Towards further understanding of the persister bacterial populations in sputum during tuberculosis therapy

Introduction

Two Mycobacterium tuberculosis (Mtb) putative persister phenotypes were explored: Mtb positive for Lipid Bodies (LB+) and those described as that Differentially Culturable (DCMtb), due to their dependence on Mtb culture supernatant (CSN) for growth. Both phenotypes are associated with antimicrobial tolerance, are detected sputum and have been linked to inferior treatment responses. The enumeration of DCMtb in sputum and the influence of Nitric Oxide (NO) on the frequency of LB positive bacilli were studied and results related to patient treatment responses.

Methods

For NO / LB studies, Mtb isolates from The Gambia were used. LB+ frequencies in the sputa from which the Mtb strains were isolated, were known. To achieve amenable inter-strain comparisons a method of growth on agar to sub-confluency was established. This enabled harvesting without producing baseline new gene expression and exposure to NO followed by assessment of LB positivity and transcriptional analyses. For DCMtb studies most probable number (with and without CSN) and colony forming unit (CFU) assays were performed. Two sample sets were investigated, one from the MARK-TB biobank and a second collected prospectively from RIFASHORT, a high dose rifampicin trial. Results were linked to treatment responses.

Results

Mtb strains were shown to express different LB+ frequencies at baseline and following NO exposure; frequencies correlated strongly ( r2 >0.9. p < 0.05) to those observed the cognate sputum samples. These stains also varied in their susceptibility to NO toxicity. In the MARK-TB study, high proportions of DCMtb in sputum samples taken at week 4 of treatment associated with unfavourable outcomes (OR 20.8). In RIFASHORT, reduction of DCMtb and CFUs were accelerated with high dose rifampicin compared to standard treatment.

Conclusion

Mtb strains from different lineages show distinct responses to NO with respect to LBs, gene expression and susceptibility. These features potentially explain observed associations between sputum LB positivity and treatment responses. DCMtb frequencies in sputum from treated individuals associated with outcome in the MARK-TB study and with rifampicin dosage in RIFASHORT. These studies open up new opportunities to better understand differing responses to TB treatment and to enable prediction of outcomes without prolonged follow up.