Characterisation of bedaquiline resistance mediated by the <i>M. tuberculosis</i> MmpS5L5 efflux pump

Abstract

Until recently, the treatment of drug-resistant tuberculosis required prolonged antibiotic treatment, often up to two years, with a toxic combination of less effective drugs. Bedaquiline, an antibiotic that targets the mycobacterial ATP synthase, has revitalised the ability to treat multidrug-resistant tuberculosis as part of new all-oral 6-month antibiotic regimens. Unfortunately, but predictably, clinical resistance to bedaquiline is already widespread, threatening the utility of this life-saving drug. Bedaquiline resistance is commonly caused by mutations in Rv0678, a transcriptional repressor of the MmpS5-MmpL5 efflux pump. MmpS5L5 can efflux bedaquiline and a spectrum of anti-TB drugs in clinical development, making this mode of drug resistance highly consequential. Despite its importance, little is known about the pump’s structure and substrate recognition. This thesis aims to characterise the MmpS5-MmpL5 efflux pump using structural and genetic approaches, focusing on two key questions: whether MmpS5 and MmpL5 form a trimeric efflux pump assembly, and how MmpL5 recognises bedaquiline and other chemically diverse substrates. In Chapter 3, I describe progress towards purification of MmpS5-L5 for structural analysis. High-resolution clear native PAGE and size exclusion chromatography showed that purified MmpL5 forms a detergent-labile oligomer in vitro. Using sequential affinity purification, I show that MmpS5 and MmpL5 can co-purify in vitro. Screening thermostable homologs of MmpL5 identified a homolog that can efflux bedaquiline, produces higher levels of recombinant expression and is more stable in vitro. Finally, binding analysis using surface plasmon resonance (SPR) revealed that both bedaquiline and TBAJ-587 bind to purified MmpL5 with a dissociation constant (KD) of approximately 2 µM. In Chapter 4, I leverage recent advances in protein structure prediction to produce a high-confidence model of trimeric MmpS5-L5. This model aligns well with both the experimental data from Chapter 3 and photobleaching data from the literature. Several features of this model are notable, including how MmpS5 wraps around MmpL5, contacting each protomer. The coiled-coil domain of MmpL5 is an ~18 nm alpha-helical barrel with a 7–9 Å lumen that is lined with methionine residues. Based on this structure, I propose that substrates are transported across the periplasm via this tube. Sequence conservation at the tip of the coiled-coil domain hints at the possibility of an interaction with an unknown outer membrane protein. In Chapter 5, I establish a genetic system to test the function of the MmpS5L5 complex in Mycobacterium marinum and Mycobacterium tuberculosis. Using this system, I show that MmpS5L5 is a multidrug efflux pump able to transport bedaquiline, clofazimine, PBTZ-169, TBAJ-587 and OPC-167832. This system is used to test key aspects of the trimeric model introduced in Chapter 4 including the requirement for “proton-relay” residues and the requirement of the coiled-coil domain for efflux. The paralog MmpS4L4, whilst having high sequence similarity to MmpS5L5, cannot efflux antibiotics. Using this observation, MmpL5 is mutated and residues that are important for antibiotic efflux are identified. Genetic suppressor selections to restore function of these mutants identified a suppressor mutation that enhances the activity of MmpL5. Further selections using error-prone PCR libraries of MmpS5L5 identified two additional activity variants that increase the efflux of bedaquiline. Three independent methods converge on a region that likely represents the site of bedaquiline recognition by MmpL5. In Chapter 6, I discuss the implications of this work on anti-TB therapy and recommend that genotypic bedaquiline resistance sequencing be revised to include MmpS5L5. Informed by the observations in this thesis, strategies to inhibit MmpL5 and modifications to bedaquiline that might circumvent efflux are explored. MmpS5L5 represents a novel type of efflux system, and this discovery is placed in the context of currently known efflux systems. The chapter concludes by discussing unanswered questions regarding the transport mechanism, including kinetics, functional rotation, and substrate movement through the coiled-coil domain, and proposes experimental strategies to address these questions.