New molecular test reveals hidden drug interactions in TB treatment

Tuberculosis treatment relies on combinations of antibiotics, but choosing the best mixtures in preclinical studies is hard because standard measures give limited insight into how drugs interact. Colony forming units (CFU) count live bacteria but do not always predict how two or three drugs will work together. To address that gap, corresponding author Nicholas D. Walter and colleagues tested a new approach called SEARCH-TB, which uses targeted in vivo transcriptional profiling to read how Mycobacterium tuberculosis (Mtb) changes its physiology when exposed to drugs. The team focused on the widely discussed three-drug BPaL regimen (bedaquiline, pretomanid, linezolid) and broke it down to see what each drug does alone and in combination. They tested treatments in a BALB/c high-dose aerosol mouse infection model and sampled bacterial transcriptional responses after 2, 7, and 14 days. By looking at gene expression instead of just CFU, the researchers aimed to see whether one drug can change Mtb in ways that help or hurt the activity of a second drug, information that CFU counts alone can miss.

SEARCH-TB is a targeted in vivo transcriptional profiling tool that measures how drugs affect Mtb physiology inside infected animals. In this study, the team measured transcriptional responses to monotherapy, pairwise combinations, and the full 3-drug BPaL regimen at 2, 7, and 14 days. Monotherapy with each drug rapidly induced drug-specific Mtb transcriptional responses by day 2, and those responses continued to evolve over 14 days. In pairwise combinations, bedaquiline dominated the transcriptional signature when combined with either pretomanid or linezolid, meaning the bedaquiline-induced physiological changes in Mtb were most prominent. The pretomanid-linezolid pair produced a blended, intermediate transcriptional profile between the two drugs. When all three drugs were combined in BPaL, the addition of both pretomanid and linezolid to bedaquiline produced a greater transcriptional response than would have been predicted from the pairwise results. These findings show SEARCH-TB can detect complex, non-additive interactions among bedaquiline, pretomanid, and linezolid that CFU measures alone might miss.

The study demonstrates that drug-induced perturbations of Mtb physiology can be altered in unexpected ways when drugs are combined, and that transcriptional profiling provides a far more granular readout of those effects than CFU burden. SEARCH-TB proved capable of disaggregating the contributions of bedaquiline, pretomanid, and linezolid in vivo, revealing dominance, blending, and synergy patterns among the drugs. This proof of concept suggests that molecular measures of Mtb physiology could become a complementary pharmacodynamic marker in preclinical testing, helping scientists understand why some combinations succeed while others fail. If adopted alongside traditional measures, targeted in vivo transcriptional profiling like SEARCH-TB could guide regimen selection by revealing physiologic interactions early, potentially improving the design of safer, more effective TB drug combinations before costly clinical trials.