Science

The issue of antibiotic resistance is a critical concern for public health, prompting numerous investigations into the impact of treatment strategies on preventing or slowing down the emergence of resistance. While existing studies have predominantly focused on chromosomal resistance mutations, the consequences of often clinically more relevant plasmid-conferred resistance remain insufficiently explored. To address this gap, we conducted three extensive in vitro experiments utilizing a liquid-handling platform. These experiments evaluated the efficacy of five distinct treatment strategies using two antibiotics (tetracycline and ceftazidime) along with two horizontally transmissible clinical resistance plasmids conferring the respective resistances. Among the experimentally investigated treatment strategies, combination therapy proved to be the most effective in preventing the emergence of double resistance while minimizing the number of infections. To verify the reliability of these findings, we constructed a computational model of our experiments that we parameterized using the experimental data. We employed this model to augment the experimental data by conducting an in silico parameter sensitivity analysis. The sensitivity analysis corroborated our experimental results, demonstrating that combination therapy consistently outperformed other treatment strategies across a range of parameter values.

Bacterial population decline at antibiotic concentrations above the minimum inhibitory concentration (MIC) remains poorly characterized. This is because colony-forming units (CFU), the standard method to quantify inhibition, are slow, labor-intensive, and costly. Luminescence assays are widely used to quantify population dynamics at subinhibitory concentrations, yet their limitations and reliability at high concentrations remain underexplored. Here, we compared luminescence- and CFU-based rates in Escherichia coli across 20 antimicrobials. In our experiments, luminescence- and CFU-based rates did not differ significantly for half of them. For the other half, CFU-based decline rates were consistently higher. The estimates differed for two main reasons: First, because light intensity tracks biomass more closely than population size, luminescence declined more slowly than the population when bacteria filamented. Second, CFU-based estimates indicated a steeper decline when treatment reduced the number of colonies formed per plated bacterium. This can result from changes in clustering behavior, physiological changes that impair culturability, or antimicrobial carryover. Thus, the suitability of luminescence to quantify bacterial decline depends on the physiological effects of the antimicrobial and whether the quantity of interest is cell number or biomass. Within these limitations, luminescence can serve as an efficient, high-throughput alternative for quantifying bacterial dynamics at super-MIC concentrations.

Assessing whether drug combinations synergise or antagonise is difficult for several reasons: (i) measuring bacterial death rates at clinically relevant inhibitory drug concentrations is methodologically challenging, (ii) there is no unifying definition of what constitutes synergy or antagonism, and (iii) both synergy and antagonism may be concentration- and mixing-ratio-dependent. To assess how well sub-inhibitory measurements predict inhibitory behaviour, we quantified drug interactions for 15 pairwise drug combinations on a concentration checkerboard covering a wide range of inhibitory and sub-inhibitory concentrations. To this end, we tracked the population dynamics of 8640 bioluminescent E. coli cultures by recording their light-intensity trajectories. To handle time-varying treatment effects and allow fair comparisons between drugs with distinct killing dynamics, we used a time-weighted net growth rate ψ to summarise each trajectory and assigned interaction labels (synergistic/independent/antagonistic) based on Bliss independence and Loewe additivity. We found that the interaction label depends on both the concentration and the mixing ratio, frequently changing between the sub-inhibitory and inhibitory regimes. Characterising drug combinations at a single sub-inhibitory concentration is therefore not sufficient. Instead, their combined effects should be assessed at the conditions of their intended use.

Since the discovery of penicillin, antibiotics have been a cornerstone of modern medicine. This achievement is now under threat, as bacteria have evolved resistance to antibiotics across all major drug classes. Slowing the rise of resistance will require changes in how existing antibiotics are deployed. In this thesis, we investigate the pharmacodynamics of drug combinations and how multidrug treatment strategies shape the dynamics of plasmid-mediated resistance.

In time-critical clinical emergencies such as sepsis, therapy cannot wait for phenotypic susceptibility testing and therefore relies on predefined empirical strategies. To assess how these strategies affect the clearance probability and the plasmid-mediated emergence of double resistance, we conducted large-scale automated in vitro experiments that mimic hospital-like transmission dynamics. Across most scenarios, treating with two antibiotics simultaneously (combination therapy) was the most effective strategy.

Because the effectiveness of combination therapy is shaped by drug interactions (i.e. synergy, antagonism, or independence), we next set out to quantify the treatment effects of several drug combinations. To assess treatment effects under clinically relevant inhibitory conditions in high throughput, we evaluated whether bioluminescence-based light intensity is a suitable proxy for cell number dynamics.

For 20 antimicrobials, we compared bioluminescence trajectories to colony-forming unit (CFU) counts and supplemented these experiments with microscopy imaging. We found that light intensity dynamics align more closely with biomass dynamics than with cell number dynamics. Accordingly, rates inferred from bioluminescence and from cell numbers tend to align when cell size remains approximately constant. Furthermore, we observed that CFU-based estimates can be biased by drug-induced changes in culturability and by antibiotic carry-over.

Using the bioluminescence method, we quantified antibiotic interactions for 15 drug pairs across checkerboards spanning sub-inhibitory to inhibitory concentrations. We found that interaction types at sub-inhibitory concentrations frequently differ from those at inhibitory concentrations. In addition, interaction types can vary with mixing ratio and depend on the chosen reference model. Together, these results highlight the potential of combination therapy, provide methodological insights to optimise it, and caution against uncritical extrapolation of findings across the measured concentration space.

The rise of multidrug-resistant bacteria increasingly threatens the achievements of antibiotic therapy: fewer new antibiotics are being developed, and resistance appears ever earlier. Clinical practice routinely tests bacteria for resistance to single antibiotics, but rarely to combinations — even though combinations could contribute substantially to containing resistance, and no published method exists for estimating how likely resistance to a given combination is. This thesis develops and tests such a method. Two drugs are mixed at varying ratios inside a microfluidic structure and diluted stepwise. Bacteria grow against the resulting concentration gradient until they reach the minimum inhibitory concentration (MIC); once resistance mutations occur, the MIC boundary shifts toward higher concentrations. The time this takes is a useful indicator of the likelihood of future resistance.

The objective of the ensuing Bachelor thesis was to develop a test bench for a flow visualization in the left ventricular assist device "HeartMate 3" by Thoratec. In order to achieve this, the HeartMate 3 was reverse engineered and reproduced on a 3:1 scale. Flow visualization was implemented by using the particle image velocimetry method. Due to a design error, the results cannot directly be transferred from the model to the original pump. Nevertheless the test rig proved to be suitable. It could be operated comfortably and the experiments performed successfully. The experiments provided valuable information on the existence of ring vortex structures in the upper gap of the pump. These factors have a major influence on the residence times of the blood components in the upper slit. For this reason the detection of these vortex structures and the associated explanation constitute an important contribution to the further development of heart support systems.