Uncomplicated drug-sensitive tuberculosis (TB) must be treated with up to four drugs to prevent emergence of drug resistance. Yet, resistance is on the rise, with multi-drug and extensively-drug resistant cases increasing globally. Poor compliance due to a large pill burden and lengthy therapy has been largely blamed for the escalation of drug-resistant TB. While this is probably a contributing factor, it is likely that fators intrinsic to the host and pathogen also drive the emergence of drug resistance. Our proposed research will investigate the contribution of the following four factors to the development of drug resistance at the level of the TB lesion: (i) clinical Mycobacterium tuberculosis (Mtb) strains wit mutations that produce small increases in MIC which are well below MIC cutoffs that usually identify drug resistance;(ii) inter-patient variability in drug exposure;(iii) heterogeneous drug distribution within lesion compartments;and (iv) drug induced mutagenesis. We propose to determine how these factors interact to create de facto monotherapy in selected niches in vitro and in the rabbit model of active cavitary TB. In vitro, we will establish the routes by which Mtb strains evolve from fully susceptible to fully resistant following a stepwise decline of susceptibility, with Mtb strains that have defined mutations likely to emulate several different clinical scenarios of drug-resistance risk. Surprisingly, the potential of fluoroquinolone-mediated DNA damage to increase the frequency of resistance-conferring mutations has not been explored for Mtb, while it is well-characterize for other bacterial species. Since fluoroquinolones are being considered as the new backbone for TB therapy, both for drug susceptible and drug resistant disease, we will fill this gap and translate our in vitro findings in the rabbit model, uing single drug treatment and combination therapy. We will use an optimized rabbit model of chronic active TB which recapitulates the major pathological features of human TB: solid cellular granulomas, necrotic fibrotic lesions and cavities. The model is ideally suited to query lesion-centric read-outs of microbiology, genetic resistance, drug penetration, and strain-specific polymorphisms. We will focus our studies on the three-drug combination currently viewed as among the most promising in clinical development for drug-sensitive and drug-resistant TB. This regime combines the fluoroquinolone moxifloxacin, pyrazinamide and the nitroimidazole PA-824. Using pharmacometrics approaches, we will build a pharmacokinetic-pharmacodynamic model that integrates small MIC increases, plasma and lesion PK variability, drug-induced mutagenesis and spatial drug distribution in relation to emergence of resistance. By predicting the risk of de facto monotherapy, the model will inform in silico predictions of safe individualize human combinations that prevent acquisition of drug resistance during treatment of active TB. It is our intent that future use of this approach will shift the paradigm from empirical to rational design of drug/dosing regimens with decreased risk for the development of drug resistance.