Open Access

Bacterial gene expression can be regulated at mRNA level by cis-acting mRNA elements termed riboswitches. Riboswitches operate by conformational switching between a ligand-free and a ligand-bound state with different structures that either activate or inhibit gene expression. This PhD thesis contributes to the molecular level understanding of full-length purine riboswitches. It presents biophysical investigations on the ligand-dependent folding of the full-length translation-regulating add adenine riboswitch from the gram-negative human pathogenic marine bacterium Vibrio vulnificus (Asw). Asw has the typical bipartite riboswitch architecture with a 5’ ligand-sensing aptamer domain and a 3’ regulatory domain termed expression platform. According to the working hypothesis, Asw employs a unique thermodynamically-controlled 3-state conformational switching mechanism between an apoB, an apoA and a holo conformation to regulate translation initiation in a temperature-compensated manner. The two apo conformations are the putative translation-OFF states and the holo conformation is the putative translation-ON state of Asw. In the main project of this PhD thesis, an integrated nuclear magnetic resonance (NMR) and smFRET spectroscopic study of the full-length 112-nucleotide Asw (112Asw) was performed. The adenine-dependent folding of 112Asw was monitored at the level of base pairing interactions by NMR of the RNA imino protons, and at the level of three long-range intramolecular distances by smFRET of immobilized molecules. The integrated NMR and smFRET spectroscopic study of 112Asw yielded two major findings. First, NMR and smFRET both revealed that adenine binding to 112Asw impedes apoB formation by stabilizing the apoA secondary structure in the holo conformation without modulating tertiary structural interactions between the two riboswitch domains. This highlights the central role of competitive P1 and P4 helix formation at the interface of the aptamer and the expression platform for switching the accessibility of the ribosome binding site of 112Asw. Moreover, it strongly corroborates the hypothesis that purine riboswitches in general operate according to the key principle of a spatially decoupled secondary structural allosteric switch that proceeds without ligand-induced tertiary structural interactions between the aptamer domain and the expression platform. Second, it was uncovered by smFRET that the apoA and the holo conformation of 112Asw do not adopt a single folding state at near-physiological Mg2+ concentration. Instead, apoA and holo exhibit a persistent dynamic equilibrium between substates with an undocked (U), a short-lived docked (D1; ~s) and a Mg2+-bound long-lived docked (D2; ~10 s) aptamer kissing loop motif. In the holo conformation, the fractional population of the long-lived docked substate is ~2-fold increased compared to the apoA conformation, but undocked and docked substates are still comparably stable. The here described multiple folding states of the apoA and the holo conformation might have regulatory properties that are in between the apoB translation-OFF state and the holo-D2 translation-ON state. Additonally, an integrated NMR and smFRET analysis of 127-nucleotide Asw (127Asw) is presented. Compared to 112Asw, 127Asw is 3’-elongated by 15 nucleotides of the adenosine deaminase encoding sequence of the add gene from Vibrio vulnificus. 127Asw was chosen as mRNA template for future investigations of the interaction between Asw and the 30S ribosomal subunit. The NMR spectra of 127Asw demonstrated that 127Asw has the same overall secondary structure as 112Asw. Like for 112Asw, the combined NMR and smFRET analysis of 127Asw showed that adenine binding impedes apoB formation and stabilizes a long-lived docked aptamer kissing loop fold. However, compared to 112Asw, 127Asw has a destabilized aptamer kissing loop motif and a stabilized P4 helix in the expression platform. Finally, ligand-observed studies of the transient encounter complex between Asw and the near-cognate ligand hypoxanthine are described. By competition binding WaterLOGSY NMR experiments with hypoxanthine and the adenine analogue 2,6-diaminopurine, it could be shown that hypoxanthine binds to the same binding site of 112Asw as the cognate ligand adenine. The hypoxanthine binding constant measured with the WaterLOGSY method is in the low mM range (1.8 mM) and substantially exceeds the physiological hypoxanthine concentration in E. coli (~0.3 mM), thus ruling out that hypoxanthine binding can significantly impact the translational regulation of Asw in vivo. Also, preliminary FTIR difference spectra of 13C,15N-labelled and unlabelled hypoxanthine in complex with the pbuE adenine riboswitch aptamer and the xpt guanine riboswitch aptamer are discussed. These spectra showed a pattern of multiple IR bands that appeared to be characteristic for the respective complex.

The fact that the interaction of oligonucleotides follows strict rules has been utilized to create two- or three-dimensional objects made of DNA. With computer-assisted design of DNA sequences, any arbitrary structure on the nanometer- to micrometer-scale can be generated just by hybridization of the needed strands. As astonishing these structures are, without any modification of the DNA strands involved no function can be assigned to them. Many different ways of functionalizing DNA-nanostructures have been developed with light-responsive nanostructures having a rather subordinated role. Almost all light responsive DNA-nanostructures involve the acyclic azobenzene-linking system tAzo based on D-threoninol which is known to work best at elevated temperatures to ensure optimal switching. As the structure of DNA-constructs is mainly maintained by hydrogen-bonding, variation of the temperature should be avoided in order to keep the structure intact. To develop a light-responsive nanostructure model system with low-temperature operating azobenzene C-nucleosides, DNA-minicircles have been utilized. Those minicircles bear a lariat-like protrusion with a 10 base long single-stranded overhang, which is responsible for the dimerization with a ring bearing a complementary binding region. DNA-minicircles have been produced in a sequential manner by building and purifying the single stranded minicircle first by splint ligation and prepratative PAGE or RP-HPLC, followed by annealing it to the outer ring and subsequent purification by molecular-weight cut-off. Imaging of DNA-minicircles by atomic force microscopy (AFM) was possible with several methods of sample preparation leading to images of varying quality. With the help of AFM, qualitative analysis of the minicircles was possible. It could be shown, that theoretical and empirical size dimensions of the rings and their interactions were in great accordance. Designing the interaction site of the minicircles proved to be the main task in this project. The amount of C-nucleosidic modifications was identified by screening, followed by a screening of their optimal position and binding partners in the counterstrand. Two azobenzene C-nucleosides in a 10mer binding region and abasic sites opposing them appeared to give the best compromise between absolute dimerization ratio and photocontrolled change of it, as identified by native PAGE. In the following, the dimerization ratios of minicircles containing azobenzene C-nucleosides were compared with minicircles containing tAzo and unmodified minicircles. It could be shown, that the tAzo-modification leads to an elevated binding affinity compared to the unmodified minicircles, but the change upon irradiation is relatively humble compared to the C-nucleosides. For the C-nucleosidic modifications dimerization ratios reached a maximum of 40% in favored trans-state, but could be almost completely turned-off when switching into cis-state. In addition, arylazopyrazole-modified C-nucleosides could be switched into trans-state by irradiating at 530 nm, which is an improvement compared to standard azobenzene, as it shifts irradiation wavelength closer to the phototherapeutic window. The utilization of DNA-analogous C-nucleosides bring two drawbacks with them: the ribose units include the flexibility of the sugar conformation and it is reasonable to think, that upon isomerization of the azobenzene, part of the steric stress generated is compensated by the sugar reconfiguration, which is lost for duplex destabilization. In addition, the combination of the ribosidic linker end the end-to-end distance of trans-azobenzene causes the chromophore to penetrate deep into the base stack of the opposing strand, causing a serious destabilization even in favored trans-state. The goal was to find a linker system, that combines the benefits of the azobenzene C-nucleoside without the possibility to change sugar conformation and the strong destabilization in the trans-state. For this reason locked azobenzene C-nucleosides in analogy to LNA nucleosides have been synthesized. The synthesis of LNA analogous azobenzene C-nucleosides (LNAzo) was possible over a 16-step synthesis, with the critical step being the addition of in situ lithiated azobenzene to protected sugar aldehyde. Both anomers of LNAzo and mAzo as reference where incorporated into different oligonucleotide test systems by solid phase synthesis for thorough evaluation. It could be shown, that LNAzo β has a similar performance to mAzo in DNA with overall slightly increased TM- and ΔTM-values. Performance of LNAzo β was similar to mAzo even if steric stress is reduced by using abasic sites in the counterstrand opposing the azobenzene. Only in a RNA context, the true potential of LNAzo β could be observed. In a DNA/RNA duplex, photocontrol could be improved by almost 50%, in a RNA/RNA duplex even by over 100%. Although the primary goal was the improvement of the azobenzene C-nucleoside for a DNA-nanostructure context, LNAzo β proved not to give a sufficient improvement in regard to the cost-value ratio. Never the less, the invention of the locked azobenzene C-nucleoside was a huge success for reversible photoregulation of RNA hybridization. With this, a new way to regulate RNA hybridization has been found, which could be used to create RNA therapeutics in an antisense-approach. As LNAzo β improved duplex stability only in a limited amount in DNA, further improvements on the backbone have been declared futile and focus shifted onto optimization of the chromophore. First, the azobenzene as it is installed on the ribosidic linker decreases duplex stability by forcing its distal aromat deep into opposing base stacking region. It would be an improvement, if in favored trans-state the distal aromat would be positioned in the less confined space of either major or minor groove and only upon isomerization would shift into base pairing region. Second, the azobenzene itself is not able to contribute to attractive interactions aside from relatively weak π-interactions to adjacent nucleobases, which could be improved, if it could partake in hydrogen bonding. For those apparent reasons, 2-phenyldiazenyl-modified purines have been selected as targets. They combine the ability to contribute to hydrogen bonding of nucleobases with the photochomicity of azobenzenes. Both 2’-deoxyadenosine- and 2’-deoxyguanosine-analogue photoswitches dAAzo and dGAzo have been synthesized and incorporated into 10mer DNA test systems by solid phase synthesis. It could be shown, that duplex stability could be increased compared to established azobenzene C-nucleoside. The improvement was stronger for dAAzo than for dGAzo as in the case for guanosine the amino function on the C2-position had to be replaced by the phenyldiazenyl function, reducing its ability to form hydrogen bonds. Unfortunately, photocontrol of duplex stability caused by 2-phenyldiazenyl purines was rather limited. A reason for this could be the positioning of the distal aromat within the duplex, which can be close to the opposing nucleobase (endo-helical) or in greater distance (exo-helical). The exo-helical conformation of the trans-isomer can only switch to the exo-P-cis-conformation, which relocates the distal aromat in the minor groove, without significant impact on duplex stability.

Die Plasmamembran eukaryotischer Zellen dient als Barriere zwischen dem Inneren einer Zelle und ihrer Umgebung. Eine wichtige Aufgabe von Proteinen, die sich in der Plasmamembran befinden, besteht in der Erkennung der Umgebung, der Übermittlung dieser Informationen über die Plasmamembran in das Innere einer Zelle und der Einleitung einer zellulären Antwort. Membranrezeptoren binden Liganden, was zu ihrer Aktivierung und der Rekrutierung von intrazellulären Proteinen führt. Funktionelle Signalkomplexe werden gebildet und leiten einen Informationstransfer durch die Zellmembran ein, so dass die Expression bestimmter Gene stimuliert oder unterdrückt wird. Eine Störung der Signalinitiierung und -übertragung tritt bei vielen Krankheiten auf, so dass Membranproteine ein wichtiges Ziel in der Medikamentenentwicklung sind. In dieser Arbeit wird die Fragestellung bearbeitet, wie der Tumornekrosefaktor-Rezeptor 1 (TNFR1) in funktionelle Komplexe in der Plasmamembran einer intakten Zelle organisiert ist. TNFR1 besitzt vier cysteinreiche Domänen (CRDs) in seiner extrazellulären Region. Die erste und von der Plasmamembran am weitesten entfernte CRD ist die Pre-Ligand Assembly Domain (PLAD). Kristallstrukturen zeigten, dass sich in einem TNFR1-Dimer zwei PLAD in unmittelbarer Nähe befinden. Crosslinking-Experimente berichteten über mehrere oligomere Zustände von TNFR1; die Ergebnisse unterschieden sich nach Art und Konzentration des Crosslinkers. In der nativen Umgebung einer intakten Zelle wurde der oligomere Zustand von TNFR1 bisher nicht bestimmt. Der kanonische Ligand für TNFR1 ist der Tumornekrosefaktor alpha (TNF), ein Homotrimer, welches in löslicher oder membrangebundener Form vorliegt. Nach der Bindung von TNF an TNFR1 bilden sich Rezeptortrimere. Diese Proteinkomplexe rekrutieren intrazellulär Proteine und bilden einen funktionellen Membrankomplex, der intrazelluläre Signalkaskaden aktiviert. Die kanonische Signalweiterleitung erfolgt durch den nuclear factor kappa-light-chain-enhancer of activated B-cells (NF-B), welcher Zellteilung oder Entzündung induziert. TNFR1 kann auch andere Signalwege wie beispielsweise Apoptose durch einen zytosolischen Komplex und die Procaspase-8, oder Nekroptose durch das Nekrosom und die mixed lineage kinase domain-like (MLKL)-Domäne einleiten. Die Dysregulation von TNFR1 ist bei einer Vielzahl von Krankheiten zu finden. Erhöhte TNFR1-Expressiosraten treten bei acquired immune deficiency syndrome (AIDS), multipler Sklerose und verschiedenen Krebsarten auf. In einem zweiten Projekt wurde in Zusammenarbeit mit Prof. Dr. Michael Lanzer (Heidelberg, Germany) der Expressionsgrad des Proteins VAR2CSA in membranassoziierten knobs bestimmt, welche in Erythrozyten vorkommen, die mit dem Parasiten Plasmodium falciparum infizierten wurden. VAR2CSA gehört zur Proteinfamilie des Plasmodium falciparum erythrocyte membrane protein 1 (pfEMP1). Nach einer Infektion wird VAR2CSA zur Wirtszellmembran transportiert und in knobs eingelagert. Patienten, die Sichelzellenanämie-Erythrozyten (HbAS) aufweisen, sind im Gegensatz zu Patienten mit gesunden Erythrozyten (HbAA) immun gegen Malaria. Während die beiden Erythrozytentypen eine unterschiedliche Morphologie der knobs aufweisen, blieb ihre Zusammensetzung in Bezug auf VAR2CSA bisher ungeklärt. Das Verständnis der Proteinfunktion erfordert eine Beschreibung der molekularen Organisation funktioneller Einheiten in der zellulären Umgebung. Hierfür ist die Fluoreszenzmikroskopie eine geeignete Methode, da sie eine gezielte Markierung von Zielproteinen ermöglicht. Die hohe Sensitivität ermöglicht die Visualisierung einzelner Proteine. Eine Einschränkung in der konventionellen Fluoreszenzmikroskopie ist die Auflösungsgrenze. Strukturelle Elemente, die kleiner als etwa die halbe Anregungswellenlänge sind (für die meisten Anwendungen 200 bis 300 nm) können nicht aufgelöst werden. Die Entwicklung der hochauflösenden Fluoreszenzmikroskopie ermöglichte es, diese Auflösungsgrenze zu umgehen und eine räumliche Auflösung von wenigen Nanometern zu erreichen, was die Visualisierung und Charakterisierung einzelner Proteinkomplexe ermöglichte. Eine Art der hochauflösenden Fluoreszenzmikroskopie ist die single-molecule localization microscopy (SMLM), die auf der Detektion einzelner Fluorophore, einer genauen Bestimmung ihrer Position (Lokalisation) und der Erzeugung eines rekonstruierten Bildes unterhalb der optischen Auflösungsgrenze basiert. Da die meisten Proben in der Fluoreszenzmikroskopie eine zu hohe räumliche Dichte an Fluorophoren aufweisen, um den Nachweis von einzelnen Fluorophoren zu ermöglichen, werden Verfahren zur Kontrolle der Emission von Fluorophoren eingesetzt. Eine Möglichkeit ist der Einsatz von Fluorophoren, die optisch zwischen einem nicht-fluoreszierenden und einem fluoreszierenden Zustand geschaltet werden können, z.B. photoschaltbare fluoreszierende Proteine in photoactivated localization microscopy (PALM) oder organische Farbstoffe in (direct) stochastic optical reconstruction microscopy ((d)STORM). SMLM erreicht eine räumliche Auflösung von 20 nm, was in den meisten Fällen ausreicht, um einzelne Proteinkomplexe in einer Zelle aufzulösen. Diese räumliche Auflösung ist jedoch nicht ausreichend, um Untereinheiten innerhalb eines Proteinkomplexes zu visualisieren. Zu diesem Zweck wurde SMLM erweitert und die verfügbare kinetische Information genutzt, die bei der Detektion einzelner Fluorophore ausgelesen wird. Viele Fluorophore weisen metastabile Dunkelzustände auf, die eine Lebensdauer von bis zu Sekunden aufweisen. Diese Übergänge erscheinen als "Blinken" der Fluoreszenzemission. In Kombination mit kinetischen Modellen kann aus der Anzahl an Blink-Ereignissen die Anzahl der Fluorophore ermittelt werden. Angewendet auf hochaufgelöste Proteinkomplexe kann die Auflösungsgrenze von hochauflösender Mikroskopie umgangen werden, und die Anzahl der Protein-Untereinheiten in einem hochaufgelösten Proteincluster ermittelt werden. Hierzu wird beispielsweise das photoschaltbare fluoreszierende Protein mEos2 an ein Zielprotein funsioniert (quantitative PALM (qPALM)). ...