Certain populations of Mycobacterium tuberculosis go undetected by standard diagnostics but can be enumerated using limiting dilution assays. These differentially detectable M. tuberculosis (DD M. tuberculosis) populations may have relevance for persistence due to their drug tolerance. It is unclear how well DD M. tuberculosis from patients is modeled by a recently developed in vitro model in which M. tuberculosis starved in phosphate-buffered saline is incubated with rifampin to produce DD M. tuberculosis (the PBS-RIF model). This study attempted to answer this question. We selected 14 genes that displayed differential expression in the PBS-RIF model and evaluated their expression in patient sputa containing various proportions of DD M. tuberculosis. The expression of 12/14 genes correlated with the relative abundance of DD M. tuberculosis in patient sputa. Culture filtrate (CF), which promotes recovery of DD M. tuberculosis from certain patient sputa, improved these correlations in most cases. The gene whose reduced expression relative to M. tuberculosis 16S rRNA showed the greatest association with the presence and relative abundance of DD M. tuberculosis in patient sputa, icl1, was recently shown to play a functional role in restraining DD M. tuberculosis formation in the PBS-RIF model. Expression of icl1, combined with two additional DD M. tuberculosis-related genes, showed strong performance for predicting the presence or absence of DD M. tuberculosis in patient sputa (receiver operating characteristic [ROC] area under the curve [AUC] = 0.88). Thus, the in vitro DD M. tuberculosis model developed by Saito et al. (K. Saito, T. Warrier, S. Somersan-Karakaya, L. Kaminski, et al., Proc Natl Acad Sci U S A 114:E4832-E4840, 2017, https://doi.org/10.1073/pnas.1705385114) bears a resemblance to DD M. tuberculosis found in tuberculosis (TB) patients, and DD M. tuberculosis transcriptional profiles may be useful for monitoring DD M. tuberculosis populations in patient sputum. IMPORTANCE Differentially detectable M. tuberculosis (DD M. tuberculosis), which is detectable by limiting dilution assays but not by CFU, is present and enriched for in TB patient sputum after initiation of first-line therapy. These cryptic cells may play a role in disease persistence due to their phenotypic tolerance to anti-TB drugs. A recently developed in vitro model of DD M. tuberculosis (the PBS-RIF model) has expanded our understanding of these cells, though how well it translates to DD M. tuberculosis in patients is currently unknown. To answer this question, we selected 14 genes that displayed differential expression in the PBS-RIF model and evaluated their expression in TB patient sputa. We found that 12/14 of these genes showed a similar expression profile in patient sputa that correlated with the relative abundance of DD M. tuberculosis. Further, the expression of three of these genes showed strong performance for predicting the presence or absence of DD M. tuberculosis in patient sputa. The use of DD M. tuberculosis transcriptional profiles may allow for easier monitoring of DD M. tuberculosis populations in patient sputum in comparison to limiting dilution assays.
Keywords: Mycobacterium tuberculosis; clinical methods; diagnostics; differentially detectable bacteria; molecular methods; multidrug resistance; persistence; viable but nonculturable.