Elucidating the Gating Mechanism of Cys-Loop Receptors ÖZGE YOLUK Doctoral Thesis Stockholm, Sweden 2016 TRITA FYS 2016-26 ISSN 0280-316X KTH School of Engineering Sciences ISRN KTH/FYS/–16:26–SE SE-100 44 Stockholm ISBN 978-91-7729-009-4 SWEDEN Akademisk avhandling som med tillstånd av Kungl Tekniska högskolan framlägges till offentlig granskning för avläggande av teknologie doktorsexamen i biologisk fysik måndagen den 13 juni 2016 klockan 14.00 i F3, Lindstedtsvägen 26, KTH Campus, Kungl Tekniska högskolan, Stockholm. © Özge Yoluk, June 2016 Tryck: Universitetsservice US-AB iii Abstract Cys-loop receptors are membrane proteins that are key players for the fast synaptic neurotransmission. Their ion transport initiates new nerve signals after activation by small agonist molecules, but this function is also highly sensitive to allosteric modulation by a number of compounds such as anes- thetics, alcohol or anti-parasitic agents. For a long time, these modulators were believed to act primarily on the membrane, but the availability of high- resolution structures has made it possible to identify several binding sites in the transmembrane domains of the ion channels. It is known that lig- and binding in the extracellular domain causes a conformational earthquake that interacts with the transmembrane domain (and the allosteric modula- tor sites), which leads to channel opening. The investigations carried out in this thesis aim at understanding the connection between ligand binding and channel opening with molecular modeling and computer simulations. I present new models of the mammalian GABAA receptor based on the eukaryotic structure GluCl co-crystallized with an anti-parasitic agent, and show how these models can be used to study receptor-modulator interactions. I also show how removal of the bound modulator leads to gradual closing of the channel in molecular dynamics simulations. In contrast, simulations of the receptor with both the agonist and the modulator remain stable in an open-like conformation. This makes it possible to extract several key interac- tions, and I propose mechanisms for how the extracellular domain motion is initiated. The rapid increase in the number of cys-loop receptor structures the last few years has further made it possible to use principal component analysis (PCA) to create low-dimensional descriptions of the conformational land- scape. By performing PCA on the crystal structure ensemble, I have been able to divide the structures into functional clusters. Sampling of these clus- ters with molecular dynamics simulations reveals transitions that occupy the intermediate positions in the principal component space. The studies presented in this thesis contribute to our understanding of the gating mechanism and the functional clustering of the cys-loop recep- tor structures, which both are important to design new allosteric modulator drugs that influence the channel function, in particular to treat neurological disorders. Keywords: cys-loop receptor, ion channel, neuroscience, molecular dy- namics, activation, modulation, modeling, gating, GABAA iv Sammanfattning Cys-loop-receptorer är membranproteiner som spelar en nyckelroll för snabb synaptisk neurotransmission. Deras jontransport initierar nya nerv- signaler efter aktivering av små agonistmolekyler, men denna funktion är också mycket känslig för allosterisk modulering av ett antal föreningar såsom bedövningsmedel, alkohol eller antiparasitmedel. Under en lång tid troddes dessa modulatorer främst verka på membranet, men tillgången till högupp- lösta strukturer har gjort det möjligt att identifiera flera bindningsställen i transmembrandomänerna hos jonkanalerna. Det är känt att ligandbindning i den extracellulära domänen orsakar en jordbävningsliknande stor konfor- mationsförändring som interagerar med transmembrandomänen och regioner specifika för de allosteriska modulatorerna, vilket leder till att kanalen öpp- nas. De studier som presenteras i denna avhandling syftar till att förstå sam- bandet mellan ligandbindning och kanalöppning med molekylmodellering och datorsimuleringar. Jag presenterar nya modeller för GABAA-receptor för däggdjur som är baserade på en eukaryot struktur av GluCl vars struktur bestämts bunden till ett antiparasitmedel, och visar hur dessa modeller kan användas för att studera interaktioner mellan receptor och modulator. Jag visar också hur kanalen gradvis stänger i molekyldynamiksimuleringar om den bundna mo- dulatorn tas bort. I simuleringar med både agonist och modulator bundna förblir kanalen i en relativt öppen konformation. Detta gör det möjligt att identifera flera viktiga interaktioner, och jag föreslår mekanismer för hur den extracellulära domänens rörelse initieras. Den snabba ökningen av antalet cys-loop-receptorstrukturer de senas- te åren har vidare gjort det möjligt att använda principalkomponentanalys (PCA) för att skapa lågdimensionella beskrivningar av hela det tillgängli- ga rummet av strukturer. Genom att utföra PCA på en stor grupp av kri- stallstrukturer har jag kunnat dela in strukturerna i funktionella grupper. Molekyldynamiksimuleringar av dessa grupper gör det möjligt att identifiera övergångar som motsvarar mellantillstånd i rummet av principalkomponenter. De studier som presenteras i denna avhandling bidrar till vår förståelse av kanalöppningsmekanismen och den funktionella grupperingen av cys-loop receptorstrukturer. Denna kunskap är viktig för att kunna utforma nya läke- medel som genom allosterisk modulering påverkar funktionen hos proteinet, i synnerhet för att behandla neurologiska störningar. Nyckelord: cys-loop-receptor, jonkanal, neurovetenskap, molekyldynamik, aktivering, modulering, modellering, grind, GABAA “Feet, what do I need you for when I have wings to fly?” Frida Kahlo v List of Publications This thesis is based on following publications, which are referred to in the text by their roman numerals: Paper I Edward J Bertaccini, Özge Yoluk, Erik R Lindahl, and James R Trudell. Assessment of Homology Templates and an Anesthetic Binding Site within the γ-Aminobutyric Acid Receptor. Anesthesiology, 2013 vol. 119 (5) pp. 1087–1095. Paper II Özge Yoluk, Torben Broemstrup, Edward J Bertaccini, James R Trudell, and Erik Lindahl. Stabilization of the GluCl Ligand-Gated Ion Channel in the Presence and Absence of Ivermectin. Biophys. J., 2013 vol. 105 (3) pp. 640–647. Paper III Özge Yoluk, Erik Lindahl, and Magnus Andersson. Conformational Gating Dynamics in the GluCl Anion-Selective Chloride Channel. ACS Chem Neurosci. 2015 vol. 6 (8) pp. 1459–1467. Paper IV Stephanie A. Heusser, Özge Yoluk, Göran Klement, Erika A. Riederer, Erik Lindahl, Rebecca J. Howard. Functional Characterization of Neuro- transmitter Activation and Modulation in a Nematode Model Ligand-gated Ion Channel. J. Neurochem., in press. doi: 10.1111/jnc.13644 Paper V Laura Orellana, Özge Yoluk, Oliver Carrillo, Modesto Orozco and Erik Lindahl. Prediction and Validation of Protein Intermediate States from Struc- turally Rich Ensembles and Coarse-Grained Simulations. Revision submitted to Nat. Commun. Additional publications: Mieke Nys, Ana Farinha, Özge Yoluk, Magnus Andersson, Eveline Wijck- mans, Marijke Brams, Radovan Spurny, Erik Lindahl, Chris Ulens. A Novel Al- losteric Binding Site in a Cys-Loop Receptor Ligand-Binding Domain Unveiled in the Crystal Structure of ELIC in Complex with Chlorpromazine. Revision submit- ted to Proc. Natl. Acad. Sci. U.S.A. vii viii Author Contributions to the Papers For paper I E.B. came up with the idea, did most of the calculations, and most of the manuscript writing together with J.R.T and E.L.. Ö.Y. created and analyzed the Rosetta model. For paper II, E.L. developed the idea. Ö.Y. did all the calculations and the analysis. T.B. parametrised the ivermectin molecule. Ö.Y. and E.L. did most of the manuscript writing in collaboration with the other co-authors. Paper III was initiated by M.A. All the calculations and the analysis were performed by Ö.Y. in discussion with M.A. and E.L. Manuscript writing was done by all co-authors. The mutational analysis in Paper IV was designed by Ö.Y. based on the re- sults in paper III. Ö.Y. generated the mutated constructs. S.A.H. did the functional studies in oocytes, and the data analysis in collaboration with G.K. and R.J.H., ex- cept for the modulation data which was obtained by E.A.R and R.J.H. Manuscript writing was done mostly by S.A.H. and R.J.H. in collaboration with the author co-authors. The code for eBDIMS in Paper V was written by L.O. and O.C. L.O. came up with the idea, did most of the calculations and the manuscript writing. Ö.Y. helped with the ensemble preparations and analysis, ran some of the transition methods, and helped with figure preparations. All authors contributed to the manuscript writing and discussions. Contents Contents ix Abbreviations xi 1 Introduction 1 2 Molecular Neurobiology 5 2.1 Neurons . 5 2.1.1 Sensory Neurons . 7 2.1.2 Interneurons . 7 2.1.3 Motor Neurons . 7 2.2 Neuronal Activity . 8 2.2.1 Electrical synapses . 9 2.2.2 Chemical synapses . 10 2.3 The Ion Channel Superfamily . 11 3 Cys-Loop Receptors 13 3.1 The Discovery of Cys-Loop Receptors . 13 3.2 Channelopathies . 14 3.2.1 Hyperekplexia . 14 3.2.2 Epilepsy . 15 3.3 Anesthesia and Substance Abuse . 15 3.4 The First Structure . 16 3.5 Origins of Cys-Loop Receptors . 17 3.5.1 ELIC . 18 ix x CONTENTS 3.5.2 GLIC . 18 3.5.3 Are Prokaryotic Models Relevant? . 19 4 Modeling and Channel Dynamics in GluCl 21 4.1 Glutamate-Gated Chloride Channel . 21 4.1.1 Topology . 22 4.1.2 GluCl-based Models Capture Modulator Properties . 24 4.2 Channel Dynamics of GluCl . 26 4.2.1 Simulations of GluCl can Capture Closing Dynamics . 27 4.2.2 From Agonist Binding to Channel Opening . 30 4.3 From in silico to in vitro ......................... 33 5 A Gating Mechanism 35 5.1 Principal Component Analysis . 35 5.2 Gating Intermediates in the GLIC ensemble . 36 5.3 Functional Clusters of Eukaryotic Structures . 37 5.4 Cys-loop Receptor Dynamics Derived From Long-Scale Simulations . 39 6 Methods 45 6.1 Homology Modeling .