Immortal genome assumption significantly underestimates replication and death rates of Mycobacterium tuberculosis in mice and monkeys

Immune correlates of protection against infection with Mycobacterium tuberculosis (Mtb) or against tuberculosis (TB) remain poorly defined. The ratio of colony-forming units (CFUs) to chromosomal equivalents (CEQs), Z = CFUs/CEQs , recovered either from the whole lung (mice) or individual lesions (monkeys or rabbits) of Mtb-infected animals has been used as a metric for how effectively Mtb is killed in vivo . However, the contribution of bacterial killing to changes in the CFU/CEQ ratio during an infection has not been rigorously investigated. We developed alternative mathematical models to study the dynamics of CFUs, CEQs, and their ratio during an Mtb infection. We find that the ratio Z alone cannot be used to infer the death rate of bacteria, unless the dynamics of CEQs and CFUs are entirely uncoupled, which is biologically unreasonable and inconsistent with the view that CEQs reflect an accumulated burden of both viable and non-viable bacteria. Importantly, we estimate a decay rate of 3.6% per day (a half-life of about 20 days) of Mtb H37Rv CEQs in B6 mice that is similar to 4% per day, previously reported for Mtb Erdman in cynomolgus macaques. While the estimated Mtb DNA decay rate seems small, we found that estimated rates of Mtb replication and death/killing are still extremely sensitive even to slow decay of Mtb DNA, in part, because Mtb replication and death rates are also small, especially during chronic phases of infection. By applying our models to data on Mtb dynamics during the first 3 weeks of infection in macaques, we provide evidence of substantial killing of Mtb bacteria, prior to the arrival of adaptive immunity to the site of infection, challenging the previously established notion of non-dying bacteria in the absence of T-cell immunity and granuloma formation. We also propose experiments that will allow more accurate measurement of the rate of Mtb DNA loss, helping more rigorously quantify the impact of immunity on within-host Mtb dynamics. IMPORTANCE Mycobacterium tuberculosis (Mtb) is responsible for an estimated 1.23 million deaths annually. Vaccine development against Mtb is challenging, in part due to a lack of reliable indicators of immune protection. To quantify bacteria killing by the immune system, counts of Mtb genomes isolated from tissues from infected animals have been used to estimate a total population of live plus dead bacteria, on the grounds that Mtb genome decay is negligible after bacteria death. We estimate that Mtb genomes decay in the lungs of mice at a rate of 3.6% per day, similar to a previous estimate in monkeys. Using mathematical modeling, we show that this decay rate is sufficient to cause substantial underestimation of both Mtb replication and death rates in vivo . We reanalyze data from untreated animals and argue that substantially more bacteria replication and death occurs in infected mice and monkeys than was previously estimated.