Doctoral Thesis
Refine
Year of publication
- 2022 (238) (remove)
Document Type
- Doctoral Thesis (238) (remove)
Language
- English (135)
- German (101)
- Multiple languages (2)
Has Fulltext
- yes (238)
Is part of the Bibliography
- no (238)
Keywords
- ACLF (2)
- CBM (2)
- Digitalisierung (2)
- HADES (2)
- Leberzirrhose (2)
- Nierentransplantation (2)
- Saccharomyces cerevisiae (2)
- mTOR (2)
- 177Lu-PSMA (1)
- ACKR (1)
Institute
- Medizin (76)
- Biochemie, Chemie und Pharmazie (39)
- Biowissenschaften (32)
- Physik (25)
- Informatik und Mathematik (15)
- Geowissenschaften / Geographie (11)
- Neuere Philologien (8)
- Biochemie und Chemie (6)
- Psychologie und Sportwissenschaften (5)
- Gesellschaftswissenschaften (3)
Heart development is a dynamic process modulated by various extracellular and intracellular cues. Cardiac progenitors in vertebrates such as the zebrafish, migrate over to the midline after differentiation from the epiblast (Bakkers, 2011; Rosenthal & Harvey, 2010; Stainier et al., 1996; Trinh & Stainier, 2004). These progenitors form a cardiac disc at the midline which elongates into the linear heart tube. The differentiation and migration of cardiac precursors is modulated by signaling interactions between cardiac precursor cells and their extracellular environment known as the Extracellular Matrix (ECM). Studies have shown that Cell-ECM interactions play a crucial role in sculpting the heart during early morphogenic events (Davis CL, 1924; Männer & Yelbuz, 2019; Rosenthal & Harvey, 2010). One key factor to these processes is the presence of a specialized ECM known as the Basement Membrane (BM). Extracellular basement membrane proteins such as Fibronectin have been shown to modulate these very early migration processes of the cardiomyocyte progenitors (Trinh & Stainier, 2004). As the heart develops further, the linear heart tube is composed of myocardial cells with an inner endothelial cell lining separated by a layer of thick jelly like substance called the cardiac jelly (Barry A, 1948; Davis CL, 1924; Little et al., 1989). The cardiac jelly also called the cardiac basement membrane, has been shown to regulate distinct developmental events during cardiogenesis. This early CJ contains components of the basal lamina such as laminins, fibronectin, hyaluronan as well as non-fibrillar collagens such as Collagen IV (Little et al., 1989). In this study, I aimed to identify ECM molecules of the Basement Membrane in the heart and identify their role in the modulation of cardiac development and regeneration using the zebrafish as my model organism.
I identified genes belonging to the Zebrafish Matrisome expressed during cardiac developmental and regeneration and performed CRISPR/Cas9 sgRNA mediated mutagenesis. I also developed overexpression tools for these genes.
Agrinp168 mutants exhibited no obvious gross morphology defects during cardiac development and were adult viable. Adult mutants exhibited reduced cardiomyocyte proliferation, but no significant difference in cardiomyocyte dedifferentiation post cardiac cryoinjury.
Decorin overexpression through mRNA injections led to increased myocardial wall thickness and DN dcn overexpression through mRNA injections led to loss of cardiac looping during early development.
Mutants for Small Leucine Rich Proteoglycan (SLRP) prelp generated using CRISPR/Cas9 mutagenesis exhibited cardiovascular defects. Close observation of prelp mutant hearts revealed a reduced heart rate and impaired fractional shortening of the ventricle. prelp mutants exhibited an enlarged atrium at 48 hpf and 72 hpf as well as a reduced ventricle size at 72 hpf. Chamber size in the mutant hearts were enlarged irrespective of contractility of the heart. Mutants showed an increased number of Atrial cardiomyocytes, but no change in cell size. On the molecular level, extracellular Laminin localization was disrupted in prelp mutants along with an increase in thickness and volume of the cardiac HA in the CJ suggesting a potential compensatory role, or retention of immaturity of the cardiac jelly in the prelp mutants. Transcriptomics analysis on the prelp mutant hearts revealed downregulation of ECM organization and ECM-Receptor interaction processes in the mutants. Gene Ontology analysis on prelp mutants hearts transcriptome revealed increased MAPK signaling. Interestingly, genes related to degradation of cardiac HA and maturation of cardiac jelly were downregulated, and genes related to epithelial identity of cardiomyocytes were upregulated. Analysis of the mutant hearts at single cell resolution revealed increased number of mutants exhibiting rounded up cardiomyocytes and loss of apical Podocalyxin. Truncated forms of prelp were generated to identify domain specific roles for Prelp, and reintroduction of N-terminal truncated Prelp into the mutants rescued the basal lamina localization and cardiac jelly volume phenotypes. Myocardium specific re-establishment of prelp expression revealed a marked rescue of the mutant cardiovascular phenotype suggesting that tissue specific expression of prelp is not required so long as Prelp is secreted into the CJ. With these data, I’ve elucidated the role of ECM SLRPs in modulation of cardiac chamber morphogenesis process and regeneration of the heart.
Mechanism of the MHC I chaperone TAPBPR and its role in promoting UGGT1-mediated quality control
(2022)
Information about the health status of most nucleated cells is provided through peptides presented on major histocompatibility complex I (pMHC I) on the cell surface. T cell receptors of CD8+ T cells constantly monitor these complexes and allow the immune system to detect and eliminate infected or cancerous cells. Antigenic peptides displayed on MHC I are typically derived from the cellular proteome and are translocated into the lumen of the endoplasmic reticulum (ER) by the ATP-binding cassette (ABC) transporter associated with antigen processing (TAP), which is part of the peptide-loading complex (PLC). In a process called peptide editing, the MHC I-dedicated chaperone tapasin (Tsn) selects peptides for their ability to form stable complexes with MHC I. While initial peptide loading is catalyzed in the confines of the PLC, the second quality control is mediated by TAPBPR, operating in the peptide-depleted cis-Golgi network. TAPBPR was shown to have a more fine-tuning effect on the presented peptide repertoire rather than initial peptide selection. The fundamental mechanism of peptide editing was illuminated by two crystal structures of TAPBPR in complex with peptide-receptive MHC I. Notably, one of these structures reported a structural element that inserted into the peptidebinding pocket. The so-called scoop loop was assumed to be involved in mediating peptide exchange but the underlying mechanism remained undefined. Additionally, latest results suggested that TAPBPR mediates the interaction of the glucosyltransferase UGGT1 with peptide-receptive MHC. To expand the current knowledge of quality control processes in the antigen presentation pathway, the contribution of the scoop loop in peptide editing and the role of TAPBPR in UGGT1-mediated quality control needs to be elucidated. In the first part of this study, TAPBPR proteins with various loop lengths were designed to scrutinize the contribution of the scoop loop in chaperoning peptidereceptive MHC I. In a light-driven approach, the ability of TAPBPR variants to form stable complexes with peptide-free MHC I was tested. These results demonstrated that in a peptide-depleted environment, the scoop loop is of critical importance for TAPBPR to chaperone intrinsically unstable, peptidereceptive MHC I clients. Moreover, fluorescence polarization-based assays allowed the pursuit of peptide exchange in different, native-like environments. Peptide displacement activities of TAPBPR variants illustrated that catalyzed peptide editing is primarily induced by structural elements outside the scoop loop. In a peptide-depleted environment, the scoop loop occupies the position of the peptide C-terminus and acts as an internal peptide surrogate. By combining complex formation and fluorescence polarization experiments, the scoop loop of TAPBPR was shown to be critically important in stabilizing empty MHC I and functions as an internal peptide selector. In the second part of this study, a novel in-vitro glucosylation assay was established to examine the role of TAPBPR in UGGT1-catalyzed re-glucosylation of TAPBPR-bound MHC I clients. Therefore, a peptide-free MHC I-TAPBPR complex with defined glycan species was designed which served as physiological substrate for UGGT1. By subjecting the recombinantly expressed HLA-A*68:02- TAPBPR complex and UGGT1 proteins to the new in-vitro system, UGGT1 was shown to catalyze the transfer of a glucose residue to the N-linked glycan of TAPBPR-bound Man9GlcNAc2-HLA-A*68:02. Moreover, a high-affinity, photocleavable peptide was applied to dissociate the MHC I-chaperone complex. However, in the absence of TAPBPR, no glucosyltransferase activity was observed. Generation of peptide-free MHC I through UV illumination also showed no activity, and only the addition of TAPBPR could restore UGGT1-mediated reglucosylation of the empty MHC I. Independent of the peptide status of HLAA*68:02, the combination of protein glycoengineering and LC-MS analysis implicated that UGGT1 exclusively acts on TAPBPR-chaperoned HLA-A*68:02. The newly established system provided insights into the function of TAPBPR during UGGT1-catalyzed re-glucosylation activity and quality control of MHC I. Taken together, the scoop loop allows TAPBPR to function as MHC I chaperone through stabilizing peptide-receptive MHC I. In a peptide-depleted environment, the loop structure serves as an internal peptide surrogate and can only be dislodged by a high-affinity peptide. Based on these findings, TAPBPR fulfills a dual function in the second level of quality control. On the one hand, TAPBPR functions as peptide editor, shaping the repertoire of presented peptides. On the other hand, TAPBPR mediates peptide-receptive MHC I clients to the folding sensor UGGT1. Here, TAPBPR is essential to promote UGGT1-catalyzed reglucosylation of the N-linked glycan, giving MHC I a second chance to be loaded with an optimal peptide cargo in the peptide loading complex.
Die digitale Revolution stellt viele traditionelle Industrien vor große Herausforderungen. Auf dem Finanzmarkt werden innovative Geschäftsmodelle geschaffen, die die Rahmenbedingungen, unter denen Finanzprodukte und -dienste angeboten werden, drastisch verändern. Infolgedessen entstehen rechtliche Unsicherheiten sowohl für die Marktakteure als auch für die Aufsicht. Diese Unsicherheiten weisen auf die Notwendigkeit hin, den Rechtsrahmen an die technologische und ökonomische Entwicklung anzupassen. Im Rahmen dieser Dissertation werden die Herausforderungen für das Aufsichts- und Wettbewerbsrecht untersucht, die die digitale Transformation des Finanzmarktes verursacht. Der Finanzmarkt wird vor allem durch die Entstehung von FinTechs, durch das Eintreten von BigTechs in den Finanzbereich und durch die Veränderung der Produkte und Dienstleistungen traditioneller Anbieter auf der Grundlage moderner Technologien transformiert. Die Arbeit gibt einen Überblick über die zentralen innovativen Geschäftsmodelle, der mit den zahlreichen praxisrelevanten Beispielen begleitet wird. Anschließend folgt eine Beurteilung des Umfanges und der Effizienz der vorhandenen und vorgeschlagenen aufsichtsrechtlichen Vorschriften. Diese Erkenntnisse dienen als Grundlage für die wettbewerbsrechtliche Analyse des Finanzmarktes mit dem Fokus auf die plattform- und algorithmenbasierten Geschäftsmodelle im zweiten Teil der Dissertation. Da werden wettbewerbsrechtliche Konstellationen betrachtet, die zwischen konkurrierenden innovativen Geschäftsmodellen entstehen. Die Verfasserin befasst sich mit der Problematik der Definition des relevanten Marktes bei den algorithmen- und plattformbasierten Geschäftsmodellen, der Beurteilung ihrer Marktmacht, den möglichen Auswirkungen der Netzwerkeffekte auf die mehrseitigen Plattformen. Ferner werden solche Aspekte wie Datenzugriff als wettbewerbsrechtlicher Faktor und die Rolle der Algorithmen für die Durchführung wettbewerbswidrigen Praktiken analysiert. Infolgedessen wird ein Überblick über mögliche kartellrechtliche Probleme gegeben, die im Rahmen der Digitalisierung im Finanzmarkt auftreten können, sowie die Vorstellungen darüber, wie sich Wettbewerbsfaktoren verändern sollten, wenn die Anwendung der traditionellen Konzepte des Wettbewerbsrechts nicht immer möglich oder sinnvoll ist. Es werden die Bereiche identifiziert, in denen das europäische und deutsche Wettbewerbsrecht derzeit nicht in der Lage ist, die von den innovativen Geschäftsmodellen stammenden Herausforderungen effektiv zu bewältigen. Im dritten Teil der Dissertation wird erörtert, wie sich das Aufsichts- und das Wettbewerbsrecht ergänzen und zusammen ein System der Ex-ante- und Ex-post-Regulierung bilden. Es wird dargestellt, wie die aufsichtsrechtlichen Anforderungen an die Finanzinstitute, FinTechs und BigTechs mit den ausgewählten Wettbewerbsergebnissen im Finanzsektor zusammenhängen. Darüber hinaus erklärt die Verfasserin, inwiefern einige aufsichtsrechtliche Vorschriften wettbewerbsnachteilig oder wettbewerbsfreundlich auswirken können und warum die Koordination zwischen den aufsichts- und wettbewerbsrechtlichen Instrumenten und Ansätzen essenziell ist. Schließlich werden die Vor- und Nachteile unterschiedlicher Optionen zur Regulierung innovativer Geschäftsmodelle auf dem Finanzmarkt auf einer abstrakteren Ebene erwogen.
People can describe spatial scenes with language and, vice versa, create images based on linguistic descriptions. However, current systems do not even come close to matching the complexity of humans when it comes to reconstructing a scene from a given text. Even the ever-advancing development of better and better Transformer-based models has not been able to achieve this so far. This task, the automatic generation of a 3D scene based on an input text, is called text-to-3D scene generation. The key challenge, and focus of this dissertation, now relate to the following topics:
(a) Analyses of how well current language models understand spatial information, how static embeddings compare, and whether they can be improved by anaphora resolution.
(b) Automated resource generation for context expansion and grounding that can help in the creation of realistic scenes.
(c) Creation of a VR-based text-to-3D scene system that can be used as an annotation and active-learning environment, but can also be easily extended in a modular way with additional features to solve more contexts in the future.
(d) Analyze existing practices and tools for digital and virtual teaching, learning, and collaboration, as well as the conditions and strategies in the context of VR.
In the first part of this work, we could show that static word embeddings do not benefit significantly from pronoun substitution. We explain this result by the loss of contextual information, the reduction in the relative occurrence of rare words, and the absence of pronouns to be substituted. But we were able to we have shown that both static and contextualizing language models appear to encode object knowledge, but require a sophisticated apparatus to retrieve it. The models themselves in combination with the measures differ greatly in terms of the amount of knowledge they allow to extract.
Classifier-based variants perform significantly better than the unsupervised methods from bias research, but this is also due to overfitting. The resources generated for this evaluation are later also an important component of point three.
In the second part, we present AffordanceUPT, a modularization of UPT trained on the HICO-DET dataset, which we have extended with Gibsonien/telic annotations. We then show that AffordanceUPT can effectively make the Gibsonian/telic distinction and that the model learns other correlations in the data to make such distinctions (e.g., the presence of hands in the image) that have important implications for grounding images to language.
The third part first presents a VR project to support spatial annotation respectively IsoSpace. The direct spatial visualization and the immediate interaction with the 3D objects should make the labeling more intuitive and thus easier. The project will later be incorporated as part of the Semantic Scene Builder (SeSB). The project itself in turn relies on the Text2SceneVR presented here for generating spatial hypertext, which in turn is based on the VAnnotatoR. Finally, we introduce Semantic Scene Builder (SeSB), a VR-based text-to-3D scene framework using Semantic Annotation Framework (SemAF) as a scheme for annotating semantic relations. It integrates a wide range of tools and resources by utilizing SemAF and UIMA as a unified data structure to generate 3D scenes from textual descriptions and also supports annotations. When evaluating SeSB against another state-of-the-art tool, it was found that our approach not only performed better, but also allowed us to model a wider variety of scenes. The final part reviews existing practices and tools for digital and virtual teaching, learning, and collaboration, as well as the conditions and strategies needed to make the most of technological opportunities in the future.
Protein biosynthesis is a fundamental process across all domains of life. Polypeptides are produced by translating the genetic information of the messenger RNA (mRNA) into amino acids. This elaborate procedure is divided into the four distinct phases: initiation, elongation, termination, and ribosome recycling. The phases are controlled and regulated by a multitude of translation factors. During initiation, the ribosome assembles on the mRNA. Initiation factors (IFs) bind to the small ribosomal subunit (SSU) and assist the recruitment of mRNA and initiator transfer RNA (tRNA), which delivers the first amino acid methionine. After positioning the SSU at the start codon of the mRNA, additional IFs support the joining of the large ribosomal subunit (LSU). Next, elongation factors (EFs) deliver amino-acylated tRNAs (aa-tRNAs) to the translating ribosome and assist kinetic proofreading and ribosome subunit translocation after the catalytic transfer of the polypeptide onto the aa-tRNA. When a stop codon is reached, translation is terminated by release factors (RFs) that hydrolyze the peptidyl-tRNA to release the nascent protein chain. Afterwards, the ribosome is recycled in Eukaryotes and Archaea by the conserved and essential factor ABCE1, which splits the ribosome into the LSU and SSU. ABCE1 remains bound to the SSU forming the post-splitting complex (post-SC). mRNA translation closes into a cycle by recruitment of IFs to the post-SC and the start of a new round of initiation. The post-SC presents the platform for translation initiation. However, the role of ABCE1 in initiation remains elusive. Therefore, the main goal of my thesis was to unravel the molecular mechanism of ABCE1 on the post-SC and during initiation complex (IC) assembly.
Using a reconstituted system, the high-resolution structure of the archaeal post-SC was solved by cryogenic electron microscopy (cryo-EM) following the native splitting route. It was the first complete model of an archaeal SSU at atomic resolution and revealed a previously undescribed ribosomal protein, which we termed eS21. The hinge 2 region of ABCE1 was identified to be the major interaction interface that anchors to the SSU. Functional characterization of single residue mutations in hinge 2 unraveled essential interactions with the ribosomal RNA backbone of the SSU. Sensing of SSU-binding was found to be allosterically transmitted to the nucleotide-binding sites (NBSs) for integration into the ATPase cycle of ABCE1.
Reconstitution of the archaeal translation apparatus allowed for dissection of IC assembly in the presence of ABCE1. Three different ICs were resolved by cryo-EM. The results were in accordance with recent structural findings of eukaryotic translation initiation and highlighted that the involvement of ABCE1 is conserved.
In a semi-native approach, recombinant ABCE1 was pulled-down from crenarchaeal cell lysates. Mass spectrometric analysis of co-immunoprecipitated ribosomal complexes identified the association of numerous translation factors to the post-SC in a cellular context. The establishment of the genetic toolbox of the acidothermophilic Sulfolobus acidocaldarius allowed the homologous expression of ABCE1. Pull-down of native ABCE1 revealed similar ribosomal complexes as the semi-native and reconstituted approaches. Together, my results gave first physiological relevance of ABCE1 involvement in mRNA translation initiation in Archaea. Native archaeal ABCE1-ICs were vitrified for structural analysis by cryo-EM. Thereby, future structural analysis will allow to analyze the interactions of ABCE1 on native ICs and identify its role in IC assembly.
To address the molecular process of IC assembly, the binding affinity of aIF1 to the SSU was determined by fluorescence polarization. Similar studies will allow for a detailed functional analysis on IF recruitment to the SSU in presence of ABCE1.
mRNA surveillance and ribosome-associated quality control (RQC) mechanisms evolved to ensure cell viability. The pathways overcome ribosome stalling and defective translation components. Stalled ribosomes are terminated by special RFs, which do not hydrolyze the peptidyl-tRNA, but allow dissociation of the ribosome by ABCE1. Faulty messages are degraded via mRNA decay pathways and the LSU is rescued by RQC factors. Recently, the bacterial RQC factor MutS2 was identified to specifically target collided di- and polysomes but its molecular mechanism remains unknown. In this thesis, initial functional analyses showed tri-phosphate specific nucleotide binding of MutS2. While the dissociation of collided disomes by MutS2 could not be observed, the results pave the way for future in vitro studies of bacterial RQC factors acting on specific ribosome populations.
In the future, mRNA translation research must focus on complex quality control processes to comprehensively understand this fundamental cellular process in a holistic context.
Einleitung: Knochenersatzmaterialien gewinnen in der regenerativen Medizin immer weiter an Bedeutung. Polylactid-co-Glycolid (PLGA) wird als etabliertes synthetisches Biomaterial bereits vielseitig in der medizinischen Versorgung eingesetzt, häufig in Form von Mikropartikeln oder Platzhaltern beladen mit Antibiotika oder immunaktiven Zellen zur topischen Antibiotikatherapie und Freisetzung von regenerativen Zellen in einer Defektzone. Ziel dieser Studie war die Charakterisierung einer neuen Darbietungsform von PLGA auf ihre physikalischen Eigenschaften, ihre immunologische Signatur und Zytokompatibilität. Dieses Material soll den Einsatz von PLGA in Form einer porösen Membran zur Abdeckung von großen Knochendefekten ermöglichen.
Material und Methoden: Das PLGA-Gemisch wurde in Form von scheibenförmigen Prüfkörpern und Membranen aufbereitet. Es wurde eine biomechanische Analyse des Materials auf Bruchfestigkeit, Biegesteifigkeit und seine Diffusionskapazität durchgeführt. Die Scaffolds wurden mit Knochenmarkstammzellen (BMC und MSC) besiedelt. Mittels MTT-Tests wurde die Lebensdauer und die Viabilität der Zellen auf der Prüfkörperoberfläche ermittelt. Die Zellen wurden durch PCR-Analyse auf osteogene Differenzierung untersucht. Zur Erfassung der immunologischen Signatur erfolgte ein Vollblutstimulationstest mit anschließender umfassender Auswertung des Sekretoms mittels Proteom Profiler-Analyse.
Ergebnisse: Das Material wies eine geringe mechanische Stabilität auf. Die PLGA-Membran zeigte sich für Proteine bis 11,4 kDa durchlässig. Es konnte eine stetige Degeneration des Materials in wässrigem Milieu festgestellt werden. In der Sekretom-Analyse des Vollblutstimulationstests zeigte sich eine für die PLGAMembran charakteristische immunologische Signatur mit Aktivierung von proinflammatorischen Faktoren. Das Zellüberleben auf der Prüfkörperoberfläche nahm während des Erfassungszeitraums konstant ab, während die Zellviabiliät auf dem Wellboden stieg. Es konnte keine signifikante osteogene Differenzierung der Stammzellen festgestellt werden.
Diskussion: Die schnelle Degeneration des Materials in wässriger Umgebung beeinflusste sowohl die Stabilität der Membran als auch die Zelladhäsion auf den Scaffolds negativ. Das Material wirkte dabei weder toxisch auf die Zellen, noch induzierte es eine osteogene Differenzierung. Die Analyse der Immunsignatur ließ eine frühe und starke Entzündungsreaktion nach Implantation des Materials vermuten.
Schlussfolgerung und Ausblick: Während das vorgelegte PLGA-Gemisch nicht die mechanischen Eigenschaften eines gewichttragenden Knochenersatzstoffes aufweist, so lässt es sich dennoch erfolgreich mit Stammzellen beladen, die nach Implantation gezielt in der Defektzone freigesetzt werden können. Die Anregung einer starken Immunantwort kann gerade in der frühen Heilungsphase hilfreich sein. Weitere in vivo-Untersuchungen am Tiermodell sind notwendig zur Optimierung möglicher therapeutischer Anwendungen.
Das akut-auf-chronische Leberversagen (ACLF) entsteht durch eine akute Dekompensation einer vorbestehenden Leberzirrhose mit begleitendem Multiorganversagen und ist durch eine sehr hohe Kurzzeitmortalität gekennzeichnet. Die Kriterien der European Association for the Study of Chronic Liver Failure (EASL-CLIF) definieren in diesem Zusammenhang ein Lungenversagen als Abfall des Horovitz-Quotienten unter 200mmHg oder als Abfall des Quotienten aus SpO2/FiO2 unter 214. Während Lungenfunktionsstörungen als unabhängiger Risikofaktor für Mortalität bei Patienten mit Leberzirrhose belegt sind, wurde die Bedeutung von mechanischer Beatmung und Lungenversagen bisher nicht isoliert untersucht. Ziel der Studie ist die Rolle von mechanischer Beatmung und Lungenversagen im ACLF zu analysieren und eine mögliche Unterinterpretation in den etablierten prädiktiven Modellen zu evaluieren.
Es wurden Daten aus 775 Hospitalisationen mit intensivmedizinischer Behandlung am Universitätsklinikum Frankfurt von insgesamt 498 Patienten mit Leberzirrhose im Zeitraum März 2015 bis Juli 2019 retrospektiv erfasst und ausgewertet. ACLF und Organversagen wurden gemäß EASL-CLIF Kriterien definiert. Es erfolgte die statistische Analyse verschiedener Kohorten, wobei ACLF-Patienten mittels Propensity Score hinsichtlich CLIF-C ACLF Score und Geschlecht gematcht und anhand der pulmonalen Beeinträchtigung in eine dreiarmige Testkohorte bestehend aus 49 Patienten mit Lungenversagen, 49 Patienten mit Schutzintubation und 49 Patienten ohne pulmonale Beeinträchtigung aufgeteilt wurden.
216 Patienten zeigten ein ACLF bei Aufnahme, 121 wurden mechanisch beatmet und 68 hatten ein Lungenversagen. In der gematchten Kohorte (n=147) konnten Lungenversagen (HR 3,0) und mechanische Beatmung (HR 1,7) als unabhängige Risikofaktoren für eine hohe 28-Tage-Mortalität identifiziert werden. Der CLIF-C ACLF Score konnte als bestes Modell für die Vorhersage der Kurzzeitmortalität bestätigt werden (AUROC 0,81), zeigte allerdingt deutliche Schwächen in der Subkohorte der Patienten mit Lungenversagen (AUROC 0,49) und mechanischer Beatmung (AUROC 0,68). Eine Kalibrierung des CLIF-C ACLF Scores für diese beiden Risikofaktoren liefert auch in der Gesamtkohorte aller zirrhotischer Patienten eine signifikant verbesserte prädiktive Performance (AUROC 0,87; p=0,001).
Mechanische Beatmung und Lungenversagen sind im Rahmen eines ACLF mit einer erhöhten Kurzzeitmortalität assoziiert und nur ungenügend in den etablierten Prädiktionsmodellen abgebildet. Die Kalibrierung des CLIF-C ACLF Scores für diese Risikofaktoren führt zu einer verbesserten Vorhersagegenauigkeit.
Untersuchungen zur Bedeutung selektiver Autophagie für Alterungsprozesse von Podospora anserina
(2022)
Das Ziel der vorliegenden Arbeit war, die Funktion und die Rolle von Autophagie-assoziierten Proteinen im Alternsmodell Podospora anserina zu untersuchen und einen Einblick in die nicht-selektive Autophagie, die Mitophagie und die Bildung und den Abbau von Autophagosomen im Zusammenhang zur Alterung von P. anserina zu analysieren. Dabei wurden folgende Erkenntnisse erhalten:
1. Die Untersuchungen zu ΔPaAtg8 bestätigen, dass die PaATG8-abhängige Autophagosomenbildung zur Aufrechterhaltung der Lebensspanne benötigt wird. In ΔPaAtg8 kommt es zu einem Verlust der nicht-selektiven Autophagie. Die Mitophagie hingegen ist auch ohne PaATG8 partiell möglich und es liegt ein PaATG8-unabhängiger Abbau von mitochondrialen Proteinen in P. anserina vor.
2. In P. anserina ist PaATG11 an der nicht-selektiven Autophagie beteiligt und auch die Mitophagie erfolgt in Abhängigkeit dieses Gerüstproteins. Während der PaAtg11-Deletionsstamm unter Normalbedingungen keinen zum Wildtyp veränderten Phänotyp zeigt, führt eine Kultivierung auf M2-Medium mit Glycerin als einziger Kohlenstoffquelle zu einer starken Verkürzung der Lebensspanne. Eine mikroskopische Untersuchung der Mitochondrien zeigte, dass im juvenilen Altersstadium von ΔPaAtg11 stark fragmentierte Mitochondrien vorliegen. Während der Alterung normalisiert sich die Mitochondrienmorphologie wieder. Der mitochondriale Funktionsverlust wird möglicherweise von den fragmentierten Mitochondrien ausgelöst, denn eine Kultivierung von älteren ΔPaAtg11-Stämmen auf M2-Medium mit Glycerin führt zu einer Normalisierung der Lebensspanne.
3. Die initialen Untersuchungen zur ΔPaAtg11/ΔPaAtg24-Doppelmutante zeigen, dass es bei der Kultivierung unter Normalbedingungen zu einem additiven Effekt der beiden Genverluste kommt. Bei der Anzucht auf M2-Medium mit Glycerin hingegen kann eine im Vergleich zum ΔPaAtg11-Stamm längere Lebensspanne festgestellt werden. Die Mikroskopie der Mitochondrien in ΔPaAtg11/ΔPaAtg24 zeigt, dass im juvenilen Alter zum Wildtyp vergleichbare filamentöse Mitochondrien vorhanden sind.
4. In P. anserina ist PaATG24 kein Mitophagierezeptorprotein, da im PaAtg24-Deletionsstamm eine Beeinträchtigung der nicht-selektiven Autophagie vorliegt. Auch die Mitophagie ist in diesem Stamm geschädigt. Die mikroskopische Betrachtung der Mitochondrien zeigt keinen Unterschied zum Wildtyp. Bei der Untersuchung zur Mitochondrienfunktion durch M2-Medium mit Glycerin ist wie unter Normalbedingungen eine verkürzte Lebensspanne feststellbar.
5. Der Abbau von GFP::PaATG8 ist in der PaAtg24-Deletionsmutante signifikant verringert und es kommt zu einer Akkumulation von Autophagosomen, somit liegt in diesem Stamm eine Beeinträchtigung des autophagosomalen Flusses vor. Bei der mikroskopischen Untersuchung von PaATG24 zeigt sich, dass dieses Protein in P. anserina im Bereich der Vakuolen lokalisiert ist. Die Analyse der Vakuole-Autophagosomen-Fusion zeigt jedoch, dass dieser Mechanismus unabhängig von PaATG24 ist. Die Vakuolenmorphologie und Vakuolengröße ist in ΔPaAtg24 beeinträchtigt und dadurch kommt es zu dem beobachteten Defekt der nicht-selektiven und selektiven Autophagie.
Mutational analysis of ribosomal DNA and maturation-scheme analysis of ribosomal RNA in A. thaliana
(2022)
Ribosome biogenesis is a fundamental cellular process beginning with long precursor rRNA transcription from multi-copies of repetitive 45S ribosomal DNAs. At the subunit level, the primary pre-rRNA transcript encapsuled in 90S protein-RNA complex undergoes decisive splitting in two chief ways for further maturation into large (LSU) and small (SSU) ribosomal subunit. The usage of specific rDNA copies from defined chromosomes and their selective role during growth and development have been a topic of interest owing to its contribution to specialized ribosome theory which proposes non-monolithic functions for ribosomes and thereby their mRNA translation potential. Dual-guide CRISPR/Cas9 mediated disruption of rDNA regions resulted in stable disruption of up to 2.5% and 5% of all rDNA copies in hetero- and homozygous (ploop KD) conditions, respectively. At the RNA level, the mutation excised a critical structural element, P-loop on the LSU 25S rRNA. Mutation caused a dosage dependent defect with homozygosity leading to severe developmental defects through vegetative and reproductive growth phases which is manifested in their proteome by means of disregulation through both increase and decrease of several gene ontological categories of proteins in mutants. Interestingly, the mutation on chromosome 4 triggered dosage compensation through rRNA expression from chromosome 2 further compounded by ectopic rRNA biogenesis defects. The mutated copies however are not incorporated in the translating ribosomes and as a direct or indirect consequence led to elevated basal autophagic levels in the mutants.
The primary 35S transcript is known to undergo two modes of initial cleavages at the pre-rRNA level that aid in their subsequent maturation. Root cell culture (RCC) studies shows that these cells contain a novel ITS2-first cleaved precursor even under control growth conditions, P-C2 adding a third maturation means for the 35S pre-rRNA. This maturation path is further known to be triggered under elevated growth temperature forming a novel adaptive response in Arabidopsis and two other crop plants, tomato, and rice. Taken together, the pulse-chase labeling analysis of control and stressed tissues uncovers the fine-tuned pre-rRNA schematics with crossovers between multiple maturation paths.
In the human brain, the incoming light to the retina is transformed into meaningful representations that allow us to interact with the world. In a similar vein, the RGB pixel values are transformed by a deep neural network (DNN) into meaningful representations relevant to solving a computer vision task it was trained for. Therefore, in my research, I aim to reveal insights into the visual representations in the human visual cortex and DNNs solving vision tasks.
In the previous decade, DNNs have emerged as the state-of-the-art models for predicting neural responses in the human and monkey visual cortex. Research has shown that training on a task related to a brain region’s function leads to better predictivity than a randomly initialized network. Based on this observation, we proposed that we can use DNNs trained on different computer vision tasks to identify functional mapping of the human visual cortex.
To validate our proposed idea, we first investigate a brain region occipital place area (OPA) using DNNs trained on scene parsing task and scene classification task. From the previous investigations about OPA’s functions, we knew that it encodes navigational affordances that require spatial information about the scene. Therefore, we hypothesized that OPA’s representation should be closer to a scene parsing model than a scene classification model as the scene parsing task explicitly requires spatial information about the scene. Our results showed that scene parsing models had representation closer to OPA than scene classification models thus validating our approach.
We then selected multiple DNNs performing a wide range of computer vision tasks ranging from low-level tasks such as edge detection, 3D tasks such as surface normals, and semantic tasks such as semantic segmentation. We compared the representations of these DNNs with all the regions in the visual cortex, thus revealing the functional representations of different regions of the visual cortex. Our results highly converged with previous investigations of these brain regions validating the feasibility of the proposed approach in finding functional representations of the human brain. Our results also provided new insights into underinvestigated brain regions that can serve as starting hypotheses and promote further investigation into those brain regions.
We applied the same approach to find representational insights about the DNNs. A DNN usually consists of multiple layers with each layer performing a computation leading to the final layer that performs prediction for a given task. Training on different tasks could lead to very different representations. Therefore, we first investigate at which stage does the representation in DNNs trained on different tasks starts to differ. We further investigate if the DNNs trained on similar tasks lead to similar representations and on dissimilar tasks lead to more dissimilar representations. We selected the same set of DNNs used in the previous work that were trained on the Taskonomy dataset on a diverse range of 2D, 3D and semantic tasks. Then, given a DNN trained on a particular task, we compared the representation of multiple layers to corresponding layers in other DNNs. From this analysis, we aimed to reveal where in the network architecture task-specific representation is prominent. We found that task specificity increases as we go deeper into the DNN architecture and similar tasks start to cluster in groups. We found that the grouping we found using representational similarity was highly correlated with grouping based on transfer learning thus creating an interesting application of the approach to model selection in transfer learning.
During previous works, several new measures were introduced to compare DNN representations. So, we identified the commonalities in different measures and unified different measures into a single framework referred to as duality diagram similarity. This work opens up new possibilities for similarity measures to understand DNN representations. While demonstrating a much higher correlation with transfer learning than previous state-of-the-art measures we extend it to understanding layer-wise representations of models trained on the Imagenet and Places dataset using different tasks and demonstrate its applicability to layer selection for transfer learning.
In all the previous works, we used the task-specific DNN representations to understand the representations in the human visual cortex and other DNNs. We were able to interpret our findings in terms of computer vision tasks such as edge detection, semantic segmentation, depth estimation, etc. however we were not able to map the representations to human interpretable concepts. Therefore in our most recent work, we developed a new method that associates individual artificial neurons with human interpretable concepts.
Overall, the works in this thesis revealed new insights into the representation of the visual cortex and DNNs...