Filtern
Volltext vorhanden
- ja (29)
Gehört zur Bibliographie
- ja (29)
Erscheinungsjahr
Dokumenttyp
- Dissertation (29)
Schlagworte
- Thrombozyt (3)
- Tissue Engineering (3)
- Bioinformatik (2)
- Cytoskeleton (2)
- Metagenom (2)
- Schlaganfall (2)
- Thrombose (2)
- Zellskelett (2)
- metagenomics (2)
- 3 D bioprinting (1)
Institut
- Graduate School of Life Sciences (29) (entfernen)
Sonstige beteiligte Institutionen
- Biomedical Center Munich, Department of Physiological Chemistry, Ludwig-Maximilians-Universität München (1)
- Chair of Experimental Biomedicine I (1)
- Department of Veterinary Sciences, Experimental Parasitology, Ludwig-Maximilians-Universität München (1)
- GEOMAR Helmholtz-Zentrum für Ozeanforschung Kiel (1)
- Research Center for Infectious Diseases, University of Würzburg (1)
- University of Stellenbosch, Division of Medical Virology (1)
Chlamydia trachomatis, an obligate intracellular human pathogen, is the world’s leading cause of infection related blindness and the most common, bacterial sexually transmitted disease. In order to establish an optimal replicative niche, the pathogen extensively interferes with the physiology of the host cell. Chlamydia switches in its complex developmental cycle between the infectious non-replicative elementary bodies (EBs) and the non-infectious replicative reticulate bodies (RBs). The transformation to RBs, shortly after entering a host cell, is a crucial process in infection to start chlamydial replication. Currently it is unknown how the transition from EBs to RBs is initiated. In this thesis, we could show that, in an axenic media approach, L glutamine uptake by the pathogen is crucial to initiate the EB to RB transition. L-glutamine is converted to amino acids which are used by the bacteria to synthesize peptidoglycan. Peptidoglycan inturn is believed to function in separating dividing Chlamydia. The glutamine metabolism is reprogrammed in infected cells in a c-Myc-dependent manner, in order to accomplish the increased requirement for L-glutamine. Upon a chlamydial infection, the proto-oncogene c-Myc gets upregulated to promote host cell glutaminolysis via glutaminase GLS1 and the L-glutamine transporter SLC1A5/ASCT2. Interference with this metabolic reprogramming leads to limited growth of C. trachomatis. Besides the active infection, Chlamydia can persist over a long period of time within the host cell whereby chronic and recurrent infections establish. C. trachomatis acquire a persistent state during an immune attack in response to elevated interferon-γ (IFN-γ) levels. It has been shown that IFN-γ activates the catabolic depletion of L-tryptophan via indoleamine 2,3-dioxygenase (IDO), resulting in the formation of non-infectious atypical chlamydial forms. In this thesis, we could show that IFN-γ depletes the key metabolic regulator c-Myc, which has been demonstrated to be a prerequisite for chlamydial development and growth, in a STAT1-dependent manner. Moreover, metabolic analyses revealed that the pathogen de routs the host cell TCA cycle to enrich pyrimidine biosynthesis. Supplementing pyrimidines or a-ketoglutarate helps the bacteria to partially overcome the persistent state. Together, the results indicate a central role of c-Myc induced host glutamine metabolism reprogramming and L-glutamine for the development of C. trachomatis, which may provide a basis for anti-infectious strategies. Furthermore, they challenge the longstanding hypothesis of L-tryptophan shortage as the sole reason for IFN-γ induced persistence and suggest a pivotal role of c-Myc in the control of the C. trachomatis dormancy.
Chlamydia trachomatis (Ct) is an obligate intracellular human pathogen. It causes blinding trachoma and sexually transmitted disease such as chlamydia, pelvic inflammatory disease and lymphogranuloma venereum. Ct has a unique biphasic development cycle and replicates in an intracellular vacuole called inclusion. Normally it has two forms: the infectious form, elementary body (EB); and the non-infectious form, reticulate body (RB). Ct is not easily amenable to genetic manipulation. Hence, to understand the infection process, it is crucial to study how the metabolic activity of Ct exactly evolves in the host cell and what roles of EB and RB play differentially in Ct metabolism during infection. In addition, Ct was found regularly coinfected with other pathogens in patients who got sexually transmitted diseases (STDs). A lack of powerful methods to culture Ct outside of the host cell makes the detailed molecular mechanisms of coinfection difficult to study.
In this work, a genome-scale metabolic model with 321 metabolites and 277 reactions was first reconstructed by me to study Ct metabolic adaptation in the host cell during infection. This model was calculated to yield 84 extreme pathways, and metabolic flux strength was then modelled regarding 20hpi, 40hpi and later based on a published proteomics dataset. Activities of key enzymes involved in target pathways were further validated by RT-qPCR in both HeLa229 and HUVEC cell lines. This study suggests that Ct's major active pathways involve glycolysis, gluconeogenesis, glycerolphospholipid biosynthesis and pentose phosphate pathway, while Ct's incomplete tricarboxylic acid cycle and fatty acid biosynthesis are less active. EB is more activated in almost all these carbohydrate pathways than RB. Result suggests the survival of Ct generally requires a lot of acetyl-CoA from the host. Besides, both EB and RB can utilize folate biosynthesis to generate NAD(P)H but may use different pathways depending on the demands of ATP. When more ATP is available from both host cell and Ct itself, RB is more activated by utilizing energy providing chemicals generated by enzymes associated in the nucleic acid metabolism. The forming of folate also suggests large glutamate consumption, which is supposed to be converted from glutamine by the glutamine-fructose-6-phosphate transaminase (glmS) and CTP synthase (pyrG).
Then, RNA sequencing (RNA-seq) data analysis was performed by me in a coinfection study. Metatranscriptome from patient RNA-seq data provides a realistic overview. Thirteen patient samples were collected and sequenced by our collaborators. Six male samples were obtained by urethral swab, and seven female samples were collected by cervicovaginal lavage. All the samples were Neisseria gonorrhoeae (GC) positive, and half of them had coinfection with Ct. HISAT2 and Stringtie were used for transcriptomic mapping and assembly respectively, and differential expression analysis by DESeq2, Ballgown and Cuffdiff2 are parallelly processed for comparison. Although the measured transcripts were not sufficient to assemble Ct's transcriptome, the differential expression of genes in both the host and GC were analyzed by comparing Ct positive group (Ct+) against Ct-uninfected group. The results show that in the Ct+ group, the host MHC class II immune response was highly induced. Ct infection is associated with the regulation of DNA methylation, DNA double-strand damage and ubiquitination. The analysis also shows Ct infection enhances host fatty acid beta oxidation, thereby inducing mROS, and the host responds to reduce ceramide production and glycolysis. The coinfection upregulates GC's own ion transporters and amino acid uptake, while it downregulates GC's restriction and modification systems. Meanwhile, GC has the nitrosative and oxidative stress response and also increases the ability for ferric uptake especially in the Ct+ group compared to Ct-uninfected group.
In conclusion, methods in bioinformatics were used here in analyzing the metabolism of Ct itself, and the responses of the host and GC respectively in a coinfection study with and without Ct. These methods provide metabolic and metatranscriptomic details to study Ct metabolism during infection and Ct associated coinfection in the human microbiota.
Ein Schlüsselereignis, welches dem prognosebestimmenden Organversagen bei systemi-schen Entzündungsprozessen und Sepsis vorangeht, ist die Entwicklung einer mikrovas-kulären endothelialen Schrankenstörung. Das vaskuläre endotheliale (VE-) Cadherin als mechanischer Stabilisator der Endothelbarriere spielt dabei eine wichtige Rolle. In der Inflammation werden Spaltprodukte von VE-Cadherin (sVE-Cadherin) gebildet. Ge-genstand der vorliegenden Arbeit war die Untersuchung der Hypothese ob diese Spalt-produkte selbst an der Störung der endothelialen Barrierefunktion beteiligt sind.
Es wurde hierfür humanes sVE-Cadherin bestehend aus den extrazellulären Domänen EC1-5 (sVE-CadherinEC1-5) generiert. In Messungen des transendothelialen elektrischen Widerstands (TER), mit Immunfluoreszenzfärbungen und Western Blot Analysen wird gezeigt, dass sVE-Cadherin dosisabhängig die Barriere Integrität in primären humanen dermalen Endothelzellen stört. Dies führt zu einer Reduktion von VE-Cadherin und den assoziierten Proteinen α-, γ- und δ-Catenin und ZO-1, die nach der Applikation von sVE-Cadherin an den Zellgrenzen reduziert sind. Die Interaktion zwischen VE-PTP und VE-Cadherin wird durch sVE-CadherinEC1-5 reduziert. Durch pharmakologische Hem-mung der Phosphataseaktivität von VE-PTP mittels AKB9778 wird der durch sVE-CadherinEC1-5-induzierte Verlust der Endothelbarriere aufgehoben. Dagegen zeigt die direkte Aktivierung von Tie-2 mittels Angiopoetin-1 keinen protektiven Effekt auf die durch sVE-CadherinEC1-5 gestörte Endothelbarriere. Weitere Analysen zeigen eine erhöh-te Expression von GEF-H1 durch sVE-CadherinEC1-5. Diese ist ebenfalls durch AKB9778 hemmbar.
Zusätzlich zu diesen Untersuchungen wurden die Konstrukte EC1-4 und EC3-5 in ver-schiedene Vektoren kloniert, um zu bestimmen, ob die extrazelluläre Domäne 5 von VE-Cadherin die dominante Rolle bei den sVE-Cadherin-vermittelten Effekten spielt.
Zusammenfassend zeigen diese Untersuchungen zum ersten Mal, dass sVE-CadherinEC1-5 unabhängig von proinflammatorischen Auslösern über die Aktivierung des VE-PTP/RhoA-Signalweges den Zusammenbruch der Endothelbarriere mitversursacht. Dies stellt einen neuen pathophysiologischer Mechanismus dar, der zum Gesamtverständnis der entzündungsinduzierten Barriereveränderungen des Endothels beiträgt.
Platelets, small anucleated blood cells responsible for hemostasis, interact at sights of injury with several exposed extracellular matrix (ECM) proteins through specific receptors. Ligand binding leads to activation, adhesion and aggregation of platelets. Already megakaryocytes (MKs), the immediate precursor cells in bone marrow (BM), are in constant contact to these ECM proteins (ECMP). The interaction of ECMP with MKs is, in contrast to platelets, less well understood. It is therefore important to study how MKs interact with sinusoids via the underlying ECMP. This thesis addresses three major topics to elucidate these interactions and their role in platelet biogenesis.
First, we studied the topology of ECMP within BM and their impact on proplatelet formation (PPF) in vitro. By establishing a four-color immunofluorescence microscopy we localized collagens and other ECMP and determined their degree of contact towards vessels and megakaryocytes (MKs). In in vitro assays we could demonstrate that Col I mediates increased MK adhesion, but inhibits PPF by collagen receptor GPVI. By immunoblot analyses we identified that the signaling events underyling this inhibition are different from those in platelet activation at the Src family kinase level.
Second, we determined the degree of MK-ECM interaction in situ using confocal laser scanning microscopy of four-color IF-stained femora and spleen sections. In transgenic mouse models lacking either of the two major collagen receptors we could show that these mice have an impaired association of MKs to collagens in the BM, while the MK count in spleen increased threefold. This might contribute to the overall unaltered platelet counts in collagen receptor-deficient mice.
In a third approach, we studied how the equilibrium of ECMP within BM is altered after irradiation. Collagen type IV and laminin-α5 subunits were selectively degraded at the sinusoids, while the matrix degrading protease MMP9 was upregulated in MKs. Platelet numbers decreased and platelets became hyporesponsive towards agonists, especially those for GPVI activation.
Taken together, the results indicate that MK-ECM interaction differs substantially from the well-known platelet-ECM signaling. Future work should further elucidate how ECMP can be targeted to ameliorate the platelet production and function defects, especially in patients after BM irradiation.
The role of miR-21 in the pathophysiology of neuropathic pain using the model of B7-H1 knockout mice
(2017)
The impact of microRNA (miRNA) as key players in the regulation of immune and neuronal gene expression and their role as master switches in the pathophysiology of neuropathic pain is increasingly recognized. miR-21 is a promising candidate that could be linked to the immune and the nociceptive system. To further investigate the pathophysiological role of miR-21 in neuropathic pain, we assesed mice deficient of B7 homolog 1 (B7-H1 ko), a protein with suppressive effect on inflammatory responses.
B7-H1 ko mice and wildtype littermates (WT) of three different age-groups, young (8 weeks), middle-aged (6 months), and old (12 months) received a spared nerve injury (SNI). Thermal withdrawal latencies and mechanical withdrawal thresholds were determined. Further, we investigated anxiety-, depression-like and cognitive behavior. Quantitative real time PCR was used to determine miR-21 relative expression in peripheral nerves, dorsal root ganglia and white blood cells (WBC) at distinct time points after SNI.
Naïve B7-H1 ko mice showed mechanical hyposensitivity with increasing age. Young and middle-aged B7-H1 ko mice displayed lower mechanical withdrawal thresholds compared to WT mice. From day three after SNI both genotypes developed mechanical and heat hypersensitivity, without intergroup differences. As supported by the results of three behavioral tests, no relevant differences were found for anxiety-like behavior after SNI in B7-H1 ko and WT mice. Also, there was no indication of depression-like behavior after SNI or any effect of SNI on cognition in both genotypes. The injured nerves of B7-H1 ko and WT mice showed higher miR-21 expression and invasion of macrophages and T cells 7 days after SNI without intergroup differences. Perineurial miR-21 inhibitor injection reversed SNI-induced mechanical and heat hypersensitivity in old B7-H1 ko and WT mice.
This study reveals that reduced mechanical thresholds and heat withdrawal latencies are associated with miR-21 induction in the tibial and common peroneal nerve after SNI, which can be reversed by perineurial injection of a miR-21 inhibitor. Contrary to expectations, miR-21 expression levels were not higher in B7-H1 ko compared to WT mice. Thus, the B7-H1 ko mouse may be of minor importance for the study of miR-21 related pain. However, these results spot the contribution of miR-21 in the pathophysiology of neuropathic pain and emphasize the crucial role of miRNA in the regulation of neuronal and immune circuits that contribute to neuropathic pain.
Lungenkrebs ist weltweit für die meisten krebsassoziierten Tode verantwortlich. Ursache dafür ist unter anderem, dass viele Medikamente in der klinischen Anwendung, aufgrund nicht übertragbarer Ergebnisse aus der Präklinik, scheitern. Zur Entwicklung neuer Therapiestrategien werden deshalb Modelle benötigt, welche die in vivo Situation besser widerspiegeln. Besonders wichtig ist es dabei, zu zeigen, für welche Fragestellungen ein neues Testsystem valide Ergebnisse liefert.
In dieser Arbeit ist es mit Hilfe des Tissue Engineering gelungen, ein humanes 3D in vitro Lungentumor-Testsystem weiter zu entwickeln und für verschiedene Fragestellungen zu validieren. Zudem konnten sowohl für die Herstellung als auch für die Behandlung der Tumormodelle SOPs etabliert werden. Hier wurde zunächst beobachtet, dass die Auswerteparameter für die Beurteilung von Behandlungseffekten eine geringe Varianz aufweisen und das 3D Modell deshalb als Testsystem geeignet ist.
Ein Vergleich der Morphologie, des EMT-Status und der Differenzierung der Tumorzelllinien im 3D Modell mit Tumorbiopsaten von Adenokarzinompatienten verdeutlichte, dass die 3D Modelle tumorrelevante Merkmale besitzen. So sind die Zelllinien auf der biologischen Matrix, verglichen mit der jeweiligen 2D Kultur, durch eine reduzierte Proliferationsrate gekennzeichnet, welche eher der in vivo Situation entspricht. Für die Etablierung und Validierung des 3D Modells als Testsystem war es notwendig, klinisch relevante Therapien in dem Modell anzuwenden und die Ergebnisse der Behandlung in vitro mit denen im Patienten zu vergleichen. Dabei konnte zunächst bestätigt werden, dass eine zielgerichtete Therapie gegen den EGFR in dem 3D System zu einer verstärkten Induktion der Apoptose im Vergleich zu 2D führt. Dies entspricht klinischen Beobachtungen, bei denen EGFR-mutierte Patienten gut auf eine Therapie mit Tyrosin-Kinase-Inhibitoren (TKI) ansprechen. Anschließend wurde in dieser Arbeit erstmals in vitro gezeigt, dass die Behandlung mit einem HSP90-Inhibitor bei KRAS-Mutation wie in behandelten Patienten keine eindeutigen Vorteile bringt, diese jedoch in Experimenten der 2D Zellkultur mit den entsprechenden Zelllinien vorhergesagt werden. Die Ergebnisse aus dem in vitro Modell spiegeln damit verschiedene klinische Studien wider und unterstreichen das Potenzial des 3D Lungentumor-Testsystems die Wirkung zielgerichteter Therapien vorherzusagen. Durch die Messung von Signalwegsaktivierungen über Phospho-Arrays und Western Blot konnten in dieser Arbeit Unterschiede zwischen 2D und 3D nach Behandlung gezeigt werden. Diese lieferten die Grundlage für bioinformatische Vorhersagen für Medikamente.
Mit fortschreitender Erkrankung und dem Entstehen invasiver Tumore, die möglicherweise Metastasen bilden, verschlechtert sich die Prognose von Krebspatienten. Zudem entwickeln Patienten, die zunächst auf eine Therapie mit TKI ansprechen, bereits nach kurzer Zeit Resistenzen, die ebenfalls zur Progression des Tumorwachstums führen. Zur Wirkungsuntersuchung von Substanzen in solchen fortgeschrittenen Erkrankungsstadien wurde das bestehende Testsystem erweitert. Zum einen wurde mit Hilfe des Wachstumsfaktors TGF-β1 eine EMT ausgelöst. Hier konnte beobachtet werden, dass sich die Expression verschiedener EMT- und invasionsassoziierter Gene und Proteine veränderte und die Zellen vor allem in dynamischer Kultur verstärkt die Basalmembran der Matrix überquerten. Zum anderen wurde die Ausbildung von Resistenzen gegenüber TKI durch die Generierung von resistenten Subpopulationen aus einer ursprünglich sensitiven Zelllinie und anschließender Kultivierung auf der Matrix abgebildet. Dabei zeigte sich keine der klinisch bekannten Mutationen als ursächlich für die Resistenz, sodass weitere Mechanismen untersucht wurden. Hier konnten Veränderungen in der Signaltransduktion sowie der Expression EMT-assoziierter Proteine festgestellt werden.
Im letzten Teil der Arbeit wurde eine neuartige Behandlung im Bereich der Immuntherapie erfolgreich in dem 3D Modell angewendet. Dafür wurden T-Zellen, die einen chimären Antigen-Rezeptor (CAR) gegen ROR1 tragen, in statischer und dynamischer Kultur zu den Tumorzellen gegeben und der Therapieeffekt mittels histologischer Färbung und der Bestimmung der Apoptose evaluiert. Zusätzlich konnten Eigenschaften der T-Zellen, wie deren Proliferation sowie Zytokinausschüttung quantifiziert und damit eine spezifische Wirkung der CAR transduzierten T-Zellen gegenüber Kontroll-T-Zellen nachgewiesen werden.
Zusammenfassend ist es in dieser Arbeit gelungen, ein humanes 3D Lungentumor-Testsystem für die Anwendung in der präklinischen Entwicklung von Krebsmedikamenten sowie der Grundlagenforschung im Bereich der Tumorbiologie zu etablieren. Dieses Testsystem ist in der Lage relevante Daten zu Biomarker-geleiteten Therapien, zur Behandlung fortgeschrittener Tumorstadien und zur Verbesserung neuartiger Therapiestrategien zu liefern.
Marine sponges (phylum Porifera) are simple, sessile, filter-feeder animals. Microbial symbionts are commonly found in the sponge internal tissue, termed the mesohyl. With respect to the microbial content, sponges are classified as either low-microbial abundance sponges (LMA), or high-microbial abundance sponges (HMA). The HMA/LMA dichotomy was explored in this Thesis using the Red Sea sponges as experimental models. A range of methods encompassing transmission electron microscopy, 16S rRNA gene deep sequencing, and metatranscriptomics was employed towards this goal. Here, particular emphasis was placed on the functional analysis of sponge microbiomes.
The Red Sea sponges Stylissa carteri, Xestospongia testudinaria, Amphimedon ochracea, and Crella cyathophora were classified as HMA or LMA sponges using transmission electron microscopy. The diversity, specificity, and transcriptional activity of microbes associated with the sponges S. carteri (LMA) and X. testudinaria (HMA) and seawater were investigated using 16S rRNA amplicon pyrosequencing. The microbial composition of S. carteri was more similar to that of seawater than to that of X. testudinaria, which is consistent with the observation that the sequence data set of S. carteri contained many more possibly seawater sequences (~24%) than the X. testudinaria data set (~6%). The most abundant operational taxonomic units (OTUs) were shared between all three sources (S. carteri, X. testudinaria, seawater), while rare OTUs were unique to any given source. Despite this high degree of overlap, each sponge species contained its own specific microbiota. S. carteri microbiomes were enriched of Gammaproteobacteria and members of the genus Synechococcus and Nitrospira. Enriched members of X. testudinaria microbiomes included Chloroflexi, Deferribacteres, and Actinobacteria. The transcriptional activity of sponge-associated microorganisms was assessed by comparing 16S rRNA gene with transcript amplicons, which showed a good correlation.
The microbial functional gene repertoire of sponges and seawater from the Red Sea (X. testudinaria, S. carteri) and the Mediterranean (Aplysina aerophoba, Dysidea avara) were investigated with the environmental microarray GeoChip 4. Amplicon sequencing was performed alongside in order to assess microbial diversity. The typical microbial diversity patterns characteristic of HMA (abundance of Gammaproteobacteria, Chloroflexi, Acidobacteria, Deferribacteres, and others) and LMA sponges (abundance of Alpha-, Beta-, Gammaproteobacteria, Cyanobacteria, and Bacteroidetes) were confirmed. The HMA/LMA dichotomy was stronger than any possible geographic pattern based on microbial diversity (amplicon) and functional genes (GeoChip). However upon inspection of individual genes detected by GeoChip, very few specific differences were discernible, including differences related to microbial ammonia oxidation, ammonification (higher gene abundance in sponges over seawater) as well as denitrification (lower gene abundance). Furthermore, a higher abundance of a gene, pcc, representative of archaeal autotrophic carbon fixation was noted in sponges over seawater. Thirdly, stress-related genes, in particular those related to radiation, were found in lower abundances in sponge microbiomes than in seawater. With the exception of few documented specific differences, the functional gene repertoire between the different sources appeared largely similar.
The most actively expressed genes of S. carteri microbiomes were investigated with metatranscriptomics. Prokaryotic mRNA was enriched from sponge total RNA, sequenced using Illumina HiSeq technology, and annotated with the metagenomics Rapid Annotation using Subsystem Technology (MG-RAST) pipeline. High expression of archaeal ammonia oxidation and photosynthetic carbon fixation by members of the genus Synechococcus was detected. Functions related to stress response and membrane transporters were among the most highly expressed by S. carteri symbionts. Unexpectedly, gene functions related to methylotrophy were highly expressed by gammaproteobacterial symbionts. The presence of seawater-derived microbes is indicated by the phylogenetic proximity of organic carbon transporters to orthologs of members from the SAR11 clade. In summary, the most expressed functions of the S. carteri-associated microbial community were revealed and linked to the dominant taxonomic members of the microbiome.
In conclusion, HMA and LMA Red Sea sponges were used as models to gain insights into relevant themes in sponge microbiology, i.e. diversity, specificity, and functional activities. Overall, my Thesis contributes to a better understanding of sponge-associated microbial communities, and the implications of this association to marine ecology.
Zielsetzung
Die Entwicklung von Präventionsstrategien zur Senkung der Morbidität und Mortalität aufgrund von kardiovaskulären Erkrankungen (KVE) in der Bevölkerung stellt eine Hauptaufgabe der Epidemiologie und Public Health Forschung dar.
In den vergangenen 20 Jahren rückte die Hochrisikoprävention im Zuge der Weiterentwicklung der Scoringsysteme für das KVE Hochrisiko-Screening in den Fokus der Leitlinien zur KVE Prävention.
Jedoch sind die größten Erfolge aus einer komplementären Strategie aus Hochrisiko- und Populationsprävention mit Priorität auf der Reduktion der Exposition von Risikofaktoren für KVE in der gesamten Population zu erwarten.
Die Grundvoraussetzung für die Entwicklung effizienter, populationsweiter Präventionsprogramme ist das Verständnis einerseits der Rolle von Risikofaktoren bei der Krankheitsentstehung und andererseits der Bedeutung der Risikofaktoren auf Populationsebene.
Der Populations-assoziierte Risikoanteil (PAF) ist das bevorzugte statistische Maß zur Quantifizierung des Effekts von Risikofaktoren auf Populationsebene, da er neben der Effektstärke eines Risikofaktors auch dessen Prävalenz berücksichtigt.
In der Praxis erfolgt die Berechnung des PAF in multifaktoriellen Situationen mithilfe von Adjustierungsansätzen oder Partialisierungsansätzen.
Partialisierungsansätze, zu denen auch der gemittelt sequenzielle PAF (gsPAF) gehört, erfüllen die Additivitätseigenschaft.
Insbesondere der gsPAF kommt daher in der praktischen Anwendung zunehmend häufiger zum Einsatz.
Das Ziel der vorliegenden Arbeit ist die Charakterisierung des gsPAF am Beispiel der Epidemiologie von KVE.
Methoden
In Projekt 1 erfolgt die theoretische Abgrenzung des gsPAF von anderen Adjustierungs- und Partialisierungsverfahren in Bezug auf Intention, Definition, Modellvoraussetzungen und -annahmen und Interpretation. Diese verschiedenen Konzepte werden in einer einheitlichen mathematischen Symbolik dargestellt, um das Verständnis zu erleichtern und Abweichungen in den Definitionen hervorzuheben. Anschließend wird in Projekt 2 der praktische Vergleich von modellbasierten Punktschätzern vorgenommen. Im Rahmen der Sekundäranalyse der ProsCIS-Studie über den Populationseinfluss von Risikofaktoren auf schlechtes Outcome nach Schlaganfall werden dem gsPAF ein additiver und ein multiplikativer Adjustierungsansatz gegenübergestellt und die Schätzergebnisse hinsichtlich Übereinstimmung der Größenordnung und Rangfolgen analysiert. In Projekt 3 werden im Rahmen einer Simulationsstudie nach dem proof-of-concept-Prinzip die asymptotischen Eigenschaften existierender modellfreier und modellbasierter Schätzer des gsPAF in Verbindung mit resamplingbasierten Konfidenzschätzern in einer Situation mit einem binären Outcome und drei binären Risikofaktoren unter insgesamt 296 Modellsituationen charakterisiert. Dabei wird die Abhängigkeit von der Stichprobengröße, der Prävalenz des Outcomes, der Prävalenz und Effektstärke der Risikofaktoren, der stochastischen Abhängigkeit der Risikofaktoren und ihrer Effekte auf das Outcome, der Vollständigkeit des statistischen Modells sowie des Outcome-Mechanismus untersucht. Abschließend erfolgt in Projekt 4 die Demonstration der gsPAF-Schätzung exemplarisch im Rahmen der Sekundäranalyse des deutschen Arms der EUROASPIRE IV-Studie. Hier wird der Einfluss von Baselinefaktoren auf das Auftreten rekurrenter kardiovaskulärer Ereignisse nach erstmaliger Hospitalisierung auf Populationsebene modelliert. Die Ergebnisse werden anschließend einer umfassenden Methodenkritik unterzogen. Dazu wird die Modellanpassung der Regressionsmodelle überprüft, die Performanz der gsPAF-Schätzung mit Hilfe der zuvor entwickelten Simulationsstudie evaluiert, eine exemplarische Stichprobenumfangsplanung durchgeführt sowie die Angemessenheit der Modellannahmen des gsPAF diskutiert.
Ergebnisse
%Die Möglichkeiten der statistischen Modellierung von PAF sind nahezu unbegrenzt.
Projekt 1: Adjustierungs- und Partialisierungsmethoden beantworten verschiedene Fragestellungen. Dies resultiert aus dem unterschiedlichen Umgang beider Methoden mit Subgruppen, die bezüglich mehrerer Risikofaktoren gleichzeitig exponiert sind, und führt infolgedessen auch zu unterschiedlichen Interpretationen. Der PAF beschreibt den Anteil an der Ereigniswahrscheinlichkeit, der mit dem Vorliegen eines Risikofaktors assoziiert ist. Für den gsPAF muss zusätzlich betont werden, dass der Effekt in Subgruppen mit mehreren Risikofaktoren auf additive Weise zerlegt und der Anteil des Zusammenwirkens der beteiligten Risikofaktoren (Surplus) zu gleichen Anteilen den Risikofaktoren zugewiesen wird.
Dahinter steckt die Annahme, dass dieser Teil nur durch das Zusammenwirken überhaupt entstehen konnte, wofür beide Risikofaktoren gleichermaßen verantwortlich gemacht werden. Im Gegensatz zu Adjustierungsmethoden erfüllen Partialisierungsmethoden zwar die Additivitätseigenschaft, gehen jedoch gleichzeitig mit spezifischen Modellannahmen einher, die Kenntnisse über die kausalen Verläufe der Risikofaktoren voraussetzen. Im Falle des gsPAF ist dies die Annahme, dass unter den betrachteten Risikofaktoren keine hierarchischen Abhängigkeiten herrschen.
Die theoretische Basis des gsPAF ist derzeit nur für dichotome Outcomes umfangreich erarbeitet und deckt hier alle Ansprüche für den Praxiseinsatz ab: Modellfreie und modellbasierte Punktschätzer, zugehörige Varianzschätzer mit und ohne Berücksichtigung von Störgrößen und Konfidenzschätzer stehen zur Verfügung. Mathematische Eigenschaften wie Symmetrie, Dummyeigenschaft, Additivität und (internen) marginalen Rationalität des gsPAF und anderer Partialisierungsansätze wurden erörtert. Die verfügbare Software stellt derzeit nur Ausschnitte des Methodenspektrums zur Schätzung des gsPAF bereit und ist deshalb für den Einsatz in der empirischen Forschung zu KVE nur begrenzt nützlich. Eine erfolgreiche und effiziente Recherche zum gsPAF wird durch die uneinheitliche Verwendung der Fachtermini ''partieller'' und ''gemittelt sequenzieller'' PAF erschwert.
Projekt 2: Der Vergleich von Ergebnissen aus einem Adjustierungsansatz mit Ergebnissen aus einem Partialisierungsansatz ist über den kombinierten PAF möglich, da der unterschiedliche Umgang mit Subgruppen, die bezüglich mehrerer Risikofaktoren gleichzeitig exponiert sind, nicht zum Tragen kommt, solange nur der kombinierte Risikofaktor im statistischen Modell berücksichtigt wird. Anhand des Datenbeispiels der ProsCIS-Studie wurde für diesen Parameter keine Abweichung der Ergebnisse des multiplikativen Ansatzes (Faktor 1,0) und nur eine geringe Abweichung des additiven Ansatzes (Faktor 1,1) vom gsPAF beobachtet. Die Größenordnungen der Schätzwerte einzelner Risikofaktoren sowie deren Summe sind zwischen Adjustierungs- und Partialisierungsmethoden nicht vergleichbar. Die Ergebnisse aus dem multiplikativen Regressionsmodell weichen bis zu einem Faktor von 1,3 von den Schätzwerten des gsPAF ab. Die Abweichungen aus dem additiven Regressionsmodell gehen deutlich darüber hinaus. Der gsPAF liefert nahezu additive Schätzergebnisse, während die Summe der risikofaktorspezifischen Schätzwerte aus den beiden Adjustierungsmethoden den kombinierten PAF übersteigt. Im Gegensatz zu vorangegangenen Studien wird die Rangfolge der Risikofaktoren im Datenbeispiel nicht wesentlich von der Schätzmethode beeinflusst.
Projekt 3: Die Simulationsstudie charakterisiert die modellfreien und modellbasierten Punktschätzer des gsPAF und belegt deren Konsistenz und (asymptotische) Erwartungstreue, sofern das statistische Modell korrekt spezifiziert ist. Es zeigt sich, dass in kleinen Stichproben oder bei kleinen Ereigniswahrscheinlichkeiten der modellbasierte Schätzer erwartungstreu und damit dem modellfreien Schätzer überlegen ist. Die Berechnungszeit des modellbasierten Schätzers steigt jedoch superlinear mit steigender Stichprobengröße und mit steigender Anzahl von Variablen im Regressionsmodell an. Resamplingbasierte Methoden wie Bootstrap Normal, Perzentil und Jackknife eignen sich für die Schätzung von Konfidenzintervallen des gsPAF. Auch hier ist ein superlinearer Anstieg der Berechnungszeit insbesondere in Verbindung mit dem modellbasierten Schätzer mit steigender Stichprobengröße und mit steigender Anzahl der Risikofaktoren im statistischen Modell zu beobachten.
Biologische Interaktionen von Risikofaktoren im Outcome-Mechanismus verändern die Wahrscheinlichkeit für Ereignisse in Subgruppen mit mehreren Risikofaktoren weg von einem stochastisch unabhängigen und hin zu einem stochastisch abhängigen Szenario. Diese Ereigniswahrscheinlichkeiten werden durch die Anpassung der Parameter im binär-logistischen Regressionsmodell angenähert. Modelle ohne Interaktionsterme repräsentieren aus statistischer Sicht immer einen Outcome-Mechanismus mit stochastischer Abhängigkeit. Interaktionsterme sind nur dann als biologische Interaktionen zu interpretieren, wenn der biologische Outcome-Mechanismus korrekt durch die logistische Regressionsfunktion beschrieben wird. Anderenfalls dienen die Interaktionsterme nur der Modellanpassung und spiegeln nicht die An- oder Abwesenheit biologischer Interaktionen wider. Die Vernachlässigung von relevanten Interaktionstermen führt zu ernstzunehmenden Verzerrungen der Modellparameter und infolgedessen zu stark verzerrten gsPAF-Schätzungen. Dies ist jedoch durch eine gewissenhafte Überprüfung der Modellanpassung während der Auswertung vermeidbar. Grundsätzlich liefert die modellbasierte Schätzung des gsPAF mit allen Interaktionstermen immer unverzerrte Ergebnisse.
Die benötigte Stichprobengröße für eine aussagekräftige Schätzung des gsPAF übersteigt die für relative Maße und steigt mit der Anzahl zu betrachtender Variablen im Modell und mit sinkender Prävalenz des Outcomes an. Während für den PAF steigende Effektgrößen der Risikofaktoren die benötigte Stichprobengröße verkleinern, wurde in der Simulationsstudie ein umgekehrter Zusammenhang für den gsPAF beobachtet.
Projekt 4: Die in den Projekten 1 und 3 gewonnenen Erkenntnisse wurden im Rahmen der Datenanalyse der EUROASPIRE IV-Studie am Praxisbeispiel untersucht und diskutiert. Das Regressionsmodell ohne Interaktionsterme lieferte verzerrte gsPAF-Schätzungen, was durch die Berücksichtigung von Interaktionstermen korrigiert werden konnte. Die resamplingbasierten Konfidenzintervalle überdeckten große Teile des Wertebereiches des gsPAF und liefern somit keine nützlichen Informationen für die epidemiologische Interpretation der Studienergebnisse. Die Validierung der gsPAF-Schätzungen mit Hilfe der Simulationsstudie machte auf die mangelnde Performanz der Punkt- und Konfidenzintervalle aufgrund der verhältnismäßig kleinen Stichprobengröße für die betrachtete Anzahl der Risikofaktoren aufmerksam. Die benötigte Stichprobengröße für eine performante Schätzung des gsPAF in einer Datensituation wie in der EUROASPIRE IV-Studie beobachtet wurde mit Hilfe der Simulationsstudie ermittelt.
Dabei wurde deutlich, dass etwa das Zehnfache der vorliegenden Stichprobengröße benötigt würde, um den modellfreien Schätzer des gsPAF zusammen mit resamplingbasierten Konfidenzintervallen mit einer ausreichenden Performanz schätzen zu können.
Da unter den in EUROASPIRE IV betrachteten Risikofaktoren hierarchische Abhängigkeiten vorliegen könnten, sind die Voraussetzungen für die Schätzung des gsPAF nicht erfüllt. Anstelle des gsPAF könnte im vorliegenden Beispiel ein adjustierter Schätzer zum Einsatz kommen, oder, sofern genügend Informationen über die kausalen Zusammenhänge unter den Risikofaktoren vorliegen, auch sequenzielle oder proportionale Partialisierungsansätze. Die durchgeführte Methodenkritik in Projekt 4 ermöglicht es, weitere Schritte zur Steigerung der Aussagekraft der Studienergebnisse zu unternehmen, beispielsweise durch die Wahl geeigneter statistischer Methoden und die Erhöhung des Stichprobenumfangs.
Schlussfolgerungen
Die Grundvoraussetzungen für die Gewinnung qualitativ hochwertiger Daten sind bekanntermaßen die Wahl eines der Forschungsfrage angemessenen Studiendesigns sowie die sorgfältige Studienplanung. Aufgrund der hohen Anzahl der Risikofaktoren und Störgrößen für kardiovaskuläre Erkrankungen sowie der Komplexität ihrer kausalen Verläufe erfordern Beobachtungsstudien zu KVE große Stichproben, um eine unverzerrte und valide Schätzung der Effekte von Risikofaktoren zu ermöglichen.
Doch die gewonnenen Erkenntnisse eignen sich nur dann für Schlussfolgerungen im epidemiologischen und Public Health Kontext dann, wenn auch die statistische Analyse der Studiendaten mit einer ebenso hohen Qualität erfolgt.
Eine qualitativ hochwertige Datenanalyse zeichnet sich aus durch
(1) die Auswahl der statistischen Methoden passend zur Forschungsfrage,
(2) die Berücksichtigung aktueller methodischer Forschungsergebnisse,
(3) die sorgfältige Überprüfung der Modellannahmen und Modellanpassung,
(4) die Sicherstellung und Überprüfung einer guten Performanz der Punkt- und Konfidenzschätzer und
(5) die realistische Interpretation der Ergebnisse unter Berücksichtigung der Modellvoraussetzungen und -annahmen.
Ein gewissenhafter Umgang mit den statistischen Methoden ist erforderlich, um belastbare Schlussfolgerungen aus Beobachtungsstudien ziehen zu können. Dies gilt insbesondere im Kontext von Sekundärdatenanalysen, die einen beträchtlichen Anteil der Publikationen darstellen. Simulationsstudien sind ein schlagkräftiges Werkzeug für die Validierung der verwendeten statistischen Methoden und ermöglichen die Einschätzung des Informationsgehaltes von Analyseergebnissen. Sie sind ausgesprochen flexibel und lassen sich an beliebige Datensituationen anpassen. Das macht sie zu einem unverzichtbaren Qualitätskriterium für die Publikation empirischer Studien. Jeder Validierungsschritt trägt wesentlich zu einer verbesserten Qualität der Publikationen bei. Damit entsteht eine solide Basis, um die kausalen Verläufe der Risikofaktoren aufzudecken und die Entwicklung von Präventionsprogrammen zur Verbesserung des Gesundheitsstatus in der Population durch Reduktion der Morbidität und Mortalität von KVE voranzubringen.
The microbial communities that live inside the human gastrointestinal tract -the human gut
microbiome- are important for host health and wellbeing. Characterizing this new “organ”,
made up of as many cells as the human body itself, has recently become possible through
technological advances. Metagenomics, the high-throughput sequencing of DNA directly from
microbial communities, enables us to take genomic snapshots of thousands of microbes living
together in this complex ecosystem, without the need for isolating and growing them.
Quantifying the composition of the human gut microbiome allows us to investigate its
properties and connect it to host physiology and disease. The wealth of such connections was
unexpected and is probably still underestimated. Due to the fact that most of our dietary as well
as medicinal intake affects the microbiome and that the microbiome itself interacts with our
immune system through a multitude of pathways, many mechanisms have been proposed to
explain the observed correlations, though most have yet to be understood in depth.
An obvious prerequisite to characterizing the microbiome and its interactions with the host is
the accurate quantification of its composition, i.e. determining which microbes are present and
in what numbers they occur. Historically, standard practices have existed for sample handling,
DNA extraction and data analysis for many years. However, these were generally developed for
single microbe cultures and it is not always feasible to implement them in large scale
metagenomic studies. Partly because of this and partly because of the excitement that new
technology brings about, the first metagenomic studies each took the liberty to define their own
approach and protocols. From early meta-analysis of these studies it became clear that the
differences in sample handling, as well as differences in computational approaches, made
comparisons across studies very difficult. This restricts our ability to cross-validate findings of
individual studies and to pool samples from larger cohorts. To address the pressing need for
standardization, we undertook an extensive comparison of 21 different DNA extraction methods
as well as a series of other sample manipulations that affect quantification. We developed a
number of criteria for determining the measurement quality in the absence of a mock
community and used these to propose best practices for sampling, DNA extraction and library
preparation. If these were to be accepted as standards in the field, it would greatly improve
comparability across studies, which would dramatically increase the power of our inferences
and our ability to draw general conclusions about the microbiome.
Most metagenomics studies involve comparisons between microbial communities, for example
between fecal samples from cases and controls. A multitude of approaches have been proposed
to calculate community dissimilarities (beta diversity) and they are often combined with
various preprocessing techniques. Direct metagenomics quantification usually counts
sequencing reads mapped to specific taxonomic units, which can be species, genera, etc. Due to
technology-inherent differences in sampling depth, normalizing counts is necessary, for
instance by dividing each count by the sum of all counts in a sample (i.e. total sum scaling), or by
subsampling. To derive a single value for community (dis-)similarity, multiple distance
measures have been proposed. Although it is theoretically difficult to benchmark these
approaches, we developed a biologically motivated framework in which distance measures can
be evaluated. This highlights the importance of data transformations and their impact on the
measured distances.
Building on our experience with accurate abundance estimation and data preprocessing
techniques, we can now try and understand some of the basic properties of microbial
communities. In 2011, it was proposed that the space of genus level variation of the human gut
microbial community is structured into three basic types, termed enterotypes. These were
described in a multi-country cohort, so as to be independent of geography, age and other host
properties. Operationally defined through a clustering approach, they are “densely populated
areas in a multidimensional space of community composition”(source) and were proposed as a
general stratifier for the human population. Later studies that applied this concept to other
datasets raised concerns about the optimum number of clusters and robustness of the
clustering approach. This heralded a long standing debate about the existence of structure and
the best ways to determine and capture it. Here, we reconsider the concept of enterotypes, in
the context of the vastly increased amounts of available data. We propose a refined framework
in which the different types should be thought of as weak attractors in compositional space and
we try to implement an approach to determining which attractor a sample is closest to. To this
end, we train a classifier on a reference dataset to assign membership to new samples. This way,
enterotypes assignment is no longer dataset dependent and effects due to biased sampling are
minimized. Using a model in which we assume the existence of three enterotypes characterized
by the same driver genera, as originally postulated, we show the relevance of this stratification
and propose it to be used in a clinical setting as a potential marker for disease development.
Moreover, we believe that these attractors underline different rules of community assembly and
we recommend they be accounted for when analyzing gut microbiome samples.
While enterotypes describe structure in the community at genus level, metagenomic sequencing
can in principle achieve single-nucleotide resolution, allowing us to identify single nucleotide
polymorphisms (SNPs) and other genomic variants in the gut microbiome. Analysis
methodology for this level of resolution has only recently been developed and little exploration
has been done to date. Assessing SNPs in a large, multinational cohort, we discovered that the
landscape of genomic variation seems highly structured even beyond species resolution,
indicating that clearly distinguishable subspecies are prevalent among gut microbes. In several
cases, these subspecies exhibit geo-stratification, with some subspecies only found in the
Chinese population. Generally however, they present only minor dispersion limitations and are
seen across most of our study populations. Within one individual, one subspecies is commonly
found to dominate and only rarely are several subspecies observed to co-occur in the same
ecosystem. Analysis of longitudinal data indicates that the dominant subspecies remains stable
over periods of more than three years. When interrogating their functional properties we find
many differences, with specific ones appearing relevant to the host. For example, we identify a
subspecies of E. rectale that is lacking the flagellum operon and find its presence to be
significantly associated with lower body mass index and lower insulin resistance of their hosts;
it also correlates with higher microbial community diversity. These associations could not be
seen at the species level (where multiple subspecies are convoluted), which illustrates the
importance of this increased resolution for a more comprehensive understanding of microbial
interactions within the microbiome and with the host.
Taken together, our results provide a rigorous basis for performing comparative metagenomics
of the human gut, encompassing recommendations for both experimental sample processing
and computational analysis. We furthermore refine the concept of community stratification into
enterotypes, develop a reference-based approach for enterotype assignment and provide
compelling evidence for their relevance. Lastly, by harnessing the full resolution of
metagenomics, we discover a highly structured genomic variation landscape below the
microbial species level and identify common subspecies of the human gut microbiome. By
developing these high-precision metagenomics analysis tools, we thus hope to contribute to a
greatly improved understanding of the properties and dynamics of the human gut microbiome.
LIM and SH3 protein 1 (LASP1) is a nucleocytoplasmic scaffolding protein. LASP1 interacts with various cytoskeletal proteins via its domain structure and is known to participate in physiological processes of cells. In the present study, a detailed investigation of the expression pattern of LASP1 protein in normal skin, melanocytic nevi and melanoma was carried out and the melanocyte–specific function of LASP1 was analyzed. LASP1 protein was identified in stratum basale of skin epidermis and a very high level was detected in nevi, the benign tumor of melanocyte. In the highly proliferative basal cells, an additional distinct nuclear localization of the protein was noted. In different tumor entities, an elevated LASP1 expression and nuclear localization, correlated positively with malignancy and tumor grade. However, LASP1 level was determined to be very low in melanoma and even reduced in metastases. Melanoma is distinguished as the first tumor tested to date – that displayed an absence of elevated LASP1 expression. In addition no significant relation was observed between LASP1 protein expression and clinicopathological parameters in melanoma.
The epidermal melanin unit of skin comprises of melanocytes and keratinocytes. Melanocytes are specialized cells that synthesize the photo protective coloring pigment, melanin inside unique organelles called melanosomes. The presence of LASP1 in melanocytes is reported for the first time through this study and the existence was confirmed by immunoblotting analysis in cultured normal human epidermal melanocyte (NHEM) and in melanoma cell lines, along with the immunohistostaining imaging in normal skin and in melanocytic nevi. LASP1 depletion in MaMel2 cells revealed a moderate increase in the intracellular melanin level independently of de novo melanogenesis, pointing to a partial hindrance in melanin release. Immunofluorescence images of NHEM and MaMel2 cells visualized co-localization of LASP1 with dynamin and tyrosinase concomitant with melanosomes at the dendrite tips of the cells. Melanosome isolation experiments by sucrose density gradient centrifugation clearly demonstrated the presence of LASP1 and the melanosome specific markers tyrosinase and TRP1 in late stage melanosomes.
The study identified LASP1 and dynamin as novel binding partners in melanocytes and provides first evidence for the existence of LASP1 and dynamin (a protein well–known for its involvement in vesicle formation and budding) in melanosomes. Co-localization of LASP1 and dynamin along the dendrites and at the tips of the melanocytes indicates a potential participation of the two proteins in the membrane vesicle fission at the plasma membrane.
In summary, a possible involvement of LASP1 in the actin–dynamin mediated membrane fission and exocytosis of melanin laden melanosome vesicles into the extracellular matrix is suggested.
In several countries, a decline in mortality, case-fatality and recurrence rates of stroke was observed. However, studies investigating sex-specific and subtype-specific (pathological and etiological) time trends in stroke mortality, case-fatality and recurrence rates are scarce, especially in Germany. The decline in ischemic stroke mortality and case-fatality might be associated with the high quality of acute care of ischemic stroke, but the exact determinants of early outcome remains unknown for Germany.
Therefore, as first step of this thesis, we investigated the time trends of subtype- and sex-specific age- standardized stroke mortality rates in Germany from 1998 to 2015, by applying joinpoint regression on official causes of death statistics, provided by the Federal Statistical Office. Furthermore, a regional comparison of the time trends in stroke mortality between East and West was conducted. In the second step, time trends in case-fatality and stroke recurrence rates were analyzed using data from a population- based stroke register in Germany between 1996 and 2015. The analysis was stratified by sex and etiological subtype of ischemic stroke. In the third step, quality of stroke care and the association between adherence to measures of quality of acute ischemic stroke care and in-hospital mortality was estimated based on data from nine regional hospital-based stroke registers in Germany from the years 2015 and 2016.
We showed that in Germany, age-standardized stroke mortality declined by over 50% from 1998 to 2015 both, in women and men. Stratified by the pathological subtypes of stroke, the decrease in mortality was larger in ischemic stroke compared to hemorrhagic stroke. Different patterns in the time trends of stroke were observed for stroke subtypes, regions in Germany (former Eastern part of Germany (EG), former Western part of Germany (WG)) and sex, but in all strata a decline was found. By applying joinpoint regression, the number of changes in time trend differed between the regions and up to three changes in the trend in ischemic stroke mortality were detected. Trends in hemorrhagic stroke were in parallel between the regions with up to one change (in women) in joinpoint regression. Comparing the regions, stroke mortality was higher in EG compared to WG throughout the whole observed time period, however the differences between the regions started to diminish from 2007 onwards.
Further it was found that, based on the population-based Erlangen Stroke Project (ESPro), case-fatality and recurrence rates in ischemic stroke patients are still high in Germany. 46% died and 20% got a recurrent stroke within the first five years after stroke. Case-fatality rates declined statistically significant from 1996 to 2015 across all ischemic stroke patients and all etiological subtypes of ischemic stroke. Based on Cox regression no statistically significant decrease in stroke recurrence was observed.
Based on the pooled data of nine regional hospital-based stroke registers from the years 2015 and 2016 covering about 80% of all hospitalized stroke patients in Germany, a high quality of care of acute ischemic stroke patients, measured via 11 evidence-based quality indicators (QI) of process of care, was observed. Across all registers, most QI reached the predefined target values for good quality of stroke care. 9 out of 11 QI showed a significant association with 7-day in-hospital mortality. An inverse linear association between overall adherence to QI and 7-day in-hospital mortality was observed.
In conclusion, stroke mortality and case-fatality showed a favorable development over time in Germany, which might partly be due to improvements in acute treatment. This is supported by the association between overall adherence to quality of care and in-hospital mortality. However, there might be room for improvements in long-term secondary prevention, as no clear reduction in recurrence rates was observed.
Der Meniskus, ein scheibenförmiger Faserknorpel, spielt im Kniegelenk eine bedeutende Rolle, weil er Kräfte und Druck im Kniegelenk gleichmäßig verteilt, Stöße dämpft sowie der Kraftübertragung und Stabilisierung dient. Durch die Entfernung des Gewebes, der sogenannten Totalmeniskektomie, nach einer Meniskusverletzung oder einem Riss, verändern sich die mechanischen Eigenschaften des Gelenks stark und verursachen durch die erhöhte Belastung der Gelenkflächen Arthrose. Arthrose ist weltweit die Häufigste aller Gelenkerkrankungen. Der Erhalt der körperlichen Leistungsfähigkeit und Mobilität bis ins hohe Alter sowie die Bewahrung der Gesundheit von Herz-Kreislauf- und Stoffwechselorganen zählen aufgrund des demografischen Wandels zu den großen medizinischen Herausforderungen. Die Erkrankung des muskuloskelettalen Systems stellte 2010 im Bundesgebiet die am häufigsten vorkommende Krankheitsart dar.
Während Risse in den äußeren Teilen des Meniskus aufgrund des Anschlusses an das Blutgefäßsystem spontan heilen können, können sie dies in tieferen Zonen nicht. Durch die begrenzte Heilungsfähigkeit des Knorpels bleibt langfristig der Einsatz eines Ersatzgewebes die einzige therapeutische Alternative.
In der vorliegenden Arbeit wurde als therapeutische Alternative erfolgreich ein vaskularisiertes Meniskusersatzgewebe mit Methoden des Tissue Engineering entwickelt. Es soll in Zukunft als Implantat Verwendung finden. Tissue Engineering ist ein interdisziplinäres Forschungsfeld, in dem Gewebe außerhalb des Körpers generiert werden. Schlüsselkomponenten sind Zellen, die aus einem Organismus isoliert werden, und Trägerstrukturen, die mit Zellen besiedelt werden. Die Biomaterialien geben den Zellen eine geeignete Umgebung, die die Extrazelluläre Matrix (EZM) ersetzen soll, um die Funktion der Zellen beizubehalten, eigene Matrix zu bilden. Zum Erhalt eines funktionelles Gewebes werden oftmals dynamische Kultursysteme, sogenannte Bioreaktoren, verwendet, die natürliche Stimuli wie beispielsweise den Blutfluss oder mechanische Kompressionskräfte während der in vitro Reifungsphase des Gewebes, zur Verfügung stellen. Das Gewebekonstrukt wurde auf Basis natürlicher Biomaterialien aufgebaut, unter Verwendung ausschließlich primärer Zellen, die später direkt vom Patienten gewonnen werden können und damit Abstoßungsreaktionen auszuschließen sind. Da der Meniskus teilvaskularisiert ist und die in vivo Situation des Gewebes bestmöglich nachgebaut werden sollte, wurden Konstrukte mit mehreren Zelltypen, sogenannte Ko-Kulturen aufgebaut. Neben mikrovaskulären Endothelzellen (mvEZ) und Meniskuszellen (MZ) erfolgten Versuche mit mesenchymalen Stammzellen (MSZ).
Zur Bereitstellung einer zelltypspezifischen Matrixumgebung, diente den mvEZ ein Stück Schweinedarm mit azellularisierten Gefäßstrukturen (BioVaSc®) und den MZ diente eine geeig- nete Kollagenmatrix (Kollagen Typ I Hydrogel). Die Validierung und Charakterisierung des aufgebauten 3D Meniskuskonstrukts, welches in einem dynamischen Perfusions-Bioreaktorsystem kultiviert wurde, erfolgte mit knorpeltypischen Matrixmarkern wie Aggrekan, Kollagen Typ I, II und X sowie mit den Transkriptionsfaktoren RunX2 und Sox9, die in der Knorpelentstehung von großer Bedeutung sind. Zusätzlich erfolgten Auswertungen mit endothelzellspezifischen Markern wie vWF, CD31 und VEGF, um die Vaskularisierung im Konstrukt nachzuweisen. Analysiert wurden auch die Zellvitalitäten in den Konstrukten.
Aufgrund einer nur geringen Verfügbarkeit von MZ wurden Kulturansätze mit alternativen Zellquellen, den MSZ, durchgeführt. Dafür erfolgte zunächst deren Isolation und Charakterisierung und die Auswahl einer geeigneten 3D Kollagenmatrix. Die beste Zellintegration der MSZ konnte auf einer eigens hergestellten elektrogesponnenen Matrix beobachtet werden. Die Matrix besteht aus zwei unterschiedlichen Kollagentypen, die auf insgesamt fünf Schichten verteilt sind. Die Fasern besitzen weiter unterschiedliche Ausrichtungen. Während die Kollagen Typ I Fasern in den äußeren Schichten keiner Ausrichtung zugehören, liegen die Kollagen Typ II Fasern in der mittleren Schicht parallel zueinander. Der native Meniskus war für den Aufbau einer solchen Kollagen-Trägerstruktur das natürliche Vorbild, das imitiert werden sollte. Nach der Besiedelung der Matrix mit MSZ, konnte eine Integration der Zellen bereits nach vier Tagen bis in die Mittelschicht sowie eine spontane chondrogene Differenzierung nach einer insgesamt dreiwöchigen Kultivierung gezeigt werden. Das Biomaterial stellt in Hinblick auf die Differenzierung der Zellen ohne die Zugabe von Wachstumsfaktoren eine relevante Bedeutung für klinische Studien dar.
Zur Kultivierung des 3D Meniskuskonstrukts wurde ein Bioreaktor entwickelt. Mit diesem können neben Perfusion der Gefäßsysteme zusätzlich Kompressionskräfte sowie Scherspannungen auf das Ersatzgewebe appliziert und die Differenzierung von MZ bzw. MSZ während der in vitro Kultur über mechanische Reize stimuliert werden. Ein anderes Anwendungsfeld für den neuartigen Bioreaktor ist seine Verwendung als Prüftestsystem für die Optimierung und Qualitätssicherung von Gewebekonstrukten.
Platelets are important players in haemostasis and their activation is essential to limit post-traumatic blood loss upon vessel injury. On the other hand, pathological platelet activation may lead to thrombosis resulting in myocardial infarction and stroke. Platelet activation and subsequent thrombus formation are, therefore, tightly regulated and require a well-defined interplay of platelet surface receptors, intracellular signalling molecules, cytoskeletal rearrangements and the activation of the coagulation cascade.
In vivo thrombosis and haemostasis models mimic thrombus formation at sites of vascular lesions and are frequently used to assess thrombotic and haemostatic functions of platelets. In this dissertation, different in vivo models were used in mice to address the question at what level a reduced platelet count (PC) compromises stable thrombus formation. To study this, mice were rendered thrombocytopenic by low-dose anti-GPIbα antibody treatment and subjected to a tail bleeding time assay as well as to four different in vivo thrombosis models. Haemostasis and occlusive thrombus formation in small vessels were only mildly affected even at severe reductions of the PC. In contrast, occlusive thrombus formation in larger arteries required higher PCs demonstrating that considerable differences in the sensitivity for PC reductions exist between these models.
In a second part of this study, mice were rendered thrombocytopenic by injection of high-dose anti-GPIbα antibody which led to the complete loss of all platelets from the circulation for several days. During recovery from thrombocytopenia, the newly generated platelet population was characterised and revealed a defect in immunoreceptor tyrosine-based activation motif (ITAM)-signalling. This defect translated into impaired arterial thrombus formation.
To further investigate ITAM-signalling in vivo, genetically modified mice were analysed which display a positive or negative regulation of platelet ITAM-signalling in vitro. Whereas mice lacking the adapter Grb2 in platelets showed a delayed thrombus formation in vivo after acetylsalicylic acid treatment, Clp36ΔLIM bone marrow chimeric mice and SLAP/SLAP2-deficient mice displayed pro-thrombotic properties in vivo. Finally, mice lacking the adapter protein EFhd2 were analysed in vitro and in vivo. However, EFhd2-deficient platelets showed only a minor increase in the procoagulant activity compared to control.
Traditionally, ischemic stroke has been regarded as the mere consequence of cessation of cerebral blood flow, e.g. due to the thromboembolic occlusion of a major brain supplying vessel. However, the simple restoration of blood flow via thrombolysis and/or mechanical recanalization alone often does not guarantee a good functional outcome. It appears that secondary detrimental processes are triggered by hypoxia and reoxygenation, which are referred to as ischemia/reperfusion (I/R) injury. During recent years it became evident that, beside thrombosis inflammation and edema formation are key players in the pathophysiology of cerebral ischemia. The contact-kinin system represents an interface between thrombotic, inflammatory and edematous circuits. It connects the intrinsic coagulation pathway with the plasma kallikrein-kinin system (KKS) via coagulation factor FXII.
The serine protease inhibitor C1-inhibitor (C1-INH) has a wide spectrum of inhibitory activities and counteracts activation of the contact-kinin system at multiple levels. The first part of the thesis aimed to multimodally interfere with infarct development by C1-INH and to analyze modes of actions of human plasma derived C1-INH Berinert® P in a murine model of focal cerebral ischemia. It was shown that C57BL/6 mice following early application of 15.0 units (U) C1-INH, but not 7.5 U developed reduced brain infarctions by ~60% and less neurological deficits in the model of transient occlusion of the middle cerebral artery (tMCAO). This protective effect was preserved at more advanced stages of infarction (day 7), without increasing the risk of intracerebral bleeding or affecting normal hemostasis. Less neurological deficits could also be observed with delayed C1-INH treatment, whereas no improvement was achieved in the model of permanent MCAO (pMCAO). Blood-brain-barrier (BBB) damage, inflammation and thrombosis were significantly improved following 15.0 U C1-INH application early after onset of ischemia. Based on its strong antiedematous, antiinflammatory and antithrombotic properties C1-INH constitutes a multifaceted therapeutic compound that protects from ischemic neurodegeneration in ‘clinically meaningful’ settings.
The second part of the thesis addresses the still elusive functional role of macrophages in the early phase of stroke, especially the role of the macrophage-specific adhesion molecule sialoadhesin (Sn). For the first time, sialoadhesin null (Sn-/-) mice, homozygous deficient for Sn on macrophages were subjected to tMCAO to assess the clinical outcome. Neurological and motor function was significantly improved in Sn-/- mice on day 1 after ischemic stroke compared with wildtype (Sn+/+) animals. These clinical improvements were clearly detectable even on day 3 following tMCAO. Infarctions on day 1 were roughly the same size as in Sn+/+ mice and did not grow until day 3. No intracerebral bleeding could be detected at any time point of data acquisition. Twenty four hours after ischemia a strong induction of Sn was detectable in Sn+/+ mice, which was previously observed only on perivascular macrophages in the normal brain. Deletion of Sn on macrophages resulted in less disturbance of the BBB and a reduced number of CD11b+ (specific marker for macrophages/microglia) cells, which, however, was not associated with altered expression levels of inflammatory cytokines. To further analyze the function of macrophages following stroke this thesis took advantage of LysM-Cre+/-/IKK2-/- mice bearing a nuclear factor (NF)-ϰB activation defect in the myeloid lineage, including macrophages. Consequently, macrophages were not able to synthesize inflammatory cytokines under the control of NF-ϰB. Surprisingly, infarct sizes and neurological deficits upon tMCAO were roughly the same in conditional knockout mice and respective wildtype littermates. These findings provide evidence that macrophages do not contribute to tissue damage and neurological deficits, at least, not by release of inflammatory cytokines in the early phase of cerebral ischemia. In contrast, Sn which is initially expressed on perivascular macrophages and upregulated on macrophages/microglia within the parenchyma following stroke, influenced functional outcome.
In eukaryotes, the enormously long DNA molecules need to be packaged together with histone proteins into nucleosomes and further into compact chromatin structures to fit it into the nucleus. This nuclear organisation interferes with all phases of transcription that require the polymerase to bind to DNA. During transcription – the process in which the hereditary information stored in DNA is transferred to many transportable RNA molecules - nucleosomes form a physical obstacle for polymerase progression. Thus, transcription is usually accompanied by processes mediating nucleosome destabilisation, including post-translational histone modifications (PTMs) or exchange of canonical histones by their variant forms. To the best of our knowledge, acetylation of histones has the highest capability to induce chromatin opening. The lysine modification can destabilise histone-DNA interactions within a nucleosome and can serve as a binding site for various chromatin remodelers that can modify the nucleosome composition. For example, H4 acetylation can impede chromatin folding and can stimulate the exchange of canonical H2A histone by its variant form H2A.Z at transcription start sites (TSSs) in many eukaryotes, including humans. As histone H4, H2A.Z can be post-translationally acetylated and as acetylated H4, acetylated H2A.Z is enriched at TSSs suggested to be critical for transcription. However, thus far, it has been difficult to study the cause and consequence of H2A.Z acetylation.
Even though, genome-wide chromatin profiling studies such as ChIP-seq have already revealed the genomic localisation of many histone PTMs and variant proteins, they can only be used to study individual chromatin marks and not to identify all factors important for establishing a distinct chromatin structure. This would require a comprehensive understanding of all marks associated to a specific genomic locus. However, thus far, such analyses of locus-specific chromatin have only been successful for repetitive regions, such as telomeres.
In my doctoral thesis, I used the unicellular parasite Trypanosoma brucei as a model system for chromatin biology and took advantage of its chromatin landscape with TSSs comprising already 7% of the total T. brucei genome (humans: 0.00000156%). Atypical for a eukaryote, the protein-coding genes are arranged in long polycistronic transcription units (PTUs). Each PTU is controlled by its own ~10 kb-wide TSS, that lies upstream of the PTU. As observed in other eukaryotes, TSSs are enriched with nucleosomes containing acetylated histones and the histone variant H2A.Z. This is why I used T. brucei to particularly investigate the TSS-specific chromatin structures and to identify factors involved in H2A.Z deposition and transcription regulation in eukaryotes. To this end, I established an approach for locus-specific chromatin isolation that would allow me to identify the TSSs- and non-TSS-specific chromatin marks. Later, combining the approach with a method for quantifying lysine-specific histone acetylation levels, I found H2A.Z and H4 acetylation enriched in TSSs-nucleosomes and mediated by the histone acetyltransferases HAT1 and HAT2. Depletion of HAT2 reduced the levels of TSS-specific H4 acetylation, affected targeted H2A.Z deposition and shifted the sites of transcription initiation. Whereas HAT1 depletion had only a minor effect on H2A.Z deposition, it had a strong effect on H2A.Z acetylation and transcription levels. My findings demonstrate a clear link between histone acetylation, H2A.Z deposition and transcription initiation in the early diverged unicellular parasite T. brucei, which was thus far not possible to determine in other eukaryotes. Overall, my study highlights the usefulness of T. brucei as a model system for studying chromatin biology. My findings allow the conclusion that H2A.Z regardless of its modification state defines sites of transcription initiation, whereas H2A.Z acetylation is essential co-factor for transcription initiation. Altogether, my data suggest that TSS-specific chromatin establishment is one of the earliest developed mechanisms to control transcription initiation in eukaryotes.
Transcription describes the process of converting the information contained in DNA into RNA. Although, tremendous progress has been made in recent decades to uncover this complex mechanism, it is still not fully understood. Given the advances and reduction in cost of high-throughput sequencing experiments, more and more data have been generated to help elucidating this complex process. Importantly, these sequencing experiments produce massive amounts of data that are incomprehensible in their raw form for humans. Further, sequencing techniques are not always 100% accurate and are subject to a certain degree of variability and, in special cases, they might introduce technical artifacts. Thus, computational and statistical methods are indispensable to uncover the information buried in these datasets.
In this thesis, I worked with multiple high throughput datasets from herpes simplex virus 1 (HSV-1) and human cytomegalovirus (HCMV) infections. During the last decade, it has became clear that a gene might not have a single, but multiple sites at which transcription initiates. These multiple transcription start sites (TiSS) demonstrated to have regulatory effects on the gene itself depending on which TiSS is used. Specialized experimental approaches were developed to help identify TiSS (TiSS-profiling). In order to facilitate the identification of all potential TiSS that are used for cell type- and condition-specific transcription, I developed the tool iTiSS. By using a new general enrichment-based approach to predict TiSS, iTiSS proved to be applicable in integrated studies and made it less prone to false positives compared to other TiSS-calling tools. Another improvement in recent years was made in metabolic labeling experiments such as SLAM-seq. Here, they removed the time consuming and laborious step of physically separating new from old RNA in the samples. This was achieved by inducing specific nucleotide conversions in newly synthesized RNA that are later visible in the data. Consequently, the separation of new and old RNA is now done computationally and, hence, tools are needed that accurately quantify these fold-changes. My second tool that I developed, called GRAND-SLAM proved to be capable to accomplish this task and outperform competing programs. As both of my tools, iTiSS and GRAND-SLAM are not specifically tailored to my own goals, but could also facilitate the research of other groups in this field, I made them publicly available on GitHub.
I applied my tools to datasets generated in our lab as well as to publicly available data sets from HSV-1 and HCMV, respectively. For HSV-1, I was able to predict and validate TiSS with nucleotide precision using iTiSS. This has lead to the most comprehensive annotation for HSV-1 to date, which now serves as the fundamental basis of any future transcriptomic research on HSV-1. By combining both my tools, I was further able to uncover parts of the highly complex gene kinetics in HCMV and to resolve the limitations caused by the densely packed genome of HCMV.
With the ever-increasing advances in sequencing techniques and their decrease in cost, the amounts of data produced will continue to rise massively in the future. Additionally, more and more specialized omics approaches are appearing, calling for new tools to leverage their full information potential. Consequently, it has become apparent that specialized computational tools such as iTiSS and GRAND-SLAM are needed and will become an essential and indispensable part of the analysis.
Since the advent of high-throughput sequencing technologies in the mid-2010s, RNA se-
quencing (RNA-seq) has been established as the method of choice for studying gene
expression. In comparison to microarray-based methods, which have mainly been used to
study gene expression before the rise of RNA-seq, RNA-seq is able to profile the entire
transcriptome of an organism without the need to predefine genes of interest. Today,
a wide variety of RNA-seq methods and protocols exist, including dual RNA sequenc-
ing (dual RNA-seq) and multi RNA sequencing (multi RNA-seq). Dual RNA-seq and
multi RNA-seq simultaneously investigate the transcriptomes of two or more species, re-
spectively. Therefore, the total RNA of all interacting species is sequenced together and
only separated in silico. Compared to conventional RNA-seq, which can only investi-
gate one species at a time, dual RNA-seq and multi RNA-seq analyses can connect the
transcriptome changes of the species being investigated and thus give a clearer picture of
the interspecies interactions. Dual RNA-seq and multi RNA-seq have been applied to a
variety of host-pathogen, mutualistic and commensal interaction systems.
We applied dual RNA-seq to a host-pathogen system of human mast cells and Staphylo-
coccus aureus (S. aureus). S. aureus, a commensal gram-positive bacterium, can become
an opportunistic pathogen and infect skin lesions of atopic dermatitis (AD) patients.
Among the first immune cells S. aureus encounters are mast cells, which have previously
been shown to be able to kill the bacteria by discharging antimicrobial products and re-
leasing extracellular traps made of protein and deoxyribonucleic acid (DNA). However,
S. aureus is known to evade the host’s immune response by internalizing within mast
cells. Our dual RNA-seq analysis of different infection settings revealed that mast cells
and S. aureus need physical contact to influence each other’s gene expression. We could
show that S. aureus cells internalizing within mast cells undergo profound transcriptome
changes to adjust their metabolism to survive in the intracellular niche. On the host side,
we found out that infected mast cells elicit a type-I interferon (IFN-I) response in an
autocrine manner and in a paracrine manner to non-infected bystander-cells. Our study
provides the first evidence that mast cells are capable to produce IFN-I upon infection
with a bacterial pathogen.
Sponges (phylum Porifera) are evolutionary ancient, sessile filter-feeders that harbor a largely diverse microbial community within their internal mesohyl matrix. Throughout this thesis project, I aimed at exploring the adaptations of these symbionts to life within their sponge host by sequencing and analyzing the genomes of a variety of bacteria from the microbiome of the Mediterranean sponge Aplysina aerophoba. Employed methods were fluorescence-activated cell sorting with subsequent multiple displacement amplification and single-cell / ‘mini-metagenome’ sequencing, and metagenomic sequencing followed by differential coverage binning. These two main approaches both aimed at obtaining genome sequences of bacterial symbionts of A. aerophoba, that were then compared to each other and to references from other environments, to gain information on adaptations to the host sponge environment and on possible interactions with the host and within the microbial community.
Cyanobacteria are frequent members of the sponge microbial community. My ‘mini-metagenome’ sequencing project delivered three draft genomes of “Candidatus Synechococcus spongiarum,” the cyanobacterial symbiont of A. aerophoba and many more sponges inhabiting the photic zone. The most complete of these genomes was compared to other clades of this symbiont and to closely related free-living cyanobacterial references in a collaborative project published in Burgsdorf I*, Slaby BM* et al. (2015; *shared first authorship). Although the four clades of “Ca. Synechococcus spongiarum” from the four sponge species A. aerophoba, Ircinia variabilis, Theonella swinhoei, and Carteriospongia foliascens were approximately 99% identical on the level of 16S rRNA gene sequences, they greatly differed on the genomic level. Not only the genome sizes were different from clade to clade, but also the gene content and a number of features including proteins containing the eukaryotic-type domains leucine-rich repeats or tetratricopeptide repeats. On the other hand, the four clades shared a number of features such as ankyrin repeat domain-containing proteins that seemed to be conserved also among other microbial phyla in different sponge hosts and from different geographic locations. A possible novel mechanism for host phagocytosis evasion and phage resistance by means of an altered O antigen of the lipopolysaccharide was identified.
To test previous hypotheses on adaptations of sponge-associated bacteria on a broader spectrum of the microbiome of A. aerophoba while also taking a step forward in methodology, I developed a bioinformatic pipeline to combine metagenomic Illumina short-read sequencing data with PacBio long-read data. At the beginning of this project, no pipelines to combine short-read and long-read data for metagenomics were published, and at time of writing, there are still no projects published with a comparable aim of un-targeted assembly, binning and analysis of a metagenome. I tried a variety of assembly programs and settings on a simulated test dataset reflecting the properties of the real metagenomic data. The developed assembly pipeline improved not only the overall assembly statistics, but also the quality of the binned genomes, which was evaluated by comparison to the originally published genome assemblies.
The microbiome of A. aerophoba was studied from various angles in the recent years, but only genomes of the candidate phylum Poribacteria and the cyanobacterial sequences from my above-described project have been published to date. By applying my newly developed assembly pipeline to a metagenomic dataset of A. aerophoba consisting of a PacBio long-read dataset and six Illumina short-read datasets optimized for subsequent differential coverage binning, I aimed at sequencing a larger number and greater diversity of symbionts. The results of this project are currently in review by The ISME Journal. The complementation of Illumina short-read with PacBio long-read sequencing data for binning of this highly complex metagenome greatly improved the overall assembly statistics and improved the quality of the binned genomes. Thirty-seven genomes from 13 bacterial phyla and candidate phyla were binned representing the most prominent members of the microbiome of A. aerophoba. A statistical comparison revealed an enrichment of genes involved in restriction modification and toxin-antitoxin systems in most symbiont genomes over selected reference genomes. Both are defense features against incoming foreign DNA, which may be important for sponge symbionts due to the sponge’s filtration and phagocytosis activity that exposes the symbionts to high levels of free DNA. Also host colonization and matrix utilization features were significantly enriched. Due to the diversity of the binned symbiont genomes, a within-symbionts genome comparison was possible, that revealed three guilds of symbionts characterized by i) nutritional specialization on the metabolization of carnitine, ii) specialization on sulfated polysaccharides, and iii) apparent nutritional generalism. Both carnitine and sulfated polysaccharides are abundant in the sponge extracellular matrix and therefore available to the sponge symbionts as substrates. In summary, the genomes of the diverse community of symbionts in A. aerophoba were united in their defense features, but specialized regarding their nutritional preferences.
Development and application of computational tools for RNA-Seq based transcriptome annotations
(2019)
In order to understand the regulation of gene expression in organisms, precise genome annotation is essential. In recent years, RNA-Seq has become a potent method for generating and improving genome annotations. However, this Approach is time consuming and often inconsistently performed when done manually. In particular, the discovery of non-coding RNAs benefits strongly from the application of RNA-Seq data but requires significant amounts of expert knowledge and is labor-intensive. As a part of my doctoral study, I developed a modular tool called ANNOgesic that can detect numerous transcribed genomic features, including non-coding RNAs, based on RNA-Seq data in a precise and automatic fashion with a focus on bacterial and achaeal species. The software performs numerous analyses and generates several visualizations. It can generate annotations of high-Resolution that are hard to produce using traditional annotation tools that are based only on genome sequences. ANNOgesic can detect numerous novel genomic Features like UTR-derived small non-coding RNAs for which no other tool has been developed before. ANNOgesic is available under an open source license (ISCL) at https://github.com/Sung-Huan/ANNOgesic.
My doctoral work not only includes the development of ANNOgesic but also its application to annotate the transcriptome of Staphylococcus aureus HG003 - a strain which has been a insightful model in infection biology. Despite its potential as a model, a complete genome sequence and annotations have been lacking for HG003. In order to fill this gap, the annotations of this strain, including sRNAs and their functions, were generated using ANNOgesic by analyzing differential RNA-Seq data from 14 different samples (two media conditions with seven time points), as well as RNA-Seq data generated after transcript fragmentation. ANNOgesic was
also applied to annotate several bacterial and archaeal genomes, and as part of this its high performance was demonstrated. In summary, ANNOgesic is a powerful computational tool for RNA-Seq based annotations and has been successfully applied to several species.
Antikörper, die gegen eine klinisch relevante Gruppe von Rezeptoren innerhalb der Tumornekrosefaktor-Rezeptor-Superfamilie (TNFRSF) gerichtet sind, darunter CD40 und CD95 (Fas/Apo-1), benötigen ebenfalls eine Bindung an Fc-Gamma-Rezeptoren (FcγRs), um eine starke agonistische Wirkung zu entfalten. Diese FcγR-Abhängigkeit beruht weitgehend auf der bloßen zellulären Verankerung durch die Fc-Domäne des Antikörpers und benötigt dabei kein FcγR-Signalling. Ziel dieser Doktorarbeit war es, das agonistische Potenzial von αCD40- und αCD95-Antikörpern unabhängig von der Bindung an FcγRs durch die Verankerung an Myelomzellen zu entfalten. Zu diesem Zweck wurden verschiedene Antikörpervarianten (IgG1, IgG1-N297A, Fab2) gegen die TNFRSF-Mitglieder CD40 und CD95 genetisch mit einem einzelkettig kodierten B-Zell-aktivierenden Faktor (scBaff) Trimer als C-terminale myelom-spezifische Verankerungsdomäne fusioniert, welche die Fc-Domäne-vermittelte FcγR-Bindung ersetzt. Diese bispezifischen Antikörper-scBaff-Fusionsproteine wurden in Bindungsstudien und funktionellen Assays mit Tumorzelllinien untersucht, die einen oder mehrere der drei Baff-Rezeptoren exprimieren: BaffR, Transmembran-Aktivator und CAML-Interaktor (TACI) und B-Zell-Reifungsantigen (BCMA). Zelluläre Bindungsstudien zeigten, dass die Bindungseigenschaften der verschiedenen Domänen innerhalb der Antikörper-scBaff-Fusionen gegenüber der Zielantigene vollständig intakt blieben. In Ko-Kulturversuchen von CD40- und CD95-responsiven Zellen mit BaffR-, BCMA- oder TACI-exprimierenden Verankerungszellen zeigten die Antikörper-Fusionsproteine einen starken Agonismus, während in Ko-Kulturen mit Zellen ohne Expression von Baff-interagierenden Rezeptoren nur eine geringe Rezeptorstimulation beobachtet wurde. Die hier vorgestellten αCD40- und αCD95-Antikörper-scBaff-Fusionsproteine zeigen also Myelom-spezifische Aktivität und versprechen im Vergleich zu herkömmlichen CD40- und CD95-Agonisten geringere systemische Nebenwirkungen.
Neurobiology is widely supported by bioinformatics. Due to the big amount of data generated from the biological side a computational approach is required. This thesis presents four different cases of bioinformatic tools applied to the service of Neurobiology.
The first two tools presented belong to the field of image processing. In the first case, we make use of an algorithm based on the wavelet transformation to assess calcium activity events in cultured neurons. We designed an open source tool to assist neurobiology researchers in the analysis of calcium imaging videos. Such analysis is usually done manually which is time consuming and highly subjective. Our tool speeds up the work and offers the possibility of an unbiased detection of the calcium events. Even more important is that our algorithm not only detects the neuron spiking activity but also local spontaneous activity which is normally discarded because it is considered irrelevant. We showed that this activity is determinant in the calcium dynamics in neurons and it is involved in important functions like signal modulation and memory and learning.
The second project is a segmentation task. In our case we are interested in segmenting the neuron nuclei in electron microscopy images of c.elegans. Marking these structures is necessary in order to reconstruct the connectome of the organism. C.elegans is a great study case due to the simplicity of its nervous system (only 502 neurons). This worm, despite its simplicity has taught us a lot about neuronal mechanisms. There is still a lot of information we can extract from the c.elegans, therein lies the importance of reconstructing its connectome. There is a current version of the c.elegans connectome but it was done by hand and on a single subject which leaves a big room for errors. By automatizing the segmentation of the electron microscopy images we guarantee an unbiased approach and we will be able to verify the connectome on several subjects.
For the third project we moved from image processing applications to biological modeling. Because of the high complexity of even small biological systems it is necessary to analyze them with the help of computational tools. The term in silico was coined to refer to such computational models of biological systems. We designed an in silico model of the TNF (Tumor necrosis factor) ligand and its two principal receptors. This biological system is of high relevance because it is involved in the inflammation process. Inflammation is of most importance as protection mechanism but it can also lead to complicated diseases (e.g. cancer). Chronic inflammation processes can be particularly dangerous in the brain. In order to better understand the dynamics that govern the TNF system we created a model using the BioNetGen language. This is a rule based language that allows one to simulate systems where multiple agents are governed by a single rule. Using our model we characterized the TNF system and hypothesized about the relation of the ligand with each of the two receptors. Our hypotheses can be later used to define drug targets in the system or possible treatments for chronic inflammation or lack of the inflammatory response.
The final project deals with the protein folding problem. In our organism proteins are folded all the time, because only in their folded conformation are proteins capable of doing their job (with some very few exceptions). This folding process presents a great challenge for science because it has been shown to be an NP problem. NP means non deterministic Polynomial time problem. This basically means that this kind of problems cannot be efficiently solved. Nevertheless, somehow the body is capable of folding a protein in just milliseconds. This phenomenon puzzles not only biologists but also mathematicians. In mathematics NP problems have been studied for a long time and it is known that given the solution to one NP problem we could solve many of them (i.e. NP-complete problems). If we manage to understand how nature solves the protein folding problem then we might be able to apply this solution to many other problems. Our research intends to contribute to this discussion. Unfortunately, not to explain how nature solves the protein folding problem, but to explain that it does not solve the problem at all. This seems contradictory since I just mentioned that the body folds proteins all the time, but our hypothesis is that the organisms have learned to solve a simplified version of the NP problem. Nature does not solve the protein folding problem in its full complexity. It simply solves a small instance of the problem. An instance which is as simple as a convex optimization problem. We formulate the protein folding problem as an optimization problem to illustrate our claim and present some toy examples to illustrate the formulation. If our hypothesis is true, it means that protein folding is a simple problem. So we just need to understand and model the conditions of the vicinity inside the cell at the moment the folding process occurs. Once we understand this starting conformation and its influence in the folding process we will be able to design treatments for amyloid diseases such as Alzheimer's and Parkinson's.
In summary this thesis project contributes to the neurobiology research field from four different fronts. Two are practical contributions with immediate benefits, such as the calcium imaging video analysis tool and the TNF in silico model. The neuron nuclei segmentation is a contribution for the near future. A step towards the full annotation of the c.elegans connectome and later for the reconstruction of the connectome of other species. And finally, the protein folding project is a first impulse to change the way we conceive the protein folding process in nature. We try to point future research in a novel direction, where the amino code is not the most relevant characteristic of the process but the conditions within the cell.
Kritische Knochendefekte stellen heutzutage ein ungelöstes Problem in der klinischen Praxis dar, da die verfügbaren prothetischen Optionen oft die mechanische Anpassung an das Gewebe nicht gewährleisten oder zu wichtigen immunologischen und Implantat-bedingten Komplikationen führen.
In diesem Kontext ermöglichen Tissue Engineering-Ansätze neue Strategien, um in vitro Zell-Material Interaktionen zu untersuchen und so die Implantatmaterialien zu optimieren.
In dieser Arbeit habe ich Zell-Material Interaktionen eines neuen Kollagen-basierten Scaffolds untersucht, das langfristig als Trägerstruktur für eine zellbasierte Therapie für kritische Knochendefekte entwickelt werden soll. Im Rahmen der Dissertation konnte ich belegen, dass die Kollagen-basierten makroporöse Mikrocarrier für die Zellvermehrung humaner mesenchymaler Stammzellen (MSC) und deren osteogene Differenzierung unter GMP Bedingungen verwendet werden können. Außerdem habe ich die die Kokultur von hämatopoietischen Stammzellen des Knochenmarks und multiplen Myelomzellen funktionell charakterisiert. Ich konnte erstmals Kulturbedingungen etablieren, die die Langzeitkultur ohne die Verwendung von Zytokinen ermöglicht. Mittels dieser Kokultur konnte ich ein Knochenmarknischen-Modell etablieren und die Untersuchung der Expression von zentralen Signalkaskaden der Homöostase dieser Nische untersuchen. Ich konnte die Expression von zwei verschiedenen Isoformen von Osteopontin nachweisen, die in Tiermodellen nicht gefunden werden. Diese Isoformen des Osteopontins habe ich kloniert und die rekombinanten Isoformen exprimiert und ihre Rollen in der Homöostase der Knochenmarknische untersucht.
Critical size bone defects represent nowadays an unresolved problem in the clinical practice, where the available prosthetic options often lack adequate mechanical matching to the host tissue or lead to important immunological and implant-related complications.
In this context, Tissue Engineering approaches promise more effective strategies to study cell-material interactions in vitro and consequently optimize implant materials.
In this work, I investigated the cell-scaffold interactions of a new collagen-based scaffold for a putative cell-based therapy for critical size defects to be developed. In the context of this thesis, I could demonstrate that the collagen-based macroporous microcarriers could be employed for the expansion and osteogenic differentiation of human mesenchymal stromal cells (MSCs) under GMP-compliant conditions. Moreover, I functionally characterized the co-culture of bone marrow hematopoietic stem cells and multiple myeloma cells. I was for the first time able to establish culture conditions allowing their long-term culture in absence of externally supplemented cytokines. Using this co-culture, I was able to establish a bone marrow niche model to investigate the expression of key signaling pathways involved in the niche´s homeostasis. I was able to demonstrate the expression of two different isoforms of Osteopontin, that could not previously be detected in animal models. Finally, I cloned these Osteopontin isoforms, expressed recombinant versions of the isoforms, and investigated their roles in the homeostasis of the bone marrow niche.
Background: Integrase strand transfer inhibitors (INSTIs) are the latest addition to the array of antiretroviral compounds used to treat an infection with Human Immunodeficiency Virus (HIV). Due to their high efficacy and increased tolerability, INSTIs have become an integral part of first-line therapy in most high-income countries over the past years. However, little is known about HIV-1’s genetic inter- and intra-subtype diversity on the Integrase (IN)-gene and its impact on the emergence of INSTI-resistance. In the absence of a functional cure, long-term efficacy of first-line compounds remains paramount for reducing virological failure and curbing on-going HIV transmissions. South Africa, harbouring more than 20% of the global HIV burden (7.7 / 37.9 million people), requires international attention in order to globally pursue UNAIDS’ (Joint United Nations Programme on HIV/AIDS) 90-90-90 goals and the road to ending the HIV/AIDS (Acquired immunodeficiency syndrome) pandemic by 2030.
Methods: In this study, the prevalence of INSTI-resistance associated mutations (RAM) was investigated in a cohort of 169 archived drug-naïve blood samples from multiple collection sites around Cape Town, South Africa. Viral RNA was isolated from plasma samples, the integrase fragment amplified by RT-PCR and subsequently sequenced by Sanger-sequencing. Additionally, all publicly available drug-naïve, South African IN sequences, isolated before the availability of the first INSTIs in 2007, were retrieved from the Los Alamos HIV sequence database (n=284). All sequences were analysed for RAMs using the Stanford HIV Drug resistance database. The identification of polymorphism in the South African subtype C IN consensus sequence allowed for comparative analyses with global subtype B, as well as subtype C sequences, from countries other than South Africa.
Results: The IN gene could be amplified and sequenced in 95/169 samples (56%). Phylogenetic inference revealed close homology between three sequence-pairs, warranting the exclusion of 3/95 sequences from further analyses. Of the 92 samples used for mutational analyses, 86/92 (93.5%) belonged to subtype C, 5/92 (5.4%) to subtype B and 1/92 (1.1%) to subtype A. The prevalence of major and accessory INSTI RAMs was 0/92 (0%) and 1/91 (1.1%), respectively, similar to the observed rates of 8/284 (2.8%) and 8/284 (2.8%) in the database sequences (p = 0.2076 and p = 0.6944, Fisher’s exact test). Compared to subtype B IN sequences, 15 polymorphisms were significantly enriched in South African subtype C sequences (corrected p<0.0015. Fisher’s exact test, Bonferroni post-hoc procedure).
Compared to subtype C IN sequences isolated outside South Africa, four polymorphisms were significantly enriched in this study cohort (corrected p<0.0014, Fisher’s exact test, Bonferroni post-hoc procedure). The highest prevalence margin was observed for the polymorphism Met50Ile being present in 60.1% of South African subtype C sequences, compared to 37% in non-South African subtype C sequences.
Conclusions: The low prevalence of major and minor RAMs in all South African Integrase sequences predicts a high susceptibility to INSTIs, however, the presence of natural polymorphisms, in particular Met50Ile, in the majority of sequences warrants further monitoring under therapeutic pressure, as their role in mutational pathways leading to INSTI- resistance is yet to be determined. Additionally, this study revealed the presence of substantial inter- and intra-subtype diversity within the HIV-1 Subtype C IN-gene. These results implicate the need for more research on a regional, potentially patient-specific level, as mutational insights from other diverse backgrounds may not accurately represent the South African context. The implementation of a national pre-treatment INSTI-resistance screening program may provide necessary insights into the development of mutational pathways leading to INSTI-resistance under therapeutic pressure for the South African context and thereby bring South Africa one step closer to achieving UNAIDS 90-90-90 goals and ending the AIDS epidemic by 2030.
Pertussis is a highly contagious acute respiratory disease of humans which is mainly caused by the gram-negative obligate human pathogen Bordetella pertussis. Despite the availability and extensive use of vaccines, the disease persists and has shown periodic re-emergence resulting in an estimated 640,000 deaths worldwide in 2014. The pathogen expresses various virulence factors that enable it to modulate the host immune response, allowing it to colonise the ciliated airway mucosa. Many of these factors also directly interfere with host signal transduction systems, causing damage to the ciliated airway mucosa and increase mucous production. Of the many virulence factors of B. pertussis, only the tracheal cytotoxin (TCT) is able to recapitulate the pathophysiology of ciliated cell extrusion and blebbing in animal models and in human nasal biopsies. Furthermore, due to the lack of appropriate human models and donor materials, the role of bacterial virulence factors has been extrapolated from studies using animal models infected with either B. pertussis or with the closely related species B. bronchiseptica which naturally causes respiratory infections in these animals and produces many similar virulence factors. Thus, in the present work, in vitro airway mucosa models developed by co-culturing human airway epithelia cells and fibroblasts from the conduction zone of the respiratory tract on a decellularized porcine small intestine submucosa scaffold (SISser®) were used, since these models have a high correlation to native human conducting zone respiratory epithelia. The major aim was to use the engineered airway mucosa models to elucidate the contribution of B. pertussis TCT in the pathophysiology of the disease as well as the virulence mechanism of B. pertussis in general. TCT and lipopolysaccharide (LPS) either alone or in combination were observed to induce epithelial cell blebbing and necrosis in the in vitro airway mucosa model. Additionally, the toxins induced viscous hyper-mucous secretion and significantly disrupted barrier properties of the in vitro airway mucosa models. This work also sought to assess the invasion and intracellular survival of B. pertussis in the polarised epithelia, which has been critically discussed for many years in the literature. Infection of the models with B. pertussis showed that the bacteria can adhere to the models and invade the epithelial cells as early as 6 hours post inoculation. Invasion and intracellular survival assays indicated the bacteria could invade and persist intracellularly in the epithelial cells for up to 3 days. Due to the novelty of the in vitro airway mucosa models, this work also intended to establish a method for isolating individual cells for scRNA-seq after infection with B. pertussis. Cold dissociation with Bacillus licheniformis subtilisin A was found to be capable of dissociating the cells without inducing a strong fragmentation, a problem which occurs when collagenase and trypsin/EDTA are used. In summary, the present work showed that TCT acts possibly in conjunction with LPS to disrupt the human airway mucosa much like previously shown in the hamster tracheal ring models and thus appears to play an important role during the natural B. pertussis infection. Furthermore, we established a method for infecting and isolating infected cells from the airway mucosa models in order to further investigate the effect of B. pertussis infection on the different cell populations in the airway by single cell analytics in the future.
Additive manufacturing processes such as 3D printing are booming in the industry due to their high degree of freedom in terms of geometric shapes and available materials. Focusing on patient-specific medicine, 3D printing has also proven useful in the Life Sciences, where it exploits the shape fidelity for individualized tissues in the field of bioprinting. In parallel, the current systems of bioreactor technology have adapted to the new manufacturing technology as well and 3D-printed bioreactors are increasingly being developed. For the first time, this work combines the manufacturing of the tissue and a tailored bioreactor, significantly streamlining the overall process and optimally merging the two processes. This way the production of the tissues can be individualized by customizing the reactor to the tissue and the patient-specific wound geometry. For this reason, a common basis and guideline for the cross-device and cross-material use of 3D printers was created initially. Their applicability was demonstrated by the iterative development of a perfusable bioreactor system, made from polydimethylsiloxane (PDMS) and a lignin-based filament, into which a biological tissue of flexible shape can be bioprinted. Cost-effective bioink-replacements and in silico computational fluid dynamics simulations were used for material sustainability and shape development. Also, nutrient distribution and shear stress could be predicted in this way pre-experimentally.
As a proof of functionality and adaptability of the reactor, tissues made from a nanocellulose-based Cellink® Bioink, as well as an alginate-based ink mixed with Me-PMeOx100-b-PnPrOzi100-EIP (POx) (Alginate-POx bioink) were successfully cultured dynamically in the bioreactor together with C2C12 cell line. Tissue maturation was further demonstrated using hMSC which were successfully induced to adipocyte differentiation. For further standardization, a mobile electrical device for automated media exchange was developed, improving handling in the laboratory and thus reduces the probability of contamination.
The platelet cytoskeleton ensures normal size and discoid shape under resting conditions and undergoes immediate reorganization in response to changes in the extracellular environment through integrin-based adhesion sites, resulting in actomyosin-mediated contractile forces. Mutations in the contractile protein non-muscle myosin heavy chain IIA display, among others, macrothrombocytopenia and a mild to moderate bleeding tendency in human patients. It is insufficiently understood which factors contribute to the hemostatic defect found in MYH9-related disease patients. Therefore, a better understanding of the underlying biophysical mechanisms in thrombus formation and stabilization is warranted.
This thesis demonstrates that an amino acid exchange at the positions 702, 1424 and 1841 in the heavy chain of the contractile protein non-muscle myosin IIA, caused by heterozygous point mutations in the gene, resulted in macrothrombocytopenia and increased bleeding in mice, reflecting the clinical hallmark of the MYH9-related disease in human patients. Basic characterization of biological functions of Myh9 mutant platelets revealed overall normal surface glycoprotein expression and agonist-induced activation when compared to wildtype platelets. However, myosin light chain phosphorylation after thrombin-activation was reduced in mutant platelets, resulting in less contractile forces and a defect in clot retraction. Altered biophysical characteristics with lower adhesion and interaction forces of Myh9 mutant platelets led to reduced thrombus formation and stability. Platelets from patients with the respective mutations recapitulated the findings obtained with murine platelets, such as impaired thrombus formation and stiffness.
Besides biological and biophysical characterization of mutant platelets from mice and men, treatment options were investigated to prevent increased bleeding caused by reduced platelet forces. The antifibrinolytic agent tranexamic acid was applied to stabilize less compact thrombi, which are presumably more vulnerable to fibrinolysis. The hemostatic function in Myh9 mutant mice was improved by interfering with the fibrinolytic system. These results show the beneficial effect of fibrin stabilization to reduce bleeding in MYH9-related disease.
Studies on the role of cytoskeletal-regulatory and -crosslinking proteins in platelet function
(2023)
Cytoskeletal reorganization in platelets is highly regulated and important for proper platelet function during activation and aggregation at sites of vascular injury. In this thesis, the role of three different cytoskeletal-regulatory and -crosslinking proteins was studied in platelet physiology using megakaryocyte- and platelet-specific knockout mice. The generation of branched actin filaments is regulated by nucleation promoting factors (NPF) and the Arp2/3 complex.
(1.) The WAVE complex is a NPF, which upregulates the Arp2/3 complex activity at the plasma membrane. As shown in this thesis, the loss of the WAVE complex subunit Cyfip1 in mice did not alter platelet production and had only a minor impact on platelet activation. However, Cyfip1 played an essential role for branching of actin filaments and consequently for lamellipodia formation in vitro. The importance of lamellipodia for thrombus formation and stability has been controversially discussed. Cyfip1-deficient platelets were able to form a stable thrombus ex vivo and in vivo and a hemostatic plug comparable to controls. Moreover, Cyfip1-deficient mice maintained vascular integrity at the site of inflammation. These data show that platelet lamellipodia formation is not required for hemostatic function and pathophysiological thrombus formation.
(2.) The WASH complex is another NPF, which mediates actin filament polymerization on endosomal vesicles via the Arp2/3 complex. Loss of the WASH complex subunit Strumpellin led to a decreased protein abundance of the WASH protein and to a 20% reduction in integrin αIIbβ3 surface expression on platelets and megakaryocytes, whereas the expression of other surface receptors as well as the platelet count, size, ex vivo thrombus formation and bleeding time remained unaltered. These data point to a distinct role of Strumpellin in maintaining integrin αIIbβ3 expression and provide new insights into regulatory mechanisms of platelet integrins.
(3.) MACF1 has been described as a cytoskeletal crosslinker of microtubules and F-actin. However, MACF1-deficient mice displayed no alterations in platelet production, activation, thrombus formation and hemostatic function. Further, no compensatory up- or downregulation of other proteins could be found that contain an F-actin- and a microtubule-binding domain. These data indicate that MACF1 is dispensable for platelet biogenesis, activation and thrombus formation. Nevertheless, functional redundancy among different proteins mediating the cytoskeletal crosstalk may exist.
Adrenocortical carcinoma (ACC) is a rare, but highly aggressive endocrine malignancy. Tumor-related hypercortisolism is present in 60 % of patients and associated with worse outcome. While cancer immunotherapies have revolutionized the treatment of many cancer entities, the results of initial studies of different immune checkpoint inhibitors in ACC were heterogeneous. Up to now, five small clinical trials with a total of 121 patients have been published and demonstrated an objective response in only 17 patients. However, one of the studies, by Raj et al., reported a clinically meaningful disease control rate of 52 % and a median overall survival of almost 25 months suggesting that a subgroup of ACC patients may benefit from immunotherapeutic approaches. Following the hypothesis that some ACCs are characterized by a glucocorticoid-induced T lymphocytes depletion, several studies were performed as part of the presented thesis. First, the immune cell infiltration in a large cohort of 146 ACC specimens was investigated. It was demonstrated for the first time, and against the common assumption, that ACCs were infiltrated not only by FoxP3+ regulatory T cells (49.3 %), but also that a vast majority of tumor samples was infiltrated by CD4+ TH cells (74 %) and CD8+ cytotoxic T cells (84.3 %), albeit the immune cell number varied heterogeneously and was rather low (median: 7.7 CD3+ T cells / high power field, range: 0.1-376). Moreover, the presence of CD3+-, CD4+- and CD8+ ACC-infiltrating lymphocytes was associated with an improved recurrence-free (HR: 0.31 95 % CI 0.11-0.82) and overall survival (HR: 0.47 96 % CI 0.25-0.87). Particularly, patients with tumor-infiltrating CD4+ TH cells without glucocorticoid excess had a significantly longer overall survival compared to patients with T cell-depleted ACC and hypercortisolism (121 vs. 27 months, p = 0.004). Hence, the impact of glucocorticoids might to some extent be responsible for the modest immunogenicity in ACC as hypercortisolism was reversely correlated with the number of CD4+ TH cells. Accordingly, CD3+ T cells co-cultured with steroidogenic NCI-H295R ACC cells demonstrated in vitro an enhanced anti-tumoral cytotoxicity by secreting 747.96 ±225.53 pg/ml IFN-γ in a therapeutically hormone-depleted microenvironment (by incubation with metyrapone), versus only 276.02 ±117.46 pg/ml IFN-γ in a standard environment with glucocorticoid excess.
Other potential biomarkers to predict response to immunotherapies are the immunomodulatory checkpoint molecules, programmed cell death 1 (PD-1) and its ligand PD-L1, since both are targets of antibodies used therapeutically in different cancer entities. In a subcohort of 129 ACCs, expressions of both molecules were heterogeneous (PD-1 17.4 %, range 1-15; PD-L1 24.4 %, range 1 - 90) and rather low. Interestingly, PD-1 expression significantly influenced ACC patients´ overall (HR: 0.21 95 % CI 0.53-0.84) and progression- free survival (HR: 0.30 95 % CI 0.13-0.72) independently of established factors, like ENSAT tumor stage, resection status, Ki67 proliferation index and glucocorticoid excess, while PD-L1 had no impact.
In conclusion, this study provides several potential explanations for the heterogeneous results of the immune checkpoint therapy in advanced ACC. In addition, the establishment of PD-1 as prognostic marker can be easily applied in routine clinical care, because it is nowadays anyway part of a detailed histo-pathological work-up. Furthermore, these results provide the rationale and will pave the way towards a combination therapy using immune checkpoint inhibitors as well as glucocorticoid blockers. This will increase the likelihood of re-activating the immunological anti-tumor potential in ACC. However, this will have to be demonstrated by additional preclinical in vivo experiments and finally in clinical trials with patients.
Introduction.
Mobile health (mHealth) integrates mobile devices into healthcare, enabling remote monitoring, data collection, and personalized interventions. Machine Learning (ML), a subfield of Artificial Intelligence (AI), can use mHealth data to confirm or extend domain knowledge by finding associations within the data, i.e., with the goal of improving healthcare decisions. In this work, two data collection techniques were used for mHealth data fed into ML systems: Mobile Crowdsensing (MCS), which is a collaborative data gathering approach, and Ecological Momentary Assessments (EMA), which capture real-time individual experiences within the individual’s common environments using questionnaires and sensors. We collected EMA and MCS data on tinnitus and COVID-19. About 15 % of the world’s population suffers from tinnitus.
Materials & Methods.
This thesis investigates the challenges of ML systems when using MCS and EMA data. It asks: How can ML confirm or broad domain knowledge? Domain knowledge refers to expertise and understanding in a specific field, gained through experience and education. Are ML systems always superior to simple heuristics and if yes, how can one reach explainable AI (XAI) in the presence of mHealth data? An XAI method enables a human to understand why a model makes certain predictions. Finally, which guidelines can be beneficial for the use of ML within the mHealth domain? In tinnitus research, ML discerns gender, temperature, and season-related variations among patients. In the realm of COVID-19, we collaboratively designed a COVID-19 check app for public education, incorporating EMA data to offer informative feedback on COVID-19-related matters. This thesis uses seven EMA datasets with more than 250,000 assessments. Our analyses revealed a set of challenges: App user over-representation, time gaps, identity ambiguity, and operating system specific rounding errors, among others. Our systematic review of 450 medical studies assessed prior utilization of XAI methods.
Results.
ML models predict gender and tinnitus perception, validating gender-linked tinnitus disparities. Using season and temperature to predict tinnitus shows the association of these variables with tinnitus. Multiple assessments of one app user can constitute a group. Neglecting these groups in data sets leads to model overfitting. In select instances, heuristics outperform ML models, highlighting the need for domain expert consultation to unveil hidden groups or find simple heuristics.
Conclusion.
This thesis suggests guidelines for mHealth related data analyses and improves estimates for ML performance. Close communication with medical domain experts to identify latent user subsets and incremental benefits of ML is essential.