Quantitative Approaches for Drug Testing in Chronic Toxoplasmosis: Leveraging New Insights into Bradyzoite Biology within Tissue Cysts in Vivo.

Toxoplasmic encephalitis (TE) is one of the HIV-AIDS defining opportunistic infections that is triggered by the reactivation of a chronic Toxoplasma infection in the brain. The chronic infection is mediated by bradyzoites which reside within tissue cysts. Dogma has held that bradyzoites are non-replicative metabolically inert entities. Historically, under this paradigm, the assessment of drug efficacy has been limited to tissue cyst number, typically presented as a percentage reduction in cyst burden. The actual impact at the level of individual bradyzoites has been largely ignored. Our recent work (Watts et al. MBio 2015) has directly challenged the dogma of bradyzoites being metabolically inert having demonstrated that tissue cysts are nonuniform and contain metabolically heterogeneous bradyzoites that are dynamic and capable of replication. Enumeration of bradyzoites within tissue cysts was made possible by our development of BradyCount 1.0, an imaging based application that established the intracyst bradyzoite burden by quantifying the number of nuclei within a defined optical section of an imaged tissue cyst. An additional marker for cell cycle progression TgIMC3 captured not only active cytokinesis but also provided a time-stamp to “birthdate” individual bradyzoites within the cyst. Given that replication is an energy intensive process, we examined the activity of mitochondria and capacity for stored glucose in amylopectin granules (AG). Consistent with the heterogeneity in replication potential and recency, mitochondrial activity and AG distribution are also found to be heterogeneous across encysted bradyzoites. Unlike nuclei, which are discrete round to ovoid structures, mitochondrial profiles and the AG distribution patterns have varied morphologies. This is also true of the profiles of individual bradyzoites and the developing progeny within them for which not only the shape, but also the signal intensity informs on the cell cycle status of individual bradyzoite. These aspects cannot be captured using BradyCount 1.0. For this reason we propose to develop BradyCount 2.0 which will have the added functionality of being able to detect and discriminate between a spectrum of morphological forms. We intend to incorporate trainable machine learning approaches driven by the range of observed phenotypes to develop an efficient screening platform to inform on these parameters at the level of individual bradyzoites within cysts. This advance will immediately improve the resolution and potential for mechanistic insights as each tissue cyst can yield between several hundred to thousands of individual data points. The collection of these functional physiological data will drive the development of a computational modeling approach, based on a Markov architecture which is ideal given the opportunistic nature of replication within encysted bradyzoites. We propose to test the validity of the imaging approaches (focused on cell cycle progression/replication, mitochondrial activity and stored energy reserves (AG)) singly and in combination by using established drugs known to have no impact on cyst number (though not necessarily intracyst burden as is the case for pyrimethamine and sulfadiazine) or a reduction in overall cyst burden (atovaquone and endochin-like quinolones which target mitochondrial function). Finally, the consequence of drug treatment will be modeled with the Markov models being refined using the empirically derived biological data. We believe that a combination of assessing drug effects on individual bradyzoites using quantifiable physiologically relevant metrics we can establish a new paradigm to robustly test new drugs in vivo. In doing so we will provide the framework to fill a significant gap in the targeted treatment of the chronic infection in immune suppressed.