Overview

Project Highlights:

  • Do low concentrations of antibiotics, heavy metals or disinfectants select for resistance in the environment?
  • Is horizontal gene transfer an efficient mechanisms for maintaining resistance genes in the absence of antibiotic selection?
  • Which potential mitigations strategies would be likely to reduce the level of resistance?

Overview

The widespread use of antibiotics in human and veterinary medicine and as growth promoters in agriculture has not only selected for resistance genes but also for plasmids carrying them. Mixtures of antimicrobials select for insertion sequence elements as they facilitate the accumulation of multiple resistances, increasing the potential for pathogens to acquire new resistances and new combinations of resistances. Multidrug resistant pathogens have brought us back to the pre-antibiotic era.

Hotspots of resistance genes in the environment are wastewater treatment plants and animal manures and slurries on farms, which are being spread on arable land. Runoff from these fields and effluent from wastewater treatment enter rivers and river sediments, also untreated sewage is discharged into rivers through storm overflow drains.

As partner of an international JPIAMR-funded consortium, we developed a mathematical model of AMR dynamics in wastewater treatment that makes a number of important predictions (Fig. 1). Our partner at Newcastle University (Prof Graham) is testing these predictions. We will be developing this single compartment model into a model of multiple compartments to understand AMR transmission from residential or hospital sewage via wastewater treatment into receiving rivers. Other partners are making measurements in Denmark, Spain and the UK. We are also partner in a NERC-funded project investigating AMR dynamics in dairy farms that can provide further data on slurries as a source of AMR genes.

The group of Prof Wellington in collaboration with CEH and Thames Water has been sampling 69 river sites along the Thames including tributaries to do qPCR and 16S rRNA amplicon and metagenome sequencing. The sequence data is being analysed in collaboration with Dr Quince. Both will be co-supervisors of the student. Amos et al. (2015) fitted a mostly statistical model to the data. Our aim is to use a mechanistic, mathematical model instead. Also, there are now a lot more data available.

In the project, the student will further develop our mathematical model to include AMR dynamics in rivers to generate a more complete catchment scale model and use a Bayesian framework to select appropriate model variants and infer parameters including their uncertainty. This can then inform risk analysis.

Figure 1: Our mathematical model of resistance plasmid transmission in the activated sludge stage of wastewater treatment predicts a selective window at much lower concentrations of an antibiotic than previously thought based on laboratory experiments. The conventional selective window would be at higher antibiotic concentrations, between the MICs of the susceptible and the resistant bacteria.

Methodology

The mathematical model will be based on ordinary differential equations and potentially include an agent-based sub-model to capture the nested organization of resistance genes on plasmids in host bacteria. We are using this approach to extend the Activated Sludge Model 1 with resistance plasmid transmission, antibiotic turnover and including enteric bacteria from faeces to model exchange of plasmids between enteric and indigenous wastewater bacteria. We are currently analysing the one compartment model (manuscript in preparation) and will develop the multi-compartment model and the Approximate Bayesian Computation (ABC) statistical model selection and inference methodology in 2019. We have used ABC before to identify the best model to describe predator prey interactions as this method can be used when the likelihood function is not known.

Training and Skills

The Doctoral Researcher (DR) will acquire a broad set of mathematical modelling, statistical data analysis and programming skills.

This will be enhanced by interdisciplinary collaboration with the Wellington and Quince groups as the DR will learn to communicate and collaborate with experimental researchers. The modelling will also guide future experimental effort by identifying the most important parameters and processes.

Moreover, the DR will have the opportunity for public outreach activities to inform the AMR debate with the results of our project. Project management and communication skills will also be gained.

Timeline

Year 1:

Learn AMR dynamics modelling and Bayesian inference from postdoc on JPIAMR grant. Then extend this framework by including river and farm compartments to create a complete model of AMR transmission on the scale of a catchment. Identify the best way to utilize the rich and growing data from the Thames catchment in collaboration with another CENTA DR, James Delaney, supervised by Wellington & Quince, who is working with this dataset.

Year 2:

Improving on the mostly statistical analysis of a fraction of these data published by Amos et al. (2015) by using the mechanistic mathematical model to infer the parameters describing the underlying processes in AMR transmission rather than just fit a statistical model to the outcome of these processes (paper 1).

Year 3:

Bringing together the data from JPIAMR grant on the upstream sewage and wastewater treatment compartments with the Thames data on rivers and the NERC grant on farm slurries to create a complete model of AMR transmission in a catchment and infer parameters (paper 2).

Simulate the effect of various mitigation strategies on AMR transmission (paper 3).

If time remains, use the model and its inferred parameters to quantify the risk of resistance transmission in different compartments and overall (paper 4).

Partners and collaboration (including CASE)

The Kreft group has >15 years’ experience in mathematical modelling, including modelling plasmid dynamics, investigating the fate of resistance on a dairy farm and in urban waters in Denmark, Spain and the UK.

The Wellington lab has for many years driven forward the research on AMR dynamics in the Thames catchment with collaborators from CEH, Thames Water and others.

The Quince group is at the forefront of developing more rigorous statistical and bioinformatic algorithms e.g. for reconstructing genomes from metagenomes.

There is strong potential for conversion to CASE that will be pursued before project start (Thames Water, Environment Agency or DEFRA, AstraZeneca).

Further Details

Dr Jan-Ulrich Kreft

School of Biosciences & Institute of Microbiology and Infection & Centre for Computational Biology

The University of Birmingham

Edgbaston, Birmingham, B15 2TT, UK

Tel: +44 (0)121 41-48851

Email: j.kreft@bham.ac.uk

Web: www.tinyurl.com/kreftlab

 

Professor E M H Wellington

School of Life Sciences

The University of Warwick

Coventry CV4 7AL, UK

Tel: +44 (0)2476 523184

Email: e.m.h.wellington@warwick.ac.uk

Web: http://www2.warwick.ac.uk/fac/sci/lifesci/people/ewellington

 

Dr Christopher Quince

Warwick Medical School – Microbiology and Infection

The University of Warwick

Coventry CV4 7AL, UK

Tel: +44 (0)2476 522317

Email: C.Quince@warwick.ac.uk

Web: https://warwick.ac.uk/fac/sci/med/staff/cquince/