Membrane proteins are of undeniable importance for cell physiology across all domains of life and a loss of their function, e.g., due to mutations in their coding sequence, is almost always linked to disease. In humans, mutations in the gene coding for the cystic fibrosis transmembrane conductance regulator (CFTR), an ATP-gated anion channel in epithelia, give rise to cystic fibrosis (CF). Over 2100 mutations of the CFTR gene are known, however, their disease liability remains mostly undetermined. Causal therapies, i.e., small-molecule drugs that target CFTR itself, have improved the lives of people with the most common mutations (e.g. ΔF508, G551D) over the last decade. In contrast, many rare CF-phenotypic mutations are not eligible for these novel treatments and would benefit from in vitro evaluation of their molecular consequences. In vitro studies of membrane proteins are often complicated by the intrinsic hydrophobicity and aggregation susceptibility of this protein group. However, this can be avoided by using short membrane protein fragments corresponding to the smallest in vivo folding unit of the respective protein at the ER membrane. These model proteins can be easily genetically modified, expressed and purified, making them a suitable tool to pinpoint the effects of mutations. This thesis demonstrates the utility of such a reductionist model system: TM3/4, the second helical hairpin of CFTR’s transmembrane domain 1, was used to study protein folding with a focus on disease-causing missense mutations of CFTR, which may cause CFTR misfolding in vivo. TM3/4 purification was first optimized by using a thioredoxin tag, which allowed heat purification of the fusion protein even after initial purification steps. Optimal heat treatment for maximal protein purity and recovery were determined for TM3/4 and another helical hairpin, ATP synthase subunit c. Moreover, tertiary folding of a CF-phenotypic loop mutation, E217G, introducing a non-native GXXXG interaction motif was analyzed by single-molecule Förster resonance energy transfer (smFRET) in different lipid bilayer conditions, showing unusually increased stability in comparison to wild type (WT) TM3/4. Furthermore, smFRET was used in tandem with circular dichroism and fluorescence spectroscopy to assess the effect of a specific membrane lipid, cholesterol, on TM3/4 variants showing significant changes on secondary but not tertiary structure. Lastly, a mutant library of 13 TM3/4 mutants was established to perform drug screenings with CFTR correctors – a class of small molecules rescuing or preventing misfolding of CFTR. This screening study demonstrated that (i) not all CF-phenotypic missense mutations are locally misfolded at a lipid bilayer comparable to the ER membrane; and (ii) in vitro restoration of a native WT-like conformation of locally misfolded TM3/4 mutants is not only possible but different drug-mutant pairings can be identified related to folding rescue efficiency of a given corrector on a respective mutant. The latter identified drug-mutant pairings may lead to drug repurposing if the effect can be confirmed in cell culture experiments. In conclusion, the TM3/4 minimal model of CFTR and biophysical methods, such as smFRET, proved as versatile tools not only for investigation of mutation and lipid effects on membrane protein folding but also for drug screenings in a disease context.:1 INTRODUCTION 2 THEORETICAL BACKGROUND 2.1 MEMBRANE PROTEINS AND THEIR NATIVE ENVIRONMENTS 2.1.1 Membrane protein families and their role in human health 2.1.2 Fundamental folding models of α-helical membrane proteins 2.1.3 Co-translational folding at the ER supported by the translocon 2.1.4 Folding-relevant interactions within membrane proteins 2.1.5 Biological membranes and lipid classes 2.1.6 Physical properties of lipid bilayers impacting membrane proteins 2.1.7 Membrane models for in vitro studies 2.2 CYSTIC FIBROSIS AND CFTR 2.2.1 Pathology of cystic fibrosis 2.2.2 Structure and function of the CFTR channel 2.2.3 A minimal model of CFTR to study rare CF mutations 2.2.4 Missense mutations within the CFTR segmental model TM3/4 2.2.5 Novel modulator therapies for the treatment of cystic fibrosis 2.3 IN VITRO ASSESSMENT OF MEMBRANE PROTEIN FOLDING 2.3.1 Expression and purification of membrane proteins 2.3.2 Single-molecule FRET in single- and multi-well mode for protein folding 3 HEAT PURIFICATION OF TRX MEMBRANE PROTEIN FUSIONS 3.1 PREAMBLE AND SUMMARY 3.2 RESULTS AND DISCUSSION 4 IMPACT OF A CFTR LOOP MUTATION WITH ATYPICAL STABILITY 4.1 PREAMBLE AND SUMMARY 4.2 RESULTS AND DISCUSSION 5 EFFECTS OF CHOLESTEROL ON LOCAL CFTR FOLDING 5.1 PREAMBLE AND SUMMARY 5.2 RESULTS 5.2.1 Folding of TM3/4 hairpins in the presence of cholesterol 5.2.2 Folding of TM3/4 hairpins in the presence of Lumacaftor 5.2.3 Impact of Lumacaftor on membrane fluidity 5.3 DISCUSSION 6 CFTR CORRECTOR SCREENINGS WITH SINGLE-MOLECULE FRET 6.1 PRESCREENING TO IDENTIFY MISFOLDED TM3/4 VARIANTS 6.2 SCREENING OF MISFOLDED TM3/4 VARIANTS WITH CFTR CORRECTORS 7 CONCLUSIONS 8 OUTLOOK 9 MATERIALS AND METHODS 9.1 CONSTRUCT DESIGN OF HELICAL TRANSMEMBRANE HAIRPINS 9.2 PROTEIN EXPRESSION AND PURIFICATION 9.3 HEAT TREATMENT OF HELICAL TRANSMEMBRANE CONSTRUCTS 9.4 SINGLE-MOLECULE FRET EXPERIMENTS 9.4.1 Labeling of TM3/4 constructs 9.4.2 Liposome preparation and reconstitution of labeled protein constructs 9.4.3 Single-molecule FRET measurements in manual mode 9.4.4 Single-molecule FRET measurements in multi-well screening mode 9.5 CIRCULAR DICHROISM SPECTROSCOPY 9.5.1 Circular dichroism to determine protein heat stability 9.5.2 Circular dichroism to study protein structure in different lipid bilayers 9.6 FLUORESCENCE SPECTROSCOPY 9.6.1 Vesicle leakage assay to test lipid bilayer stability 9.6.2 Examining lipid bilayer fluidity with fluorescent probes 10 APPENDIX 10.1 GENERATION OF A TM3/4 MUTANT LIBRARY 10.2 TM3/4 SCREENINGS WITH CFTR CORRECTORS 10.2.1 SmFRET control screenings and supporting data 10.2.2 Extracted closed state fractions from smFRET screenings 10.2.3 DLS to measure vesicle integrity after corrector addition 11 REFERENCES 12 ACKNOWLEDGEMENTS 13 ERKLÄRUNG GEMÄß §5 ABS. 1 S. 3 DER PROMOTIONSORDNUNG / Membranproteine sind für die Zellphysiologie aller biologischen Domänen von unbestreitbarer Bedeutung und ein Verlust ihrer Funktion, z.B. durch Mutationen in ihrer kodierenden Sequenz, ist fast immer Auslöser von Krankheiten. Beim Menschen führen Mutationen im Gen für den Cystic Fibrosis Transmembrane Conductance Regulator (CFTR), einen ATP-abhängigen Anionenkanal in Epithelien, zu Mukoviszidose (CF). Über 2100 Mutationen des CFTR-Gens sind bekannt – ob jedoch alle Mutationen tatsächlich CF auslösen, ist weitgehend ungeklärt. Kausale Therapien, d.h. niedermolekulare Medikamente, die auf CFTR selbst abzielen, haben in den letzten zehn Jahren die Lebensqualität von Menschen mit den häufigsten Mutationen (z.B. ΔF508, G551D) verbessert. Demgegenüber stehen jedoch viele seltene CF-phänotypische Mutationen, für welche diese neuartigen Behandlungen nicht zugelassen sind, wodurch diese Mutationen von einer In-vitro-Analyse ihrer molekularen Konsequenzen profitieren würden. In-vitro-Untersuchungen von Membranproteinen werden oft durch die intrinsische Hydrophobizität und Aggregationsanfälligkeit dieser Proteine erschwert. Dies kann jedoch vermieden werden, indem kurze Membranproteinfragmente verwendet werden, die der kleinsten in vivo Faltungseinheit des jeweiligen Proteins an der ER-Membran entsprechen. Diese Modellproteine können routiniert genetisch verändert, exprimiert und aufgereinigt werden, was sie zu einem geeigneten Werkzeug macht, um die Auswirkungen von Mutationen zu genau festzustellen. Diese Dissertation demonstriert die Nützlichkeit eines solchen reduktionistischen Modellsystems: TM3/4, das zweite helikale Haarnadel-Motiv der Transmembrandomäne 1 von CFTR, wurde verwendet, um Proteinfaltung mit Schwerpunkt auf krankheitsverursachende Missense-Mutationen von CFTR zu untersuchen, welche eine CFTR-Fehlfaltung in vivo verursachen können. Die TM3/4-Aufreinigung wurde zunächst durch die Verwendung eines Thioredoxin-Tags optimiert, der eine Hitzeaufreinigung des Fusionsproteins auch nach anfänglichen Reinigungsschritten ermöglichte. Die optimale Hitzebehandlung für maximale Proteinreinheit und -ausbeute wurde für TM3/4 und ein weiteres helikales Haarnadelprotein, die ATP-Synthase-Untereinheit c, bestimmt. Weiterhin wurde die tertiäre Faltung einer CF-phänotypischen Mutation, E217G, die ein nicht-natives GXXXG-Interaktionsmotiv einführt, mittels einzelmolekularem Förster-Resonanzenergietransfer (smFRET) in verschiedenen Lipiddoppelschichten analysiert, welche eine ungewöhnlich erhöhte Stabilität im Vergleich zum TM3/4-Wildtyp (WT) zeigte. Darüber hinaus wurde smFRET in Verbindung mit Circulardichroismus und Fluoreszenzspektroskopie verwendet, um die Wirkung eines spezifischen Membranlipids, Cholesterin, auf TM3/4-Varianten zu untersuchen, welches signifikante Auswirkungen auf die sekundäre, aber nicht auf die tertiäre Proteinstruktur hatte. Schließlich wurde eine Mutantenbibliothek von 13 TM3/4-Mutanten eingerichtet, um Wirkstoffscreenings mit CFTR-Korrektoren durchzuführen – einer Klasse kleiner Moleküle, die die Fehlfaltung von CFTR verhindern können. Diese Screening-Studie zeigte, dass (i) nicht alle CF-phänotypischen Missense-Mutationen lokal an einer Lipiddoppelschicht fehlgefaltet sind, die mit der ER-Membran vergleichbar ist; und (ii) die In-vitro-Wiederherstellung einer nativen WT-ähnlichen Konformation von lokal fehlgefalteten TM3/4-Mutanten ist nicht nur möglich, sondern es können auch verschiedene Wirkstoff-Mutanten-Paare identifiziert werden, die mit der Faltungsrettungseffizienz eines Korrektors auf eine bestimmte Mutante zusammenhängen. Die letztgenannten Wirkstoff-Mutanten-Paare können zu Drug-Repurposings führen, wenn die Wirkung in Zellkulturexperimenten bestätigt werden kann. Im Allgemeinen, haben sich das TM3/4-Minimalfaltungsmodell von CFTR sowie biophysikalische Methoden, wie z.B. smFRET, als vielseitige Werkzeuge nicht nur für die Untersuchung von Mutations- und Lipideffekten auf die Membranproteinfaltung, sondern auch für das Screening von Medikamenten im Krankheitskontext erwiesen.:1 INTRODUCTION 2 THEORETICAL BACKGROUND 2.1 MEMBRANE PROTEINS AND THEIR NATIVE ENVIRONMENTS 2.1.1 Membrane protein families and their role in human health 2.1.2 Fundamental folding models of α-helical membrane proteins 2.1.3 Co-translational folding at the ER supported by the translocon 2.1.4 Folding-relevant interactions within membrane proteins 2.1.5 Biological membranes and lipid classes 2.1.6 Physical properties of lipid bilayers impacting membrane proteins 2.1.7 Membrane models for in vitro studies 2.2 CYSTIC FIBROSIS AND CFTR 2.2.1 Pathology of cystic fibrosis 2.2.2 Structure and function of the CFTR channel 2.2.3 A minimal model of CFTR to study rare CF mutations 2.2.4 Missense mutations within the CFTR segmental model TM3/4 2.2.5 Novel modulator therapies for the treatment of cystic fibrosis 2.3 IN VITRO ASSESSMENT OF MEMBRANE PROTEIN FOLDING 2.3.1 Expression and purification of membrane proteins 2.3.2 Single-molecule FRET in single- and multi-well mode for protein folding 3 HEAT PURIFICATION OF TRX MEMBRANE PROTEIN FUSIONS 3.1 PREAMBLE AND SUMMARY 3.2 RESULTS AND DISCUSSION 4 IMPACT OF A CFTR LOOP MUTATION WITH ATYPICAL STABILITY 4.1 PREAMBLE AND SUMMARY 4.2 RESULTS AND DISCUSSION 5 EFFECTS OF CHOLESTEROL ON LOCAL CFTR FOLDING 5.1 PREAMBLE AND SUMMARY 5.2 RESULTS 5.2.1 Folding of TM3/4 hairpins in the presence of cholesterol 5.2.2 Folding of TM3/4 hairpins in the presence of Lumacaftor 5.2.3 Impact of Lumacaftor on membrane fluidity 5.3 DISCUSSION 6 CFTR CORRECTOR SCREENINGS WITH SINGLE-MOLECULE FRET 6.1 PRESCREENING TO IDENTIFY MISFOLDED TM3/4 VARIANTS 6.2 SCREENING OF MISFOLDED TM3/4 VARIANTS WITH CFTR CORRECTORS 7 CONCLUSIONS 8 OUTLOOK 9 MATERIALS AND METHODS 9.1 CONSTRUCT DESIGN OF HELICAL TRANSMEMBRANE HAIRPINS 9.2 PROTEIN EXPRESSION AND PURIFICATION 9.3 HEAT TREATMENT OF HELICAL TRANSMEMBRANE CONSTRUCTS 9.4 SINGLE-MOLECULE FRET EXPERIMENTS 9.4.1 Labeling of TM3/4 constructs 9.4.2 Liposome preparation and reconstitution of labeled protein constructs 9.4.3 Single-molecule FRET measurements in manual mode 9.4.4 Single-molecule FRET measurements in multi-well screening mode 9.5 CIRCULAR DICHROISM SPECTROSCOPY 9.5.1 Circular dichroism to determine protein heat stability 9.5.2 Circular dichroism to study protein structure in different lipid bilayers 9.6 FLUORESCENCE SPECTROSCOPY 9.6.1 Vesicle leakage assay to test lipid bilayer stability 9.6.2 Examining lipid bilayer fluidity with fluorescent probes 10 APPENDIX 10.1 GENERATION OF A TM3/4 MUTANT LIBRARY 10.2 TM3/4 SCREENINGS WITH CFTR CORRECTORS 10.2.1 SmFRET control screenings and supporting data 10.2.2 Extracted closed state fractions from smFRET screenings 10.2.3 DLS to measure vesicle integrity after corrector addition 11 REFERENCES 12 ACKNOWLEDGEMENTS 13 ERKLÄRUNG GEMÄß §5 ABS. 1 S. 3 DER PROMOTIONSORDNUNG

Chronic Pseudomonas aeruginosa infections play an important role in the progress of lung disease in patients suffering from cystic fibrosis (CF). Recent studies indicate that polymicrobial microbiome profiles in the airway are associated with less inflammation. Thus, the hypothesis was raised that certain commensal bacteria might protect the host from inflammation. We therefore performed a screening study with commensals isolated from CF airway microbiome samples to identify potential beneficial commensals. We isolated more than 80 aerobic or facultative anaerobic commensal strains, including strains from genera Streptococcus, Neisseria, Actinomyces, Corynebacterium, Dermabacter, Micrococcus and Rothia. Through a screening experiment of co-infection in human epithelial cell lines, we identified multiple commensal strains, especially strains belonging to Streptococcus mitis, that reduced P. aeruginosa triggered inflammatory responses. The results were confirmed by co-infection experiments in ex-vivo precision cut lung slices (PCLS) from mice. The underlying mechanisms of the complex host-pathogen-commensal crosstalk were investigated from both the host and the bacterial sides with a focus on S. mitis. Transcriptome changes in the host in response to co-infection and mono-infection were evaluated, and the results indicated that several signalling pathways mediating inflammatory responses were downregulated by co-infection with S. mitis and P. aeruginosa compared to P. aeruginosa mono-infection, such as neutrophil extracellular trap formation. The genomic differences among S. mitis strains with and without protective effects were investigated by whole genome sequencing, revealing genes only present in the S. mitis strains showing protective effects. In summary, through both in vitro and ex vivo studies, we could identify a variety of commensal strains that may reduce host inflammatory responses induced by P. aeruginosa infection. These findings support the hypothesis that CF airway commensals may protect the host from inflammation.

Cystic fibrosis (CF) is caused by mutations in the gene that codes for the chloride channel cystic fibrosis transmembrane conductance regulator (CFTR). Recent advances in CF treatment have included use of small-molecule drugs known as modulators, such as Lumacaftor (VX-809), but their detailed mechanism of action and interplay with the surrounding lipid membranes, including cholesterol, remain largely unknown. To examine these phenomena and guide future modulator development, we prepared a set of wild type (WT) and mutant helical hairpin constructs consisting of CFTR transmembrane (TM) segments 3 and 4 and the intervening extracellular loop (termed TM3/4 hairpins) that represent minimal membrane protein tertiary folding units. These hairpin variants, including CF-phenotypic loop mutants E217G and Q220R, and membrane-buried mutant V232D, were reconstituted into large unilamellar phosphatidylcholine (POPC) vesicles, and into corresponding vesicles containing 70 mol% POPC +30 mol% cholesterol, and studied by single-molecule FRET and circular dichroism experiments. We found that the presence of 30 mol% cholesterol induced an increase in helicity of all TM3/4 hairpins, suggesting an increase in bilayer cross-section and hence an increase in the depth of membrane insertion compared to pure POPC vesicles. Importantly, when we added the corrector VX-809, regardless of the presence or absence of cholesterol, all mutants displayed folding and helicity largely indistinguishable from the WT hairpin. Fluorescence spectroscopy measurements suggest that the corrector alters lipid packing and water accessibility. We propose a model whereby VX-809 shields the protein from the lipid environment in a mutant-independent manner such that the WT scaffold prevails. Such ‘normalization’ to WT conformation is consistent with the action of VX-809 as a protein-folding chaperone.