Technische Universität Dresden: Qucosa
Not a member yet
16344 research outputs found
Sort by
Resonance states of the three-disk scattering system
No full textResonances of scattering problems appear in many branches of physics and mathematics. The structure of resonance states is central for the understanding of these systems. For a paradigmatic example of chaotic scattering, the three-disk scattering system, we investigate the structure of resonance states in the semiclassical limit. We confirm a recent conjecture on the factorization of resonance states into a product of a multifractal density of classical origin and universal exponentially distributed fluctuations for this system. For the first factor, we show explicitly how it is described by classical dynamics in the semiclassical limit. This is also applied to the left-right Husimi representation of resonance states, which combines left and right resonance states. Furthermore, we study higher-order corrections to the distribution of the universal fluctuations and observe the recently described ray-segment scarring also in the three-disk scattering system. This is enabled by a new numerical method for the computation of resonances, which allows for going much further into the semiclassical limit than previously possible. As a consequence, we are also able to study the distribution of resonance poles and confirm the fractal Weyl law over a correspondingly large range.:1. Introduction
2. Three-disk scattering system
2.1. Classical and quantum chaotic scattering
2.2. Classical dynamics
2.2.1. Time-continuous dynamics and boundary map
2.2.2. Invariant sets
2.2.3. Fractal dimension of the chaotic saddle
2.2.4. Natural measure
2.2.5. Perron-Frobenius operator and Ulam’s method
2.3. Quantization and resonance states
3. Distribution of resonance poles
3.1. Spectrum
3.2. Fractal Weyl law
3.3. Distribution of decay rates
3.4. Semiclassical computation of resonance poles
4. Factorization of resonance states
4.1. Eigenstates of closed chaotic systems
4.1.1. Quantum ergodicity theorem
4.1.2. Random wave model
4.1.3. Factorization of eigenstates
4.2. Factorization of resonance states
4.2.1. Resonance states of the three-disk scattering system
4.2.2. Factorization
4.2.3. Average resonance states
4.2.4. Universal fluctuations
4.3. Deviations due to finite size effects
4.3.1. Random wave model on finite domain
4.3.2. Influence on the fluctuations of resonance states
4.3.3. Phase rigidity from boundary integral
4.4. Ray-segment scars
4.5. Application to left-right Husimi representation
4.5.1. Left-right Husimi representation
4.5.2. Factorization
4.5.3. Average left-right Husimi representation
4.5.4. Universal left-right fluctuations
5. Semiclassical limit of resonance states
5.1. Previous work
5.2. Classical measures
5.2.1. Generalized Ulam’s method
5.2.2. Selection criterion
5.2.3. Number of Ulam cells
5.3. Quantum-classical comparison
5.3.1. Average resonance states
5.3.2. Individual resonance states
5.3.3. Semiclassical limit
5.4. Application to left-right Husimi representation
5.4.1. Construction of left-right measure
5.4.2. Quantum-classical comparison
6. Summary and outlook
Appendix
A. Classical measures in position space
B. Parameters for computation
C. Jensen-Shannon divergence
List of figuresResonanzen von Streuproblemen treten in vielen Bereichen der Physik und Mathematik auf. Dabei ist die Struktur von Resonanzzuständen zentral für das Verständnis dieser Systeme. Für ein paradigmatisches Beispielsystem chaotischer Streuung, das Drei-Scheiben-Billard, untersuchen wir die Struktur von Resonanzzuständen im semiklassischen Grenzfall. Wir bestätigen eine kürzlich aufgestellte Vermutung zur Faktorisierung von Resonanzzuständen in ein Produkt aus einer multifraktalen Dichte klassischen Ursprungs und universell exponentiell verteilten Fluktuationen für dieses System. Für den ersten Faktor zeigen wir explizit, wie er im semiklassischen Grenzfall durch die klassische Dynamik beschrieben wird. Dies wird auch auf die Links-Rechts-Husimi-Darstellung von Resonanzzuständen angewendet, die linke und rechte Resonanzzustände kombiniert. Des Weiteren untersuchen wir höhere Korrekturen zur Verteilung der universellen Fluktuationen und beobachten kürzlich erstmals beschriebene Narben entlang von Abschnitten klassischer Trajektorien im Drei-Scheiben-Billard. Dies wird durch eine neue numerische Methode zur Berechnung von Resonanzen ermöglicht, die es erlaubt, viel weiter in den semiklassischen Grenzfall zu gehen als bisher möglich. Als Folge sind wir auch in der Lage, über einen entsprechend großen Bereich die Verteilung von Resonanzpolen zu untersuchen und das fraktale Weyl-Gesetz zu bestätigen.:1. Introduction
2. Three-disk scattering system
2.1. Classical and quantum chaotic scattering
2.2. Classical dynamics
2.2.1. Time-continuous dynamics and boundary map
2.2.2. Invariant sets
2.2.3. Fractal dimension of the chaotic saddle
2.2.4. Natural measure
2.2.5. Perron-Frobenius operator and Ulam’s method
2.3. Quantization and resonance states
3. Distribution of resonance poles
3.1. Spectrum
3.2. Fractal Weyl law
3.3. Distribution of decay rates
3.4. Semiclassical computation of resonance poles
4. Factorization of resonance states
4.1. Eigenstates of closed chaotic systems
4.1.1. Quantum ergodicity theorem
4.1.2. Random wave model
4.1.3. Factorization of eigenstates
4.2. Factorization of resonance states
4.2.1. Resonance states of the three-disk scattering system
4.2.2. Factorization
4.2.3. Average resonance states
4.2.4. Universal fluctuations
4.3. Deviations due to finite size effects
4.3.1. Random wave model on finite domain
4.3.2. Influence on the fluctuations of resonance states
4.3.3. Phase rigidity from boundary integral
4.4. Ray-segment scars
4.5. Application to left-right Husimi representation
4.5.1. Left-right Husimi representation
4.5.2. Factorization
4.5.3. Average left-right Husimi representation
4.5.4. Universal left-right fluctuations
5. Semiclassical limit of resonance states
5.1. Previous work
5.2. Classical measures
5.2.1. Generalized Ulam’s method
5.2.2. Selection criterion
5.2.3. Number of Ulam cells
5.3. Quantum-classical comparison
5.3.1. Average resonance states
5.3.2. Individual resonance states
5.3.3. Semiclassical limit
5.4. Application to left-right Husimi representation
5.4.1. Construction of left-right measure
5.4.2. Quantum-classical comparison
6. Summary and outlook
Appendix
A. Classical measures in position space
B. Parameters for computation
C. Jensen-Shannon divergence
List of figure
Adherence of the rotating vortex lattice in the noncentrosymmetric superconductor Ru7B3 to the London model
No full textThe noncentrosymmetric superconductor Ru7B3 has in previous studies demonstrated remarkably unusual behaviour in its vortex lattice (VL), where the nearest neighbour directions of the vortices dissociate from the crystal lattice and instead show a complex field-history dependence, and the VL rotates as the field is changed. In this study, we look at the VL form factor of Ru7B3 during this field-history dependence, to check for deviations from established models, such as the London model. We find that the data is well described by the anisotropic London model, which is in accordance with theoretical predictions that the alterations to the structure of the vortices due to broken inversion symmetry should be small. From this, we also extract values for the penetration depth and coherence length
Evaluation of Small Molecule Neurotrophin Mimetics in Models of Neurodegeneration and Neuroinflammation and Novel Insights into the Regulation of Microglial Function
No full textNeurodegenerative and neuroinflammatory diseases are conditions of increasing prevalence and pose a global burden on the healthcare system and society. However, treatment of neurodegenerative diseases presents a difficult task owing to the complex pathophysiology of central nervous system (CNS) disorders, and the limited accessibility of the CNS for pharmacological interventions. Microglia, the resident immune cells of the CNS, exert important functions during development in neuronal and synaptic maturation, and are major contributors to the maintenance of CNS homeostasis. In disease conditions, microglia drive pro-inflammatory responses by secretion of cytokines, pro-inflammatory metabolites, reactive oxygen species (ROS) and nitric oxide (NO), and phagocytose pathogens, dead cells and debris. Recent research has shown that microglia activation does not occur simply as an on/off system, but rather by acquisition of numerous activation states dependent on spatiotemporal cues. Chronic activation of microglia can be detrimental for neuronal health and may impede recovery from neurodegenerative diseases. Therefore, gaining deeper insight into the mechanisms regulating microglial activation is of great importance for development of better therapeutics against neurodegenerative and neuroinflammatory diseases.
Previous research has shown that the nerve growth factor (NGF), as well as the neurosteroid dehydroepiandrosterone (DHEA) can modulate neuroinflammation by activation of the tropomyosin receptor kinase A (TRKA) receptor on microglia. In this work, we synthesized a novel C-17-spiro-cyclopropyl DHEA derivative, ENT-A010, and showed that it activates the TRKA receptor and its downstream target AKT. Pre-treatment with ENT-A010 enhanced amyloid β phagocytosis and promoted a protective phenotype in lipopolysaccharide (LPS)-treated microglia. Additionally, pre-treatment with ENT-A010 partially preserved the homeostatic gene signature and ramified morphology of microglia in the hippocampi of LPS treated mice. Thus, we have identified ENT-A010, a novel TRKA ligand with the ability to modulate microglial function, as a molecule of interest for further research advancing therapeutic strategies against neurodegenerative disorders.
Innate immune responses require increased energy expenditure coupled with cellular metabolic rewiring. However, the cellular metabolic changes in activated microglia remain relatively poorly understood. Here we studied the cellular metabolic changes in brain microglia in mice under LPS treatment. We observed rewiring of microglia metabolism, featured by an increase in OxPhos and induction of aconitate decarboxylase (ACOD1) and itaconate production. Similarly to macrophages, ACOD1 deficiency increased Interleukin 1 beta (Il-1b) production, while treatment with itaconate reduced Il-1b expression in mouse primary microglia. Moreover, ACOD1 deficiency also enhanced production of argininosuccinate, an intermediate metabolite of arginine biosynthesis, while argininosuccinate pre-treatment in mice exacerbated the LPS-induced pro-inflammatory activation of microglia, suggesting that argininosuccinate functions as a pro-inflammatory metabolite.
Concluding, our findings revealed that argininosuccinate accumulation enhanced the pro-inflammatory activation of microglia and thus identified suppression of argininosuccinate accumulation as a novel mechanism by which ACOD1-mediated itaconate production can exert anti-inflammatory effects in microglia
Ab initio study of time-reversal invariant superconducting Weyl semimetals
No full textWeyl semimetals are a topological phase of matter characterized by robust electronic features in both the bulk and at the surface. In this work, we investigate two Weyl semimetals—PtBi2 and BeAu—that exhibit exotic superconducting behavior, such as surface-only superconductivity (in PtBi2) and multi-gap superconductivity (in BeAu). Through detailed theoretical characterization of their electronic structures, we aim to understand the relationship between their topological properties and superconductivity.
The first part of this work focuses on PtBi2. We analyze its metallic structure with an emphasis on the presence of topologically protected crossings in the bulk (Weyl points) and the associated Fermi arcs. Additionally, we examine the system’s evolu- tion under external perturbations such as pressure and doping. Our findings indicate that a structural phase transition to a topologically trivial system is energetically close, consistent with recent experimental reports. Based on our results, we also develop an effective model that reproduces the number and distribution of Weyl points in the real compound while preserving the correct symmetries. By comparing our density functional theory calculations with new experimental data from our collaborators, we confirm the crucial role of topological surface states in shaping the surface electronic properties of PtBi2. Furthermore, we provide key insights into the nature of its surface superconducting state.
The second part of this work focuses on BeAu. We identify several topological features characteristic of chiral systems, including multifold fermions and isolated Weyl points, and study their dependence on external hydrostatic pressure. Motivated by recent proposals on topological superconductivity in chiral superconductors, we further investigate BeAu’s superconducting properties. We establish the conventional electron-phonon origin of its superconductivity by computing the critical tempera- ture in agreement with experimental measurements. In doing so, we derive a simple expression to account for the effect of external hydrostatic pressure on phonon modes, revealing strong similarities between this system and hydrogen-based super- conductors. Additionally, we identify substantial portions of BeAu’s Fermi surface that carry nontrivial chirality, suggesting the potential emergence of a topological superconducting phase driven by the interplay between Weyl physics and multi-gap superconductivity.
Our findings reinforce the significance of Weyl semimetals as promising systems not only for realizing topological superconductivity but also for uncovering novel phenomena, such as the robust surface superconducting state observed in PtBi2
Insights into the interface during ice adhesion measurements
No full textWhen evaluating published results from ice adhesion measurements to characterise the applicability of surface coatings, there is usually a large variation of the published properties and a high scattering of these values. Moreover ice adhesion is known as a highly susceptible parameter regarding the measurement temperature which additionally leads to deviations. This paper is a contribution to the evaluation of ice adhesion results and provides a correlation of the measurements with the surface characteristics. In the paper a novel instrumentalised method of a quasistatic ice adhesion test is proposed in order to measure additional information about force and displacement behaviour of adhered ice. The evaluation of several measurements reveals insights into nanoscopic processes in the interface during the adhesion process. Different modes of the adhesion process of ice, like clean breaks, sliding with almost no force, sliding with high forces, multiple breaks or multiple sticking, not only we found to correlate to the material, but also to methodological test parameters. The overlapping influence of surface characteristics and test parameters is highlighted
EMT induces characteristic changes of Rho GTPases and downstream effectors with a mitosis-specific twist
No full textEpithelial-mesenchymal transition (EMT) is a key cellular transformation for many physiological and pathological processes ranging from cancer over wound healing to embryogenesis. Changes in cell migration, cell morphology and cellular contractility were identified as hallmarks of EMT. These cellular properties are known to be tightly regulated by the actin cytoskeleton. EMT-induced changes of actin-cytoskeletal regulation were demonstrated by previous reports of changes of actin cortex mechanics in conjunction with modifications of cortex-associated f-actin and myosin. However, at the current state, the changes of upstream actomyosin signaling that lead to corresponding mechanical and compositional changes of the cortex are not well understood. In this work, we show in breast epithelial cancer cells MCF-7 that EMT results in characteristic changes of the cortical association of Rho-GTPases Rac1, RhoA and RhoC and downstream actin regulators cofilin, mDia1 and Arp2/3. In the light of our findings, we propose that EMT-induced changes in cortical mechanics rely on two hitherto unappreciated signaling paths—i) an interaction between Rac1 and RhoC and ii) an inhibitory effect of Arp2/3 activity on cortical association of myosin II
Targeting fibroblast growth factor receptors in head and neck cancers: Identifying therapeutic vulnerabilities for radiochemosensitization and overcoming adaptive resistance
No full textBackground
Head and neck squamous cell carcinoma (HNSCC) remains a critical oncological challenge, particularly in human papillomavirus (HPV)-negative cases. These tumors are characterized by poor patient survival rates and high recurrence rates, attributed mainly to the development of therapy resistances. The mechanisms underlying these resistances are complex and multifactorial, including phenotypic plasticity driven by epithelial-to-mesenchymal transition (EMT), and cooperative pro-survival signaling via receptor tyrosine kinases (RTKs) and integrin adhesion receptors. While dual inhibition of the frequently overexpressed epidermal growth factor receptor (EGFR) and β1 integrin has demonstrated promising preclinical efficacy, non-responsiveness in certain models suggests the involvement of additional oncogenic signaling networks.
Hypothesis
This study aims to identify novel RTK/β1 integrin-based vulnerabilities in HPV-negative HNSCC. It is hypothesized that combined inhibition of a specific RTK and β1 integrin elicits superior cytotoxic and radiochemosensitizing effects in HPV-negative HNSCC models compared to monotherapy. Furthermore, it is hypothesized that targeting kinases upregulated in response to initial therapy effectively circumvents emerging resistance mechanisms.
Methods
The cytotoxic and radiochemosensitizing efficacy of targeting 10 distinct RTKs, alone or in conjunction with β1 integrin, was evaluated in up to 20 three-dimensional laminin-rich extracellular matrix (3D lrECM)-cultured HNSCC cell models. RNA interference and pharmacological-based targeting approaches were utilized in cell viability and clonogenic survival assays complemented by single doses of 6 Gy X-rays and/or cisplatin treatment. Molecular profiling was performed by RNA sequencing and protein-based biochemical assays. Cell viability-based drug screening with 36 kinase inhibitors was employed to identify resistance-deactivating targets. Clinical relevance was assessed by transferring the resistance signature to HPV-negative HNSCC cohorts using machine learning and pathway analysis. Network analyses integrating drug screening and clinical cohort data were conducted to identify key determinants of resistance.
Results
Among the examined RTKs, depletion of the fibroblast growth factor receptor (FGFR) family exhibited the most pronounced cytotoxic and radiosensitizing effects. Dual pharmacological targeting, using the pan-FGFR inhibitor erdafitinib (FGFRi) and the anti-β1 integrin antibody AIIB2, enhanced radiochemosensitization across a heterogeneous panel of HNSCC cell models. However, a subset of models exhibited a profound radioprotective resistance to FGFRi. Transcriptomic analyses revealed that FGFRi treatment elicited mesenchymal-to-epithelial transition (MET) in sensitive UT-SCC 33 cells, whereas FGFRi-induced resistance in UM-SCC 10a cells was associated with an EGFR-driven partial EMT (pEMT) phenotype. Mechanistically, a feedback loop involving ERK1/2 hyperphosphorylation upon FGFR3 deactivation was implicated in this adaptive pEMT resistance. Kinase inhibitor screening confirmed that EGFR inhibition effectively abrogated FGFRi resistance in UM-SCC 10a cells. Additionally, novel kinase targets, including Fyn, PAK1-3, and PKCα, were identified, demonstrating the capacity to either reverse FGFRi resistance or enhance FGFRi efficacy in other HNSCC models. Transferring the pEMT-related resistance signature to patient cohorts confirmed its prognostic significance and highlighted overexpressed and druggable genes, including PLAU, HBEGF, and ITGA5. Integrating drug screening and clinical signature results into network analyses underscored their close interconnection and emphasized the central role of EGFR as a key mediator of pEMT-related resistance, thus offering targetable vulnerabilities for therapeutic intervention.
Conclusions
This study highlights the therapeutic potential of targeting the FGFR family in HPV-negative HNSCC, while also revealing the risk of adaptive resistance. FGFR inhibition induced a spectrum of responses, ranging from beneficial radiochemosensitization, enhanced by concurrent β1 integrin blockade, to pronounced cyto- and radioprotective effects. Notably, the dynamics of EMT were significantly influenced by FGFRi treatment across this spectrum. The adaptive pEMT-related resistance appears to be of clinical significance and can be therapeutically exploited through rational combination strategies targeting FGFR, EGFR, and associated kinases. As FGFR inhibitors gain clinical traction in other cancer types, further investigation into predictive biomarkers and adaptive resistance mechanisms will be essential for their optimal application in HNSCC. The identified FGFRi resistance network provides a foundation for the development of personalized and adaptable therapeutic approaches aimed at achieving durable treatment responses in HNSCC.Hintergrund
Das Plattenepithelkarzinom des Kopf-Hals-Bereichs (HNSCC) stellt nach wie vor eine schwierige onkologische Herausforderung dar, insbesondere bei humanen Papillomavirus (HPV)-negativen Fällen. Diese Tumore zeichnen sich durch niedrige Überlebenschancen und hohe Rezidivraten aus, was im Wesentlichen auf die Entwicklung von Therapieresistenzen zurückzuführen ist. Die zugrundeliegenden Resistenzmechanismen sind komplex und vielschichtig und basieren unter anderem auf der durch epithelial-mesenchymale Transition (EMT) angetriebene phänotypische Plastizität, sowie einer kooperativen und überlebensfördernden Signalübertragung durch Rezeptortyrosinkinasen (RTKs) und Integrin-Adhäsionsrezeptoren. Die duale Hemmung des häufig überexprimierten epidermalen Wachstumsfaktorrezeptors (EGFR) und β1-Integrin zeigte eine vielversprechende präklinische Wirksamkeit. Jedoch deutet das Nichtansprechen in bestimmten Modellen auf die Beteiligung weiterer onkogener Signalnetzwerke hin.
Fragestellung/Hypothese
Ziel dieser Studie ist es neue RTK/β1-Integrin-basierte therapeutische Angriffspunkte bei HPV-negativem HNSCC zu identifizieren. Es wird die Hypothese aufgestellt, dass die kombinierte Hemmung eines spezifischen RTKs und β1-Integrin in HPV-negativen HNSCC-Modellen eine stärkere zytotoxische und radiochemosensibilisierende Wirkung zeigt als die jeweilige Einzeltherapie. Darüber hinaus wird angenommen, dass die Inaktivierung von Kinasen, welche als Reaktion auf die primäre Therapie hochreguliert werden, aufkommende Resistenzmechanismen wirkungsvoll unterbindet.
Material und Methode
Die zytotoxische und radiochemosensibilisierende Wirksamkeit der Inhibition von 10 verschiedenen RTKs, allein oder in Kombination mit β1-Integrin, wurde in bis zu 20 HNSCC Zellmodellen, kultiviert in dreidimensionaler lamininreicher extrazellulärer Matrix (3D lrECM), untersucht. RNA-Interferenz- und pharmakologische Targeting-Ansätze wurden in Zellviabilitäts- und klonogenen Überlebenstests eingesetzt, ergänzt durch Einzeldosen von 6 Gy Röntgenstrahlung und/oder Cisplatin-Behandlung. Molekulare Charakterisierungen wurden mittels RNA-Sequenzierung und proteinbasierten biochemischen Assays durchgeführt. Ein auf Zellviabilität basierendes Medikamentenscreening mit 36 Kinaseinhibitoren wurde verwendet, um resistenzdeaktivierende Ziele zu identifizieren. Die klinische Relevanz wurde durch Übertragung der Resistenzsignatur auf HPV-negative HNSCC-Kohorten mittels maschinellen Lernens und Signalwegsanalyse bewertet. Die Daten aus dem Medikamentenscreening und klinischen Kohorten wurden in Netzwerkanalysen, zur Bestimmung der wichtigsten Resistenzfaktoren, zusammengeführt.
Ergebnisse
Unter den untersuchten RTKs zeigte die Depletion der Fibroblasten-Wachstumsfaktor-Rezeptor (FGFR)-Familie die ausgeprägtesten zytotoxischen und radiosensibilisierenden Effekte. Die duale pharmakologische Hemmung mit dem pan-FGFR-Inhibitor Erdafitinib (FGFRi) und dem anti-β1-Integrin-Antikörper AIIB2 verstärkte die Radiochemosensibilisierung über viele heterogene HNSCC-Zellmodelle hinweg. Eine Untergruppe von Modellen wies jedoch eine starke radioprotektive Resistenz gegenüber FGFRi auf. Transkriptomanalysen ergaben, dass die FGFRi-Behandlung in sensitiven UT-SCC 33-Zellen eine mesenchymale-zu-epitheliale Transition (MET) auslöste, während die FGFRi-induzierte Resistenz in UM-SCC 10a-Zellen mit einem EGFR-gesteuerten partiellen EMT-Phänotyp (pEMT) assoziiert war. Mechanistisch kann für die adaptive pEMT-Resistenz eine Feedback-Schleife, ausgelöst durch FGFR3-Deaktivierung und anschließender ERK1/2 Hyperphosphorylierung, angenommen werden. Das Medikamentenscreening bestätigte, dass die EGFR-Hemmung die FGFRi-Resistenz in UM-SCC 10a-Zellen effektiv aufhob. Darüber hinaus wurden neue Kinase-Zielstrukturen wie Fyn, PAK1-3 und PKCα identifiziert, deren Inhibition ebenfalls die FGFRi-Resistenz neutralisierte oder die FGFRi-Wirksamkeit in anderen HNSCC-Modellen verbesserte. Die Übertragung der pEMT-bezogenen Resistenzsignatur auf Patientenkohorten bestätigte ihre prognostische Bedeutung und hob überexprimierte und medikamentös angreifbare Gene wie PLAU, HBEGF und ITGA5 hervor. Die Integration der Ergebnisse aus dem Medikamentenscreening und den klinischen Signaturen in Netzwerkanalysen unterstrich deren enge Verknüpfung und betonte die zentrale Rolle von EGFR als Hauptvermittler der pEMT-bezogenen Resistenz, wodurch sich gezielte Schwachstellen für therapeutische Interventionen ergeben.
Schlussfolgerungen
Diese Studie verdeutlicht das therapeutische Potenzial einer gezielten Inhibition der FGFR-Familie bei HPV-negativem HNSCC, zeigt jedoch auch das Risiko adaptiver Resistenzen auf. Die FGFR-Hemmung induzierte ein Spektrum von Reaktionen, das von vorteilhafter Radiochemosensibilisierung, verstärkt durch gleichzeitige β1-Integrin-Blockade, bis hin zu ausgeprägten zyto- und radioprotektiven Effekten reichte. Bemerkenswerterweise wurde die EMT-Dynamik durch die FGFRi-Behandlung über dieses Spektrum hinweg wesentlich beeinflusst. Die adaptive pEMT-assoziierte Resistenz scheint von klinischer Bedeutung zu sein und kann durch rationale Kombinationsstrategien, die auf FGFR, EGFR und damit verbundene Kinasen abzielen, therapeutisch genutzt werden. Da FGFR-Inhibitoren in anderen Krebsarten klinisch an Bedeutung gewinnen, wird eine weitere Untersuchung prädiktiver Biomarker und adaptiver Resistenzmechanismen entscheidend für ihre optimale Anwendung in HNSCC sein. Das identifizierte FGFRi-Resistenznetzwerk bietet eine Grundlage für die Entwicklung personalisierter und anpassungsfähiger Therapieansätze, mit welchen dauerhafte Behandlungserfolge in HNSCC erzielt werden könnten
Umkleiden für Alle: Gestaltungsempfehlungen für die inklusive und gemeinsame Nutzung von Umkleidebereiche
No full textSportanlagen, die von inklusiven Sportgruppen genutzt werden benötigen einerseits ungehinderten Zugang zu den Sportflächen und andererseits Umkleidebereiche, welche die Bedürfnisse der Sportler:innen erfüllen. Zur Planung solcher Räume stehen Richtlinien zur Verfügung, welche sich auf räumliche Abmaße und körperliche Anforderungen beschränken. Um die Gestaltung nach den Bedürfnissen der Nutzenden auszurichten, sind Vorgaben über technische Abmaße hinaus erforderlich. [... aus dem Text
Gezielte Immuntherapie des metastasierten Nierenzellkarzinoms mit Hilfe Antigen-spezifischer Nanopartikel
No full textie Entwicklung neuartiger Immuntherapien hat in den letzten Jahren enorme Fortschritte bei der Behandlung maligner Erkrankungen erzielt. Viele Behandlungen zielen darauf ab, die meist ineffektiven und erschöpften bzw. supprimierten Immun-Effektor-Funktionem zu reaktivieren und so eine anti-tumorale Immunantwort zu initiieren. Neben den Erfolgen treten jedoch auch Rückschläge auf, wie beispielsweise immunvermittelte Nebenwirkungen oder die Entstehung von Resistenzen, sodass weiterhin ein großer Bedarf an innovativen Therapiekonzepten besteht. Der Einsatz von Bio-Immunkonjugaten (BICs) im Nanopartikelformat (NANO:BICs) eröffnet eine vielversprechende Strategie für die zielgerichtete Immuntherapie maligner Erkrankungen. Aufbauend auf den in der Arbeitsgruppe von Prof. Temme etablierten RICIA zur gezielten Stimulation von EGFRvIII-positiven Tumorzellen wurden in dieser Arbeit PSCA-spezifische NANO:BICs entwickelt. Diese sollten über eine CpG-Oligodesoxynukleotid (ODN)-vermittelte TLR9-Stimulation tumorinfiltrierenden Immunzellen in PSCA-positiven Nierenzellkarzinom-Metastasen aktivieren. Die Partikel bestanden aus monobiotinylierten Einzelkettenantikörpern gegen PSCA bzw. PSMA als Negativ-Kontrolle, dem Biotin-bindenden Linkerprotein NeutrAvidin sowie monobiotinylierten CpG-ODNs. Untersucht wurden CpG-ODNs, die den humanen oder den murinen TLR9 in unterschiedlichen Assays aktivieren sollten. Darüber hinaus erfolgte die Analyse unterschiedlicher Zucker-Phosphatrückgrate. Zum einen wurden CpG-ODNs mit dem natürlich vorkommenden Phosphodiester (PD)-Rückgrat untersucht, welches jedoch anfällig für Nukleasen und damit potenziell instabil ist. Zum anderen wurden CpG-ODNs mit dem synthetisch modifizierten Phosphothioat (PTO)-Rückgrat verwendet, welches stabiler ist, aber auch unspezifische Interaktionen auslösen kann. Außerdem erfolgte die Analyse von CpG-ODNs mit einem Phosphodiester CpG-ODN und einer Phosphothioat-Modifikation (teil-PTO, tPTO), um die vorteilhaften Eigenschaften beider Rückgrate zu vereinen. Die Komponenten wurden, basierend auf den Ergebnissen unterschiedlicher Titrationsstudien, in definierter molekularer Stöchiometrie assembliert. Ferner wurde das Vorhandensein von ungebundenen CpG-ODNs in der Immunkonjugat-Lösung durch eine Aufreinigung ausgeschlossen. Im ersten Teil der Arbeit erfolgte die Assemblierung der NANO:BICs. Dafür wurden zunächst die Einzelkettenantikörper mithilfe von Produktionszelllinien hergestellt und aufgereinigt. Anschließend wurde ein geeignetes stöchiometrisches Verhältnis für die Konjugation an NeutrAvidin im NANO:BIC über Titrationsstudien ermittelt. Die unterschiedlichen CpG-ODNs wurden ebenfalls titriert und eine geeignete Behandlungskonzentration evaluiert. Zuletzt erfolgte die Aufreinigung der Komplexe, um ungebundene, unbiotinylierte CpG-ODNs zu entfernen.
Im zweiten Teil erfolgte die Funktions-Evaluierung der NANO:BICs in vitro. Ergebnisse des HEK-BlueTM-Reportergen-Assays zeigten, dass PTO-CpG-ODN-beladene NANO:BICs die stärkste und PD-CpG-ODNs die schwächste TLR9-spezifische Stimulation hervorriefen. Ferner führte die Behandlung von TLR9-positiven RPMI-8226-PSCA-Zellen mit NANO:BICs zur Freisetzung verschiedener Immunmodulatoren, wobei NANO:BICs mit PD-CpG-ODNs die ausgeprägtesten Rezeptor-spezifischen Effekte verursachten. Mikroskopische Analysen
verwiesen auf eine Internalisierung und unspezifische Aufnahme von PTO-CpG-ODN-NANO:BICs und eine Adhäsion von PD- und tPTO-CpG-ODN-NANO:BICs an PSCA-positiven Renca-Luc-PSCA-Tumorzellen. Murine Immunzellen nahmen die Partikel unabhängig der angesteuerten Zielstruktur auf. Darüber hinaus wurde festgestellt, dass diese nach NANO:BICs-Stimulation eine erhöhte Sekretion an immunstimulierenden Molekülen aufwiesen und eine verstärkte Expression von Aktivierungsmarkern zeigten. Basierend auf den Ergebnissen vorheriger Studien wurde deutlich, dass tPTO-modifizierte CpG-ODNs im Vergleich zu PD- und PTO-CpG-ODNs die potenziell effektivsten Therapieergebnisse erzielten und wurden folglich ausschließlich für weitere Versuche verwendet. Ko-Kultivierungsversuche bestätigten ferner, dass an Tumorzellen adhärierte NANO:BICs von Immunzellen aufgenommen und diese dadurch aktiviert werden konnten.
Im letzten Teil der Arbeit wurden die Funktionen der NANO:BICs in vivo in dem syngenen Balb/c-Tiermodell evaluiert. Die initial eingesetzte Dosis der NANO:BICs führte nicht zu einer signifikanten antitumoralen Wirkung und zeigte lediglich Trends zu einer verstärkten myeloiden Zellrekrutierung. Auch eine Dosissteigerung und Anpassung des Behandlungsschemas blieben ohne deutliche Verbesserung. Distributionsanalysen deuteten darauf hin, dass sich die Partikel in den Metastasen unzureichend anreicherten. Für zukünftige Studien empfiehlt sich daher eine Anpassung des Injektionsschemas an die Metastasengröße sowie eine mögliche Erhöhung der therapeutischen Injektionsintervallen, um die Partikelakkumulation zu optimieren und so das anti-tumorale Potenzial der NANO:BICs voll auszuschöpfen. Ferner bietet es sich an, ein syngenes, orthotopes Modell in weiterführenden Studien heranzuziehen, um die natürlichen Metastasierungwege besser nachbilden zu können.The development of novel immunotherapies has made tremendous progress in the treatment of malignant diseases in recent years. Many strategies aim to reactivate the often ineffective, exhausted, or suppressed effector functions of the immune system and thereby initiate an antitumor immune response. Despite these successes, difficulties such as immune-mediated side effects or the emergence of resistance continue to occur, underlining the ongoing need for innovative therapeutic concepts. Bio-immune conjugates in nanoparticle form (NANO:BICs) offer a particularly promising approach for targeted immunotherapy of malignancies. Building on the RICIA platform established in Prof. Temme’s group - which selectively stimulates EGFRvIII-positive tumor cells via TLR3 - this work developed PSCA-specific, TLR9-targeting NANO:BICs. These were designed to activate tumor-infiltrating immune cells in PSCA-positive renal cell carcinoma metastases through CpG-oligodeoxynucleotide (ODN) - mediated TLR9 stimulation. Each nanoparticle was assembled from three components in defined stoichiometry: a monobiotinylated single-chain antibody against PSCA (or PSMA as a negative control), the biotin-binding linker protein NeutrAvidin and monobiotinylated CpG-ODNs as TLR9 agonists. CpG-ODNs capable of activating either human or murine TLR9 were tested in various assays. In addition, different sugar-phosphate backbones were compared: the naturally occurring phosphodiester (PD) - which is nuclease-sensitive and potentially unstable, the fully modified phosphorothioate (PTO) - which is more stable but can provoke nonspecific interactions and a partially PTO-modified backbone (tPTO) that combines advantages of both. Unbound CpG-ODNs were removed by purification to avoid nonspecific immune activation. In the first part of the study, the NANO:BICs were assembled. Monobiotinylated single-chain
antibodies were produced and purified from scFv-producing cell lines. Titration studies then determined the optimal stoichiometry for conjugation to NeutrAvidin, and suitable treatment concentrations of the various CpG-ODNs were identified. Finally, complexes were purified to eliminate unbiotinylated CpG-ODNs.
The second part evaluated NANO:BIC functionality in vitro. A HEK-Blue™ reporter assay revealed that PTO-CpG-ODN-loaded NANO:BICs induced the strongest TLR9-specific activation, whereas PD-CpG-ODNs elicited the weakest response. Treatment of TLR9-positive RPMI-8226-PSCA cells with NANO:BICs triggered the release of multiple immunomodulatory factors, with PD-CpG-ODN particles producing the most pronounced receptor-specific effects. Microscopic analyses showed that PTO-CpG-ODN-NANO:BICs underwent nonspecific uptake, while PD- and tPTO-CpG-ODN-NANO:BICs adhered specifically to PSCA-positive Renca-Luc-PSCA tumor cells. Murine immune cells internalized the particles regardless of
targeting moiety and, upon stimulation, exhibited increased secretion of immunostimulatory molecules and upregulation of activation markers. Based on earlier findings, tPTO-modified CpG-ODNs emerged as the most promising and were therefore used exclusively in subsequent experiments. Co-culture experiments further confirmed that tumor-bound NANO:BICs could be taken up by immune cells, leading to their activation. In the final part of this work, NANO:BIC efficacy was tested in vivo in a syngeneic Balb/c mouse model. The initial dosing regimen did not yield a significant antitumor effect, showing only trends toward enhanced myeloid cell recruitment. Increasing the dose and modifying the treatment schedule likewise failed to improve outcomes. Distribution analyses indicated suboptimal accumulation of particles in tumor metastases. For future studies, adjusting injection schemes to metastasis size and increasing the number of therapeutic administrations
are recommended to optimize nanoparticle accumulation and fully harness the antitumor potential of NANO:BICs. Additionally, employing a syngeneic, orthotopic model could better recapitulate natural metastatic pathways
Advances in Machine Learning Modeling and Optimization for Small Tabular Data in Materials Design
No full textAdvancing materials is essential for driving technological progress with the constant objective of achieving maximum performance at reduced resource use. To this end, machine learning (ML) offers transformative opportunities for materials design. It enables knowledge extraction from datasets and rapid screening to optimize material properties. However, applying ML in this field faces challenges, including interdisciplinary data integration, as well as sparse and limited data. This dissertation addresses these challenges through advancements in data management, ML-based modeling, and optimization of materials designs.
A research data management system is developed to facilitate cross-process and cross-laboratory data integration, focusing on data organization, metadata standardization, and usability. It provides a user-friendly, unified data storage that allows researchers to focus on data provision rather than routine data alignment tasks, thus improving collaboration within interdisciplinary research networks.
For ML modeling, the suitability of various methods for small datasets is evaluated. To assess the predictive performance of ML models, nested cross-validation with random sampling emerges as the most accurate method. For optimization tasks, additional novel data sampling approaches by Euclidean distance or the dataset’s Pareto front improve the assessment of ML model extrapolation capabilities. In model training, automated ML almost consistently outperforms manual approaches, achieving superior prediction accuracy with minimal training time, typically within 15 minutes. Additionally, Multi-task learning (MTL) is explored as a solution for effectively leveraging integrated datasets. MTL with regressor chains achieves up to 15% higher prediction accuracy than single-task methods, particularly in data-scarce scenarios, demonstrating its value for materials design applications.
Applications of ML-driven optimization are demonstrated through two simulation use cases and a drop-tower experimental use case, showcasing the workflow’s capability to achieve almost Pareto-optimal solutions using only a maximum of 200 data points. The findings highlight the potential to optimize materials performance and sustainability, balancing complex trade-offs such as mechanical properties and global warming potential.
This dissertation advances the data-driven design of materials by overcoming data limitations in applying ML. These advancements enhance research efficiency and drive the development of high-performance and sustainable materials.:List of Figures IV
List of Tables IX
Symbols and Abbreviations XI
1 Introduction 1
1.1 Motivation 1
1.2 Scope and limitations 2
1.3 Thesis structure 3
2 Theoretical Foundations 5
2.1 Material classes and properties 5
2.1.1 Concrete 7
2.1.2 Reinforced concrete 8
2.2 Data-driven methods 9
2.2.1 Machine-learning models 9
2.2.2 Evaluating machine learning model performance 12
2.2.3 Hyperparameter optimization 14
2.2.4 Ensemble learning 14
2.2.5 Dimensionality reduction 16
3 State of the Art 18
3.1 Machine-learning application in materials design 18
3.1.1 Machine learning workflows 20
3.1.2 Usability of machine learning 21
3.2 Research data management 22
3.2.1 Research data infrastructures 23
3.2.2 Metadata standards and schemas 24
3.2.3 Technical languages 25
3.2.4 Data Integration 26
3.3 Data-driven materials modeling 26
3.3.1 Assessing the predictive performance of machine learning models 27
3.3.2 Machine learning with single-task learning 28
3.3.3 Machine learning with multi-task learning 31
3.4 Optimization Strategies 32
3.4.1 Multi-objective optimization in materials-design 33
3.4.2 Evaluation of model extrapolation performance 35
3.5 Conclusion 36
4 Objectives and Research Questions 38
5 Research Data Management 43
5.1 Problem definition and challenges 43
5.2 Requirements definition 46
5.3 Research data management concept and infrastructure 47
5.3.1 Status quo in research network GRK 2250 47
5.3.2 Research data infrastructure 48
5.4 Developing a metadata process modeler 50
5.5 Developing a metadata thesaurus 53
5.6 Implementation 55
5.7 Dataset description 59
5.7.1 Research network GRK 2250 datasets 59
5.7.2 Public materials engineering datasets 62
6 Data-driven Modeling 66
6.1 Assessing the predictive performance of machine learning models 66
6.1.1 Methods 67
6.1.2 Comparison of evaluation strategies on synthetic datasets 69
6.1.3 Representative data-sampling in real datasets 69
6.1.4 Comparison of evaluation strategies with different data sampling methods on real datasets 70
6.1.5 Conclusion 72
6.2 Benchmarking automated machine learning frameworks in single-task learning 73
6.2.1 Selection of automated machine learning frameworks 74
6.2.2 Implementation and experimental setup 77
6.2.3 Performance comparison 78
6.2.4 Impact of dataset size on model performance 81
6.2.5 Discussion of the frameworks 81
6.2.6 Conclusion 85
6.3 Single-task learning for time series data 86
6.3.1 Featurization of time series data 86
6.3.2 Time series datasets 86
6.3.3 Performance comparison 88
6.3.4 Discussion of the featurization 90
6.3.5 Conclusion 91
6.4 Multi-task Learning 91
6.4.1 Multi-task learning methods 92
6.4.2 Performance of multi-task learning in comparison to single-task learning 97
6.4.3 Discussion of multi-task learning performance 102
6.4.4 Conclusion 105
6.5 Case study: modeling drop-tower experiments 106
7 Data-driven Optimization 109
7.1 Methods for data-driven optimization 110
7.1.1 Machine learning models for optimization 110
7.1.2 Evaluating model performance in optimization 110
7.1.3 Multi-objective optimization 112
7.2 Use cases 113
7.2.1 Simulation of electrospun polystyrene/polyacrylonitrile fibers 114
7.2.2 Simulation of the textile reinforced concrete cantilever beam 114
7.2.3 Drop-tower experiments with hybrid reinforced concrete 116
7.3 Results and analysis of optimization performance 117
7.3.1 Dataset creation and optimization targets 117
7.3.2 Machine learning model performance across evaluation strategies 118
7.3.3 Performance of the optimizers 119
7.3.4 Optimization performance of the machine learning models 120
7.4 Discussion of data-driven optimization 123
7.4.1 Overall performance insights 123
7.4.2 Comparison of optimization methods 124
7.4.3 Comparison of machine learning models 125
7.4.4 Comparison of evaluation strategies 126
7.5 Case study: sustainable designs for hybrid reinforced mineral-bonded protective layers 127
7.6 Conclusion 131
8 Conclusion and Outlook 133
8.1 Conclusion 133
8.2 Outlook 135
References 137
List of Publications 154
A Appendix 157
A.1 Code and data availability 157
A.2 Data-driven modeling 158
A.2.1 Hyperparameter search spaces 158
A.2.2 Detailed results of multi-task learning compared to single-task learning 160
A.3 Data driven optimization 161
A.3.1 Optimizer comparison 161
A.3.2 Model comparison 162
A.3.3 Material parameters in textile-reinforced concrete simulation 16