In vitro evolution of uropathogenic Escherichia coli to fosfomycin resistance in a 3D cultured human bladder microtissue model

Technical Summary: In vitro evolution of uropathogenic Escherichia coli to fosfomycin resistance in a 3D cultured human bladder microtissue model

Problem Statement
Current in vitro studies of antimicrobial resistance (AMR) predominantly rely on laboratory growth media, which fail to accurately recapitulate the in vivo infection environment. This divergence limits the translational potential of such research, as the conditions governing AMR acquisition and evolution in the laboratory may not reflect those found in clinical settings. While longitudinal clinical isolate collections exist, they are often retrospective, expensive, and difficult to obtain. Although 3D culture models exist for studying host-pathogen interactions, their application to the study of AMR acquisition and evolution remains limited. There is a specific need for physiologically relevant models that can simulate the human bladder microenvironment to better understand how resistance evolves during treatment.

Methodology
The authors utilized the 3D Urine-tolerant Human Urothelial (3D-UHU) model, a system that mimics the human bladder urothelium by stratifying into 7–8 cell layers (basal, intermediate, and umbrella) and tolerating 100% human urine for up to four weeks.

• Strains: Two uropathogenic E. coli clinical isolates, UTI-34 (ST73) and UTI-59 (ST127), were selected based on their susceptibility to common UTI antibiotics.
• Evolution Experiment: The researchers performed an experimental evolution study within the 3D-UHU model. Bacteria were exposed to increasing concentrations of fosfomycin (0.25x, 0.5x, and 1x MIC) over 72 hours in the apical chamber containing pooled human urine supplemented with glucose-6-phosphate (to facilitate fosfomycin uptake via the UhpT transporter).
• Selection: Variants were recovered on Mueller Hinton (MH) agar containing 4x and 8x MIC of fosfomycin. Five independent variants were generated for each ancestor strain (one variant, 34D, was excluded as a contaminant).
• Genomic Analysis: Whole-genome sequencing (Illumina and Oxford Nanopore) was performed on the ancestral strains and the selected variants. Mutations were called using Snippy against the respective ancestral genomes.
• Clinical Relevance Screening: A bioinformatics pipeline was developed to screen the observed mutations against a curated dataset of 14,163 E. coli genomes isolated from human urine (derived from the BakRep database). The screening employed a three-tier evidence system: exact match, regional match (within ±20 amino acids), and gene-level match.
• Phenotypic Characterization: Minimum Inhibitory Concentrations (MICs) were determined in both MH broth and conditioned urine. Collateral susceptibility to cephalexin, nitrofurantoin, and trimethoprim was assessed via disc diffusion. Fitness was evaluated through growth curve analysis in MH broth.

Key Results

• Resistance Evolution: All successfully recovered fosfomycin-selected variants exhibited resistance to fosfomycin.
• Mutational Landscape:
  • UTI-34 Variants: Showed diverse mutational patterns. Variants contained mutations in glpT (deletions and missense), uhpT (nonsense), uhpC (stop-loss), cyaA (missense), and ptsI (frameshift).
  • UTI-59 Variants: Showed more constrained evolution. All variants (except 59D) harbored single base-pair deletions in glpT and uhpT causing frameshifts. Variant 59D contained a 9 bp in-frame deletion in uhpA.
  • Commonality: All variants possessed mutations in genes known to be linked to clinical fosfomycin resistance (glpT, uhpT, uhpA, uhpC).
• Clinical Relevance: Screening against the human urine E. coli dataset confirmed that the observed mutations were representative of clinical isolates.
  • Exact matches were found for in-frame deletions in glpT and uhpA, stop-loss in uhpC, and nonsense mutations in uhpT.
  • Regional and gene-level matches were confirmed for frameshifts in ptsI and glpT, and missense mutations in cyaA.
  • The incidence of these mutations varied, with frameshifts in glpT and uhpT being relatively common in the dataset.
• Collateral Effects:
  • No statistically significant collateral resistance or susceptibility to cephalexin or nitrofurantoin was observed in the variants.
  • A statistically significant increase in susceptibility to trimethoprim was observed in UTI-34 variants, but not in UTI-59 variants.
• Fitness:
  • UTI-34 variants generally showed a significant decrease in carrying capacity and area under the curve (AUC), with an increased maximum growth rate, except for variant 34E.
  • UTI-59 variants largely showed no significant difference in fitness parameters compared to the ancestor, with the exception of variants 59B and 59C which showed reduced AUC and carrying capacity.

Significance and Claims
The paper posits that the 3D-UHU model serves as a viable platform for studying AMR acquisition under physiologically relevant conditions that closely mimic the human bladder. The study demonstrates that this model can select for clinically relevant mutations in uropathogenic E. coli that are indistinguishable from those found in clinical isolates from urine.

The authors claim that using such 3D microtissue models improves the clinical relevance of in vitro AMR studies, offering a better understanding of how resistance is acquired and expressed in an infection context. They suggest this approach has the potential to inform new strategies to combat AMR. The study is presented as a "proof-of-concept," acknowledging limitations such as the use of monocultures rather than polymicrobial communities and the use of a limited panel of isolates. The authors do not claim that the model replaces clinical data but rather that it offers a controlled, prospective alternative to retrospective clinical studies for investigating the mechanisms of resistance evolution.