Abstract:
The Review on Antimicrobial Resistance predicts that in thirty years infections with antibiotic-resistant microorganisms will become one of the leading causes of death. The discovery of new antibiotics has so far been too slow to ensure continuous use of antibiotics in the face of growing resistance. Therefore, efforts to curb resistance emergence gain in importance. These efforts comprise two complementary strategies. The first focuses on the mechanisms of resistance emergence, in the hope that it would enable development of pharmacological agents constraining resistance emergence. The second aims at improving antibiotic use practices, based on studies of the impact of antibiotics on resistance emergence within patient populations. Antibiotic resistance emerges in bacterial cells, negatively influences the human gut microbiome, and transfers between people. Hence, antibiotic resistance has impacts across several levels of biological organization.
This thesis describes four projects, which concerned various aspects of antibiotics resistance. The first two projects deal with basic resistance emergence mechanisms, on the level of bacterial strains and bacterial consortia, whereas the other two deal with finding better practices for antibiotic use on a population level.
During the first project, I analyzed changes in genomes of MRSA strains isolated from several patients throughout antibiotic therapies and developing MRSA infections. I observed changes in number and types of virulence factors responsible for interacting with the human body, which are attributed to mobile genetic elements. In the second project, I showed that, prompted by antibiotic therapy, within the human gut microbiome resistance transfers from bacterial genomes onto plasmids, prophages, and free phages. Hence, resistance emergence depends not only on the antibiotic therapy but also on the state of the gut microbiome, which again results from the patients’ overall health and previous antibiotic therapies.
The third project, SATURN, employed machine learning methods for a large set of data regarding patients’ demographics, comorbidities, antibiotic therapies, surgeries, and colonization with multi-drug resistant bacteria. The final classifiers were made available on the AskSaturn website where the doctors can compare antibiotic therapies based on the probability of colonization with multi-drug resistant bacteria. The fourth project, Tübiom, focused on the antibiotic-influenced gut microbiomes of the healthy population.
The first two projects rely on genome and metagenome sequencing data. For them, I designed specialized bioinformatics analysis pipelines. The latter two projects use mixed data, which were analyzed with machine learning algorithms. These projects also involved web development and data visualization. Although each of the projects requires different data and methods, each of them provides a crucial part in a pipeline aiming at utilizing gut microbiome information in medical practice to constrain resistance emergence.
Print-on-Demand