High-content dual-live fluorescence imaging, determination of M. tuberculosis intrabacterial pH-GFP fluorescence ratio, and M. tuberculosis-associated LysoTracker intensity.

For high-content live-cell imaging, cells were infected with fluorescent mycobacteria producing ratiometric pH-GFP at an MOI of 1 as described above. After 24 h, the culture medium was replaced by fresh medium only or fresh medium containing 50 nM ConA (Sigma-Aldrich, C9705) for iPSDM and 100 nM ConA for MDM. After 24 h, infected cells were washed once with PBS buffer (pH 7.4) and stained with complete medium containing 200 nM LysoTracker Red DND-99 (Invitrogen, L7528) in a humidified 37°C incubator with 5% CO₂. After 30 min, staining medium was removed and replaced with fresh medium containing NucRed Live 647 ReadyProbes (Invitrogen, {"type":"entrez-nucleotide","attrs":{"text":"R37106","term_id":"794562","term_text":"R37106"}}R37106) following the manufacturer’s recommendations to facilitate cell detection. Live-cell imaging was further performed using the OPERA Phenix microscope with a 63× water-immersion objective. Image acquisition was performed with the confocal mode using the default autofocus function and a binning of 1. Mtb pH-GFP signal was detected using excitation wavelength (λ_ex) 405 nm/emission wavelength (λ_em) 500 to 550 nm and λ_ex 488 nm/λ_em 500 to 550 nm, LysoTracker signal was detected using λ_ex 561 nm/λ_em 570 to 630 nm, and NucRed Live signal was detected using λ_ex 640 nm/λ_em 650 to 760 nm. Laser power for all channels was set between 20% and 30% with an exposure time of 200 ms. Each channel was imaged independently, and a minimum of 3 to 4 distinct focal z-planes spaced with 0.5 to 1 μm was acquired. Multiple fields of view (323 μm by 323 μm) from each individual well were imaged with a set overlap of 10% in between fields. Segmentation and analysis were performed using the Harmony software (Perkin-Elmer, version 4.9). Briefly, cellular region was detected based on the fluorescent signals in the far-red emission channel using the “Find Image Region” building block and the “Absolute Threshold” function. Intracellular Mtb pH-GFP was detected based on the GFP signal obtained into both λ_ex 405 nm/λ_em 500 to 550 nm and λ_ex 488 nm/λ_em 500 to 550 nm channels using the “Find Image Region” building block and the “Absolute Threshold” function. Signals from the two GFP channels were merged using the “Calculate Image” building block and the function “By Formula,” where a channel A+B operation was applied. This combined image was filtered to reduce background noise using the “Filter Image” building block and a sliding parabola function. This M. tuberculosis mask was used to quantify Mtb pH-GFP mean fluorescent signal per single object from both 405-nm/510-nm and 488-nm/510-nm channels. Ratiometric signals were obtained by dividing the mean intensity quantified at λ_ex 405 nm/λ_em 500 to 550 nm by the one obtained at λ_ex 488 nm/λ_em 500 to 550 nm for each object. To quantify M. tuberculosis-associated LysoTracker intensity, the M. tuberculosis mask was slightly extended using the “Find Surrounding Region” building block using method A with an individual threshold value of 0.8 and conservation of the input region. When assessing the spatiotemporal mode of action of PZA, human macrophages were infected for 24 h and further treated with increasing concentrations of PZA ranging from 0 to 400 mg/L in the absence or presence of ConA for an additional 4, 16, 24, or 72 h before acquisition was performed as described above. When assessing anti-TB drug-mediated pH disruption, M. tuberculosis-infected cells were left untreated or pulsed with PZA (100 mg/L), BDQ (2.5 mg/L), INH (5 mg/L), or RIF (5 mg/L) for 24 h before imaging. Determination of absolute changes of pH-GFP ratio relative to the control condition (also referred as Δintrabacterial pH) was done by subtracting the value obtained under each experimental condition from its corresponding control condition. All the results were exported as CSV files, imported in the R Studio software (The R Project for Statistical Computing, version 1.3.1073), and most of the graphs, displayed as boxplots, scatterplots, or raincloud plots, were plotted with the ggplot2 package (version 3.3.2).