This document outlines the detailed methodologies employed in a study investigating Mycobacterium abscessus and Mycobacterium tuberculosis, covering sample collection, bacterial growth, drug susceptibility, and advanced single-cell analysis techniques.
Sample Collection and Ethics
Mycobacterium abscessus ATCC-19977, avirulent M. tuberculosis H37Ra, and auxotrophic M. tuberculosis mc27000 were utilized. Clinical M. abscessus isolates were obtained from respiratory samples of patients across the UK, Ireland, Sweden, Denmark, the Netherlands, and Australia. All work adhered to BSL-2 regulations and received ethical approval from relevant boards.
Bacterial Growth Assessments
Planktonic growth rates and lag times for M. abscessus were determined by inoculating bacteria in Middlebrook 7H9 medium, incubating at 37°C with shaking, and measuring optical densities (OD 565). Gompertz functions were fitted to background-corrected OD readings to estimate growth rates and lag times using specific formulas.
Drug Susceptibility Testing (DST)
Antibiotic resistance was quantified using Minimum Inhibitory Concentrations (MICs) according to Clinical Laboratory Standards Institute (CLSI) guidelines. M. abscessus isolates were prepared and tested against a panel of antibiotics in 96-well plates incubated at 30°C. M. tuberculosis DST followed a similar protocol, with different antibiotic panels and incubation times.
CFU-Based Time–Kill Kinetics
M. tuberculosis mc27000 was cultured to mid-log phase, starved, and exposed to various antitubercular drugs at maximum therapeutic blood concentrations (Cmax) in 96-well plates. Colony-forming units (CFUs) were determined at multiple time points by serial dilution and plating on agar to assess bacterial killing.
In Vivo Data Analysis
Classifications of M. tuberculosis drug regimens as 'good or worse than SOC' or 'better than SOC' were sourced from mouse models and phase 2a/2b clinical trials. For M. abscessus, patient treatment outcomes (persisting vs. cleared infections) were assessed. Logistic regression and AUC-ROC were used to evaluate predictive performance of drug regimens.
Atypical Single-Cell Time-Kill (ASCT) Experimental Setup
ASCT uses a dual-layer approach: an agar pad immobilizing bacteria and a drug-containing solution. M. abscessus and M. tuberculosis isolates were prepared in growth or starvation conditions, centrifuged, and resuspended to achieve single-cell suspensions. The agar pad solution contained specific media components, propidium iodide (PI), and bacteria. Drug solutions were added to 1,536-well plates, followed by centrifugation and incubation before imaging.
ASCT Image Analyses
Image processing involved five steps: background control (BaSiC), cell segmentation (ilastik), cell classification (PI+ or PI-), drift correction, and custom-scripted cell tracking. Segmentation accuracy was validated, and PI classification accuracy was compared with manual annotations. Drift correction optimized tracking, linking individual objects across frames based on centroid distances, area changes, orientation, and convex hull area.
ASCT Data Analyses
Bacterial growth in ASCT was defined as a >3.1-fold increase in total object area during antibiotic exposure. Reproducibility of M. abscessus time-kill kinetics was assessed by coefficient of variation, and outliers were detected using principal component analysis. Live-cell fractions were normalized to 0-h extrapolated values. Overall antibiotic killing was quantified as the area under the time-kill curve.
Drug Tolerance Analyses
The relationship between M. abscessus drug tolerance phenotypes, growth rates, lag times, and MICs was assessed using Pearson and Spearman correlations. A Spearman correlation matrix was generated from time-kill curve data of clinical M. abscessus isolates, and principal component analysis visualized drug clustering.
Single-Cell Growth Assessment
A backward tracking approach was used to identify single cells capable of forming microcolonies after antibiotic washout. Objects 5–15 times the median single-cell size were categorized as microcolonies and tracked back to their originating single cells using homology indices based on object area and centroids.
Whole-Genome Sequencing
M. abscessus isolates underwent DNA extraction, library construction, and multiplexed paired-end sequencing. De novo genome assemblies were quality-checked. Sequence reads were mapped to the ATCC-19977 genome to identify Single Nucleotide Polymorphisms (SNPs), INDELs, and large deletions, which were then annotated. Maximum-likelihood phylogenetic trees were generated.
Heritability Estimations
Unitigs were extracted, and a similarity matrix from phylogenetic distances was used to correct for population structure. FaST linear mixed models estimated narrow-sense heritability (h2) for M. abscessus drug tolerance phenotypes, assessing the proportion of variance attributable to genetic variation.
Phenogenomic Analysis
Genome-wide association studies (GWAS) analyzed approximately 300,000 M. abscessus genetic variants (SNPs, INDELs, deletions) in relation to drug tolerance phenotypes. Linear mixed models accounted for population structure, and associations were quantified using the Wald test with a Bonferroni threshold.
GFP-Induction Experiments
M. abscessus ATCC-19977 was transformed to express GFP under a tetracycline-inducible promoter. After antibiotic treatment and washout, GFP expression was induced on the ASCT platform. Brightfield and fluorescence images (PI and GFP) were acquired to assess viability and recovery, with data visualized as hexbin density plots.
Drug Interaction Assessments
M. tuberculosis drug combinations were tested in pairwise checkerboards. Bacterial fitness (OD 600) was assessed, and dose-response curves were fitted. Bliss interaction scores were determined for pairwise and high-order combinations. Logistic regression models, using interaction and MIC parameters, predicted in vivo outcomes in mouse models.
Resazurin Assay
As an alternative to ASCT, resazurin reduction was measured in M. tuberculosis mc27000 exposed to drug regimens. Fluorescence (RFUs) was quantified over 14 days, and antibiotic killing (RFU change) was compared with ASCT-derived killing and in vivo outcomes.
Generation of MAB_0233 Knockout Mutant
MAB_0233 knockout mutants (ΔMAB_0233) were generated in M. abscessus ATCC-19977 using ORBIT technology. This involved transformation with specific plasmids (pKM444, pKM496), electroporation, and plating on selective media. Knockout presence was verified by PCR and Sanger sequencing.
Gene Complementation
The ΔMAB_0233 mutant was complemented by amplifying MAB_0233 and cloning it into the pMV261 vector under a constitutive promoter. Cured knockout strains were transformed, and gene expression in the complemented strain was verified by quantitative real-time PCR (qPCR) using housekeeping genes as controls.
Reporting Summary
Further details on the research design are available in the Nature Portfolio Reporting Summary.