ABSTRACT Though very few new tuberculosis (TB) drugs have been introduced over the past forty years, the pivotal questions relevant to TB therapeutic development are well articulated across the field and are being pursued with great vigor in the biomedical community. Can we find a way to shorten treatment for from the current six months? Which novel mechanisms of action will support complete sterilization when combined with existing standard of care treatment? Are persister populations completely eliminated through a specific treatment regimen? In order to succeed in the clinic, first we must address these questions preclinically through the use of predictive animal models. Evaluation of the efficacy of experimental tuberculosis therapeutics is commonly performed using mouse models, due to the relatively small size and low cost of mice. However, many mouse models fail to reproducibly develop human-like pathology, specifically hypoxic necrotic granulomas, and thereby may have limited predictive value for certain compounds. Mouse models which do develop human-like pathology may suffer from substantial heterogeneity or complex protocols required to limit significant morbidity. Our recent work has overcome these significant limitations, and here we propose to refine, validate and characterize an innovative mouse model of TB that is more predictive of outcomes in patients. We have completed a series of studies in mice that strengthen our ability to advance a more predictive model of TB, one that consistently produces human-like pathology in the form of hypoxic necrotic granulomas. This new model builds on work that demonstrated the development of hypoxic granulomas in Nos2-/-mice, but adds significant value by simplifying the protocol and utilizing the natural aerosol route to infect mice with the partially attenuated Mycobacterium tuberculosis (Mtb) strain R1Rv. We have paired this novel model with fluorescent reporter strains that specifically mark drug and immune tolerant Mtb cells. When deployed together to support pharmacological evaluation of potential TB drug regimens, these two tools can provide unique insights into why some therapeutic approaches may work well to shorten chemotherapy in patients, while others will not do so. In this proposal, we seek to establish the predictive value of the model to determine drug efficacy responses, and to reveal new insights into the biology of drug-tolerant Mtb. In the first two Specific Aims we seek to optimize the novel mouse model and validate it as a more rigorous measure of in vivo anti-tubercular drug efficacy, and in the third Aim we employ the model to identify the niche occupied by drug tolerant cells, to better understand their potential vulnerabilities. Completion of the three Aims of this proposal will deliver a more predictive and well-characterized mouse model of Mtb infection that will have the potential to become a new standard for small animal testing of TB drugs.