015; Lipfert et al. 2007; Zhou, Kini, and Sivaraman 2011). For example, when analyzing membrane proteins by mass spectrometry, excess detergent micelles saturate peptide binding sites during preparative reverse phase chromatography, and when eluted can lead to ion-suppression (Gundry et al. 2009).  Excess micelles also prevent characterization in structural studies; in X-ray crystallography, uneven concentration of free micelles along with protein makes it difficult to form reproducible crystal conditions (Wiener 2004).  When attempting structure determination by cryo-electron microscopy (Cryo-EM), free detergent micelles decrease contrast and can interfere with vitrification of sample grids (Singh and Sigworth 2015).  In highly sensitive techniques such as isothermal calorimetry, exchange of detergent monomers decreases signal to noise ratio in binding curves, preventing analysis without carefully controlled subtraction of detergent artifacts (Zhou, Kini, and Sivaraman 2011).  Perhaps most importantly, micelles are labile structures, resulting in a more dynamic and loosely constrained environment than a lipid bilayer.  As a result, membrane proteins often experience significant shifts in enzymatic activity, thermostability and structure when characterized in a detergent environment compared to their native lipid environment (Bao et al. 2013; Palazzo, Lopez, and Mallardi 2010).  Furthermore, the extensive interactions formed by membrane proteins with surrounding lipids can become destabilized, leading to loss of annular lipids that may be important for enzymatic activity or 8  stability (Gupta et al. 2017).  It can therefore be difficult to relate biochemical data gathered from a detergent environment in a biological context.  This has lead to the adoption of new detergent free reconstitution systems over the past decade.    1.2.2 Protein based reconstitution systems  Two protein based reconstitution systems are currently available; the nanodisc system and more recently developed Salipro method (Ritchie et al. 2009; Frauenfeld et al. 2016).  Both methods use amphipathic protein scaffolds that have been designed based on naturally occurring lipid binding proteins, ApoA1 and saposin-A, respectively.   1.2.2.1   Nanodiscs  Interest in ApoA1 (Apolipoprotein A- 1) was first stimulated because the protein is essential for the biogenesis of high density lipid particles (HDL) that traffic cholesterol in the bloodstream (Brunham and Hayden 2015, Segrest and Anantharamaiah, 1994).  The structure of ApoA1 consists of a globular domain, followed by a series of amphipathic helices that are connected by flexible proline linkers (Mei and Atkinson 2015).  In its lipid free form ApoA1 is soluble and monodisperse; however, upon addition of small amounts of phospholipid, it will associate into particles composed of a lipid core encircled by two molecules of ApoA1 (Mei and Atkinson 2015; Segrest et al. 1999).  Detailled analysis of these particles revealed that lipids were captured in a bilayer arrangement, with each phospholipid leaflet shielded from surrounding solution by an individual ApoA1 monomer, in effect creating a small disc of soluble membrane (Bibow et al. 2017, Chromy et al. 2007).     Dr. Stephen Sligar at University of Illinois was the first to recognize how the lipid binding capabilities of certain naturally occurring proteins could be advantageous for the solubility of membrane proteins; his group optimized the ApoA1 sequence to form the membrane scaffold proteins (MSPs) (Ritchie et al. 2009).  MSPs contain only the amphipathic helical region of ApoA1 and stabilize pure lipid systems in self assembled, discoidal nanometer sized particles called nanodiscs (Ritchie et al. 2009).  The self-assembly reconstitution process t simply consists of mixing MSP with lipids solubilized in detergent, followed by removal of the detergent by adsorption to polystyrene beads (Ritchie et al. 2009; Hagn, Nasr, and Wagner 2018).  During 9  detergent removal, the MSPs captures lipids into small discoidal particles of high homogeneity, approximately 9.7nm in diameter when using the scaffold MSP1D1.  The method can be also used to capture purified membrane proteins (Fig. 1-3, Route 1), and even more to form membrane protein libraries, in which the entire membrane proteome is captured in nanodiscs (Fig. 1-3, Route 2).        Figure 1-3. Methods for nanodisc formation.  The standard method for self-assembling an membrane protein (MP) into a nanodisc is shown in route 1 (left): after detergent solubilization and purification, the target MP (green) is mixed with the membrane scaffold protein (MSP, blue) and lipids at the correct stoichiometry, followed by detergent removal though incubation with hydrophobic beads. Often, however, the MP is not stable in detergent for the extended times needed for purification. Alternatively (route 2, right), the starting membrane or tissue can be directly solubilized with excess lipid and scaffold protein. The rapid detergent removal results in placement of the target MP (green), together with other MPs (gray) from the tissue, into the nanodisc. Subsequent purification, often with an affinity tag, is performed. This 10  latter route can be used to generate a soluble MP library that faithfully represents the MPs in the starting tissue.  Figure and caption reproduced with permission from reference (Denisov and Sligar 2016).  It has been demonstrated in some detail that nanodisc diameter is directly proportional to scaffold length and amount of lipid added during reconstitution (Grinkova, Denisov, and Sligar 2010).  Recognizing the power of a soluble and monodisperse lipid bilayer, the system was initially applied to the Cytochrome P450 3A4.  Addition of the Cytochrome P450 during the self-assembly process resulted in its 1:1 incorporation into the nanodisc, rendering the protein entirely soluble in detergent free buffer, while maintaining functional ligand binding activity (Grinkova, Denisov, and Sligar 2010; Denisov and Sligar 2011; Bayburt and Sligar 2002).  Since then, the nanodisc has gradually become an invaluable tool for membrane protein research.  Control over scaffold diameter, lipid content, and protein concentration in the self-assembly process means the nanodisc can be used to test effect of lipids on protein structure and activity or else to capture protein of different diameters or oligomeric states (Boldog, Li, and Hazelbauer 2007; Bao et al. 2013; Dalal et al. 2012).  However, whereas the nanodisc has performed admirably as a tool for the functional characterization of membrane proteins, its use in structural determination of membrane proteins has been met with mixed results.       At the time of this writing, no crystal structure of a membrane protein in a nanodisc exists, despite the technologies adoption over the past 15 years.  It has been suggested that the heterogeneity of incorporated lipids, and translational mobility of a protein incorporated into a fluid bilayer are the root cause of the nanodisc’s resistance to crystallization (Denisov and Sligar 2016).  On the other hand, these factors have not prevented structure determination by nuclear magnetic resonance (NMR) or Cryo-EM (Chaptal et al. 2017; Hagn et al. 2013; Mi et al. 2017).  Progressive shortening of the MSP to achieve a nanodisc with diameter that almost exactly matches the incorporated membrane protein can be used to make highly monodiserpse nanodisc preparations suitable for NMR (Hagn et al. 2013; Hagn, Nasr, and Wagner 2018).  In Cryo-EM, the individual lipids and MSPs of the nanodisc are poorly resolved, but incorporated membrane proteins display good stability and their structure can be determined to moderate resolutions (<7Å).  However, it should be noted that of the Cryo-EM structures determined in the nanodisc, with the exception of the polycystic K+ channel, many of these proteins have extended soluble domains that protrude from the membrane (Gao et al. 2016; Frauenfeld et al. 2011; Stam and Wilkens 2017; Shen et al. 2016; Katayama et al. 2010).  This becomes more relevant when 11  contrasted to a Cryo-EM structure of a membrane protein in the Salipro system, which is discussed next.  1.2.2.2        saposin Lipoparticles     The patented Salipro system follows the same self-assembly principle as a nanodisc; solubilized lipid, membrane protein, and the amphipathic Salipro scaffold are mixed together, followed by removal of detergent micelles by dilution or adsorbents. Salipro is based on the particular lipid binding properties of saposin A (Frauenfeld et al. 2016). Sphingolipid activator proteins (SAPs) are nonenzymatic proteins that increase lipid accessibility to lysosomal, degradative enzymes.  saposin A functions in this capacity to accelerate degradation of galactosylceramide by β-galactosylceramidase, producing ceramide and galactose (Popovic et al. 2012).  In solution, saposin A folds into a compact, globular structure with a buried hydrophobic core (Chien et al. 2018, Popovic et al. 2012). Upon contact with lipids and in acidic buffer, the saposin unfolds into a lipid-dependant structure with an extended, hydrophobic lipid binding face (Frauenfeld et al. 2016; Vaccaro et al. 1995; Popovic et al. 2012).  This unfolding effect can be also be accomplished at neutral pH through the addition of non-ionic detergents such as DDM or Lauryldimethylamine N-oxide (LDAO) (Popovic et al. 2012, Chien et al. 2018).  Similar to MSPs, the hydrophobic face of saposin A contacts the alkyl chain portion of lipids (Popovic et al. 2012).  However, the region of contact is comparatively short compared to the multiple alpha-helices of MSP.  The small contact surface of saposin allows it to form so-called “picodiscs” containing only 20-30 lipids (Fig 1-4), in contrast to nanodiscs which contain 90-130 lipids (Popovic et al. 2012; Jun Li et al. 2016).    12   Figure 1-4:  Modelled examples of saposin reconstituted nanoparticles.  Incubation of saposin A with lipids and membrane proteins leads to self-assembly in nanoparticles. The saposin adjusts to the size of the target molecule. Models of saposin-lipid complexes were adapted from Protein Data Bank (PDB) IDs 4DDJ, 4APS, 4NCO, 3DIN and 2DOB;   Figure reproduced with permission (Frauenfeld et al. 2016).      When applied for membrane protein stabilization, multiple saposins binds to the target membrane protein (Frauenfeld et al. 2016).  In the case of the polyamine transporter POT4, saposins stabilize the transporter in aqueous solution, enabling its structure determination by cryo-EM.  POT4 is almost entirely integral to the membrane, yet cryo-EM structure analysis is able to clearly identify the individual transmembrane helices and bound saposins (Frauenfeld et al. 2016).  This is in contrast to cryo-EM structures of the nanodisc, where the scaffold proteins at the surface of the target cannot be resolved.  Here, direct contacts between the individual 13  saposin monomers and transmembrane domains of the transporter could be visualized, suggesting that the highly ordered saposins may be advantageous for structure determination at high resolution (Fraunfeld et al 2016).  There is however still little data for how membrane protein function is affected by Salipro, and why certain lipids are apparently excluded in the final assembly (Frauenfeld et al. 2016, Flayhan et al. 2018).  While the saposin in complex with detergents has been solved by X-ray crystallography, no report of crystallization of a Salipro-particle containing an embedded membrane protein has been realized.      1.2.3 Synthetic polymer based reconstitution systems   Protein scaffolds are often limited in their scalability, purity and homogeneity because they must be produced in recombinant cell culture (Faas et al 2017).  In addition, protein scaffolds still rely on detergent to release the target protein from the membrane, (Hagn, Nasr, and Wagner 2018; Frauenfeld et al. 2016).  To work around these limitations, researchers have developed various synthetic amphipathic scaffolds for supporting membrane proteins in solution, which I review in the following sections.     Figure 1-5. Chemical structure of synthetic polymer scaffolds for detergent free stabilization of membrane proteins.  In the model amphipol A8-35, the stoichiometry of each carboxylate, octylamine, and isopropylamine are  x ≅ 25, y ≅ 17, and z ≅ 28, respectively.  In the styrene maleic acid block co-polymer, each block contains a styrene group (where m = 2 or 3, depending on polymer formulation) and a maleic acid group, where n = 1.    14  1.2.3.1   Styrene maleic Acid lipid nanoparticles    This group of synthetic scaffolds is based on the styrene maleic anhydride block co-polymer.  Simple hydrolysis of the anhydride bond produces styrene maleic acid co-polymer (SMA) which is soluble in aqueous solution (Lee et al. 2016; Gulati et al. 2014; Dörr et al. 2014).  SMA is able to directly solubilize lipid membranes, encapsulating membrane proteins into discoidal styrene-maleic acid lipid nanoparticles (SMALPs). The SMA polymer is thought to wrap around a patch of the lipid bilayer, with the hydrophobic styrene groups intercalating between the alkyl chains and hydrophilic maleic acid groups facing the aqueous solution (Dörr et al. 2014; Swainsbury et al. 2014).  While solubilization of the membrane requires excess SMA, SMALPs are not dependant on a CMC and are therefore stable in aqueous solution (Dörr et al. 2014; Swainsbury et al. 2014).    The ability of a protein to go from the insoluble, biological membrane, directly to a soluble and monodisperse nanoparticle without encountering detergents is a definitive advantage that is only provided by SMALPs.  Furthermore, direct solubilization of a lipid bilayer by SMA is thought to immobilize native lipids in the particle, meaning that little or no delipidation of a solubilized membrane protein is occuring (Swainsbury et al. 2014).  Indeed, comparison of the lipid profiles of purified membrane proteins show significantly more variety of lipids that are trapped in SMALP than in conventional non-ionic detergents (Swainsbury et al. 2014).  Furthermore, proteins incorporated in SMALPs have increased thermostability compared to detergent, suggesting that proteins are more stable due to the native lipid environment provided by the SMALP (Dörr et al. 2014; Swainsbury et al. 2014).  However, recent studies on lipid transfer between SMALPs and a supported monolayer indicate that lipid transfer still occurs in these particles (Cuevas Arenas et al. 2017).  Other experiments also demonstrate that SMA is not a “one-size fits all” scaffold; the polymer preferentially solubilizes membrane proteins of defined size and membrane composition (Cuevas Arenas et al. 2017; Swainsbury et al. 2017).  The conclusions from these experiments are counter to the main advantage of utilizing the SMA polymer as a purification method, which is the unbiased preservation of the native lipid bilayer around a purified membrane protein.      SMALPs can have significant effects on enzyme activity, and there has been only limited success in structural studies.  Maleic acid is a strong chelator, meaning that addition of di- or tri-valent cations leads to destabilization of a SMALP and incompatibility with many enzymatic 15  assays that require ions as co-factors (Oluwole et al. 2017; Ravula et al. 2018).  While it is possible to chemically modify the polymer to reduce chelation (Oluwole et al. 2017; Ravula et al. 2018), whether the chemical environment provided by these polymers is truly “near-native” becomes questionable.  An additional issue is that most SMA polymer and variants are sourced from industrial preparations designed for use in the pigment or cleaning industries (Dorr et al 2014, Oluwale et al 2017).  These polymer preparations have a high degree of heterogeneity in length, and the order of incorporation of the styrene or maleic acid moieties in each block of the co-polymer is not controlled (Lee et al. 2016).  New synthesis techniques can hopefully produce better polymers, with higher degree of monodispersity and composition in the future (Lee et al. 2016a; Craig et al. 2016).  The recent structure determination of AcrB in a SMALP by cryo-EM nevertheless proves that current preparations of the polymer are good enough for structural determination (Parmar et al. 2018). The structure of AcrB in SMALPs is a landmark achievement, since the protein was purified without the use of any detergents.  However, in comparison to AcrB structures determined in detergent or nanodiscs, resolution of the transmembrane domains of AcrB was of lower quality in the SMALPs (Parmar et al. 2018).  As of today, the most obvious advantage of a SMALP is avoidance of detergents in the protein purification process, and concomitant increase in thermostability of the purified proteins.   1.2.3.2    Amphipols   Amphipols are another group of synthetic polymer that have been developed by the group of Dr. Popot at the Institut de Biologie Physico-Chimique in Paris, France. The overall chemical arrangement of an amphipol is similar to the SMA polymer.  Amphipols contain repeated hydrophobic functional groups, that are balanced by hydrophilic carboxylate groups (Fig. 1-5), the best characterized amphipol for use with membrane proteins research is the A8-35 variant (Fig 1-5).  As such, like the SMA, amphipols are sensitive to acidic buffers (<pH7), divalent cations, and the polymer can be heterogeneous in composition.  Also similar to SMA, amphipols A8-35 can be used to solubilize membranes directly from a lipid bilayer.  However, the overall quality of captured proteins is better when the membrane is first solubilized in detergent, and the protein later exchanged into amphipols (Popot 2010).  After transfer, membrane protein-16  amphipol complexes are soluble and display enhanced thermostability (Popot 2010; Calabrese et al. 2015).    Lipid substrates can be delivered to membrane proteins in solutions of excess amphipols, presumably due to collision and exchange between the individual amphipol complexes (Popot et al 2010)  (Popot 2010) A major inconvenience is that upon removal of excess amphipols, the membrane protein-amphipol complex starts to aggregage especially in acidic conditions.  Like SMALPs, protein inactivation has also been reported after capture in amphipols (Picard et al. 2006). Similar to saposins, amphipols have the flexibility needed to adapt to proteins of various size, fold and shape.  Membrane proteins ranging from 1-70 transmembrane helices have been stabilized in amphipols, and several high resolution cryo-EM structures of membrane proteins have been determined (Althoff et al. 2011; Baker, Fan, and Serysheva 2015). Amphipols and SMALPs have been utilized to deliver membrane protein into lipid mesophase for crystallization of bacteriorhodopsin (Polovinkin et al. 2014; Broecker, Eger, and Ernst 2017).     1.2.4 Peptide based reconstitution systems   Amphipathic, peptide based reconstitution systems have been proposed as a flexible alternative to protein based reconstitution methods.  This is because peptides can offer a balance between the synthetic polymers and protein scaffolds.  Similar to amphipathic polymers, peptides can be synthesized to a high degree of homogeneity and produced in scalable quantities.  Like protein scaffolds, peptides are composed of highly controlled sequence of naturally occurring amino acids and therefore maintain a more natural protein-lipid environment.  Several different methods to produce stable protein lipid particles have been developed; including peptergents, lipopeptide detergents, nano-structured beta sheet peptides, and ApoA1 mimetic peptide nanodiscs, which I will discuss in the following sections.      1.2.4.1      Peptergents   Peptergents are a class of amphipathic peptides that are proposed to act much like traditional detergent. They are composed of repeating units of hydrophobic alanines, capped by a positively charged residue, typically either Lysine or Aspartate (Fig. 1-5).  The N-terminus is acetylated or C-terminus amidated to ensure a strong hydrophobic moment throughout the peptide.  This 17  causes peptergents to form spherical micelles, thus they can be used to directly solubilize membrane to extract proteins.  By varying the length of the peptide, as well as the charge of the N or C-terminus, a large suite of peptergents has been described (Fig 1-5).  These alternative peptergents form micelles with varying aggregation number and surface charge have provided researchers with multiple solubilization options.  Importantly, peptergents lead to an increase in protein stability compared to conventional detergents after extraction from a biological membrane (Corin et al. 2011).  However, due to the large excess of peptides necessary to directly solubilize a biological membrane, their use can rapidly become economically unfeasible due to high peptide synthesis costs.  In addition, their high hydrophobicity can make it difficult to resuspend these peptides in solution, often requiring careful titration of base and disruptive sonication for full dissolution.  As a result, while a powerful tool in the right circumstances, peptergents have not been widely adopted as more ubiquitous tools, such as the nanodisc, for membrane protein study.   18  Figure 1-6. Molecular models of peptide detergents at neutral pH.  A) Ac-AAAAAAD-COOH. B) Ac-AAAAAAK-CONH2. C) DAAAAAA-CONH2. D) KAAAAAA-CONH2. E) Ac-VVVD-COOH. F) Ac-VVVK-CONH2. G) Ac-IIID-COOH. H) Ac-IIIK-CONH2. I) Ac-LLLD-COOH. J) Ac-LLLK-CONH2.. The hydrophobic tails of the peptide detergents consist of alanine (A), valine (V), isoleucine (I) and leucine (L). Each peptide is ∼2–2.5 nm long, which is similar size to biological phospholipids. Color code: teal, carbon; red, oxygen; blue, nitrogen and white, hydrogen.  Figure and caption modified with permission from reference (Corin et al. 2011).  1.2.4.2    Lipopeptides      Lipopeptides consist of a single, amphipathic a-helix coupled to two lipid groups located at the N- and C-terminus of the peptide (McGregor et al. 2003).  In this arrangement, it is expected that the covalently attached lipids lie parallel to the hydrophobic face of the peptide, which is formed by a row of alanines.  In contrast to traditional detergents which form a spherical micelle, lipopeptides are thought to form a cylindrical micelle, which can approximate the bilayer membrane structure (Figure 1-7).  Similar to peptergents, lipopeptides stabilize membrane proteins better than conventional detergents, and can be synthesized in a range of peptide lengths to alter the width of the cylindrical micelle. Lipopeptides have been applied to stabilize E. coli inner and outer membrane proteins LacY and PagP, respectively (Mcgregor et al 2003).  One construct, LPD-16, was found to stabilize the inner membrane protein LacY compared to even mild detergents such as DDM.  H1N15-NMR was also utilized to characterize the size of the outer membrane protein PagP stabilized in a shorter construct, LPD-14.  From this study, it was concluded that approximately 10-15 lipopeptides associated with a single monomer of PagP.         Although they have a low CMC, excess lipopeptide must still be supplied in the reconstitution buffer to maintain protein stability (McGregor et al. 2003). While a powerful stabilization tool compared to traditional detergents, lipopeptides are complex and expensive to synthesize, therefore membrane proteins are typically buffer exchanged into the lipopeptide micelle after initial purification in a traditional non-ionic detergent (Privé 2009).    19   Figure 1-7.  Structural comparison of traditional detergent micelle, lipid bilayer, and lipopeptide micelle. (A) Expected conformation arrangement for a 12 amino acid long lipopeptide. The α-helical peptide is shown in the gray Cα trace, and includes the ornithine side chains at positions 2 and 24. Each ornithine δ-amine group is coupled to a dodecanoic acid moiety (indicated with black carbon atoms and a red oxygen atom) by the formation of an amide bond. (B–D) Schematic representations of individual molecules and their assemblies as idealized geometric solids. In this series, red and green represent hydrophilic and hydrophobic surfaces, respectively. (B) A “traditional” detergent such as DDM is represented as a cone that assembles into a sphere. Note that a cone is a poor approximation of the monomer shape, because in reality, the volume of the alkyl chain does not taper. (C) A lipopeptide can be seen as a wedge that assembles into a cylinder. In this case, the volume of the hydrophobic alkyl chain is uniform over the length of the chain. (D) A phospholipid is shown as a box, which packs laterally with other boxes to form even larger boxes, but does not form a closed particle. Thus, the lipopeptide design combines features of both traditional detergents and bilayer-forming lipids, since they can form closed particles with relatively uniform acyl chain packing.  Figure has been modified with permission from reference (McGregor et al. 2003).  1.2.4.3      Nanostructured beta-sheet peptides     Nanostructured [beta]-sheet peptides are formed by small, 8 amino acid peptides that form a “picket fence” around a membrane protein (Fig. 1-8ab).  Each vertical peptide is linked to the 20  other by hydrogen bonding of the peptide backbones, forming an extended beta-sheet wrapping around a reconstituted membrane protein.  To allow these peptides to bind with a high affinity to the membrane protein template, and to increase the overall hydrophobicity, the peptides are modificed by extended alkyl chains (Fig. 1-7a). Similar to amphipols and lipopeptide detergents, membrane proteins must be first solubilized in detergent before insertion into nanostructured [beta]-sheet peptides (Tao et al. 2013).  Although the nanostructured [beta]-sheet peptides are well suited for preparing monodisperse membrane protein preparations, they contain extensive amino acid modifications, which can be prohibitively expensive to synthesize.  Furthermore, these peptides make extended filaments composed of [beta]-sheets when rehydrated in the absence of a membrane protein template, which render them difficult to solubilize in aqeuous solution (Fig 1-7cd).     Figure 1-8.  Structures of β-strand peptides (BP) employed to stabilize integral membrane proteins (IMPs). (a) The BP sequences (BP-1, -2, and -3) feature facial amphiphilicity with alternate hydrophobic (red) and hydrophilic (blue) residues, and differ in the number of N-Methyl amino acids (green). (b) Cartoon representation of proposed β-barrel architecture assembled from BPs (blue strands) by interstrand H-bonding in which the hydrophobic alkyl chains (space filling spheres) associate with and sequester the IMP surfaces (orange α-helices). (c) CD spectra of BP-1, -2 and -3 indicate secondary structures with β-sheet character. (d) Electron micrographs of negatively stained BP-1 show self-assembled long filamentous structures of ~3 nm in diameter (inset). The scale bar represents 30 nm.  Figure and caption modified with permission from reference (Tao et al. 2013).   21  1.2.4.4    ApoA1 mimetic peptides  Similar to MSPs, peptide scaffolds that mimic the charge distribution of ApoA1 have been designed.    Pioneering work from the Segrest lab identified that a small, 18 mer peptide (termed 18A) designed to mimic the charge distribution typically present in the amphipathic helices of ApoA1 is able to stabilize lipids into discoidal particles (Anantharamaiah et al. 1985).  Originally, these ApoA1 mimetic peptides were designed and utilized for treatments of cholesterol efflux in models of atherosclerosis (Anantharamaiah et al. 1985; Islam et al. 2018; Van Lenten et al. 2009).  These peptides can be used for self-assembly of lipid particles to form lipid nanodiscs (Midtgaard et al. 2014; Larsen et al. 2016; Anantharamaiah et al. 1985).  Since the original work in the group of Segrest, the 18A peptide has been modified in multiple ways to modulate its peptide-peptide and peptide-lipid interactions properties.  To increase hydrophobicity of the peptide, leucine residues on the hydrophobic side of the helix have been substituted to phenylalanine, increasing the overall affinity of the peptide for lipid membranes (Mishra et al. 2008).  Modifications of the N and C-terminus by acetylation and amidation, respectively, also increase affinity of the peptide for lipids (Getz and Reardon 2011). Two 18A peptides which have been linked together increase thermodynamic disc stability (Getz and Reardon 2011; Larsen et al. 2016).  Addition of a helix breaking residue in the middle of these bi-helical peptides, such as proline or glycine residue, increase allows the researchers to form different sized lipid-peptide nanodiscs by modifying the lipid to peptide ratio during reconstitution (Larsen et al. 2016; Midtgaard et al. 2014).  It was found that peptide nanodiscs containing a glycine linker displayed the highest homogeneity.       Application of ApoA1 mimetic peptides has mainly focused on the stabilization of small patches of lipids.  Naturally, given the success of the nanodisc for membrane protein stabilization, there have been recent efforts to apply these peptides for membrane protein stabilization. Earlier results show it is possible to stabilize bacteriorhodopsin in solution using the 18A peptide, or the bi-helical derivative of 18A, provided lipids are also added (Midtgaard et al. 2014; Larsen et al. 2016).  However, the procedure required for incorporation of bacteriorhodospin into lipid-peptide discs is more complicated than even the self-assembly procedure employed in nanodisc.  First, the lipid-peptide nanodiscs must be formed by mixture in organic methanol solvent at specific ratios, dried down, and hydrated overnight to form soluble, lipid peptide nanodiscs.  Next, lipid-peptide nanodiscs must be solubilized in detergent 22  before addition of the bacteriorhodospin , and excess detergent and aggregated protein must be removed (Larsen et al. 2016).  Bacteriorhodospin appear soluble and monodisperse as measured by gel-filtration.  However, the method provides little advantage over traditional nanodisc methods. The peptide nanodisc size must be optimized by addition of lipids, and these sample handling steps introduced during disc preparation.  Furthermore, over time the formed nanodiscs become unstable and aggregate, possibly due to instability of the discs upon displacement of lipids by an inserted membrane protein (Midtgaard et al. 2014).  Peptide nanodiscs formed from the bi-helical 18A derivates, termed “beltides”, display less aggregation than peptide nanodiscs formed with the peptide 18A, although even these optimized beltide nanodiscs are only structurally stable at low temperature <4°C, whereas at room temperature they can rapidly aggregate (Larsen et al. 2016).    In summary, while peptides have been explored for use in membrane protein stabilization, to date there has been little adoption by the biochemical community.  In the case of peptergents and lipopeptides, there are few advantages of using these peptide based detergents over traditional detergents due to their high cost.  ApoA1 mimetic peptides show promise, but current methods to apply these constructs for reconstitution of membrane proteins present many of the same drawbacks as reconstitution of membrane proteins into MSP nanodiscs, and the formed nanodiscs are unstable.  Thus, there is more exploration needed for how to apply these peptides for membrane protein stabilization.  In chapter 2, I will present the design of the peptidisc, a simple method for generating large quantities of monodisperse membrane proteins using ApoA1 mimetic peptides.  The peptidisc avoids many of the issues for each of the different membrane protein reconstitution systems presented above, and therefore may become a universal tool for membrane protein stabilization. 1.3 Methods to examine protein-protein interaction networks    Protein-protein interaction networks allow for the limited number of encoded proteins to accomplish an impressive number of functions in the cell.  This is because functional flexibility of complexes is often accomplished by the removal, swapping, or fine-tuning of interactions between different complex members.  This regulation of protein-protein interaction networks provides a second level of regulatory control over complex function beyond simple expression of 23  a protein constituent.  Recognition of the importance of these protein-protein interaction networks (“interactomes”), has lead to growing interest in new techniques to characterize their effects.  Because protein interaction networks consist of thousands of individual proteins, high-throughput methods must be adopted to study interactomes, which can be broadly grouped into either genetic or proteomic approaches.    1.3.1 Genetic approaches     There are several, high-throughput, genetic methods for detecting protein interaction networks, the most popular being two-hybrid (2H) and protein fragment complementation (PCA) approaches.  2H screens involve co-expression of tagged  “bait” and “prey” proteins.  If the “bait” and “prey” proteins interact, they drive transcription of a reporter gene.  In model organisms, such as yeast or E. coli it is relatively simple, albeit laborious, to individually tag and express the thousands of open reading frames in the cell (Rajagopola et al. 2010).  In combination with robotic culturing, interactions of the entire proteome can be screened once a 2H library is developed (Rajagopola et al. 2014).   However, the method obviously suffers from certain drawbacks.  First, tagging and expressing every protein in a genome is a monumental task, becoming effectively impossible in organisms that lack tools for genetic manipulation. Second, If the protein library is heterologous expressed in yeast (Y2H) , the proteins are made in a non-native environment, and can be degraded, modified, or improperly folded.   Furthermore, the method is known to suffer from high false positive and false negative identification rates (Huang et al. 2007).  This can be caused by under-sampling, or simply because gene over-expression can lead to complex misassembly.  Furthermore, even with native protein expression levels, the two-hybrid assay is not truly examining native protein interactions.  Fusion of a reporter gene to a protein can alter protein stability, expression levels, and disrupt complex formation,  which together bias identifications of certain groups of proteins interactions (Huang et al. 2007).  Finally, there is no guarantee that there will be proper assembly of the reporter construct if the two protein interact, which can lead to a weak reporter signal and missed positive identifications.  To avoid expression issues, protein constructs can be expressed directly from the chromosome, although there will always be some interference by addition of tags.  Similarly, PCA approaches also use tagged bait and prey proteins. However, each protein contains a 24  fragment of a reporter gene, such as GFP or luciferase (Rochette et al. 2015).  If the two proteins interact, the reporter construct will be re-assembled to produce a measurable signal.  This method suffers from many of the issues reported with two-hybrid approaches.      1.3.2 Proteomic approaches    Proteomic methods to identify protein interactions can be broadly classified into three categories; affinity purification (AP) methods, bi-dimensional gel (2D-gel) and protein elution correlation profile (PCP).  In the following sections, I introduce and expand on the strengths and weaknesses of targeted AP-MS, low-throughput (2D-gel) and high-throughput (PCP) correlation methods. 25   Figure 1-9. Protein pppurification strategies for MS-based PPI identification. (A–D) Strategies to enrich a protein of interest from a crude lysate using endogenous antibodies (A), epitope tagging (B), biotinylated peptide tags (C), or proximity ligation methods (D). (E,F) Protein complex separation strategies from a crude lysate for global interactome profiling using biochemical fractionation (E) or perturbation analysis (F). Abbreviations: Ab, antibody; APEX, engineered ascorbate peroxidase for proximity-ligation; AP-MS, affinity purification MS; BioID, proximity-dependent biotin identification; IEX, ion-exchange chromatography; PPI, protein–protein interaction; QUICK, quantitative immunoprecipitation combined with knockdown; SEC, size-exclusion chromatography; TAP-MS, tandem affinity-purification MS; TPP, thermal proteome profiling.  Figure and caption reproduced with permission from reference (Smits and Vermeulen, 2016).   26  1.3.2.1 Interaction identification by co-affinity purification (AP-MS)     In AP, target proteins of interest are fused with an affinity tag.  The cells are lysed, and the expressed protein of interest is purified via its affinity tag.  Any proteins in complex with the tagged “bait” protein are coincidentally also enriched.  The entire mixture can then be analyzed via MS to identify proteins enriched by the tagged bait protein relative to a background control ( Dunham, Mullin, and Gingras 2012).  AP methods involve protein purification using a variety of affinity resins.  His-tags, biotinylated peptide tags, antibody epitope tags, or even antibodies specific to the protein target of interest can be utilized (Fig 1-9 A, B and C).  However, most AP-methods carry many of the same limitations in regards to scalability as two-hybrid approaches; each target must be genetically tagged, which can alter protein interactions or localization (Hoffmann et al. 2005; Stadler et al. 2013).  In addition, protein stability can be compromised by tagging (Booth et al. 2018).  AP methods can also be plagued by numerous false-positives due to non-specific affinity of certain proteins for affinity resins and overabundant proteins contaminating the experiment (Dunham, Mullin, and Gingras 2012; Mellacheruvu et al. 2013).  Many of the most common AP contaminants have now been listed in the contaminant repository for affinity purification, otherwise known by its tongue in cheek moniker, the “CRAPome” (Mellacheruvu et al. 2013).  Affinity purification of post-translationally labeled proteins represents an alternative strategy to identify protein interacting partners by AP-MS.  The BioID system utilizes a mutant form of the protein BirA genetically fused to a protein of interest (Fig. 1-8D)  (Roux, Kim, and Burke 2013).  Upon expression, the BirA mutant releases 5’-Biotin-AMP, which can diffuse into the surrounding millieu and react with nearby primary amines (Cronan et al. 2005).  After biotinylation, the cells are lysed and biotinylated peptides purified by a streptavidin-coated affinity matrix and later identified by MS.  Because the activated biotin can also react with water, it has a very short half-life in the cell, and generally only labels proteins within 10-15nm away from the originating BirA source (Kim et al. 2014).  As a result, the technique has primarily been utilized for identifying the location of soluble proteins sequestered in cellular compartments, as other highly abundant proteins in the compartment are within proximity to the protein of interest and become biotinylated (Cronan et al. 2005; Kim and Roux 2016).    27  1.3.2.2  Protein Correlation Profiling and 2D-gel analysis methods    The second method for determining protein-protein interactions by proteomic methods relies on the fact that proteins which are in a stable complex co-fractionate.  By extension, measuring how protein are correlated to each other during fractionation can be an indicator of complex formation (Fig 1-8E).  This has been done through two main methods, either by 2D-gel or protein correlation profiling.  In 2D-gel experiments, protein complexes are separated in one dimension, often blue-native page or isoelectric separation, and then separated a second time on a denaturing gel.  Staining of the 2D gel can then be used to visualize denatured proteins that line up in the same “channel” from the non-denaturing gel.  Excision of the spots, followed by digestion and proteomic identification allows identification of proteins that are possibly in a complex.  This method has been used with some success over the years, but it is impractical for high-throughput analysis of the entire interactome (Bunai and Yamane 2005). Protein correlation profiling combines high resolution fractionation with quantitative proteomic analysis (Kristensen, Gsponer and Foster, 2012).  This allows determination of the elution and co-elution profiles of thousands of proteins in a single experiment.  The success of the method depends on a suitable fractionation method to provide enrichment of specific complexes against overabundant proteins, as well as accurate and reproducible quantitation of each protein to build comprehensive elution profiles.  Post-digestion labelling of peptides can be for quantitation however, labelling efficiency can vary and therefore reduce accuracy. To avoid these issues, methods such as label-free quantification (LFQ), or stable amino acid isotopologue labelling in cell culture (SILAC) can be employed.  In this thesis, quantification of cell envelope proteins is accomplished by SILAC, which is further discussed in the following section.  1.4 Stable amino acid isotopologue labelling in cell culture (SILAC)   SILAC is a robust method for labelling the proteome of two or more cell cultures and without altering the chemistry of incorporated amino acids.  Cell growth in the presence of amino acids that contain stable nitrogen and carbon isotopes results in generation of a “heavy” proteome.  Since only heavy amino acids are present during cell growth, a proteome can be labelled with near 100% efficiency, allowing it to be easily discerned from other unlabeled proteome by MS (Ong, Foster, and Mann 2003).  This is because the inclusion of heavy amino acid isotopologues 28  into tryptic peptides results in defined mass shifts in the mass spectrogram (Fig 1-10). As the amino acid isotopologues are chemically identical, the peptides will also act identically during the preparative steps leading up to their identification by MS. Isotopologues can be used under different experimental conditions so that the corresponding peptide intensities precisely reflect the difference in protein abundance between conditions (G. Zhang and Neubert 2009).  Compared to label-free quantitation, which requires internal standards and separate experiments for each condition, SILAC quantitation provides higher quantification accuracy without internal standards as well as 2 to 3 fold less instrument runs due to sample multiplexing (analysis of two different conditions in the same MS run).  However, a disadvantage of SILAC is the requirement of amino acid auxotrophy, so that the proteome can be labelled by the heavy amino acid with near 100% efficiency, as tools are not available to genetically alter all organisms.  In addition, while SILAC is a powerful approach for relative quantitation, absolute quantitation with this method is challenging (Ong, Foster, and Mann 2003).  Nonetheless, the SILAC approach has been used to identify soluble interactors of membrane proteins in AP-MS experiments (Zhang et al. 2012), analyze tissue and cell wide changes in the proteome due to cell differentiation (Ong et al. 2002), and lastly, determine elution profiles with high resolution during PCP experiments (Kristensen, Gsponer, and Foster 2012; Scott et al. 2017).  In summary, PCP-SILAC is an exceedingly powerful method for quantitative measurements of native protein levels by proteomics.  In chapter 3, we utilize SILAC to build enrichment profiles for the purposes of PCP, which are analyzed by a suite of bioinformatics algorithms termed Prediction of interactomes from co-elution data (PrInCE) (Stacey et al. 2017).  29    Figure 1-10.  Overview of SILAC.  Cells growing in normal, light medium are subcultured in medium containing heavy, stable isotope-labeled amino acids like lysine with six 13C6 (highlighted in red). Cell growth, protein synthesis and turnover lead to metabolic incorporation of the SILAC amino acid throughout the proteome. Digestion of proteins with lys-C or trypsin results in peptides bearing lysine-13C6 and the residue-specific mass shift of 6 Da from the light peptide, thereby distinguishing the two forms for quantification by MS.  Figure reproduced with permission from reference (Ong and Mann 2006).   1.5 Prediction of interactomes from co-elution data (PrInCE)      PrInCE is a bionformatics pipeline developed by Stacey et al. that utilizes  co-fractionation data to build lists of binary interactions.  Once given a set of elution profiles from a single experiment, the data is “cleaned” by removal of incomplete chromatagrams (<5 30  consecutive fractions) and approximation of missing points (For a missing point between two fractions, an averaged value derived from the flanking fractions is input).  The “cleaned” chromatograms then go through 5 separate steps included in the PrInCE pipeline (Fig 1-11c).  First, to resolve discrete populations of proteins (i.e. dimers vs monomers vs oligomers), Gaussian-like peaks are identified in the co-fractionation profiles (Stacey et al. 2017; Kristensen, Gsponer, and Foster 2012).  Second, the replicate experiments are aligned to correct the curves for slight differences in the elution dimension.  Third, if the experiment includes a third label to compare between different conditions (multiplexing), the program will detect the relative changes in protein quantities between conditions.  Fourth, protein-protein interactions are predicted for each condition.  Five different distance scores together provide an “interaction score” value that reflects the similarity between elution profiles.  For a pair of co-fractionation profiles (defined as ci and cj) these distance measures are taken into account:  ● One minus correlation coefficient, 1 − R corr: One minus the Pearson correlation coefficient between ci and cj . ● Correlation p-value, p corr: Corresponding p-value to 1 − R corr. ● Euclidean distance between co-fractionation profiles ci and cj, E. ● Peak location, P: Calculated as the difference, in fractions, between the locations of the maximum values of ci and cj . ● Co-apex score, CA: Euclidean distance between the closest (μ, σ) pairs, where μ and σ are Gaussian parameters fitted to ci and c j.  Thus chromatograms with at least one pair of similar Gaussian peaks will have a low (similar) Co-apex score.  These combination of distance measurements let PrInCE classify proteins as interacting pairs (Fig1-11a), or non-interacting pairs (Fig1-11b).  Raw elution profiles obtained from high resolution data can be sufficient for PrInCE to adequately identify individual protein complexes.  However, additional source of information can be included to increase confidence of the prediction, such as co-expression data, annotation similarity, or AP-MS experiments.  Finally, PrInCE incorporates the complex building module based on the clusterONE algorithm (Nepusz, Hu and Paccarno, 2012) which is used to cluster interaction data into discrete complexes.  Importantly, rather than set arbitrary distance cut-offs by the researcher,  PrInCE uses machine 31  learning and a gold-standard data set to optimize the distance restraints of each module.  It does this by finding the best gold standard data set of experimentally determined protein interactions in the raw elution profiles.  Specifically, a Naïve Bayes classifier is trained to evaluate the resemblance between the distance measures for a candidate protein-protein pair and the distance measures for the gold standard reference interactions  (Stacey et al. 2017).  This method allows comparison of the predicted interactome with a reference data set, so that a precision value (i.e probability of a predicted interaction) can be assigned.  However, the sole reliance on a gold-standard can make interaction predictions difficult if i) a suitable reference does not exist, or ii) the reference data-set is not well identified in the data-set (Stacey et al. 2017).        Figure 1-11.  PrInCE Pipeline overview. a. Co-fractionation profiles from known interactors, ribosomal proteins P61247 (black) and P62899 (grey). b. Co-fractionation profiles from non- interacting protein pair, Q6IN85 (black) and E9PGT1 (grey). c. Pipeline workflow. Raw data consists of co-fractionation profiles grouped by replicate and condition. In pre-processing, Gaussian mixture models are fit to each co-fractionation profile to obtain peak height, width, and center. If there are multiple replicates, the Alignment 32  module adjusts profiles such that Gaussian peaks for the same protein occur in the same fraction across replicates. Changes in protein amounts between conditions, i.e. fold changes, are computed in the FoldChange module. Interactions between pairs of proteins are predicted by first calculating distance measures between each pair of proteins and feeding these values into a Naive Bayes supervised learning classifier. Known interactions from a reference database, e.g. IntAct, are used for training. Finally, the list of predicted pairwise interactions is processed by an optimized ClusterONE algorithm to predict protein complexes.  Figure and caption are reproduced with permission from reference (Stacey et al. 2017).  1.6 On the structure and interactome of the E. coli cell envelope    Compartmentalization is a fundamental characteristic of life.  All living organisms are composed of cellular units which are delineated from the surrounding millieau by a plasma membrane.  In cells, the cellular membrane is composed of a phospholipid bilayer that surrounds the cytoplasm.  However, in prokaryotes, the boundary around the cytoplasm consists of more than a single bilayer.  Instead, prokaryotes contain a multi-layered structure termed the cell envelope (CE) (Silhavy, Kahne, and Walker 2010).  The Gram stain, developed over a 100 years ago, allowed segregation of bacteria based on fundamental differences between the cell envelopes.  In Gram positive the plasma membrane is bounded to a large, multilayered sacculus made of oligosacharides cross-linked by specialized peptides termed peptidoglycan.  In Gram-negative bacteria, a much thinner peptidoglycan sacculus surrounds the plasma membrane, followed by an outer membrane (Silhavy, Kahne, and Walker 2010).  The CE plays important roles in cell division, environmental sensing, antibiotic resistance, energy generation and nutrient retention.  These crucial functions are supported and maintained by the actions of a number of different proteins that.  Accordingly, up to 30% of the E. coli proteome is thought to reside in the CE (Stenberg et al. 2005), and like soluble proteins, many CE proteins reside in multi-protein complexes that act together to complete their function (Babu et al. 2018).  However, unlike soluble proteins, the majority of CE proteins are insoluble in aqueous buffers and are therefore recalcitrant to many techniques for characterizing these interactions.  Recently, the global interactome of the E. coli cell envelope was described (Babu et al. 2018).  In the work, AP-MS was validated by label-free PCP approaches, to build a high confidence E. coli interactome.  While the predicted interaction network (interactome) was exceedingly robust, as it was constructed and validated by multiple different approaches, the in vitro assays were all accomplished in detergent are time, labour and resource intensive to repeat under different 33  conditions for comparative interactomics.  In chapter 4, I report a high-throughput method for characterization of the SILAC labelled E. coli cell envelope interactome using detergent-free fractionation and tandem liquid chromatography mass spectrometry (LC-MS/MS).   1.7 Channels and Transporters      Biological membranes separate the inside and outside of cellular compartments, controlling material flow by their selective permeability.  This selective permeability is largely accomplished by the controlled actions of integral membrane proteins that span the membrane.  These membrane proteins have been largely divided into two different classes; passive channels, which facilitate molecule movement down their gradients of concentration, and transporters, which release energy from ATP or other sources to actively move molecules against their concentration gradients.  Researchers have broadly classified proteins into one of these two categories, providing detailed mechanistic models to explain the protein movements required to catalyze each of these processes.  However, with growing structural and biochemical understanding of membrane proteins, the line between channels and transporters is becoming increasingly blurred.  In this section, I will utilize the maltose transporter MalFGK2, and through several non-destructive mutations, convert it into a bona-fide chloride channel.  The study provides evidence for the expanded alternating access model, discussed below, and has implications for the genetic evolution of transporters and channels.              1.7.1 Canonical alternating access model     About 50 years ago at the University of Cambridge, Dr. Jardetzky described the alternating access model as a mechanism for the energetically uphill movement of small molecules and ions catalyzed by transporters (Jardetzky 1966).  In the alternating access mechanism, the coordinated actions of two gating elements allows for movement of substrate (Fig 1-12b).  By ensuring one gate is always closed during transport, diffusion of the substrate through the transporter is prevented (Jardetzky 1966; Gadsby 2009).  This is the primary structural distinction between how we describe a channel and a transporter.  In a channel, substrate movement is controlled by one gate (Fig. 1-12a) (Gadsby 2009).  If the gate is open, substrate is free to travel bi-34  directionally through the channel along its concentration gradient (Gadsby 2009).  Because movement through a channel is only limited by substrate diffusion, the rate of transport is magnitudes higher than observed in transporters, and can be detected as electrical currents if the transported substrate is ionic in nature (Gadsby 2009).  However, the two gates of a transporter theoretically make free diffusion of a substrate impossible; the substrate eventually will be occluded in the transporter (Fig 1-12C), limiting transport rates below those necessary to manifest as electric currents (Gadsby 2009).  This very observable difference in transport rates is the primary mechanistic basis to differentiate channels and transporters and easily observable by conductance measurements (Gadsby 2009).  The one gate versus two gates concept is structurally and functionally well represented in crystal structures of both channels and transporters, although there are some exceptions, which are described in the following section.       35    Figure 1-12. The one vs two gate model distinction between channels and transporters.  A) Schematic representation of a channel as membrane-spanning pore through which movement of dissolved substrate (red circles) is controlled by a single gate, cartooned here as a hinged door. B) Transporter as membrane-spanning pore with two gates that open and close alternately. Coupling of an energy source to switch the relative binding affinity for red vs. blue ions between the left-and right-hand states enables active exchange of red for blue ions across the membrane. C) Occluded states, with both gates closed around bound ions, preclude inadvertent opening of the second gate before closure of the first gate, which would otherwise allow ions to flow down their electrochemical potential gradient several orders of magnitude faster than the 36  pump can move them against that gradient; pumped ion movement is rate limited by the gating reactions, rather than by electrodiffusion.  Figure modified with permission from reference (Gadsby, 2009). 1.7.2 Ion and water leakage in alternating access transporters   Probably the most well known example of a group of structurally almost identical proteins that can function as both transporters and channels are the CLC chloride channels.  Select members of this family, with almost identical substrate translocation pathways in their respective crystal structures, can function as either a bonafide chloride channel or H+/Cl- exchange -pump (Gadsby 2009).  This is surprising, as it demonstrates that the structural delineation between what constitutes a channel or transporter is not as easily defined as a simple one gate-two gate concept.  In addition, such a high degree of similarity in structure, yet completely different mechanisms of transport, suggest that these proteins are evolutionarily linked and that a channel may evolve from a transporter (Chen and Hwang 2008; Ashcroft, Gadsby, and Miller 2009).  The blurred boundary between channel and transporter is also apparent in the cystic fibrosis transmembrane conductance regulator (CFTR), which upon mutation, is the causative agent of cystic fibrosis (Gadsby, Vergani, and Csanády 2006).  CFTR is our only example of an ABC transporter that naturally acts as a chloride channel with physiological activity (Li, Ramjeesingh, and Bear 1996; Bear et al. 1992).  While the channel activity in transporters could be thought of as a rare oddity of structure, there are actually multiple documented cases of transporters with suspected water or ion leakage (DeFelice and Goswami 2007), despite clear occlusion of the substrate translocation pathway in their open or closed crystal structures (Li et al 2013).  This has lead to the proposal of the expanded alternating access model (Li et al. 2013).   1.7.3   The expanded alternating access model   In the expanded alternating access model, a membrane transporter is hypothesized to transition through multiple, short-lived intermediates as it moves from the inward to outward conformation.  During these transient intermediates, the gates of the transporter are not perfectly sealed, allowing ion and water, but not substrate, leakage to occur (Jing Li et al. 2013).  However, as the states are exceedingly transient, they are almost impossible to detect without the use of molecular dynamics simulations (Li et al 2013).  In this chapter, I will investigate whether 37  a transporter can be converted into a channel by destabilizing its resting state so that it more frequently samples the intermdiate states hypothesized to allow ion and water conduction posed by the expanded alternating access model (Figure 1-12B).  To examine this question, I utilize the maltose transporter MalFGK2, the structure of which has been solved in both its inward facing and outward facing conformations.     Figure 1-13.  Models for the conformational movements of an ABC transporter due to nucleotide binding as described by the alternating access model.  A) The canonical alternating access model modeled with an ABC transporter.  Nucleotide binding causes an immediate conformational switch the outward-facing conformation, which is sealed for ion or water transport.  ATP hydrolysis switches the transporter back to the inward facing state, which is also sealed.  B) The expanded alternating access model modelled with an ABC transporter.  During or slightly before ATP binding, conformational flexibility of the transporter allows micro-states to occur where the transporter is not entirely sealed.  The transient intermediate can allow water/ion leakage to occur.  This figure is modified from reference (Carlson et al. 2016)      1.8 The type I ATP Binding Cassette (ABC) transporter MalFGK2 and its periplasmic binding protein MalE    The maltose transporter, MalFGK2 is encoded by the genes malF, malG, and malK contained in an operon at 91.5min of the E. coli chromosome, in the malB region of the E. coli chromosome.  38  These genes make up a Type I ABC transporter.  The membrane domains are formed by MalF and MalG, while the nucleotide binding domains are formed by two copies of MalK (J. Chen 2013).  Together, these proteins have become an invaluable model system for how ABC transporter’s function, in large part due to the high resolution crystal structures determined by a group out of Rockefeller university lead by Dr. Jue Chen.  MalE, a soluble substrate binding protein (SBP) located in the periplasm, is responsible for binding the transported substrate maltose and stimulating transport activity by the transporter (Davidson, Shuman, and Nikaido 1992).  In addition to procuring specificity, the coupling of the substrate binding protein to transporter stimulation can be important for prevention of unnecessary ATP hydrolysis (Davidson, Shuman, and Nikaido 1992; Bao and Duong 2012).  However, to fully understand how the substrate binding protein can be so important for regulation of ATPase activity of the transporter, we can turn to the crystal structures of the transporter in its apo and ATP bound states (Figure 1-14), which are reviewed in the following section.    Figure 1-14. A) Inward (apo,) and B) outward (ATP-bound) structures of the E. coli maltose importer MalFGK2.  The ribbon models shown were modelled in chimera using the deposited crystal structures (PDB 39  accession codes 3FH6 and 2R6G for inward facing and outward facing conformations, respectively).  Colour is used to distinguish the subunits as described; MalF (green), MalG (purple), MalK (cyan), and MalE (orange). 1.8.1 Coupling of conformational changes with enzyme activity of MalFGK2   In the apo state, MalFGK2 is inward facing, and the large P2 loop that binds MalE is flexible and therefore unresolved in the X-ray diffraction (Fig. 1-14A).  Binding of ATP causes the transporter to shift to its outward state, where the substrate binding pocket becomes available to bind maltose (Bao and Duong 2013) .  At this point, the substrate binding protein can stably bind the transporter through interactions with exposed periplasmic loops (Fig. 1-14B), stabilizing the outward state and catalyzing ATP hydrolysis (Bao et al. 2015).  This view of ATP turnover in the maltose transporter suggests a mechanism that relies on a delicate balance between its inward and outward facing conformations.  Simple binding of bound nucleotide in the cytoplasm, or MalE in the periplasm is enough to tilt conformational equilibrium from the inward to outward state, and vice versa.  In fact, slow swapping between the inward and outward states allows the transporter to go through full cycles of ATP hydrolysis without the addition of maltose or MalE, called the basal ATPase activity of the transporter ((Bao and Duong 2012; Bao et al. 2013; Böhm et al. 2013).  This equilibrium between the inward and outward conformations can be directly tied to conformational flexibility of the transporter (Bao and Duong 2012; Bao et al. 2013).  In a detergent environment, where the transporter experiences the most flexibility, basal ATP hydrolysis is high, and is not significantly stimulated by MalE or maltose (Bao et al. 2013).  In a more restrained nanodisc environment, there is still a high basal ATPase activity, but the stimulatory effect of MalE becomes more pronounced (Bao et al. 2013).  However, in the most restrained environment provided by a lipid bilayer, the transporter loses almost all basal ATPase activity unless in the presence of MalE and maltose (Bao and Duong 2012; Bao et al. 2013).    1.8.2 Mutants of MalFGK2 which decouple ATPase activity of the transporter from stimulation by MalE         The ABC transporter MalFGK2 is a highly coordinated molecular machine for high affinity transport of maltose.  However, substantial mutational work on the transporter has identified 40  multiple mutations that can deregulate the transporter from stimulation by its substrate binding protein (Covitz et al. 1994).  These substrate binding protein independent (SBP-independant) mutants generally have a very high basal ATPase activity and a gain-of-function ability to transport maltose independent of MalE (Covitz et al. 1994; Amy L. Davidson 2002).  Of these mutants, the MalF500 mutant is probably the best characterized.  Two mutations in the MalF-MalG interface destabilize the transporter, which allows it to rapidly cycle through its inward and outward conformations.  In Chapter 4, I will utilize the MalF500 mutations to increase the dwell time and frequency of transient states experienced by MalFGK2 described in the expanded alternating access model.    1.9                  Overview of Objectives  The work presented herein is unified by its focus on an underappreciated portion of cell biology, membrane proteins.  Critically important to cell function, study of these proteins represents a particularly challenging problem for scientists, both technically and conceptually.  To contribute a better understanding of these important proteins, and better yet, enable other researchers to continue studying these proteins through development of new methods, was the driving forces behind the studies detailed in this thesis.   My efforts to contribute to the scientific community are outlayed in the following chapters, with the objective for each chapter summarized as follows: In Chapter 2, I aim to design and validate a method that can rapidly and flexibly incorporate membrane proteins into detergent free buffer.  It is important that the designed method be scalable, so that large quantities of membrane proteins can be incorporated into the scaffold.  The method must also display little preference for reconstitution of discrete populations or sizes of membrane proteins.  In other words, I aimed to develop a “one-size fits all” reconstitution scaffold.  It was also imperative that the method results in highly monodisperse membrane protein reconstitutions, and be relatively cheap, as previous scaffolds that did not satisfy these requirements have been met with mixed adoption by the biochemical community.  In addition, these specifications were required so that I could apply the peptidisc to identification of the cell envelope interactome of E. coli, which was the main objective of Chapter 3.  In Chapter 3, my experimental objectives were to fractionate the cell envelope of E. 41  coli in detergent free buffer so that it can be analyzed by the bionformatics pipeline PrInCE.  I test two different separation methods, and perform validation experiments (computational and in vitro) to complement the identified interactions in PrInCE.  In Chapter 4, my work focused on investigating the alternating access mechanism, utilizing the well-documented ABC transporter MalFGK2.  My objectives here were to demonstrate that the MalF500 mutant biases the transporter to more frequently sample the transition state, and then measure whether ion flux could be detected.  I also investigated the role of the periplasmic gate in regulating this channel activity.             Chapter 2:  Design and validation of the Peptidisc method, a simple yet efficient procedure for stabilizing membrane proteins in detergent-free buffer  2.1 Introduction        Membrane proteins play essential roles, such as membrane transport, signal transduction, cell homeostasis, and energy metabolism. Despite their importance, obtaining these proteins in a stable and non-aggregated state remains problematic. Membrane proteins are generally purified in detergent micelles, but these small amphipathic molecules are quite often detrimental to protein structure and activity, in addition to interfering with downstream analytical methods. This drawback has led researchers to develop detergent-free alternatives such as amphipols (Popot 2010), SMALPs (Dörr et al. 2014), saposin A lipoparticles (Frauenfeld et al. 2016), and the popular nanodisc system (Denisov and Sligar 2016; Ritchie et al. 2009). In the later case, two amphipathic membrane scaffold proteins (MSPs) derived from ApoA1 stabilize a small patch of 42  lipid bilayer containing the membrane protein. (Ritchie et al. 2009; Grinkova, Denisov, and Sligar 2010). However, in spite of an apparent simplicity, the formation of a nanodisc depends on several critical factors such as lipid to protein ratio, scaffold length, type of lipids, rate of detergent removal and overall amenability of the target for re-assembly into lipid bilayer (Ritchie et al. 2009; Grinkova, Denisov, and Sligar 2010; Bayburt, Grinkova, and Sligar 2006). The method is not trivial and small deviations from optimal conditions often leads to low-efficiency reconstitution, or else liposome formation or protein aggregation (Ritchie et al. 2009; Bayburt, Grinkova, and Sligar 2006).        Peptides have been considered as an alternative to scaffold proteins and synthetic polymers.  Peptergents (Corin et al. 2011),  lipopeptide detergents (McGregor et al. 2003), nanostructured [beta]-sheet peptides (McGregor et al. 2003), and bi-helical derivatives of the ApoA1-mimetic peptide 18A, termed “beltides” (Larsen et al. 2016), have all been proposed as alternatives to detergents for solubilization of membrane proteins.  However, these systems have not been widely adopted for structural or functional characterization of membrane proteins as scaffold proteins or synthetic polymers, for a variety of reasons.  Peptergents and lipopeptide detergents are able to solubilize membrane proteins directly from a lipid bilayer, but the formed particles are still mixed micelles, which become unstable if removed from excess peptide (McGregor et al. 2003; Corrin et al. 2011).  Beltides were shown to trap bacteriorhodopsin in solution, but the method required prior incubation with lipids and generated particles which were reportedly unstable at physiological temperatures (Larsen et al. 2016).  To increase sequence hydrophobicity, nanostructured [beta]-sheet peptides contain extended alkyl chains covalently linked to glycine residues, while lipopeptide detergents are covalently linked to a lipid molecule (McGregor et al. 2003; Tao et al. 2013).  Additional amino acid modifications can substantially increase complexity and cost of peptide synthesis, providing another barrier to adoption (Privé 2009).  In addition, such peptides are often difficult to work with from a solubility perspective.  Peptergents require careful titration of base and sonication to become soluble in aqueous solution (Corin et al. 2011), while nanostructured [beta]-sheet peptides can form extended filaments in the absence of detergents (Tao et al. 2013).        We therefore develop a peptide based reconstitution method that is flexible, cost-effective, and unhindered by issues of low scaffold solubility which we call the peptidisc. The peptidisc is made by multiple copies of an amphipathic bi-helical peptide (NSPr) wrapping around a target 43  membrane protein. No additional lipids are necessary, except perhaps those that have co-purified with the target membrane protein (i.e. endogenous lipids). The peptide we present (called NSPr) is a reversed version of the original nanodisc scaffold peptide (NSP), which consists of two repeats of the ApoA1-derived 18A peptide joined by a flexible linker proline (Kariyazono et al. 2016), in addition to two leucine residues which are substituted by phenylalanines to increase lipid affinity and stability (Kariyazono et al. 2016) (Supplementary Table 3).  While the original NSP was shown to stabilize lipid particles (Kariyazono et al. 2016), issues with solubility as well as applicability to membrane proteins was not tested. We show here the general applicability of NSPr for stabilizing both α-helical and β-barrel membrane proteins of different size, topology, and complexity.  2.2 Materials and Methods 2.2.1  Plasmids and biological reagents      Tryptone, yeast extract, NaCl, imidazole, Tris-base, acrylamide 40%, bis-acrylamide 2% and TEMED were obtained from Bioshop, Canada. Isopropyl β-D-1-thiogalactopyranoside (IPTG), ampicillin, and arabinose were purchased from GoldBio. Detergents n-dodecyl-β-d-maltoside (DDM) and octyl-β-D-glucoside (β-OG) were from Anatrace. Detergent N,N-dimethyldodecylamine N-oxide (LDAO) was from Sigma. Total E. coli lipids were purchased from Avanti Polar Lipids.  Resource 15Q, Fast Flow S, Superdex 200HR 10/300 GL and 5/150 GL were obtained from GE Healthcare. Ni2+-NTA chelating Sepharose was obtained from Qiagen. All other chemicals were obtained from Fischer Scientific Canada. Peptide NSP and NSPr were obtained from A+ peptide Co., Ltd. (Purity >80%) or Genscript (purity >80%). NSPrbio was obtained from KareBay (purity >80%).  To aid accessibility, bulk NSPr peptide is now available from www.peptidisc.com/ 2.2.2  Preparation of the NSP peptides  For solubility test experiments, lyophilized NSP and NSPr (purity of 82% and 85%, respectively) were resuspended in dH20 at room temperature to final concentrations of 15 mg/mL and 25mg/mL, respectively. Peptide concentration was determined by absorbance at 280nm.  44  Residual trifluoroacetic acid from peptide synthesis results in a low pH solution (pH 2-3).  For all other experiments, peptides were solubilized in dH2O at 6mg/mL.  Solubilized peptides were stored at 4°C for up to 5 weeks.  Immediately before use, the pH of the peptide solution was modified by addition of 20mM Tris-HCl, pH 8 to form the so-called Assembly Buffer.  Immediately before use in peptidisc reconstitutions, peptide concentration in Assembly Buffer was verified by Bradford assay (Bradford 1976).      2.2.3  On-column reconstitution   MalFGK2 (300 µg) in Buffer E + 0.02% DDM was mixed with NSP (480 µg) in Assembly Buffer in a total volume of 100µL. The mixture was immediately injected onto a 100µL loop connected to a Superdex 200HR 5/200 GL column running at 0.4 ml/min in Buffer AC (50mM Tris-HCl, pH 8; 100mM NaCl).  Fractions were collected, pooled, concentrated using a 100kDa polysulfone filter (Pall Corporation), and stored at 4°C.  For on-column reconstitution of FhuA, 500µL of the protein (1mg) was mixed with NSP (1.8mg) in Buffer A + 0.05% LDAO, and injected onto a 500 µL loop connected to a Superdex 200HR 10/300 GL column running at 0.5 ml/min in Buffer AC.   2.2.4  In-gel reconstitution   The target membrane protein (~1.25 µg) was mixed with increasing concentrations of NSPr (0-2.5 µg) and allowed to incubate for 1-2 minutes at room temperature.  The mixture was then supplemented with Buffer A to bring the final detergent concentration below its CMC (0.008% and 0.01% for DDM and LDAO, respectively) while keeping the final volume to 15 µL. A solution of glycerol was added to 10% final to facilitate loading on 4-12% clear native polyacrylamide gel elecrtrophoresis (CN-PAGE). The electrophoresis was set constant at 25mA for 1 hour at room temperature.  Bands were visualized by Coomassie Blue G250 staining.       45  2.2.5  On-bead reconstitution   Crude membranes (10 mL at 7.5 mg/ml total protein content) containing overexpressed MalFGK2 were solubilized in Buffer A + 1% DDM for 1 hour at 4 °C before removal of insoluble aggregate by ultracentrifugation (100,000 x g, 1 hour, 4°C).  The solubilized membrane proteins were incubated with 200 µl of Ni-NTA resin (Qiagen) pre-equilibriated in Buffer A + 0.02% DDM for 1 hour at 4ºC.  The Ni-NTA beads were collected by low-speed centrifugation (3,000 x g, 3 min), washed twice with 10 CV of Buffer B supplemented with 0.02% DDM.  Post-washing, 10 CV of Assembly Buffer (1 mg/mL NSPr in 20 mM Tris-HCl pH 8) was added to the beads and allowed to incubate for 5 minutes on ice.  The Assembly Buffer was removed and the beads loaded into a gravity column with 10 CV of Buffer B (50 mM Tris-HCl, pH 8; 200 mM NaCl, 10% glycerol, 15 mM imidazole).  The assembled peptidiscs were subsequently treated with 500 µL of Buffer C (50 mM Tris-HCl, pH 8; 100 mM NaCl;10% glycerol; 400 mM imidazole) to elute the peptidisc from the affinity resin. The same procedure was done in parallel, except the NSPr was omitted from the Assembly Buffer and 0.02% DDM was included in Buffer A, B, and C.   2.2.6 Reconstitution of MalFGK2 and BRC into proteoliposomes   Proteoliposomes were prepared at a molar protein:lipid ratio of 1:2000.  Total E. coli lipids were dissolved in chloroform, dried under nitrogen and resuspended in Buffer A + 0.8% β-OG.  Purified MalFGK2 was added to the solubilized lipids, and the detergent was removed by overnight incubation at 4ºC with Amberlite XAD-2 adsorbent beads (Supelco). The proteoliposomes were isolated by ultracentrifugation (100,000 × g, 60 min at 4°C) and resuspended in 20 mM Tris–HCl, pH 8 before use in ATPase assays.  The same procedure was employed for the BRC, but a lipid mixture of DOPC:DOPG (80:20 mol/mol) was utilized in place of total E. coli lipids.   46  2.2.7  Reconstitution of BRC in peptidiscs, low-lipid nanodiscs, styrene maleic acid nanoparticles, DDM and SDS.   The purified BRC complex (1mg/mL) was mixed at a 1:1.8 (µg/µg) ratio with NSPr followed by 10-fold dilution in Buffer A to decrease the LDAO concentration to 0.003%.  For formation of low-lipid nanodiscs, the purified BRC complex was instead mixed at a 1:2 (mol/mol) ratio with MSP1D1 before dilution.  Alternatively, an equivalent amount of BRC was diluted in Buffer A supplemented with 0.03% LDAO, 0.02% DDM, 0.1% SMA or 0.1% SDS as described.  After incubation for 10 minutes on ice, aggregated proteins were removed by centrifugation (13,000 x g, 10 min at 4°C).  Peptidisc formation was confirmed by analysis on CN-PAGE.    2.2.8 FhuA binding assay   FhuA-MSPL156 nanodiscs were prepared as previously described (Mills et al. 2014).  FhuA-NSPr was prepared by on-column peptidisc reconstitution.  About 2 µg of FhuA reconstituted into either MSPL156 or NSPr was incubated with TonB23-329 (2 µg) or ColM (5 µg) in the presence or absence of ferricrocin for 5 minutes at room temperature. The protein complexes were separated by CN-PAGE and visualized by Coomassie blue staining. Neither monomeric TonB nor ColM migrate on CN-PAGE due to their isoelectric points > pH 8.8.   2.2.9 Absorbance spectroscopy   Absorption spectra were recorded using a Hitachi U-3010 spectrophotometer. A blank measurement was recorded in Buffer A (+0.03% LDAO for detergent purified BRC). Samples were incubated in a PCR thermocycler at the indicated temperature, and then measured at the desired time points in a quartz cuvette at room temperature.  Spectra were collected between 600 nm and 1100 nm (scan time ∼20 sec) at intervals of 1.5 min.  For comparisons of spectra between conditions, spectra were normalized to a value of 1.0 at 804 nm.    47  2.2.10   NSPr quantification The MalFGK2 and BRC peptidiscs were prepared by on-column reconstitution on a Superdex 5/25 column equilibrated in Buffer A, followed by one additional gel filtration step to ensure full removal of free NSPr.  MalFGK2 (1 µg), FhuA (2 µg), and BRC (2 µg) peptidiscs were analyzed by 15% SDS-PAGE.  Gels were stained with Coomassie Blue G-250, and destained overnight before fluorescence measurement (excitation 680nm, emission 700nm) on a LICOR Odyssey scanner. The band corresponding to the NSPr peptide was quantified by densitometry using Image J and compared to a standard curve of NSPr (0-2 µg) loaded on the same gel. The determined NSPr amount was then subtracted from the total amount of protein loaded on the gel to determine the amount of reconstituted membrane protein in the peptidisc.  Membrane protein content in peptidisc (g) = total protein in peptidisc (g) - measured NSPr content (g). We used these calculated mass measurements and the molecular weight (MW) for NSPr (4.5kDa), MalFGK2 (173kDa), FhuA (80kDa) and BRC (94kDa) to calculate NSPr stoichiometry as follows;   NSPr Stoichiometry = MW Membrane protein (g/mol)  x          Measured NSPr content   (g)                                      MW NSPr (g/mol)        Membrane protein content in peptidisc (g)   Each experiment was repeated in triplicate on three different gels.  We note that detergent-purified FhuA co-purified with a contaminant, thought to be short chain lipopolysaccharides, that migrated to the same position as NSPr, therefore FhuA was reconstituted using NSPr labelled with a biotin group (NSPrbio). To quantify NSPrbio, western blots were incubated with streptavidin conjugated to Alexafluor 680 in phosphate buffered saline (PBS), followed by several washes in PBS + 0.1% Tween. Western blots were imaged on a LICOR Odyssey scanner fluorescence (excitation 680nm, emission 700nm), and the bands corresponding to NSPrbio quantified in Image J software suite.   2.2.11   Lipid extraction and quantification   The MaFGK2 and BRC peptidiscs were prepared on-bead, and the FhuA peptidisc was prepared on-column.  MalFGK2 (40 µg), FhuA (40 µg), and BRC (80 µg) peptidiscs were diluted to a 48  final volume of 200 µL of Buffer A, then mixed with 800 µL of a 2:1 solution of methanol:chloroform for 10 minutes at 25°C in glass screw cap vials.  200 µL of chloroform and 200 µL of distilled water were added sequentially, vortexed briefly, and the resulting two phase system separated by low speed centrifugation (3,000 r.p.m., 10 minutes).  The organic phase was dried under nitrogen, and stored at -20°C. Total phosphate content was determined by a modified version of the malachite green assay (Lanzetta et al. 1979).  Briefly, malachite green reagent was prepared as follows: ammonium molybdate (4.2 g) was dissolved in 100 mL of 4M HCl, then mixed with 300mL malachite green (135 mg) dissolved in distilled water.  The solution was mixed for 1 hour at 4°C, filtered, and stored at 4°C before use.  Dried lipid extracts were subsequently incubated with 1 mL of 70% perchloric acid for 3 hours at 130°C, and then 20 µL of the resulting solution mixed with 500 µL of the malachite green reagent for 5 minutes at room temperature before absorbance measurement at 660 nm. Phosphate standards (KH2PO4) were diluted into perchloric acid and used to prepare a standard curve with phosphate concentrations ranging from 0.01nmol to 1 nmol PO4.  For thin layer chromatography (TLC) analysis, dried lipids were resuspended in 30 µL of chloroform, and 10 µL were dotted onto a TLC Silica gel 60 (Millipore).  The TLC was developed in a solution of 35:25:3:28 chlorofrom:triethylamine:dH2O:ethanol.  Plates were dried in an oven for 5 minutes at 150°C.  Lipids were visualized by lightly wetting plates in a solution of 10% Cu2S04 in 8.5% phosphoric acid, followed by heating for 5 minutes at 150°C.      2.3      Results 2.3.1  Solubility comparison of NSP and NSPr   The original NSP is able to directly capture lipids from a lipid bilayer into discoidal particles of varying size because of its high hydrophobicity (Vollmer, von Rechenberg, and Höltje 1999; Kariyazono et al. 2016).  However, NSP was difficult to solubilize; the solution was cloudy upon visual inspection at a concentration of 15 mg/mL after resuspension in water.  To increase peptide solubility, we reversed the amino acid sequence of NSP to form the NSPr.  In NSPr the two amphipathic helices have a slightly altered orientation, which leads to a decreased overall hydrophobic moment and increased electropotential of the peptide (Supplementary Figure 2-49  1AC).  To decrease complexity of peptide synthesis, we also forewent acetylation and amidation of the peptide. This small modification allowed for NSPR to be fully dissolved in distilled water at concentrations up to 25 mg/ml.  As NSPR fully dissolved, light scattering was negligible compared to a pure water control (Supplementary Figure 2-1B).  In contrast, the NSP resuspension resulted in significant light scattering, indicating incomplete solubilization of the peptide construct (Supplementary Figure 2-1C).   2.3.2  On column reconstitution of MalFGK2   With the soluble NSPr peptide, we develop “on-column”, “on-bead”, and “in-gel” methods for formation of peptidiscs.  We initially developed the “on-column” reconstitution method using the ABC transporter MalFGK2 (Fig. 2-1). The NSPr peptide was mixed with MalFGK2 in dodecyl maltoside and the mixture applied directly onto a size exclusion column equilibrated in a detergent-free buffer (Fig. 2-1A). The collected particles (MalFGK2-NSPR) were soluble and monodisperse, as shown by clear-native (CN) PAGE and blue-native (BN) PAGE (Fig. 2-1B).   2.3.3  Structure, composition and functional activity of MalFGK2 peptidisc   The approximate peptide content of the particles was determined using SDS-PAGE (Supplementary Figure 2-2A). Also, since some lipids (i.e. annular lipids) can remain tightly bound to membrane proteins during purification (Bechara et al. 2015), we also determined the lipid content by thin layer chromatography and photocolorimetric methods (Supplementary Figure 2-3AB).  The initial analysis indicated a stoichiometry of 10 ±2 peptide per MalFGK2 (Supplementary Fig 2-2A, Supplementary Table 1), and 41 ±10 lipids (Supplementary Figure 2A, Supplementary Table 2-1). This stoichiometry was used to calculate the mass of the MalFGK2 peptidisc (251 ± 12kDa; Supplementary Table 1). Negatively charged lipids are known to play a critical role in regulation of the transporter by stabilizing interactions with the regulatory protein EIIA during conditions of high glucose (Bao and Duong 2013).  Interestingly, the lipids identified in the TLC analysis of the MalFGK2 peptidisc were predominantly the negative phospholipids cardiolipin and phosphatidylglycerol (Supplementary Figure 2-2B), To corroborate the peptide and lipid stoichiometry, we determined the molecular weight of the intact 50  complex by native MS (247 ± 24 kDa; Supplementary Fig. 2-3A), and size exclusion chromatography coupled multi-angle light scattering (SEC-MALS) (250 ±17 kDa; Supplementary Figure 5A). The SEC-MALS analysis showed that the peptidisc is stable, even after extended storage (i.e. 3 days at 4°C). We then examined the particles by single particle negative-stain electron microscopy (Fig. 2-1C). The 2D-class averages revealed a structure very similar to MalFGK2 in nanodiscs (Fabre et al. 2017), with distinctly visible elements such as the MalK2 dimer, the periplasmic P2 loop and a large density corresponding to the NSPR peptides wrapping the MalFG membrane domain. The measured diameter of the peptidisc was 11.7 ± 1.4 nm, which is consistent with a stoichiometry of 12 ± 2 peptides per MalFGK2 complex when arranged in the double-belt model (Supplementary Table 2-2). Finally, the ATPase activity of MalFGK2 in peptidisc was similar to that reported in proteoliposomes and nanodiscs,  in contrast to the high and unregulated ATPase activity reported in detergent micelles (Fig. 2-2D).  Importantly, the structural integrity of the peptidisc was maintained at physiological temperatures, with ~ 80% of MalFGK2 peptidisc remaining stable after incubation at 30ºC for 3 hours (Supplementary Figure 2-5B).   51   Figure 2-1. The “on-column” reconstitution of MalFGK2. A) Typical size-exclusion chromatography of MalFGK2 in peptidisc (MalFGK2-NSPR) using the “on-column” method.  B) Clear-native (CN) PAGE analysis of MalFGK2 in detergent micelle (DDM), nanodisc (MSP1D1), and peptidisc (NSPR). Blue-native (BN) PAGE analysis of MalFGK2 in nanodisc (MSP1D1) and peptidisc (NSPR). C) Top panel: Field of view of particles stained with uranyl formate. Bottom panel: Selected class averages representing three characteristic views of MalFGK2 in peptidisc. The nucleotide binding domains (MalK2), the transmembrane domain (MalFG), and periplasmic P2-loop are indicated with yellow, red and blue arrows, respectively. D) Maltose-dependent ATPase activity of MalFGK2 (0.5 µM) reconstituted in detergent (DDM), proteoliposomes (PL), peptidiscs (NSP), and nanodiscs (MSP1D1) obtained at 30°C in the presence or absence of MalE (2.5 µM). Error bars represent standard deviations from 3 separate experiments.     52  2.3.4 Composition and functional activity of FhuA peptidisc   The ability of the peptide to trap β-barrel membrane proteins was next tested using the outer membrane transporter FhuA as a model (Supplementary Fig 2-6A).  Analysis of FhuA-peptide particles by MS (Supplementary Fig. 2-3E and 2-3F) indicated a molecular weight of 138 ±17 kDa.  Following the same approach applied to MalFGK2 above, we quantified the individual peptide and lipid components of the FhuA peptidisc (Supplemental Figure 2-2B and 2-2A), resulting in an average of 8 ±3 phospholipids and 10 ±2 NSPR per FhuA peptidisc (Supplementary Table 2-1). These measurements were used to determine a molecular mass of 131 ±9 kDa for the FhuA peptidisc (Supplementary Table 2-1).  Binding analysis on CN-PAGE further showed that FhuA in peptidisc is a functional receptor for TonB and colicin M (Supplementary Fig. 2-6B).  The binding of TonB and colicin M was modulated by the ligand ferricrocin, as previously reported in vivo and in vitro (Supplementary Fig. 2-6B). (Fabre et al. 2017; Mills et al. 2014; Wayne, Frick, and Neilands 1976). The peptidisc is therefore suitable for the functional reconstitution of both α-helical and β-barrel membrane proteins.   2.3.5  In gel reconstitution screening of optimal reconstitution ratios (RR50) for MalFGK2, OmpF3, Sec(YEG)n, and FhuA   The “on-gel” method (Fig. 2-2) was developed to determine optimal reconstitution conditions in a time- and cost-effective manner.  Small amounts of peptides (0-2.5 µg) were mixed with the target protein (~1.25 µg) in detergent solution, and the resulting mixture directly loaded on native gel. Removal of the non-ionic detergent occurred during electrophoresis when the protein-peptide mixture entered the detergent-free environment of the gel.  At the correct NSP ratio, the target membrane protein did not aggregate at the top of the gel but instead migrated in a soluble form to its expected molecular weight position. This simple reconstitution method allowed us to estimate the effective NSP concentrations required to trap 4 different integral membrane complexes into a peptidisc: MalFGK2 (Fig. 2-2A), FhuA (Fig. 2-2B), the trimeric OmpF porin (Fig. 2-2C), and the membrane translocon SecYEG (Fig. 2-2D). This titration analysis showed that membrane proteins were generally reconstituted in peptidiscs at similar peptide concentrations, with a half-maximal molar ratio (RR50) of 20 (Fig. 2-2E and Fig. 2-2F). This is 53  significantly higher than the measured stoichiometry of ~10 peptides per protein complex (Supplementary Table 2-1), suggesting that excess peptide is needed to achieve efficient reconstitution. This analysis also showed that SecYEG can be trapped in peptidisc as a dimer and higher-order oligomeric assemblies (Fig. 2-2C), probably due to the self-association of this membrane complex in detergent solution (Bessonneau et al. 2002). This later observation further differentiates the peptidisc from the nanodisc because in the nanodisc, the selective reconstitution of the SecYEG monomer and SecYEG dimer requires use of MSP proteins of different lengths (Supplementary Figure 2-7). (Dalal et al. 2012)  54   Figure 2-2: Express “in-gel” method for determining optimal reconstitution ratio. A) NSP and MalFGK2 were mixed at the indicated molar ratio for 2 minutes in Buffer A containing a low amount of detergent (~0.008% DDM) before loading onto CN-PAGE. Peptidisc reconstitution occurs during migration in the detergent-free gel environment. The same experiment was performed with B) FhuA, C) OmpF3, D) SecYEG. E) Reconstitution efficiency of FhuA as a function of the NSP concentration. The protein band FhuA-NSP in B) was quantified with Image J and the data plotted as log (mol NSP/mol FhuA).  The data were fitted with a Boltzmann sigmoidal function to generate a curve describing the reconstitution efficiency and the half-maximal reconstitution ratio (RR50). F) The RR50 was determined for other target proteins as in (E).  Error bars represent the standard deviation from 3 separate reconstitution experiments.   55  2.3.6  On bead reconstitution of MalFGK2 into peptidisc    The effectiveness of the NSPR peptide prompted us to develop an “on-beads” reconstitution method (Fig. 2-3), wherein affinity-purification of the target protein and incorporation into peptidisc are carried out simultaneously. As illustrated in Fig. 2-3A, an excess of NSP peptide was added when the target protein was still bound to the beads in the detergent environment. The detergent micelle was then progressively diluted during the washing steps (Fig. 2-3A, step 3). Finally, the peptidisc assembly was eluted in a detergent-free buffer (Fig. 2-3A, step 5). We tested the method using the his-tagged MalFGK2 complex (Fig. 3B). Analysis of the eluted complex by BN-PAGE and CN-PAGE showed that MalFGK2 is readily incorporated into water-soluble peptidiscs (Fig. 2-3C). The purity of the complex was equal to that obtained by conventional detergent-based chromatography (Fig. 2-3C).  Another, significant advantage of the “on-beads” method is removal and recycling of NSP for next reconstitution since the necessary amount of peptide interacts with bound membrane protein.   56   Figure 2-3: Direct “on-beads” reconstitution during membrane protein purification. A) Principle of the “on-beads” reconstitution. Step 1: the tagged protein is extracted from the membrane with sufficient detergent to solubilize the membrane (>> CMC) and incubated with the affinity resin. Step 2: The beads are washed twice with the detergent buffer near its critical micelle concentration (~ CMC). Step 3: The beads are incubated with buffer containing excess NSP and limited amount of detergent (< CMC). Step 4: The beads are washed in detergent-free buffer to remove unbound NSP and residual detergent. Step 5: The protein captured in peptidiscs is eluted from the column in detergent-free solution. B) SDS-PAGE and C) native-PAGE analysis of the his-tagged MalGFK2 complex purified following conventional detergent method and “on-beads” peptidisc method.   57  2.3.7  The peptidisc increases thermostability of BRC   Finally, we reconstituted the photosynthetic bacterial reaction center (BRC) from Rhodobacter sphaeroides, given its potential for biotechnological application. The BRC is employed in bio-hybrid solar cells due to its ability to absorb light in the near-infrared spectrum with high quantum efficiency (Blankenship et al. 2011; Ravi and Tan 2015; Yaghoubi et al. 2017). However, sustained heat and light exposure lead to irreversible loss of pigments and protein denaturation (Hughes et al. 2006; Swainsbury et al. 2014). Thermal stability and solubility at high protein concentration are therefore important parameters for successful application. The molecular weight of the BRC peptidisc measured by native MS is 138±18 kDa (Supplementary Table 1, Supplementary Fig. 4C and Fig. 4D).  It has been shown that purification of BRC in LDAO delipidates the complex (Swainsbury et al. 2014), and accordingly the BRC peptidisc contained only 4 ±1 phospholipids (Supplementary Fig 3A). Analysis by SDS-PAGE indicated a stoichiometry of 9 ±1 NSP per BRC (Supplementary Fig 2C). The calculated molecular weight was 138 ±5 kDa, in excellent agreement with native MS data (Supplementary Table 1).  We next measured the stability of the BRC pigments in peptidiscs or in LDAO detergent (Supplementary Fig. 2-8A and Fig. 2-8B). The spectral properties of the BRC complex were similar in both environments (Fig. 2-4A). However, the BRC in peptidisc resisted denaturation 65°C for 1 hour, while it was fully denatured in less than 4 minutes in LDAO (Fig. 4B). This difference corresponds to ~100 fold increase of the half-life of the BRC complex at elevated temperatures (Fig. 2-4C).    Figure 2-4: Thermostability of the BRC complex in peptidiscs. A) Absorbance scans of the BRC (1 µM) in detergent solution (0.03% LDAO, red trace) and in peptidisc (black trace).  Scans were normalized to the value measured at 803 nm (the absorbance peak of the accessory bacteriochlorophylls). B) Decrease in 58  absorbance of the BRC at 803 nm after incubation at 65°C for the indicated time.  C) Calculated half-life of the BRC in peptidisc and LDAO at 65°C. The data in B) were fit with an exponential decay function to determine the corresponding half-life. Error bars represent the standard deviation from 3 separate experiments.   2.3.8  Comparison of peptidisc to other membrane mimetics for BRC stabilization   We also compared the thermal stability of the BRC complex when reconstituted without additional lipids in SMA polymer and nanodiscs (Supplementary Fig. 8C and 8D). Without added lipid, the diameter of the BRC is too small for the MSP1D1 belt, resulting in some protein aggregation and heterogenous nanodisc preparation (Supplementary Figure 8F, left panel, lane 4).  Reconstitution into the SMA polymer without lipids was monodisperse (Supplementary Figure 2-8F, left panel, lane 5) and the thermostability was higher than in LDAO (Supplementary Figure 2-8E).  This observation is contrary to previous reports which suggest that the lipid environment in the SMA particle is what increases BRC thermostability (Swainsbury et al. 2014).  In our hands, reconstitution of the BRC into peptidiscs, proteoliposomes, low-lipid nanodiscs and SMA particles all result in comparable thermostability (Supplementary Figure 2-8E). From these results, it appears more important to remove detergents than include lipids to increase thermal tolerance of the BRC.   2.4  Discussion   Detergents remain a most effective way to extract and purify membrane proteins, yet these surfactants have many undesired effects on protein stability and downstream biochemical analysis. To circumvent these difficulties, and to handle these proteins like their water-soluble counterparts, membrane proteins are more often reconstituted with amphipathic scaffolds.  However, current methods of reconstitution are difficult because each scaffold system has specific properties and limitations, and each requires substantial optimization. The aim of our work was to develop a “one-size fits all” method to streamline the capture of membrane proteins in detergent-free solution.   59  2.4.1  The peptidisc methodology is simple and seamlessly integrates into existing membrane protein purification schemes  We show that the peptidisc is a simple and efficient way to replace detergent.  The system works with membrane proteins of different size, fold and complexity, and none require addition of exogenous lipids. The peptidisc captures membrane proteins regardless of their initial lipid content, and therefore the use of exogenous lipids to match the diameter of the scaffold such as in the nanodisc system is avoided.  Since the binding of the peptide is essentially guided by the size and shape of the protein template, the peptide stoichiometry is also self-determined.  As a direct consequence, the preparation of peptidisc is possible through rapid detergent removal techniques such as “in-gel”, “on-column” and “on-bead” methods.  Each of these methods has considerable advantages over overnight dialysis and “biobeads” techniques traditionally used for scaffold reconstitution. The “in-gel” method is fast, high throughput and requires only 1-2 µg of the precious target membrane protein in order to identify the optimal peptide ratio (Figure 2-2D). The “on-column” method allows direct preparation of large quantities of peptidiscs with simultaneous removal of salts, detergent and excess peptides. However, gel filtration requires concentrated protein samples and unbound peptide is wasted. The “on-bead” method is therefore ideal for the peptidisc reconstitution because protein purification, detergent removal, trapping in peptidisc and recovering of unbound peptide are carried out simultaneously in the same tube. Additionally, because lengthy exposure to detergent is avoided, this method may be especially suited for labile membrane proteins and complexes.   2.4.2 The peptidisc stabilizes membrane proteins in their functional states   Beside these methodological considerations, we show that the peptidisc maintain proteins in a functional state. For instance, both FhuA and MalFGK2 retain their ability to interact with their soluble binding partners. In contrast to the detergent, the ATPase activity of MalFGK2 in peptidisc regains dependence to substrate and maltose-binding protein MalE, indicating a return to the transporter’s native membrane conformation (Bao and Duong 2012). Negative-stain electron microscopy of MalFGK2 shows good resolution for the periplasmic P2 loop and cytosolic ABC domains, whereas the transmembrane domains MalFG are surrounded by an extra 60  density corresponding to the peptide belt and naturally present lipids.  Lipid quantitation indeed indicates that ~40 molecules, mostly acidic, are captured with MalFGK2.  In the case of the BRC complex however, there must be direct protein-peptide contact because no lipids are detected in the final assembly (Supplementary Figure 2-3A). Despite the absence of lipids, the thermal stability of the BRC in peptidisc is still much higher than in detergent, with a melting temperature similar to that observed in proteoliposomes (Supplementary Figure 2-8E).  Clearly, the peptidisc is more than a simple surrogate of the detergent molecules; the peptide assembly forms an environment that stabilize and protect the transmembrane domains of a membrane protein from aqueous solution.   2.4.3   The peptidisc fills a unique role in comparison to other membrane mimetic systems   The peptidisc offers distinct advantages when compared to other amphipathic scaffolds.  Unlike beltides, peptergents, lipopeptides, and nanostructured [beta]-sheet peptides, the NSPr does not require modifications at its N- or C-termini.  Both ends of NSPr remain available for modifications, such as biotinylation, which we show does not affect peptidisc assembly (Supplementary Figure 2-2B). Importantly, peptergents and lipopeptides have critical micelle concentrations and dynamically exchange in solution. Because of this instability, an excess of these peptides is always needed to keep membrane proteins in solution (Corin et al. 2011; McGregor et al. 2003). This is not the case for the peptidisc because NSPr binds strongly to its target, allowing free peptides to be removed without compromising peptidisc stability. Finally, at the difference of the nanodisc, the length of the NSPr does not need to be adjusted to the diameter of the target protein. This property may be especially advantageous when capturing macromolecular membrane protein complexes or proteins that exist in oligomeric state, as is the case with the SecYEG complex (Supplementary Figure 7). The NSPr is synthetically made and can be obtained in large quantity, high purity and free of immunogens usually found in recombinant cell expression systems. The structural homogeneity of NSPr is also high compared to other synthetic scaffolds, such as amphipols and styrene-maleic acids (SMA). This is because peptide synthesis is sequential whereas addition of carboxylate and other repeating units in amphipols and industrially prepared SMA polymers is randomly distributed along the polymer 61  chain during assembly (Popot 2010; Lee et al. 2016). Due to the “one-pot” synthesis, these synthetic polymers are often polydisperse mixtures of different lengths (Popot 2010; Smith et al. 2017).  This compositional heterogeneity can be problematic for functional and structural studies. Finally, due to positive and negative charged amino acids, the solubility of the NSPr remains high at various pH or with divalent cations. Both of these conditions can destabilize assemblies formed by synthetic polymers because they largely depend on charged carboxylate groups for solubility (Popot 2010). 2.4.4  On the association of the peptide in the peptidisc   How exactly the peptidisc wraps around the membrane protein template remains an important question. In the nanodisc system, the two MSPs arrange themselves in an anti-parallel “double belt” configuration (Denisov and Sligar 2016). If NSPr were also arranged in a double belt, then the peptide to protein ratio would expectedly varies by factor 2, as an odd number of scaffold would leave part of the protein exposed to the environment. However, the MS profiles for FhuA and BRC indicate each protein is reconstituted into populations of peptidisc which can vary in mass by 1 peptide only (Supplementary Figure 2-4BC), suggesting an arrangement that is flexible.  Possibly, the NSPr could be arranged in an orthogonal “picket fence” orientation as proposed for lipopeptides, nanostructured [beta]-sheet peptides, and single helix ApoA1 mimetic peptides (McGregor et al. 2003; Tao et al. 2013, Islam et al 2018). However, the length of NSPr (37 amino acids) is too long to be orthogonal while maintaining contact with hydrophobic parts of the protein or alkyl chains of annular lipids. Thus, the NSPr perhaps simply lies in a tilted orientation.  A tilted orientation would facilitate optimal binding and stoichiometry as the peptide shifts in angle of association, adapting to fit the target membrane protein template.   2.4.5   Perspective on the peptidisc method     In conclusion, the peptidisc offers several advantages. The method is cheap, fast and seamlessly integrated in existing protein purification protocols, such as size exclusion and affinity chromatography. The peptide is relatively simple to synthesize and it can be recycled via the “on-bead” method to decrease consumption. The peptidisc is not hindered by issue of buffer instability or heterogeneity as observed with other synthetic scaffold.  Since the peptide self-62  associates to its template and without added lipids, this could be advantageous for structural studies which are affected by compositional heterogeneity.  There are of course applications where other scaffold systems are better suited, such as direct protein solubilization with SMA polymers, or control over lipid environment offered by nanodiscs. Nevertheless, the current advantages of the peptidisc surely diminish the challenges associated with biochemical, structural and pharmacological characterization of purified membrane proteins, hopefully making the peptidisc a most practical way for stabilizing them without detergents.                 63  Chapter 3:  A high-throughput approach to identify protein complexes of the E. coli cell envelope in detergent-free buffer. 3.1 Introduction    Proteins control biological systems in a cell. While many perform their functions independently, the majority of proteins interact with others for proper biological activity. Characterizing protein interaction networks (the interactome) has traditionally been accomplished by targeted methods such as affinity purification (AP-MS) (Arifuzzaman et al. 2006; Hu et al. 2009; Babu et al. 2012, Babu et al 2017), PCA (Rochette et al. 2015; Tarassov et al. 2008), or Y2H (Rajagopala et al. 2014).  These methods are technically limited in their scope by poor scalability, as bait proteins must be each independently tagged, making experiments time and labour intensive.  Tags can also disrupt binding sites and create false negatives.  Protein-correlation-profiling (PCP), in combination with quantitative proteomics methods, such as label free quantitation (LFQ) or stable isotope labelling of amino acids in cell culture (SILAC), has emerged as an attractive alternative for the rapid characterization of protein complexes under native expression conditions and without genetic manipulation (Kristensen, Gsponer, and Foster 2012; Scott et al. 2017, Havugimanna et al. 2012, Wan et al. 2015).  Deep fractionation of a proteome, followed by quantitative proteomic analysis of co-fractionation profiles, allows for identification of protein complexes through a “guilt-by-association” principle (McBride et al, 2017).  This approach can generate thousands of potential interactions in a single experiment.  Furthermore, incorporation of SILAC multiplexing extends the method, allowing simultaneous comparison of multiple states of the interactome (Kristensen, Gsponer, and Foster 2012; Scott et al. 2017).     While PCP has been routinely applied to soluble proteomes, large scale PCP characterizations of the cell envelope proteomes have only recently been reported (Babu et al 2017).  This is in large part due to the hydrophobic nature of membrane proteins and their sequestration into lipid 64  membranes.  To extract cell envelope proteins, it is necessary to solubilize the lipid bilayer with the aid of non-denaturing detergents or amphipathic styrene maleic acid co-polymers (SMA). Previously, cell envelope complexes have been profiled by solubilization of an isolated cell membrane with a non-denaturing detergent, followed by separation by size exclusion chromatography, density gradient centrifugation (McBride et al. 2017) or non-denaturing blue-native gel electrophoresis (Scott et al. 2017). Chemical crosslinking to covalently link neighboring proteins can also be used to conduct PCP under denaturing conditions (Larance et al. 2016).  However, detergents are known to decrease protein stability (Yang et al. 2014), disrupt protein-protein interactions, and must be carefully removed prior to analysis by MS through precipitation or adsorbents (Yeung and Stanley 2010; Antharavally et al. 2011).  Prolonged exposure to detergents or acidic blue dyes can alter protein conformation and delipidate proteins, which can have severe implications on cell envelope protein complex assembly and stability (Bechara et al. 2015; Bao et al. 2013; Yang et al. 2014).  Thus, an ideal separation method would occur in the complete absence of detergents or ionic dyes.   We have recently described the peptidisc as a flexible system for detergent free membrane protein stabilization (Carlson et al. 2018).  The peptidisc is formed when multiple copies of a 4.5kDa (37 amino acids) amphipathic peptide scaffold self-assemble around solubilized membrane proteins. This self-assembly occurs upon removal of detergent, incorporating endogenous lipids and solubilized membrane proteins into a detergent free particle.  The number of wrapping peptide scaffolds adapts to fit the membrane protein target.  The end result of reconstitution is a membrane protein that displays increased stability, is free of detergent effects, and soluble in aqueous solution.  Our previous work has shown that the peptidisc is an effective tool for solubilizing affinity-purified membrane proteins of the inner and outer membrane of E. coli.  Because the peptidisc can adapt to membrane protein diameter and endogenous lipid content, and does not require addition of exogenous lipids, it is optimally suited for reconstituting heterogenous mixtures of membrane proteins.  This is in contrast to the preparations of crude membranes in the nanodisc, which require titration of exogenous lipids and selectively reconstitute membrane protein’s according to protein diameter and scaffold length (Carlson et al 2018).     Here, we expand use of the peptidisc to reconstitute the heterogeneous mixture of membrane proteins found in the E. coli cell envelope. The peptidisc is able to incorporate membrane 65  proteins from a solubilized crude membrane extract and the resulting protein library is soluble in detergent free buffer and amenable to size exclusion chromatography (SEC).  Stable isotope labelled amino acids in cell culture (SILAC), in combination with protein correlation profiling (PCP) and tandem MS (PCP-SILAC), is then used to identify the peptidisc library after high resolution SEC fractionation. To contrast to another detergent-free method, we compare the co-fractionation profiles of large membrane protein complexes solubilized in SMA, finding that large membrane protein complexes are better preserved in the peptidisc.   In the peptidisc solubilized cell envelope, we quantify 1209 different proteins, of which 591 are predicted to be membrane associated.  Of the identified proteins, we predict 4911 binary interactions out of > 700,000 possible interactions at 50% precision in relation to our gold standard data set, the Intact E. coli membrane complex database.  We benchmarked our interactions against the recently published E. coli cell envelope interactome (Babu et al. 2018); interacting pairs in the peptidisc interactome were enriched for shared gene ontology terms, PFAM binding domains, and correlation of growth phenotype (Babu et al. 2018; Erickson et al. 2017).  Clustering of binary interactions was resolved into 202 protein complexes using the ClusterONE algorithm.  We also identify and validate several known interactors of the bacterial translocon SecYEG, as well as novel interactors of the periplasmic scaffold protein MipA identified in our deposited interaction list.  Within the family of ABC transporters we identify an outlier substrate binding protein complex, the methionine importer MetNI-MetQ, which suggests lipid-anchors can act to stabilize complex formation.  We have made our raw and processed data accessible via Open Source Framework (Supplemental table 3-1) to serve as a useful resource for identification or secondary validation of membrane protein interactions. The peptidisc therefore provides a rapid method for rendering membrane protein complexes amenable to separation by high resolution SEC chromatography in the absence of detergent, which allows for high-throughput determination of the membrane protein interactome.  66  3.2 Materials and Methods 3.2.1 Plasmids and biological reagents   Tryptone, yeast extract, Na2HPO4, KH2PO4, NaCl, imidazole, Tris-base, acrylamide 40%, bis-acrylamide 2% and TEMED were obtained from Bioshop, Canada. Amino acid isotopologues were purchased from Cambridge Isotope Laboratories.  Isopropyl β-D-1-thiogalactopyranoside (IPTG), ampicillin, kanamycin, and arabinose were purchased from GoldBio. Detergents n-Dodecyl-β-D-Maltoside (DDM) and Octyl-β-D-glucoside (β-OG) were from Anatrace. Detergent N,N-dimethyldodecylamine N-oxide (LDAO) was from Sigma. Biosep 4000 GFC/SEC columns were purchased from Phenomenex.  Ni2+-NTA chelating Sepharose was obtained from Qiagen.  All other chemicals were obtained from Fisher Scientific Canada.  NSPr and NSPrbio was obtained from KareBay (purity >80%).  The sequence for yfgM or ppiD were inserted into pBAD33 with an 6x C-terminal His tag via the polymerase incomplete primer extension (PIPE) method (Klock and Lesley 2009) to form pBad33-yfgM and pBAD33-ppiD, respectively. To create pBad33-YfgM-PpiD, the sequence of PpiD without a 6x His tag was amplified and inserted into pBad33-YfgM using the PIPE method.  The sequence for mipA was also inserted into pBAD22 with an 6x C-terminal His tag via PIPE to form the construct pBAD22-mipA.  All construct sequences were confirmed by Sanger sequencing (Genewiz).  The sequence encoding for the gene msbA was inserted with a 6X N-terminal his-tag into the vector pET28 to form the plasmid pet28-msbA. The plasmids pBad22-HA-EYG and pBad22-his-EYG have been previously described.  (Tam et al. 2005; Maillard et al. 2007). 3.2.2 Preparation of SILAC labelled, E. coli cell envelope.   For preparation of heavy and light labelled crude membranes, E. coli strain JW2806 (ΔlysA763::kan) was labelled with light or heavy isotopologues of lysine and arginine, as previously described (Zhang et al. 2012).  Cells were picked from a single colony and grown overnight in 5 mL of LB + 25 µg/mL kanamycin at 37°C.  The overnight culture was isolated by low speed centrifugation (5,000 x g, 6 minutes), and resuspended in an equivalent volume of M9 minimal media.  The culture was pelleted and washed 2 more times to ensure full removal of residual LB media.  Unless otherwise stated, the cells were subsequently diluted 1/2000 into two flasks  containing 250 mL M9 minimal media + 0.1% glucose + 100 µg/mL thiamine. The flasks 67  were supplemented with either 0.06 mg/mL lysine and 0.035 mg/mL arginine or 0.06 mg/mL D4-lysine and 0.035 mg/mL 13C615N4-arginine to form the light and heavy labelled cultures, respectively. A control culture without supplemented amino acid was also inoculated; however no growth was observed due to the inability of JW2806 to produce the essential amino acid lysine. Cells were grown at 37°C for 16 hours until OD reached ~ 0.9-1.1.  3.2.3 Protein expression and Purification   Unless otherwise stated, all proteins were expressed in E. coli BL21(DE3) (New England Biolabs) for 3 hours at 37ºC after induction at an OD of 0.4-0.7 in LB medium supplemented with required antibiotic.  Cells were harvested by low speed centrifugation (10,000 x g, 6 min) and resuspended in SEC buffer (50 mM Tris-HCl, pH 7.2; 50 mM Na-acetate; 50 mM K-acetate).  Resuspended cells were treated with 1mM phenylmethylsulfonyl fluoride (PMSF) and lysed using a french press at 10,000 psi.  Unbroken cell debris and other aggregates were removed by an additional low speed centrifugation. Cytosolic and crude membrane fractions containing the overexpressed protein of interest were subsequently isolated by ultracentrifugation (100,000 x g, 45 minutes) and crude membrane fraction resuspended in SEC Buffer. Crude membrane (20 mg/mL) containing His-tagged MsbA was incorporated into peptidisc libraries as described below.  Peptidisc-MsbA was subsequently isolated by Ni2+-chelating chromatography in SEC buffer, washed in 10 column volumes (CV) of Wash Buffer (20 mM Tris-HCl: pH 7.1; 50 mM K-acetate; 50 mM Na-acetate; 15 mM imidazole), and then eluted in ½ CV Elution Buffer (20 mM Tris-HCl: pH 7.1; 50 mM K-acetate; 50 mM Na-acetate; 400 mM imidazole).  For purification of MsbA in DDM, the procedure was repeated except there was no addition of NSPr to the solubilized crude membrane and a concentration of 0.02% DDM was maintained in all buffers during the dilution and purification steps.  His-tagged MalFGK2 was expressed as previously described from plasmid pBAD22-malFGK2 (Bao and Duong 2012).  Crude membranes containing overexpressed MalFGK2 were subsequently solubilized in either DDM or reconstituted into peptidiscs, as described below.  pBad33, pBad22, pBad33-ppiD, pBad33-yfgM, and pBad22-mipA were chemically transformed into E. coli JW2806.  Cells were grown overnight in LB + the required antibiotic (25 µg/mL chloramphenicol for pBAD33 and 100µg/mL ampicillin for pBAD22).  Overnight cultures were pelleted by low speed centrifugation (5,000 x g, 6 minutes) and washed 3x in equivalent volumes of label free M9 68  minimal media.  Cells were inoculated 1/100 into M9 minimal media supplemented with antibiotics as specified above.  Cultures containing empty vectors were supplemented with 0.06mg/mL of D4-lysine and 0.035mg/mL of 13C615N4-arginine. Cultures containing pBad33-yfgM, pBad33-ppiD, or pBad22-mipA were supplemented with 0.06mg/mL of D4-lysine and 0.035mg/mL of 13C6 14N4-arginine.  Cells were induced at OD 0.4-0.6 by addition of 0.01% arabinose, and expression allowed to take place for 1.5 hours before cell harvest.  Crude membranes containing YfgM, PpiD, and MipA were isolated by ultracentrifugation after cell lysis.  Membranes were resuspended in Buffer A at a concentration of 5mg/mL and solubilized in 1% DDM for 1 hour at 4°C.  Solubilized membranes were clarified by ultracentrifugation (100,000 x g, 15 min) before incubation overnight with Ni-NTA affinity resin.  Beads were washed in 10 CV Buffer A + 25 mM imidazole + 0.02% DDM, then eluted in Buffer A + 250 mM imidazole + 0.02% DDM.  Eluted proteins were snap frozen in liquid nitrogen and stored at -80°C for later use. Plasmids pBad33-YfgM, pBad33-PpiD, and pBad33-YfgM PpiD were transformed into chemically competent BL21DE3 cells harbouring the plasmid pBad22-HA-EYG. For expression of his-tagged SecYEG complex only, plasmid pBad22-his-EYG was transformed into BL21DE3 cells. Overnight cultures were prepared in LB media supplemented with appropriate antibiotics at the concentrations specified above. After an overnight incubation, the cultures were diluted 1:100 into fresh LB media with antibiotics. Protein expression was induced at OD 0.4-0.6 by addition of 0.1% arabinose, and cultures were grown for a further two hours before harvesting. Cells were resuspended in TSG buffer (50 mM Tris HCl pH 8; 50 mM NaCl; 10% glycerol) before being lysed as described above. Membranes were prepared as above, except all steps were performed using TSG buffer. Membranes in TSG buffer at a concentration of 5mg/mL were solubilized in 1% DDM for 1 hour at 4°C.  Solubilized material was clarified by ultracentrifugation (100,000 x g, 15 min) before incubation for 30 minutes with Ni-NTA affinity resin.  Beads were washed in 10 CV TSG buffer + 0.02% DDM, then eluted in TSG buffer + 300 mM imidazole + 0.02% DDM.  Eluted proteins were analyzed by 15% SDS-PAGE followed by either Coomassie Blue staining or a Western blot using a SecY-specific antibody as previously described (Dalal and Duong 2009; Dalal et al. 2012).  69  3.2.4 Preparation of E. coli membrane solubilized in peptidisc libraries Cells grown to OD 0.9-1.2 were pelleted by low speed centrifugation (5,000 x g, 6 minutes), and resuspended in 2mL SEC buffer.  Cells were lysed by french press (10,000 psi, 2 passages) and cell debris removed by an additional low speed centrifugation step (10,000 x g, 10 minutes).  Crude membrane was isolated by ultracentrifugation (100,000 x g, 45 minutes), and resuspended in SEC buffer to a protein concentration of 20mg/mL. The crude membrane was solubilized in 0.8% DDM, and isolated by ultracentrifugation. Solubilized crude membrane (100µL at 10mg/mL) was mixed with NSPrbio peptide (350µL at 6mg/mL), and the mixture diluted to 10mL in SEC buffer ([DDM] ≅ 0.008%).  The mixture was concentrated over a 100kDa polysulfone filter (Sarstedt) to 500µL, then diluted again to 5mL in Sec Buffer ([DDM] ≅ 0.0008%).  The library was concentrated to 250 µL ([Total protein] ≅ 6mg/mL) and left on ice until fractionation.  For pulldown experiments, the libraries were concentrated to ≈1 mg/mL and placed on ice. 3.2.5 Preparation of E. coli cell membrane solubilized in SMA. The SMA polymer containing 2:1 styrene to maleic acid ratio was prepared following the procedure described in reference (Dörr et al. 2014).  In brief, 10% of SMA 2000 (Cray Valley), was refluxed for 3 hours at 80°C in 1M KOH, resulting in complete solubilization of the polymer.  Polymer was then precipitated by dropwise addition of 6 M HCl accompanied by stirring and pelleted by centrifugation (1,500 x g for 5 min).  The pellet was then washed 3 times with 50 mL of 25 mM HCl, followed by a third wash in ultrapure water and subsequent lyophilization.  SMA (pre and post-hydrolysis) was analyzed by Fourier Transform-Infrared Spectroscopy (FT-IR) to confirm full hydrolysis of the anhydride group.  The hydrolyzed SMA was later resuspended at 10% wt/vol in 25 mM Tris-HCl, and the pH of the solution adjusted to 8 with 1 M NaOH.  Cells were pelleted by low speed centrifugation (5,000 x g, 6 min), and resuspended in 2 mL SEC buffer.  Cells were lysed by french press (10,000 psi, 2 pass) and cell debris removed by an additional low speed centrifugation step (10,000 x g, 10 min).  Crude membrane was isolated by ultracentrifugation (100,000 x g, 45 min), and resuspended in SEC buffer to a protein concentration of 20mg/mL.  Crude membranes were solubilized by addition of 3% SMA2000. for 1 hour at 4°C, clarified by ultracentrifugation  (100,000 x g, 15 min, 4°C), then placed on ice for subsequent use.     70  3.2.6 Fractionation of cell envelope libraries in detergent free buffer    Cell envelope protein libraries were fractionated by high resolution size exclusion chromatography as previously described (Zhang et al. 2012; Scott et al. 2017b; Kristensen, Gsponer, and Foster 2012).  In brief, 200 µL of prepared libraries were separated by fractionation over two tandem BioSep4000 columns (Phenomenex) pre-equilibrated in SEC buffer at 8°C.  Fractionation occurred at a flow rate of 0.5 mL/min.  Fractions were collected from 10-22 mL elution volume.   3.2.7 In solution digestion of fractionated samples.   Without peptidisc reconstitution, detergent was removed from protein samples prior to mass spectrometry by acetone precipitation.  In brief, protein sample was mixed with 80% ice cold acetone, then left overnight on ice to precipitate.  The precipitated proteins were pelleted by low speed centrifugation (10,000 x g, 10 minutes, 4°C), washed with an equivalent volume of ice cold, 100% acetone and pelleted again (10,000 x g, 10 minutes, 4°C).  The supernatant was aspirated away and pellet air-dried at 42°C for 10 minutes before storage at -20°C until digestion.  For peptidiscs libraries, detergent was removed during peptidisc assembly, so no acetone precipitation was necessary.  We used a modified protocol to digest protein samples into tryptic peptides (Scott et al. 2017).  In brief, samples were first denatured in 6M urea.  When NSPrbio was present in the sample, the denatured proteins were incubated with streptavidin coated agarose beads (2 µg beads/µl pre-washed in SEC Buffer) for 30 minutes at 25°C and the supernatant removed to deplete the peptide.  Denatured proteins were incubated with 5 mM DTT for 1 hour at 25°C to reduce any cysteines.  Free cysteines were alkylated by addition of 20 mM iodoacetmaide for 1 hour at 25°C in the dark, the reaction was then quenched by addition of 40 mM DTT.  Samples were pre-cleaved by addition of 0.1 µg Lys-C for 1.5 hours at 25°C, followed by dilution to 1M urea in 50 mM ammonium bicarbonate, pH 8.3. Proteomics grade trypsin (1 µg; Promega) was added to each sample, and the reactions left to digest overnight at 25°C.  Digested samples were acidified to <pH 2.5 by addition of 1% trifluoroacetic acid and the resulting peptide supernatant purified using self‐made Stop‐and‐go‐extraction tips (StageTips) composed of C18 Empore material (3M) packed in to 200 μl pipette tips (Ishihama, Rappsilber, 71  and Mann 2006; Rappsilber, Ishihama, and Mann 2003; Rappsilber, Mann, and Ishihama 2007). Prior to addition of the peptide solution, StageTips were conditioned with methanol and equilibrated with 0.5% acetic acid (Buffer A3). Peptide supernatants were loaded onto columns and washed with three bed volumes of Buffer A3. Peptide samples were eluted with 80% MeCN, 0.5% acetic acid (Buffer B3) into microfuge tubes, dried down using a vacuum concentrator, and stored at -20°C.  3.2.8 LC-MS/MS analysis Purified peptides were analyzed using a quadrupole – time of flight mass spectrometer (Impact II; Bruker Daltonics) on-line coupled to an Easy nano LC 1000 HPLC (ThermoFisher Scientific) using a Captive spray nanospray ionization source (Bruker Daltonics) including a 2 cm-long, 100 μm-inner diameter fused silica fritted trap column, 75 μm-inner diameter fused silica analytical column with an integrated spray tip (6-8 μm diameter opening, pulled on a P-2000 laser puller from Sutter Instruments).  The trap column is packed with 5 μm Aqua C-18 beads (Phenomenex, www.phenomenex.com) while the analytical column is packed with 1.9 μm-diameter Reprosil-Pur C-18-AQ beads (Dr. Maisch, www.Dr-Maisch.com).  Buffer A consisted of 0.1% aqueous formic acid in water, and buffer B consisted of 0.1% formic acid in acetonitrile.  Samples were resuspended in buffer A and loaded with the same buffer.  Standard 45 min gradients were run from 0% B to 35% B over 90 min, then to 100% B over 2 min, held at 100% B for 15 min.  Before each run the trap column was conditioned with 20 μL buffer A, the analytical – with 4 μL of the same buffer and the sample loading was set at 20 μL.  When one column system was used the sample loading was set at 8μL + sample volume.  The LC thermostat temperature was set at 7°C.  The Captive Spray Tip holder was modified similarly to an already described procedure (Beck et al. 2015) – the fused silica spray capillary was removed (together with the tubing which holds it) to reduce the dead volume, and the analytical column tip was fitted in the Bruker spray tip holder using a piece of 1/16” x 0.015 PEEK tubing (IDEX), an 1/16” metal two way connector and a 16-004 Vespel ferrule.  The sample was loaded on the trap column at 850Bar and the analysis was performed at 0.25 μL/min flow rate.  The Impact II was set to acquire in a data-dependent auto-MS/MS mode with inactive focus fragmenting the 20 most abundant ions (one at the time at 18Hz rate) after each full-range scan from m/z 200 Th to m/z 2000 Th (at 5 72  Hz rate).  The isolation window for MS/MS was 2 to 3 Th depending on parent ion mass to charge ratio and the collision energy ranged from 23 to 65 eV depending on ion mass and charge (Beck et al. 2015).  Parent ions were then excluded from MS/MS for the next 0.4 min and reconsidered if their intensity increased more than 5 times.  Singly charged ions were excluded since in ESI mode peptides usually carry multiple charges.  Strict active exclusion was applied.  Mass accuracy: error of mass measurement is typically within 5 ppm and is not allowed to exceed 10 ppm.  The nano ESI source was operated at 1900 V capillary voltage, 0.20 Bar nano buster pressure, 3 L/min drying gas and 150 °C drying temperature.    Analysis of MS data was performed using MaxQuant 1.5.3.30 (Cox and Mann 2008; Cox et al. 2014; Tyanova, Mann, and Cox 2014). The search was performed against a database comprised of the protein sequences from the source organism (E. coli K12) plus common contaminants using the following parameters: peptide mass accuracy 40 parts per million; fragment mass accuracy 0.05 Da; trypsin enzyme specificity, fixed modifications - carbamidomethyl, variable modifications - methionine oxidation, deamidated N, Q and N-acetyl peptides.  Proteins were quantified from 1 peptide identification.  Only those peptides exceeding the individually calculated 99% confidence limit (as opposed to the average limit for the whole experiment) were considered as accurately identified.     3.2.9 Binary interaction identification by PrInCE and complex assignment  Protein-protein interactions (PPIs) were predicted using PrInCE (Stacey et al. 2017), a co-fractionation data analysis pipeline that assigns PPIs based on the similarity of co-fractionation profiles.  Since PrInCE employs a naive Bayes classifier, a set of known interacting and non-interacting protein pairs are required to train the classifier, i.e. a gold standard set.  We constructed a gold standard set of protein complexes by combining the 30S ribosome with membrane protein complexes given by the IntAct protein complex database (www.ebi.ac.uk/complexportal/). True positive interactions (TP) are between proteins present in the same gold standard complex, and false positive interactions (FP) are interactions between proteins present in the gold standard set but which are not members of the same complex.  PrInCE calculates an interaction score for each protein pair, which is equal to the precision (TP / 73  (TP +FP)) of all predicted interactions.  For full implementation of PrInCE see (Stacey et al. 2017).   Unlike other co-fractionation analyses, which associate protein pairs using external datasets such as gene co-citation (Larance et al. 2016), PrInCE is designed to use only information derived from the experimental dataset.  This increases the likelihood of predicting novel PPIs, since the use of external, published data biases predicted PPI towards well-studied proteins and previously-predicted interactions.  However, since there are still a considerable number of annular lipids retained in peptidiscs, the molecular weight of protein complexes can vary and broaden elution peaks (Carlson et al 2018) and thereby increase false positives. Therefore, we struck a balance between predicted interactome size and PPI novelty by including a single external dataset, the M3D database (Many Microbes Microarray Database: (Faith et al. 2008).  For each protein pair observed in our experimental dataset, we calculated the Pearson correlation between expression profiles from M3D.  Protein pairs not in the M3D database were imputed as the mean correlation value.    3.3 Results 3.3.1 Capture of the E. coli cell envelope proteome in peptidisc  The peptidisc-based SEC-PCP-SILAC workflow is presented in Figure 3-1.  To start this analysis, three different non-ionic detergents (DDM, LDAO and β-OG) and one ionic detergent (DOC) were tested using E. coli crude membranes enriched for the overxpressed, His-tagged control protein MsbA.  Upon removal of aggregate, the solubilized cell envelope proteome was incubated with the nanodisc scaffold peptide NSPr at a ratio peptide to protein ~2:1 (g/g).  The formation of the peptidisc library was initiated by detergent dilution followed by removal of micelles and unbound peptide using a spin column filter with a cut-off of 100kDa.  The overall protein content of the library was compared to the original detergent extract on SDS-PAGE (Fig. 3-2A).  The peptidisc capture most, if not all, of the proteins solubilized in detergent (Fig 3-2A).  However, the marker MsbA is preferentially solubilized in DDM, therefore this detergent was selected in the subsequent studies (Fig 3-2A).  74    75  Figure 3-1.  Overview of the peptidisc-based SEC-PCP-SILAC workflow.  A) Identical E. coli cultures are labelled in SILAC media (i), lysed with french press and crude membrane fraction isolated by ultracentrifugation (ii).  Membranes are solubilized in non-ionic detergent (DDM) (iii), transferred into NSPrbio solution to form peptidiscs, and then filtered to remove excess peptide and detergent (iv).  B) The light and heavy peptidisc libraries are separated by high-resolution SEC in detergent free buffer.  The light fraction are pooled and aliquoted into the heavy fractions as an internal quantification standard.  C) Proteins in each fraction are denatured, depleted of NSPrbio, digested, and analyzed by LC-MS/MS.  Maxquant is used to identify peptides and to quantify heavy protein enrichment in each fraction.  Binary protein interactions are identified from the co-elution data using the prediction of interactomes bioinformatics pipeline (PrinCE).  Binary interactions are subsequently segregated into predicted complexes using the ClusterONE algorithm.       The peptidisc library and the DDM-solubilized cell envelope proteome were fractionated by gel filtration (Fig 3-2B and Fig 3-2C) and the richest protein fraction as determined by UV absorbance on the size exclusion column was analyzed by MS (i.e. fraction #12; Fig. 3-2B and 3-2C).  A total of 125 proteins and 162 proteins were identified in the detergent and peptidisc fraction respectively, with ~85% overlap between the two (Fig. 3-2D).  The absence of detergent in the peptidisc fraction enabled us to omit acetone precipitation, which may be the reason why a higher number of proteins is identified.  To ensure that the cell envelope proteome is captured into distinct peptidiscs we isolated MsbA from the peptidisc library.  MsbA is a homodimeric ABC transporter of 120 kDa and therefore this protein is expected to be isolated as such. Analysis of the entire peptidisc library by non-denaturing CN-PAGE reveals a major band containing MsbA (Fig. 3-2E). This MsbA-peptidisc can be isolated from the library by Ni-NTA chromatography (Fig. 3-2F) and its molecular weight is consistent with the MsbA dimer (Fig 3-1G).  Together, this preliminary analysis provides strong evidence that the DDM-solubilized cell envelope proteome is efficiently captured in peptidiscs.  76    Figure 3-2. .  The peptidisc captures detergent solubilized membrane proteins with high efficiency.  A) SDS-PAGE analysis of detergent solubilized E. coli crude membrane before and after reconstitution into peptidiscs. The crude membrane preparation was solubilized in either 1% n-dodecyl-beta-maltoside 77  (DDM), 3% ß-octyl glucoside (ß-OG), 1% sodium deoxycholate (DOC), or 1% lauryldimethylamine-N-oxide (LDAO), followed by reconstitution into peptidiscs by dilution and buffer exchange.    B)  SDS-PAGE analysis of native E. coli membranes incorporated into peptidisc after fractionation by size exclusion chromatography in detergent free buffer.  C)As in C, with membranes solubilized in DDM and fractionated in buffer supplemented with DDM.  D) Protein number and overlap after SEC-fractionation of DDM extract and peptidiscs library prepared from DDM extract. A total of 20 fractions were collected, and the fraction containing the highest concentration of protein (fraction 12) analyzed by electrospray MS in triplicate. The MS data was searched together in Maxquant.  E-G) Purification of MsbA from the peptidisc library E)  CN-PAGE analysis of crude membrane solubilized in DDM (Lane 1) or in peptidiscs (Lane 2). F)  The peptidisc library containing overexpressed MsbA (Lane 1) was bound to Ni-NTA beads, washed in Buffer A (Lane 2), and eluted in Buffer A + 250mM imidazole (Lane 3).  Samples were analysed by SDS-PAGE.  G) CN-PAGE analysis of MsbA purified in DDM (Lane 1) or purified in peptidiscs (Lane 2).    3.3.2 Fractionation of the SILAC-labeled peptidisc library   The cell envelope proteome was labeled with light and heavy isotopes before capture in peptidisc.  The resulting libraries were separated by high-resolution size exclusion chromatography using two BioSep4000 columns placed in tandem.  The light fractions were pooled and equivalent volumes added to each heavy fraction before trypsin digestion and electrospray MS analysis.  Peptides were assigned and enrichment values were calculated in Maxquant for each fraction. The experiment was repeated twice, resulting in a combined total of 1209 proteins identified across the 54 fractions.  Of these 1209 proteins, 790 (65%) were detected in both replicates.  As expected, a large fraction of these proteins (591 proteins) are predicted based on the gene ontology term to be associated with the cell envelope (Supplementary Table 3-2).  To compare the peptidisc library with another detergent-free fractionation method, we solubilized the E. coli membrane with the styrene maleic acid co-polymer (SMA).  The SMA polymer directly solubilizes membrane lipids and captures proteins into styrene maleic acid lipid-protein nanoparticles (SMALPs) (Dörr et al. 2014).  The high resolution SEC-MS analysis of the SMALPs library allowed identification of 1576 proteins across 66 fractions.  There was good reproducibility between replicates 1210 proteins were quantified in both replicates (63% overlap) and a large part of the identified proteins (705 proteins) were predicted to be associated with the cell envelope by gene ontology term (Supplemental Table 3-2).  Importantly, the overlap between the SMALPs and peptidisc library was excellent, with 1026 proteins shared between the two libraries. Furthermore, the overall distribution of proteins in each library, as classified according to their gene ontology term and their originating compartment, was nearly 78  identical (Fig. 3-3A). Thus, SMALPs and peptidiscs are both suitable for solubilization and detergent-free fractionation of the E. coli cell membrane. In addition, the similar repertoire of proteins identified in both the SMALPs and peptidiscs libraries indicate that detergent solubilization followed by immediate reconstitution into peptidisc results in comparable capture efficiency of the cell envelope proteome as direct solubilization in the SMA polymer.             3.3.3 Large membrane protein complexes are captured in the peptidisc library   The success of the membrane profiling method depends on the capture of intact membrane complexes into peptidiscs.  To verify the ability of the peptidisc to stably isolate membrane protein complexes, we compared the co-fractionation profiles for three well characterized membrane protein complexes after encapsulating the membrane library either in SMALPs or peptidiscs (Fig. 3-3C and 3-3D, Supplementary Fig. 3-1).  In all cases, the complexes we assessed appeared more stable in the peptidisc library.  In the SMALPs library, the ATP synthase complex was dissociated, causing its protein subunits to elute separately (Fig. 3-3C).  In contrast, the ATP synthase was preserved in peptidiscs (Fig. 3-3D).  The BamABCD was solubilized with the SMA polymer, but here also the elution profiles for the individual subunits showed low correlation, suggesting partial dissociation of the complex  (Supplementary Fig. 3-1A).  In contrast, each BAM subunit presented an almost identical co-fractionation pattern in peptidisc ((Supplementary Fig. 3-1B).  Similarly, the individual subunits of the respiratory chain complex (NuoABCDGI) showed a higher degree of correlation in their fractionation profiles in peptidisc versus SMALPs (Supplementary Fig. 3-1D and 3-1C, respectively).  From these results, it is apparent that the SMA polymer does not stabilize large membrane protein complexes. The peptidisc, on the other hand, appears better suited for stabilization of large, multiprotein complexes.  79    Figure 3-3.  Proteomic analysis of soluble, SILAC labelled E. coli membrane proteins in SMALPs or peptidisc libraries.  Gene ontology analysis of identified proteins; annotated cellular compartment of identified proteins in SMALP library (A), or peptidisc library (B).  Co-fractionation profiles for quantified subunits of the ATP synthase complex in SMALPs (C) or peptidisc (D).             3.3.4 Prediction of binary interactions by PrInCE  Binary protein interactions in the peptidisc library were predicted using PrInCE, a software package designed for analyzing PCP-SILAC datasets (Predicting Interactomes via Co-Elution) (Stacey et al. 2017). PrInCE predicts which protein pairs are interacting or not according to the similarity or dissimilarity of their fractionation profiles (Fig. 3-4A and 3-4B, respectively). As described in details in the method section, a naive Bayes classifier was trained (10-fold cross-validation) using multiple pairwise similarity measures based on either the entire co-fractionation profile (Pearson correlation, Euclidean distance) and whether proteins shared an elution peak. To refine prediction, we also incorporated a single measure of pairwise similarity from the M3D expression database (Faith et al. 2008), since this allowed the classifier to distinguish between true interacting protein pairs and pairs whose fractionation profiles were only spuriously similar. 80  Training labels for the classifier were generated from the gold standard complexes, with “interacting” label applied to protein pairs in the same gold standard complex and “non-interacting” label applied to pairs present in the gold standard list but not in the same gold standard complex.  Using this approach we predicted 4911 pairwise interactions at an estimated precision (TP/[TP + FP]) of 50% (Fig. 3-4C; Supplementary Table 1E). These pairwise interactions were supported by co-fractionation data, and co-fractionation profiles of predicted interactions displayed good correlation in at least one replicate (average Pearson correlation coefficient 0.78) compared to random pairs (average correlation 0.16). The list also captures the majority of gold standard pairwise interactions (recall = 0.80, Fig. 6D).    Figure 3-4.  Interaction properties of peptidisc PCP-SILAC.  A) Typical elution profiles of interacting protein pair as classified by PrinCE.  B) Typical elution profiles of non-interacting protein pair as classified by PrinCE.  C) Precision vs. accumulated number of interactions D) Precision-recall properties of peptidisc PCP-SILAC.    81  3.3.5 Computational validation of binary interactions We used the set of gold standard complexes to estimate the error rate (precision) of our peptidisc interactome (Fig. 3-4C and 3-4D). However, the set of gold standard interactions in E. coli is relatively small compared to the set of gold standard interactions derived from mammalian studies (Ruepp et al. 2008, Rajagopala et al. 2014).  Therefore, to further estimate the accuracy of our interactome we determined whether predicted PPIs were enriched over predicted non-PPIs using three measures of biological association: i) whether protein pairs share at least one GO term; ii) whether protein pairs share a domain interface, as listed in  the 3D interacting domains database (3did; (Mosca et al. 2014); and iii) whether protein pairs are positively correlated to growth phenotypes in stress conditions, as listed in the Tolerome database (Erickson et al. 2017). These enrichment values allowed us to benchmark our interactome against the E. coli cell envelope (CE) interactome recently published by Babu and collaborators (Babu et al. 2018).  As shown Fig. 3-5A, our predicted interactome was richer for all three measures. We also correlated the predicted PPIs, random protein pairs, and E. coli CE interactome using the expression profiles listed in the M3D database (Farah et al. 2008). This M3D expression database was used to train the PrInCE classifier, therefore our interactome is greatly enriched for high correlation over random (Fig 3-5B). Finally, we confirmed that our predicted peptidisc interactome overlapped significantly with the CE interactome (Babu et al. 2018). Of the 4911 interactions predicted here, 424 are present in both interactomes, a highly significant overlap (p<0.001, permutation test) (Fig. 3-5C). 82   Figure 3-5  Comparison between the peptidisc interactome and the Babu et al. 2018 cell envelope interactome. A)  Enrichment (≥1) of shared gene ontology term (GO), interacting PFAM domains, and Pearson correlation for growth phenotype in response to stress conditions (Tolerome), for interacting vs non-interacting protein pairs in this study (black bars) and as reported in Babu et al 2017 (grey bars).  B) M3D co-expression correlation values for peptidisc interactome (black), Babu 2017 interactome (grey) and random pairs from peptidisc (dashed black) or Babu et al. 2018 interactomes (dashed grey). C)  Number of overlapping interactions between the peptidisc and Babu et al. interactomes (black arrow) and scrambled peptidisc interactome (100 iterations, grey bars).   3.3.6 Assignment and validation of protein complexes   We used a two-stage algorithm to cluster the identified pairwise interactions into complexes (Drew et al. 2017; Wan et al. 2015).  A first stage clustering was performed using ClusterONE (Nepusz, Yu, and Paccanaro 2012), an algorithm that allows moonlighting proteins 83  present in multiple protein complexes (Nepusz, Yu, and Paccanaro 2012). However, because ClusterONE tends to collapse biologically distinct protein groups into the same protein complex, we performed a second stage refinement using the MCL algorithm (Enright, Van Dongen, and Ouzounis 2002).  The combination of these two algorithms ensured that the same protein can be assigned to multiple complexes. In addition, both ClusterONE and MCL have tunable parameters, we therefore performed a grid search optimization to find the parameter set which maximizes the matching ratio value between predicted complexes and our set of gold standard complexes. This procedure produced 202 complexes with a median size of 5 proteins (Supplementary Table 1G).  As for the pairwise interactions above, we employed GO terms as evidence for biologically meaningful complex, and we reported that 36 of the 202 complexes were significantly enriched for at least one GO term (hypergeometric test, Benjamini-Hochberg correction), a highly significant number (p<0.001, permutation test). Because clustering method removes pairwise interactions that are inconsistent with the predicted complexes, the subset of pairwise interactions clustered into complexes should be scoring higher than not clustered pairwise interactions (Drew et al. 2017). This was indeed the case: the 3490 pairwise interactions clustered in complexes had a significantly higher interaction score than the 1421 unclustered interactions (mean interaction score 0.64 vs 0.59, p=3e-72, Wilcoxon rank-sum test).    3.3.7  Validation of binary interactions by AP-MS   The inner membrane anchored proteins YfgM and PpiD were picked as validation targets.  These two proteins were previously shown to form complex by BN-PAGE and size exclusion chromatography (Maddalo et al. 2011) and predicted here to interact with each other in our PrInCE list and also with the Sec translocation machinery, which we have previously characterized in some detail (Dalal and Duong, 2012).  YfgM and PpiD were separately overexpressed in SILAC labeling conditions and recovered with the crude membrane fraction.  After  membrane solubilization with DDM, each protein was affinity purified by Ni-NTA chromatography (Supplementary Figure 3-2A and 3-2B) and the co-purified interactors were identified by quantitative proteomics.  As a control, affinity purification was performed using SILAC labelled membranes prepared from cells transformed with an empty vector.  The overall goal of this AP-MS assay was also to compare the 84  list of interactions detected by the PCP method, which examines natively expressed proteins, to the list of pulldown interactors, which utilize overexpressed and tagged bait proteins.    As expected, YfgM and PpiD were enriched >1000 fold in their corresponding AP-MS datalist (Supplementary Table 3-1G).  However, only a fraction of the predicted interactors identified by PrInCE were significantly enriched in the AP-MS datalist (Fig. 3-6A and 3-6B); of the 24 predicted interactors for YfgM identified by PrInCE at or above 80% precision, 10 were identified in the AP-MS experiments, but only PpiD was significantly enriched (>99.6% percentile, Fig. 3-6D).  A similar observation was made with PpiD; of the 10 predicted PpiD interactors identified by PrInCE at or above 75% precision, 6 were identified by AP-MS but only YfgM and SecF was significantly enriched (>95.4% percentile, Fig 3-6D). The protein SecF is component of the bacterial holotranslocon, here detected by PrInCE to interact with PpiD at the precision threshold of 65%.  In addition to SecF, PrInCE also detected the holotranslocon components SecD, SecY and SecG as potential interactors of YfgM and PpiD at or above 66% precision (Supplementary Table 3-1E).  However, neither SecD, SecY nor SecG were significantly enriched in the YfgM and PpiD AP-MS datalists (Supplementary Table 3-1G).  We note that AP-MS experiments were performed under overproduction conditions, therefore a stoichiometric imbalance and/or an improper assembly of subunits can explain this higher number of false negatives interactions, especially when complexes are unstable or just transient. In support to this hypothesis, we find that the simultaneous over-production of the SecYEG core translocon with both PpiD and YfgM is necessary for the detection of an interaction (Supplementary Figure 3-3). Taken together, this comparative analysis indicate that the peptidisc PCP-SILAC method has potential to reveal interactors that would otherwise remains undetected in the classical, detergent solubilized, “single protein bait” AP-MS workflow.      The protein MipA was chosen as an additional validation target.  It has been proposed that MipA (for MltA interacting protein) is a general protein binding scaffold, bridging the outer membrane lipoprotein transglycosylase MltA and the inner membrane anchored transpeptidase MrcB(Vollmer, von Rechenberg, and Höltje 1999).  MipA was overexpressed in SILAC labeling conditions and recovered with the crude membrane fraction, although some MipA was  present in insoluble aggregates (data not shown).  Following membrane solubilization, MipA-His was isolated by Ni-NTA affinity chromatography (Supplementary Figure 3-2B).  Of the 13 predicted MipA interactors identified by PrInCE at 75% precision, 6 could be detected in the AP-MS datalist 85  (Fig 3-6C, red squares) with 2 of them, YajC and AtpF, being significantly enriched (>95.4% percentile, Fig. 3-6D).  There was also an enrichment of MrcB and MltA in the datalist (Fig. 8C), in agreement with previous AP-MS reports ((Vollmer, von Rechenberg, and Höltje 1999).  However, these two last proteins were not quantified in the PCP-SILAC experiments, perhaps due to low abundance in the cell membrane. Importantly, the interactions of MipA with YajC and AtpF, here validated by both PCP-SILAC and AP-MS, has not been reported before.      86    Figure 3-6.  Validation of predicted interactors by AP-MS.   A)  Enrichment matrix of heavy and light peptides for each protein identified in the PpiD AP-MS pulldown.  Each black dot is a protein quantified in the pulldown experiments, red squares are interacting partners identified by PrInCE.  The size of the red square indicates precision threshold of the interaction list where the protein pair is identified.  B) As in C, but for proteins identified in the YfgM AP-MSpulldown.  C) As in C, but for proteins identified in the MipA AP-MS pulldown.  Blue squares indicate complexes previously identified in reference (Vollmer, von Rechenberg, and Höltje 1999). Proteins identified in the PCP-SILAC but not in the AP-MS experiments are not represented.   E)  Enrichment analysis of quantified proteins in AP-MS experiments of PpiD (top histogram), YfgM (middle histogram), and MipA (bottom histogram).  To calculate significant enrichment, normal distributions were calculated and the top 2 standard deviations considered significantly enriched (95.4%).  Select interaction partners also identified by PrInCE are indicated with black arrows.      87  3.3.9  Identification of a unique Type I ABC transporter complex      ATP Binding Cassette transporters are a broad group of membrane proteins that utilize energy derived from ATP hydrolysis to catalyze transport of their substrate across the membrane.  In our data, we detect both ABC importers and exporters co-eluting as soluble complexes after reconstitution into peptidiscs (Fig 3-7).  ABC importers in bacteria are segregated into Type I and Type II importers (Davidson et al. 2008).  In our data, the majority of detected ABC importers are of the Type I variety.  In Type I importers, a substrate binding protein (SBP) is required to catalyze uptake of the substrate.  The SBP is an independant, soluble protein that resides in the periplasm that dynamically associates with its cognate type I transporter in response to binding of nucleotide.  In contrast, the substrate binding protein associated with a type II transporter is stably bound with high affinity to the transporter regardless of nucleotide addition.  This difference can be visualized in the raw enrichment chromatograms; the nucleotide binding domain FepC, of the type II ABC transporter FepCD, co-migrates with its binding protein FepB (Fig. 3-7A), wherease the SBPs of detected type I transporters are clearly migrating as separate entities from their transporters (Fig. 3-7C and 3-7D).  Interestingly, the type I ABC transporter MetNI appears to exist in a stable complex with its SBP, MetQ (Fig. 3-7B).  Importantly, MetQ contains a lipidation moeity, tethering it to the external leaflet of the inner membrane.  To our knowledge, MetQ is the only SBP with a putative lipidation signal in E. coli.  Previous attempts to stabilize a complex in vitro between ATP bound MetNI and lipid-free, recombinantly expressed MetQ require inactivating mutations in both MetNI and MetQ (Nguyen et al. 2016).  Similar mutations in MalE and MalK were also necessary to crystallize the complex of the Type I maltose importer MalFGK2 and its SBP MalE (Oldham et al. 2007).  Identification of MetQ-MetNI complex in the absence of these stabilizing mutations or presence of ATP suggests that SBP-lipidation may be an alternative strategy for stable complex formation of a type I transporter with its SBP.  These examples together show that high-throughput peptidisc PCP-SILAC has potential to reveal general but also unique insights into the cell envelope proteome.   88   Figure 3-7.  Co-fractionation profiles of ABC transporters and SBPs in peptidisc. A) Profile for Type II transporter FepC and SBP FepB.  B) Profile for Type I transporter LivFGM (black, blue and red trace, respectively) and corresponding SBPs LivJ and LivK (green and magenta trace, respectively). C) Profile for Type I transporter MetNI (black and blue trace, respectively) and SBP MetQ (magenta trace).  D) Profile for Type I transporter HisPQ (black and blue trace, respectively), and SBP HisJ (magenta trace).  E) Elution profiles for select SBPs. 3.4 Discussion    Despite important progresses in the field of proteomics, the characterization of the membrane interactome has lagged behind due to the reliance on surfactants to maintain membrane protein solubility. In this study, we show that the peptidisc is well suited for the capture of a cell 89  envelope proteome for study by high-throughput proteomics.  The peptidisc can entrap membrane proteins directly out of a crude detergent membrane extract and the resulting membrane protein library is stable and soluble enough to be characterized by SEC-PCP-SILAC and LC-MS/MS; high-throughput methods to detect membrane protein interactions.  We discuss below the advantages and limitations of the peptidisc-proteomic workflow, along our initial biological findings using the E. coli cell envelope proteome as a test bed. 3.4.1 Comparison to other detergent-free membrane solubilization methods.   The first detergent-free membrane protein libraries were developed using the nanodisc and applied with some success for small molecule inhibitors screen and detection of novel protein-protein interactions (Roy et al. 2015; Wilcox et al. 2015; Marty et al. 2013).   Although conceptually similar, the peptidisc library offers significant advantages, especially in the context of proteome wide analysis.  Because it is flexible, the peptidisc adapts to a wide variety of membrane protein sizes. Also, the peptidisc does not require addition of exogenous lipids to match the scaffold diameter, which simplifies the optimization process and eliminates potential bias in the capture process caused by added lipids (Roy et al. 2015).  The process of self-assembly in peptidisc is also very rapid as it does not require dialysis or removal of detergent by adsorbents.  The library formation simply occurs by detergent dilution, which limits exposure to surfactants to only a few minutes. This in turn decrease chances of detergent-mediated complex dissociation or aggregation.  Of course, the initial detergent membrane solubilization cannot be avoided.  In addition, as we show here using the model protein MsbA, the exact library content can vary depending on the nature of the detergent employed.  To address to what extent the DDM detergent affects the quality of the library, we compared the stability of several large membrane protein complexes when extracted from the membrane with the SMA polymer.  The SMA polymer can directly solubilize proteins from the biological membrane into stable lipid-protein nanoparticles. (Dörr et al. 2014).  Here, our MS analysis indicates a high degree of overlap between proteins captured in peptidisc and SMALPs libraries (85.5% similarity), showing that both methods effectively stabilize the cell envelope proteome. Surprisingly however, the SMALPs library were devoid of certain large membrane protein assemblies, such as the ATP synthase, Bam, and respiratory chain I complexes (Figure 5, Supplementary Figure 2).  In contrast, these membrane protein assemblies were readily detected in the peptidisc library 90  (Figure 5, Supplementary Figure 2).  This comparison highlights an important detail when performing high-throughput interactome studies in different solubiliization agents, solubility is not a guarantee of complex stability.  In contrast to the SMA polymer, we found that the peptidisc was able to support both stability and solubility of these large membrane protein complexes, and was therefore better suited for interactomics studies.       3.4.2 Quality of the interactome    The computational analysis of the peptidisc library co-fractionation data set via PrInCE generated ~4900 binary interactions at 50% precision out of >700,000 potential random interactions.  In addition to novel interactions, a significant number of the interactions (~8%) were also identified in a recent large scale E. coli cell membrane interactome study (Babu et al. 2018).  Although there were a lower number of interactions identified in this study, the binary interaction list published here is equally or better supported by various enrichment measurements than our reference E. coli interactome (Babu et al. 2018; Fig. 3-5A).  Altogether, we conclude that peptidisc combined with SEC-PCP-SILAC is a promising novel approach for generating high-throughput high-quality membrane protein interactomes. 3.4.3  Methods to increase number of predicted interactions by peptidisc PCP-SILAC      There is of course a large amount of data analyzed in high-throughput protein correlation profiling experiments, which can lead to spurious associations along with truly interacting protein pairs. Therefore, it is crucial to estimate and control for false positives, both in the prediction of binary interactions and the assignment list of proteins to complexes. This is especially important in this study because the co-elution peaks are rather broad, perhaps due to the relatively large size and variable amounts of lipids potentially present in the peptidisc assemblies.  The peak broadening reduces the resolution of the size exclusion chromatography and can lead to false positive interactions. To counter this, we incorporated information from a co-expression dataset (M3D) (Faith et al. 2008) to help increase the chance of identifying true interacting protein pairs rather than co-eluting, but not interacting, protein pairs.  Incorporating additional external datasets can limit false positives even further, but also biases predicted interactions toward protein pairs that are already supported in the literature at the expenses of 91  detecting novel interactions.  Therefore, further experimental work will be needed to increase the resolution of the fractionation, such as separation over sucrose gradient (McBride et al. 2017), ion exchange column (McBride et al. 2017; Maddalo et al. 2011), and clear native PAGE.  Each of these technique is feasible with the peptidisc and together should greatly increase the accuracy of the interaction list   3.4.4 Advantages of the peptidisc PCP-SILAC method          There are other technical advantages for use of the peptidisc PCP-SILAC based workflow.  PCP based interactome studies require an order of magnitude less instrument time when compared to conventional low-throughput techniques such as AP-MS or Y2H (Kristensen, Gsponer, and Foster 2012) making large scale interactomics studies feasible in practice.    SILAC allows for non-invasive method for labelling of the proteome so that the method profiles natively expressed membrane protein complexes, which can limit false negatives.  Furthermore, the peptidisc prevents dissociation of complexes that may occur upon extended exposure to detergents or amphipathic polymers.  These advantages may allow detection of interactions in a native membrane that are fragile and can be disturbed by epitope tags or uneven expression, such as the the YfgM-PpiD-SecYEG interaction.  3.4.5 Peptidisc PCP-SILAC can detect interactions not available in detergent AP-MS experiments   Like every proteomic-based discovery method, it is important to benchmark the predicted interaction list against other databases, experimental datasets, and validate detected interactions.  Accordingly, the predicted interactions published here are significantly enriched in multiple indicators of interaction, including gene ontology annotations, shared binding domains, and correlation of shared growth phenotypes (Fig 3-5A).  However, in contrast to mammalian systems, there are few commercially available antibodies for the purification of native E. coli membrane proteins for validation experiments.  Therefore, to investigate individual interactions detected by PrInCe, we utilized two systems we have worked with in the past: the bacterial translocon SecYEG, and comparison of complexes between ABC importers and their cognate SBPs. 92    Previous studies have shown using 2D gel electrophoresis and co-immunoprecipitation that SecYEG-YfgM-PpiD form a transient complex (Götzke et al. 2014).  Similarly, we report here that the interactions between SecY, YfgM, and PpiD are readily detected by PrInCE (Supplementary Table 1E). However, overexpression of YfgM or PpiD is insufficient to enrich the bacterial translocon SecYEG in our AP-MS experiments (Supplementary Table 1G). Similarly, we do not observe co-purification of SecYEG with either YfgM or PpiD in our affinity purification experiments.  Indeed, it was not until all the subunits were over-expressed that the SecYEG-YfgM-PpiD complex could be isolated by affinity purification (Supplementary Figure 3-3). This observation suggests that SecYEG-YfgM and SecYEG-PpiD interactions are relatively transient, and as a consequence are not detected by affinity purification.  By contrast, the SecYEG-YfgM-PpiD interaction seems to be stable enough to be isolated by affinity purification. Because the PCP-SILAC method requires no genetic manipulation or presence of epitope tags, it seems that this technique has the potential to capture native membrane protein complexes that would otherwise be lost in AP-MS experiments under artificial expression conditions, potentially increasing chances at discovery of novel protein interactions.    3.4.6 Comparison between ABC transporters in E. coli reveals a potential new mechanism for stable complex formation.       PCP-SILAC also facilitates targeted comparisons between protein complexes in a proteome.  This study identified 13 Type I substrate binding proteins (SBPs).  With the exception of one SBP, all elute separately from their cognate ABC transporter.  This separate elution is expected as stable association of SBPs with the transporters’ membrane domain typically depends on the presence of non-hydrolysable ATP analogs, which were not included in our SEC buffers (Bao and Duong 2012; Oldham and Chen 2011; Nguyen 2016).  In the case of the methionine transporter, MetNI eluted together with its SBP MetQ (Figure 3-7C), which to our knowledge is the only lipidated SBP in E. coli.   Given that it is also the only SBP we observe co-eluting with a Type I ABC transporter, it is unlikely that the reconstitution method is causing specific formation of the MetNI-MetQ interaction.   It is possible that the lipid anchor of MetQ somehow allows stable formation of a complex with MetNI in the absence of nucleotides.  The potential role of lipidation  to stabilize complex formation between MetQ and MetNI provides an 93  explanation for this outlier in the E. coli genome. Lipidation of SBP is commonly found in Gram-positive bacteria to prevent diffusion away from the cell (van der Heide and Poolman 2002).  In Gram negative bacteria, association of SBP to transporters usually involves extended loops (Daus, Grote, and Schneider 2009), and sometimes covalent attachment to the transmembrane domains (Lycklama a Nijeholt et al. 2018).  Perhaps lipidation of MetQ may also play an important role in maintaining SBP association during the transporter’s catalytic cycle.   3.4.7 SEC-PCP-SILAC of peptidisc solubilized membranes is a high-throughput approach for generating high quality membrane interactomes.                In conclusion, the peptidisc library offer a unique method for replicating the cell envelope proteome in detergent free buffer.  In this study, we outline how to best prepare these cell envelope libraries for analysis of the contained interactome by SEC-PCP-SILAC.  Although PCP-SILAC can be used to identify novel interactions, the most powerful aspect of the technique is for comparative analysis of interactomes by inclusion of a third amino acid isotopologue label.  This allows rapid profiling of the changes in the global interaction landscape under different conditions.  In combination with other conditional high-throughput experiments, such as co-expression arrays, and proper support by low-throughput interaction reference studies, such as AP-MS experiments, this information can be used to rapidly describe cell-wide changes that lead to observed phenotypes.   Stabilization of the cell envelope proteome in peptidiscs described here may therefore offer a powerful new tool to facilitate comparative membrane interactomics.  In addition, the interaction and complex lists identified by PCP-SILAC and datasets describing the solubilization profiles of both peptidiscs and SMALPs in this study will undoubtedly serve future researchers to better understand these two detergent-free systems.  We have therefore made our datasets available as a general resource to supplement current interactomes and solubilization profiles in the literature.  Datasets are available at the Open Source Framework via https://osf.io/dhnm6/?view_only=538182ab0e4e4e3d8dffffc08f93dfcf.       94  Chapter 4.1:  Identification of Chloride Channel Activity in the Maltose Transporter MalFGK2   4.1  Introduction   Most membrane transporters facilitate the movement of substrate molecules using an alternating access mechanism (Jardetzky 1966; ter Beek, Guskov, and Slotboom 2014; van der Does and Tampé 2004; Slotboom 2014; Jing Li et al. 2015). This conserved mode of transport corresponds to a cycle of conformational changes coupled to the movement of gating elements or rigid-body structures on either side of the membrane (Jing Li et al. 2015; Khare et al. 2009; Krishnamurthy, Piscitelli, and Gouaux 2009). This movement results in the alternate exposure of the substrate-binding site to the extracellular and intracellular environments (ter Beek, Guskov, and Slotboom 2014; Khare et al. 2009). Intermediate states, which normally occur only transiently during transport, lie between the inward- and outward-facing conformations (Forrest and Rudnick 2009; Jing Li et al. 2013).   The alternating access mechanism ensures that the free flow of ions and water molecules is restricted during transport (Jardetzky 1966; Gadsby 2009). Accordingly, since only one gate is open at a given time, the transporter is switching conformations without ever producing a membrane channel (Forrest and Rudnick 2009). Space-filling models derived from the crystal structures of inward- and outward-facing transporters indicate that the gates are sufficiently tight to prevent the passage of ions and water molecules (Kaback et al. 2007; Krishnamurthy, Piscitelli, and Gouaux 2009; ter Beek, Guskov, and Slotboom 2014). However, crystal structures represent static conformations, it is therefore unknown if the dynamics of the gates, especially during conformational transitions, is coordinated enough to prevent the passage of ions and water molecules.   A recent molecular dynamics simulation on various alternating access transporters reported that water passage might be possible through short-lived intermediates states (Jing Li et al. 2013). It was proposed that these water-conducting states are inherent characteristic to the alternating access mechanism. Water conductance has also been reported in the family of ion co-transporter (DeFelice and Goswami 2007; MacAulay et al. 2002; Duquette, Bissonnette, and 95  Lapointe 2001). In the ABC transporter family, with the exception of CFTR (T.-Y. Chen and Hwang 2008), there is however no such evidence. This may be due to the highly transient nature of the intermediate states, which makes detection inherently difficult (Jing Li et al. 2013).   In this study, we utilize the MalFGK2 transporter to explore the gating mechanism. The substrate translocation pathway is comprised of two membrane proteins, MalF and MalG. The nucleotide-binding domain, which controls the conformation of MalFG, is composed of the homodimeric MalK subunit. Transport also requires the periplasmic binding protein MalE to traffic maltose to the transporter and to stimulate its ATPase activity (Amy L. Davidson et al. 2008; Bao and Duong 2014). In this study, to facilitate the detection of ion conduction, we employed the mutant transporter MalF500 (Bao et al. 2013; Amy L. Davidson 2002). This mutant hydrolyzes large amounts of ATP independent from MalE and maltose. This high ATPase activity is because the mutant rests in intermediate conformations near the transition state (Bao et al. 2013; Davidson, Shuman, and Nikaido 1992; Daus, Grote, and Schneider 2009). Our results demonstrate that MalF500 forms an ion conducting channel which, when overproduced, is deleterious to the cell. In contrast, wild type MalFGK2 rests in the inward-facing conformation which is impermeable to ions.  4.2  Materials and Methods 4.2.1  Plasmids and Biological Reagents   The detergent n-dodecyl-β-D-maltoside (DDM) was purchased from Anatrace.  The lipids 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC), 1,2-dioleoyl-sn-glycero-3-phospho-1′-rac-glycerol (DOPG) and E. coli polar lipid extract were purchased from Avanti Polar Lipids. Superdex 200 10/300 GL, Resource 15Q, and Ni2+-NTA chelating Sepharose columns were obtained from GE Healthcare. All other chemicals were obtained from Sigma. Proteins MalE and MalFGK2 were purified as previously described (23), using plasmids pBAD33-MalE, pTrc-MalFGK2 and pBAD22-MalFGK2. Cysteine mutations were introduced into MalF (position A394 and V442) and MalG (position V230 and T182) using plasmid pTrc-MalFGK2 as template. Glycine mutations were introduced into MalF (V442) and MalG (V230, T228) using plasmid pBAD22-MalFGK2. Mutations were introduced through polymerase incomplete primer 96  extension (Klock and Lesley 2009) and verified by sequencing.  The plasmids encoding for the SecYEG translocation channel (pBAD22-SecEYG) with deletions in the plug domain (pBAD22-SecEY∆61-70G) were previously described (Maillard et al. 2007).   4.2.2  Cysteine cross-linking   Cross-linking reactions were performed in buffer A (50 mM Tris-HCl, pH 8.0, 100 mM NaCl, 10% glycerol, 5 mM MgCl2-, 0.02% DDM) for 6 min at 37°C using 5 μM MalFGK2 and 100 µM copper-o-phenanthroline (CuPhe3). Reactions were stopped with N-ethylmaleimide (5 mM) before analysis by 15% SDS-PAGE. The MalFGK2 complex carrying the cysteine mutations on the periplasmic gate (MalFV442C-MalGV230C) was reduced with 4 mM DTE and dialyzed in buffer A plus 40 μM DTE before cross-linking assays.   4.2.3  Incorporation of proteins in proteoliposomes   Total E. coli lipids dissolved in chloroform were dried under a stream of nitrogen. The lipids were resuspended in TSG buffer (50 mM Tris-HCl, pH 7.8, 100 mM NaCl, 10% glycerol) containing 0.5% DDM. The lipids were mixed with the purified MalFGK2 complex at protein: lipid ratio of 1:2000 in TSG buffer plus 0.1% DDM. Detergent was removed using BioBeads (one-half volume) and gentle shaking overnight at 4°C. Proteoliposomes were isolated by centrifugation (100,000 × g, 1h, 4°C) and resuspended in TSG buffer at a final concentration of 3 μM MalFGK2.  For incorporation of ergosterol into proteoliposomes, a mixture of DOPC:DOPG: ergosterol (ratio 60:20:20) was dissolved in chloroform, air dried, and resuspended in S buffer (50 mM HEPES pH 7.4, NaCl 150 mM, 0.5% DDM) before addition of MalFGK2. To incorporate nystatin, proteoliposomes were first frozen in liquid N2 after which they were thawed on ice in the presence of nystatin (75 μg/mL) and then briefly sonicated (15 sec, three times) (26).   97  4.2.4  Spheroplast membrane permeability assays  E. coli strain KM9 (unc-) was transformed with pBAD-MalFGK2 or pBAD-SecEYG.  Cells were grown in LB medium to OD600 ~0.4 before induction with 0.2 % arabinose.  Cells were harvested 60 min after induction (3000 x g, 10 min), washed with 1 volume of 5 % sucrose buffer (20 mM Tris-SO4 pH 7.8) and resuspended in 1/20 cell culture volume of 18 % sucrose buffer.  Cells were converted to spheroplasts by addition of 0.1 mg/mL lysozyme and 2 mM EDTA for 10 min on ice. Conversion to spheroplasts was considered complete when cell lysis was total and immediate upon dilution into water. To measure membrane permeability, spheroplasts were diluted 20-fold in 500 µl of buffer L (293 mM KCl, 20 mM Tris-SO4 pH 7.8) in the presence or absence of valinomycin (5 μM). Cell lysis was measured every 5 seconds at 540 nm. Tips with a large diameter were employed at all time to prevent pressure-induced lysis during pipetting.   4.2.5  Planar lipid bilayer experiments    The electrical currents were recorded on a Planar Lipid Bilayer Workstation (BLM; Warner instruments) composed of a Digidata 1440 Low-noise Data Acquisition System and a BC-535 Bilayer Clamp amplifier.  Unless otherwise stated, the recorded data were sampled at 1 kHz. The lipid bilayers were painted across a 150 μm hole aperture using a flame smoothed glass applicator stick dipped into a mixture of DOPC:DOPG (ratio 70:30 at 15 mg/mL) in hexadecane. Lipid bilayers were considered ready for protein insertion when their capacitance reached 70-90 pFa. The chambers on the cis-side and trans-side of the bilayer were adjusted to 650 mM KCl and 150 mM KCl, respectively.  The two chambers were connected to Ag/AgCl2 electrodes using an agarose salt bridge (2% agarose in 1M KCl). The electrical current across the lipid bilayer was stable for at least 10 min before addition of proteoliposomes in the cis-chamber. Fusion of the proteoliposomes to the lipid bilayer was facilitated using a stirring magnet. The bilayers containing channel activity were sometimes broken but could be repainted from the cis-chamber. All bilayer measurements were performed at room temperature. Data were recorded and analyzed using the Axoclamp pClamp software suite, version 10.2.  98  4.3  Results 4.3.1  MalF500 can readily access the transition state   In the absence of nucleotide, the conformation of the maltose transporter is inward-facing (Khare et al. 2009; Bao and Duong 2014). The transporter becomes outward-facing upon binding of ATP (Bao and Duong 2013). Its basal ATPase activity is also stimulated by MalE and maltose. In contrast, the ATPase activity of MalF500 (bearing mutations MalFN505I and MalGG338R) is very high and independent of MalE and maltose (Amy L. Davidson 2002; Bao et al. 2013; Daus, Grote, and Schneider 2009). It is proposed that mutations in MalF500 destabilize the inward-facing state, thereby diminishing the energy barrier of the transition state (Khare et al. 2009). To assess the conformation of MalF500, we performed a crosslinking analysis using the cysteine pairs MalF442C-MalG230C and MalF394C-MalG182C (Bao and Duong 2014, 2013). The cysteine pair MalF442C-MalG230C reports on the inward-facing conformation of the transporter (Fig. 4-1a).  Without ATP, wild type MalFGK2 is inward-facing and the crosslink efficiency is maximal (taken as 100%; Fig 4-1c). In the presence of ATP, MalFGK2 converts to outward-facing and the crosslink efficiency diminishes to ~43%. The same analysis shows that MalF500 does not depend on ATP because its conformation remains essentially unchanged. Also, the maximal crosslink efficiency is 4-fold lower than the wild type (i.e. ~20-25%). Thus, MalF500 is resting in a conformation different than inward-facing. We next employed the cysteine pair MalF394C-MalG182C, which reports on the outward-facing state of the transporter (Fig 4-1a). With MalFGK2, the crosslink efficiency is low because the transporter is inward-facing (Fig 4-b). In the presence of ATP, the crosslink efficiency increases up to ~84% because the transporter becomes outward-facing. In contrast, the crosslink efficiency with MalF500 is maximal and this conformational change is independent from ATP. Together, these data show that i) MalF500 rests in a conformation away from the resting state and ii) MalF500 reaches the outward-facing conformation independent from ATP. Thus, MalF500 access spontaneously the transition state and hydrolyze larger amounts of ATP in the absence of MalE and maltose. 99   Figure 4-1.  Conformation of MalF500 (a) The disulfide bond between MalFV442C and MalGV230C (stars) indicate that the transporter is inward-facing. The disulfide bond between MalFA394C and MalGT182C indicates that the transporter is outward-facing. (b) The MalFGK2 complex in proteoliposome (4.5 μM) was treated with CuPhe3 (100 μM), with or without ATP (2 mM), followed by addition of NEM (5 mM) at the indicated time. The MalF-MalG crosslink products were detected by SDS-PAGE (15%) and Coomassie blue staining. The crosslink intensity was quantified using Image J. The crosslink efficiency of MalFGK2 (V230C-V442C) without ATP is taken as reference point (100%) in the upper panel. C) Repeated as in B with the diagnostic T182C-A394C cross linking mutations.  The crosslink efficiency of MalF500 (T182C-A394C) with ATP is taken as reference point (100%) in the lower panel. Each crosslink experiment was repeated at least 3 times. This figure was reproduced from Carlson et al. 2016. 4.3.2  MalF500 is deleterious to the cell   Overproduction of MalF500 in the membrane results in a significant growth delay (Fig. 4-2a). 100  This may be caused by the high basal ATPase activity of MalF500. It is also possible that the inherent conformational flexibility of MalF500 may increase membrane permeability and compromise cell viability. In an attempt to differentiate the possibilities, we introduced the mutation E159Q into the Walker A motif of MalK (Covitz et al. 1994). This mutation almost fully abolishes ATP hydrolysis (Fig. 4-2b), yet the growth of MalF500E159Q is still impaired compared to the wild type (Fig. 4-2a). We then introduced a His6-tag at the N-terminus of MalF. The His6-tag decreases the ATPase activity of the wild-type complex through stabilization of the inward facing conformation (Bao and Duong 2014). Addition of the His6-tag on MalF500 reduces its ATPase activity to wild type levels (Fig. 4-2b), yet the mutant still displays a significant growth defect. These results suggest that the conformation of MalF500, not just its high basal ATPase activity, is detrimental to the cell.  101    Figure 4-2. Activity of MalF500 (a) Effect of MalF500 on cell growth. E. coli BL21 was transformed with pBAD22-derived plasmids containing the MalFGK2 complex under control of the arabinose promoter. Cell were grown in LB media to OD600nm ~ 1.0, serial diluted and plated on LB-agar plate in the presence of 0.02% arabinose to induce expression of the MalFGK2 complex. Plates were incubated at 37°C for 16 h. (b) ATPase activity. Proteoliposomes (0.5 µM) and maltose (1 mM) were incubated at 37°C in the presence or absence of MalE (1 µM) as indicated. Standard deviation was obtained from three separate experiments. ND: not detected.  This figure was reproduced from Carlson et al. 2016.   102  4.3.3  MalF500 is permeable to chloride    We employed a spheroplast lysis assay to determine whether MalF500 affects cell membrane permeability. This method has been developed to study the ion conductance of the SecYEG protein translocation channel (Park and Rapoport 2011). Briefly, spheroplasts were diluted into an iso-osmotic solution of KCl in the presence of the K+ ionophore, valinomycin. If the membrane is permeable to Cl-, the spheroplasts swell and eventually lyse due to the rapid influx of water. As expected, a low degree of lysis was observed in the absence of valinomycin (Fig. 4-3a). In the presence of valinomycin, spheroplasts containing the MalF500 complex lysed immediately (Fig. 4-3b, red trace). The initial lysis rate is very high, more than 7-fold higher than spheroplasts containing wild type MalFGK2 complex and near half of spheroplasts containing the SecYEG channel with an altered plug domain (SecEYΔ61-70G; purple trace). When the experiment is performed with the mutant MalF500his6, the rate of lysis is diminished by ~50% (Fig. 4-3b, orange trace). This is consistent with the observation that a His6-tag at the N-terminus of MalF only partially restores cell viability.       103   Figure 4-3. Permeability of MalF500 in spheroplast assays E. coli KM9 (unc-) overproducing the indicated complex was converted to spheroplasts, and then diluted into an iso-osmotic solution of KCl (293 mM) in the absence (a) or presence (b) of valinomycin (15 µM). Cell lysis was monitored every 5 seconds at 540nm. The rate of lysis is determined from the linear part of the curve (i.e. first 30 sec after dilution in KCl).  This figure was reproduced from Carlson et al. 2016. 4.3.4  Ion channel activity of MalF500    The MalF500 complex was purified and incorporated into planar lipid bilayers in order to determine its ion conductance properties (Fig. 4-4a). The data show that MalF500 has a channel-like activity, and distinct single channel opening and closing events can be resolved. Quantitative analysis of the recordings (Fig. 4-4b and Fig. 4-4c) further show that (i) MalF500 creates a membrane channel that is voltage insensitive given the linear voltage-current relationship; (ii) is anion selective, with a reversal potential (-38mV) that is close to the calculated Nernst potential for chloride (-37mV); and (iii) has rapid gating kinetics, with an average dwell time for the opening state of only ~2-20mS. Such quick opening and closing kinetics suggests that the 104  transporter is conformationally flexible. This is in contrast to wild type MalFGK2 which, throughout the experiment, remained stably closed (Fig. 4-4a). Although bilayers often contained more than one channel, single channel opening events were easily identified and were used to calculate the ion conductance of MalF500. The histogram of current amplitudes and conductance magnitude were plotted as a current-voltage curve; the slope representing the single channel conductance in pS (Fig. 4-4c).   Figure 4-4. Ion channel activity of MalF500 in planar lipid bilayers (a) The electrical currents were recorded across a planar lipid bilayer (70% DOPC, 30% DOPG) at a holding membrane potential of +50mV. Current traces were filtered at 500Hz. The traces for the wild type MalFGK2 were recorded after >15 fusion events using nystatin/ergosterol as a reporter system indicating protein delivery. (b) Histogram of current amplitudes. The number of channel events obtained at +50mV was determined using the Clampfit analysis program and the single channel search function. The currents were plotted as a function of their intensity. (c) Current-voltage curve for MalF500. Current amplitudes (pA) were plotted according to the applied holding voltage (mV). The slope of the curve represents the channel conductance in pS. The reversal potential is -38mV as indicated by the x-intercept of the curve.  This figure was reproduced from Carlson et al. 2016.   105  4.3.5  The periplasmic gate seals MalF500    The resting maltose transporter (i.e. inward-facing) is sealed by gating loops on the periplasmic side on the membrane (Khare et al. 2009). The amino acyl side chain MalFV442, MalGT228 and MalGV230 form this interface (Fig. 4-5a). These residues were replaced by the short chain amino acid glycine. According to the space-filling model (Fig. 4-5a), such mutations produce a wide-open pore in lieu of the gate. Surprisingly, results from the cell growth (Fig. 4-5b), ATPase activity (Fig. 5c), and cell membrane permeability assays (Fig. 4-5d) reveal that MalFGKGGG behaves much like the wild type. The glycine residues were then introduced into MalF500 to produce the mutant MalF500GGG. In this case, we observed an immediate and strong growth defect upon protein production (Fig. 4-5b) together with a dramatic increase in cell membrane permeability (Fig. 4-5d). We note that the ATPase activity of MalF500GGG is ~ 4-fold lesser than MalF500 (Fig. 4-5c), suggesting that ATP consumption is not the primary reason for the higher growth defect of MalF500GGG. Apparently, the nature of the gating residues is particularly important when the transporter rests in a conformation near to the transition state.   106   107  Figure 4-5. The periplasmic gate seals MalF500 (a) The three gating residues (red, MalFV442, MalGT228 and MalGV230) forming the periplasmic gate on MalF (yellow) and MalG (pink) were mutated to glycine residues. The space filling representation was created with Pymol using the inward-facing crystal structure of MalFGK2 (PDB accession code 3FH6). (b) Growth curve of E. coli overproducing the indicated complex. Cells were grown in LB liquid media at 37°C for 150 min before induction with 0.2% arabinose. (c) ATPase activity. Proteoliposomes containing the indicated mutant complex (0.5 µM) were incubated with maltose (1 mM) at 37°C with or without MalE (1 µM) as indicated. Standard deviation was obtained from three separate experiments. (d) Spheroplast assays. Spheroplasts were diluted into an iso-osmotic solution of KCl in the presence or absence of valinomycin (15 µM). Cell lysis was monitored every 5 seconds at 540nm.  This figure was reproduced from Carlson et al. 2016.  4.3.6  MalF500GGG forms a quasi-permanently open channel    The mutant MalF500GGG was purified and inserted into planar lipid bilayers (Fig. 4-6a). In contrast to MalF500, which is equally distributed between open and closed states, MalF500GGG behaves like an open-state channel (Fig. 4-6b). The recordings (Fig. 4-6a) show that the frequency of a gating event is very slow (every ~1-2 sec, compared ~20 msec for to MalF500). This increased open pore duration is consistent with both the dramatic increase in cell membrane permeability (Fig. 4-5d) and the very strong effect on bacterial growth (Fig. 4-5b). Single channel recordings could not be captured with the MalF500GGG mutant; however, isolated closing events could be detected in bilayers containing multiple copies of the mutant (Fig. 4-6a). The rate of these closing events was such that only one channel out of ~ 25 was observed closing at a time.  108   Figure 4-6. Ion channel activity of MalF500GGG (a) Typical electrical currents. Traces were obtained at a holding potential of -50mV. (b) All point histogram of current amplitudes. The number of channels present in the MalF500GGG bilayer (≈25 channels incorporated, -250pA) results in a higher total current amplitude than observed for MalF500 (≈3 channels incorporated, -15pA). For MalF500GGG the observed channels are predominantly open, therefore concurrent closing events are rare and the all-point histogram shows only two current distributions despite the presence of multiple channels. The number of channels incorporated is determined by dividing the maximum observed current by the amplitude of one opening event as observed in the traces represented in (a).  This figure was reproduced from Carlson et al. 2016. 109  4.4 Discussion 4.4.1  Insights into the expanded alternating access model from chloride leakage in MalF500   The two major conformational states of the maltose transporter, inward-facing and outward-facing, appear sealed against ions and water molecules (Khare et al. 2009; Jing Li et al. 2013; Oldham et al. 2007). Yet, whether the impermeability is maintained when the transporter cycles between these two states is unknown. These intermediate conformations are structurally difficult to characterize and, in the case of MalF500, also difficult to crystallize (Amy L. Davidson 2002). Recent computer simulations suggest that water molecules can permeate through the substrate passageway when the transporter pass through intermediate conformations; however the pore is very narrow and it is open for only a few nanoseconds (Jing Li et al. 2013).     Here, we have augmented access to the intermediate conformations using the mutant MalF500, which carries the mutations MalFG338R and MalFN505I. These mutations destabilize the inward-facing conformation, thereby decreasing the energy barrier for the transition (Khare et al. 2009; Bao et al. 2013; Daus, Grote, and Schneider 2009). As a result, the MalF500 mutant is capable of spontaneously adopting the outward-facing conformation independently of ATP (Fig. 4-). Through the use of this mutant, we show that increased access to the intermediate conformations allows for a significant degree of ion conduction (Fig. 4-4). The currents measured are high (>107 ions/sec), indicative of a movement of ions through a channel-like structure (DeFelice and Goswami 2007; Jayaram et al. 2008). In addition, the frequency of channel closing and opening is very fast: every few milliseconds. This rapid cycling is consistent with the decreased energy barrier between the inward- and outward-facing conformations, and with the exceptional ability of this mutant to hydrolyze a large amount of ATP. The results also show mutations in MalF500 that stabilize its conformation, and therefore diminish its basal ATP activity, also diminish ion conduction (Fig. 4-3). Interestingly, the conductance and spheroplast assays indicate that MalF500 is selective for Cl- over K+, although there is no obvious structural characteristic in MalFGK2 to explain this selectivity. Notwithstanding, our data show that MalFGK2 must rest away from the conformations adopted by MalF500 in order to preserve the 110  membrane barrier. This membrane barrier is particularly important in bacteria as ion gradients represent a main energy source (Maloney, Kashket, and Wilson 1974). It is therefore not surprising that overproduction of MalF500, and especially MalF500GGG, results in a significant bacterial growth defect (Fig. 4-5b). This growth defect may reflect the energetic cost associated with the use of counter-acting pumps required to compensate for the leakage of chloride ions.   4.4.2  Role of the periplasmic gate of MalFGK2 in prevention of chloride conductance    The periplasmic gate on MalFGK2 is formed at the interface of four helices (Khare et al. 2009). Surprisingly, we find that alteration of this interface (i.e. introduction of the mutation MalFV442G, MalGT228G, and MalGV230G) does not alter the function of the gate: the mutant is viable and we do not detect increased ion conductance (Fig. 4-5d). This result was rather unexpected since a space-filling computer model suggests that the glycine residues leave an open pore over the transport pathway (Fig. 4-5a). It is therefore possible that an energetic pressure is forcing the mutated gate structure to adopt a conformation that re-creates this essential membrane seal. This phenomenon has been reported in the case of the SecYEG translocon after the deletion of the plug domain (W. Li et al. 2007). In contrast, introduction of the glycine residues into MalF500 leads to a quasi-permanently open pore (Fig. 4-6). The residues forming the periplasmic gate are therefore critical when the transporter rests in a conformation close to the transition state.   4.4.3  Insights into the evolution of the ABC transporter turned chloride channel CFTR.  Active transporters and ion channels function by different mechanisms (W. Li et al. 2007; Gadsby 2009), yet our results show that the maltose transporter can acquire a channel-like activity after just a few pertinently located mutations. In the case of CFTR, it is proposed that the stabilization of an intermediate conformation is at the origin of the conversion of this transporter into a chloride channel (Jordan et al. 2008). Specifically, comparisons made with the closely related ABCA4 transporter suggest that the mutation of a salt bridge positioned far away from 111  the gate region might have been crucial for conversion of the ancestral transporter into a channel-like state. After this initial conformational shift, the modern pore and gate regions in CFTR were acquired through additional mutations (Jordan et al. 2008). Consistent with this hypothesis, our results reveal that a simple disruption to the gate region, as in the case of MalFGKGGG, does not suffice to create an ion channel. It is rather the mutations that promote access of the transporter to intermediate conformations, like in MalF500, that are responsible for the ion channel activity.      Chapter 5:  General Conclusions and Future Work    The study of membrane proteins is of critical importance in the laboratory and clinic, as these proteins are the primary contact point for how a cell senses and interacts with its molecular environment.  There are still critical roadblocks that reduce the effectiveness of biochemical techniques for the analysis of membrane proteins, which are largely due to researcher’s reliance on non-native, denaturing detergent environments to analyze these proteins.  The findings presented in this thesis largely focus on developing new tools for analysis of membrane proteins in the absence of detergents.  In addition to presentation of new methods, this thesis also investigates the fundamental link between two of the most well recognized classes of membrane protein, transporters and channels.  The thesis can be broken down into three distinct sections:  First, to develop and validate the peptidisc as a universal reconstitution scaffold for membrane proteins in detergent free buffer; second, to investigate the cell envelope interactome of E. coli by SEC-PCP-SILAC in peptidiscs; and third, to better understand how a model ABC transporter may acquire ion leakage, framed within the context of the alternating access mechanism.             In Chapter 2, we design and validate the peptidisc method.  We demonstrate that the peptidisc can capture membrane proteins from both the inner and outer membrane (Fig. 2-2), upon addition of peptide in excess of a universal reconstitution ratio (RR50).  We find that the peptidisc 112  is likely localized around the transmembrane regions of a reconstituted membrane protein, as in all cases the reconstituted membrane proteins were able to interact with their soluble binding partners and ligands (Fig. 2-1, Supplementary Figure 2-6).  We also determine the peptide stoichiometry in reconstitutions of three different membrane proteins, BRC, MalFGK2, and FhuA (Supplementary Figure 2-2 and Supplementary Figure 2-3).  Interestingly, despite significant variations in each protein’s structure, number of bound lipids, and diameter, there was a similar number of peptides bound to each protein (Supplementary Table 2-1).  The calculated and observed molecular weights of each peptidisc are in good agreement (Supplementary Table 2-1), suggesting little to no detergent remains in the disc.  However, the similar number of peptides in each disc are difficult to reconcile with a double-belt model, as a reduction in number of peptides would be expected when reconstituting the small, delipidated BRC as compared to the larger, lipidated MalFGK2.  It is therefore possible that the multiple peptides can arrange in more than one layer around a membrane protein (like an onion) or are not in a double-belt conformation at all.     When linked into a long protein scaffold, the amphipathic helices of ApoA1 arrange in a “double-belt” model around a lipid bilayer.  However, recent molecular simulations of single-helix, ApoA1 mimetic peptides in complex with lipids suggest a “picket-fence” conformation, whereupon each peptide lies perpendicular to the plane of the lipid bilayer (Islam et al. 2018).  The NSP is a bi-helical peptide, therefore it is likely too large to lie in a “picket-fence” orientation.  However, we show by intact MS that there are clearly populations of peptidiscs that contain an odd number of peptides (Supplementary Figure 2-4B and 2-4C), which would seemingly exclude a double-belt model.  We propose here that the peptide may adopt a structure that may lie between these two examples, adopting a tilted orientation.  However, without a high-resolution structure a peptidisc, this hypothesis remains speculative.  High-resolution Cryo-EM has been utilized to determine the stoichimoemtry and arrangement of saposins in a stabilized saposin-lipid-transporter complex and may also prove fruitful for determining the arrangement of peptides in the peptidisc (Frauenfeld et al. 2016).  More recently, NMR studies were able to determine the conformations of MSPs in a nanodisc (Bibow et al. 2017).  Similar approaches may be taken in the future with the peptidisc, as understanding how the peptide wraps around its membrane protein template can be critical for optimizing future iterations of the method.  113      One interesting potential application for the peptidisc is towards detergent free crystallography.  To date, most membrane protein structures have been purified in detergents and crystallized via the hanging drop method.  Detergents often remain loosely bound to proteins in the crystal lattice, which can result in stabilization of unexpected, possibly non-native conformations of membrane proteins (Ward et al. 2007).  Synthetic polymer scaffolds have been utilized to economically solubilize and deliver detergent-free membrane proteins into lipid cubic mesophase, resulting in high resolution crystal structures (Polovinkin et al. 2014; Broecker, Eger, and Ernst 2017).  However, the synthetic scaffold has not been visualized in any of these structures, indicating that the protein is more likely to have exchanged scaffold for lipid as it enters into the mesophase.  Therefore, there has been little success crystallizing membrane proteins in membrane mimetics outside of lipid mesophase enabled approaches.  In contrast with synthetic polymer scaffolds, the peptide scaffolds are structurally homogeneous and may therefore be better suited for structural studies by detergent-free hanging drop methods, where homogeneity and purity are significant predictors for chances of determining a high resolution crystal structure (McPherson 2004).             Peptidisc formation occurs by addition of excess peptide, which self-assembles around a minimal template (Fig. 2-2).  Peptidisc assembly is thought to be driven by the hydrophobic face of the peptide binding the TMs and exposed lipids of a target membrane protein upon removal of detergent.  Because the peptide can also form inter-peptide interactions, there is an equilibrium of peptide bound to other peptides, and peptide assembled around a membrane protein of interest in a reconstitution reaction, which manifests as two peaks in the on-column reconstitutions of MalFGK2 and FhuA (Fig 2-1A and Supplementary Figure 2-3A).  The reconstitution process can therefore be represented by schematic 5.1:   5.1                 Detergent + NSP + MP →  (NSP-NSP)X + (NSP-MP)Y                                    -Detergent    From our results, it is apparent that after removal of detergent, the NSP either binds other peptides in a peptide micelle, or associates to a membrane protein (MP) in a peptidisc.  We hypothesize that the amount of peptide micelle and formed peptidisc are proportional to each other and constant, represented by X and Y in schematic 5.1, respectively.   In this scenario, 114  when NSP is limiting in the reconstitution, a lower number of peptidiscs will be formed to maintain the micelle to peptidisc ratio on the right side of the reaction, resulting in larger peptidiscs and aggregation of the target membrane protein.  In contrast, if peptide is added in excess, the MP becomes the limiting, and is reconstituted with 100% efficiency.  This view of the reconstitution process, albeit simple, best explains why we found that the RR50 for peptidisc formation to be a universal value, irregardless of protein diameter or lipid content (Fig. 2-2).  Understanding peptidisc formation can provide a roadmap for optimizing the peptide sequence.  Kariyazono et al, who first described the NSP peptide, highlighted the asymmetry present in the NSP peptide can decrease inter-peptide interactions (Kariyazono et al, 2016).  In our proposed reconstitution process, increasing asymmetry of the peptide sequence may therefore lead to higher reconstitution efficiency and a decrease in the RR50.         In contrast to the nanodisc, the peptidisc method only contains endogenous lipids which survive the detergent purification step. This is an advantage from a sample-handling perspectie, as there is no optimization of lipid content or scaffold length to make homogenous peptidiscs.  A disadvantage of the method is that the lipid environment of a peptidisc cannot be controlled by the method presented here.  Fundamentally, there is little drive for exogenous lipid to be incorporated into the peptidisc if it is not already associated to the protein in the detergent micelle.  This is because peptidisc formation is flexible, and will proceed to reconstitute the smallest possible particle with the amount of peptide present.  As a result, detergent solubilized lipid will preferentially incorporate into small, lipid-peptide particles rather than be incorporated into the larger peptidisc.   5.2 Detergent + NSPN +MPM + LipidK  →   (NSP-NSP)X + [(NSP-Lipid)Z + (NSP+MP)V]y        -Detergent                                 Where X : (Z+V) in Eq. 5.2 is ≈ X:Y in Eq. 5.1     Indeed, we found that addition of detergent solubilized lipid, even in small quantities, during peptidisc reconstitution results in aggregation of the reconstituted protein and laddering on native page gels (Carlson and Duong, unpublished results).  The laddering observed after addition of lipids appears identical to the laddering observed when reconstituting membrane proteins below their RR50 (Figure 2-2).  It is likely that addition of detergent solubilized lipid acts to deplete 115  available NSP by forming small lipid-peptide particles; as the peptide is depleted, it drops below the RR50, and protein incorporation into larger peptidiscs occurs.       Interestingly, the beltides-nanodisc reconstitution method, which utilizes an analagous scaffold to the NSP found in the peptidisc, inserts membrane protein into preformed lipid-peptide nanodiscs (Larsen et al 2016).  However, the formed particles experience instability upon raised temperatures, suggesting that there is still a fundamental issue with the resulting protein-beltide-lipid nanodiscs.   Perhaps, in contrast to the lipids which are stably bound to the membrane protein template in a peptidisc (ie annular lipids), the excess lipids contained in a beltide nanodisc are unstable and can be shed as small lipid-peptide particles, leading to the aggregation observed by Larsen et al.  For future experiments to incorporate exogenous lipids into the peptidisc, it may be necessary to deliver exogenous lipid during the initial solubilization step.  This would ensure that the exogenous lipids are incorporated into the solubilized protein before peptidisc assembly, thereby bypassing lipid sequesteration into peptide-lipid only particles.    There is more work to be done to understand the formation of the peptidisc.  However, the lack of clarity on the finer details of assembly do not preclude its use for the structural and functional study of membrane proteins.  In Chapter 2, we achieve multiple examples of functional, monodisperse reconstitutions of membrane proteins into the peptidisc, demonstrating the simplicity and robust flexibility of the method.  In Chapter 3, we utilize this method to reconstitute the cell envelope proteome into peptidiscs.  For reconstitution of the cell envelope proteome, we briefly solubilize the membrane in detergent, followed by rapid dilution into an excess of peptide.  The excess peptide micelles and free detergent are removed by the buffer exchange step over a 100kDa spin filter.  Peptidisc libraries can also be formed by direct on-column reconstitution, but the high concentrations of detergent used to solubilize a membrane are not fully removed by this method.  This is important in the context of PCP-SILAC, as the overabundance of free peptide and detergent micelles in an on-column reconstitution elute in later fractions and can interfere with identification rates in nanospray LC-MS/MS (Carlson and Duong, unpublished results).    Reconstitution of the E. coli crude membrane into peptidiscs, and subsequent identification of the complexes contained within, provides a high-throughput approach for analyzing membrane protein interactions in a stable, detergent-free environment.  As the method relies on initial 116  detergent solubilization to release membrane proteins from the lipid bilayer, a gentle, non-ionic detergent is recommended.  However, we show that the method is applicable to membranes solubilized in a variety of different detergents (Supplementary Figure 3-2A).  This is important, as the method can be utilized to target various membrane environments by selective solubilization in different detergents.  Although we show that the peptidisc is able to specifically capture an overexpressed membrane protein complex, the ABC transporter MsbA (Fig. 3-2C-E), it is also possible that during assembly of the peptidisc that transient interactions are also captured.  Capture of transient SecEYG oligomers and dimers in the peptidisc was apparent during “in-gel” peptidisc reconstitutions (Fig. 2-2D).  These higher order oligomers of SecYEG had previously been demonstrated only with very low, radiolabelled quantities of SecYEG and were dependant on specific concentrations of detergent (Bessonneau et al. 2002).  However, in the peptidisc, these dynamic complexes are robustly captured (Fig. 2-2D).  From this perspective, reconstitution of the E. coli crude membrane into peptidiscs may stabilize interactions that are otherwise unstable in a detergent environment or when expressed under non-native conditions.  We also find that the peptidisc is a comparatively more gentle method for separation of large membrane protein complexes compared to the SMA polymer (Fig. 3-3), another well known membrane mimetic.       The ability to isolate native interactions by peptidisc-SEC-PCP-SILAC that would otherwise go undetected by other methods was well demonstrated in the case of the YfgM-PpiD-SecYEG complex (Supplementary Figure 3-3).  YfgM-PpiD is thought to form an ancillary component of the SecEYG translocon (Gotzke et al. 2014).  In a stable peptidisc environment, the interactions between YfgM, Ppid, SecY, and SecG, were detected in the PCP-SILAC experiments.  However, unless all components of the complex were overexpressed, co-affinity purification of SecEYG with either YfgM or PpiD did not occur. This is because peptidisc-SEC-PCP-SILAC measures interactions between native membrane proteins, and is not subject to the high false-negative rate associated with AP-MS experiments.   Indeed, computational validation of the interactors detected in the SEC-PCP-SILAC experiments show that the interaction list detected by this method is comparable to other high-throughput, AP-MS interaction lists which require 10 fold more mass-spectrometry experiments (Fig. 3-5).  Our interaction list, and the interaction list generated by AP-MS methods, display a small, albeit statistically significant, amount of overlap (approximately 424 proteins) (Fig 3-117  5C).  This low amount of overlap is likely due to the inclusion of contaminating soluble proteins in the cell envelope preparation, as the reference study is focused entirely on targeted membrane proteins.  In addition, our list also contains a significant number of potential novel interactions.  Many of these novel interactions will be false-positives, as the list is generated at 50% precision score, however, it is also likely that a large number of these interactions are true-postive novel interactions that are not easily detected by other methods.  For example, here we were able to identify by PCP-SILAC and validate by AP-MS experiments two new possible interactors to the membrane bound protein MipA, YajC and AtpF.  Interestingly, although unreported in the high confidence data, MipA was also identified as interacting with another member of the ATP synthase (AtpD) in our comparative reference AP-MS study (M. Babu, correspondence).  Thus, our interaction list can function as both a potential method for capturing high confidence E. coli membrane protein interactions, or as a potential high-throughput validation tool when identifying new interactions by parallel techniques.  Although not performed here, SILAC also enables multiplexing of samples for comparison of interactomes between conditions (Scott et al. 2017; Kristensen, Gsponer, and Foster 2012).  If this were to be attempted by low-throughput AP-MS experiments, the number of MS samples would be impractical (between 5,000-10,000 MS samples per condition).  However, using PCP-SILAC the entire study can occur in a single experiment, in this case, 110 MS samples.  The potential for this kind of high-throughput analysis is massive.  Already, studies have been performed to analyze how protein interactomes change during the apoptosis cascade, or in response to growth factor stimulation (Kristensen, Gsponer, and Foster 2012; Scott et al. 2017).     We utilized a two-step clustering method to break our binary interactions into distinct complexes.  This allowed for separation of the binary interactions, into 202 complexes (Supplementary Table 3-1F).  As expected, the predicted complexes were enriched for shared gene ontology terms, and displayed higher overall interaction scores than proteins not included in complexes.  Utilizing a 50% precision cut-off for the binary interaction landscape can be a double-edged sword for the purposes of complex assignment, as complexes can contain large numbers of proteins that can be difficult to validate in vitro.  However, the trade-off of a lower precision list is that more novel interactions are captured, thus there is a decrease of false-negatives.  Our predicted complex list can be useful for comparative interactomics, where gain or 118  loss of subunits or complexes can be quickly measured, or as a reference for future, more targeted studies.    In our binary interaction list, we identify a possible outlier in the ABC transporter family, MetNI-MetQ, which appears to from a stable SBP-Type I ABC transporter complex in the absence of nucleotide (Fig. 3-7C).  Previous in vitro studies examining the interactions between MetQ and MetNI have been observed exclusively with soluble MetQ, which only forms a stable complex with MetNI with the addition of incactivating mutations.  In contrast, our results indicate that MetQ is tethered to the membrane and forms a complex with native MetNI.  MetQ is the only known SBP in E. coli that contains a putative lipidation site, and we have confirmed that this protein largely co-fractionates with the membrane (Carlson, Wason and Duong, unpublished results).  This is an interesting observation, as SBP lipidation is a known mechanism to prevent SBP diffusion away from the cell in Gram positive bacteria, which do not have an outer membrane.  However, in E. coli there is an outer membrane and SBP lipidation is therefore not necessary.   In Chapter 3, we propose that the lipidation of MetQ occurs, causing it to form a stable complex with MetNI.  However, how this effects MetNI function, and whether these proteins can be isolated as a complex in preparative quantities, are still to be determined and will be the focus of future studies.     In Chapter 4, we utilize the SBP-independant mutant MalF500 to increase access to the transition state of an alternating access transporter.  We demonstrate through site-specific cross-linking that the the MalF500 mutations increase the transporter’s ability to cycle between the inward and outward conformations (Fig. 4-1).  Accordingly, we find that the mutant also displays heightened basal ATPase activity, and is not stimulated by addition of MalE (Fig 4-2B).  Upon overexpression, the MalF500 mutant clearly has a negative effect on cell growth (Fig. 4-2A).  We show that this growth defect is not due to the high basal ATPase activity of MalF500 as mutations that aboilsh ATP hydrolysis do not abolish this growth defect (Fig. 4-2).  Rather, we find that significant chloride leakage is occuring due to increased access of the transporter to its transition state, which is an alternative explanation for the observed growth defect.  Importantly, we show that this chloride leakage is a bonafide channel activity occurring through MalF500 mutant, as we could detect transient currents through the transporter in planar lipid bilayers.  Identification of a channel activity in MalF500 is consistent with the expanded alternating access model, which hypothesizes that transporters evolve to prevent leakage of a 119  specific substrate during transport, but that often small ion and water leakage can occur.  Accordingly, while we could detect chloride leakage in spheroplasts overexpressing MalF500, the membranes remained impermeable to a larger sulfate ion.     We also mutated the periplasmic gate of MalFGK2 to identify its role in preventing ion transport.  We found that disruption of the periplasmic gate on its own was insufficient to elucidate a channel response in the transporter.  However, if the gate was disrupted in the MalF500 mutant, the transporter became quasi-permanently open, whereas the MalF500 mutant alone would rapidly cycle between the open and closed channel states.  We relate these findings to the importance of membrane impermeability during alternating access transport.  As proposed by the alternating access model, transporters must passage through their leaky transition states relatively quickly, or risk compromising the membrane permeability barrier.  Furthermore, we also draw comparisons of our findings to evolution of the cystic fibrosis transmembrane conductance regulator protein  (CFTR).  In CFTR, much of the biochemical investigation has focused on the gating region of the channel, presumably because loss of the second vestigial gate is what caused this ancient transporter to adopt channel activity (Miller 2010; Gadsby 2009).  However, our results clearly show that channel like activity is inherent to an alternating access transporter, and by simply biasing transporter conformation, channel activity can result.  This is in agreement with evolutionary analysis of CFTR, which suggests an early mutational divergence from the ABCA4 family occured not by mutations in the pore/gate region, rather in residues that are critical for stabilization of the conformation of the transporter (Jordan et al. 2008).  Thus, our work supports this potential mechanism for how a transporter may evolve to become a channel; first, the conformational equilibria of the transporter is disturbed, increasing transient leakage proposed by the expanded alternating access mode; second, additional mutations in the pore and gate regions allow fine-tuning of this channel activity.     The work on chloride conductance of the MalFGK2 transporter was enabled by the exceptionally strong catalogue of mutational, structural and functional analysis of this transporter.  However, this is just one example of a Type I ABC transporter that uses alternating access transport.  Similar studies must be accomplished in other alternating access transporter families, such as the major facilitator superfamily.  This would provide additional in vitro evidence necessary to support the expanded alternating access model outside of molecular dynamics simulations. 120    In conclusion, we have provided considerable biophysical evidence for the expanded alternating access model, which had only been previously explored by molecular dynamics simulations.  The method we used to identify the transient conducting states of MalFGK2 can be used as a template for future studies on other alternating access transporters.  We also contribute a new tool for the biochemical analysis of membrane proteins, the peptidisc.  Not only do we demonstrate the wide applicability of the peptidisc, but our studies give insight into the mechanism of peptidisc formation, which may be utilized to optimize the technique in later studies.  Furthermore, we demonstrate that various model membrane proteins reconstituted into the peptidisc are functional, have increased stability and are homogenous.  This study therefore provides benchmarks so that the technique can be applied by other researchers to new systems.  We also apply the peptidisc, in combination with SEC-PCP-SILAC, as a high-throughput approach to catalogue the E. coli membrane interactome.  We deposit our high quality interaction list in the Open Source Framework (https://osf.io/dhnm6/?view_only=538182ab0e4e4e3d8dffffc08f93dfcf)  and details of the peptidisc method to enable comparative interactomics of membrane proteins, and eventual discovery of new protein complexes that reside in this physiochemically unique environment of the cell.                  121    References Althoff, Thorsten, Deryck J. Mills, Jean-Luc Popot, and Werner Kühlbrandt. 2011. “Arrangement of Electron Transport Chain Components in Bovine Mitochondrial Supercomplex I1III2IV1.” The EMBO Journal 30 (22): 4652–64. Anantharamaiah, G. M., J. L. Jones, C. G. Brouillette, C. F. Schmidt, B. H. Chung, T. A. Hughes, A. S. Bhown, and J. P. Segrest. 1985. “Studies of Synthetic Peptide Analogs of the Amphipathic Helix. Structure of Complexes with Dimyristoyl Phosphatidylcholine.” The Journal of Biological Chemistry 260 (18): 10248–55. Antharavally, Babu S., Krishna A. Mallia, Michael M. Rosenblatt, Ashok M. Salunkhe, John C. Rogers, Paul Haney, and Navid Haghdoost. 2011. “Efficient Removal of Detergents from Proteins and Peptides in a Spin Column Format.” Analytical Biochemistry 416 (1): 39–44. Arifuzzaman, Mohammad, Maki Maeda, Aya Itoh, Kensaku Nishikata, Chiharu Takita, Rintaro Saito, Takeshi Ara, et al. 2006. “Large-Scale Identification of Protein-Protein Interaction of Escherichia Coli K-12.” Genome Research 16 (5): 686–91. Ashcroft, F., D. Gadsby, and C. Miller. 2009. “Introduction. The Blurred Boundary between Channels and Transporters: We Dedicate This Volume to the Memory of Peter Lauger, a Pioneer of the Link between Channels and Pumps.” Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences 364 (1514): 145–47. Babu, Mohan, Cedoljub Bundalovic-Torma, Charles Calmettes, Sadhna Phanse, Qingzhou Zhang, Yue Jiang, Zoran Minic, et al. 2017. “Global Landscape of Cell Envelope Protein Complexes in Escherichia Coli.” Nature Biotechnology, November. https://doi.org/10.1038/nbt.4024. Babu, Mohan, James Vlasblom, Shuye Pu, Xinghua Guo, Chris Graham, Björn D. M. Bean, Helen E. Burston, et al. 2012. “Interaction Landscape of Membrane-Protein Complexes in Saccharomyces Cerevisiae.” Nature 489 (7417): 585–89. Baker, Mariah R., Guizhen Fan, and Irina I. Serysheva. 2015. “Single-Particle Cryo-EM of the Ryanodine Receptor Channel.” European Journal of Translational Myology 25 (1). https://doi.org/10.4081/bam.2015.1.35. Bao, Huan, Kush Dalal, Eric Cytrynbaum, and Franck Duong. 2015. “Sequential Action of MalE and Maltose Allows Coupling ATP Hydrolysis to Translocation in the MalFGK2 Transporter.” The Journal of Biological Chemistry 290 (42): 25452–60. Bao, Huan, Kush Dalal, Victor Wang, Isabelle Rouiller, and Franck Duong. 2013a. “The Maltose ABC Transporter: Action of Membrane Lipids on the Transporter Stability, Coupling and ATPase Activity.” Biochimica et Biophysica Acta 1828 (8): 1723–30. Bao, Huan, and Franck Duong. 2012. “Discovery of an Auto-Regulation Mechanism for the Maltose ABC Transporter MalFGK2.” PloS One 7 (4): e34836. ———. 2013a. “ATP Alone Triggers the Outward Facing Conformation of the Maltose ATP-Binding Cassette Transporter.” The Journal of Biological Chemistry 288 (5): 3439–48. ———. 2013b. “Phosphatidylglycerol Directs Binding and Inhibitory Action of EIIAGlc Protein on the Maltose Transporter.” The Journal of Biological Chemistry 288 (33): 23666–74. ———. 2014. “Nucleotide-Free MalK Drives the Transition of the Maltose Transporter to the Inward-Facing Conformation.” The Journal of Biological Chemistry 289 (14): 9844–51. Bayburt, Timothy H., Yelena V. Grinkova, and Stephen G. Sligar. 2006. “Assembly of Single 122  Bacteriorhodopsin Trimers in Bilayer Nanodiscs.” Archives of Biochemistry and Biophysics 450 (2): 215–22. Bayburt, Timothy H., and Stephen G. Sligar. 2002. “Single-Molecule Height Measurements on Microsomal Cytochrome P450 in Nanometer-Scale Phospholipid Bilayer Disks.” Proceedings of the National Academy of Sciences of the United States of America 99 (10): 6725–30. Bear, C. E., C. H. Li, N. Kartner, R. J. Bridges, T. J. Jensen, M. Ramjeesingh, and J. R. Riordan. 1992. “Purification and Functional Reconstitution of the Cystic Fibrosis Transmembrane Conductance Regulator (CFTR).” Cell 68 (4): 809–18. Bechara, Chérine, Anne Nöll, Nina Morgner, Matteo T. Degiacomi, Robert Tampé, and Carol V. Robinson. 2015a. “A Subset of Annular Lipids Is Linked to the Flippase Activity of an ABC Transporter.” Nature Chemistry 7 (3): 255–62. Beck, Scarlet, Annette Michalski, Oliver Raether, Markus Lubeck, Stephanie Kaspar, Niels Goedecke, Carsten Baessmann, et al. 2015. “The Impact II, a Very High-Resolution Quadrupole Time-of-Flight Instrument (QTOF) for Deep Shotgun Proteomics.” Molecular & Cellular Proteomics: MCP 14 (7): 2014–29. Beek, Josy ter, Albert Guskov, and Dirk Jan Slotboom. 2014. “Structural Diversity of ABC Transporters.” The Journal of General Physiology 143 (4): 419–35. Bessonneau, Pascal, Véronique Besson, Ian Collinson, and Franck Duong. 2002. “The SecYEG Preprotein Translocation Channel Is a Conformationally Dynamic and Dimeric Structure.” The EMBO Journal 21 (5): 995–1003. Bibow, Stefan, Yevhen Polyhach, Cédric Eichmann, Celestine N. Chi, Julia Kowal, Stefan  Albiez, Robert A. McLeod, et al. 2017. “Solution Structure of Discoidal High-Density Lipoprotein Particles with a Shortened Apolipoprotein A-I.” Nature Structural & Molecular Biology 24 (2): 187–93. Blankenship, Robert E., David M. Tiede, James Barber, Gary W. Brudvig, Graham Fleming, Maria Ghirardi, M. R. Gunner, et al. 2011. “Comparing Photosynthetic and Photovoltaic Efficiencies and Recognizing the Potential for Improvement.” Science 332 (6031): 805–9. Böhm, Simon, Anke Licht, Steven Wuttge, Erwin Schneider, and Enrica Bordignon. 2013. “Conformational Plasticity of the Type I Maltose ABC Importer.” Proceedings of the National Academy of Sciences of the United States of America 110 (14): 5492–97. Boldog, Thomas, Mingshan Li, and Gerald L. Hazelbauer. 2007. “Using Nanodiscs to Create Water-Soluble Transmembrane Chemoreceptors Inserted in Lipid Bilayers.” Methods in Enzymology 423: 317–35. Booth, William T., Caleb R. Schlachter, Swanandi Pote, Nikita Ussin, Nicholas J. Mank, Vincent  Klapper, Lesa R. Offermann, Chuanbing Tang, Barry K. Hurlburt, and Maksymilian Chruszcz. 2018. “Impact of an N-Terminal Polyhistidine Tag on Protein Thermal Stability.” ACS Omega 3 (1): 760–68. Bradford, M. M. 1976. “A Rapid and Sensitive Method for the Quantitation of Microgram Quantities of Protein Utilizing the Principle of Protein-Dye Binding.” Analytical Biochemistry 72 (May): 248–54. Broecker, Jana, Bryan T. Eger, and Oliver P. Ernst. 2017. “Crystallogenesis of Membrane Proteins Mediated by Polymer-Bounded Lipid Nanodiscs.” Structure  25 (2): 384–92. Brunham, Liam R., and Michael R. Hayden. 2015. “Human Genetics of HDL: Insight into Particle Metabolism and Function.” Progress in Lipid Research 58 (April): 14–25. Bunai, Keigo, and Kunio Yamane. 2005. “Effectiveness and Limitation of Two-Dimensional Gel  Electrophoresis in Bacterial Membrane Protein Proteomics and Perspectives.” Journal of Chromatography. B, Analytical Technologies in the Biomedical and Life Sciences 815 (1-2): 227–36. 123  Calabrese, Antonio N., Thomas G. Watkinson, Peter J. F. Henderson, Sheena E. Radford, and Alison E. Ashcroft. 2015. “Amphipols Outperform Dodecylmaltoside Micelles in Stabilizing Membrane Protein Structure in the Gas Phase.” Analytical Chemistry 87 (2): 1118–26. Carlson, Michael L., Huan Bao, and Franck Duong. 2016. “Formation of a Chloride-Conducting  State in the Maltose ATP-Binding Cassette (ABC) Transporter.” The Journal of Biological Chemistry 291 (23): 12119–25. Chaptal, Vincent, Frédéric Delolme, Arnaud Kilburg, Sandrine Magnard, Cédric Montigny, Martin Picard, Charlène Prier, et al. 2017. “Quantification of Detergents Complexed with Membrane Proteins.” Scientific Reports 7 (February): 41751. Chen, Jue. 2013. “Molecular Mechanism of the Escherichia Coli Maltose Transporter.” Current Opinion in Structural Biology 23 (4): 492–98. Chen, Tsung-Yu, and Tzyh-Chang Hwang. 2008. “CLC-0 and CFTR: Chloride Channels Evolved from Transporters.” Physiological Reviews 88 (2): 351–87. Chien, Chih-Ta Henry, Lukas R. Helfinger, Mark J. Bostock, Andras Solt, Yi Lei Tan, and Daniel Nietlispach. 2017. “An Adaptable Phospholipid Membrane Mimetic System for Solution NMR Studies of Membrane Proteins.” Journal of the American Chemical Society 139 (42): 14829–32. Corin, Karolina, Philipp Baaske, Deepali B. Ravel, Junyao Song, Emily Brown, Xiaoqiang Wang, Christoph J. Wienken, et al. 2011a. “Designer Lipid-like Peptides: A Class of Detergents for Studying Functional Olfactory Receptors Using Commercial Cell-Free Systems.” PloS One 6 (11): e25067. Covitz, K. M., C. H. Panagiotidis, L. I. Hor, M. Reyes, N. A. Treptow, and H. A. Shuman. 1994. “Mutations That Alter the Transmembrane Signalling Pathway in an ATP Binding Cassette (ABC) Transporter.” The EMBO Journal 13 (7): 1752–59. Cox, Jürgen, Marco Y. Hein, Christian A. Luber, Igor Paron, Nagarjuna Nagaraj, and Matthias Mann. 2014. “Accurate Proteome-Wide Label-Free Quantification by Delayed Normalization and Maximal Peptide Ratio Extraction, Termed MaxLFQ.” Molecular & Cellular Proteomics: MCP 13 (9): 2513–26. Cox, Jürgen, and Matthias Mann. 2008. “MaxQuant Enables High Peptide Identification Rates, Individualized P.p.b.-Range Mass Accuracies and Proteome-Wide Protein Quantification.” Nature Biotechnology 26 (12): 1367–72. Craig, Andrew F., Emily E. Clark, Indra D. Sahu, Rongfu Zhang, Nick D. Frantz, M. Sameer Al-Abdul-Wahid, Carole Dabney-Smith, Dominik Konkolewicz, and Gary A. Lorigan. 2016. “Tuning the Size of Styrene-Maleic Acid Copolymer-Lipid Nanoparticles (SMALPs) Using RAFT Polymerization for Biophysical Studies.” Biochimica et Biophysica Acta 1858 (11): 2931–39. Cronan, John E. 2005. “Targeted and Proximity-Dependent Promiscuous Protein Biotinylation by a Mutant Escherichia Coli Biotin Protein Ligase.” The Journal of Nutritional Biochemistry 16 (7): 416–18. Cuevas Arenas, Rodrigo, Bartholomäus Danielczak, Anne Martel, Lionel Porcar, Cécile Breyton, Christine Ebel, and Sandro Keller. 2017. “Fast Collisional Lipid Transfer Among Polymer-Bounded Nanodiscs.” Scientific Reports 7 (April): 45875. Dalal, Kush, Catherine S. Chan, Stephen G. Sligar, and Franck Duong. 2012. “Two Copies of the SecY Channel and Acidic Lipids Are Necessary to Activate the SecA Translocation ATPase.” Proceedings of the National Academy of Sciences of the United States of America 109 (11): 4104–9. Dalal, Kush, and Franck Duong. 2009. “The SecY Complex Forms a Chennel Capable of Ionic Discrimination.”  EMBO Reports 10(7): 762-68. Daus, Martin L., Mathias Grote, and Erwin Schneider. 2009a. “The MalF P2 Loop of the ATP-Binding Cassette Transporter MalFGK2 from Escherichia Coli and Salmonella Enterica 124  Serovar Typhimurium Interacts with Maltose Binding Protein (MalE) throughout the Catalytic Cycle.” Journal of Bacteriology 191 (3): 754–61. Davidson, A. L., H. A. Shuman, and H. Nikaido. 1992. “Mechanism of Maltose Transport in Escherichia Coli: Transmembrane Signaling by Periplasmic Binding Proteins.” Proceedings of the National Academy of Sciences of the United States of America 89 (6): 2360–64. Davidson, Amy L. 2002. “Mechanism of Coupling of Transport to Hydrolysis in Bacterial ATP-Binding Cassette Transporters.” Journal of Bacteriology 184 (5): 1225–33. Davidson, Amy L., Elie Dassa, Cedric Orelle, and Jue Chen. 2008. “Structure, Function, and Evolution of Bacterial ATP-Binding Cassette Systems.” Microbiology and Molecular Biology Reviews: MMBR 72 (2): 317–64, table of contents. DeFelice, Louis J., and Tapasree Goswami. 2007. “Transporters as Channels.” Annual Review of Physiology 69 (1): 87–112. Denisov, Ilia G., and Stephen G. Sligar. 2011. “Cytochromes P450 in Nanodiscs.” Biochimica et Biophysica Acta (BBA) - Proteins and Proteomics 1814 (1): 223–29. ———. 2016. “Nanodiscs for Structural and Functional Studies of Membrane Proteins.” Nature Structural & Molecular Biology 23 (6): 481–86. Does, Chris van der, and Robert Tampé. 2004. “How Do ABC Transporters Drive Transport?” Biological Chemistry 385 (10): 927–33. Dominguez Pardo, Juan J., Jonas M. Dörr, Aditya Iyer, Ruud C. Cox, Stefan Scheidelaar, Martijn C. Koorengevel, Vinod Subramaniam, and J. Antoinette Killian. 2016. “Solubilization of Lipids and Lipid Phases by the Styrene–maleic Acid Copolymer.” European Biophysics Journal: EBJ 46 (1): 91–101. Dörr, Jonas M., Martijn C. Koorengevel, Marre Schäfer, Alexander V. Prokofyev, Stefan Scheidelaar, Elwin A. W. van der Cruijsen, Timothy R. Dafforn, Marc Baldus, and J. Antoinette Killian. 2014. “Detergent-Free Isolation, Characterization, and Functional Reconstitution of a Tetrameric K+ Channel: The Power of Native Nanodiscs.” Proceedings of the National Academy of Sciences of the United States of America 111 (52): 18607–12. Drew, Kevin, Chanjae Lee, Ryan L. Huizar, Fan Tu, Blake Borgeson, Claire D. McWhite, Yun Ma, John B. Wallingford, and Edward M. Marcotte. 2017. “Integration of over 9,000 Mass Spectrometry Experiments Builds a Global Map of Human Protein Complexes.” Molecular Systems Biology 13 (6): 932. Dunham, Wade H., Michael Mullin, and Anne-Claude Gingras. 2012. “Affinity-Purification Coupled to Mass Spectrometry: Basic Principles and Strategies.” Proteomics 12 (10): 1576–90. Duquette, P. P., P. Bissonnette, and J. Y. Lapointe. 2001. “Local Osmotic Gradients Drive the Water Flux Associated with Na(+)/glucose Cotransport.” Proceedings of the National Academy of Sciences of the United States of America 98 (7): 3796–3801. Enright, A.J., S. Van Dongen, and C.A. Ouzounis. 2002. “An Efficient Algorithm for Large-Scale Detection of Protein Families.” Nucleic Acids Research 30(7): 1575-84. Erickson, Keesha E., James D. Winkler, Danh T. Nguyen, Ryan T. Gill, and Anushree Chatterjee. 2017. “The Tolerome: A Database of Transcriptome-Level Contributions to Diverse Escherichia Coli Resistance and Tolerance Phenotypes.” ACS Synthetic Biology 6 (12): 2302–15. Fabre, Lucien, Huan Bao, James Innes, Franck Duong, and Isabelle Rouiller. 2017. “Negative Stain Single-Particle EM of the Maltose Transporter in Nanodiscs Reveals Asymmetric Closure of MalKand Catalytic Roles of ATP, MalE, and Maltose.” The Journal of Biological Chemistry 292 (13): 5457–64. Faith, Jeremiah J., Michael E. Driscoll, Vincent A. Fusaro, Elissa J. Cosgrove, Boris Hayete, Frank S. Juhn, Stephen J. Schneider, and Timothy S. Gardner. 2008. “Many Microbe Microarrays Database: Uniformly Normalized Affymetrix Compendia with Structured Experimental Metadata.” Nucleic Acids Research 36 (Database issue): D866–70. 125  Flayhan, Ali, Haydyn D. T. Mertens, Yonca Ural-Blimke, Maria Martinez Molledo, Dmitri I. Svergun, and Christian Löw. 2018. “saposin Lipid Nanoparticles: A Highly Versatile and Modular Tool for Membrane Protein Research.” Structure  26 (2): 345–55.e5. Forrest, Lucy R., and Gary Rudnick. 2009. “The Rocking Bundle: A Mechanism for Ion-Coupled Solute Flux by Symmetrical Transporters.” Physiology  24 (December): 377–86. Frauenfeld, Jens, James Gumbart, Eli O. van der Sluis, Soledad Funes, Marco Gartmann, Birgitta Beatrix, Thorsten Mielke, et al. 2011. “Cryo-EM Structure of the ribosome–SecYE Complex in the Membrane Environment.” Nature Structural & Molecular Biology 18 (5): 614–21. Frauenfeld, Jens, Robin Löving, Jean-Paul Armache, Andreas F-P Sonnen, Fatma Guettou, Per Moberg, Lin Zhu, et al. 2016. “A saposin-Lipoprotein Nanoparticle System for Membrane Proteins.” Nature Methods 13 (4): 345–51. Gadsby, David C. 2009. “Ion Channels versus Ion Pumps: The Principal Difference, in Principle.” Nature Reviews. Molecular Cell Biology 10 (5): 344–52. Gadsby, David C., Paola Vergani, and László Csanády. 2006. “The ABC Protein Turned Chloride Channel Whose Failure Causes Cystic Fibrosis.” Nature 440 (7083): 477–83. Gao, Yuan, Erhu Cao, David Julius, and Yifan Cheng. 2016. “TRPV1 Structures in Nanodiscs Reveal Mechanisms of Ligand and Lipid Action.” Nature 534 (7607): 347–51. Garner, Ashley E., D. Alastair Smith, and Nigel M. Hooper. 2008. “Visualization of Detergent Solubilization of Membranes: Implications for the Isolation of Rafts.” Biophysical Journal 94 (4): 1326–40. Getz, Godfrey S., and Catherine A. Reardon. 2011. “Apolipoprotein A-I and A-I Mimetic Peptides: A Role in Atherosclerosis.” Journal of Inflammation Research 4 (June): 83–92. Grinkova, Yelena V., Ilia G. Denisov, and Stephen G. Sligar. 2010. “Engineering Extended Membrane Scaffold Proteins for Self-Assembly of Soluble Nanoscale Lipid Bilayers.” Protein Engineering, Design & Selection: PEDS 23 (11): 843–48. Gotzke, Hansjorg, Isolde Palombo, Claudio Muheim, Else Perrody, Pierre Genevaux, Renuka Kudva, Matthias Muller, and Daniel O. Daley. 2014. “YfgM Is an Ancillary Subunit of the SecYEG Translocon Escherichia Coli.”  The Journal of Biological Chemistry 289 (27):19089-97. Gulati, Sonali, Mohammed Jamshad, Timothy J. Knowles, Kerrie A. Morrison, Rebecca Downing, Natasha Cant, Richard Collins, et al. 2014. “Detergent-Free Purification of ABC (ATP-Binding-Cassette) Transporters.” Biochemical Journal 461 (2): 269–78. Gundry, Rebekah L., Melanie Y. White, Christopher I. Murray, Lesley A. Kane, Qin Fu, Brian A. Stanley, and Jennifer E. Van Eyk. 2009. “Preparation of Proteins and Peptides for Mass Spectrometry Analysis in a Bottom-Up Proteomics Workflow.” In Current Protocols in Molecular Biology. Gupta, Kallol, Joseph A. C. Donlan, Jonathan T. S. Hopper, Povilas Uzdavinys, Michael Landreh, Weston B. Struwe, David Drew, Andrew J. Baldwin, Phillip J. Stansfeld, and Carol V. Robinson. 2017. “The Role of Interfacial Lipids in Stabilizing Membrane Protein Oligomers.” Nature 541 (7637): 421–24. Hagn, Franz, Manuel Etzkorn, Thomas Raschle, and Gerhard Wagner. 2013. “Optimized Phospholipid Bilayer Nanodiscs Facilitate High-Resolution Structure Determination of Membrane Proteins.” Journal of the American Chemical Society 135 (5): 1919–25. Hagn, Franz, Mahmoud L. Nasr, and Gerhard Wagner. 2018. “Assembly of Phospholipid Nanodiscs of Controlled Size for Structural Studies of Membrane Proteins by NMR.” Nature Protocols 13 (1): 79–98. Heide, T. van der and B. Poolman. 2002. “ABC Transporters: One, Two or Four Extracytoplasmic Substrate-Binding Sites?” EMBO Reports 3 (10):938-43. Hoffmann, Carsten, Guido Gaietta, Moritz Bünemann, Stephen R. Adams, Silke Oberdorff- 126  Maass, Björn Behr, Jean-Pierre Vilardaga, Roger Y. Tsien, Mark H. Ellisman, and Martin J. Lohse. 2005. “A FlAsH-Based FRET Approach to Determine G Protein-Coupled Receptor Activation in Living Cells.” Nature Methods 2 (3): 171–76. Hohn, Michael, Grant Tang, Grant Goodyear, P. R. Baldwin, Zhong Huang, Pawel A. Penczek,  Chao Yang, Robert M. Glaeser, Paul D. Adams, and Steven J. Ludtke. 2007. “SPARX, a New Environment for Cryo-EM Image Processing.” Journal of Structural Biology 157 (1): 47–55. Huang, Hailiang, Bruno M. Jedynak, and Joel S. Bader. 2007. “Where Have All the Interactions  Gone? Estimating the Coverage of Two-Hybrid Protein Interaction Maps.” PLoS Computational Biology 3 (11): e214. Hughes, Arwel V., Paul Rees, Peter Heathcote, and Michael R. Jones. 2006. “Kinetic Analysis of the Thermal Stability of the Photosynthetic Reaction Center from Rhodobacter Sphaeroides.” Biophysical Journal 90 (11): 4155–66. Hu, Pingzhao, Sarath Chandra Janga, Mohan Babu, J. Javier Díaz-Mejía, Gareth Butland, Wenhong Yang, Oxana Pogoutse, et al. 2009. “Global Functional Atlas of Escherichia Coli Encompassing Previously Uncharacterized Proteins.” PLoS Biology 7 (4): e96. Hvorup, Rikki N., Birke A Goetz, Martina Niederer, Kaspar Hollenstein, Eduardo Perozo, and Kaspar P. Locher. 2007.  “Asymmetry in the Structure of the ABC Transporter-Binding Protein Complex BtuCD-BtuF.”  Science 317 (5843):1387-90. Ishihama, Yasushi, Juri Rappsilber, and Matthias Mann. 2006. “Modular Stop and Go Extraction Tips with Stacked Disks for Parallel and Multidimensional Peptide Fractionation in Proteomics.” Journal of Proteome Research 5 (4): 988–94. Islam, Rafique M., Mohsen Pourmousa, Denis Sviridov, Scott M. Gordon, Edward B. Neufeld, Lita A. Freeman, B. Scott Perrin, Richard W. Pastor, and Alan T. Remaley. 2018. “Structural Properties of Apolipoprotein A-I Mimetic Peptides That Promote ABCA1-Dependent Cholesterol Efflux.” Scientific Reports 8 (1). Nature Publishing Group: 2956. Jardetzky, O. 1966. “Simple Allosteric Model for Membrane Pumps.” Nature 211 (5052): 969–70. Jayaram, Hariharan, Alessio Accardi, Fang Wu, Carole Williams, and Christopher Miller. 2008. “Ion Permeation through a Cl--Selective Channel Designed from a CLC Cl-/H+ Exchanger.” Proceedings of the National Academy of Sciences of the United States of America 105 (32): 11194–99. Jordan, I. King, Karthik C. Kota, Guiying Cui, Christopher H. Thompson, and Nael A. McCarty. 2008. “Evolutionary and Functional Divergence between the Cystic Fibrosis Transmembrane Conductance Regulator and Related ATP-Binding Cassette Transporters.” Proceedings of the National Academy of Sciences of the United States of America 105 (48): 18865–70. Jun, D., R. G. Saer, J. D. Madden, and J. T. Beatty. 2013. “Use of New Strains of Rhodobacter Sphaeroides and a Modified Simple Culture Medium to Increase Yield and Facilitate Purification of the Reaction Centre.” Photosynthesis Research 120 (1-2): 197–205. Kaback, H. Ronald, R. Dunten, S. Frillingos, P. Venkatesan, I. Kwaw, W. Zhang, and Natalia Ermolova. 2007. “Site-Directed Alkylation and the Alternating Access Model for LacY.” Proceedings of the National Academy of Sciences of the United States of America 104 (2): 491–94. Kariyazono, Hirokazu, Ryo Nadai, Rin Miyajima, Yuki Takechi-Haraya, Teruhiko Baba, Akira Shigenaga, Keiichiro Okuhira, Akira Otaka, and Hiroyuki Saito. 2016. “Formation of Stable Nanodiscs by Bihelical Apolipoprotein A-I Mimetic Peptide.” Journal of Peptide Science: An Official Publication of the European Peptide Society 22 (2): 116–22. Katayama, H., J. Wang, F. Tama, L. Chollet, E. P. Gogol, R. J. Collier, and M. T. Fisher. 2010. 127  “Three-Dimensional Structure of the Anthrax Toxin Pore Inserted into Lipid Nanodiscs and Lipid Vesicles.” Proceedings of the National Academy of Sciences of the United States of America 107 (8): 3453–57. Khare, Dheeraj, Michael L. Oldham, Cedric Orelle, Amy L. Davidson, and Jue Chen. 2009. “Alternating Access in Maltose Transporter Mediated by Rigid-Body Rotations.” Molecular Cell 33 (4): 528–36. Kim, Dae In, K. C. Birendra, Wenhong Zhu, Khatereh Motamedchaboki, Valérie Doye, and Kyle J. Roux. 2014. “Probing Nuclear Pore Complex Architecture with Proximity-Dependent Biotinylation.” Proceedings of the National Academy of Sciences of the United States of America 111 (24): E2453–61. Kim, Dae In, and Kyle J. Roux. 2016. “Filling the Void: Proximity-Based Labeling of Proteins in Living Cells.” Trends in Cell Biology 26 (11): 804–17. Klock, Heath E., and Scott A. Lesley. 2009a. “The Polymerase Incomplete Primer Extension (PIPE) Method Applied to High-Throughput Cloning and Site-Directed Mutagenesis.” Methods in Molecular Biology  498: 91–103. Krishnamurthy, Harini, Chayne L. Piscitelli, and Eric Gouaux. 2009. “Unlocking the Molecular Secrets of Sodium-Coupled Transporters.” Nature 459 (7245): 347–55. Kristensen, Anders R., Joerg Gsponer, and Leonard J. Foster. 2012a. “A High-Throughput Approach for Measuring Temporal Changes in the Interactome.” Nature Methods 9 (9): 907–9. Lanzetta, P. A., L. J. Alvarez, P. S. Reinach, and O. A. Candia. 1979. “An Improved Assay for Nanomole Amounts of Inorganic Phosphate.” Analytical Biochemistry 100 (1): 95–97. Larance, Mark, Kathryn J. Kirkwood, Michele Tinti, Alejandro Brenes Murillo, Michael A. J. Ferguson, and Angus I. Lamond. 2016a. “Global Membrane Protein Interactome Analysis Using In Vivo Crosslinking and Mass Spectrometry-Based Protein Correlation Profiling.” Molecular & Cellular Proteomics: MCP 15 (7): 2476–90. Larsen, Andreas N., Kasper K. Sørensen, Nicolai T. Johansen, Anne Martel, Jacob J. K. Kirkensgaard, Knud J. Jensen, Lise Arleth, and Søren Roi Midtgaard. 2016. “Dimeric Peptides with Three Different Linkers Self-Assemble with Phospholipids to Form Peptide Nanodiscs That Stabilize Membrane Proteins.” Soft Matter 12 (27): 5937–49. Lee, Sarah C., Tim J. Knowles, Vincent L. G. Postis, Mohammed Jamshad, Rosemary A. Parslow, Yu-Pin Lin, Adrian Goldman, et al. 2016a. “A Method for Detergent-Free Isolation of Membrane Proteins in Their Local Lipid Environment.” Nature Protocols 11 (7): 1149–62. Li, Canhui, Mohabir Ramjeesingh, and Christine E. Bear. 1996. “Purified Cystic Fibrosis Transmembrane Conductance Regulator (CFTR) Does Not Function as an ATP Channel.” The Journal of Biological Chemistry 271 (20): 11623–26. Li, Jing, Saher A. Shaikh, Giray Enkavi, Po-Chao Wen, Zhijian Huang, and Emad Tajkhorshid. 2013. “Transient Formation of Water-Conducting States in Membrane Transporters.” Proceedings of the National Academy of Sciences of the United States of America 110 (19): 7696–7701. Li, Jing, Po-Chao Wen, Mahmoud Moradi, and Emad Tajkhorshid. 2015. “Computational Characterization of Structural Dynamics Underlying Function in Active Membrane Transporters.” Current Opinion in Structural Biology 31 (April): 96–105. Li, Jun, Michele R. Richards, Dhanashri Bagal, Iain D. G. Campuzano, Elena N. Kitova, Zi Jian Xiong, Gilbert G. Privé, and John S. Klassen. 2016. “Characterizing the Size and Composition of saposin A Lipoprotein Picodiscs.” Analytical Chemistry 88 (19): 9524–31. Lipfert, Jan, Linda Columbus, Vincent B. Chu, and Sebastian Doniach. 2007. “Analysis of Small-Angle X-Ray Scattering Data of Protein–detergent Complexes by Singular Value Decomposition.” Journal of Applied Crystallography 40 (s1): s235–39. Li, Weikai, Sol Schulman, Dana Boyd, Karl Erlandson, Jon Beckwith, and Tom A. Rapoport. 2007. “The Plug Domain of the SecY Protein Stabilizes the Closed State of the 128  Translocation Channel and Maintains a Membrane Seal.” Molecular Cell 26 (4): 511–21. Lycklama A Nijeholt, Jelger A., Ruslan Vietrov, Gea K. Schuurman-Wolters, and Bert Poolman. 2018. “Energy Coupling Efficiency in the Type I ABC Transporter GlnPQ.” Journal of Molecular Biology, February. https://doi.org/10.1016/j.jmb.2018.02.001. MacAulay, Nanna, Ulrik Gether, Dan A. Klaeke, and Thomas Zeuthen. 2002. “Passive Water and Urea Permeability of a Human Na(+)-Glutamate Cotransporter Expressed in Xenopus Oocytes.” The Journal of Physiology 542 (Pt 3): 817–28. Maddalo, Gianluca, Filippa Stenberg-Bruzell, Hansjörg Götzke, Stephen Toddo, Patrik Björkholm, Hanna Eriksson, Peter Chovanec, et al. 2011a. “Systematic Analysis of Native Membrane Protein Complexes in Escherichia Coli.” Journal of Proteome Research 10 (4): 1848–59. Maillard, Antoine P., Shifana Lalani, Filo Silva, Dominique Belin, and Franck Duong. 2007. “Deregulation of the SecYEG Translocation Channel upon Removal of the Plug Domain.” The Journal of Biological Chemistry 282 (2): 1281–87. Maire, M. le, P. Champeil, and J. V. Moller. 2000. “Interaction of Membrane Proteins and Lipids with Solubilizing Detergents.” Biochimica et Biophysica Acta 1508 (1-2): 86–111. Maloney, P. C., E. R. Kashket, and T. H. Wilson. 1974. “A Protonmotive Force Drives ATP Synthesis in Bacteria.” Proceedings of the National Academy of Sciences of the United States of America 71 (10): 3896–3900. Marty, Michael T., Kyle C. Wilcox, William L. Klein, and Stephen G. Sligar. 2013. “Nanodisc-Solubilized Membrane Protein Library Reflects the Membrane Proteome.” Analytical and Bioanalytical Chemistry 405 (12): 4009–16. Marty, Michael T., Andrew J. Baldwin, Erik G. Marklund, Georg K. A. Hochberg, Justin L. P. Benesch, and Carol V. Robinson. 2015. “Bayesian Deconvolution of Mass and Ion Mobility Spectra: From Binary Interactions to Polydisperse Ensembles.” Analytical Chemistry 87 (8): 4370–76. McBride, Zachary, Donglai Chen, Christy Reick, Jun Xie, and Daniel B. Szymanski. 2017. “Global Analysis of Membrane-Associated Protein Oligomerization Using Protein Correlation Profiling.” Molecular & Cellular Proteomics: MCP 16 (11): 1972–89. McGregor, Clare-Louise, Lu Chen, Neil C. Pomroy, Peter Hwang, Sandy Go, Avijit Chakrabartty, and Gilbert G. Privé. 2003. “Lipopeptide Detergents Designed for the Structural Study of Membrane Proteins.” Nature Biotechnology 21 (2): 171–76. Mei, Xiaohu, and David Atkinson. 2015. “Lipid-Free Apolipoprotein A-I Structure: Insights into HDL Formation and Atherosclerosis Development.” Archives of Medical Research 46 (5): 351–60. Mellacheruvu, Dattatreya, Zachary Wright, Amber L. Couzens, Jean-Philippe Lambert, Nicole A. St-Denis, Tuo Li, Yana V. Miteva, et al. 2013. “The CRAPome: A Contaminant Repository for Affinity Purification-Mass Spectrometry Data.” Nature Methods 10 (8): 730–36. Midtgaard, Søren Roi, Martin Cramer Pedersen, Jacob Judas Kain Kirkensgaard, Kasper Kildegaard Sørensen, Kell Mortensen, Knud J. Jensen, and Lise Arleth. 2014. “Self-Assembling Peptides Form Nanodiscs That Stabilize Membrane Proteins.” Soft Matter 10 (5): 738–52. Mills, Allan, Hai-Tuong Le, James W. Coulton, and Franck Duong. 2014. “FhuA Interactions in a Detergent-Free Nanodisc Environment.” Biochimica et Biophysica Acta (BBA) - Biomembranes 1838 (1): 364–71. Mindell, Joseph A., and Nikolaus Grigorieff. 2003. “Accurate Determination of Local Defocus and Specimen Tilt in Electron Microscopy.” Journal of Structural Biology 142 (3): 334–47. Mishra, Vinod K., Mayakonda N. Palgunachari, N. Rama Krishna, John Glushka, Jere P. Segrest, and G. M. Anantharamaiah. 2008. “Effect of Leucine to Phenylalanine Substitution on the Nonpolar Face of a Class A Amphipathic Helical Peptide on Its Interaction with Lipid.” The Journal of Biological Chemistry 283 (49): 34393–402. 129  Mi, Wei, Yanyan Li, Sung Hwan Yoon, Robert K. Ernst, Thomas Walz, and Maofu Liao. 2017. “Structural Basis of MsbA-Mediated Lipopolysaccharide Transport.” Nature 549 (7671): 233–37. Mosca, Roberto, Arnaud Céol, Amelie Stein, Roger Olivella, and Patrick Aloy. 2014. “3did: A Catalog of Domain-Based Interactions of Known Three-Dimensional Structure.” Nucleic Acids Research 42 (Database issue): D374–79. Nepusz, Tamas, Haiyuan Yu, and Alberto Paccanaro. 2012. “Detecting Overlapping Protein Complexes in Protein-Protein Interaction Networks.”  Nature Methods 9(5): 471-472. Nguyen, Phong T., Qi Wen Li, Neena S. Kadaba, Jeffrey Y. Lai, Janet G. Yang, and Douglas C. Rees. 2015. “The Contribution of Methionine to the Stability of the Escherichia Coli MetNIQ ABC Transporter-Substrate Binding Protein Complex.” Biological Chemistry 396 (9-10): 1127–34. Oldham, Michael L., Dheeraj Khare, Florante A. Quiocho, Amy L. Davidson, and Jue Chen. 2007a. “Crystal Structure of a Catalytic Intermediate of the Maltose Transporter.” Nature 450 (7169): 515–21. Oluwole, Abraham Olusegun, Bartholomäus Danielczak, Annette Meister, Jonathan Oyebamiji Babalola, Carolyn Vargas, and Sandro Keller. 2017. “Solubilization of Membrane Proteins into Functional Lipid-Bilayer Nanodiscs Using a Diisobutylene/Maleic Acid Copolymer.” Angewandte Chemie  56 (7): 1919–24. Ong, Shao-En, Blagoy Blagoev, Irina Kratchmarova, Dan Bach Kristensen, Hanno Steen, Akhilesh Pandey, and Matthias Mann. 2002. “Stable Isotope Labeling by Amino Acids in Cell Culture, SILAC, as a Simple and Accurate Approach to Expression Proteomics.” Molecular & Cellular Proteomics: MCP 1 (5): 376–86. Ong, Shao-En, Leonard J. Foster, and Matthias Mann. 2003. “Mass Spectrometric-Based Approaches in Quantitative Proteomics.” Methods  29 (2): 124–30. Ong, Shao-En, and Matthias Mann. 2006. “A Practical Recipe for Stable Isotope Labeling by  Amino Acids in Cell Culture (SILAC).” Nature Protocols 1 (6): 2650–60. Orwick-Rydmark, Marcella, Thomas Arnold, and Dirk Linke. 2016. “The Use of Detergents to Purify Membrane Proteins.” In Current Protocols in Protein Science, 4.8.1–4.8.35. Palazzo, Gerardo, Francesco Lopez, and Antonia Mallardi. 2010. “Effect of Detergent Concentration on the Thermal Stability of a Membrane Protein: The Case Study of Bacterial Reaction Center Solubilized by N,N-Dimethyldodecylamine-N-Oxide.” Biochimica et Biophysica Acta (BBA) - Proteins and Proteomics 1804 (1): 137–46. Park, Eunyong, and Tom A. Rapoport. 2011. “Preserving the Membrane Barrier for Small Molecules during Bacterial Protein Translocation.” Nature 473 (7346): 239–42. Parmar, Mayuriben, Shaun Rawson, Charlotte A. Scarff, Adrian Goldman, Timothy R. Dafforn, Stephen P. Muench, and Vincent L. G. Postis. 2018. “Using a SMALP Platform to Determine a Sub-Nm Single Particle Cryo-EM Membrane Protein Structure.” Biochimica et Biophysica Acta 1860 (2): 378–83. Picard, Martin, Tassadite Dahmane, Manuel Garrigos, Carole Gauron, Fabrice Giusti, Marc le Maire, Jean-Luc Popot, and Philippe Champeil. 2006. “Protective and Inhibitory Effects of Various Types of Amphipols on the Ca2+-ATPase from Sarcoplasmic Reticulum: A Comparative Study.” Biochemistry 45 (6): 1861–69. Polovinkin, V., I. Gushchin, M. Sintsov, E. Round, T. Balandin, P. Chervakov, V. Shevchenko, et al. 2014. “High-Resolution Structure of a Membrane Protein Transferred from Amphipol to a Lipidic Mesophase.” The Journal of Membrane Biology 247 (9-10): 997–1004. Popot, Jean-Luc. 2010. “Amphipols, Nanodiscs, and Fluorinated Surfactants: Three Nonconventional Approaches to Studying Membrane Proteins in Aqueous Solutions.” Annual Review of Biochemistry 79 (1): 737–75. Popovic, Konstantin, John Holyoake, Régis Pomès, and Gilbert G. Privé. 2012. “Structure of 130  saposin A Lipoprotein Discs.” Proceedings of the National Academy of Sciences of the United States of America 109 (8): 2908–12. Prehna, Gerd, Guijin Zhang, Xiandi Gong, Marek Duszyk, Mark Okon, Lawrence P. McIntosh, Joel H. Weiner, and Natalie C. J. Strynadka. 2012. “A Protein Export Pathway Involving Escherichia Coli Porins.” Structure  20 (7): 1154–66. Privé, Gilbert G. 2009. “Lipopeptide Detergents for Membrane Protein Studies.” Current Opinion in Structural Biology 19 (4): 379–85. Qasem-Abdullah, Hiba, Michal Perach, Nurit Livnat-Levanon, and Oded Lewinson. 2017. “ATP Binding and Hydrolysis Disrupt the High-Affinity Interaction between the Heme ABC Transporter HmuUV and Its Cognate Substrate-Binding Protein.” The Journal of Biological Chemistry 292 (35): 14617–24. Rajagopala, Seesandra V., Patricia Sikorski, Ashwani Kumar, Roberto Mosca, James Vlasblom, Roland Arnold, Jonathan Franca-Koh, et al. 2014. “The Binary Protein-Protein Interaction Landscape of Escherichia Coli.” Nature Biotechnology 32 (3): 285–90. Rappsilber, Juri, Yasushi Ishihama, and Matthias Mann. 2003. “Stop and Go Extraction Tips for Matrix-Assisted Laser Desorption/ionization, Nanoelectrospray, and LC/MS Sample Pretreatment in Proteomics.” Analytical Chemistry 75 (3): 663–70. Rappsilber, Juri, Matthias Mann, and Yasushi Ishihama. 2007. “Protocol for Micro-Purification, Enrichment, Pre-Fractionation and Storage of Peptides for Proteomics Using StageTips.” Nature Protocols 2 (8): 1896–1906. Ravi, Sai Kishore, and Swee Ching Tan. 2015. “Progress and Perspectives in Exploiting Photosynthetic Biomolecules for Solar Energy Harnessing.” Energy & Environmental Science 8 (9): 2551–73. Ravula, Thirupathi, Nathaniel Z. Hardin, Sudheer Kumar Ramadugu, Sarah J. Cox, and Ayyalusamy Ramamoorthy. 2018. “Formation of pH-Resistant Monodispersed Polymer-Lipid Nanodiscs.” Angewandte Chemie  57 (5): 1342–45. Reißer, Sabine, Erik Strandberg, Thomas Steinbrecher, and Anne S. Ulrich. 2014. “3D Hydrophobic Moment Vectors as a Tool to Characterize the Surface Polarity of Amphiphilic Peptides.” Biophysical Journal 106 (11): 2385–94. Rice, Austin J., Aekyung Park, and Heather W. Pinkett. 2014. “Diversity in ABC Transporters: Type I, II and III Importers.”  Critical Reviews in Biochemistry and Molecular Biology 49 (5): 426-37. Ritchie, T. K., Y. V. Grinkova, T. H. Bayburt, I. G. Denisov, J. K. Zolnerciks, W. M. Atkins, and S. G. Sligar. 2009. “Reconstitution of Membrane Proteins in Phospholipid Bilayer Nanodiscs.” In Methods in Enzymology, 211–31. Rochette, Samuel, Guillaume Diss, Marie Filteau, Jean-Baptiste Leducq, Alexandre K. Dubé, and Christian R. Landry. 2015. “Genome-Wide Protein-Protein Interaction Screening by Protein-Fragment Complementation Assay (PCA) in Living Cells.” Journal of Visualized Experiments: JoVE, no. 97 (March). https://doi.org/10.3791/52255. Roux, Kyle J., Dae In Kim, and Brian Burke. 2013. “BioID: A Screen for Protein-Protein Interactions.” Current Protocols in Protein Science / Editorial Board, John E. Coligan ... [et Al.] 74 (November): Unit 19.23. Roy, Jahnabi, Holly Pondenis, Timothy M. Fan, and Aditi Das. 2015. “Direct Capture of Functional Proteins from Mammalian Plasma Membranes into Nanodiscs.” Biochemistry 54 (41): 6299–6302. Ruepp, Andreas, Barbara Brauner, Imtraud Dunger-Kaltenbach, Goar Frishman, Corinna Montrone, Michael Stransky, Brigitte Waegele, et al. 2008. “CORUM: The Comprehensive Resource of Mammalian Protein Complexes.”  Nucleic Acids Research 36 (Database issue):D646-50. Scheres, Sjors H. W. 2012. “RELION: Implementation of a Bayesian Approach to Cryo-EM Structure Determination.” Journal of Structural Biology 180 (3): 519–30. 131  Scott, Nichollas E., Lindsay D. Rogers, Anna Prudova, Nat F. Brown, Nikolaus Fortelny, Christopher M. Overall, and Leonard J. Foster. 2017a. “Interactome Disassembly during Apoptosis Occurs Independent of Caspase Cleavage.” Molecular Systems Biology 13 (1): 906. Segrest, J. P., M. K. Jones, A. E. Klon, C. J. Sheldahl, M. Hellinger, H. De Loof, and S. C. Harvey. 1999. “A Detailed Molecular Belt Model for Apolipoprotein A-I in Discoidal High Density Lipoprotein.” The Journal of Biological Chemistry 274 (45): 31755–58. Shen, Peter S., Xiaoyong Yang, Paul G. DeCaen, Xiaowen Liu, David Bulkley, David E. Clapham, and Erhu Cao. 2016. “The Structure of the Polycystic Kidney Disease Channel PKD2 in Lipid Nanodiscs.” Cell 167 (3): 763–73.e11. Silhavy, Thomas J., Daniel Kahne, and Suzanne Walker. 2010. “The Bacterial Cell Envelope.” Cold Spring Harbor Perspectives in Biology 2 (5): a000414. Singh, Satinder K., and Fred J. Sigworth. 2015. “Cryo-EM: Spinning the Micelles Away.” Structure  23 (9): 1561. Slotboom, Dirk J. 2014. “Structural and Mechanistic Insights into Prokaryotic Energy-Coupling Factor Transporters.” Nature Reviews. Microbiology 12 (2): 79–87. Smith, Anton A. A., Henriette E. Autzen, Tomas Laursen, Vincent Wu, Max Yen, Aaron Hall, Scott D. Hansen, Yifan Cheng, and Ting Xu. 2017. “Controlling Styrene Maleic Acid Lipid Particles through RAFT.” Biomacromolecules 18 (11): 3706–13. Stacey, R. Greg, Michael A. Skinnider, Nichollas E. Scott, and Leonard J. Foster. 2017a. “A Rapid and Accurate Approach for Prediction of Interactomes from Co-Elution Data (PrInCE).” BMC Bioinformatics 18 (1): 457. Stadler, Charlotte, Elton Rexhepaj, Vasanth R. Singan, Robert F. Murphy, Rainer Pepperkok,  Mathias Uhlén, Jeremy C. Simpson, and Emma Lundberg. 2013. “Immunofluorescence and Fluorescent-Protein Tagging Show High Correlation for Protein Localization in Mammalian Cells.” Nature Methods 10 (4): 315–23. Stam, Nicholas J., and Stephan Wilkens. 2017. “Structure of the Lipid Nanodisc-Reconstituted Vacuolar ATPase Proton Channel: DEFINITION OF THE INTERACTION OF ROTOR AND STATOR AND IMPLICATIONS FOR ENZYME REGULATION BY REVERSIBLE DISSOCIATION.” The Journal of Biological Chemistry 292 (5): 1749–61. Swainsbury, David J. K., Stefan Scheidelaar, Nicholas Foster, Rienk van Grondelle, J. Antoinette Killian, and Michael R. Jones. 2017. “The Effectiveness of Styrene-Maleic Acid (SMA) Copolymers for Solubilisation of Integral Membrane Proteins from SMA-Accessible and SMA-Resistant Membranes.” Biochimica et Biophysica Acta 1859 (10): 2133–43. Swainsbury, David J. K., Stefan Scheidelaar, Rienk van Grondelle, J. Antoinette Killian, and Michael R. Jones. 2014. “Bacterial Reaction Centers Purified with Styrene Maleic Acid Copolymer Retain Native Membrane Functional Properties and Display Enhanced Stability.” Angewandte Chemie  53 (44): 11803–7. Tam, Patrick C. K., Antoine P. Maillard, Kenneth K. Y. Chan, and Franck Duong. 2005. “Investigating the SecY Plug Movement at the SecYEG Translocation Channel.” The EMBO Journal 24 (19): 3380–88. Tang, G., L. Peng, D. Mann, C. Yang, P. A. Penczek, G. Goodyear, P. A. Baldwin, W. Jiang, and S. J. Ludtke. 2006. “EMAN2: Software for Image Analysis and Single Particle Reconstruction.” Microscopy and Microanalysis: The Official Journal of Microscopy Society of America, Microbeam Analysis Society, Microscopical Society of Canada 12 (S02): 388–89. Tao, Houchao, Sung Chang Lee, Arne Moeller, Rituparna Sinha Roy, Fai Yiu Siu, Jörg Zimmermann, Raymond C. Stevens, Clinton S. Potter, Bridget Carragher, and Qinghai Zhang. 2013. “Engineered Nanostructured β-Sheet Peptides Protect Membrane Proteins.” Nature Methods 10 (8): 759–61. 132  Tarassov, K., V. Messier, C. R. Landry, S. Radinovic, M. M. S. Molina, I. Shames, Y. Malitskaya, J. Vogel, H. Bussey, and S. W. Michnick. 2008. “An in Vivo Map of the Yeast Protein Interactome.” Science 320 (5882): 1465–70. Tyanova, Stefka, Matthias Mann, and Jürgen Cox. 2014. “MaxQuant for in-Depth Analysis of Large SILAC Datasets.” Methods in Molecular Biology  1188: 351–64. Vaccaro, A. M., F. Ciaffoni, M. Tatti, R. Salvioli, A. Barca, D. Tognozzi, and C. Scerch. 1995. “pH-Dependent Conformational Properties of saposins and Their Interactions with Phospholipid Membranes.” The Journal of Biological Chemistry 270 (51): 30576–80. Van Lenten, Brian J., Alan C. Wagner, G. M. Anantharamaiah, Mohamad Navab, Srinivasa T. Reddy, Georgette M. Buga, and Alan M. Fogelman. 2009. “Apolipoprotein A-I Mimetic Peptides.” Current Atherosclerosis Reports 11 (1): 52–57. Vollmer, W., M. von Rechenberg, and J. V. Höltje. 1999. “Demonstration of Molecular Interactions between the Murein Polymerase PBP1B, the Lytic Transglycosylase MltA, and the Scaffolding Protein MipA of Escherichia Coli.” The Journal of Biological Chemistry 274 (10): 6726–34. Wayne, R., K. Frick, and J. B. Neilands. 1976. “Siderophore Protection against Colicins M, B, V, and Ia in Escherichia Coli.” Journal of Bacteriology 126 (1): 7–12. Wan, Cuihon, Blake Borgeson, Sadhna Phanse, Fan Tu, Kevin Drew, Greg Clark, Xuejian Xiong, et al. 2015.  “Panorama of Ancient Metazon Macromolecular Complexes.” Nature 525 (7569):339-44. Wiener, Michael C. 2004. “A Pedestrian Guide to Membrane Protein Crystallization.” Methods  34 (3): 364–72. Wilcox, Kyle C., Matthew R. Marunde, Aditi Das, Pauline T. Velasco, Benjamin D. Kuhns, Michael T. Marty, Haoming Jiang, Chi-Hao Luan, Stephen G. Sligar, and William L. Klein. 2015. “Nanoscale Synaptic Membrane Mimetic Allows Unbiased High Throughput Screen That Targets Binding Sites for Alzheimer’s-Associated Aβ Oligomers.” PloS One 10 (4): e0125263. Yaghoubi, Houman, Michael Schaefer, Shayan Yaghoubi, Daniel Jun, Rudy Schlaf, J. Thomas Beatty, and Arash Takshi. 2017. “A ZnO Nanowire Bio-Hybrid Solar Cell.” Nanotechnology 28 (5): 054006. Yang, Zhengrong, Chi Wang, Qingxian Zhou, Jianli An, Ellen Hildebrandt, Luba A. Aleksandrov, John C. Kappes, et al. 2014a. “Membrane Protein Stability Can Be Compromised by Detergent Interactions with the Extramembranous Soluble Domains.” Protein Science: A Publication of the Protein Society 23 (6): 769–89. Yeung, Yee-Guide, and E. Richard Stanley. 2010. “Rapid Detergent Removal from Peptide Samples with Ethyl Acetate for Mass Spectrometry Analysis.” Current Protocols in Protein Science / Editorial Board, John E. Coligan ... [et Al.] Chapter 16 (February): Unit 16.12. Yin, Hang, and Aaron D. Flynn. 2016. “Drugging Membrane Protein Interactions.” Annual  Review of Biomedical Engineering 18 (July): 51–76. Zhang, Guoan, and Thomas A. Neubert. 2009. “Use of Stable Isotope Labeling by Amino Acids in Cell Culture (SILAC) for Phosphotyrosine Protein Identification and Quantitation.” Methods in Molecular Biology  527: 79–92, xi. Zhang, Xiao X., Catherine S. Chan, Huan Bao, Yuan Fang, Leonard J. Foster, and Franck Duong. 2012. “Nanodiscs and SILAC-Based Mass Spectrometry to Identify a Membrane Protein Interactome.” Journal of Proteome Research 11 (2): 1454–59. Zhou, Xingding, R. Manjunatha Kini, and J. Sivaraman. 2011. “Application of Isothermal Titration Calorimetry and Column Chromatography for Identification of Biomolecular Targets.” Nature Protocols 6 (2): 158–65. 133  Appendices Appendix 2A:  Additional Methods - Chapter 2  2A.1 Protein expression and purification      Unless otherwise stated, all proteins were expressed in E. coli BL21(DE3) (New England Biolabs) for 3 hours at 37ºC after induction at an OD of 0.4-0.7 in LB medium supplemented with required antibiotic.  Cells were harvested by low speed centrifugation (10,000 x g, 6 min) and resuspended in Buffer A (50mM Tris-HCl: pH 8; 100 mM NaCl; 10% glycerol).  Resuspended cells were treated with 1mM phenylmethylsulfonyl fluoride (PMSF) and lysed using a microfluidizer (Microfluidics) at 10,000 psi.  Unbroken cell debris and other aggregates were removed by an additional low speed centrifugation. Cytosolic and crude membrane fractions containing the overexpressed protein of interest were subsequently isolated by ultracentrifugation (100,000 x g, 45 minutes) and crude membrane fraction resuspended in Buffer A (50mM Tris-HCl: pH8, 100mM NaCl, 10% glycerol).  MalE and His-tagged MalFGK2 were purified as previously described, (Bao and Duong. 2012) expressed from plasmids pBAD33-MalE and pBAD22-FGKhis, respectively. Crude membrane containing His-tagged MalFGK2 were solubilized at 4°C overnight in Buffer A + 1% DDM and clarified by ultracentrifugation.  Solubilized MalFGK2 was isolated by Ni2+-chelating chromatography in Buffer  A + 0.02% DDM, washed in 5 column volumes (CV) of Buffer B (50mM Tris-HCl: pH 8; 200mM NaCl; 15mM imidazole; 10% glycerol) + 0.02% DDM, and then eluted in Buffer C (50mM Tris-HCl: pH 8; 100mM NaCl; 400mM imidazole; 10% glycerol) + 0.02% DDM. Protein MalE was isolated on Resource 15Q column, concentrated using a 30 kDa polysulfone filter (Pall Corporation), and then further purified on Superdex 200HR 10/300 GL column equilibrated in Buffer EQ (50mM Tris-HCl: pH 8, 50mM NaCl, 10% glycerol).  His-tagged-MSPL156 and His-tagged TonB23-329 were purified by Ni2+-chelating chromatography as previously described (Mills et al. 2014).  His-tagged Colicin M was expressed and purified according to established protocols from plasmid pMLD189 in the E. coli strain BW25113 (Mills et al. 2014).  His-tagged FhuA, encoded by plasmid pHX405, was expressed in E. coli strain AW740 (ΔompF, ΔompC) in M9 minimal media, and was purified in LDAO as previously described  (Mills et al. 2014).  OmpF was expressed from 134  E. coli JW2203 (ΔOmpC) as previously described (Prehna et al. 2012).  Prepared crude membrane was resuspended in Buffer A and the inner membrane solubilized by addition of 1% Triton X-100.  The outer membrane fraction (OM) was isolated by ultracentrifugation, resuspended in Buffer A + 1% LDAO at a concentration of 3mg/mL, and incubated overnight at 4ºC with gentle rocking. Insoluble material was removed by an additional ultracentrifugation step, and the clarified lysate was applied onto a Resource 15Q column pre-equilibriated in Buffer EQ + 0.1% LDAO.  OmpF was eluted by a linear 20mL gradient of 50-700mM NaCl, and further purified by Superdex 200HR 10/300 in Buffer A + 0.1% LDAO.  Expression and purification of His-tagged SecYEG was performed from the plasmid pBad22-His-EYG as previously described. (Dalal et al. 2012) Crude membranes were solubilized for one hour at 4°C in Buffer A + 1% DDM.  Solubilized material was clarified by ultra-centrifugation and passed over a 5mL Ni2+-NTA column. After extensive washing in Buffer A + 0.02% DDM, SecYEG was eluted in Buffer A + 0.02% DDM over a 20mL gradient of 0-600mM imidazole. The most concentrated fractions were pooled and diluted 5-fold in Buffer O (50mM Tris-HCl: pH 8, 10% glycerol + 0.02% DDM) before being applied to a 5mL Fast Flow S cation exchange column pre-equilibrated in Buffer EQ + 0.02% DDM. Bound protein was eluted over a 20mL gradient from 50-600mM NaCl in Buffer EQ + 0.02% DDM.  Plasmids pET28 encoding his-tagged MSPD1 and MSP1D1E3 proteins were transformed into BL21 cells and protein expression and purification was performed as previously described (Dalal et al 2012).  All proteins, with the exception of BRC, were flash frozen in liquid nitrogen immediately after purification and stored at -80°C for later use.  BRC was purified as previously described (Jun et al. 2013).  In brief, His-tagged BRC was expressed in Rhodobacter Sphaeroides RcX (ΔpuhA, ΔpufQBALMX, ΔrshI, ΔppsR) using plasmid pIND4-RC1.  A preculture of 10mL in RLB media (LB medium; 810µM MgCl2; 510 µM CaCl2) + 25 µg/mL kanamycin was transferred into 100ml of RLB-kan and grown overnight at 30ºC before transfer into 1 L of  freshly prepared RLB-kan.  After growth for 8 hours at 30ºC, BRC production was induced with 1mM IPTG for an additional 16 hours.  During growth and purification, light exposure was kept to a minimum.  Cells were harvested by low speed centrifugation, resuspended in Buffer A and lysed by French press (10,000 psi).  Unbroken cells and cell debris were removed by low speed centrifugation, and the supernatant treated with 1% LDAO overnight at 4ºC.  After removal of insoluble material by ultracentrifugation, the supernatant was supplemented with 10mM imidazole and the BRC purified by Ni2+-chelating affinity chromatography.  BRC bound to affinity resin was washed overnight at 135  4ºC in 20 column volumes of Buffer B + 0.03% LDAO, before elution in Buffer C + 0.03% LDAO.  The complex was further purified on a Superdex 200HR 10/300 GL in Buffer A + 0.03% LDAO, and stored in the dark at 4ºC before use in thermostability assays.  2A.2 Native gel electrophoresis   Equal volumes of 4% and 12% acrylamide solutions were prepared in advance (Supplementary Table 2). Linear gradient gels were formed by gradual mixing of the two solutions (35 mL each) at a flow rate of 2ml/min using a 100mL gradient mixer (Sigma). The cross-linking agents, TEMED and ammonium persulfate, were added immediately before gradient mixing.  Once poured, plastic wells (Biorad) were inserted and gels allowed to cure for 90 minutes before storage at 4°C.  For clear-native PAGE, anode and cathode buffers consisted of Buffer N (37mM Tris-HCl; 35mM Glycine; pH 8.8). For blue-native PAGE, anode buffer consisted of Buffer N + 180 µM Coomassie Blue G-250, and cathode buffer contained Buffer N only. 2A.3 Dynamic and static light scattering analysis  Aliquots of MalFGK2-NSP were analyzed by static light scatteringStatic light scattering analysis was performed using a WTC-050S5 column (Wyatt Technologies) connected to a miniDAWN light scattering detector and interferometry refractometer (Wyatt Technologies). Data were recorded in real time and the molecular masses were calculated using the Debye fit method using the ASTRA software (Wyatt Technology).   2A.4 Sample preparation and EM image acquisition MalFGK2-peptidisc sample (0.035 mg/mL) reconstituted by on-column method was applied onto negatively glow-discharged carbon-coated grids (400 mesh, copper grid) for 1 min, and excess liquid was removed by blotting with filter paper. Freshly prepared 1.5% uranyl formate (pH 5) was 136  added (5µl) for 1 min and then blotted. Around 200 digital micrographs were collected using a FEI Tecnai G2 F20 microscope operated at 200 kV and equipped with a Gatan Ultrascan 4k x 4k Digital CCD Camera. The images were recorded at defocus between 0.7 and 1.4µm at a magnification of 67,000X at the camera and a pixel size of 2.24 Å.  2A.5 EM data processing and image analysis Contrast transfer function parameters were determined using CTFFIND3 (Mindell and Grigorieff. 2008). We selected 31188 protein particles using e2boxer from the EMAN2 software suite (Tang et al. 2008) and extracted with a box size of 96x96 pixels. Particles were classified using a likelihood 2D classification with 16 seeds with the RELION-1.3 software suite (Scheres. 2012). A 2D variance of particles contained in side view (3175 particles) was computed with SPARX to estimate the NSP diameter variation (Hohn et al. 2007). The measurements were done using e2display from the EMAN2 software suite (Tang et al. 2008) on the side views shown in Figure 3-1C.  2A.6 Mass Spectrometry BRC peptidisc, MalFGK2 peptidisc, and FhuA peptidisc were prepared by “on-column” reconstitution in 100mM ammonium acetate, pH 7.0 at protein to NSP (g/g) ratios of 1:1.8, 1:1.6, and 1:1.8, respectively.  Mass spectrometry measurements were performed in positive ion mode on a Synapt G2S quadrupole-ion mobility separation-time-of-flight (Q-IMS-TOF) mass spectrometer (Waters, Manchester, UK) with a nanoflow electrospray ionization ESI (nanoESI) source. Borosilicate capillaries (1.0 mm o.d., 0.68 mm i.d.) were pulled in-house using a P-1000 micropipette puller (Sutter Instruments, Novato, CA). A voltage of ~1.0 kV was applied to a platinum wire was inserted into the nanoESI tip. A source temperature of 60 °C and a Cone voltage of 30 V were used. Argon was used in the Trap and Transfer ion guides, at pressures of 2.77 x 10-2 mbar and 2.84 x 10-2 mbar, respectively, and the Trap and Transfer voltages were 5 V and 2 V, respectively. All data was processed using MassLynx software (v4.1). Spectral deconvolution was performed with the UniDec (Marty et al. 2015) deconvolution algorithm using the following parameters: m/z range – 7000 to 9500 (MalFGK2 peptidisc), 5500 to 9000 (BRC peptidisc), 5000 137  to 10000 (FhuA peptidisc); Subtract minimum - 50.0; Gaussian Smoothing - 10.0; Bin every 1.0; Linear m/z (constant delta m/z); Charge Range - 20 to 40 (MalFGK2 peptidisc), 10 to 30 (BRC peptidisc), 10 to 30 (FhuA peptidisc); Mass range - 200,000 to 300,000 (MalFGK2 peptidisc),  100,000 to 180,000 (BRC peptidisc), 100,000 to 170,000 (FhuA peptidisc); Sample Mass Every 1.0 Da; Peak FWHM (Th) 4.0; Peak Shape Function - Gaussian; Charge Smooth Window - 1.0; Mass Difference - 4474.0; Mass Smooth Window - 1.0; Maximum number of iterations - 1000.  Spectral files were loaded as text files containing intensity and m/z values.    2A.7 Other methods The MalFGK2 ATPase activity was determined by monitoring the release of inorganic phosphate using the malachite green method. (Lanzetta et al, 1976)  Protein and peptide concentrations were determined by Bradford assay.[6]   SMA polymer containing 2:1 styrene to maleic acid ratio was prepared following the procedure described by Dorr et al. (Dorr et al, 2014)  In brief, 10% of SMA 2000 (Cray Valley), was refluxed for 3 hours at 80°C in 1M KOH, resulting in complete solubilization of the polymer.  Polymer was then precipitated by dropwise addition of 6M HCl accompanied by stirring and pelleted by centrifugation (1500g x 5 minutes).  The pellet was then washed 3 x with 50mL of 25mM HCl, followed by a third wash in ultrapure water and subsequent lyophilization.  Lyophilized SMA was later re-suspended at 10% wt/vol in 25mM Tris-HCl, and the pH of the solution adjusted to 8 with 1M NaOH.    Appendix 2B:  Supplementary Figures and Tables - Chapter 2  SUPPLEMENTAL FIGURE 2-1 138    Supplemental Figure 2-1 – Solubility test of NSP and NSPr.  A) Peptide models computed by the 3D-hydrophobic moment peptide calculator. The direction of hydrophobic moment is indicated by a red line.  Peptides are oriented with their N to C-terminus from bottom (red) to top (blue).  B)  Turbidity measurement of peptide suspension.  The absorbance of light at 550nm for NSP (blue squares, 15mg/mL) and NSPr (red circles, 25mg/mL), re-suspended in distilled water (dH2O) were compared to a dH2O control (green triangles).  C)  Calculated electropotential and hydrophobic moment of peptide variants.  Calculations were performed using the 3D-hydrophobic moment peptide calculator as described in material and methods.           SUPPLEMENTAL FIGURE 2-2   139    Supplemental Figure 2-2 – Quantification of NSPr in peptidiscs.  A) left panel; 15% SDS-PAGE analysis of MalFGK2 in peptidisc or DDM.  NSPr runs at the bottom of the gel and can be visualized with Coomassie blue staining.  Dye fluorescence was measured on a LICOR Odyssey scanner and quantified by Image J.  Right panel; Standard curve derived from NSPr titration measurement (black dots), and average intensity of NSPr fluorescence from MalFGK2 peptidisc (red dot).  B)  Left Panel: Western Blot of FhuA-peptidisc reconstituted into NSPrbio, and visualized by incubation with Streptavidin-Alexa 680.  Fluorescence of the Alexa 680 dye was measured on a LICOR Odyssey scanner (700nm, excitation 680nm) and quantified in Image J.  Right Panel: Standard curve as in A.  C) 15% SDS-PAGE analysis of BRC in peptidisc.  The MLH subunits of BRC partially resist denaturation by SDS, resulting in a higher molecular weight band located above the single subunits.  Each gel was repeated in triplicate with independent standard curves to calculate the values reported in Supplementary Table 1. 140     SUPPLEMENTAL FIGURE 2-3  Supplemental Figure 2-3 - Quantification of phospholipids trapped in peptidiscs.  A)  Calculated number of phospholipids per peptidisc.  Phospholipid content was determined by Malachite green assay after acid digestion of lipid extracts. Error bars represent standard deviation derived from three separate measurements.  B)  TLC analysis of lipid extracts obtained from 10µg MalFGK2 peptidisc, 10µg FhuA peptidiscs and 20µg BRC peptidiscs, as well as pure lipid standards Cardiolipin (CL), 1,2-dioleoyl-sn-glycero-3-phosphoglycerol (PG), and 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (PE).             SUPPLEMENTAL FIGURE 2-4 141    Supplemental Figure 2-4 - Native MS of intact peptidiscs.  Panels A, C, E are mass spectra acquired in positive ion mode for aqueous ammonium acetate solutions (100 mM, pH 7, 22 °C) of MalFGK2-NSPr, BRC-NSPr and FhuA-NSPr, respectively. Panels B, D, F are deconvoluted mass spectra of the peptidiscs shown in A, C, E, respectively    SUPPLEMENTAL FIGURE 2-5 142     Supplemental Figure 2-5 – Stability of MalFGK2 peptidisc.  A) Multi-angle light scattering analysis of MalFGK2 reconstituted in peptidisc. MalFGK2-NSPr (100µg) was left for 3 days at 4°C before analysis by SEC-MALS.  Protein sample was injected and protein concentration tracked through differential refractive interferometry (dRI, black trace). Molecular weight was calculated for the fractions corresponding to the peak of MalFGK2-NSP (red trace). B) Structural stability of MalFGK2-NSPr.  The MalFGK2 peptidisc was incubated at 30°C in Buffer A for the indicated time, then analyzed by BN-PAGE.      143  SUPPLEMENTAL FIGURE 2-6    Supplemental Figure 2-6 - Binding activity of FhuA in nanodiscs and peptidiscs.  A) Typical SEC profile of FhuA reconstituted in peptidisc (FhuA-NSPr) using an “on-column” reconstitution protocol as described in Material and Methods. B) The FhuA transporter reconstituted in nanodiscs (FhuA-MSPL156) or peptidiscs (FhuA-NSPr) was incubated with the C-terminal TonB23-329 fragment (2µg) or with colicin M (5µg), with or without ferricrocin as indicated.  Samples were analysed by CN-PAGE and Coomassie-blue staining of the gel.   144  SUPPLEMENTAL FIGURE 2-7    Supplemental Figure 2-7 - Capture of SecYEG monomer and dimer in peptidisc and nanodisc. A) Left panel; the SecYEG complex (2.5µg) was incubated with the indicated scaffolds proteins (1.25µg each) in Buffer A + 0.02% DDM. The sample was diluted 3-fold in detergent-free Buffer A and immediately analysed by CN-PAGE. Right panel; the experiment on left panel was performed without addition of SecYEG.  B) Illustration representing possible reconstitution products of the SecYEGn complex into 145  MSP1D1 and NSPr.  The subunits SecY, SecE, and SecG are represented as blue, pink, and magenta, respectively.   SUPPLEMENTAL FIGURE 2-8    Supplemental Figure 2-8 - Effect of peptidisc on BRC stability.  A) Absorbance scans of the BRC complex (1µM) in peptidisc after incubation at 65°C for up to 1 hour (blue traces). Incubation at 90°C leads to full release of the bacteriochlorophyll pigment (magenta trace). B) Absorbance scans of the BRC (1µM) after incubation at 65°C in 0.03% LDAO for up to 4 min. C) Fluorescence of the BRC in peptidisc (green trace), 0.1% LDAO (red trace), 0.02% DDM (blue trace), and 0.1% SDS (black trace). The BRC (1 µM) was incubated for 5 minutes at the indicated temperature before fluorescence was measured (700 nm; excitation at 680 nm). D)  The experiment repeated as in C), with BRC reconstituted into MSP1D1 (1:2 BRC:MSP1D1 molar ratio, brown trace), SMA (0.1%, grey trace), Proteoliposomes (1:1600:400 BRC:DOPC:DOPG), and 146  peptidiscs (green trace).  Fluorescence were normalized to 100% after denaturation for 5 minutes at 90 degrees.  E) Data were fitted using a Boltzmann sigmoidal function to calculate the melting temperature (Tm).  F)  Analysis of reconstituted BRC fractions on BN-PAGE (left panel) and SDS-PAGE (right panel)   SUPPLEMENTAL TABLE 2-1 Peptidisc Molecular Weight Measured by ESI-MS (kDa) Measured NSPr stoichiometry (NSPr/disc) Measured  Lipid stoichiometry (Lipid/Disc)  Calculated  Molecular Weight* (kDa) MalFGK2-NSPr 247 ± 24 10 (± 2) : 1 41 (± 10) : 1 251 ± 12 BRC-NSPr 138 ± 18 9 (± 1) : 1 4 (± 1) : 1 138 ± 5 FhuA-NSPr  137 ± 18 10 (± 2) : 1 8 (± 3) : 1 131 ± 9  Supplemental Table 2-1 - Molecular weight and scaffold stoichiometry of peptidiscs measured by native MS *The formula for the calculated molecular weight is as follows: MWpeptidisc = MW(protein) + n(MWNSPr) + m(MWLipid);  where n is the measured NSPr stoichiometry, m is the measured lipid stoichiometry, MWLipid = 0.8kDa, MWNSPr = 4.5kDa, and MWprotein = 173kDa, 80kDa, and 94kDa for MalFGK2, FhuA, and BRC, respectively.  For NSPr and Lipid stoichiometry, the standard deviation is derived from three separate measurements.    SUPPLEMENTAL TABLE 2-2  Measured Molecular Weight (kDa) a  Measured Diameter b Calculated Stoichiometry (scaffold/disc) c  MalFGK2 Peptidisc   250 ±17  11.7 ±1.4  12 (±2):1  Supplemental Table 2-2 - Molecular weight, diameter, and scaffold stoichiometry of MalFGK2 reconstituted in peptidisc. aMolecular weight calculated from SEC-MALS data (Fig. S1).  The standard error is derived from three independent SEC-MALS experiments. bDiameter of MalFGK2-peptidiscs determined by negative stain electron microscopy Fig. 1B, assuming a perfectly circular shape.   cStoichiometry (n), based on the measured diameter of the particles, was calculated with the following formula: π(ddisc-2dα-helix) = (n/2)LNSPr; where dα-helix represents the diameter of an alpha-helix (0.5nm), ddisc represents the measured disc diameter, and LNSPr represents length of the NSPr peptide. 147  SUPPLEMENTAL TABLE 2-3  Scaffold  Primary Sequence (N-ter to C-ter)  Length aa Å NSPr FAEKFKEAVKDYFAKFWDPAAEKLKEAVKDYFAKLWD  37 55.5 NSP DWLKAFYDKVAEKLKEAAPDWFKAFYDKVAEKFKEAF 37 55.5 NSPrbio Biotin-FAEKFKEAVKDYFAKFWDPAAEKLKEAVKDYFAKLWD  37 55.5 MSPL156 GHHHHHHHDYDIPTTENLYFQ//GSTFSKLREQLGPVTQ EFWDNLEKETEGLRQEMSKDLEEVKAKVQPYLDDFQK KWQEEMELYRQKVEPLRAELQEGARQKLHELQEKLSP LGEEMRDRARAHVDALRTHLAPYSDELRQRLAARLEA LKENGGAR  135 203 MSP1D1 GHHHHHHHDYDIPTTENLYFQ//GSTFSKLREQLGPVTQE FWDNLEKETEGLRQEMSKDLEEVKAKVQPYLDDFQKKW QEEMELYRQKVEPLRAELQEGARQKLHELQEKLSPLGEE MRDRARAHVDALRTHLAPYSDELRQRLAARLEALKENGG ARLAEYHAKATEHLSTLSEKAKPALEDLRQGLLPVLESFKV SFLSALEEYTKKLNTQ  190 285 MSP1D1E3 MGHHHHHHHDYDIPTTENLYFQ//GSTFSKLREQLGPVTQE FWDNLEKETEGLRQEMSKDLEEVKAKVQPYLDDFQKKW QEEMELYRQKVEPLRAELQEGARQKLHELQEKLSPLGEEM RDRARAHVDALRTHLAPYLDDFQKKWQEEMELYRQKVEP LRAELQEGARQKLHELQEKLSPLGEEMRDRARAHVDALRT HLAPYSDELRQRLAARLEALKENGGARLAEYHAKATEHLST LSEKAKPALEDLRQGLLPVLESFKVSFLSALEEYTKKLNTQ  256 384  Supplementary Table 2-3 - Amphipathic scaffolds used in this study.  The length of the scaffold proteins was calculated by multiplying the number of amino acids (aa) by 1.5Å, which is the rise given by an amino acid structured in an alpha-helix. For the MSPs scaffolds, the number of amino acids was from the TEV cleavage site (ENYLFQ//GXXX) to the C-terminus of the proteins.  148  SUPPLEMENTAL TABLE 2-4 Reagent 4% 12% Acrylamide (40%) 4.9 mL 14.6 mL Bis-Acrylamide (2%) 2.7 mL 8 mL 1.5M Tris-pH 8.8 12.5 mL 12.5 mL Glycerol (10%) 0 mL 10 mL dH2O Up to 50 mL Up to 50 mL APS (10%) 145 µL 58 µL TEMED 14.5 µL 5.8 µL  Supplementary Table 4 - Native Gel Buffer Recipes.                           149  Appendix 3A: Supplementary Figures and Tables – Chapter 3  SUPPLEMENTAL FIGURE 3-1  Supplementary Figure 3-1:   Fractionation profiles for select E. coli membrane protein complexes solubilized in SMA and peptidisc. Co-fractionation profiles for quantified subunits of the Bam complex (A), and respiratory chain complex (C) solubilized in SMALPs.  Co-fractionation profiles for quantified subunits of the Bam complex (B), and respiratory chain complex (D) solubilized in peptidiscs.  Average Pearson correlation and standard deviation between subunits is reported for the displayed complex.  * Due to incomplete chromatograms, NuoE and NuoE plus NuoF were omitted from calculations of the average correlation coefficient for SMA and peptidisc, respectively.        150  SUPPLEMENTAL FIGURE 3-2    Supplementary Figure 3-2: Purification of YfgM, PpiD, and MipA.  A)  Ni-NTA affinity pulldown of DDM solubilized crude membranes enriched for YfgM or PpiD, and empty vector control. The initial solubilized membrane (Start), first and final washes (Wash), and eluted protein (Elution) are analyzed by Coomassie stained SDS-PAGE.  (B) As in A, but using crude membranes enriched for MipA and corresponding empty vector control.                        151   SUPPLEMENTAL FIGURE 3-3    Supplementary Figure 3-3:  Co-elution of SecYEG with YfgMHis-PpiD. E. coli membranes containing HA-tagged SecYEG co-expressed with either YfgMHis ,PpiD His or YfgMHis-PpiD were solubilized in 1% DDM followed by pulldown using Ni-NTA resin. The his-tagged SecYEG complex was used as a pulldown control.  Eluted proteins were analyzed by 15% SDS-PAGE followed by either Coomassie staining or an immunoblot using an anti-SecY antibody.                 152   SUPPLEMENTAL TABLE 2-1   File Filename Contents a                                          Run_1_Peptidisc_CM_Replicates_54fxns.txt Protein enrichment in SEC fractionated peptidiscs (Biological Replicate 1) b                                          Run_2_Peptidisc_CM_Replicates_54fxns.txt Protein enrichment in SEC fractionated peptidiscs (Biological Replicate 2) c Run_1_SMA_CM_Replicates_35fxns.txt Protein enrichment of SEC fractionated SMALPs (Biological Replicate 1) d Run_1_SMA_CM_Replicates_66fxns.txt Protein enrichment of SEC fractionated SMALPs (Biological Replicate 2) e 50% precision binary interactions peptidisc PCP-SILAC.csv  Annotated, binary interaction list of membrane proteins identified in the peptidisc library by PCP-SILAC. f Complex assignment of 50% precision peptidisc binary interactions.csv   Predicted complexes from identified binary interactions in peptidisc library at 50% precision. g Enrichment Values for AP-MS  Enrichment of  proteins in AP-MS pulldowns of YfgM, PpiD, and MipA 153  Link to files  (https://osf.io/dhnm6/?view_only=538182ab0e4e4e3d8dffffc08f93dfcf)      Supplementary Table 3-1:  Co-fractionation data, interaction lists and assigned complex.  The full lists have been deposited and can be accessed at the link indicated below.   SUPPLEMENTAL TABLE 2-2   Gene ontology term (Associated with cell envelope) Anchored component of membrane Anchored component of external side of membrane Anchored component of periplasmic side of outer membrane Extrinsic component of periplasmic side of plasma membrane Gram-negative bacterium cell wall Extrinsic component of plasma membrane Integral component of membrane Cell envelope Cell outer membrane Integral component of cell outer membrane Integral component of plasma membrane Integral component of membrane membrane Cell wall External side of cell outer membrane Intrinsic component of cell outer membrane Intrinsic component  membrane Intrinsic component of plasma membrane Plasma membrane Extrinsic component of cell outer membrane Extrinsic component of membrane Intrinsic component of external side of  plasma membrane Intrinsic component of periplasmic side of plasma membrane Intrinsic component of periplasmic side of cell outer membrane Outer-membrane bounded periplasmic space Periplasmic space Plasma membrane  Intrinsic component of cytoplasmic side of plasma membrane Outer membrane  Periplasmic side of outer membrane Peptidoglycan based cell wall      Supplementary Table 3-2: List of GO terms used to predict protein association with the E. coli cell envelope.   

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0371907/manifest

Comment

Related Items