Refine
Year of publication
- 2013 (15) (remove)
Document Type
- Doctoral Thesis (15) (remove)
Language
- English (15) (remove)
Has Fulltext
- yes (15)
Is part of the Bibliography
- no (15)
Keywords
- Copper decoration (1)
- Etching (1)
Institute
- Biochemie und Chemie (15) (remove)
Channelrhodopsin-2 (ChR2) is a light-gated cation selective channel from the unicellular alga Chlamydomonas reinhardtii, which is involved in phototaxis and photophobic responses. As other rhodopsins, ChR2 comprises a seven-transmembrane helix (TMH) motif and a retinal as the light-sensitive chromophore. The chromophore is covalently attached via a protonated Schiff base to the conserved lysine residue Lys257 located in TMH7. Based on its primary sequence and the all-trans configuration of the retinal in the ground state, ChR2 is assigned to the type I rhodopsins, also referred to as microbial-type rhodopsins. Upon light activation, the retinal isomerizes from the all-trans to the 13-cis form. This photoisomerization, which is accompanied by conformational changes of the protein, eventually leads to the opening of the channel and cation translocation. Cation flux during the conductive state leads to depolarization of the cell membrane and subsequent triggering of action potentials when expressed in neurons. Therefore, ChR2 has become the most versatile optogenetic tool, enabling a non-invasive investigation of neural circuits at high spatial and temporal resolution. With the rapidly increasing importance of ChR2 as a tool in neurobiology and cell biology, structural information is the prerequisite to an unambiguous understanding of the molecular mechanisms of this unique light-activated ion channel. The coupling between isomerization and structural alterations is well understood for other microbial-type rhodopsins, like bacteriorhodopsin (bR), halorhodopsin (HR) and sensory rhodopsin II (SRII). In case of ChR2, the first data on light-induced conformational changes came from spectroscopic studies and structural information is still missing. However, in order to fully understand the mechanism of light transduction by ChR2, it is necessary to determine the changes in the protein structure at specific steps in the photocycle.
By the time I started my PhD thesis, there was no structural information of ChR2 available. Therefore, the objective of this thesis was to obtain structural information of the transmembrane domain containing the first 315 amino acids of ChR2 by cryo electron crystallography. Besides revealing the structure of membrane proteins, cryo-EM of two-dimensional (2D) crystals is ideal for investigating conformational changes in membrane proteins induced by different stimuli. Therefore, the second objective of my thesis was the investigation of light-induced conformational changes in the slow C128T ChR2 mutant. The ~1,000 times longer lifetime of the open state of the C128T mutant compared to the wild-type allowed to trap different intermediates that accumulate during the photocycle.
In 2012, the X-ray structure of a channelrhodopsin-1/channelrhodopsin-2 chimaera (C1C2) at 2.3 Å resolution in the closed dark-adapted state was published (Kato et al., 2012). The structure revealed the essential molecular architecture of C1C2, including the retinal-binding pocket and the putative cation conduction pathway. Together with biochemical, spectroscopic, mutagenesis experiments, and the high-resolution model, some functionally important residues of ChR2 have been identified. However, unambiguous explanation of the molecular determinants that contribute to activation (gating) and transport were still mostly unknown.
RESULTS AND CONCLUSIONS
The first half of my theses dealt with 2D crystallization of ChR2. I succeeded in obtaining 2D crystals of ChR2 of four different types, which differed in size, crystal packing, crystal contacts and resolution, yielding structure factors up to 6 Å resolution. The crystals were grown by reconstituting the protein with different lipids at various lipid-to-protein ratios. The best crystals formed with the synthetic lipid DMPC and EPL upon detergent removal by dialysis. The projection maps calculated from these crystals revealed the overall structure of C128T ChR2 at 6 Å resolution and were published in 2011 (Müller et al., 2011). Surprisingly, ChR2 was found to be a dimer in all crystal types. The ChR2 dimer was stable both in detergent solution and in the presence of lipids for 2D crystallization. The monomers clearly showed the expected densities for the seven TMHs.
The arrangement of the ChR2 dimers on the four 2D lattices was different. However, comparison of the individual rojection maps revealed no significant differences within the ChR2 interface in the four crystal forms. The observation that the structure of the dimer was the same in all four crystal forms and in different lipids suggested strong specific contacts between the two protomers and implied that the protein was also dimeric in the native membrane. These findings were in agreement with Western blot analysis of plasma membranes from oocytes expressing ChR2 and laser-induced liquid bead ion desorption mass spectrometry, which both showed ChR2 as a dimer. The unusual stability of the ChR2 dimer contrasts with other microbial rhodopsins, which exist in different oligomeric states, i.e. monomers, trimers or dimers. These observations raised the question whether the functional unit is the monomer or the dimer.
The comparison of the projection map of the light-driven proton pump bR at the same resolution showed similar overall dimensions. Based on this comparison, the densities which became evident in the ChR2 projection maps could be assigned to the corresponding seven densities in bR. The shape of the densities near the dimer interface suggested that TMHs 2, 3, and 4 are oriented more or less perpendicular to the membrane plane, while the other four helices appear to be more tilted, as in bR.
Based on the high-resolution bR structure and the projection structures obtained, I have built a homology model. On the basis of this homology model, several residues found in the dimer interface were selected for mutational studies in order to disrupt the dimer interface.
The investigation of light-induced conformational changes in C128T ChR2 was the second part of my thesis. I designed an experimental setup for trapping light-induced conformational changes in C128T ChR2. In addition, I optimized the sample preparation in a way that the different illumination conditions did not alter the quality of the crystals. I have trapped two different functional states, namely the conductive open state and the non-conductive closed dark-adapted state.
In order to visualize the location and the extent of conformational changes, projection difference maps were calculated between the open and the closed state. Visual inspection of the difference maps between the open and the two closed states revealed three difference peaks that map to the TMHs 2, 6, and 7, indicating significant and specific rearrangements of these helices. The strong pair of positive/negative peaks at TMH6 suggests an outward tilt movement of approximately 2 Å. Close comparison of similar work on bR revealed that this movement is likely to occur at the cytoplasmic end of TMH6. A second highly significant negative peak is observed at TMH7, indicating a less pronounced tilt compared to TMH6. The third negative peak at TMH2 indicates a loss of density in this region. No significant differences were recorded at the TMH1, 5 and at the dimer interface formed by TMH3 and 4.
I succeeded in trapping and characterizing the open and closed state in the photocycle of ChR2 and could demonstrate that the transition from the closed to the open state is linked to significant light-induced tilt movements of TMH6 and 7, plus a loss of order in TMH2. These conformational changes are likely to create a large water-filled conducting pore, which seems to be required for the conductance of up to 2,000 ions per photocycle. The previously mentioned spectroscopic studies support the difference structures I obtained. This approach sets the stage for studying structural changes accompanying the formation and decay of other photocycle intermediates in ChR2. Future studies will aim at three-dimensional maps of the open and closed state at higher resolution.
Structural determinants for substrate specificity of the promiscuous multidrug efflux pump AcrB
(2013)
Opportunistic Gram-negative pathogens such as Escherichia coli, Klebsiella pneumoniae, Acinetobacter Baumanii and Pseudomonas aeruginosa are becoming more and more multiresistant against many commonly available antibiotics [39, 40]. An important resistance mechanism of Gram-negative bacteria is the efflux of noxious compounds by tripartite systems [39, 41-44]. The best studied and most clinically relevant tripartite system is the AcrA-AcrB-TolC system of Escherichia coli, where substrate recognition and energy transduction takes place in the inner membrane protein AcrB. AcrB has a remarkably huge substrate spectrum and can recognize structurally diverse molecules, such as hexan in contrast to erythromycin, as its substrates [45]. Therefore, overproduction of the tripartite system can render a Gram-negative pathogen resistant against multiple antibiotics at once. The mechanisms of how AcrB is able to recognize such an enormous spectrum of molecules as substrates, without compromising its specificity (e.g. by neglecting essential compounds like lipids or gluclose as its susbtates), remained puzzling. Structural insight into substrate specificity was so far limited to two co-crystal structures of AcrB, where minocycline and doxorubicin, respectively, were identified bound to an internal binding pocket of AcrB. This binding pocket is particularly deeply buried into internal parts of the T monomer of AcrB and was, therefore, denoted deep binding pocket (DBP). Analysis of several AcrB co-crystal structures with substrate molecules bound to the DBP [4, 23, 25] indicated that the substrate promiscuity involved multisite binding modes within the DBP. Multisite binding modes, where different substrate molecules can bind to slightly different positions and orientations to the same binding pocket, is a common feature of multidrug recognizing proteins such as QacR or BmrR [27-29]. Nevertheless, AcrB's substrate spectrum is much broader than substrate spectra of most other multidrug recognizing proteins. Therefore, it is likely that additional mechanisms are involved in mediating the observed high substrate promiscuity of AcrB. In our recently published high-resolution AcrB/doxorubicin co-crystal structure (pdb entry: 4DX7 [23]) we were able to identify two additional substrate binding pockets in the L monomer of AcrB: i) the access pocket (AP), with an opening towards the periplasm, and ii) a putative binding site in a groove between transmembrane helices 8 and 9 (TM8/TM9 groove), accessible from the lipid layer of the inner membrane. Both binding pockets are likely to be access sites for substrates towards AcrB. Furthermore, each of the binding pockets are possibly specialized to recognize a specific subset of the entire substrate spectrum of AcrB, i.e. highly hydrophobic substrates (e.g. n-dodecyl-ß-d-maltoside or sodium dodecylsulfate) might access AcrB towards the TM8/TM9 groove and water soluble substrates (e.g. berberine) might access AcrB towards the AP. Since substrates will accumulate in the membrane or the periplasm according to their hydrophilic or hydrophobic nature, substrates will be "pre-selected" by the medium, rather than by the protein itself, and guided to their appropriate access site. This process is proposed to be called "medium- mediated pre-selection". The AcrB/doxorubicin co-crystal structure (pdb entry: 4DX7 [23]) furthermore revealed that the AP and DBP are in next neighborhood to each other and are separated by a switch loop. This switch loop adopts distinct conformations in the L, T and O monomers. Specific switch loop conformations are strongly involved in coordinating the selective occupation of both binding pockets, the AP and the DBP. The conformation of the switch loop in the L monomer (L-switch loop) opens the AP and closes the DBP, whereas the conformation of the switch loop in the T monomer (T-switch Loop) opens the DBP and closes the AP. An analysis of all asymmetric AcrB structures indicated that the L-switch loop is able to adopt multiple distinct conformations, whereas the conformation of T-switch loop remained largely congruent in all crystal structures. Moreover, each distinct switch loop conformation, observed in co-crystal structures of AcrB with occupied AP [4, 23], was perfectly adapted to the bound substrate molecule. Therefore, the putatively flexible switch loop is likely to act as an adaptive module and mediates a high binding pocket plasticity without altering the global protein structure. This binding mode is called adaptor-mediated binding mechanism, where an flexible adaptive module (like the switch loop) is able to adapt the surface shape of an binding pocket to different substrate molecules. Furthermore, structural and biochemical analyses of an AcrB G616N variant, revealed the involvement of specific switch loop conformations in the substrate specificity of AcrB. A substitution of G616, located on the switch loop, to N616 was able to alter the conformation of the switch loop exclusively in the L monomers of AcrB, whereas the switch loop conformations in T and O monomers remained congruent to the conformations observed in crystal structures of wildtype AcrB. Moreover, cells producing the AcrB G616N and MexB, both bearing the G616N amino acid substitution, exhibited a reduced resistance against certain substrates, whereas the resistance against most other substrates remained on the level of wildtype AcrB. Correlations of the phenotypes with minimal projection areas, a novel 2-spatiodimensional parameter which approximates the size of a substrate molecule, revealed that AcrB variants with a G616N substitution have a reduced efflux activity for exclusively large substrate molecules. The rejection of large substrates is most likely connected with altered L-switch loop conformations....
In the past century, scientists have realized that venoms are a source of a number of natural substances presenting a wide range of pharmacological properties and often displaying a high specificity for their targets. Thus, the field of toxinology came into being, which is defined as the study of toxic substances of biological origin. Toxins are found in a wide variety of animals, including fish, cone snails, scorpions, snakes, and even some mammals. To be classified as venom, these must contain substances, i.e. toxins, which disturb physiological processes and must be deliberately delivered to the target animal. Snakes have evolved one of the most sophisticated mechanisms for venom delivery. Envenomation by snakebite can induce and inhibit aggregation/agglutination of platelets as well as inhibit/activate hemostasis, but also disrupt other physiological functions via neurotoxins and angioneurin growth factors. Snake venoms contain a substantial amount of C-type lectin-related proteins (CLRPs) which are known to function, notably, as integrin inhibitors. CLRPs are heterodimers composed of homologous α and β subunits which can assemble either covalently or noncovalently to oligomers, resulting in αβ, (αβ)2 and (αβ)4 structures. Some of the main targets of CLRPs are membrane receptors, coagulation factors, and proteins essential to hemostasis. The platelet collagen receptors GPVI and α2β1 integrin as well as the von Willebrand factor receptor GPIb play important roles in platelet activation and aggregation and are considered main targets of antithrombotic drugs. In this thesis, the integrin α2β1 is particularly considered as it is the sole collagen-binding integrin on platelets. Reduced expression of this platelet receptor results in dysfunction of platelet responses. Equivalently, overexpression of α2β1 integrin results in an increased risk of thrombosis. As a result, selective inhibitors of the collagen-α2β1 interaction could give rise to effective antithrombotic drugs. Integrins are large receptors which mediate cell-cell contacts and the binding of cells to the extracellular matrix (ECM). Therefore, they play a role in physiological processes, e.g. hemostasis and immunity, as well as in pathological processes, e.g. tumor angiogenesis and atherosclerosis. 18 α and 8 β integrin subunits, with nine α subunits containing an additional A domain, associate non-covalently to form 24 heterodimers with distinct binding specificities. Integrin collagen receptors are a subclass of four receptors which all utilize the β1 subunit. The α2β1 integrin is a collagen-binding receptor expressed not only on platelets, but also on endothelial and epithelial cells. Consequently, this integrin is also essential for cell adhesion and migration playing a role in angiogenesis as well as tumor metastasis. To date, there are five known antagonists of α2β1 integrin: EMS16, rhodocetin, vixapatin, and most recently rhinocetin and flavocetin-A. The first four have been shown to be specific for the integrin α2A domain, the major collagen-binding domain. All these antagonists are CLRPs and present new leads for drug design. In the past few years, many insights into the structure and function of rhodocetin were obtained. Monoclonal antibodies proved to be advantageous in disclosing this information, making them not only useful as therapeutic agents, but also as tools for protein characterization. The venom of the Vipera palaestinae snake was recently shown to contain an α2β1 integrin inhibitor, which prevented the integrin from binding collagen. This inhibitor, called vixapatin, was the initial focus of this dissertation. Vixapatin’s interaction with the α2β1 integrin needed further characterization on a molecular and cellular level to assess its medical potential and monoclonal antibodies were to be used as a tool. Originally, vixapatin had been isolated by reversed-phase high-performance liquid chromatography. To avoid the stringency of this method, for this study, it was replaced with gentler chromatographic methods. First, the α2β1 integrin inhibitor was isolated from the crude snake venom with affinity chromatography using the α2A domain as bait, establishing a method to quickly screen venoms for α2β1-binding proteins which affect the collagenintegrin interaction. The applicability of this method to other snake venoms was shown by isolating an α2A domain-specific toxin from the venom of Trimeresurus flavoviridis. To allow further characterization of both these toxins, gel filtration and ion exchange chromatography were employed to purify the protein without the α2A domain. These classical protein purification methods resulted in similar separation patterns of both the V. palaestinae and T. flavoviridis venom proteins. Purified proteins exhibiting the potential of inhibiting integrinbinding to collagen were analyzed by two-dimensional gel electrophoresis. Both VP-i and flavocetin-A, the integrin inhibitors from V. palaestinae and T. flavoviridis, respectively, were shown to have more complex structures than was evident from the purification. Each consisted of four low-molecular-weight proteins which assembled into two bands (for VP-i) or one single band (for flavocetin-A) under non-reducing conditions. Mass spectrometry analyses revealed VP-i to belong to the family of CLRPs, just like vixapatin does. However, these two proteins differed in their primary sequences and only showed homology to one another. The toxin purified from T. flavoviridis revealed this toxin to be flavocetin-A, a heterodimeric CLRP which had so far only been shown to have GPIb-binding activity. At the time of flavocetin-A’s purification, flavocetin-B was co-purified; flavocetin-B consists of the same two α and β subunits, plus an additional γ subunit. As no sequence information is known to date for the γ subunit, it may be one of the additional proteins purified here, along with an additional δ subunit. Therefore, the toxin isolated here may actually consist of four different subunits forming a tetramer of two different heterodimers, generating an (αβ)2(γδ)2 structure. This proposed (αβ)2(γδ)2 flavocetin-A structure has binding sites for both α2β1 integrin and GPIb, with no sterical overlap, as shown by affinity chromatography using the α2A domain and the extracellular domain of the GPIb receptor. The potential of VP-i and flavocetin-A to inhibit integrin-binding to type I collagen was shown during purification: Both toxins efficiently bind to the integrin α2A domain; also, VP-i and vixapatin bind to the A domain with the same affinity. Surface plasmon resonance showed the interaction of flavocetin-A with the α2β1 integrin to be extremely strong and association to be very fast. Furthermore, both toxins were shown to inhibit binding of the wildtype integrin to collagen: VP-i and flavocetin-A acted antagonistically on cell adhesion and cell migration. Initially, the interaction between VP-i and α2β1 integrin was to be further characterized with the help of monoclonal antibodies. However, this proved problematic, the procedure requiring various optimizations. Although, after expert consultation, some monoclonal antibodies could be obtained, the cells were extremely sensitive and gave unsatisfactory results when tested as detection tools in Western blot and immunoassays. Concluding, two novel α2β1 integrin inhibitors were discovered: VP-i and flavocetin-A, which were purified using the same procedure and which have similar functions. Both are Ctype lectin-related proteins which effectively inhibit cell adhesion and migration. This underlines that nature has instrumentalized CLRPs to specifically inhibit α2β1 integrin. Further characterization of VP-i and flavocetin-A will be able to provide leads for future drug development.
In this thesis the integral membrane protein diacylglycerol kinase (DAGK) from E.coli is investigated with solid-state NMR. The aim is to gain an insight into the enzyme’s mechanism through integration of kinetic, structural and dynamic data. The biological function of DAGK is the transfer of the γ-phosphate group from Mg*ATP to diacylglycerol (DAG) building phosphatidic acid (PA)[6] as port of the membrane-derived oligosaccharide cycle[31,34]. Surprisingly, DAGK does not share structural or sequential similarities with other kinases[12]. Typical sequence motives found in other kinases, which catalyze phosphoryl transfer reactions, are not found[13]. In its physiological form DAGK is a homo-trimer with nine transmembrane helices, three catalytic centers and a size of 39.6 kDa.
First, the set-up of a real-time 31P MAS NMR experiment is shown. This experiment allows measuring in real-time the simultaneous ATP hydrolysis in the aqueous phase and lipid substrate phos-phorylation in the membrane phase with atomic resolution under magic angle spinning[56]. After fast transfer of the sample into the NMR spectrometer the enzymatic reaction is started with a temperature jump. This approach of real-time MAS NMR in a dual-phase system was demonstrated for the lipid substrate analogs dioleoyl- (DOG) and dibutyrylglycerol (DBG), with a C8 and C4 aliphatic chain, respectively. The combination of 31P direct and cross polarization functions as a dynamic filter. In the 31P direct polarized experiment nuclei in both phases are detected, while in the 31P cross polar-ized experiment, only nuclei in the membrane phase are detected. Rates for substrate turnover, i.e. degradation of γP-, βP, αP-ATP and build-up of βP-, αP-ADP, free phosphate as side reaction, and PA are obtained, which reveal a Michaelis-Menten behavior with regard to Mg*ATP and DBG. Here Mg*ATP and DBG follow a random-equilibrium model, where every substrate can bind indepen-dently from the other substrate. Analyses of the peak integrals from educts and products of the enzymatic reaction, revealed the stoichiometry of the reaction: 1.5 ATP molecules are used to phos-phorylate one DBG molecule. The excess of ATP is attributed to the basal ATPase activity. Further-more, experiments with ATPγS, usually regarded as a non-hydrolysable ATP-analog, where carried out. Surprisingly, DAGK hydrolyzes ATPγS and also transfers the thio-phosphate group to the lipid acceptor DBG, which points to a certain degree of plasticity in the active center. A phosphorylated enzyme intermediate was not detected. These results suggest the building of a ternary complex of Mg*ATP, DBG and DAGK performing a direct-phosphoryl transfer reaction, without passing through a phosphorylated enzyme intermediate. Experiments with the transition state analog ortho-vanadate (Vi) showed a decoupling of the ATP hydrolysis activity from lipid substrate phosphorylation. This indicates a specific transfer site for the γ-phosphate group from ATP to DAG, which can be blocked by Vi.
A general disadvantage of NMR spectroscopy compared to other spectroscopic methods is its inherent low sensitivity. One possible starting point for the improvement of signal-to-noise per unit time is the reduction of the spin-lattice relaxation time of protons[209]. Usually 95 % of the experi-mental time is required for the relaxation of the 1H to equilibrium. The addition of paramagnetic species can be used to reduce the 1H T1[233]. In a comprehensive study four different paramagnetic agents were tested: Cu2+-EDTA, Cu2+-EDTA-tag, Gd3+-TTAHA and Gd3+-DOTA. The titration of these paramagnetic complexes showed the principle feasibility of this approach, but differences between the tested species exist. The most promising complex is Gd3+-DOTA which, at a concentration of 2 mM, causes a 10-time improvement of signal-to-noise ratio per unit time. This allowed measuring 2D 13C-13C correlation spectra of proteoliposomes in one tenth of the usual required experimental time (i.e. 10 hours vs. 4 days) with good signal-to-noise.
For the investigation of structural or dynamic changes in the protein upon substrate interaction with MAS NMR, the spectral properties CP efficiency and resolution of the DAGK in liposomes needed to be improved. The most critical step during sample preparation is the reconstitution of the membrane protein from detergent micelles into a membrane of synthetic lipids under detergent removal. For this procedure the important criteria are enzymatic activity, measured in a coupled ATPase assay[55], and homogeneity of the proteoliposomes, which was tested e.g. on a discontinuous sucrose step gradient. Therefore an extensive study was carried out, in which different detergents, lipids and lipid mixtures, techniques for detergent removal and different protein-to-lipid ratios were tested. A direct correlation between high ATPase activity and good resolution was not found. Moreover, active DAGK in a mixture of DMPC and cholesterol, which emulates the membrane features of a membrane containing DAG, showed the best CP efficiency and resolution.
The assignment of the protein backbone and amino acid side chains the first mandatory step towards the investigation of structural and dynamical features influencing and defining the enzymatic mechanism by MAS NMR. As the assignment procedure is very time consuming for a total protein, a special labeling scheme for DAGK was developed, which allows assigning most of the protein areas presumably involved in enzyme catalysis. The assignment of DAGK with solution NMR[132] was not transferable to the MAS NMR spectra. Most important for the assignment process were the unique pairs[335], two consecutive amino acids which only appear once in the amino acid sequence. These unique pairs served as anchor points. Five different multinuclear MAS NMR experiments (DARR, NCO, NCA, NCACX, NCOCX) were required for the sequential assignment. It was possible to assign 35 % of the total amino acid sequence with one sample and 8 experiments acquired at 850 MHz. The secondary structure analysis showed subtle differences to the DAGK assignment with solution NMR[132], which can be attributed to the different environment in lipid bilayers and detergent micelles.
Data about structural and dynamical changes under substrate interaction can reveal details about the enzymatic mechanism. Therefore changes in chemical shift in 2D heteronuclear correlation experiments in the apo-state and under substrate saturated conditions with the substrates Mg*AMP-PNP, a non-hydrolysable ATP-analog, DOG, a mixture of Mg*AMP-PNP and DOG as well as inhibited by Vi were recorded. The most significant peak changes were observed at the interface membrane-cytoplasm as well as the the N-terminal amphipathic helix. The residues revealing chemical shift perturbations correlate with conserved residues or such residues, for which importance for catalysis and/or folding could be shown in mutation studies[8]. Especially noticeable were the changes at the amino acids Asn 72, Lys 64, His 87, Tyr 86 and Asp 95.
Beside changes of the chemical shift, changes of line width or signal doubling were observable. These changes can point to a correlation with dynamic reorientations in the μs-ms time regime, which are most relevant for enzymatic processes. The protein backbone dynamics in the apo-state as well as saturated with the substrates or inhibited with Vi were investigated with a 15N-CODEX experiment, which is based on the reorientation of the CSA tensor upon dynamical changes[350]. Specific effects of the different substrates or analogs on the protein backbone dynamic were revealed complementing the structural data and the chemical shift perturbation experiments.
Retroviral vectors are powerful tools in clinical gene therapy as they integrate permanently into the target cell genome and thus guarantee long-term expression of transgenes. Therefore, they belong to the most frequently used application platforms in clinical gene therapy involving a broad range of different target cells and tissues. However, stable genomic integration of retroviral vectors can be oncogenic, as reported in several animal models and in clinical trials. In particular, γ-retroviral vectors, which derive from naturally mutagenic γ-retroviruses, integrate semirandomly into the host genome with regard to the target sequence, but have a preference for regions of active transcription and regulatory elements of transcriptionally active genes. The integration can result in overexpression of adjacent genes or disruption of ‘target’ gene expression. Moreover, γ-retroviral integration can cause modified transcripts and proteins through alternative or aberrant splicing or through premature termination of transcription.
Initially, the event of insertional mutagenesis and subsequent induction of leukemia by the genotoxicity of a γ-retroviral vector was described in a mouse model after genetic modification of hematopoietic stem cells (HSCs). Vector-related activation and overexpression of the oncogene ecotropic viral integration site-1 (Evi1) fostered clonal outgrowth and leukemogenesis. Additional genotoxic events of γ-retroviral vectors were observed in clinical HSC gene therapy trials for X-linked severe combined immune deficiency (SCID-X1), chronic granulomatous disease (X-CGD), and Wiskott-Aldrich Syndrome (WAS). But, genotoxicity induced by γ-retroviral vectors has never been described in clinical gene therapy trials involving adoptive transfer of genetically modified mature T lymphocytes. This fact is surprising, since T cells are long-lived and have a high capacity of self-renewal.
In a previous study, the susceptibility towards oncogenic transformation of mature T cells and HSCs after genetic modification was compared. It could be demonstrated that T-cell receptor (TCR)-polyclonal mature T cells are far less prone to transformation after γ-retroviral transfer of (proto-)oncogenes in vivo than HSCs. Additional experiments revealed that TCR-oligoclonal (OT-I and P14) mature T cells are transformable in the same setting and give rise to mature T-cell lymphomas (MTCLs).
In the present thesis, the susceptibility of mature T cells towards insertional mutagenesis was investigated. Within the first part of the thesis, retroviral integration sites (RISs) from 33 murine MTCLs were retrieved and subsequently analyzed in terms of integration pattern, detection of common integration sites (CIS) and gene ontology (GO). As these bioinformatic results demonstrated that insertional mutagenesis most likely contributed to mature T-cell lymphomagenesis, the susceptibility of mature T cells was directly assessed in a mouse model. Therefore, murine TCR-oligoclonal OT-I T cells were transduced with an enhanced green fluorescent protein (EGFP) encoding γ-retroviral vector and gene-modified T cells were transplanted into RAG1-/- mice. After 16 months, including one round of serial transplantation, a case of MTCL emerged. Tumor cells were characterized by CD3, CD8, TCR and ICOS expression. Integration site analysis via ligation-mediated polymerase chain reaction (LM-PCR) revealed a proviral insertion in the Janus kinase 1 (Jak1) gene. Subsequent overexpression of Jak1 could be demonstrated on transcriptional and protein level. Furthermore, T-cell lymphoma cells were characterized by an activated Jak/STAT-pathway as signal transducer and activator of transcription 3 (STAT3) was highly phosphorylated. The overexpression of Jak1 was causally implicated in tumor growth promotion as specific pharmacological inhibition of Jak1 using Ruxolitinib significantly prolonged survival of mice transplanted with these Jak1-activated tumor cells. A concluding systematic metaanalysis of available gene expression data on human mature T-cell lymphomas/leukemias confirmed the relevance of Jak/STAT overexpression in sporadic human T-cell tumorigenesis.
This was the first reported case of an insertional mutagenesis event in mature T cells in vivo. Thus, the results obtained in this thesis underline the importance of long-term monitoring of genetically modified T cells in vivo and the evaluation of vector toxicology and safety in T-cell based gene therapies. In particular, the transduction of T cells with a recombinant TCR or CAR (chimeric antigen receptor) bears a risk enhancement, as normal T-cell homeostasis is perturbed besides the general risk of insertional mutagenesis.
In the absence of apparent mutations, alteration of gene expression patterns represents the key mechanism by which normal cells evolve to cancer cells.
Gene expression is tightly regulated by posttranscriptional processes. Within this context, RNA-binding proteins (RBPs) represent fundamental factors, since they control mechanisms, such as mRNA-stabilization, -translation and -degradation. Human antigen R (HuR) was among the first RBPs that have been directly associated to carcinogenesis. HuR modulates the stability and translation of mRNAs which encode proteins facilitating various ‘hallmarks of cancer’, namely proliferation, evasion of growth suppression, angiogenesis, cell death resistance, invasion and metastasis. Furthermore, it is well established that tumor-promoting inflammation contributes to tumorigenesis. In this process, monocytes are attracted to the site of the tumor and educated towards a tumor-promoting macrophage phenotype. While HuR has been extensively studied in various tumor cell types, little is known about HuR in hepatocellular carcinoma (HCC). Thus, the aim of my work was to characterize the contribution of HuR to the development of cancer characteristics in HCC. I was particularly interested to investigate if HuR facilitates tumor-promoting inflammation, since a role for HuR has not been described in this context. To this end, I depleted HuR in HepG2 cells (HuR k/d) and used a co-culture model of HepG2 tumor spheroids and infiltrating monocytes to study the impact of HuR on the tumor microenvironment. I could show that depletion of HuR resulted in the reduction of cell numbers. Additionally, the expression of proliferation marker KI-67 and proto-oncogene c-Myc was reduced, supporting a proliferative role of HuR. Furthermore, exposure to cytotoxic staurosporine elevated apoptosis in HuR k/d cells compared to control cells. Concomitantly, the expression of the anti-apoptotic mediator B-cell lymphoma protein-2 (Bcl-2) was markedly reduced in the HuR k/d cells, pointing to an involvement of HuR in cell survival processes.
Accordingly, a pro-survival function of HuR was also observed in tumor spheroids, since HuR k/d spheroids exhibited a larger necrotic core region at earlier time points and showed elevated numbers of dead cells compared to control (Ctr.) spheroids. Interestingly, HuR k/d spheroids isplayed reduced numbers of infiltrated macrophages, suggesting that HuR contributes to a tumor-promoting, inflammatory microenvironment by recruiting monocytes/macrophages to the tumor site. Aiming at identifying HuR-regulated factors responsible for the recruitment of monocytes, I found reduced levels of the chemokine interleukin 8 (IL-8) in supernatants of HuR k/d spheroids, supporting a critical involvement of HuR in the chemoattraction of monocytes. Analyzing supernatants of co-cultures of macrophages and HuR k/d or Ctr. spheroids revealed additional differences in chemokine secretion patterns. Interestingly, protein levels of many chemokines were elevated in co-cultures of HuR k/d spheroids compared to control co-cultures. Albeit enhanced chemokine secretion was observed, less monocytes are recruited into HuR k/d spheroids, further underlining the necessity of HuR in cancer related monocyte/macrophage attraction and infiltration. Differences between chemokine profiles of mono- and co-cultured spheroids could be attributable to changes in spheroid-derived chemokines as a result of the crosstalk with the immune cells. Provided the chemokines originate from monocytes/macrophages, the different secretion patterns suggest that HuR contributes to the modulation of the functional phenotype of infiltrated macrophages, since the tumorenvironment is critically involved in the shaping of macrophage phenotypes. Regions of low-oxygen (hypoxia) represent another critical feature of tumors. Therefore, I next analyzed the impact of HuR on the hypoxic response. Loss of HuR attenuated hypoxia-inducible factor (HIF) 2α expression after exposure to hypoxia, while HIF-1α protein levels remained unaltered. Considering previous results of our group, showing that HIF-2α depletion (HIF-2α k/d) resulted in the enhanced expression of HIF-1α protein, I aimed to determine the involvement of HuR in the compensatory upregulation of HIF-1α protein in HIF-2α k/d cells. I could demonstrate that not only total HuR protein levels, but specifically cytoplasmic HuR was elevated in HIF-2α depleted cells pointing to enhanced HuR activity. Silencing HuR in HIF-2α deficient cells attenuated enhanced HIF-1α protein expression, thus confirming a direct role of HuR in the compensatory upregulation of HIF-1α. This as also reflected on HIF-1α target gene expression. I further investigated the mechanism underlying the compensatory HIF-1α expression in HIF-2α deficient cells. Analyzing HIF-1α mRNA expression, I excluded enhanced HIF1-α transcription and stability to account for elevated HIF-1α expression in HIF-2α k/d cells. HIF-1α promoter activity assays confirmed the mRNA data. Furthermore, HIF-1α protein half-life was not elevated in HIF-2α k/d cells compared to control cells, indicating that HIF-1α protein stability is not altered in HIF-2α k/d cells. Analysis of the association of HIF-1α with the translational machinery using polysomal fractionation finally revealed an increased istribution of HIF-1α mRNA in the heavier polysomal fractions in HIF-2α k/d cells compared to control cells. Since augmented ribosome occupancy is an indicator for more efficient translation, I propose enhanced HIF-1α translation as underlying principle of the compensatory increase in HIF-1α protein levels in HIF-2α k/d cells. In summary, my results demonstrate that HuR is critical for the development of cancer characteristics in HCC. Future work analyzing the impact of HuR on tumor-promoting inflammation, specifically macrophage attraction and activation could provide new trategies to inhibit macrophage-driven tumor progression. Furthermore, I provide evidence that HuR contributes to the hypoxic response by regulating the expression of HIF-1α and HIF-2α. Targeting single HIF-isoforms for tumor therapy should be carefully considered, because of their compensatory regulation when one α-subunit is depleted. Thus, therapeutic strategies targeting factors such as HuR that control both α-subunits and at the same time prevent compensation might be more promising.
Plants absorb sunlight via photosynthetic pigments and convert light energy intochemical energy in the process of photosynthesis. These pigments are mainly bound to antenna protein complexes that funnel the excitation energy to the photosynthetic reaction centres. The peripheral antenna of plant photosystem II (PSII) consists of the major light-harvesting complex of PSII (LHC-II) and the minor LHCs CP29, CP26 and CP24. Light intensity can change frequently and plants need to adapt to high-light conditions in order to avoid photodamage. When more photons are absorbed than can be utilised by the photosynthetic machinery, excessive excitation energy is dissipated as heat by short-term adaptation processes collectively known as non-photochemical quenching (NPQ). A decrease in PSII antenna chlorophyll (Chl) fluorescence yield and a reduction in the average Chl fluorescence lifetime are associated with NPQ. The main component of NPQ is the so-called energy-dependent quenching (qE), and it is triggered by the rapid drop in thylakoid lumenal pH resulting from the plant’s photosynthetic activity. This process is thought to take place at the PSII antenna complexes, which therefore not only capture and transfer light energy but are also involved in balancing the energy flow. The decrease in lumenal pH acivates the enzyme violaxanthin de-epoxidase (VDE), which converts the xanthophyll violaxanthin (Vio) into zeaxanthin (Zea) in the xanthophyll cycle. In addition, the PSII subunit PsbS was discovered to be essential for qE by screening qE-deficient Arabidopsis thaliana mutants. This membrane protein is considered a member of the LHC superfamily, which also includes LHC-II and the minor LHCs. Previous studies on PsbS isolated either from native source or refolded in vitro have produced inconsistent results on its pigment binding capacity. Interestingly, a pH-dependent change in the quaternary structure of PsbS under high light conditions has been reported. This observed dimer-tomonomer transition very likely follows the protonation of lumenal glutamates upon the drop in pH and is accompanied by a change in PSII supercomplex localisation. PsbS dimers are preferentially found in association with the PSII core, whereas PsbS monomers co-localise with LHC-II.Despite the identification of !pH, Zea and PsbS as key players in qE, both the nature of the quencher(s) as well as the underlying molecular mechanism leading to excess energy dissipation still remain unknown. Several models have been put forward to explain the reversible switch in the antenna from an energy-transmitting to a quenched state. Proposals include a simple pigment exchange of Vio for Zea, and aggregation or an internal conformational change of LHC-II. Charge transfer (CT)quenching in the minor LHCs or quenching by carotenoid dark state (Car S1)-Chl interactions have also been suggested. However, none of these qE models has so far been capable of accommodating all the physiological observations and available experimental data. Most importantly, the function of PsbS remains an enigma. A recent qE model suggested that monomerisation of PsbS enables the protein to transiently bind a carotenoid and form a quenching unit with a Chl of a PSII LHC. In view of the various proposed qE mechanisms, this thesis aimed at understanding the interplay of the different qE components and the contribution of the PSII subunits LHC-II, the minor LHCs and PsbS to qE. The initial approach was to investigate the properties of the PSII subunits in the most simple in vitro model system, namely in detergent solution. For this purpose, LHC-II was isolated either from native source or refolded from recombinantly produced protein. Investigation of the minor LHCs and PsbS required heterologous expression and refolding. In addition, experiments were performed on aggregated LHC-II. Aggregates of LHC-II have been used as a popular model system for qE because they exhibit highly quenched Chl fluorescence. At the final stage of this doctoral work, a more sophisticated model system to approximate the thylakoid membrane was developed by reconstitution of the PSII subunits LHC-II and PsbS into liposomes. This system not only allowed for investigation of these membrane proteins in their native environment, but also for mimicking the xanthophyll cycle by distribution of Zea within the membrane as well as !pH by outside buffer exchange. The role of Zea in qE was first investigated with detergent solubilised antenna proteins. The requirement of this xanthophyll for qE is well-known, but the specific contribution to the molecular quenching mechansim is unclear. Previous work had shown that replacement of Vio for Zea in LHC-II was not sufficient to induce Chl fluorescence quenching in Zea-LHC-II, as suggested by the so-called molecular gearshift mechanism. However, by means of selective two-photon excitation spectroscopy, an increase in electronic interactions between Car S1 and Chls was observed for LHC-II upon lowering the pH of the detergent buffer. Electronic Car S1-Chl coupling became even stronger when Zea-LHC-II was probed. The extent of Car S1-Chl coupling correlated directly with the extent of Chl fluorescence quenching, in a similar way as observed previously in live plants under high-light conditions. However, very similar results were obtained with LHC-II aggregates. This implied that the increase in electronic interactions and fluorescence quenching was independent of Zea and low pH. Further experiments on aggregates of LHC-II Chl mutants indicated that the targeted pigments were also not essential for the observed effects. It is proposed that the same molecular mechanism causes an increase in electronic Car S1-Chl interactions and Chl fluorescence quenching in Zea-LHC-II at low pH as well as in aggregated LHC-II. Most likely, surface exposed pigments form random quenching centres in both cases. On the other hand, it was possible that Zea could act as a direct quencher of excess excitation energy in the minor LHCs. However, enrichment of refolded CP29, CP26 and CP24 with Zea did not lead to a change in the Chl excited state lifetime. Formation of a carotenoid radical cation, previously implied in CT quenching, was also not observed, although artificial generation of such a radical cation was principally possible as shown for CP29. During the course of this work, a study reporting the formation of Zea radical cations in minor LHCs was published. Therefore, Zea-enriched minor LHCs were again investigated on the experimental apparatus used in the reported study. Indeed, the presence of at least one carotenoid radical cation for each minor complex was detected. It is suggested that either the preparation method of incubating the refolded minor LHCs with Zea in contrast to refolding the complexes with only Zea and lutein causes the observed differences or that the observed spectral radical cation signatures are due to experimental artifacts. While the experiments with LHC-II and the minor LHCs gave useful insights into the putative qE mechanism, the quencher site and the mode of action of Zea could still not be unambiguously identified. Most importantly, these studies could not explain the function of the qE keyplayer PsbS. Therefore, the focus of the work was shifted to PsbS protein production, purification and characterisation. In view of inconsistent reports on the pigment binding capacity of this PSII subunit, refolding trials with and without photosynthetic pigments were conducted. The formation of a specific pigmentprotein complex typical for other LHCs was not observed and neither was the earlier reported “activation” of Zea for qE by binding to this protein. Nevertheless, PsbS refolded without pigments displayed secondary structure content in agreement with previous studies, indicating pigment-independent folding. Reconstitution of pigmentfree, refolded PsbS into liposomes confirmed that the protein is stable in the absence of pigments. Zea distributed in PsbS-containing liposomes also showed no spectral alteration that would indicate its “activation”. With the ability to reconstitute PsbS, it was then possible to proceed to modelling qE in a proteoliposome system. For this purpose, PsbS was co-reconstituted with LHC-II, which has been reported to interact with PsbS. One-photon excitation (OPE) and two-photon excitation (TPE) spectroscopy measurements were performed on LHC-II- and LHC-II/PsbS-containing liposomes. This enabled both quantification of Chl fluorescence quenching as well as determination of the extent of electronic Car S1-Chl interactions. The effect of Zea was investigated by incorporating it in the proteoliposome membrane. It was shown that Zea alone was not able to induce significant Chl fluorescence quenching when only LHC-II was present. However, when LHC-II and PsbS were co-reconstituted, pronounced Chl fluorescence quenching and an increase in electronic Car S1-Chl interactions were observed and both effects were enhanced when Zea was present. Western blot analysis indicated the presence of a LHC-II/PsbS-heterodimer in these proteoliposomes. In addition to the OPE and TPE measurements, the average Chl fluorescence lifetime was determined in detergent-free buffer at neutral pH and directly after buffer exchange to low pH. No significant changes in the average lifetime were observed for LHC-II proteoliposomes when either Zea was present or after exchange for low pH buffer. This indicated that Zea alone cannot act as a direct quencher, which concurs with the OPE measurements. Moreover, the complex was also properly reconstituted as no aggregation or significant Chl fluorescence quenching were observed. The average lifetime was not significantly affected in LHC-II/PsbS-proteoliposomes, independent of Zea or pH. However, a shortlived component in the presence of a long-lived component was not resolvable with the time resolution of the fluorescence lifetime apparatus.
Implications for qE model systems and the in vivo quenching mechanism are discussed based on the experiments in detergent solution, on LHC-II aggregates and with the proteoliposome model system.
Modern computational molecular quantum chemical studies, such as the present one, typically employ a wide range of theoretical techniques. The latter are often rather complicated and one should not generally expect that an experimental scientist in the area of physical chemistry, a potential reader of this work, should be familiar with all these techniques. To simplify the reading of the Thesis and to make it self-sufficient, it is supplied with an overview of the employed theoretical methodologies (Chapter 1). The overview explains basic quantum-chemical terminology referred to throughout the Thesis, introduces theoretical foundations of the methods and outlines their properties and limitations. In Part 1.1 of Chapter 1, methods for the solution of the molecular Schrödinger equation are introduced, while in the subsequent Parts 1.2 and 1.3 methods for the solution of the electronic Schrödinger equation are presented to find the ground and excited states, respectively. Part 1.4 is dedicated to basis-set effects which are omnipresent in electronic-structure calculations. It contains a number of unusual insights and concepts proposed by the author and, thus, may be insightful also to experts in quantum chemistry.
In Chapter 2, the phenomenon of acetone-water proton exchange catalyzed by tubular as well as amorphous aggregates of calix[4]hydroquinone (CHQ) macromolecules, which has been observed previously in NMR experiments (Ref. D1D), is investigated by means of correlated quantum-chemical methods. The first part of the study (Section 2.3-2.7) considers concerted proton transfer, assisted by several initially neutral OH-groups in the hydrogen-bonded networks of CHQ aggregates. The second part of the study (Section 2.8-2.13) is dedicated to a second mechanism of proton exchange: step-wise proton transfer via formation of ionic intermediates resulting from CHQ pre-dissociation. CHQ application-specific as well as general conclusions, relevant to the main topic of the Thesis (i.e. influence of specific microsolvation on the considered proton transfer processes), are presented in Section 2.14.
The phenomenon of dual fluorescence observed in clusters of methyl 4-N,N-dimethylaminobenzoate ester (DMABME) and two water molecules in the gas phase, is studied in Chapter 3. Experimentally, the dual fluorescence was detected in experiments combining optical and ground-state ion-depletion infrared spectroscopies in ultracold molecular beams (Ref. D2D). In Section 3.3, calculated ground-state infrared spectra are presented that allow to identify the structures of those isomers, which are present in the gas-phase, as well as the structure of the isomer responsible for dual fluorescence. To further understand the reaction mechanism of dual fluorescence, excited-state potential energy surfaces of the identified isomers were computed along the relevant twisted intermolecular charge-transfer formation coordinate and the mechanism of energy dissipation in these complexes was investigated (Section 3.4-3.5) (Ref. D3D). A brief summary of the main results of this chapter and conclusions are given in Section 3.6. Finally, in Section 3.7 a complementary benchmark study of the quality of ground-state potential energy surfaces of prototypical hydrogen-bonded systems (ammonia-water and formic acid-water dimers) obtained at the level of BSSE-corrected MP2 combined with moderate basis sets, has been conducted. The quality of potential energy surfaces was tested with respect to basis-set size, level of electron correlation and anharmonicity effects and the applied methodology to identify the IR spectrum of hydrated DMABME complexes (Section 3.3) has been found to be sufficient to uniquely assign the IR spectra.
So far clinical human immunodeficiency virus (HIV) therapy is limited to non-curative treatments. However, as recently shown, alternative approaches such as HIV gene therapy have the potential to functionally cure the disease (e.g. the hematopoietic stem cell (HSC)-transplantation with a CCR5Δ32 homozygous transplant) (1). In contrast to the highly personalized medical treatment applied in the ‘Berlin case’, more broadly applicable approaches are currently under intensive investigation.
One example is the adeno-associated-virus (AAV)-mediated delivery of in vivo secreted antiviral entry inhibitors (iSAVE), the concept of which is based on the direct in vivo administration of a broadly applicable highly potent antiviral gene (here: a C46-derived entry inhibitory peptide interfering with HIV-1 membrane fusion). The AAV-based gene delivery is believed to overcome several limitations of gene therapeutic treatments based on ex vivo lentiviral trials in the past. It is (i) targeting differentiated HIV target cells (i.e. liver and differentiated lymphatic cells) reducing the risk of genotoxicity compared to stem cell-based trials, (ii) overcoming the limitation of a low number of genetically modifiable cells as in lentivirally based ex vivo transduction strategies (i.e. limited modifiable cell number due to culture conditions and lower vector titers) and (iii) using the safe AAV vector system, which has not been associated with major genotoxicity in men. (iv) Most importantly, the concept of secretable entry inhibitors does not require transduction of large amounts of cells due to the protective bystander effect. Thus, iSAVE might be a treatment principle for HIV infection that might be able to cure patients irrespective of their viral isolates or adherence.
Accordingly, the iSAVE concept could aim at two different sites in the patient for the production of antiviral transgenes, either the systemic production via suitable producer cells (e.g. hepatocytes) or the local production in the lymphatic system.
In a first approach, we are able to efficiently target hepatocytes using the natural AAV serotype 8 to express high plasma levels of secretable antiviral entry inhibitors in order to systemically suppress viral replication. In this setting we could show that iSAVE peptides are highly expressed in hepatocytes. However, plasma levels of iSAVE were insufficient when using a secretable peptide as sole antiviral transgene.
As a second treatment strategy, the iSAVE project aimed to deliver antiviral genes directly to the site of viral replication, the lymphatic system. Here, (i) a panel of naturally occurring AAV serotypes as well as (ii) AAV retargeting approaches were employed to design a highly efficient and selective AAV vector variant for gene delivery into the lymphatic system after intravenous vector administration.
In detail, (i) screening of the natural occurring serotypes revealed that the AAV serotype 1 (AAV-1) was best in targeting splenic tissue in two humanized mouse models, however at a very low level. After systemic AAV-1 vector administration neither transduction of human lymphocytes did occur nor was iSAVE expressed in the lymphatic system in a humanized mouse model.
(ii) In a second approach, we modified the well-characterized AAV-2 serotype in a tropism-defining region of its capsid gene by insertion of human peripheral blood lymphocytes (hPBL)-tropic peptide ligands. These in turn were selected by M13 in vivo phage display and by in vivo AAV peptide display. Selected variants were cloned and tested for hPBL transduction in vitro. Although the selected variants did not show increased expression efficacies compared to AAV-2 WT, it still might be possible that the selected variant are more specific for hPBLs as these conditions have not been tested.
As these selection processes required a humanized mouse model that comprises a functional lymphatic system, we established the previously described Trimera mouse model in our lab (2). We found that this mouse model could be further improved to allow engraftment of a lower number of gene-modified (gm) human T cells as in the classical Trimera model. These modified Trimera mice (mT3 mice) were conditioned by inclusion of cyclophosphamide (CTX) to the irradiation-conditioning scheme of the classical Trimera model.
Comparison of mT3 mice with established NSG and DKO mice in an adoptive gm T cell transplantation setting revealed that NSG mice were the most robust model providing high reproducibility in human T cell engraftment. MT3 mice allowed a substantial, yet more variable engraftment of gm T cells. Besides comparing engraftment kinetics, the graft quality (i.e. clonality and cytokine milieu) was analyzed. Again, NSG mice showed the most balanced homeostatic repopulation three weeks after transplantation, while mT3 mice were prone to Th1-type, oligloclonal repopulation, indicating an early onset of xenograft-versus-host disease. Finally, the lymphatic infiltration was analyzed. As expected, mT3 mice provided the most intact lymphatic structures, although the normal lymphatic morphology was not restored.
In conclusion, it was demonstrated in this work that AAV-mediated iSAVE gene therapy faces specific limitations depending on the respective targeting approach
In the systemic approach, iSAVE peptides have to be further optimized in terms of transgene design itself, as high-level accumulation in murine plasma was not feasible for the short iSAVE precursor. In the local, lymphatic targeting approach, AAV-mediated expression faces its limits in targeting specificity but foremost expression efficacy. Thus, the AAV vector itself needs further optimization for sufficient local iSAVE expression levels. Independently from the AAV-related approaches, a novel humanized mouse model was established in this work. Despite drawbacks regarding repopulation variability and set-up complexity, the novel mT3 mouse model comprised improved secondary lymphatic structures for adoptive T cell transfer, which might be an interesting platform for studies in lymphoma or leukemia therapy.
Silicon wafers such as Silicon on Insulator (SOI) and strained silicon on Insulator (sSOI) are the essential and basic materials of advanced microelectronic devices. However, they often show various kinds of crystal defects which impair the function of these devices. The most efficient method to date, for detecting such defects and for determining their density, is to delineate them by etching the wafers with a suitable etching solution and characterise them via light optical microscopy. Etch pits are formed at defect sites which are etched at a faster rate than at the perfect lattice. The standard etching solution used for SOI and sSOI is a dilute version of Secco. As Secco contains carcinogenic and environmentally hazardous chromium (VI), the use of which is or will be restricted by law in many countries, suitable chromium (VI)-free etching solutions like Organic Peracid Etches (OPE), modified Chemical Polishing Etches (CP) like CP4 mod and mixtures with organic oxidizing agents like chloranil (CA) have been developed for the successful delineation of various types of crystal defects.
However there are still nanometer-sized defects which are hard to detect or escape detection by this method. Copper decoration is a well known method to magnify these defects. It consists in applying a copper nitrate solution to the back of the SOI or sSOI wafer. On annealing, copper diffuses through the substrate and the BOX (buried oxide) to the SOI/sSOI film and on quenching to room temperature, copper precipitates as copper silicide, SiCu3, foremost at crystal defects where the lattice strain is greater than at perfect lattice sites. These silicides increase the volume in these parts of the crystal lattice and defect magnification occurs. A considerable disadvantage of this method is its tendency for artefact formation, when the copper concentration used is too high, with the copper precipitating at the film surface. The consequence is a higher density of etch pits whereby true defect etch pits cannot be differentiated from those caused by artefacts.
The aim of this thesis is to show that the processes of decorating and etching can be combined successfully to delineate all crystal defects in SOI and sSOI. An ideal result would have been to find a copper decoration procedure that decorates all existing crystal defects at a copper concentration that avoids artefact formation.
Characterization of mouse NOA1 : subcellular localizaion, G-Quadruplex binding and proteolysis
(2013)
Mitochondria contain their own protein synthesis machinery with mitoribosomes that are similar to prokaryotic ribosomes. The thirteen proteins encoded in the mitochondrial genome are members of the respiratory chain complexes that generate a proton gradient, which is the electromotoric force for ATP synthesis.
NOA1 (Nitric Oxide Associated Protein-1) is a nuclear encoded GTPase that positively influences mitochondrial respiration and ATP production. Although a role in mitoribosome assembly was assigned to NOA1 the underlying molecular mechanism is poorly understood. This work shows that the multi-domain protein NOA1 serves multiple purposes for the function of mitochondria. NOA1 is a dual localized protein that makes a detour through the nucleus before mitochondrial import. The nuclear shuttling is mediated by a nuclear localization signal and the now identified nuclear export signal. SELEX (Systemic Evolution of Ligands by Exponential Enrichment) analysis revealed a G-quadruplex binding motif that characterizes NOA1 as ribonucleoprotein (RNP). G-quadruplex binding was coupled to the GTPase activity and increased the GTP hydrolysis rate. The sequence of localization events and the identification of NOA1 being a RNP lead to the discussion of an alternative import pathway for RNPs into mitochondria. The short-lived NOA1 contains ClpX recognition motifs and is specifically degraded by the mitochondrial matrix protease ClpXP. NOA1 is the first reported substrate of ClpXP in higher eukaryotes and augments the contribution of the ClpXP protease for mitochondrial metabolism. To assess the direct action of NOA1 on the mitoribosome co-sedimentation assays were performed. They showed that the interaction of NOA1 and the mitoribosome is dependent on the GTPase function and the nascent peptide chain. In vitro, NOA1 facilitated the membrane insertion of newly translated and isotope labeled mitochondrial translation products into inverted mitochondrial inner membrane vesicles. In conclusion, NOA1 is a G-quadruplex-RNP that acts as mitochondrial membrane insertion factor for mtDNA-encoded proteins.
This thesis provides a comprehensive model of the molecular function of NOA1 and is the basis for future research. The identification of NOA1 as ClpXP substrate is a major contribution to the field of mitochondrial research.
This work presents a biochemical, functional and structural characterization of Aquifex aeolicus F1FO ATP synthase obtained using both a native form (AAF1FO) and a heterologous form (EAF1FO) of this enzyme.
F1FO ATP synthases catalyze the synthesis of ATP from ADP and inorganic phosphate driven by ion motive forces across the membrane and therefore play a key cellular function. Because of their central role in supporting life, F1FO ATP synthases are ubiquitous and have been remarkably conserved throughout evolution. For their biological importance, F1FO ATP synthases have been extensively studied for many decades and many of them were characterized from both a functional and a structural standpoint. However, important properties of ATP synthases – specifically properties pertaining to their membrane embedded subunits – have yet to be determined and no structures are available to date for the intact enzyme complex. Therefore, F1FO ATP synthases are still a major focus of research worldwide. Our research group had previously reported an initial characterization of AAF1FO and had indicated that this enzyme presents unique features, i.e. a bent central stalk and a putatively heterodimeric peripheral stalk. Based on such a characterization, this enzyme revealed promising for structural and functional studies on ATP synthases and became the focus of this doctoral thesis. Two different lines of research were followed in this work.
First, the characterization of AAF1FO was extended by bioinformatic, biochemical and enzymatic analyses. The work on AAF1FO led to the identification of a new detergent that maintains a higher homogeneity and integrity of the complex, namely the detergent trans-4-(trans-4’-propylcyclohexyl)cyclohexyl-α-D-maltoside (α-PCC). The characterization of AAF1FO in this new detergent showed that AAF1FO is a proton-dependent, not a sodium ion-dependent ATP synthase and that its ATP hydrolysis mechanism needs to be triggered and activated by high temperatures, possibly inducing a conformational switch in subunit γ. Moreover, this approach suggested that AAF1FO may present unusual features in its membrane subunits, i.e. short N-terminal segments in subunits a and c with implications for the membrane insertion mechanism of these subunits.
Investigating on these unique features of A. aeolicus F1FO ATP synthase could not be done using A. aeolicus cells, because these require a harsh and dangerous environment for growth and they are inaccessible to genetic manipulations. Therefore, a second approach was pursued, in which an expression system was created to produce the enzyme in the heterologous host E. coli. This second approach was experimentally challenging, because A. aeolicus F1FO ATP synthase is a 500-kDa multimeric membrane enzyme with a complicated and still not entirely determined stoichiometry and because its encoding genes are scattered throughout A. aeolicus genome, rather than being organized in one single operon. However, an artificial operon suitable for expression was created in this work and led to the successful production of an active and fully assembled form of Aquifex aeolicus F1FO ATP synthase. Such artificial operon was created using a stepwise approach, in which we expressed and studied first individual subunits, then subcomplexes, and finally the entire F1FO ATP synthase complex. We confirmed experimentally that subunits b1 and b2 form a heterodimeric subcomplex in the E. coli membranes, which is a unique case among ATP synthases of non-photosynthetic organisms. Moreover, we determined that the b1b2 subcomplex is sufficient to recruit the soluble F1 subcomplex to the membranes, without requiring the presence of the other membrane subunits a and c. The latter subunits can be produced in our expression system only when the whole ATP synthase is expressed, but not in isolation nor in the context of smaller FO subcomplexes. These observations led us to propose a novel mechanism for the assembly of ATP synthases, in which first the F1 subcomplex attaches to the membrane via subunit b1b2, and then cring and subunits a assemble to complete the FO subcomplex. Furthermore, we could purify the heterologous ATP synthase (EAF1FO) to homogeneity by chromatography and electro-elution. Enzymatic assays showed that the purified form of EAF1FO is as active as AAF1FO. Peptide mass fingerprinting showed that EAF1FO is composed of the same subunits as AAF1FO and all soluble and membrane subunits could be identified. Finally, single-particle electron microscopy analysis revealed that the structure of EAF1FO is identical to that of AAF1FO. Therefore, the EAF1FO expression system serves as a reliable platform for investigating on properties of AAF1FO.
Specifically, in this work, EAF1FO was used to study the membrane insertion mechanism of rotary subunit c. Subunits c possess different lengths and levels of hydrophobicity across species and by analyzing their N-terminal variability, four phylogenetic groups of subunits c were distinguished (groups 1 to 4). As a member of group 2, the subunit c from A. aeolicus F1FO ATP synthase is characterized by an N-terminal segment that functions as a signal peptide with SRP recognition features, a unique case for bacterial F1FO ATP synthases. By accurately designing mutants of EAF1FO, we determined that such a signal peptide is strictly necessary for membrane insertion of subunit c and we concluded that A. aeolicus subunit c inserts into E. coli membranes using a different pathway than E. coli subunit c. Such a property may be common to other ATP synthases from extremophilic organisms, which all cluster in the same phylogenetic group.
In conclusion, the successful production of the fully assembled and active F1FO ATP synthase from A. aeolicus in E. coli reported in this work provides a novel genetic system to study A. aeolicus F1FO ATP synthase. To a broader extent, it will also serve in the future as a solid reference for designing strategies aimed at producing large multi-subunit complexes with complicated stoichiometry.
Alzheimer’s disease (AD), which was first reported more than a century ago by Alhzeimer, is one of the commonest forms of dementia which affects >30 million people globally (>8 million in Europe). The origin and pathogenesis of AD is poorly understood and there is no cure available for the disease. AD is characterized by the accumulation of senile plaques composed of amyloid beta peptides (Ab 37-43) which is formed by the gamma secretase (GS) complex by cleaving amyloid precursor protein. Therefore GS can be an attractive drug target. Since GS processes several other substrates like Notch, CD44 and Cadherins, nonspecific inhibition of GS has many side effects. Due to the lack of crystal structure of GS, which is attributed to the extreme difficulties in purifying it, molecular modeling can be useful to understand its architecture. So far only low resolution cryoEM structures of the complex has been solved which only provides a rough structure of the complex at low 12-15 A resolution Furthermore the activity of GS in vitro can be achieved by means of cell-free (CF) expression.
GS comprises catalytic subunits namely presenilins and supporting elements containing Pen-2, Aph-1 and Nicastrin. The origin of AD is hidden in the regulated intramembrnae proteolysis (RIP) which is involved in various physiological processes and also in leukemia. So far growth factors, cytokines, receptors, viral proteins, cell adhesion proteins, signal peptides and GS has been shown to undergo RIP. During RIP, the target proteins undergo extracellular shredding and intramembrane proteolysis.
This thesis is based on molecular modeling, molecular dynamics (MD) simulations, cell-free (CF) expression, mass spectrometry, NMR, crystallization, activity assay etc of the components of GS complex and G-protein coupled receptors (GPCRs).
First I validated the NMR structure of PS1 CTF in detergent micelles and lipid bilayers using coarse-grained MD simulations using MARTINI forcefield implemented in Gromacs. CTF was simulated in DPC micelles, DPPC and DLPC lipid bilayer. Starting from random configuration of detergent and lipids, micelle and lipid bilyer were formed respectively in presence of CTF and it was oriented properly to the micelle and bilyer during the simulation. Around DPC molecules formed micelle around CTF in agreement of the experimental results in which 80-85 DPC molecules are required to form micelles. The structure obtained in DPC was similar to that of NMR structure but differed in bilayer simulations showed the possibility of substrate docking in the conserved PAL motif. Simulations of CTF in implicit membrane (IMM1) in CHAMM yielded similar structure to that from coarse grained MD.
I performed cell-free expression optimization, crystallization and NMR spectroscopy of Pen-2 in various detergent micelles. Additionally Pen-2 was modeled by a combination of rosetta membrane ab-initio method, HHPred distant homology modeling and incorporating NMR constraints. The models were validated by all atom and coarse grained MD simulations both in detergent micelles and POPC/DPPC lipid bilayers using MARTINI forcefield.
GS operon consisting of all four subunits was co-expressed in CF and purified. The presence of of GS subunits after pull-down with Aph-1 was determined by western blotting (Pen-2) and mass spectrometry (Presenilin-1 and Aph-1). I also studied interactions of especially PS1 CTF, APP and NTF by docking and MD.
I also made models and interfaces of Pen-2 with PS1 NTF and checked their stability by MD simulations and compared with experimental results. The goal is to model the interfaces between GS subunits using molecular modeling approaches based on available experimental data like cross-linking, mutations and NMR structure of C-terminal fragment of PS1 and transmembrane part of APP. The obtained interfaces of GS subunits may explain its catalysis mechanism which can be exploited for novel lead design. Due to lack of crystal/NMR structure of the GS subunits except the PS1 CTF, it is not possible to predict the effect of mutations in terms of APP cleavage. So I also developed a sequence based approach based on machine learning using support vector machine to predict the effect of PS1 CTF L383 mutations in terms of Aβ40/Aβ42 ratio with 88% accuracy. Mutational data derived from the Molgen database of Presenilin 1 mutations was using for training.
GPCRs (also called 7TM receptors) form a large superfamily of membrane proteins, which can be activated by small molecules, lipids, hormones, peptides, light, pain, taste and smell etc. Although 50% of the drugs in market target GPCRs , only few are targeted therapeutically. Such wide range of targets is due to involvement of GPCRs in signaling pathways related to many diseases i.e. dementia (like Alzheimer's disease), metabolic (like diabetes) including endocrinological disorders, immunological including viral infections, cardiovascular, inflammatory, senses disorders, pain and cancer.
Cannabinoid and adrenergic receptors belong to the class A (similar to rhodopsin) GPCRs. Docking of agonists and antagonists to CB1 and CB2 cannabinoid receptors revealed the importance of a centrally located rotamer toggle switch, and its possible role in the mechanism of agonist/antagonist recognition. The switch is composed of two residues, F3.36 and W6.48, located on opposite transmembrane helices TM3 and TM6 in the central part of the membranous domain of cannabinoid receptors. The CB1 and CB2 receptor models were constructed based on the adenosine A2A receptor template. The two best scored conformations of each receptor were used for the docking procedure. In all poses (ligand-receptor conformations) characterized by the lowest ligand-receptor intermolecular energy and free energy of binding the ligand type matched the state of the rotamer toggle switch: antagonists maintained an inactive state of the switch, whereas agonists changed it. In case of agonists of β2AR, the (R,R) and (S,S) stereoisomers of fenoterol, the molecular dynamics simulations provided evidence of different binding modes while preserving the same average position of ligands in the binding site. The (S,S) isomer was much more labile in the binding site and only one stable hydrogen bond was created. Such dynamical binding modes may also be valid for ligands of cannabinoid receptors because of the hydrophobic nature of their ligand-receptor interactions. However, only very long molecular dynamics simulations could verify the validity of such binding modes and how they affect the process of activation.
Human N-formyl peptide receptors (FPRs) are G protein-coupled receptors (GPCRs) involved in many physiological processes, including host defense against bacterial infection and resolving inflammation. The three human FPRs (FPR1, FPR2 and FPR3) share significant sequence homology and perform their action via coupling to Gi protein. Activation of FPRs induces a variety of responses, which are dependent on the agonist, cell type, receptor subtype, and also species involved. FPRs are expressed mainly by phagocytic leukocytes. Together, these receptors bind a large number of structurally diverse groups of agonistic ligands, including N-formyl and nonformyl peptides of different composition, that chemoattract and activate phagocytes. For example, N-formyl-Met-Leu-Phe (fMLF), an FPR1 agonist, activates human phagocyte inflammatory responses, such as intracellular calcium mobilization, production of cytokines, generation of reactive oxygen species, and chemotaxis. This ligand can efficiently activate the major bactericidal neutrophil functions and it was one of the first characterized bacterial chemotactic peptides. Whereas fMLF is by far the most frequently used chemotactic peptide in studies of neutrophil functions, atomistic descriptions for fMLF-FPR1 binding mode are still scarce mainly because of the absence of a crystal structure of this receptor. Elucidating the binding modes may contribute to designing novel and more efficient non-peptide FPR1 drug candidates. Molecular modeling of FPR1, on the other hand, can provide an efficient way to reveal details of ligand binding and activation of the receptor. However, recent modelings of FPRs were confined only to bovine rhodopsin as a template.
To locate specific ligand-receptor interactions based on a more appropriate template than rhodopsin we generated the homology models of FPR1 using the crystal structure of the chemokine receptor CXCR4, which shares over 30% sequence identity with FPR1 and is located in the same γ branch of phylogenetic tree of GPCRs (rhodopsin is located in α branch). Docking and model refinement procedures were pursued afterward. Finally, 40 ns full-atom MD simulations were conducted for the Apo form as well as for complexes of fMLF (agonist) and tBocMLF (antagonist) with FPR1 in the membrane. Based on locations of the N- and C-termini of the ligand the FPR1 extracellular pocket can be divided into two zones, namely, the anchor and activation regions. The formylated M1 residue of fMLF bound to the activation region led to a series of conformational changes of conserved residues. Internal water molecules participating in extended hydrogen bond networks were found to play a crucial role in transmitting the agonist-receptor interactions. A mechanism of initial steps of the activation concurrent with ligand binding is proposed.
I accurately predicted the structure and ligand binding pose of dopamine receptor 3 (RMSD to the crystal structure: 2.13 Å) and chemokine receptor 4 (CXCR4, RMSD to the crystal structure 3.21 Å) in GPCR-Dock 2010 competition. The homology model of the dopamine receptor 3 was 8 th best overall in the competition.
Biological membranes separate the cell interior from the outside and have diverse functions from signal transduction, apoptosis to transportations of ions and small molecules in and out of the cell. Most of these functions are fulfilled by proteins incorporated in the membrane. However, lipids as the main component of membrane not only serve as structural element for bilayer formation but they are also directly involved e.g. signalling processes and bilayer properties are important to mediate protein interactions. To fully understand the role of lipids, it is necessary to develop a molecular understanding of how certain membrane components modify bulk bilayer structure and dynamics. Membranes are known to have many different motions in different conditions and time scales. Temperature, pH, water content and many other conditions change membrane dynamics in a high degree. In addition to this, time scales of motions in membranes vary from ns to ms range corresponding to fast motion and slow motion, respectively. Therefore, membranes are needed to be studied systematically by varying the conditions and using methods to investigate motions in various time scales separately. The aim of this study was therefore perform a combined solid-state NMR / molecular dynamics study on model membranes. Different substrates, such as potential drugs, polarizing agents and signaling lipids were incorporated into bilayers and their location within the membrane and their effect onto the membrane was probed. NSAIDs (non-steroidal anti-inflammatory drugs), pirinixic acid derivatives, ceramides and polarizing agents were the substrates for membranes in this study. There were several experimental methods that were applied in order to investigate effects of these substrates on membrane dynamics. Different kind of phospholipids including POPC, DMPC and DPPC were used. In addition to experimental work, with the information gathered from solid state NMR experiments molecular dynamics simulations were performed to obtain more information about the membranes at the molecular level. As a result, combination of solid-state NMR with molecular dynamics simulations provides very systematic way of investigating membrane dynamics in a large range of time scales.
Pirinixic acid derivatives were special interest of this study because of their activity on peroxisome proliferator-activated receptor (PPAR) as an agonist as well as on enzymes of microsomal prostaglandin E2 synthase-1 (PGE2s) -1 and 5-lipoxygenase (5-LO) as dual inhibitor. Two potent pirinixic acid derivatives, 2-(4-chloro-6-(quinolin-6-ylamino)pyrimidin-2-ylthio)octanoic acid (compound 2) and 2-(4-chloro-6-(quinolin-6-ylamino)pyrimidin-2-ylthio)octanoate (compound 3), have been worked and their insertion depts were investigated by combining of solid state NMR and molecular dynamics simulations. Both experimental and theoretical results pointed out that compound 3 was inserted the phospholipid bilayer more deeply than 2. NSAIDs – lipid mixtures have been also studied here. It is known that consumption of NSAIDs as in mixture with lipids results much fewer side effects than consumption of the drugs alone. Thus, it is crucial to understand interactions of NSAIDs with lipids and investigate the possible complex formation of drugs with lipids. In this study, interactions of three widely used NSAIDs, ibuprofen, diclofenac and piroxicam, with DPPC were investigated by solid-state NMR. 1H and 31P NMR results depicted that ibuprofen and diclofenac had interactions with lipids, which is an indication of drug-lipid complex formation whereas piroxicam didn’t show any interactions with lipids suggesting that no complex formation occurred in the case of piroxicam. Ceramides are known to play key roles in many cell processes and many studies showed that the functions of ceramides are related with the ceramide effects on biological membranes. Therefore, in this study, influences of ceramides on biophysics of lipid bilayers were investigated by using various solid state NMR techniques and molecular dynamics simulations. Results from molecular dynamics simulations clearly showed that ceramide and lipids have strong interactions. More evidences about ceramide-lipid interactions were provided from 1H and 14N NMR results. In addition, it was indicated by both simulation and experimental methods that ceramide increased the rigidity of DMPC by increasing chain order parameters. BTbk is a biradical, which is used as polarizing agent for dynamic nuclear polarization (DNP) experiments and found to be more efficient than other widely used polarizing agents such as TOTAPOL. Since it is a hydrophobic compound, which prefers to stay inside lipid bilayer it is important to investigate the location and orientation of bTbk along the bilayer in order to understand its enhancement profile in DNP measurements. In this study, both NMR relaxation time measurements and molecular dynamics simulations revealed that bTbk tends to stay more close to hydrophobic chain of lipids than the interfacial part of lipids at bilayer surface.
In the first part of this work, a brief introduction on lipid membranes as well as a theoretical summary on both methods of solid-state NMR and molecular dynamics simulations is given. Then, in the second part methodology is introduced for both solid-state NMR spectrometer and theoretical calculations. Afterwards, results of different membrane systems are discussed in the following parts for both solid state NMR and MD. Finally, in the last part, a summary and the conclusion of the overall results together with some future plans are explained.
ATP synthases are multi-subunit membrane enzymes, which utilize the energy stored in a transmembrane electrochemical ion gradient to produce adenosine-5´-triphosphate (ATP), the universal energy carrier in biological systems. Research on these important enzymes goes back more than 50 years and has produced innumerable studies. The F-type ATP synthase consists of two functionally distinct, but tightly coupled subcomplexes, the water-soluble F1 and the membrane-embedded Fo complex. In its simplest form, F1 consists of five different subunits with a stoichiometry of α 3β3γδε, and harbors three catalytic centers in the α 3β3-headpiece, while Fo consists of three different subunits in a stoichiometry of ab2cn, where n varies between 8 to 15 depending on the species. From a mechanistic standpoint, the complex can also be divided into two different units, namely a stator, α3β3δ-ab2, and a rotor, γε-cn. The enzyme utilizes the energy stored in a transmembrane electrochemical gradient of protons, or in some cases Na+, to drive ATP synthesis. In particular, the downhill translocation of these ions across the Fo complex drives rotation of the γε-cn unit, which is then transduced to the active centers, catalyzing the phosphorylation of adenosine-5`-diphosphate (ADP) with inorganic phosphate (Pi), and the release of ATP....