Although the basic structure of biological membranes is provided by the lipid bilayer, most of the specific functions are carried out by membrane proteins (MPs) such as channels, ion-pumps and receptors. Additionally, it is known, that mutations in MPs are directly or indirectly involved in many diseases. Thus, structure determination of MPs is of major interest not only in structural biology but also in pharmacology, especially for drug development. Advances in structural biology of membrane proteins (MPs) have been strongly supported by the success of three leading techniques: X-ray crystallography, electron microscopy and solution NMR spectroscopy. However, X-ray crystallography and electron microscopy, require highly diffracting 3D or 2D crystals, respectively. Today, structure determination of non-crystalline solid protein preparations has been made possible through rapid progress of solid-state MAS NMR methodology for biological systems. Castellani et. al. solved and refined the first structure of a microcrystalline protein using only solid-state MAS NMR spectroscopy. These successful application open up perspectives to access systems that are difficult to crystallise or that form large heterogeneous complexes and insoluble aggregates, for example ligands bound to a MP-receptor, protein fibrils and heterogeneous proteins aggregates. Solid-state MAS NMR spectroscopy is in principle well suited to study MP at atomic resolution. In this thesis, different types of MP preparations were tested for their suitability to be studied by solid-state MAS NMR. Proteoliposomes, poorly diffracting 2D crystals and a PEG precipitate of the outer membrane protein G (OmpG) were prepared as a model system for large MPs. Results from this work, combined with data found in the literature, show that highly diffracting crystalline material is not a prerequirement for structural analysis of MPs by solid-state MAS NMR. Instead, it is possible to use non-diffracting 3D crystals, MP precipitates, poorly diffracting 2D crystals and proteoliposomes. For the latter two types of preparations, the MP is reconstituted into a lipid bilayer, which thus allows the structural investigation in a quasi-native environment. In addition, to prepare a MP sample for solid-state MAS NMR it is possible to use screening methods, that are well established for 3D and 2D crystallisation of MPs. Hopefully, these findings will open a fourth method for structural investigation of MP. The prerequisite for structural studies by NMR in general, and the most time consuming step, is always the assignment of resonances to specific nuclei within the protein. Since the last few years an ever-increasing number of assignments from solid-state MAS NMR of uniformly carbon and nitrogen labelled samples is being reported, mostly for small proteins of up to around 150 amino acids in length. However, the complexity of the spectra increases with increasing molecular weight of the protein. Thus the conventional assignment strategies developed for small proteins do not yield a sufficiently high degree of assignment for the large MP OmpG (281 amino acids). Therefore, a new assignment strategy to find starting points for large MPs was devised. The assignment procedure is based on a sample with [2,3-13C, 15N]-labelled Tyr and Phe and uniformly labelled alanine and glycine. This labelling pattern reduces the spectral overlap as well as the number of assignment possibilities. In order to extend the assignment, four other specifically labelled OmpG samples were used. The assignment procedure starts with the identification of the spin systems of each labelled amino acid using 2D 13C-13C and 3D NCACX correlation experiments. In a second step, 2D and 3D NCOCX type experiments are used for the sequential assignment of the observed resonances to specific nuclei in the OmpG amino acid sequence. Additionally, it was shown in this work, that biosynthetically site directed labelled samples, which are normally used to observe long-range correlations, were helpful to confirm the assignment. Another approach to find assignment starting points in large protein systems, is the use of spectroscopic filtering techniques. A filtering block that selects methyl resonances was used to find further assignment starting points for OmpG. Combining all these techniques, it was possible to assign nearly 50 % of the observed signals to the OmpG sequence. Using this information, a prediction of the secondary structure elements of OmpG was possible. Most of the calculated motifs were in good aggreement with the crystal structures of OmpG. The approaches presented here should be applicable to a wide variety of MPs and MP-complexes and should thus open a new avenue for the structural biology of MPs. / Biologische Membranen bestehen hauptsächlich aus Lipiden, ihre Funktion wird jedoch vor allem durch die eingebetteten Membranproteine (z.B. Kanäle, Ionenpumpen und Rezeptoren) bestimmt. Mutationen in dieser Proteinklasse können zum Auftreten verschiedener Krankheitsbilder führen, weshalb die Untersuchung der dreidimensionalen Struktur von Membranproteinen nicht nur von strukturbiologischem, sondern auch von pharmakologischem Interesse ist. In den letzten Jahren wurde eine Methode, die Festkörper NMR Spektroskopie, für Strukturuntersuchungen an Proteinproben im festen Aggregatzustand entwickelt. Diese Arbeit beschäftigt sich mit drei verschiedenen Präparationsarten von Membranproteinen, die eine Aufnahme von hochaufgelösten Festkörper NMR Spektren erlauben. Als Modelsystem wurde das Protein G der äußeren Membrane (outer membrane protein G, OmpG) von Escherichia coli gewählt. Eine wichtige Vorraussetzung zur Berechnung der Proteinstruktur aus den NMR-Spektren, ist die Zuordnung der einzelnen Signale zur jeweiligen Aminosäure in der Proteinsequenz. In dieser Arbeit wurde eine Methode entwickelt, die das Auffinden von Startpunkten für die sequentielle Zuordnung in großen Membranproteinen, wie zum Bsp. OmpG (281 Aminosäuren), erlaubt. Multidimensionale NMR Experimente mit verschieden spezifisch markierten Proben wurden durchgeführt und ermöglichten die Zuordnung von 50 % der NMR Signale der OmpG Proteinsequenz. Zur Überprüfung der gewonnenen Daten wurden diese zur Vorhersage von Sekundärstrukturelementen genutzt. Es konnte gezeigt werden, dass die berechneten Strukturmotive in guter Übereinstimmung zu den bisher veröffentlichten OmpG Strukturen liegen. Die in dieser Arbeit angewendeten Methoden sollten auf eine Vielzahl anderer Membranprotein anwendbar und somit einen neuen Weg zur Strukturbiologischen Untersuchung von Membranproteinen eröffnen.

Single-molecule force spectroscopy (SMFS) is a unique approach to study the mechanical unfolding of proteins. SMFS unfolding experiments yield insight into how interactions stabilize a protein and guide its unfolding and refolding pathways. In contrast to various water-soluble proteins whose unfolding and refolding patterns have been characterized, only α-helical membrane proteins have been probed by SMFS. It was shown that α-helical membrane proteins unfold via many intermediates; this differs from the two-state unfolding process usually observed in water-soluble proteins. In membrane proteins, upon mechanically pulling the peptide end of the protein, single and grouped α-helices and polypeptide loops unfold in steps until the entire protein is unfolded. Whether the α-helices and loops unfold individually or cooperatively to form an unfolding intermediate depends on the interactions established within the membrane protein and the membrane. Each unfolding event relates to an unfolding intermediate with the sequence of these intermediates defining the unfolding pathway of the protein. β-barrel-forming membrane proteins are the second major group of membrane proteins and have not yet been studied by SMFS. To fill this void this study was designed to characterize interactions, unfolding, and refolding of the β-barrel forming outermembrane protein G (OmpG).Folding of transmembrane proteins, despite the important part these proteins play in every biological process in a cell, is studied in only a few examples. Of those only a handful were β-stranded membrane proteins (Tamm et al., 2004; Kleinschmidt et al., 2006). Current models describe that transmembrane β-barrels fold into the lipid membrane via two major steps. First the unfolded polypeptide interacts with the lipid surface where it then folds and inserts into the membrane (Kleinschmidt et al., 2006; Huysmans et al., 2010). Conventionally, thermal or chemical denaturation is used to study folding of membrane proteins. In most cases membrane proteins were solubilized in detergent or exposed to urea to be studied, conditions that are not compatible with In vivo conditions. This suggests that the folding pathways described so far may not be a realistic representation of such pathways in nature. SMFS represents a unique approach to study the unfolding and refolding of membrane proteins into the lipid membrane (Kedrov et al., 2006; Kessler et al., 2006). Using SMFS makes it possible to study unfolding and refolding of membrane proteins in their nativephysiological environment with controlled pH, electrolyte, temperature, and most importantly in the absence of any chemical denaturant or detergent. In this thesis, SMFS was utilized to unfold and refold OmpG in E coli lipid extract. Bulk unfolding experiments suggested that OmpG unfolds and folds reversibly and much faster than α-helical proteins (Conlan et al., 2000). The folding process is thought to be a coupled two-state membrane partition-folding reaction. To the contrary, the mechanical unfolding of OmpG consisted of many sequential unfolding intermediates. Our SMFS refolding experiments showed that a partially unfolded OmpG molecule also refolds via several sequential steps. The predominant refolding steps are defined by individual β-hairpins that could later assemble the transmembrane β-barrel of OmpG. In conclusion, the most probable unfolding and refolding pathways of OmpG as a membrane β-barrel protein go through the β-hairpins as the structural segments or unfolding-refolding intermediates and the process is a multi step one rather than the simple two state process. We also used SMFS to study the physical interactions that switch the functional state and gating of OmpG. The structural changes that gate OmpG have been previously described by X-ray crystallography (Yildiz et al., 2006). They showed when the pH changes from neutral to acidic the flexible extracellular loop L6 folds into the pore and closes the OmpG pore. Here, SMFS was used to structurally localize and quantify the interactions that are associated with the pH-dependent closure. At an acidic pH, a pH-dependent interaction at loop L6 was detected. This interaction changed the unfolding of loop L6 and β-strands 11 and 12, which connect loop L6. All other interactions detected within OmpG were found to be unaffected by changes in pH. These results provide a quantitative and mechanistic explanation of how pHdependent interactions change the folding of a peptide loop to gate the transmembrane pore. It has also been shown how the stability of OmpG is optimized so that pH changes modify only those interactions necessary to gate the transmembrane pore and there are no global changes in protein conformation or mechanical properties. In the next step of interactions study, dynamic SMFS (DFS) was applied to quantify the parameters characterizing the energy barriers in energy landscape for unfolding of the OmpG. Some of these parameters are: free energy of activation and distance of the transition state from the folded state. The pH-dependent functional switching of OmpG directs the protein along different regions at the unfolding energy landscape. The two functional states of OmpG sequential folding take the same unfolding pathway as β-hairpins I–IV. After the initial unfolding events, the unfolding pathways diverge. In the open state, the unfolding of β-hairpin V in one step precedes the unfolding of β-hairpin VI. In the closed state, β-hairpin V and β-strand S11 with a part of extracellular loop L6 unfold cooperatively, and subsequently β-strand S12 unfolds with the remaining loop L6. These two unfolding pathways in the open and closed states join again in the last unfolding step of β-hairpin VII. Also, the conformational change from the open to the closed state witnesses a difference in Xu and κ in the energy landscape that translates to rigidified extracellular loop L6 at the gating area. Thus, a change in the conformational state of OmpG not only bifurcates its unfolding pathways but also tunes its mechanical properties for optimum function.

Single-molecule force spectroscopy (SMFS) is a unique approach to study the mechanical unfolding of proteins. SMFS unfolding experiments yield insight into how interactions stabilize a protein and guide its unfolding and refolding pathways. In contrast to various water-soluble proteins whose unfolding and refolding patterns have been characterized, only α-helical membrane proteins have been probed by SMFS. It was shown that α-helical membrane proteins unfold via many intermediates; this differs from the two-state unfolding process usually observed in water-soluble proteins. In membrane proteins, upon mechanically pulling the peptide end of the protein, single and grouped α-helices and polypeptide loops unfold in steps until the entire protein is unfolded. Whether the α-helices and loops unfold individually or cooperatively to form an unfolding intermediate depends on the interactions established within the membrane protein and the membrane. Each unfolding event relates to an unfolding intermediate with the sequence of these intermediates defining the unfolding pathway of the protein. β-barrel-forming membrane proteins are the second major group of membrane proteins and have not yet been studied by SMFS. To fill this void this study was designed to characterize interactions, unfolding, and refolding of the β-barrel forming outermembrane protein G (OmpG).Folding of transmembrane proteins, despite the important part these proteins play in every biological process in a cell, is studied in only a few examples. Of those only a handful were β-stranded membrane proteins (Tamm et al., 2004; Kleinschmidt et al., 2006). Current models describe that transmembrane β-barrels fold into the lipid membrane via two major steps. First the unfolded polypeptide interacts with the lipid surface where it then folds and inserts into the membrane (Kleinschmidt et al., 2006; Huysmans et al., 2010). Conventionally, thermal or chemical denaturation is used to study folding of membrane proteins. In most cases membrane proteins were solubilized in detergent or exposed to urea to be studied, conditions that are not compatible with In vivo conditions. This suggests that the folding pathways described so far may not be a realistic representation of such pathways in nature. SMFS represents a unique approach to study the unfolding and refolding of membrane proteins into the lipid membrane (Kedrov et al., 2006; Kessler et al., 2006). Using SMFS makes it possible to study unfolding and refolding of membrane proteins in their nativephysiological environment with controlled pH, electrolyte, temperature, and most importantly in the absence of any chemical denaturant or detergent. In this thesis, SMFS was utilized to unfold and refold OmpG in E coli lipid extract. Bulk unfolding experiments suggested that OmpG unfolds and folds reversibly and much faster than α-helical proteins (Conlan et al., 2000). The folding process is thought to be a coupled two-state membrane partition-folding reaction. To the contrary, the mechanical unfolding of OmpG consisted of many sequential unfolding intermediates. Our SMFS refolding experiments showed that a partially unfolded OmpG molecule also refolds via several sequential steps. The predominant refolding steps are defined by individual β-hairpins that could later assemble the transmembrane β-barrel of OmpG. In conclusion, the most probable unfolding and refolding pathways of OmpG as a membrane β-barrel protein go through the β-hairpins as the structural segments or unfolding-refolding intermediates and the process is a multi step one rather than the simple two state process. We also used SMFS to study the physical interactions that switch the functional state and gating of OmpG. The structural changes that gate OmpG have been previously described by X-ray crystallography (Yildiz et al., 2006). They showed when the pH changes from neutral to acidic the flexible extracellular loop L6 folds into the pore and closes the OmpG pore. Here, SMFS was used to structurally localize and quantify the interactions that are associated with the pH-dependent closure. At an acidic pH, a pH-dependent interaction at loop L6 was detected. This interaction changed the unfolding of loop L6 and β-strands 11 and 12, which connect loop L6. All other interactions detected within OmpG were found to be unaffected by changes in pH. These results provide a quantitative and mechanistic explanation of how pHdependent interactions change the folding of a peptide loop to gate the transmembrane pore. It has also been shown how the stability of OmpG is optimized so that pH changes modify only those interactions necessary to gate the transmembrane pore and there are no global changes in protein conformation or mechanical properties. In the next step of interactions study, dynamic SMFS (DFS) was applied to quantify the parameters characterizing the energy barriers in energy landscape for unfolding of the OmpG. Some of these parameters are: free energy of activation and distance of the transition state from the folded state. The pH-dependent functional switching of OmpG directs the protein along different regions at the unfolding energy landscape. The two functional states of OmpG sequential folding take the same unfolding pathway as β-hairpins I–IV. After the initial unfolding events, the unfolding pathways diverge. In the open state, the unfolding of β-hairpin V in one step precedes the unfolding of β-hairpin VI. In the closed state, β-hairpin V and β-strand S11 with a part of extracellular loop L6 unfold cooperatively, and subsequently β-strand S12 unfolds with the remaining loop L6. These two unfolding pathways in the open and closed states join again in the last unfolding step of β-hairpin VII. Also, the conformational change from the open to the closed state witnesses a difference in Xu and κ in the energy landscape that translates to rigidified extracellular loop L6 at the gating area. Thus, a change in the conformational state of OmpG not only bifurcates its unfolding pathways but also tunes its mechanical properties for optimum function.:Table of Contents INTRODUCTION:1 1.1 THE FIRST UNIT OF LIFE STARTED WITH MEMBRANE:1 1.2.1 CELL MEMBRANE STRUCTURE: 2 1.3 MEMBRANE PROTEINS:3 1.3.1 α-­‐HELICAL MEMBRANE PROTEINS:5 1.3.2 β-­‐BARREL MEMBRANE PROTEIN:5 1.4 MEMBRANE PROTEINS FOLDING:12 1.4.1 MODELS FOR α-­‐HELICAL MEMBRANE PROTEIN FOLDING:13 1.4.2 MODELS FOR β-­‐BARREL MEMBRANE PROTEIN FOLDING:15 1.5. GATING STUDY OF MEMBRANE PROTEINS:18 ATOMIC FORCE MICROSCOPY:19 2.1 ATOMIC FORCE MICROSCOPE:19 2.1.1 HISTORY:19 2.1.2 PRINCIPLE:19 2.1.3 THE CANTILEVER:20 2.1.4 AFM MODES 23 2.2 SINGLE-­‐MOLECULE FORCE SPECTROSCOPY:25 2.2.1 DYNAMIC FORCE SPECTROSCOPY,(DYNAMIC SMFS):27 2.3 WHAT IS THE ADVANTAGE OF USING ATOMIC FORCE MICROSCOPY IN MEMBRANE PROTEIN STUDIES?:29 FOLDING MECHANISM OF OMPG:31 3.1 UNFOLDING PATTERN: ONEβ-­‐HAIRPIN AFTER THE OTHER:31 3.1.1 OUTER MEMBRANE PROTEIN G (OMPG):31 3.1.2 MECHANICAL UNFOLDING PATHWAYS OF THE MEMBRANE β-­‐BARREL PROTEIN OMPG:33 3.1.3 MATERIAL AND METHODS:34 3.1.4 RESULTS AND DISCUSSION:41 3.2 REFOLDING PATTERN: ONE Β-­‐HAIRPIN AFTER THE OTHER:48 3.2.1. EXPLORING REFOLDING PATHWAYS AND KINETICS OF THE MEMBRANE Β-­‐BARREL PROTEIN OMPG:48 3.2.2 EXPERIMENTAL PROCEDURES:49 3.2.3 RESULTS:50 3.2.4 DISCUSSION:52 INTERACTION STUDIES:59 4.1 PH-­‐DEPENDENT INTERACTIONS GUIDE THE FOLDING AND GATE THE TRANSMEMBRANE PORE OF THE β-­‐BARREL TRANSMEMBRANE PROTEIN OMPG:59 4.1.2 INTRODUCTION:59 4.1.2 EXPERIMENTAL PROCEDURES:61 4.1.3 RESULTS AND DISCUSSION:62 4.2 DUAL ENERGY LANDSCAPE: THE FUNCTIONAL STATE OF THE OUTER MEMBRANE β-­‐BARREL PROTEIN OMPG MOLDS ITS UNFOLDING ENERGY LANDSCAPE:67 4.2.1 INTRODUCTION:67 4.2.2 EXPERIMENTAL PROCEDURES:71 4.2.3 RESULTS AND DISCUSSION:74 4.2.3.1 FUNCTIONAL STATE OF OMPG DIRECTS ITS UNFOLDING ROUTE:74 4.2.3.2 QUANTIFYING THE UNFOLDING ENERGY BARRIERS OF OMPG IN THE CLOSED AND OPEN CONFORMATIONS:75 4.2.3.3 TRANSITION STATE DISTANCES OF UNFOLDING ENERGY BARRIERS:77 4.2.3.4 ACTIVATION FREE ENERGY OF Β-­‐STRANDS AND Β-­‐HAIRPINS:79 4.2.3.5 MECHANICAL PROPERTIES OF OMPG:83 4.2.3.6 MAPPING THE UNFOLDING ENERGY LANDSCAPES OF OMPG IN THE OPEN AND CLOSED STATES:85 4.2.4 CONCLUSION:86 OUTLOOK:89 5.1 INTRODUCTION:89 5.2 INTERACTION STUDY AND UNFOLDING ENERGY LANDSCAPE:90 5.3 MEMBRANE PROTEINF OLDING:92 REFRENCES:96 ABBREVIATIONS:110 SYMBOLS:111 PUBLICATIONS:113 ACKNOWLEDGMENT:114 DECLARATION: 115