No description available.

No description available.

No description available.

Wnt signaling plays important role in many aspects of embryogenesis such as cell proliferation, cell fate specification, cell polarity and organogenesis(Clevers 2006, van Amerongen and Nusse 2009). Wnt ligands have been shown to activate several intra-cellular signaling cascades, including the canonical or Wnt/-catenin dependent pathway and the non-canonical or -catenin independent pathway. Dishevelled (Dvl) occupies a key position at crossroads of all branches of Wnt signaling cascade. To understand, how Dishevelled (Dvl) may channel signaling into the downstream branches, we sought to identify novel effectors for Dishevelled (Dvl) using a yeast-two hybrid screen. In this study, we used the PDZ domain of Dishevelled (Dvl) as a bait and from this screen, we identified a new binding protein of Dishevelled (Dvl)-termed as Custos. To characterize the functional role of Custos in Wnt signaling pathway, we used mammalian cell culture and zebrafish as a model vertebrate organism. We confirmed the interaction between Custos and Dvl using co-immunoprecipation and GST pull-down. Custos also interacted with -catenin in vivo and this interaction was positively regulated by Wnt stimulation. Immunofluorescence experiments in mammalian cells showed that Custos co-localizes with the nuclear envelope marker, lamin and inhibits translocation of -catenin to the nucleus. In zebrafish embryos, Custos is a maternal gene and expressed throughout development. Spatial in situ hybridization studies showed that Custos was expressed in the dorsal region of the embryo at early stages and in the nervous system in zebrafish at 24hpf. To delineate the biological role of Custos during embryogenesis, we conducted a gain of function and loss of function studies. Overexpression of exogenous Custos and morpholino knockdown of Custos revealed that Custos is critical for embryonic patterning. Knockout of Custos in zebrafish revealed that Custos delays embryonic development and exhibits defects in pigmentation suggesting a plausible role in neural crest development. Taken together, our studies demonstrate that Custos is a novel component of canonical Wnt signaling and required for -catenin translocation into the nucleus and important for embryonic patterning. / Biology

京都大学 / 新制・課程博士 / 博士(農学) / 甲第23521号 / 農博第2468号 / 新制||農||1087(附属図書館) / 学位論文||R3||N5352(農学部図書室) / 京都大学大学院農学研究科地域環境科学専攻 / (主査)教授 本田 与一, 教授 田中 千尋, 准教授 坂本 正弘 / 学位規則第4条第1項該当 / Doctor of Agricultural Science / Kyoto University / DFAM

The pig has similar features to the human in aspects such as physiology, immunology, and organ size. Because of these similarities, genetically modified pigs have been generated for xenotransplantation. Also, when using the pig as a model for human diseases (e.g. cystic fibrosis transmembrane conductance regulator), the pig exhibited similar symptoms to those that human patients present. The main goal of this work was to examine the efficacy of direct injection of the CRISPR/Cas9 system (clustered regularly interspaced short palindromic repeats/ CRISPR associated protein 9) in pigs and to overcome shortcomings that resulted after direct injection into the cytoplasm of developing zygotes. By using direct injection of CRISPR/Cas9 into developing zygotes, we successfully generated fetuses and piglets containing 9 different mutations. The total number of aborted fetuses was 20 and of live piglets was 55. Moreover, one issue that was encountered during the production of mutated pigs was that insertion or deletion (indel) mutations did not always introduce a premature stop codon because it did not interfere with the codon read. As a result of these triplet indel(s) mutations, a hypomorphic phenotype was presented; consequently, the mutated gene was partially functional. To prevent this hypomorphic phenotype, we introduced two sgRNAs to generate an intended deletion that would remove a DNA fragment on the genome by causing two double-strand breaks (DSB) during non-homologous end joining (NHEJ). The injection of two sgRNAs successfully generated the intended deletion on the targeted genes in embryos and live piglets. Results after using intended deletions, in IL2RG mutation pigs, did not show hypomorphic phenotypes even when a premature stop codon was not present. After using the intended deletion approach, function of the targeted genes was completely disrupted regardless of the presence or absence of a premature stop codon. Our next aim was to introduce (i.e. knock-in) a portion of exogenous (donor) DNA sequence into a specific locus by utilizing the homology direct repair (HDR) pathway. Because of the cytotoxicity of the linear form of the donor DNA, the concentration of the injected donor DNA was adjusted. After concentration optimization, four different donor DNA fragments targeting four different genes were injected into zygotes. Efficiency of knock-in was an average of 35%. Another donor DNA was used in this study which is IL2RG-IA donor DNA carried 3kb of exogenous cassette. It showed 15.6% of knock-in efficiency. IL2RG-IA Donor DNA injected embryos were transferred into surrogates, and a total of 7 pigs were born from one surrogate, but none of the 7 were positive for the knock-in. Future experiments need to be developed to optimize this approach. Overall, the direct injection of CRISPR/Cas9 is advantageous in cost, time, and efficiency for large animal production and for biomedical research. However, there are still unsolved challenges (off-targeting effects, low efficiency of knock-in, and monoallelic target mutation) that need to be elucidated for future application in humans and other species. / Doctor of Philosophy / The clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR associated protein 9 (Cas9) system is commonly used to make genetically modified pigs. The CRISPR/Cas9 system can break the DNA on a desired gene region. During the DNA repair process, random DNA base pairs can be inserted or deleted on the broken regions, thus generating a mutation on the desired gene. Scientists have adopted new methods to disrupt genes in many species. One of these new methods is the direct injection of CRISPR/Cas9 into a fertilized oocyte. In our first project, we used direct injection of the CRISPR/Cas9 system into the fertilized one-cell embryo. A total of 55 live pigs and aborted 20 fetuses with specifically disrupted genes were produced for biomedical research model. During these studies, one critical drawback of the direct injection method was encountered. Partial function of the gene was possible. To prevent this problem, two DNA broken regions were generated by the CRISPR/Cas9 system to remove the middle part the DNA by two DNA breaking. This method successfully removed the middle portion of the DNA targeted region in the pig embryos. Embryos injected with the CRISPR/Cas9 system to cut the two specific DNA regions were transplanted into surrogate pigs, and a total of 15 piglets were produced. All 15 pigs confirmed that a specific part of the gene had been removed by two DNA breakage. Also, no function of the desired gene was found in the 15 pigs. The objective of the last experiment was to introduce a specific exogenous DNA sequences into specific region of DNA using the CRISPR/Cas9 system. For this study, four different exogenous DNA fragments were synthesized for four different genes. When injected, one exogenous DNA along with the CRISPR/Cas9 system, the average integration efficiency of the four exogenous DNA fragments was 35% in the embryo. Another exogenous DNA, which was longer than other four DNA fragments showed 15.6% integration efficiency. The embryos injected with the long exogenous DNA fragment, along with the CRISPR/Cas9 system, were transferred into surrogate pigs. The result was that a total of 7 piglets were born, but the exogenous DNA sequence was not found in none of the seven piglets. In conclusion, the CRISPR/Cas9 system showed effective removal of the entire gene function of specific genes in the pig. However, for future application in the human and other species, some problems (un-wanted region mutation and low efficiency of exogenous DNA integration) continue to emerge and need to be addressed in future experiments.

Oomycetes are fungi-like eukaryotic microorganisms, which are actually phylogenetic relatives of diatoms and brown algae, within the kingdom Stramenopila. Many oomycete species, mainly in the genera Phytophthora, Pythium and downy mildews, are devastating plant pathogens that cause multibillion-dollar losses to agriculture annually in the world. Some oomycetes are also animal pathogens, causing severe losses in aquaculture and fisheries, and occasionally causing dangerous infections of humans. Phytophthora species, represented by the Irish Potato Famine pathogen P. infestans and the soybean pathogen P. sojae, are arguably the most destructive pathogens of dicotyledonous plants among the oomycete species and thus have been extensively studied. This dissertation focuses on the model oomycete pathogen P. sojae to investigate specific aspects of its molecular biology and establish an efficient genetic manipulation tool. Specifically, in Chapter 1, I briefly introduce the basic concepts of oomycete biology and pathology, and summarize the experimental techniques used for studies of oomycete genetics over the past two decades. Because the approach to studying fungi and oomycetes are similar (indeed they were incorrectly placed in the same taxonomic group until recently), a special section reviews the emerging genome editing technology CRISPR/Cas system in these organisms together. Chapter 2 and Chapter 3 focus on one of the most important intracellular activities, nuclear localization of proteins, and describe the characterization of nuclear localization signals (NLSs) in P. sojae. This focus stemmed from my early work on genome editing in P. sojae, when I discovered that conventional NLS signals from SV40 used to target the TAL effector nuclease (TALEN) to the nucleus worked poorly in P. sojae. In the first part of this work (Chapter 2), I used confocal microscopy to identify features of nuclear localization in oomycetes that differ from animals, plants and fungi, based on characterization of two classes of nuclear localization signals, cNLS and PY-NLS, and on characterization of several conserved nuclear proteins. In the second part (Chapter 3), I determined that the nuclear localization of the P. sojae bZIP1 transcription factor is mediated by multiple weak nuclear targeting motifs acting together. In Chapter 4 and Chapter 5, I describe my implementation of nuclease-based technology for genetic modification and control of P. sojae. In Chapter 4, I describe the first use of the CRISPR system in an oomycete, including its use to validate the function of a host specificity gene. This is of particular importance because molecular techniques such as gene knockouts and gene replacements, widely used in other organisms, were not previously possible in oomycetes. The successful implementation of CRISPR provides a major new research capability to the oomycete community. Following up on the studies described in Chapter 4, in Chapter 5, I describe the generalization and simplification of the CRISPR/Cas9 expression strategy in P. sojae as well as methods for mutant screening. I also describe several optimized methodologies for P. sojae manipulation based on my 5 years of experience with P. sojae. / Ph. D.

Les bactériophages sont des maîtres manipulateurs, prenant le contrôle complet d'une cellule bactérienne, contournant les systèmes de défense et détournant la machinerie de transcription et de traduction du génome bactérien pour la réplication virale. Les grandes étapes de la multiplication des phages sont connues, pourtant les mécanismes intrinsèques de cette prise de contrôle restent un mystère bien préservé. Malgré la petite taille et la simplicité relative des génomes de phages, seuls les gènes associés à la structure des virions sont amplement caractérisés. Toutefois, les protéines non structurales qui sont susceptibles d'être responsables de la prise de contrôle sont rarement étudiées. C'est effectivement le cas pour le phage modèle 2972, infectant la souche Streptococcus thermophilus DGCC7710 largement utilisée en industrie laitière. Son génome code pour 44 protéines putatives, dont 14 protéines sont classées comme étant non structurales et dont aucune fonction n'est encore associée. Lors de ce projet de maîtrise, le système CRISPR-Cas de type II-A, naturellement présent chez S. thermophilus, a été utilisé à des fins d'édition du génome de ce phage pour étudier le rôle des protéines non structurales. Ce système est idéal pour les manipulations génétiques des phages virulents généralement complexes à modifier. Ainsi, une connaissance approfondie des interactions entre le phage et son hôte sera un outil indispensable pour développer de meilleures méthodes de contrôle des phages en industrie laitière. / Bacterial viruses are master manipulators of bacterial cells. They are able to take complete control of a bacterium, bypassing bacterial immune systems, hijacking core transcription and translation machinery, and typically resulting in lysis of the host. Although the major steps of phage replication are well understood, very little is known about the mechanisms of the host-cell takeover. Despite phages having relatively small and 'simple' genomes, generally only the structural proteins have been well characterized. In contrast, non-structural proteins, which include those involved in host cell takeover, tend to be completely uncharacterized. This is certainly the case for the model of Streptococcus thermophilus phages, 2972, which infects the strain DGCC7710 widely used by the dairy industry. Its genome encodes for 44 putative proteins, 14 of which are non-structural and have no known function. In this master thesis, the type II-A CRISPR-Cas system naturally present in S. thermophilus was used for genome engineering purposes to investigate the role of non-structural proteins of phage 2972. This natural bacterial immune system provides an ideal means for genetic manipulation of virulent phages, which are otherwise intractable. This could lead to potentially valuable discoveries allowing us to further fine-tune the bacteria used in various biological processes.

Pandemics are periodic events characterized by rapid and widespread transmission of infectious disease affecting a significant proportion of the population over a large geographical area. Zoonotic strains of influenza viruses and coronaviruses have both caused pandemics in the recent past. Although vaccination is the often the most effective way to prevent infection or serious outcomes of infection, vaccine development, production, distribution, and deployment are all time- consuming and logistically challenges processes. Alternative readily deployed approaches must be quickly executed to mitigate the toll of future pandemics, especially during the early phases. The work described in this thesis describes some of these approaches. Firstly, I describe the process by which I performed genome-wide CRISPR-Cas9 knockout screens using SARS-CoV-2 variants of concern to discover crucial host factors as targets for broad-acting antivirals. I found that all variants rely on similar host pathways to replicate in the glial cell line used for the screen. I identified BCL-xL, a regulator of apoptosis, as a potential target for a broad- acting antiviral. I show that chemical inhibition of BCL-xL results in accelerated cell death in infected cells in vitro, but improved clinical signs and disease mortality of SARS-CoV-2 in our murine infection model. Secondly, I describe a unique mechanism for cooperative antiviral combination therapy. I demonstrate that chemical inhibition of neuraminidase by oseltamivir improved immune effector cell activation by hemagglutinin stalk-binding antibodies. Combination therapy of oseltamivir and stalk-binding antibodies also improved clinical signs and disease mortality of influenza in our murine infection model compared to monotherapy in both prophylactic and therapeutic contexts. Finally, I show that non-pharmaceutical public health interventions used to restrict the spread of COVID-19 were also effective against several other infectious diseases. I used an interrupted time- series analysis on Ontario public health administrative data during the early COVID-19 pandemic period and found a drastic and sustained decline in outpatient visits for diseases that are typically caused by viruses that transmit by droplet or aerosol. The three projects described in this thesis outlines broadly-protective and distinct strategies to curb the spread of novel respiratory viruses. These new tools may be leveraged to improve the response and to mitigate the burden of future pandemics. / Thesis / Doctor of Philosophy (PhD) / Most pandemics in recent history have been caused by viruses that infect the respiratory tract. Vaccination is often the best way to prevent the spread of these pandemic viruses, but making these vaccines takes time. Vaccines also work less well in the very young, the elderly, and those with a compromised immune system. These people are often also the most vulnerable to severe disease. My work describes three novel approaches to help combat the next pandemic, especially during the early phases when vaccines are still being developed, or for the segments of the population that respond poorly to vaccination. These include discovering and using new drugs that work against a wide range of viruses, using combinations of previously-discovered antiviral drugs, and using non-pharmaceutical methods such as physical distancing and wearing masks.

Among the defense strategies developed in microbes over millions of years, the innate adaptive CRISPR-Cas immune systems have spread across most of bacteria and archaea. The flexibility, simplicity, and specificity of CRISPR-Cas systems have laid the foundation for CRISPR-based genetic tools. Yet, the efficient administration of CRISPR-based tools demands rational designs to maximize the on-target efficiency and off-target specificity. Specifically, the selection of guide RNAs (gRNAs), which play a crucial role in the target recognition of CRISPR-Cas systems, is non-trivial. Despite the fact that the emerging machine learning techniques provide a solution to aid in gRNA design with prediction algorithms, design rules for many CRISPR-Cas systems are ill-defined, hindering their broader applications. CRISPR interference (CRISPRi), an alternative gene silencing technique using a catalytically dead Cas protein to interfere with transcription, is a leading technique in bacteria for functional interrogation, pathway manipulation, and genome-wide screens. Although the application is promising, it also is hindered by under-investigated design rules. Therefore, in this work, I develop a state-of-art predictive machine learning model for guide silencing efficiency in bacteria leveraging the advantages of feature engineering, data integration, interpretable AI, and automated machine learning. I first systematically investigate the influential factors that attribute to the extent of depletion in multiple CRISPRi genome-wide essentiality screens in Escherichia coli and demonstrate the surprising dominant contribution of gene-specific effects, such as gene expression level. These observations allowed me to segregate the confounding gene-specific effects using a mixed-effect random forest (MERF) model to provide a better estimate of guide efficiency, together with the improvement led by integrating multiple screens. The MERF model outperformed existing tools in an independent high-throughput saturating screen. I next interpret the predictive model to extract the design rules for robust gene silencing, such as the preference for cytosine and disfavoring for guanine and thymine within and around the protospacer adjacent motif (PAM) sequence. I further incorporated the MERF model in a web-based tool that is freely accessible at www.ciao.helmholtz-hiri.de. When comparing the MERF model with existing tools, the performance of the alternative gRNA design tool optimized for CRISPRi in eukaryotes when applied to bacteria was far from satisfying, questioning the robustness of prediction algorithms across organisms. In addition, the CRISPR-Cas systems exhibit diverse mechanisms albeit with some similarities. The captured predictive patterns from one dataset thereby are at risk of poor generalization when applied across organisms and CRISPR-Cas techniques. To fill the gap, the machine learning approach I present here for CRISPRi could serve as a blueprint for the effective development of prediction algorithms for specific organisms or CRISPR-Cas systems of interest. The explicit workflow includes three principle steps: 1) accommodating the feature set for the CRISPR-Cas system or technique; 2) optimizing a machine learning model using automated machine learning; 3) explaining the model using interpretable AI. To illustrate the applicability of the workflow and diversity of results when applied across different bacteria and CRISPR-Cas systems, I have applied this workflow to analyze three distinct CRISPR-Cas genome-wide screens. From the CRISPR base editor essentiality screen in E. coli, I have determined the PAM preference and sequence context in the editing window for efficient editing, such as A at the 2nd position of PAM, A/TT/TG downstream of PAM, and TC at the 4th to 5th position of gRNAs. From the CRISPR-Cas13a screen in E. coli, in addition to the strong correlation with the guide depletion, the target expression level is the strongest predictor in the model, supporting it as a main determinant of the activation of Cas13-induced immunity and better characterizing the CRISPR-Cas13 system. From the CRISPR-Cas12a screen in Klebsiella pneumoniae, I have extracted the design rules for robust antimicrobial activity across K. pneumoniae strains and provided a predictive algorithm for gRNA design, facilitating CRISPR-Cas12a as an alternative technique to tackle antibiotic resistance. Overall, this thesis presents an accurate prediction algorithm for CRISPRi guide efficiency in bacteria, providing insights into the determinants of efficient silencing and guide designs. The systematic exploration has led to a robust machine learning approach for effective model development in other bacteria and CRISPR-Cas systems. Applying the approach in the analysis of independent CRISPR-Cas screens not only sheds light on the design rules but also the mechanisms of the CRISPR-Cas systems. Together, I demonstrate that applied machine learning paves the way to a deeper understanding and a broader application of CRISPR-Cas systems. / Unter den Verteidigungsstrategien, welche sich über Millionen von Jahren in Mikroben entwickelt haben, hat sich das angeborene adaptive CRISPR-Cas Immunsystem in vielen Bakterien und den meisten Archaeen verbreitet. Flexibilität, Einfachheit und Spezifizität von CRISPR-Cas Systemen bilden die Grundlage für CRISPR-basierten genetischen Werkzeugen. Dennoch verlangt die effiziente Anwendung CRISPR-basierter genetischer Werkzeuge ein rationales Design, um die Effektivität zu maximieren und Spezifizität zu gewährleisten. Speziell die Auswahl an Leit-RNAs, oder auch „guide“ RNAs (gRNAs), welche eine essentielle Rolle in der Ziel-Erkennung des CRISPR-Cas Systems spielen, ist nicht trivial. Trotz aufkommender Techniken des maschinellen Lernens, die mit Hilfe von Vorhersage-Algorithmen eine Unterstützung im gRNA-Design darstellen, sind die Design-Regeln für viele CRISPR-Cas Systeme schlecht definiert und die breite Anwendung dadurch bisher gehindert. CRISPR Interferenz (CRISPRi), eine Methode der Genrepression, nutzt ein katalytisch inaktives Cas-Protein, um die Gen-Transkription zu verhindern und ist eine führende Technik für Gen-Funktionsstudien, der Manipulation von Stoffwechselwegen und genomweiter Screens in Bakterien. Auch wenn viele der Anwendungen vielversprechend sind, ist die Umsetzung aufgrund der wenig untersuchten Design-Regeln schwierig. Daher entwickele ich in dieser Arbeit ein hochmodernes auf maschinellem Lernen basierendes Modell für die Vorhersage der gRNA Genrepressions-Effizienz in Bakterien, wobei die Merkmalskonstruktion, Datenintegration, interpretierbare künstliche Intelligenz (KI) und automatisiertes maschinelles Lernen genutzt wurden. Zuerst untersuche ich systematisch die Einflussfaktoren, welche zum Ausmaß der Depletion in genomweiten CRISPRi-Screens zur Gen-Essentialität in Escherichia coli beitragen und demonstriere den überraschend dominanten Beitrag genspezifischer Effekte, wie z. B. dem Genexpressionslevel. Diese Beobachtungen erlaubten mir die genspezifischen Störvariablen mit einem sogenannten mixed-effect random forest (MERF) Modell zu segregieren, um eine bessere Einschätzung der gRNA Effizienz zu erreichen und durch die Integration zusätzlicher Screen-Daten noch weiter zu verbessern. Das MERF Modell übertraf dabei bereits existierende Werkzeuge in einem unabhängigen Hochdurchsatz Sättigungs-Screen. Als nächstes interpretiere ich die Modell Vorhersage, um Design-Regeln für eine solide Genrepression zu extrahieren, wie z. B. eine Präferenz für Cytosin und eine Abneigung gegenüber Guanin und Thymin innerhalb und der „protospacer adjacent motif“ (PAM) direkt umgebenden Sequenz. Weiterhin integrierte ich das MERF Modell in einem Web-basierten Werkzeug, welches unter www.ciao.helmholtz-hiri.de frei zugänglich ist. Ein Vergleich von existierenden Werkzeugen mit dem MERF Modell zeigt, dass alternative, für CRISPRi in Eukaryoten optimierte, gRNA Design-Werkzeuge schlecht abschneiden, sobald sie in Bakterien angewandt werden. Dies lässt Zweifel an einer robusten Übertragbarkeit dieser Vorhersage-Algorithmen zwischen verschiedenen Organismen. Zusätzlich haben CRISPR-Cas Systeme, trotz einiger genereller Gemeinsamkeiten, höchst diverse Wirkungsmechanismen. Die Vorhersagemuster eines Datensets sind daher schlecht generalisierbar, sobald sie auf andere Organismen oder CRISPR-Cas Techniken angewandt werden. Diese Lücke kann mit dem hier präsentierten Ansatz des maschinellen Lernens für CRISPRi geschlossen werden und als eine Vorlage für die Entwicklung effektiver Vorhersage-Algorithmen für spezifische Organismen oder CRISPR-Cas Systeme dienen. Der explizite Arbeitsablauf beinhaltet drei Hauptschritte: 1) Aufnehmen des Merkmalsets des jeweiligen CRISPR-Cas Systems bzw. der CRISPR-Cas Technik; 2) Optimierung des maschinellen Lernen Modells durch automatisiertes maschinelles Lernen; 3) Erklärung des Modells mit interpretierbarer KI. Um die Anwendbarkeit des Arbeitsablaufs und die Diversität der Ergebnisse, im Zusammenhang mit unterschiedlichen Organismen und CRISPR-Cas Systemen, zu demonstrieren, habe ich diese Arbeitsschritte zur Analyse drei unterschiedlicher genomweiter Screens angewandt. Von dem CRISPR „base editor“ Essentialitäts-Screen in E. coli, konnten die PAM Präferenzen und der Sequenzkontext innerhalb des Editierungsfensters für eine effiziente Editierung abgeleitet werden. Beispielsweise tragen ein A an der zweiten PAM Position, ein A/TT/TG an der PAM direkt nachgeschalten Position und ein TC an der vierten oder fünften gRNA Position zur effizienten Editierung bei. Im CRISPR-Cas13a Screen in E. coli, stellten wir eine starke Korrelation zwischen dem Genexpressionslevel und der gRNA-Depletion fest. Zusätzlich ist das Expressionslevel des Ziel-Gens der stärkste Vorhersagefaktor des Modells, was das Expressionslevel als Hauptdeterminante für die Cas13-induzierte Immunität hervorhebt und die bessere Charakterisierung von CRISPR-Cas13 Systemen ermöglicht. Aus dem CRISPR-Cas12a Screen in Klebsiella pneumoniae, habe ich gRNA Design Regeln für die robuste antimikrobielle Aktivität über unterschiedliche K. pneumoniae Stämme hinweg extrahiert und einen Vorhersage-Algorithmus für das gRNA Design bereitgestellt. Dies ermöglicht die Nutzung von Cas12a als eine alternative Lösung, um Antibiotikaresistenzen zu bekämpfen. Zusammengefasst präsentiert diese Thesis einen akkuraten Vorhersage-Algorithmus für die CRISPRi gRNA Effizienz in Bakterien und gibt Einblicke in die Determinanten für eine effiziente Genrepression und optimales gRNA Design. Die systematische Exploration führte zu einem robusten Ansatz des maschinellen Lernens für effektive Modell Entwicklungen in unterschiedlichen bakteriellen Spezies und CRISPR-Cas Systemen. Durch die Anwendung dieses Ansatzes auf unabhängige CRISPR-Cas Screens, konnte ich nicht nur wichtige Design Regeln ableiten, sondern auch die Mechanismen der jeweiligen CRISPR-Cas Systeme besser erleuchten. Zu guter Letzt demonstriere ich hier, dass angewandtes maschinelles Lernen den Weg zu einem tieferen Verständnis und einer breiteren Anwendung von CRISPR-Cas Systemen ebnen kann.