Refine
Has Fulltext
- yes (82)
Is part of the Bibliography
- yes (82)
Year of publication
- 2024 (82) (remove)
Document Type
- Doctoral Thesis (82) (remove)
Language
- English (82) (remove)
Keywords
- Tissue Engineering (6)
- Maschinelles Lernen (4)
- Entzündung (3)
- Induzierte pluripotente Stammzelle (3)
- Nanopartikel (3)
- Simulation (3)
- Bildgebendes Verfahren (2)
- Bioverfügbarkeit (2)
- Immuntherapie (2)
- Kernspintomografie (2)
Institute
- Graduate School of Life Sciences (41)
- Theodor-Boveri-Institut für Biowissenschaften (6)
- Institut für Informatik (5)
- Institut für Psychologie (5)
- Institut für Funktionsmaterialien und Biofabrikation (4)
- Institut für Pharmazie und Lebensmittelchemie (4)
- Institut für Physikalische und Theoretische Chemie (4)
- Neurologische Klinik und Poliklinik (4)
- Deutsches Zentrum für Herzinsuffizienz (DZHI) (3)
- Institut für Organische Chemie (3)
Sonstige beteiligte Institutionen
- Department of Mathematical Analysis, Faculty of Mathematics and Physics, Charles University in Prague (1)
- Deutsches Krebsforschungszentrum Heidelberg (1)
- Deutsches Zentrum für Luft- und Raumfahrt e.V. (1)
- European Space Agency (1)
- Experimental Physics V, University of Wuerzburg (1)
- Fraunhofer Insitut für Silicatforschung ISC (1)
- Fraunhofer-Institute for Silicate Research ISC (1)
- Helmholtz Institut für RNA-basierte Infektionsforschung (1)
- Helmholtz Institute for RNA-based Infection Research (HIRI) (1)
- LMU München (1)
We analyze the mathematical models of two classes of physical phenomena. The first class of phenomena we consider is the interaction between one or more insulating rigid bodies and an electrically conducting fluid, inside of which the bodies are contained, as well as the electromagnetic fields trespassing both of the materials. We take into account both the cases of incompressible and compressible fluids. In both cases our main result yields the existence of weak solutions to the associated system of partial differential equations, respectively. The proofs of these results are built upon hybrid discrete-continuous approximation schemes: Parts of the systems are discretized with respect to time in order to deal with the solution-dependent test functions in the induction equation. The remaining parts are treated as continuous equations on the small intervals between consecutive discrete time points, allowing us to employ techniques which do not transfer to the discretized setting. Moreover, the solution-dependent test functions in the momentum equation are handled via the use of classical penalization methods.
The second class of phenomena we consider is the evolution of a magnetoelastic material. Here too, our main result proves the existence of weak solutions to the corresponding system of partial differential equations. Its proof is based on De Giorgi's minimizing movements method, in which the system is discretized in time and, at each discrete time point, a minimization problem is solved, the associated Euler-Lagrange equations of which constitute a suitable approximation of the original equation of motion and magnetic force balance. The construction of such a minimization problem is made possible by the realization that, already on the continuous level, both of these equations can be written in terms of the same energy and dissipation potentials. The functional for the discrete minimization problem can then be constructed on the basis of these potentials.
This compilation focuses on adolescent mental disorders and their prevention. It comprises three distinct studies, each contributing to a deeper understanding of this critical topic. This work addresses a critical gap in the understanding of, and approach to, adolescent mental health, and as a result reveals a critically important and urgently needed policy implication for action. The thematic structure of these studies begins with an examination of the epidemiology of child and adolescent mental disorders. Baseline data were collected from N = 877 adolescents with a mean age of 12.43 years (SD = 0.65). Mental health problems, such as depressive symptoms, non-suicidal self-injury, suicidal ideation, symptoms of eating disorders, and gender differences, are thoroughly examined. Results revealed a significant portion of our sample displaying mental health problems as early as the 6th and 7th grades, with girls generally being more affected than boys. The findings underscore the importance of early adolescence in the emergence of mental health problems and thereby emphasize the need for preventive measures. Moving beyond prevalence estimates, the compilation delves into the etiology of these disorders, exploring their potential correlation with a COVID-19 infection. Understanding the early signs and risk factors is crucial for timely support. While numerous studies have investigated potential risk and protective factors during the pandemic, our focus shifts to adolescents’ coping when an infection with the virus was involved (N = 2,154, M = 12.31, SD = 0.67). We hypothesized that students infected or with close family members infected, would exhibit an increased psychopathology and a decreased functioning of protective factors such as self-efficacy or self-esteem. We found no connection between infection and the mental health status within our sample, but protective factors and mental well-being were positively associated. Thus, universal primary prevention appears to be the preferred approach for promoting mental health. Lastly, the compilation introduces LessStress, a noteworthy contribution to more evidence-based prevention programs. This universal approach is designed to reduce stress in schools, accompanied by a cluster-randomized trial to evaluate its effectiveness (estimated sample size N = 1,894). Existing studies have demonstrated the effectiveness of stress prevention, leading us to introduce a short and easy-to-implement prevention program. There is positive evidence for one-lesson interventions in schools for promoting well-being and health behaviors among adolescents. LessStress is designed based on a life skills approach that not only imparts psychoeducational content but also teaches skills relevant to everyday life and directly applicable. Throughout these studies, a common thread emerges: the pressing need to address mental disorders during childhood and adolescence. These formative years play a pivotal role in the development of mental health problems. These formative years play a crucial role in the development of mental health problems. They highlight the importance of epidemiological data collection and analysis based on the latest models to develop prevention interventions that are not only effective but also reach young people on a global level.
Ownership and usage of personal voice assistant devices like Amazon Echo or Google Home have increased drastically over the last decade since their market launch. This thesis builds upon existing computers are social actors (CASA) and media equation research that is concerned with humans displaying social reactions usually exclusive to human-human interaction when interacting with media and technological devices. CASA research has been conducted with a variety of technological devices such as desktop computers, smartphones, embodied virtual agents, and robots. However, despite their increasing popularity, little empirical work has been done to examine social reactions towards these personal stand-alone voice assistant devices, also referred to as smart speakers. Thus, this dissertation aims to adopt the CASA approach to empirically evaluate social responses to smart speakers. With this goal in mind, four laboratory experiments with a total of 407 participants have been conducted for this thesis. Results show that participants display a wide range of social reactions when interacting with voice assistants. This includes the utilization of politeness strategies such as the interviewer-bias, which led to participants giving better evaluations directly to a smart speaker device compared to a separate computer. Participants also displayed prosocial behavior toward a smart speaker after interdependence and thus a team affiliation had been induced. In a third study, participants applied gender stereotypes to a smart speaker not only in self-reports but also exhibited conformal behavior patterns based on the voice the device used. In a fourth and final study, participants followed the rule of reciprocity and provided help to a smart speaker device that helped them in a prior interaction. This effect was also moderated by subjects’ personalities, indicating that individual differences are relevant for CASA research. Consequently, this thesis provides strong empirical support for a voice assistants are social actors paradigm. This doctoral dissertation demonstrates the power and utility of this research paradigm for media psychological research and shows how considering voice assistant devices as social actors lead to a more profound understanding of voice-based technology. The findings discussed in this thesis also have implications for these devices that need to be carefully considered both in future research as well as in practical design.
Acceleration is a central aim of clinical and technical research in magnetic resonance imaging (MRI) today, with the potential to increase robustness, accessibility and patient comfort, reduce cost, and enable entirely new kinds of examinations. A key component in this endeavor is image reconstruction, as most modern approaches build on advanced signal and image processing. Here, deep learning (DL)-based methods have recently shown considerable potential, with numerous publications demonstrating benefits for MRI reconstruction. However, these methods often come at the cost of an increased risk for subtle yet critical errors. Therefore, the aim of this thesis is to advance DL-based MRI reconstruction, while ensuring high quality and fidelity with measured data. A network architecture specifically suited for this purpose is the variational network (VN). To investigate the benefits these can bring to non-Cartesian cardiac imaging, the first part presents an application of VNs, which were specifically adapted to the reconstruction of accelerated spiral acquisitions. The proposed method is compared to a segmented exam, a U-Net and a compressed sensing (CS) model using qualitative and quantitative measures. While the U-Net performed poorly, the VN as well as the CS reconstruction showed good output quality. In functional cardiac imaging, the proposed real-time method with VN reconstruction substantially accelerates examinations over the gold-standard, from over 10 to just 1 minute. Clinical parameters agreed on average.
Generally in MRI reconstruction, the assessment of image quality is complex, in particular for modern non-linear methods. Therefore, advanced techniques for precise evaluation of quality were subsequently demonstrated.
With two distinct methods, resolution and amplification or suppression of noise are quantified locally in each pixel of a reconstruction. Using these, local maps of resolution and noise in parallel imaging (GRAPPA), CS, U-Net and VN reconstructions were determined for MR images of the brain. In the tested images, GRAPPA delivers uniform and ideal resolution, but amplifies noise noticeably. The other methods adapt their behavior to image structure, where different levels of local blurring were observed at edges compared to homogeneous areas, and noise was suppressed except at edges. Overall, VNs were found to combine a number of advantageous properties, including a good trade-off between resolution and noise, fast reconstruction times, and high overall image quality and fidelity of the produced output. Therefore, this network architecture seems highly promising for MRI reconstruction.
Biological systems are in dynamic interaction. Many responses reside in the core concepts of biological systems interplay (competition and cooperation). In infection situation, the competition between a bacterial system and a host is shaped by many stressors at spatial and temporal determinants. Reactive chemical species are universal stressors against all biological systems since they potentially damage the basic requirements of these systems (nucleic acids, proteins, carbohydrates, and lipids). Either produced endogenously or exogenously, reactive chemical species affect the survival of pathogens including the gram-positive
Staphylococcus aureus (S. aureus). Therefore, bacteria developed strategies to overcome the toxicity of reactive species.
S. aureus is a widely found opportunistic pathogen. In its niche, S. aureus is in permanent contact with surrounding microbes and host factors. Deciphering the deterministic factors
in these interactions could facilitate pinpointing novel bacterial targets. Identifying
the aforementioned targets is crucial to develop new strategies not only to kill the pathogenic organisms but also to enhance the normal flora to minimize the pathogenicity and virulence of potential pathogens. Moreover, targeting S. aureus stress response can be used
to overcome bacterial resistance against host-derived factors. In this study, I identify a novel
S. aureus stress response factor against reactive electrophilic, oxygen, and hypochlorite species to better understand its resilience as a pathogen.
Although bacterial stress response is an active research field, gene function is a current bottleneck in characterizing the understudied bacterial strategies to mediate stress conditions. I aimed at understanding the function of a novel protein family integrated
in many defense systems of several biological systems.
In bacteria, fungi, and plants, old yellow enzymes (OYEs) are widely found. Since the first isolation of the yellow flavoprotein, OYEs are used as biocatalysts for decades to reduce activated C=C bonds in α,β-unsaturated carbonyl compounds. The promiscuity
of the enzymatic catalysis is advantageous for industrial applications.
However, the physiological function of OYEs, especially in bacteria, is still puzzling.
Moreover, the relevance of the OYEs in infection conditions remained enigmatic.
Here, I show that there are two groups of OYEs (OYE flavin oxidoreductase, OfrA and OfrB) that are encoded in staphylococci and some firmicutes. OfrA (SAUSA300_0859) is more conserved than OfrB (SAUSA300_0322) in staphylococci and is a part of the staphylococcal core genome.
A reporter system was established to report for ofrA in S. aureus background.
The results showed that ofrA is induced under electrophilic, oxidative, and hypochlorite stress. OfrA protects S. aureus against quinone, methylglyoxal, hydrogen peroxide,
and hypochlorite stress. Additionally, the results provide evidence that OfrA supports
thiol-dependent redox homeostasis. At the host-pathogen interface, OfrA promotes S. aureus fitness in murine macrophage cell line. In whole human blood, OfrA is involved in S. aureus survival indicating a potential clinical relevance to bacteraemia.
In addition, ofrA mutation affects the production of the virulence factor staphyloxanthin via the upper mevalonate pathway. In summary, decoding OfrA function and its proposed mechanism of action in S. aureus shed the light on a conserved stress response within multiple organisms.
Wireless communication networks already comprise an integral part of both the private and industrial sectors and are successfully replacing existing wired networks. They enable the development of novel applications and offer greater flexibility and efficiency. Although some efforts are already underway in the aerospace sector to deploy wireless communication networks on board spacecraft, none of these projects have yet succeeded in replacing the hard-wired state-of-the-art architecture for intra-spacecraft communication. The advantages are evident as the reduction of the wiring harness saves time, mass, and costs, and makes the whole integration process more flexible. It also allows for easier scaling when interconnecting different systems.
This dissertation deals with the design and implementation of a wireless network architecture to enhance intra-spacecraft communications by breaking with the state-of-the-art standards that have existed in the space industry for decades. The potential and benefits of this novel wireless network architecture are evaluated, an innovative design using ultra-wideband technology is presented. It is combined with a Medium Access Control (MAC) layer tailored for low-latency and deterministic networks supporting even mission-critical applications. As demonstrated by the Wireless Compose experiment on the International Space Station (ISS), this technology is not limited to communications but also enables novel positioning applications.
To adress the technological challenges, extensive studies have been carried out on electromagnetic compatibility, space radiation, and data robustness. The architecture was evaluated from various perspectives and successfully demonstrated in space.
Overall, this research highlights how a wireless network can improve and potentially replace existing state-of-the-art communication systems on board spacecraft in future missions. And it will help to adapt and ultimately accelerate the implementation of wireless networks in space systems.
Virtual humans (VHs) hold immense potential for collaboration in social virtual reality (VR). As VR technology advances, it's vital to assess the psychological effects on VH trust and user privacy to build meaningful social interactions in VR. In social VR, users must be able to trust the VHs they interact with as they navigate through socio-cultural activities. The evaluation of trustworthiness in VHs profoundly impacts interaction quality and user willingness to engage. Conversely, untrustworthy VHs can harm user experiences, privacy, and VR engagement. To address this, we conducted immersive VR studies, exploring how psychological factors influence user's VH trust evaluation under various psychological conditions. This research is pivotal for developing strategies to enhance user privacy, establish secure VR environments, and create a foundation of trust that supports immersive socio-cultural experiences in VR.
To date, there are no established interpersonal trust measurement tools specifically for VHs in VR. In study 1 (the familiarity study) of the current thesis the VR-adjusted version of the social conditioned place preference paradigm (SCPP) by Kiser et al., (2022) was identified as a potential trust measurement tool. We tested whether the familiarity of a VH influenced trust as measured with the SCPP paradigm and other self-defined outcome measures, in a Computer Augmented Virtual Environment (CAVE). The CAVE is a VR system that combines immersive VR with real-world elements. It consists of a room-sized space where the walls are used as projection screens to display virtual scenes and objects. In this within - subject design (n = 20), half of the participants were familiarized with one VH and tasked to explore and interact in a realistic looking virtual art museum environment. The participant’s evaluation of the VH’s trustworthiness was measured as well as their subsequent trust behaviours. Results revealed no significant differences in the evaluation of the VH’s trustworthiness nor any behavioural differences between conditions. The findings of the impact of a VH’s familiarity on trust is inconclusive due to the major limitations of the paradigm. We concluded that the SCPP paradigm needs further validation and the proposed proxies of trust need to be re-evaluated. The findings were considered in the following study.
The virtual maze paradigm design of Hale, (2018) was identified as a potential trust measurement tool, however several limitations are associated with its use to measure trust in VR. In study 2 (a validation study), improvements were made to the virtual maze paradigm of Hale, (2018) and a variant of this paradigm was implemented. We conducted a validation study with 70 participants in a between-subject design with VH trustworthiness as the between-subject factor. Participants wore a head-mounted display (HMD), to deliver an immersive VR experience. In our version of the virtual maze, it was the task of the users (the trustors) to navigate through a maze in VR, where they could interact with a VH (the trustee). They could choose to ask for advice and follow the advice from the VH if they wanted to. The number of times participants asked and followed advice and the time it took to respond to the given advice served as behavioural proxies/measures of trust. The two conditions (trustworthy vs. untrustworthy) did not differ in the content of the advice but in the appearance, tone of voice and engagement of the trustees (allegedly an avatar controlled by other participants). Results indicated that the experimental manipulation was successful, as participants rated the VH as more trustworthy in the trustworthy condition compared with the VH in the untrustworthy condition. Importantly, this manipulation affected the trust behaviour of participants, who, in the trustworthy condition, asked for advice and followed advice more often, indicating that the paradigm is sensitive to differences in VH’s trustworthiness. Thus, our paradigm can be used to measure differences in interpersonal trust towards VHs and may serve as a valuable research tool for researchers who study trust in VR. Therefore, study 2 fills the gap in the literature, for an interpersonal trust measurement tool specifically for VHs in VR.
Two experimental studies, with a sample size of 50 participants each, utilized the virtual maze paradigm where participants entered 12 rooms under different conditions. We examined the influence of cognitive load (CL) on trust towards VH in VR in study 3 (Cognitive load study), and the influence of emotional affect (Emotional affect study) on trust towards VH in VR in study 4 (EA study). In both studies, we assessed participant’s evaluation of a VH’s trustworthiness, along with three behavioural indicators of trust in the maze task: 1) frequency of advice asked, 2) frequency of advice followed, and 3) the time taken by participants to execute the received advice. In study 3, the CL was manipulated with the auditory 1-back task in the high cognitive load condition (HCL). In study 4, the Autobiographical Emotional Memory Task (AEMT) was used to manipulate the EA of participants in the negative emotional affect (NEA) condition. As an additional manipulation, while participants were immersed in VR, they were exposed to 12 negative pictures and sounds that was presented simultaneously to strengthen the initial manipulation. The manipulation of the within-subject factors (CL and EA) was successful in both studies, as significant differences between conditions were observed in both studies (higher CL in the HCL condition and a more negative EA in the NEA condition). However, only CL influenced participant’s evaluation of the VH’s trustworthiness. The VH were evaluated as significantly more trustworthy after the HCL condition. Despite the difference in trust evaluation, there was no difference in advice asking or following. Participants in study 4 asked and followed advice due to their trust in the VH and asked and followed advice equally often in both conditions. Importantly, significant differences were observed in the participants response times in both studies. In study 3 during the HCL condition participants followed advice quicker. The order in which the conditions were presented influenced the experience of CL. Participants experienced higher levels of CL and responded to advice significantly faster when low cognitive load (LCL) was presented as the first condition compared with LCL as the second condition. In study 4 participants in the NEA condition followed advice slower similar to the findings of study 3. The order in which the conditions were presented had a significant effect on the EA. Participants asked and followed advice less when the NEA condition was presented first compared with when it is presented second. Possible explanations for the findings are discussed in the thesis.
Overall, this thesis offers a novel tool for trust measurement (the virtual maze paradigm) and contributes to understanding the role of psychological factors in trust towards virtual humans in virtual reality.
Anxiety patients overgeneralize fear, also because of an inability to perceptually discriminate threat and safety signals. Therefore, some studies have developed discrimination training that successfully reduced the occurrence of fear generalization. The present work is the first to take a treatment-like approach by using discrimination training after generalization has occurred. Therefore, two studies were conducted with healthy participants using the same fear conditioning and generalization paradigm, with two faces as conditioned stimuli (CSs), and four facial morphs between CSs as generalization stimuli (GSs). Only one face (CS+) was followed by a loud scream (unconditioned stimulus, US). In Study 1, participants underwent either fear-relevant (discriminating faces) or fear-irrelevant discrimination training (discriminating width of lines) or a non-discriminative control training between the two generalization tests, each with or without feedback (n = 20 each). Generalization of US expectancy was reduced more effectively by fear-relevant compared to fear-irrelevant discrimination training. However, neither discrimination training was more effective than non-discriminative control training. Moreover, feedback reduced generalization of US expectancy only in discrimination training. Study 2 was designed to replicate the effects of the discrimination-training conditions in a large sample (N = 244) and examine their benefits in individuals at risk for anxiety disorders. Again, feedback reduced fear generalization particularly well for US expectancy. Fear relevance was not confirmed to be particularly fear-reducing in healthy participants, but may enhance training effects in individuals at risk of anxiety disorder. In summary, this work provides evidence that existing fear generalization can be reduced by discrimination training, likely involving several (higher-level) processes besides perceptual discrimination (e.g., motivational mechanisms in feedback conditions). Its use may be promising as part of individualized therapy for patients with difficulty discriminating similar stimuli.
RNA viruses rely entirely on the host machinery for their protein synthesis and harbor non-canonical translation mechanisms, such as alternative initiation and programmed –1 ribosomal frameshifting (–1PRF), to suit their specific needs. On the other hand, host cells have developed a variety of defensive strategies to safeguard their translational apparatus and at times transiently shut down global translation. An infection can lead to substantial translational remodeling in cells and translational control is critical during antiviral response. Due to their sheer diversity, this control is likely unique to each RNA virus and the intricacies of post-transcriptional regulation are unclear in certain viral species.
Here, we explored different aspects of translational regulation in virus-infected cells in detail. Using ribosome profiling, we extensively characterized the translational landscape in HIV-1 infected T cells, uncovering novel features of gene regulation in both host and virus. Additionally, we show that substantial pausing occurs prior to the frameshift site indicating complex regulatory mechanisms involving upstream viral RNA elements that can act as cis- regulators of frameshifting.
We also characterized the mechanistic details of trans- modulation of frameshifting by host- and virus-encoded proteins. Host antiviral protein ZAP-S binds to the SARS-CoV-2 frameshift site and destabilizes the stimulatory structure, leading to frameshift inhibition. On the other hand, EMCV 2A protein stabilizes the viral frameshift site, thereby, activating EMCV frameshifting. While both proteins were shown to be antagonistic in their mechanism, they interact with the host translational machinery. Furthermore, we showed that frameshifting can be regulated not just by proteins, but also by small molecules. High-throughput screening of natural and synthetic compounds identified two potent frameshift inhibitors that also impeded viral replication, namely trichangion and compound 25. Together, this work largely enhances our understanding of gene regulation mechanisms in virus-infected cells and further validates the druggability of viral –1 PRF site.
The goal of this thesis is to study the topological and algebraic properties of the quasiconformal automorphism groups of simply and multiply connected domains in the complex plain, in which the quasiconformal automorphism groups are endowed with the supremum metric on the underlying domain. More precisely, questions concerning central topological properties such as (local) compactness, (path)-connectedness and separability and their dependence on the boundary of the corresponding domains are studied, as well as completeness with respect to the supremum metric. Moreover, special subsets of the quasiconformal automorphism group of the unit disk are investigated, and concrete quasiconformal automorphisms are constructed. Finally, a possible application of quasiconformal unit disk automorphisms to symmetric cryptography is presented, in which a quasiconformal cryptosystem is defined and studied.
Deep Learning (DL) models are trained on a downstream task by feeding (potentially preprocessed) input data through a trainable Neural Network (NN) and updating its parameters to minimize the loss function between the predicted and the desired output. While this general framework has mainly remained unchanged over the years, the architectures of the trainable models have greatly evolved. Even though it is undoubtedly important to choose the right architecture, we argue that it is also beneficial to develop methods that address other components of the training process. We hypothesize that utilizing domain knowledge can be helpful to improve DL models in terms of performance and/or efficiency. Such model-agnostic methods can be applied to any existing or future architecture. Furthermore, the black box nature of DL models motivates the development of techniques to understand their inner workings. Considering the rapid advancement of DL architectures, it is again crucial to develop model-agnostic methods.
In this thesis, we explore six principles that incorporate domain knowledge to understand or improve models. They are applied either on the input or output side of the trainable model. Each principle is applied to at least two DL tasks, leading to task-specific implementations. To understand DL models, we propose to use Generated Input Data coming from a controllable generation process requiring knowledge about the data properties. This way, we can understand the model’s behavior by analyzing how it changes when one specific high-level input feature changes in the generated data. On the output side, Gradient-Based Attribution methods create a gradient at the end of the NN and then propagate it back to the input, indicating which low-level input features have a large influence on the model’s prediction. The resulting input features can be interpreted by humans using domain knowledge.
To improve the trainable model in terms of downstream performance, data and compute efficiency, or robustness to unwanted features, we explore principles that each address one of the training components besides the trainable model. Input Masking and Augmentation directly modifies the training input data, integrating knowledge about the data and its impact on the model’s output. We also explore the use of Feature Extraction using Pretrained Multimodal Models which can be seen as a beneficial preprocessing step to extract useful features. When no training data is available for the downstream task, using such features and domain knowledge expressed in other modalities can result in a Zero-Shot Learning (ZSL) setting, completely eliminating the trainable model. The Weak Label Generation principle produces new desired outputs using knowledge about the labels, giving either a good pretraining or even exclusive training dataset to solve the downstream task. Finally, improving and choosing the right Loss Function is another principle we explore in this thesis. Here, we enrich existing loss functions with knowledge about label interactions or utilize and combine multiple task-specific loss functions in a multitask setting.
We apply the principles to classification, regression, and representation tasks as well as to image and text modalities. We propose, apply, and evaluate existing and novel methods to understand and improve the model. Overall, this thesis introduces and evaluates methods that complement the development and choice of DL model architectures.
Dipolar merocyanines are very attractive supramolecular building blocks, as they combine interesting functional properties with strong, directional intermolecular interactions. The pyridine dioxocyano-pyridine (PYOP) chromophore (Chapter 2.2), used in this thesis, stands out because of its exceptionally high ground state dipole moment (g ~ 17 D), in combination with the option to retain good solubility also in unpolar solvents, by decoration with solubilizing groups.
The reliable binding motif of anti-parallel -stacking due to dipole-dipole interactions has allowed the design of molecular building blocks that form assemblies of predictable geometry. The intense unstructured charge transfer UV/Vis absorption band (eg ~ 10.7 D) is a result of the dominant contribution of the zwitterionic resonance structure which brings the PYOP chromophore just beyond the cyanine limit in solvents of low polarity (c2 = 0.60, 1,4 dioxane). The high sensitivity of the S0 – S1 UV/Vis absorption band to the environment manifests itself in a pronounced negative solvatochromism and strong H-type exciton coupling within -stacked PYOP assemblies. In accordance with the classical molecular exciton theory, an increasing hypsochromic shift of the dominant absorption band of these H aggregates can be observed as the stack size increases up to about six chromophores, where it levels out at about max ~ 440 nm (CHCl3). This allows a uniquely simple estimation of the number of interacting chromophores within the self-assembled structure from a single UV/Vis absorption spectrum of an aggregate.
The defined and well investigated PYOP dimer formation was employed in this thesis to probe the applicability and limitations of concentration-, temperature-, and solvent-dependent self-assembly studies (Chapter 3). Straightforward theoretical models to evaluate datasets of concentration-, temperature-, and solvent-dependent UV/Vis absorption by nonlinear regression analysis were derived for the case of dimer formation (Chapter 2.1). Although the dimer model is well known and widely applied in literature, this detailed derivation is helpful to understand assumptions and potential problems of the different approaches for the determination of thermodynamic parameters. This helps to decide on the most appropriate method to analyse a system of interest. In this regard it should be noted that covering a large portion of the self-assembly process with the experimental data is a prerequisite for the accuracy of the analysis. Additionally, many of the insights can also be transferred to other self-assembly systems like supramolecular polymerization or host-guest interactions.
The concentration-dependent analysis is the most straightforward method to investigate self-assembly equilibria. No additional assumptions, besides mass balance and mass action law, are required. Since it includes the least number of parameters (only K, if M/D are known), it is the most, or even only, reliable method, to elucidate the self-assembly mechanism of an unknown system by model comparison. To cover a large concentration range, however, the compound must be soluble enough and generally sample amounts at least in the low mg scale must be available.
The temperature-dependent analysis has the advantage that all thermodynamic parameters G0, H0 and S0 can be obtained from a single sample in one automated measurement. However, the accessible temperature-range is experimentally often quite limited and dependent on the solvent. For systems which do not show the transition from monomer to aggregate in a narrow temperature range, as given for, e.g., cooperative aggregation or processes with a high entropy contribution, often not the entire self-assembly process can be monitored. Furthermore, the assumptions of temperature-independent extinction coefficients of the individual species as well as temperature-independent H0 and S0 must be met. Monte Carlo simulations of data sets demonstrated that even minor changes in experimental data can significantly impact the optimized values for H0 and S0. This is due to the redundancy of these two parameters within the model framework and even small thermochromic effects can significantly influence the results. The G0 value, calculated from H0 and S0, is, however, still rather reliable.
Solvent-dependent studies can often cover the entire self-assembly process from monomeric (agg = 0) to the fully aggregated state (agg = 1). However, for dyes with strong solvatochromic effects, such as the dipolar merocyanines investigated in this thesis, the results are affected. Also, the assumption of a linear relation of the binding energy G0 and the fraction of denaturating solvent f, which is based on linear free energy relationships between G0 and the solvent polarity, can lead to errors. Especially when specific solvent effects are involved.
For the evaluation of experimental data by nonlinear regression, general data analysis software can be used, where user-defined fit models and known parameters can be implemented as desired. Alternatively, multiple specialized programs for analysing self-assembly data are available online. While the latter programs are usually more user-friendly, they have the disadvantage of being a “black box” where only pre-implemented models can be used without the option for the user to adapt models or parameters for a specific system.
In Chapter 3 comprehensive UV/Vis absorption datasets are presented for the dimerization of merocyanine derivative 1 in 1,4-dioxane, which allowed for the first time a direct comparison of the results derived from concentration-, temperature-, and solvent-dependent self-assembly studies.
The results for the binding constant K and corresponding G0 from the concentration- and temperature-dependent analysis were in very good agreement, also in comparison to the results from ITC. For the temperature-dependent analysis, though, multiple datasets of samples with different concentration had to be evaluated simultaneously to cover a meaningful part of the self-assembly process. Furthermore, a significant dependence of the optimized parameters H0 and S0 on the wavelength chosen for the analysis was observed. This can be rationalized by the small thermochromic shifts of both the monomer and the dimer UV/Vis absorption band. The results from the solvent-dependent evaluation showed the largest deviation, as expected for the highly solvatochromic merocyanine dye.
However, even here by evaluation at 491 and 549 nm the deviation for G0 was only 2.5 kJ mol1 (9%) with respect to the results from the concentration-dependent analysis (G0 = 29.1 kJ mol1). Thus, despite the strong solvatochromism of the dipolar chromophore, it can still be considered a reliable method for estimating the binding strength. Furthermore, multiple repetitions of the concentration-, temperature-, and solvent-dependent studies provided insight into the reproducibility of the results and possible sources of experimental errors. In all cases, the deviations of the results were small (G0 < 0.4 kJ mol1) and within the same range as the fit error from the nonlinear regression analysis.
The insights from these studies were an important basis for the in-depth investigation of a more complex supramolecular system in Chapter 4, as a single method is often not enough to capture the full picture of a more complicated self-assembly process. To elucidate the anti-cooperative self-assembly of the chiral merocyanine 2, a combination of multiple techniques had to be applied.
Solvent-dependent UV/Vis absorption studies in CH2Cl2/MCH mixtures showed the step-wise assembly of the merocyanine monomer (max(M) = 549 nm, CH2Cl2) to first a dimer (max(D) = 498 nm, CH2Cl2/MCH 15:85) by dipole-dipole interactions, and then a -stacked higher aggregate (max(H) = 477 nm, MCH), with pronounced H-type coupling.
The thermodynamic evaluation of this data, however, suffered from the severe solvatochromism, especially of the monomeric species (max(M, CH2Cl2) = 549 nm, max(M, MCH) = 596 nm). Therefore, concentration-dependent studies were performed at three different temperatures (298, 323, 353 K) to elucidate the self-assembly mechanism and determine reliable thermodynamic parameters. The studies at elevated temperatures were hereby necessary, to obtain experimental data over a larger agg--range. Due to the pronounced difference in the thermodynamic driving force for dimerization and higher aggregate formation (KD/K5 = 6500) a concentration range exists in MCH where almost exclusively the dimer species of 2 is present, before further self-assembly by dispersion interactions occurs. Therefore, the data could be evaluated independently for the two self-assembly steps. The self-assembly of dimers into the higher aggregate could not be described by the isodesmic model but was fitted satisfactorily to a pentamer model. This rather small size of about ten -stacked PYOP chromophores was, furthermore, consistently indicated by AFM, VPO and DOSY NMR measurements. Based on 1D and 2D NMR data as well as the strong bisignate CD signal of the higher aggregate in combination with TD-DFT calculations, a P-helical stack is proposed as its structure. The small size can be rationalized by the anti-cooperative self-assembly mechanism and the sterical demand of the solubilizing trialkoxyphenyl and the chiral tetralin substituents. Additionally, the aliphatic shell formed by the solubilizing chains around the polar chromophore stack, can account for the exceptionally high solubility of 2 in MCH (> 15 mg mL1). These combined studies of the self-assembly process enabled the identification of suitable conditions for the investigation of fluorescence properties of the individual aggregate species. Aggregation-induced emission enhancement was observed for the almost non-emissive monomer (Fl(M) = 0.23%), which can be rationalized by the increasing rigidification within the dimer (Fl(D) = 2.3%) and the higher aggregate (Fl(H) = 4.5%). The helical chirality of the PYOP decamer stack, furthermore, gave rise to a strong CPL signal with a large glum value of 0.011.
The important conclusion of this thesis is that the temperature- and solvent-dependent analyses are valid alternatives to the classical concentration-dependent analysis to determine thermodynamic parameters of self-assembly equilibria. Although, for a specific supramolecular system, one approach might be favourable over the others for a variety of reasons. The experimental limitations often demand a combination of techniques to fully elucidate a self-assembly process and to gain insights in the aggregate structure. The anti-cooperative merocyanine self-assembly, which was described here for the first time for the PYOP merocyanine 2, is no exception. Besides the interest in the merocyanine assemblies from a structural and functional point of view, the insights gained from the presented studies can also be transferred to other self-assembly systems and be a guide to find the most appropriate analysis technique.
Theory and simulation of ultrafast autodetachment dynamics and nonradiative relaxation in molecules
(2024)
In this thesis, theoretical approaches for the simulation of electron detachment processes in molecules following vibrational or electronic excitation are developed and applied. These approaches are based on the quantum-classical surface-hopping methodology, in which nuclear motion is treated classically as an ensemble of trajectories in the potential of quantum-mechanically described electronic degrees of freedom.
The slowly activating vacuolar SV/TPC1 channel is ubiquitously expressed in plants and provides a large cation conductance in the vacuolar membrane. Thereby, monovalent (K+, Na+) and in principle also divalent cations, such as Ca2+, can pass through the channel. The SV/TPC1 channel is activated upon membrane depolarization and cytosolic Ca2+ but inhibited by luminal calcium. With respect to the latter, two luminal Ca2+ binding sites (site 1 Asp240/Asp454/Glu528, site 2 Glu239/Asp240/Glu457) were identified to coordinate luminal Ca2+. In this work, the characteristics of the SV/TPC1 channels in terms of regulation and function were further elucidated, focusing on the TPC1s of Arabidopsis thaliana and Vicia faba. For electrophysiological analysis of the role of distinct pore residues for channel gating and luminal Ca2+ sensing, TPC1 channel variants were generated by site-directed mutagenesis and transiently expressed as eGFP/eYFP-fusion constructs in Arabidopsis thaliana mesophyll protoplasts of the TPC1 loss-of-function mutant attpc1-2.
1. As visualized by confocal fluorescence laser-scanning microscopy, all AtTPC1 (WT, E605A/Q, D606N, D607N, E605A/D606N, E605Q/D606N/D607N, E457N/E605A/D606N) and VfTPC1 channel variants (WT, N458E/A607E/ N608D) were correctly targeted to the vacuole membrane.
2. Patch-clamp studies revealed that removal of one of the negative charges at position Glu605 or Asp606 was already sufficient to promote voltage-dependent channel activation with higher voltage sensitivity. The combined neutralization of these residues (E605A/D606N), however, was required to additionally reduce the luminal Ca2+ sensitivity of the AtTPC1 channel, leading to hyperactive AtTPC1 channels. Thus, the residues Glu605/Asp606 are functionally coupled with the voltage sensor of AtTPC1 channel, thereby modulating channel gating, and form a novel luminal Ca2+ sensing site 3 in AtTPC1 at the luminal entrance of the ion transport pathway.
3. Interestingly, this novel luminal Ca2+ sensing site 3 (Glu605/Asp606) and Glu457 from the luminal Ca2+ sensing site 2 of the luminal Ca2+-sensitive AtTPC1 channel were neutralized by either asparagine or alanine in the TPC1 channel from Vicia faba and many other Fabaceae. Moreover, the VfTPC1 was validated to be a hyperactive TPC1 channel with higher tolerance to luminal Ca2+ loads which was in contrast to the AtTPC1 channel features. As a result, VfTPC1 but not AtTPC1 conferred the hyperexcitability of vacuoles. When AtTPC1 was mutated for the three VfTPC1-homologous polymorphic site residues, the AtTPC1 triple mutant (E457N/E605A/D606N) gained VfTPC1-like characteristics. However, when VfTPC1 was mutated for the three AtTPC1-homologous polymorphic site residues, the VfTPC1 triple mutant (N458E/A607E/N608D) still sustained VfTPC1-WT-like features. These findings indicate that the hyperactivity of VfTPC1 is achieved in part by the loss of negatively charged amino acids at positions that - as part of the luminal Ca2+ sensing sites 2 and 3 – are homologous to AtTPC1-Glu457/Glu605/Asp606 and are likely stabilized by other unknown residues or domains.
4.The luminal polymorphic pore residues (Glu605/Asp606 in AtTPC1) apparently do not contribute to the unitary conductance of TPC1. Under symmetrical K+ conditions, a single channel conductance of about 80 pS was determined for AtTPC1 wild type and the AtTPC1 double mutant E605A/D606A. This is in line with the three-fold higher unitary conductance of VfTPC1 (232 pS), which harbors neutral luminal pore residues at the homologous sites to AtTPC1.
In conclusion, by studying TPC1 channel from Arabidopsis thaliana and Vicia faba, the present thesis provides evidence that the natural TPC1 channel variants exhibit differences in voltage gating, luminal Ca2+ sensitivity and luminal Ca2+ binding sites.
Learning accompanies us throughout our lives, from early childhood education through
school, training and university to learning at work. However, much of what we learn is quickly
forgotten. The use of practice tests is a learning strategy that contributes to the acquisition of
sustainable knowledge, i.e. knowledge that is permanently available and can be retrieved when
it is needed. This dissertation first presents findings from previous research on testing in real
educational contexts and discusses theoretically why certain learner or situational
characteristics might influence the effectiveness of the testing effect. Furthermore, a cycle of
three experiments is presented, which were used to investigate whether the positive effect of
practice tests on retention (testing effect) depends on personal or situational characteristics and
also promotes the retention of lecture content that was not directly tested (transfer) in the context
of regular psychology lectures in teacher training courses. In an additional chapter, feedback
from students on the implementation of the study in the classroom context is examined in more
detail. Finally, the results of the three studies are discussed and placed in relation to the theories
presented. The central conclusion from the studies presented is that the testing effect appears to
be a very effective learning strategy that can be used effectively in university teaching and leads
to better learning outcomes regardless of learner characteristics. However, the practice tests
should cover the entire range of relevant content, as transfer effects to non-tested content are
not to be expected.
Hereditary spastic paraplegias (HSPs) are genetically-determined, neurodegenerative disorders characterized by progressive weakness and spasticity of the lower limbs. Spastic paraplegia type 11 (SPG11) is a complicated form of HSP, which is caused by mutations in the SPG11 gene encoding spatacsin, a protein possibly involved in lysosomal reformation. Based on our previous studies demonstrating that secondary neuroinflammation can be a robust amplifier of various genetically-mediated diseases of both the central and peripheral nervous system, we here test the possibility that neuroinflammation may modify the disease outcome also in a mouse model for SPG11. Spg11-knockout (Spg11-/-) mice develop early walking pattern and behavioral abnormalities, at least partially reflecting motor, and behavioral changes typical for patients. Furthermore, we detected a progressive increase in axonal damage and axonal spheroid formation in the white and grey matter compartments of the central nervous system of Spg11-/- mice. This was accompanied by a concomitant substantial increase of secondary inflammation by cytotoxic CD8+ and CD4+ T-lymphocytes. We here provide evidence that disease-related changes can be ameliorated/delayed by the genetic deletion of the adaptive immune system. Accordingly, we provide evidence that repurposing clinically approved immunomodulators (fingolimod/FTY720 or teriflunomide), that are in use for treatment of multiple sclerosis (MS), also improve disease symptoms in mice, when administered in an early (before neural damage) or late (after/during neural damage) treatment regime.
This work provides strong evidence that immunomodulation can be a therapeutic option for the still untreatable SPG11, including its typical neuropsychological features. This poses the question if inflammation is not only a disease amplifier in SPG11 but can act as a unifying factor also for other genetically mediated disorders of the CNS. If true, this may pave the way to therapeutic options in a wide range of still untreatable, primarily genetic, neurological disorders by repurposing approved immunomodulators.
mRNA is co- or post-transcriptionally processed from a precursor mRNA to a mature mRNA. In addition to 5'capping and splicing, these modifications also include polyadenylation, the addition of a polyA tail to the 3'end of the mRNA. In recent years, alternative polyadenylation in particular has increasingly been taken into account as a mechanism for regulating gene expression. It is assumed that approximately 70-75 % of human protein coding genes contain alternative polyadenylation signals, which are often located within intronic sequences of protein-coding genes. The use of such polyadenylation signals leads to shortened mRNA transcripts and thus to the generation of C-terminal shortened protein isoforms.
Interestingly, the majority of microRNAs, small non-coding RNAs that play an essential role in post-transcriptional gene regulation, are also encoded in intronic sequences of protein-coding genes and are co-transcriptionally expressed with their host genes. The biogenesis of microRNA has been well studied and is well known, but mechanisms that may influence the expression regulation of mature microRNAs are just poorly understood.
In the presented work, I aimed to investigate the influence of alternative intronic polyadenylation on the biogenesis of microRNAs. The human ion channel TRPM1 could already be associated with melanoma pathogenesis and truncated isoforms of this protein have already been described in literature. In addition, TRPM1 harbors a microRNA, miR211, in its sixth intron, which is assumed to act as a tumor suppressor. Since both, TRPM1 and miR211 have already been associated with melanoma pathogenesis, the shift towards truncated transcripts during the development of various cancers is already known and it has been shown that certain microRNAs play a crucial role in the development and progression of melanoma, melanoma cell lines were used as an in vitro model for these investigations.
Trust carries the capacity to shift the focus from risks to opportunities of a situation. Scientific studies from the field of trust research point out that besides situation-specific factors (i.e., stimuli of the environment), cross-situationally stable interindividual differences (i.e., personality) are involved in the emergence of trust. Stable interindividual differences are particularly influential to the subjective experience of situational conditions when crucial information is incomplete. The online shopping environment classifies as a prime example of markets with asymmetric information. Research has examined online consumer trust in the light of signaling theory to understand the effects of trust-enhancing signals. Previous research largely neglects interindividual differences in the perception, processing and reaction to these signals. Against this background, this scientific work has two primary objectives: the investigation of (1) interindividual differences in the evaluation of trust-enhancing signals and (2) a personality-based personalization of trust-enhancing signals in its effect on cognition and behavior. For this purpose, an interactive online shop setup was created, which served as realistic environmental framework. First, the results show a trust-enhancing effect of both objective and subjective personalization, with a superiority of subjective over objective personalization. Second, results suggest a particular susceptibility of the beliefs component of trust. Third, the results suggest that personalization exerts a specifically strong effect in what is, by definition, the particularly uncertain environment of credence goods. Fourth, results indicate that while the trust-enhancing effects of personalization operate (largely) independently of personality, the effect of personality on trust seems to depend on the condition of signal presentation. Taken together, the present work makes a contribution to understanding the effect of personality-adapted signaling environments on the emergence of trust and decision making in the specific context of B2C e-commerce.
Most medicines are taken orally. To enter the systemic circulation, they dissolve in the intestinal fluid, cross the epithelial barrier, and pass through the liver. Intestinal absorption is driven by the unique features of the gastrointestinal tract, including the bile colloids formed in the lumen and the mucus layer covering the intestinal epithelium. Neglecting this multifaceted environment can lead to poor drug development decisions, especially for poorly water-soluble drugs that interact with bile and mucus. However, there is a lack of a rationale nexus of molecular interactions between oral medicines and gastrointestinal components with drug bioavailability. Against this background, this thesis aims to develop biopharmaceutical strategies to optimize the presentation of oral therapeutics to the intestinal epithelial barrier.
In Chapter 1, the dynamics of bile colloids upon solubilization of the poorly-water soluble drug Perphenazine was studied. Perphenazine impacted molecular arrangement, structure, binding thermodynamics, and induced a morphological transition from vesicles to worm-like micelles. Despite these dynamics, the bile colloids ensured stable relative amounts of free drug substance. The chapter was published in Langmuir.
Chapter 2 examined the impact of pharmaceutical polymeric excipients on bile-mediated drug solubilization. Perphenazine and Imatinib were introduced as model compounds interacting with bile, whereas Metoprolol did not. Some polymers altered the arrangement and geometry of bile colloids, thereby affecting the molecularly soluble amount of those drugs interacting with bile. These insights into the bile-drug-excipient interplay provide a blueprint to optimizing formulations leveraging bile solubilization. The chapter was published in Journal of Controlled Release.
Chapter 3 deals with the impact of bile on porcine intestinal mucus. Mucus exposed to bile solution changed transiently, it stiffened, and the overall diffusion rate increased. The bile-induced changes eased the transport of the bile-interacting drug substance Fluphenazine, whereas Metoprolol was unaffected. This dichotomous pattern was linked to bioavailability in rats and generalized based on two previously published data sets. The outcomes point to a bile-mucus interaction relevant to drug delivery. The chapter is submitted.
The Appendix provides a guide for biopharmaceutical characterization of drug substances by nuclear magnetic resonance spectroscopy aiming at establishing a predictive algorithm.
In summary, this thesis deciphers bile-driven mechanisms shaping intestinal drug absorption. Based on these molecular insights, pharmaceuticals can be developed along a biopharmaceutical optimization, ultimately leading to better oral drugs of tomorrow.
Western societies are steadily becoming older undergoing a clear trend of delayed parenthood. Children of older fathers have an undeniably higher risk for certain neurodevelopmental disorders and other medical conditions. Changes in the epigenetic landscape and especially in DNA methylation patterns are likely to account for a portion of this inherited disease susceptibility. DNA methylation changes during the ageing process are a well-known epigenetic feature. These so-called age-DMRs exist in developmentally important genes in the methylome of several mammalian species. However, there is only a minor overlap between the age-DMR datasets of different studies. We therefore replicated age-DMRs (which were obtained from a genome wide technique) by applying a different technical approach in a larger sample number. Here, this study confirmed 10 age-DMRs in the human and 4 in the bovine sperm epigenome from a preliminary candidate list based on RRBS. For this purpose, we used bisulphite Pyrosequencing in 94 human and 36 bovine sperm samples. These Pyrosequencing results confirm RRBS as an effective and reliable method to screen for age-DMRs in the vertebrate genome. To decipher whether paternal age effects are an evolutionary conserved feature of mammalian development, we compared methylation patterns between human and bovine sperm in orthologous regulatory regions. We discovered that the level of methylation and the age effect are both species-specific and speculate that these methylation marks reflect the lineage-specific development of each species to hit evolutionary requirements and adaptation processes. Different methylation levels between species in developmentally important genes also imply a differing mutational burden, representing a potential driver for point mutations and consequently deviations in the underlying DNA sequence of different species. Using the example of different haplotypes, this study showed the great effect of single base variations on the methylation of adjacent CpGs. Nonetheless, this study could not provide further evidence or a mechanism for the transfer of epigenetic marks to future generations. Therefore, further research in tissues from the progeny of old and young fathers is required to determine if the observed methylation changes are transmitted to the next generation and if they are associated with altered transcriptional activity of the respective genes. This could provide a direct link between the methylome of sperm from elderly fathers and the development potential of the next generation.