Skip to main content

Microfluidics and modelling to map antibiotic resistance of individual cells and populations

Principal Supervisor: Dr. Jan-Ulrich Kreft, School of Biosciences

Co-supervisor: Dr. Daniele Vigolo, School of Chemical Engineering

PhD project title: Microfluidics and modelling to map antibiotic resistance of individual cells and populations

University of Registration: University of Birmingham

Project outline:

The rise of antimicrobial and antibiotic resistance threatens our ability to cure infections. If we do not tackle this crisis, we may return to the pre-antibiotic era. Many studies have looked at the evolution of resistance and the effect of inhibitory and sub-inhibitory concentrations of antibiotics. Most of these studies have used populations of many cells assuming that the variation between individuals is not important enough to warrant investigation. Some have studied the response of individual cells to antibiotics, e.g. to see how the response is affected by the age of the cell, but not over a range of different concentrations.

Our central hypothesis is that antibiotic susceptibility and fitness costs of resistance mutations or plasmids are affected by the growth rate and physiology of individual cells. As a consequence, individuals in a population will not be all the same, but differ in important ways. We will construct and use microfluidic devices consisting of two components, a gradient mixer and a microchemostat, to create concentration gradients of growth substrates and/or antibiotics to study the effect of growth rate and antibiotic concentration on individual cells growing under constant and defined conditions in cell-sized channels. We can observe and track many cells with a microscope and then use image analysis to measure e.g. growth rate or time to killing. Using these measurements of many individual cells, we can then use individual-based mathematical models to predict the behaviour of populations and how population responses are affected by the differences between individuals.

For example, generating a gradient of substrate concentrations that will lead to a gradient of growth rates, we can measure how growth rate affects antibiotic susceptibility. As we can have thousands of channels in a microfluidic device, we can record this for hundreds of cells and learn how much the responses differ between cells. Feeding this information into an individual-based model, we can then predict how the antibiotic affects the population. These predictions can be tested in larger culture vessels such as flasks. The model can also make predictions on the effect of antibiotics on biofilms as the growth of a biofilm generates a substrate concentration gradient, and from recording growth and antibiotic inhibition over a range of concentrations, we can predict the growth and inhibition of each cell in the biofilm depending on where in the gradient it is located.

Another example of the use of the microfluidic device and mathematical modelling is to study the cell-to-cell variation of fitness costs of resistance genes or plasmids. Fitness costs are typically measured for populations growing under optimal conditions, that is, at high growth rates. We do not know whether fitness costs will be different when cells grow more slowly but this could have huge consequences. What we do know is that bacteria grow much more slowly in their natural environment than under optimal conditions: E. coli has a doubling time of about 24 hours in the gut. It may also be that individual cells with higher fitness costs (more retarded growth) have lower antibiotic sensitivity. There is a lot we do not know and the methodology in this project will be able to address many fundamentally important and clinically relevant questions. There is a host of opportunities to go beyond the state of the art. We have developed some prototype microfluidic devices with the help of two MSc project students and optimized some procedures, but it will need some more testing and optimization before we can use them routinely in high throughput mode. We have explained how the integration of single cell measurements and individual-based modelling will advance microbial sciences in the opinion article below.

References:

  • Hellweger FL, Clegg RJ, Clark JR, Plugge CM, Kreft JU (2016). Advancing microbial sciences by individual-based modelling. Nature Reviews Microbiology 14: 461–471.
  • Hol FJH, Dekker C (2014). Zooming in to see the bigger picture: Microfluidic and nanofabrication tools to study bacteria. Science 346: 1251821.

BBSRC Strategic Research Priority: Food Security

Techniques that will be undertaken during the project:

  • Design, manufacture (i.e., soft-lithography) and operation of microfluidic devices.
  • Automated microscopic imaging (including fluorescence, phase contrast and DIC) and image analysis.
  • All optical velocimetry technique to evaluate the flow field within the microfluidic device (Ghost Particle Velocimetry).
  • General laboratory and microbiological methods.
  • Molecular microbiology and strain construction.
  • Individual-based modelling.
  • Statistics.
Contact:  Dr Jan-Ulrich Kreft, School of Biosciences