Fusobacteria are Gram-negative, anaerobic bacteria that are common members of the oral microbiota but are also recognized as opportunistic pathogens implicated in a range of human diseases. This dissertation elucidates the molecular mechanism underlying Fusobacterium nucleatum pathogenesis in colorectal cancer through the comprehensive structural and functional analysis of two key adhesins: the bifunctional type Va autotransporter Fap2 and the trimeric type Vc autotransporter CbpF. Using Escherichia coli as a heterologous expression system, a functional recombinant form of Fap2 could successfully be produced, overcoming significant technical challenges associated with the expression of large proteins and with anaerobic pathogen studies. This enables safer functional studies and provides a platform for therapeutic development, including potential vaccine applications and structure-based drug design. Cryo-electron microscopy revealed the extracellular domain of Fap2 as ~45 nm long, with an overall rod-shaped structure featuring a kinked β-helix architecture with dual receptor-binding capabilities. The membrane-distal tip binds the immune checkpoint receptor TIGIT (KD ~0.6 μM), while a lateral binding pit accommodates the cancer-associated glycan Gal-GalNAc. These binding sites are spatially separated, enabling simultaneous engagement of both receptors. Molecular docking and dynamics simulations identified key interaction residues, though mutagenesis studies revealed complex binding mechanisms requiring further investigation. For CbpF, high-resolution structures of the trimeric adhesin alone and in complex with CEACAM1 were determined. The N-terminal region forms a threefold symmetric β-roll, with CEACAM1 binding mediated by a unique protruding loop (E142–Y149) that creates a composite binding interface requiring contributions from an adjacent protomer. This interaction exhibits nanomolar affinity (KD ~6 nM), representing two orders of magnitude stronger binding than Fap2-TIGIT. Mutagenesis confirmed the critical role of the loop and demonstrated that trimeric assembly is essential for receptor recognition. Both adhesins employ distinct structural scaffolds to achieve complementary pathogenic functions: Fap2 neutralizes immune cells at a distance while simultaneously enabling the anchoring of bacteria to tumors, whereas CbpF provides high-affinity immune suppression through CEACAM1 engagement. This work establishes the first high-resolution structural framework for fusobacterial adhesin function and demonstrates how bacterial pathogens exploit host immune checkpoints and promote cancer progression. These findings reveal sophisticated molecular strategies employed by Fusobacterium nucleatum for tumor colonization and immune evasion, providing crucial insights for understanding microbiota-driven colorectal carcinogenesis and identifying novel therapeutic targets.

This dissertation traces the outlines of democratic theories and practices as they took shape among Black American leftists in the 1930s who wrote about the unfinished nature of abolition. Drawing on W. E. B. Du Bois’s framework of abolition democracy, first developed in his 1935 historical study Black Reconstruction, I explore the resonances of this framework in a range of political writings, organizing efforts, and cultural representations of its time. Through close readings of published and unpublished texts by Du Bois, Richard Wright, Marvel Cooke, William L. Patterson, Louise Thompson, and Langston Hughes, I argue that Black American leftists of the Depression decade articulated radical visions of collective self-determination that would extend the unfinished work of positively substantiating the abolition of racial slavery to encompass comprehensive social, political, and economic transformation. In doing so, I propose that these intellectuals and activists laid critical groundwork for abolitionist projects that would develop in the wake of the civil rights movement and in response to an expanding carceral state in the late twentieth and early twenty-first centuries.

Since the publication of Angela Davis’s reappraisal of abolition democracy in 2005, scholars and activists have increasingly turned to this concept as a critical framework for articulating the demands as well as the social and political visions of police and prison abolition. They have argued that contemporary formations of state violence protect and reinforce racial hierarchies entrenched through centuries of enslavement, colonialism, and imperialism, and they have highlighted the multifarious ways in which impacted communities—especially (though not limited to) Black and Indigenous communities—have practiced refusal and organized resistance to these formations across time and space. My dissertation proposes Depression-Era America as a site where such refusal and organized resistance not only took shape in extensive ways but was also theorized by those involved in activism of the Black left as an abolitionist, democratic practice, thereby echoing Du Bois’s concept as he articulated it in 1935 as well as foreshadowing later reappraisals of the concept in their critiques of racialized state violence and racial capitalism. Ultimately, this study argues that the writings of 1930s Black leftists offer instructive elucidations for what abolition democracy has been and what it will have been, providing crucial insights for contemporary movements seeking to dismantle carceral geographies and build more just futures.

The first part of this study aimed to identify the transmission sources and reservoirs of Campylobacter spp. in and around broiler farms. Therefore, samples were collected from Barns and their surrounding environments and analyzed using a semi-quantitative cultivation method and WGS.

The results revealed high levels of Campylobacter spp. in fecal samples and boot swabs collected from inside the barns. Although many areas within the barns were contaminated, lower concentrations were found on drinkers, feeders, walls, and ventilation systems. In contrast, environmental samples showed sporadic and low levels of Campylobacter spp. The positive environmental samples were primarily taken from puddles, water residues contaminated with chicken manure in wheel tracks, and nearby agricultural streams. Campylobacter-positive samples in the nearby barn environment were often visibly contaminated with manure.

WGS methods detected genetic relatedness between isolates from environmental and barn samples, with some strains persisting across multiple rearing periods. Additionally, new strains were repeatedly identified throughout different farm cycles. However, C. jejuni isolates from more distant agricultural streams showed no genetic relation to those found within the barns.

The study's second phase focused on detecting VBNC Campylobacter in transmissions at broiler farms and their surrounding environments. As a result, VBNC Campylobacter were occasionally found in small quantities, aligning with cultivation-based findings. Boot swabs from areas visibly contaminated with chicken manure are often tested positive for VBNC Campylobacter.

This prompted further analysis of C. jejuni's capacity to enter the VBNC state in naturally contaminated manure, highlighting a rapid decline in cultivability and VBNC formation. A transition towards the VBNC state was observed within 72 hours. Additional experiments using the same C. jejuni strain were conducted under laboratory-simulated environmental conditions in both soil and water. The results show that VBNC C. jejuni remained viable in water for extended durations, particularly at 4 °C. In contrast, its persistence in soil was lower, likely due to limitations induced by the matrix. The results of the second study support the hypothesis that Campylobacter may persist in the VBNC state under favorable conditions, particularly in water, such as after manure-mediated transmission. The findings from both studies highlight the role of contaminated manure in supporting both culturable and VBNC Campylobacter spp. in broiler farm environments. Additionally, water may act as a reservoir for Campylobacter in the surroundings of broiler farms.

Many phenomena in science and technology are modeled by nonlinear partial differential equations. Since smooth solutions often do not exist or are not physically meaningful—particularly in the presence of shocks or turbulence—it is necessary to consider generalized solution concepts. In this thesis, several generalized formulations based on inequality principles are investigated. For thermodynamically consistent phase-field models, entropic solutions are introduced, relying on a global energy inequality and a local entropy inequality. This approach offers notable advantages over formulations based solely on a local energy balance, like an existing general weak-strong uniqueness principle. For anisotropic fluid and geophysical models, dissipative solutions are studied, which are defined through a relative energy inequality. Finally, a novel and unifying concept of energy-variational solutions is proposed in a general framework and applied to hyperbolic conservation laws, liquid crystal systems, and various viscoelastic fluid models. In selected cases, this new solution concept is compared with more established approaches.

The thesis work is focused on developing a multimodal, label-free vibrational imaging pipeline and its application to three interrelated problems in cancer research: (1) mapping the local chemical impact and distribution of the anti-EGFR monoclonal antibody cetuximab in multi-layered tumor oral mucosa (TOM) models, (2) identifying, dose- and time-dependent molecular signatures of ionizing radiation response in U87 glioma cells, and (3) characterizing and localization of anisotropic plasmonic gold “nanobone” nanoparticles inside cells to evaluate their imaging contrast and theranostic potential. Using an integrated experimental workflow constituting spontaneous Raman microspectroscopy, stimulated Raman scattering (SRS) microscopy, surface-enhanced Raman scattering (SERS), atomic-force-microscopy based infrared nanoimaging (AFM-IR), complementary cell-biological assays, the thesis provides a comprehensive, label-free chemical view of drug localization, radiation impact, and nanoparticle interactions with biological systems. (1) In TOM models, label-free spectral mapping and multivariate spectral unmixing quantitatively resolved cetuximab-associated signatures and spatially co-localized these with protein-rich microdomains. Comparative studies of topical versus systemic administration routes demonstrated clear differences in drug-induced alterations in cellular components. (2) The subsequent study presents the integration of spontaneous Raman, stimulated Raman, and photothermal expansion (AFM-IR) spectromicroscopy techniques to investigate the effects of varying radiation doses (2–10 Gy) on U87 glioma cells ex vivo. Raman-based and AFM-IR spectral analyses detected statistically significant changes and radiation-induced modifications in cellular proteins, nucleic acids, and lipids. The label-free methods measured a 3-fold reduction in nucleic acid content and a 2-fold reduction in lipid content in cells exposed to 10 Gy compared to untreated controls. (3) Multimodal characterization of gold nanobones with tunable plasmonic properties and strong SERS activity was another research topic of this thesis. Correlative SERS/SRS/TEM imaging established subcellular localization patterns and aggregation states, while cytotoxicity assays confirmed biocompatible regimes for imaging. The nanobones provided significant plasmon-enhanced contrast and can be employed as a multifunctional biomarker for investigating diverse cellular systems and pathological conditions.

The mammalian genome is organised around nuclear landmarks. Active, gene- and GC-rich chromatin is enriched near nuclear speckles, whereas inactive, gene- and GC-poor chromatin is located close to the nuclear lamina and the nucleolus. Despite immense scientific efforts, why and how the genome is organised in this manner remains unclear. Nuclear speckles are biomolecular condensates composed of RNA and hundreds of proteins involved in various stages of the RNA life cycle. Due to their positioning and composition, they are hypothesised to act as convergence points for 3D genome organisation, transcriptional regulation, and RNA processing. However, the multifaceted roles of speckles have remained elusive, largely due to the lack of appropriate tools to investigate their function. To address this, I developed a cell line that enables the acute removal of the core components of nuclear speckles, SON and SRRM2, allowing the investigation of the direct effects of speckle removal. My findings indicate that nuclear speckles are crucial for the maintenance of the 3D genome structure by providing structural stability and dampening random chromatin movements. Re- moval of SON, the viscoelastic core of nuclear speckles, reduces their resistance to deflection, leading to the increase of random chromatin movements, and disrupts higher-order genome organisation due to the accumulation of spurious long-distance interactions. Nuclear speckles are localised near gene-dense chromatin domains enriched in GC-rich exons separated by short, GC-rich introns. This flat exon-intron GC content results in a hard-to-splice, GC-levelled architecture. My results demonstrate that nuclear speckles are essential for the proper splicing of these gene architectures, while they are unnecessary for the splicing of GC- differential exon-introns, regardless of their proximity to speckles. Failure to properly splice the GC-rich, levelled exons results in the downregulation of over 1500 genes, due to the nuclear retention and cytoplasmic decay of misspliced transcripts. These speckle-dependent exon-intron architectures originated in amniotes, coinciding with the expansion of SON’s and SRRM2’s Intrinsically Disordered Regions (IDRs). This suggests the emergence of nuclear speckles in amniotes and their absence in fish and invertebrates. Using the cell line models I confirm the lack of speckle-like condensates in fish and fruit flies, and their presence in turtles, birds, and mammals. I further verify that the expanded IDRs of SON are essential for nuclear speckle condensation in humans and that overexpression of human SON in fish cells leads to the formation of speckle-like condensates.

This thesis summarizes the contributions to aerosol property observations from the passive Multi-Spectral Imager (MSI) of the Earth Cloud, Aerosol, and Radiation Explorer (EarthCARE) satellite mission. It was launched in May 2024 with the goal of improving the understanding of aerosol, cloud and radiation interactions. Therefore, the satellite platform has two active and two passive instruments that simultaneously measure vertical and horizontal aerosol and cloud properties as well as radiative fluxes. While the lidar on board allows to monitor the vertical aerosol distribution, MSI is supposed to add a horizontal context to these observations. In particular, here, the scientific description of the operational level 2 algorithm, which estimates aerosol optical thickness, is presented and verified using synthetic test data and Moderate Resolution Imaging Spectroradiometer (MODIS) data be- fore the public release of official EarthCARE level 2 data. It relies on MSI measurements in the four available channels from the visible to shortwave infrared range, applying the optimal estimation technique. Comparisons to reference data have shown a good agreement over ocean (correlations are usually 0.9 or better) and a reasonable agreement over land surfaces due to the commonly known more difficult signal separation of surface and aerosol over land surfaces. Since the contribution of the aerosol signal to the observation at the top of the atmosphere is rather small using passive measurements, an accurate characterisation and calibration of the instrument is crucial for achieving the best possible aerosol property estimates. A considerable central wavelength variation across the 150 km swath width of MSI was encountered before launch. The impact of MSI’s spectral misalignment effect (SMILE) is investigated and quantified due to the present natural spectral variation in surface reflectance and atmospheric properties, i.e., aerosol properties, within the spectral ranges of the MSI bands. Relative errors in the measured top-of-atmosphere signal of up to 30 % for low optical thickness over a land surface are expected if there is no SMILE correction applied or if it is not accounted for in the level 2 algorithm. Overall, the studies presented within this work show the potential knowledge added by MSI to aerosol observations from EarthCARE, taking into account the need for high accuracy in the measurements.

Die Haltung von Haustieren ist in Deutschland weit verbreitet und hat in den letzten Jahren stetig zugenommen. Neben Hunden und Katzen – die vor allem in Familien mit Kindern beliebt sind – gibt es eine wachsende Zahl von Haushalten mit exotischen Haustieren. Der Kontakt zu Tieren hat zwar einen positiven Effekt auf die physische und psychische Gesundheit, birgt aber auch gesundheitliche Risiken. Denn die enge Bindung und der intensive Kontakt zwischen Menschen und Tieren begünstigen die wechselseitige Übertragung zoonotischer oder antimikrobiell resistenter Erreger. Vor allem für vulnerable Personengruppen kann daraus ein relevantes Gesundheitsrisiko entstehen, weshalb ein individuelles Risikobewusstsein unerlässlich ist. Medizinisches Fachpersonal, einschließlich Tierärztinnen und Tierärzte, spielt eine zentrale Rolle dabei, Informationen zu übermitteln und somit Tierhalterinnen und Tierhalter über Gesundheitsrisiken aufzuklären sowie Risikobewusstsein zu vermitteln. Um jedoch zielgerichtet und effektiv kommunizieren zu können, sollten sie nicht nur Wissen über Risikofaktoren haben, sondern auch über gute Kommunikationsfertigkeiten verfügen. Darüber hinaus ist es wichtig zu verstehen, wie Tierhalterinnen und Tierhalter die Kommunikation ihrer Tierärztin oder ihres Tierarztes wahrnehmen, um Kommunikationsstrategien entsprechend anzupassen. Um die Rolle der Tierärztinnen und Tierärzten bei der Prävention von Zoonosen und AMR zu untersuchen, wurde eine deutschlandweite Online-Umfrage unter Hunde-, Katzen- oder Exotenhaltenden durchgeführt. Neben Fragen zum Haustier, Risikofaktoren, der tierärztlichen Betreuung sowie der tierärztlichen Risikokommunikation zu Zoonosen und AMR wurde die Kommunikationswahrnehmung von Tierhalterinnen und Tierhaltern bewertet. Dabei wurde ein Kommunikations-Score – angelehnt an das Vier-Seiten-Modell von Friedemann Schulz von Thun – entwickelt, um die von Tierhaltenden empfundene Zufriedenheit der tierärztlichen Risikokommunikation abzubilden. Die Befragung der Tierhaltenden wurde von Dezember 2022 bis April 2023 über LimeSurvey durchgeführt und über Soziale Medien beworben. Insgesamt haben 1315 Hunde- und Katzenhaltende sowie 547 Halterinnen und Halter exoti-scher Tiere an der Befragung teilgenommen, wobei ausschließlich die vollständig abgeschlossenen Datensätze (1015 bzw. 344) für die Analyse berücksichtigt wurden. Die Ergebnisse zeigen das Vorhandensein verschiedener Risikofaktoren, darunter vulnerable Personen im Haushalt, importierte Tiere sowie Wildfänge, die Verfütterung von rohem Fleisch sowie den Nachweis von potenziell zoonotischen Erregern oder AMR. Gleichzeitig weisen die Ergebnisse auf ein hohes Maß an Zufriedenheit mit der tierärztlichen Kommunikation bei den Tierhal-tenden hin. Dabei standen höhere Kommunikations-Scores und Frequenzen der tierärztlichen Risikokommunikation im Zusammenhang mit einer längeren Beziehung und Kontinuität in der tierärztlichen Betreuung. Allerdings äußerten die befragten Tierhalterinnen und Tierhalter den Wunsch nach mehr Informationen zu Zoonosen und AMR im Kontext der tierärztlichen Betreuung. Zudem fiel die Nutzung bzw. Verbreitungsrate von Informationsmaterialien zu Zoonosen und AMR von Tierärztinnen und Tierärzten an Tierbesitzende gering aus. Daher sollten Tierärztinnen und Tierärzte für die Risikokommunikation weiter sensibilisiert werden. Infolge-dessen ist es empfehlenswert, die Risikokommunikation in das tierärztliche Kommunikationscurriculum zu integrieren sowie berufsbegleitende Fortbildungsmöglichkeiten auszubauen. Zudem sollten mehr adressatengerechte Informationsmaterialien zu Zoonosen und AMR verfügbar gemacht werden, um die mündliche Risikokommunikation in der tierärztlichen Praxis zu ergänzen und zu festigen. Die Studie weist mehrere Limitationen auf, darunter potenzielle Auswahlverzerrungen durch die internetbasierte Bewerbung, mögliche Erinnerungsverzerrungen aufgrund des Zeitraums zur retrospektiven Auskunft von bis zu 12 Monaten sowie die Subjektivität der Selbstauskünfte der Tierhaltenden. Zudem könnten Einschränkungen in der Verständlichkeit von tierärztlichen Informationen sowie die begrenzte Differenzierung der 4-Punkte-Likert-Skala die Aus-sagekraft der Ergebnisse beeinträchtigt haben. Zukünftige Studien könnten sich darauf konzentrieren, die Wirkung von Bildungsressourcen – z. B. Informationsmaterialien – zu Zoonosen und AMR zu bewerten und weiterzuentwickeln. Zudem könnten Studien zur Risikokommunikation und zum Wissen über Zoonosen und AMR in der Humanmedizin weitere Erkenntnisse liefern, um interdisziplinäre Perspektiven zu erwei-tern. Langzeituntersuchungen zur Gesundheit von Haustieren und ihren Halterinnen und Hal-tern könnten zudem wertvolle Erkenntnisse liefern, um tierärztliche Maßnahmen gezielter auszurichten und die Zusammenarbeit zwischen Human- und Veterinärmedizin nachhaltig zu stärken.

Since the advent of psychological testing, psychometricians were focused on developing statistical approaches which could be used to verify the validity and the reliability of these tests as measurement instruments. However, this approach has mostly assumed the homogeneity of the response process and ignored the influences of other variables on the measured construct. This has resulted in a so-called canonical psychometric model.

Assumptions which are implied by the canonical psychometric model are likely untrue, given that it is a known fact that there are many sources of construct-irrelevant variance which can threaten the validity of the test. One of the primary sources of construct-irrelevant variance is related to the test-takers themselves. The behavior of the test-takers and the phenomena associated with it can have a profound influence on the scores with examinees will receive on the test. Examinees can approach the test with differing levels of motivation or anxiety; they can use different strategies to solve the same tasks and they can differ in the speed with which they work. Ignoring these processes brings into question the interpretations of the test scores and creates missed opportunities for deeper understanding of the characteristics of examinees.

In recent times, psychometric tests are conducted more and more using computers. Computerized testing has produced, initially as a side effect, a large amount of data created during the testing session. This data has become known as process data. Process data can be composed of various indicators related to the test-taking process, such as response times or omissions of test items. The availability of process data has created novel opportunities to investigate examinees' test-taking behavior in ways which would not be possible with just their response data.

Despite the increase in the interest in modeling test-taking processes with process data, as evidenced by the proliferation of many psychometric modeling approaches aiming to model various phenomena associated with test-taking behavior, a large volume of these models approach the test-taking process from a static perspective. However, more recently, researchers have started becoming interested in modeling the changes in test-taking behavior over the course of a testing session using process data. This line of research explores how different aspects of test-taking behavior fluctuate over the course of the test and, additionally, how their relationship with other aspects of test-taking behavior fluctuates through the test.

The current dissertation is nested in that line of research. The aim of the current dissertation is to add to the existing toolbox of similar approaches by developing novel psychometric models which can be used to investigate the dynamic relationship of various aspects of test-taking behavior.

More specifically, two approaches were developed under this dissertation. The first modeling approach aims to investigate the speed-ability trade-off by modeling the change in latent ability and in latent speed during the test. This, in turn allows for the insight into the intra-individual relationship of ability and speed, one form of which is the speed-ability trade-off. The proposed model is evaluated in a simulation study with promising results.

Furthermore, an empirical study conducted as a part of this dissertation aimed to validate the interpretation of certain model parameters of the aforementioned model as a measure of the speed-ability trade-off. Because the model captures the intra-individual relationship of ability and speed, it is possible that not only the speed-ability trade-off, but other confounding effects, such as the fluctuations of motivation and concentration, are also present in the parameter estimates. To explore whether this is the case, these confounders were measured during a testing session and their effects on model parameters were investigated. The results show how the model can be a useful tool for researching the speed-ability trade-off.

The second approach aimed to investigate the etiology of omitted responses by exploring how behavior on previous test items, operationalized as responses and response times, are related to skipping further items. Statistically, this was achieved through relaxing the assumption of local independence and including residual correlations between responses or response times on the one side, and item omission indicators on the other side, to the model. It was demonstrated that this model can be a useful tool to detect dynamic patterns of association in the data, which can be indicative of certain mechanisms of generation of omitted responses.

Generally, the approaches presented in this dissertation demonstrate how it is important to take into account the dynamic nature of the test-taking process when making inferences based on available test data. Aspects of test taking behavior are constantly fluctuating throughout the test and, moreover, the associations which they have with other aspects of test-taking behavior are showing dynamic patterns of change during the testing session. Advantages of interpreting these dynamic processes on an individual and group level are discussed, as well as the implications which this dynamicity could have on the design and usage of tests. Lastly, this dissertation reflects on its limitations and offers certain suggestions for future work.

Protein tyrosine phosphatases (PTPs) play a critical role in the regulation of cellular functions, and SHP2 phosphatase is a member of the non-receptor protein tyrosine phosphatase family of PTPs. As SHP2 is a protein associated with a variety of cancer-related diseases and is a potential immunomodulator, controlling SHP2 activity has important therapeutic implications. In the previous studies (2015), GS493, an orthosteric inhibitor of SHP2 developed by Jörg Rademann’s group, was tested in several human cancer cell lines, such as mammary gland cancer cells and pancreatic cancer cells. The results revealed that the inhibitor showed effective inhibition of SHP2 wildtype. In addition, another study in 2018 reported that GS493 also showed inhibition of both wildtype and mutant of SHP2 in pancreatic ductal adenocarcinoma and non-small cell lung cancer, while allosteric inhibitors could not be effective against the mutant under the same condition. This project aimed to strategically modify the functional moieties of GS493 to understand the contribution of functional moieties to activity and to improve the inhibitor. Although the inhibitor can effectively inhibit cancer cells growth in animal studies, there still has room for further optimization in terms of activity, solubility, and cell permeability. Additionally, the new published study in 2020 showed a novel structure contain the bicyclic scaffold, giving much more space to the designing of new SHP2 inhibitors. To obtain more novel effective inhibitors of SHP2, our optimization strategies followed several directions: 1) modifying of the phosphotyrosine(pTyr) mimetic to hance the binding ability of the compounds to the PTP domain of the SHP2 protein, 2) replacing of the hydrazone linker and nitro functionalities to optimize the solubility and the cell permeability of the inhibitors, and 3) reducing the flexibility of the molecule by formation of a central ring structure to explore a new structure scaffold. At the start of the investigation, various phosphotyrosine mimetics were investigated by inserting into the original structure of GS493, followed by fragment ligation screening to generate the ligation products with improved solubility in DMSO, and then the inhibitions of SHP2 with these compounds were analyzed by enzyme bioassay, the measurement was repeated in triplicate. Afterwards, docking programs were applied to calculate the new binding modeling of GS493 and the analogs. A hetero-bicyclic core structure containing pyrazolo[4,5-b]pyridine moiety was synthesized as a new scaffold. The pentafluorophosphate functional group was incorporated into this novel pyrazolo[4,5-b]pyridine scaffold. Furthermore, compound 29e (SW053), with the replacement of a nitro functionality, was derived as an excellent small molecule inhibitor of SHP2 with an IC50 = 39 nM. The binding modeling of 29e (SW053) was investigated to provide structure-activity relationships for the future drug discovery. The cellular experiments showed that compounds 21 and 29e (SW053) were able to block the REF-ERK signaling pathway in HGF-activated cancer cells.

Die Studie untersucht die mittelassyrische Stadt Kār-Tukultī-Ninurta auf der Grundlage der überlieferten Keilschriftquellen und archäologischen Befunde. Ausgangspunkt ist eine systematische Neubewertung des Textbestandes, dessen Überlieferungsgeschichte durch fragmentarische Grabungsdokumentation, unklare Provenienzen und editorische Lücken gekennzeichnet ist. Durch die Unterscheidung eines Kernkorpus, eines erweiterten Korpus sowie kontextrelevanter Vergleichstexte wird eine methodisch abgesicherte Grundlage für die historische Auswertung geschaffen. Auf diese Basis analysiert die Arbeit die administrativen, organisatorischen und wirtschaftlichen Strukturen des Bauprojektes Kār-Tukultī-Ninurta. Im Zentrum stehen die Fragen nach Planung, zeitlicher Strukturierung, Personaleinsatz, Ressourcendistribution und logistische Umsetzung. Die Untersuchung verbindet textanalytische, prosopographische und funktionsgeschichtliche Ansätze mit archäologischen Befunden und ermöglicht so eine differenzierte Rekonstruktion der Genese der Stadt. Die Studie versteht Kār-Tukultī-Ninurta nicht primär als kurzlebige Residenzgründung, sondern als administrativ institutionalisierenden Prozess, in dem sich mittelassyrische Verwaltungspraktiken exemplarisch verdichten. Durch die quellennahen Analysen und die vollständige Transliteration der relevanten Texte leistet die Arbeit einen grundlegenden Beitrag zur Erforschung mittelassyrischen Verwaltung, Bau- und Wirtschaftsstrukturen.

Gegenstand dieser Studie war die Validierung eines In Vitro-Verfahrens zum Nachweis von Botulinum Neurotoxin A und B (BoNT/A und BoNT/B) aus verschiedenen Lebensmittelmatrizes zur Einführung in die Lebensmittelroutinediagnostik. Außerdem wurde die fakultativ anaerobe Begleitflora untersucht sowie mikrobiologische Studien an BoNT-produzierenden Clostridium spp. und deren fakultativ anaeroben Begleitflora durchgeführt. Die untersuchten Lebensmittelmatrizes waren Brühwürste, Rinderhackfleisch, geräuchertes Forellenfilet, Gemüse und Milch mit 1.5 % sowie mit 3.5 % Fettgehalt. Bisher existiert als Goldstandard zum Nachweis von BoNTs der Maus-Bioassay, der ethisch aber äußerst umstritten ist. Im Falle einer positiven Probe, dessen Extrakt den Mäusen intraperitoneal appliziert wird, können die Mäuse bei vollem Bewusstsein ersticken, da unter anderem eine schlaffe Lähmung der Atemmuskulatur eintritt. Daher ist es notwendig, eine Alternativmethode in der Lebensmittelroutinediagnostik einzuführen, die den Mausversuch ablöst. Der so genannte Endopeptidase-Suspensionsimmunoassay (Endopep-SIA) weist die Spaltprodukte der Substratproteine nach, die von den BoNTs gespalten werden. Diese Substratproteine (SNAP-25 und VAMP-2) sind an farbkodierte Beads gekoppelt. Durch die Zugabe von passenden monoklonalen Antikörpern, die an die gespaltenen Substratproteine binden, kann die BoNT- Konzentration anhand der Fluoreszenz abgeleitet werden. Zwei Toxin Anreicherungsstrategien wurden in dieser Arbeit für den Endopep-SIA angewendet – die Anreicherung mit monoklonalen Antikörpern und jene mit rekombinanten Rezeptorstrukturen. Die Nachweisgrenzen für die jeweiligen Lebensmittelmatrizes wurden anhand der Leerwertgrenzen bestimmt. Durch diese Bestimmung sind Unterschiede zu den experimentell gewonnenen Daten hinsichtlich der Sensitivität zu beobachten. Als Grundlage für die Versuche wurde die DIN EN ISO 16140-2:2016 verwendet, jedoch mit Abweichungen, da es sich bei dem Verfahren dieser Arbeit um eine immunologische Methode mit hochtoxischen Analyten handelt. Im Rahmen dieser Studie wurden zum einen direkt mit Toxin kontaminierte und zum anderen auch mit BoNT-produzierenden Clostridium spp. Sporen kontaminierte Proben hergestellt. Es wurden darüber hinaus Feldproben aus vorherigen Botulismus Ausbrüchen analysiert. Die Ergebnisse der Studie zeigen, dass eine Einführung dieses Verfahrens zum Nachweis von BoNT/A und BoNT/B aus unterschiedlichen Lebensmittelproben in die Routinediagnostik generell möglich und erfolgsversprechend ist. Vor allem hat die Anreicherung mit monoklonalen Antikörpern reproduzierbare Ergebnisse geliefert. Hinsichtlich der Sensitivität besteht sicherlich noch Verbesserungsbedarf, wobei dieses auch eine Frage der Art der Kontamination und der Probenextraktion ist. Parallel zu der Validierung wurden mikrobiologischen Studien an den mit Sporen kontaminierten Lebensmittelproben durchgeführt. Dabei wurde die fakultativ anaerobe Begleitflora sowie die Verteilung von BoNT-produzierenden Clostridium spp. innerhalb von geräuchertem Forellenfilet untersucht. Dabei ließ sich feststellen, dass ein gegenseitiges Beeinflussen und Unterdrücken des Wachstums von BoNT-produzierenden Clostridium spp. auf ELA-Platten und der fakultativ anaeroben Begleitflora auf MYP-Agar vorliegt. Dies führt zu der Überlegung, geeignete Schutzkulturen zur Verhinderung des Auskeimens und Wachstums von BoNT-produzierenden Clostridium spp. einzusetzen. Zusätzlich scheint die Homogenisierung einer Probe mit BoNT-produzierenden Clostridium spp. vermehrt die fakultativ anaerobe Begleitflora zu unterdrücken, als dies bei einer nicht homogenisierten Probe der Fall ist. Die gesamte Forschungsarbeit stellt einen großen Fortschritt im Bereich der Diagnostik von Botulinum Neurotoxin Typ A und B aus unterschiedlichen Lebensmittelmatrizes dar und soll ermöglichen, dass dieses Verfahren nach und nach in die Lebensmittelroutinediagnostik integriert werden kann.

This dissertation represents the first comprehensive potmark dataset for the Bronze Age in Lebanon. It analyses potmarks dating to the Early, Middle and Late Bronze Ages, and collects a sample deriving from a wider geographic area, including Levantine marked vessels found in Egypt and marked foreign vessels found in Lebanon. It processes the data in a novel way in the field of potmarking studies, with the use of computational and statistical approaches, integrated in the establishment of the Lebanon Potmark Database (LPD).

In the first part, this work presents the history of research on potmarks in the Near East, Egypt and the eastern Mediterranean, discusses other systems of notation on various media and across a wider geographic region, and touches on the debate distinguishing between graphic systems and writing. In the second part, this dissertation introduces potmarks from fifteen excavated Bronze Age sites in Lebanon and two surveyed areas, with the majority of potmarks representing unpublished material and having been processed first-hand in the work leading to the dissertation. A set of potmarks originating in Lebanon and found at several sites in Egypt, is added to the assemblage.

Potmarking practices are analysed for each of the periods of the Bronze Age - the Early, Middle and Late Bronze Ages, utilizing cross-referencing of dozens of parameters documented in the course of this work. Correlations between the vessel types, locations on vessels, wares, sizes of potmarks, types of signs, techniques used in the potmark production, complexity of signs, time of production, and contextual data of individual stratigraphic units are examined. In addition, correlations of potmarks to vessel volumes are tested, and the quantities and types of marked versus unmarked vessels are compared. The repertoire of signs from the Lebanon Potmark Database is compared to other potmarking systems in the region, as well as to other graphic systems, and previous potmark interpretations are reconsidered. Potmarks are additionally discussed in relation to writing, and meanings of potmarks are examined in comparison to iconography from other media carrying graphic systems. Possible interpretations of the use of potmarks documented in Lebanon are presented, as well as their diachronic and synchronic uses along other graphic systems, not only in the area of Lebanon, but across a wider region.

The present dissertation provides a new methodology in collection and modelling large-scale data of potmarking systems. It emphasizes the need for the use of digital approaches in such studies. Whereas such approaches are already utilized in the studies of some graphic systems, what is lacking is a comprehensive overreaching analysis, in which various graphic systems would be integrated and correlated. The evidence presented in this work points to a need of such a study, and this dissertation presents a step in this direction.

The performance in a test reflects not only a test-taker's ability but also their test-taking motivation. Test-taking motivation is quantified by the amount of effort a test-taker is willing to expend on the solution of test items. Reduced effort and increased variability in effort compromise the validity of standard test scores that do not take motivational factors into account. One important behavioral aspect of effort is persistence --- the time a test-taker is willing to engage with an item. Low persistence results in premature disengagement which manifests in observable phenomena like rapid guessing or omissions. Leveraging on the increasing availability of response time measures from computerized assessments, a large body of models was developed that jointly model responses and response times to account for persistence in tests. One such class are psychometric extensions of cognitive process models. The aim of this dissertation is the development, implementation and evaluation of two distinct psychometric process models of test-taking engagement. The models describe the item solution process in terms of an information accumulation process that is prematurely interrupted. The Differential Ballistic Accumulator model (DBA, Chapter 2) is a single-process model. It stands out from other models of persistence as the premature interruption is rooted within the solution process. It is assumed that a test-taker stops working on an item when progress becomes too slow. The investigation of the DBA establishes the model's computational feasibility, recommends a Bayesian estimator that incorporates an approximation of the model's likelihood function, and demonstrates its unique ability to account for the conditional dependence of responses and response times. Chapter 3 introduces the Linear Ballistic Accumulator Model for Persistence (LBA-P), which models the solution process as a race between accumulators for progress, misinformation, and disengagement. Its primary contribution is to jointly account for correct solutions, active errors, informed guesses, and item omissions. A key feature is an omission threshold that determines whether disengagement results in an omission or an informed guess. An empirical application to PIAAC data demonstrates that a three-accumulator model with a trait-based omission threshold provides the best fit, suggesting that incorrect responses are actively generated and that the tendency to omit varies individually. Chapter 4 reports an empirical online study with N=1,244 participants that examines the validity of the LBA-P's latent persistence trait. The validation strategy involved assessing convergent/discriminant validity against external measures and evaluating the selective influence on the persistence trait through the experimental manipulation of work pace. Results supported the trait's validity: a behavioral measure of persistence was a strong predictor of the persistence trait, but only weakly related to the ability and error-proneness traits. The speed-instruction manipulation successfully shifted the persistence trait as hypothesized, but also affected other traits, a finding consistent with validation studies of cognitive accumulator models. In a general discussion, these findings are synthesized and the theoretical contributions of the two proposed models are contrasted. It highlights their different conceptualizations of disengagement (progress-dependent in the DBA vs. an independent competing process in the LBA-P) and ability. A key strength of both frameworks is their demonstrated ability to capture complex empirical data patterns and enabling an efficiency view on ability by embedding the solution process in an accumulation framework. The dissertation concludes by discussing the primary limitations and the process assumptions that both models rely on. Finally, it outlines a future research agenda focused on further model development, an encouragement to explore different assumptions and specifications, and an increased effort to further validate the models' process assumptions by using more informative process data.

Blockchain and Artificial Intelligence (AI) are transformative technologies for enhancing safety, efficiency, and intelligence in modern railway systems integrated with the European Train Control System (ETCS). This research explores key applications: automatic train scheduling leveraging blockchain for secure and efficient resource allocation; real-time train control utilizing permissioned blockchain to enable collaborative movement authority generation; and parametric settlement of operational costs ensuring transparent and tamper-resistant payment processes via distributed oracle networks.

However, integrating these technologies into critical railway infrastructure introduces significant security and robustness challenges. Key vulnerabilities include threats to data integrity for smart contracts and oracles, potential system misinterpretations, and the inherent heterogeneity (data, device, task) in distributed edge computing environments that can compromise AI model performance and reliability.

To address these critical concerns, we investigate robust solutions. For securing data inputs and ensuring trustworthiness, we propose advanced detection mechanisms and privacy-preserving encryption techniques within the blockchain framework. To manage the challenges of distributed AI in heterogeneous environments, we develop a blockchain-based asynchronous federated learning framework specifically designed for efficient and resilient multi-task learning at the edge. These combined security measures are essential for ensuring the safe, reliable, and trustworthy deployment of blockchain and AI within next-generation railway systems.

The increasing availability of wireless interfaces and software integration facilitates attacker access to vehicular communication networks and the exploitation of vulnerabilities in Electronic Control Units (ECUs), enabling remote manipulation of critical vehicle functions. A significant enabler is the lack of intrinsic security mechanisms in the Controller Area Network (CAN). As security measures based on Message Authentication Codes (MACs) are often impractical for resource-constrained automotive systems, recent Intrusion Detection Systems (IDSs) leverage physical characteristics of analog transmission signals to identify the transmitting ECU and detect unauthorized messages. Consequently, lightweight, real-time-capable approaches that satisfy automotive cost and resource constraints are required.

This work addresses lightweight physical-based IDSs in their overall scope. First, the origins of variations in analog signal characteristics are analyzed through literature review and component data-sheet analysis. Concurrently, automotive-specific constraints and requirements are derived to define design criteria and enable systematic comparison of security approaches.

The core contribution is the development and evaluation of three lightweight IDS/IDPS approaches, each exploiting a distinct physical characteristic. VALID detects unauthorized transmissions via voltage levels on the CAN bus, achieving a 99.54% detection rate without complex sampling or feature extraction and enabling implementation on existing vehicle hardware. ASSASSIN exploits temporal features, including clock skew and signal rise/fall times, providing real-time classification, temperature robustness, and a 99.02% detection rate. SPARTA measures signal arrival differences to identify ECUs and implements active prevention, achieving a 100% detection rate while meeting real-time and resource constraints.

Evaluation on embedded platforms using traffic from a prototype CAN-FD bus and real vehicles demonstrates maturity and practical deployability. Overall, physical-based IDSs provide an efficient solution for authenticity assessment in in-vehicle networks while meeting stringent automotive requirements.

This thesis presents an in-depth description and analysis of an ultrafast electron diffraction (UED) instrument, with a focus on the optical configuration and the components that generate, propagate, and detect the electron beam. We have successfully produced a stable, high-quality electron beam with sufficient transverse coherence length, low emittance, and suitable beam diameters at both the sample and detector stages, facilitating high-resolution reciprocal-space measurements with standard samples. We employ direct electron detection with a high repetition rate laser system to achieve the same electron flux (10⁶ electrons/second) as compared to existing UED setups, but with significantly lower bunch charges, while detecting weak electron signals with a much higher signal-to-noise ratio (SNR) than traditionally employed UED detection systems. Furthermore, we have demonstrated the capabilities of direct electron detection, in which each pixel operates independently, even when some become saturated. The combined use of this detector with advanced correction algorithms enables real-time compensation for intensity and pointing fluctuations, significantly improving data quality. These advancements provide a robust foundation for performing time-resolved UED measurements of atomic-scale structural dynamics.

To evaluate the temporal resolution of our instrument, we employed laser-induced plasma lensing techniques with a temporal resolution of one-to-a-few picoseconds. Our fluence-dependent experiments have demonstrated the ability to track transient field dynamics at high repetition rates by using a narrow, convergent electron beam at the interaction site, resulting in positive signal intensity variations as the beam focuses in response to the transient field. Additionally, experiments with varying bunch charges highlight the low-noise performance of our direct electron detection system. Ultrafast dynamic phenomena with temporal resolution on the order of a few hundred femtoseconds are demonstrated through non-thermal laser-induced heating experiments, which are commonly used to elucidate transient lattice distortions in simple metallic systems. Particularly, we measured the differential diffraction signal from a photoexcited aluminum thin film. Further analysis of the mean-squared displacement (MSD) at lower diffraction orders suggests the presence of multiple long-range phonon-mediated energy-dissipation mechanisms, associated with shorter thermalization times than those observed in higher-order diffraction planes. Crucially, differentiating the power spectral density (PSD) components across diffraction planes leads us to infer that specific phonon branches interact differently at various diffraction orders, which is consistent with the Non-Thermal Lattice Model as discussed in the literature. Our study delineates the spatial range of dynamics explicitly by examining diffraction from each interplanar distance individually, with high sensitivity and temporal resolution without temporal compression.

Finally, we performed a preliminary characterization of a radio-frequency (RF) electron-pulse compression system. The RF synchronization jitter is optimized to a sub-20 fs RMS level, which is favorable for the estimated IRF (184 fs). Additionally, the RF compression is indirectly confirmed through measurements of the change in the electron beam transverse diameter as a function of the RF cavity power.

The tribe Cunonieae (Cunoniaceae) comprises five genera and ca. 214 species of shrubs and trees currently distributed in the Southern Hemisphere and the tropics, exhibiting an amphi-Pacific disjunct distribution. This doctoral dissertation presents a comprehensive molecular phylogenetic and biogeographic study of the tribe Cunonieae, a key group for understanding the assembly of Southern Hemisphere flora. Employing a multi-faceted approach with plastid markers, nuclear ITS, 2bRADseq data, and complete plastid genomes, this research reconstructs the evolutionary history of the tribe, with a particular focus on the large and ecologically important genus Weinmannia. The thesis findings provide a robust temporal and spatial framework for the Cunonieae and Weinmannia diversification. The study establishes for the first time a densely sampled phylogenetic framework for Cunonieae and supports the recent resurrection of the genus Pterophylla for most Old World species formerly placed in Weinmannia, which is now largely restricted to the Americas (ca. 90 species plus two species in the Mascarenes). However, the monophyly of Cunonia is challenged, with Cunonia capensis (South Africa) appearing distantly related to the New Caledonian species. A time-calibrated phylogeny and a biogeographical reconstruction were developed using fossil-calibration points. Results indicate the crown group of Cunonieae originated in the early Eocene (~56 Ma) with ancestral ranges in Antarctica and Patagonia. Antarctica acted as a biogeographic bridge between Oceania and South America prior to the formation of the Antarctic Ice Sheet, which led to the lineage's extinction on the continent. This was followed by diversification in the Andes, Madagascar, and New Caledonia during the Pliocene, coinciding with major geological events like the final uplift of the Andes. A major contribution is the generation of 52 new plastid genomes for Weinmannia, which provide a highly resolved phylogenomic tree that renders old infrageneric classifications obsolete. This genomic depth uncovered evidence of possible past hybridization and chloroplast capture, exemplified by the genomic relationships and distances of W. auriculata and W. auriculifera. Conflicting nuclear and plastid signals, particularly in the placement of species like W. boliviensis, further underscore the prevalence of incomplete lineage sorting and introgression. Finally, this thesis includes a thorough nomenclatural revision of Weinmannia, establishing a new classification with 105 accepted species and 21 infraspecific taxa. This checklist, the most current for the genus, provides an essential tool for future systematic, taxonomic, and ecological studies. In conclusion, this research constitutes a decisive advance in the scientific knowledge of the evolutionary systematics and biogeography of Weinmannia, a candidate to become a model group for plant sciences in the Andes.

As societies rely more on online communication, the precise and accurate detection of hate speech and other harmful content becomes a priority to provide safe means of communication online. With new generations growing up in fully digital environments, reliable algorithmic detection of hateful messages not only protects the most vulnerable but also safeguards free expression and supports education on the topic. This algorithmic prediction must be fair, efficient, and generalizable across different social settings, such as age groups. While the performance of hate speech classifiers is steadily improving, fundamental problems remain unaddressed.

The first challenge is the lack of available and representative data. Second, there is no universal definition of hate, as the concept is constantly evolving and shaped by cultural and social contexts. Yet, it remains essential to detect even subtle expressions of hate reliably across different settings. Lastly, current classifiers are rarely assessed for their effectiveness across specific age groups, limiting their applicability to diverse audiences.

This research addresses these challenges through 12 studies, making significant contributions to the field of missing datasets and hate speech detection. The studies achieve this by collecting and annotating novel corpora, identifying safe methods for gathering youth language data, and presenting a solution to capture different definitions of hate among adolescent groups. To enhance the generalizability of hate speech classifiers, the research trains and compares multilingual detection model, proposes an agent-based detection architecture, and introduces an innovative approach to integrating varying definitions of hate into a more efficient, centralized model. This contributes to the issue of fragmented, disjointed annotated datasets. Further, the studies address key challenges in analyzing youth language online. It presents a large language model based method for decoding coded youth language, evaluates strategies for keeping trained classifiers up to date with evolving language trends, and develops an algorithmic approach to classifying youth language. Finally, the research proposes a framework for detecting youth language bias within trained hate speech classifiers, further advancing fairness and accuracy in automated moderation.

The research results advance the field of algorithmic hate speech detection in online communication among adolescents by providing novel datasets and approaches for collection, increasing the generalizability of classifiers, and offering a novel evaluation framework. It addresses the underlying problem of dynamic language in hate speech detection.

This thesis explores the potential of classical higher-order logic (HOL) as a general-purpose meta-language capable of representing and formally reasoning about diverse domain-specific object languages. To demonstrate its versatility, we have developed a library composed of combinator-based modular building blocks within the Isabelle/HOL proof assistant, but easily translatable to other HOL-based systems. We propose that this library exemplifies a universal mathematical language layered over HOL, sufficiently powerful to capture and represent a broad range of logical formalisms—whether classical or non-classical—as well as their underlying classes of mathematical structures. To support this claim, we provide relevant references, primarily drawn from previous research. Moreover, we delve into the domain of multi-agent systems (MAS), introducing the concept of Argumentative Reasoning Agents (ARAs). This kind of agents, we suggest, are particularly well-suited to leverage the expressive power of the proposed library. In this manner, HOL emerges as an ideal common language—a lingua franca—for facilitating effective reasoning and interaction among argumentative reasoning agents in artificial intelligence.