Unlocking the potential of experimental evolution to study drug resistance in pathogenic fungi

Abstract

Exploring the dynamics and molecular mechanisms of antimicrobial drug resistance provides critical insights for developing effective strategies to combat it. This review highlights the potential of experimental evolution methods to study resistance in pathogenic fungi, drawing on insights from bacteriology and innovative approaches in mycology. We emphasize the versatility of experimental evolution in replicating clinical and environmental scenarios and propose that incorporating evolutionary modelling can enhance our understanding of antifungal resistance evolution. We advocate for a broader application of experimental evolution in medical mycology to improve our still limited understanding of drug resistance in fungi.

Fig. 1. Measuring competitive fitness.

A An example of a competition experiment in which the population composition changes under the selective pressure of a drug. B Strains tagged with antimicrobial resistance markers like a nourseothricin (NTC) resistance gene (NAT), or hygromycin B (HYG) resistance gene (HPH) allow differential plating on NTC or HYG after competition. C Expression of fluorescent proteins like green fluorescent protein (GFP) or red fluorescent protein (RFP) enables discrimination between subpopulations using flow cytometry. D Barcode sequences in strains allow differentiation by next-generation sequencing (NGS). With deep sequencing (E) and qPCR (F), relative population sizes are estimated by the relative quantification of unique genetic markers, like antifungal drug resistance conferring mutations, based on NGS coverage or PCR amplification. Figure created with BioRender.com.

Fig. 2. Methods of experimental evolution.

A Serial dilution of the culture into fresh media, with at each step the possibility to adapt the drug concentration or create variable conditions. Various vessels can be used, including 96-well plates for large-scale replicates. B Morbidostats or chemostats provide a constant supply of fresh medium for continuous growth without imposing large population bottle necks (←→ serial dilution) and can adjust the drug concentration based on cell growth, measured for example by monitoring optical density (OD). C Spatial gradients can be created using agar plates with varying concentrations across the surface and inoculate the whole surface from the start or inoculate a part of the surface and rely on mobility of microbial cells to migrate across the surface. D Mathematical evolutionary modelling requires experimentally obtained parameters but simulates evolution in silico. E Different in vivo models, including Galleria mellonella and Caenorhabditis elegans can also be used for in vivo experimental evolution. Here shown are murine models of systemic, oropharyngeal and gastric infection, but murine models for skin and vagina, have also been used to study fungal infections. Figure created with BioRender.com.

Fig. 3. Potentially important aspects and methodologies of experimental evolution.

A Drugs can be added sequentially or simultaneously, as a variation on single drug exposure. Combination therapy allows the study of drug interactions (B), with lines representing growth isoboles. One drug concentration is linearly increased in each axis and the isoboles represent the combined drug effect where cell growth is inhibited. A straight isobole indicates no interaction, so an additive effect. If the drug combination achieves the same growth inhibition with a lower dose than the additive case, the combination is considered synergistic, indicated with convex isoboles. When the opposite is true, the drugs interact antagonistically and the isoboles are concave. C Representation of collateral sensitivity, where cells develop resistance to drug A, leading to increased susceptibility to drug B. D Example of hypothetical patient plasma concentrations of two drugs, which can be simulated in vitro using automated drug dosing in a bioreactor. E The range of effects caused by a mutation can be depicted as a distribution of fitness effects (DFE). Typically, most mutations have harmful effects and are quickly eliminated from the population. Mutations with neutral or nearly neutral impacts occur at a frequency resembling a clock-like distribution. Only a small number of mutations are beneficial. F A variant DNA library is introduced into cells of interest to create a mutant cell library. This library undergoes the treatment of interest, where cells with functional variants are enriched, while those with detrimental variants are depleted. The ratios of the genetic variants are determined before and after the drug exposure by deep sequencing to determine relative changes. Finally, the enrichment scores are analysed to assess the functional impact of the mutations. Figure created with BioRender.com.

Fig. 4. Example of a hypothetical deterministic modelling framework for predicting the population size of strains based on their fitness-resistance profile in a ‘virtual patient’ after x days of treatment.

A The ordinary differential equations (ODE) for the susceptible (S) and resistant (R) cell population over time (t) form the basis of the deterministic model and account for cell growth, drug effects, and immune response (indicated by different colours in the equation). Only the ODE of the resistant cell population is shown for simplicity. Susceptibility (MIC) and growth parameters (g, K, F) are detailed in (B) and (C), determined by respectively the dose response curve and the growth curve of the mutants. The drug concentration depends on drug decay in the patient (h), shown in (D). In this example, exponential decay is assumed, and h is dependent on the half-life of the drug. The drug induced killing is determined by the Hill coefficient (H), indicated by a varying steepness of the dose response curve, shown in (E). The immune effect is simplified as one variable (I) for simplicity. F Hypothetical modelling output depicting the relationship between fitness-resistance profile of three hypothetical mutants (coloured dots) and the predicted population sizes (colour of heatmap) of these three mutants after x days of treatment in a ‘virtual patient’. In this case, only the mutant with low resistance and a medium fitness trade-off is expected to survive treatment, while susceptible or highly resistant but less fit mutants are predicted to go extinct. Figure created with BioRender.com.