Search

Antibiotic resistance is widely recognized as one of the biggest threats to global health.In hospitals, the susceptibility of a strain to an antibiotic is quantified by its Minimum Inhibitory Concentration (MIC): the minimal concentration of antibiotic necessary to inhibit the growth of the strain during 24 hours. This value plays a central role for treatment decisions.However, the MIC is a measure relying on a unique timepoint. Could we get a more informative assessment of antibiotic resistance by exploiting the whole growth curve, observed by optical density? This information could be available in a clinical context, which is a requirement of the approach. The problem is complex, notably because β-lactam antibiotics induce cell filamentation, which decorrelates the optical density from the number of live cells.In this thesis, we build a mathematical model of the response of bacterial populations to β-lactams, encompassing the different kinds of antibiotic resistance under a unifying framework. Bridging the three scales: molecular-, cell-, and population-level, this model provides simultaneous predictions of the optical density and the number of cells. Its core is a growth-fragmentation equation: a partial differential equation that considers explicitly the distribution of cell lengths. The PDE model is not very practical for numerical optimization, notably for parameter inference. Therefore, we describe the passage to a companion ODE model for efficient calibration.After calibrating this model on a library of clinical isolates with the help of a custom driver allowing the programmable use of a commercial plate reader, we show that we can relate several parameters to the antibiotic resistance genes and mutations present in the strains. We then propose a method to cluster the strains despite the presence of unidentifiable parameters, and show that three classes emerge: sensitive, tolerant/resilient, and resistant strains. In comparison with the classical system susceptible, intermediate, and resistant, these classes provide a richer explanation of the behaviour of the isolates, and allow a direct exploitation for treatment optimization., La résistance aux antibiotiques est connue comme l'un des plus grands dangers de santé publique. Dans les hôpitaux, la susceptibilité d'une souche à un antibiotique est quantifiée par sa Concentration Minimale Inhibitrice (CMI) : la dose minimale d'antibiotique nécessaire pour inhiber la croissance de la souche pendant 24 heures. Cette valeur joue un rôle central dans les décisions de traitements.Or, la CMI est une mesure reposant sur un unique point de temps. Pourrait-on obtenir une évaluation plus informative de la résistance d'une souche, en exploitant sa courbe de croissance entière, observée par densité optique (DO) ? Cette donnée pourrait être disponible dans un contexte clinique, ce qui est nécessaire pour la pertinence de l'approche. Le problème est complexe, notamment parce que les antibiotiques β-lactames provoquent la filamentation des cellules, ce qui décorrèle la DO du nombre de cellules vivantes.Dans cette thèse, nous développons un modèle mathématique de la réponse de populations bactériennes à des β-lactames, qui rassemble les différents types de résistance. Unifiant les échelles moléculaire, de la cellule et de la population, ce modèle offre des prédictions simultanées de la DO et du nombre de cellules. Son cœur est un modèle de croissance-fragmentation : une équation aux dérivées partielles considérant explicitement la distribution des tailles des cellules. Or, le modèle à EDP n'est pas idéal pour l'optimisation numérique, et notamment pour l'inférence de paramètres. Nous décrivons donc le passage à un modèle compagnon à équations différentielles ordinaires, pour une calibration efficace.Après calibration de ce modèle sur un ensemble d'isolats cliniques à l'aide d'un pilote sur mesure permettant l'automatisation d'un lecteur de plaques, nous montrons que nous pouvons relier plusieurs paramètres du modèle aux gènes et mutations contribuant à la résistance des souches aux antibiotiques. Nous proposons ensuite une méthode permettant de catégoriser les souches, en dépit de la présence de paramètres non identifiables, et observons l'émergence de trois classes : les souches sensibles, les souches tolérantes et résilientes, et les résistantes. En comparaison avec le système classique définissant les souches susceptibles, intermédiaire, et résistantes, ces classes fournissent une explication plus riche du comportement des isolats, et offrent un débouché direct sur l'optimisation de traitements.

Dense arrangements of binding sites within nucleotide sequences can collectively influence downstream transcription rates or initiate biomolecular interactions. For example, natural promoter regions can harbor many overlapping transcription factor binding sites that influence the rate of transcription initiation. Despite the prevalence of overlapping binding sites in nature, rapid design of nucleotide sequences with many overlapping sites remains a challenge. Here, we show that this is an NP-hard problem, coined here as the nucleotide String Packing Problem (SPP). We then introduce a computational technique that efficiently assembles sets of DNA-protein binding sites into dense, contiguous stretches of double-stranded DNA. For the efficient design of nucleotide sequences spanning hundreds of base pairs, we reduce the SPP to an Orienteering Problem with integer distances, and then leverage modern integer linear programming solvers. Our method optimally packs libraries of 20-100 binding sites into dense nucleotide arrays of 50-300 base pairs in 0.05-10 seconds. Unlike approximation algorithms or meta-heuristics, our approach finds provably optimal solutions. We demonstrate how our method can generate large sets of diverse sequences suitable for library generation, where the frequency of binding site usage across the returned sequences can be controlled by modulating the objective function. As an example, we then show how adding additional constraints, like the inclusion of sequence elements with fixed positions, allows for the design of bacterial promoters. The nucleotide string packing approach we present can accelerate the design of sequences with complex DNA-protein interactions. When used in combination with synthesis and high-throughput screening, this design strategy could help interrogate how complex binding site arrangements impact either gene expression or biomolecular mechanisms in varied cellular contexts., Author Summary: The way protein binding sites are arranged on DNA can control the regulation and transcription of downstream genes. Areas with a high concentration of binding sites can enable complex interplay between transcription factors, a feature that is exploited by natural promoters. However, designing synthetic promoters that contain dense arrangements of binding sites is a challenge. The task involves overlapping many binding sites, each typically about 10 nucleotides long, within a constrained sequence area, which becomes increasingly difficult as sequence length decreases, and binding site variety increases. We introduce an approach to design nucleotide sequences with optimally packed protein binding sites, which we call the nucleotide String Packing Problem (SPP). We show that the SPP can be solved efficiently using integer linear programming to identify the densest arrangements of binding sites for a specified sequence length. We show how adding additional constraints, like the inclusion of sequence elements with fixed positions, allows for the design of bacterial promoters. The presented approach enables the rapid design and study of nucleotide sequences with complex, dense binding site architectures.

Splitting proteins with light- or chemically-inducible dimers provides a mechanism for post-translational control of protein function. However, current methods for engineering stimulus-responsive split proteins often require significant protein engineering expertise and laborious screening of individual constructs. To address this challenge, we use a pooled library approach that enables rapid generation and screening of nearly all possible split protein constructs in parallel, where results can be read out using sequencing. We perform our method on Cre recombinase with optogenetic dimers as a proof of concept, resulting in comprehensive data on split sites throughout the protein. To improve accuracy in predicting split protein behavior, we develop a Bayesian computational approach to contextualize errors inherent to experimental procedures. Overall, our method provides a streamlined approach for achieving inducible post-translational control of a protein of interest.