Toxoplasma gondii is an obligate intracellular protozoan parasite that causes opportunistic disease, particularly in immunocompromised individuals. Central to its transmission and pathogenesis is the ability of the proliferative stage (tachyzoite) to convert into latent tissue cysts (bradyzoites). Encystment allows Toxoplasma to persist in the host and affords the parasite a unique opportunity to spread to new hosts without proceeding through its sexual stage, which is restricted to felids. Bradyzoite tissue cysts can cause reactivated toxoplasmosis if host immunity becomes impaired. A greater understanding of the molecular mechanisms orchestrating bradyzoite development is needed to better manage the disease. Here, we will review key studies that have contributed to our knowledge about this persistent form of the parasite and how to study it, with a focus on how cellular stress can signal for the reprogramming of gene expression needed during bradyzoite development.
Toxoplasma gondii is one of the most successful parasites on Earth. Despite its obligate intracellular lifestyle, this protozoan parasite has remarkable transmissibility and has permanently infected most species of mammals and birds around the world. As we will discuss, the ability of Toxoplasma to encyst inside the cells of host tissues and reconvert back into its proliferative stage was an evolutionary paradigm shift, circumventing the need for the parasite to undergo its sexual stage to be transmitted to a new host. This unusual property affords Toxoplasma a second major route of transmission: the capacity to disseminate clonally through intermediate hosts.
Toxoplasma belongs to phylum Apicomplexa, which contains many other protozoan pathogens of human and veterinary importance, such as Plasmodium spp. (malaria), Cryptosporidium spp. (cryptosporidiosis), and Eimeria spp. (poultry coccidiosis). Toxoplasmosis is a notorious opportunistic disease in patients with AIDS and other immunocompromised individuals which most commonly results from reactivation of a previous infection (Luft et al., ). Ocular infection by Toxoplasma is a major cause of retinochoroiditis in both immunocompromised and immunocompetent individuals (Wallace & Stanford, ). Additionally, Toxoplasma can cross the placental barrier and cause abortion or congenital birth defects if the mother becomes infected for the first time shortly before or during pregnancy (Jones et al., ). In utero infection is also associated with an elevated risk of ocular toxoplasmosis owing to spontaneous reactivation of disease (Wallace & Stanford, ).
Toxoplasma has a complex life cycle consisting of multiple stages that fluctuate between proliferative and latent stages (Fig. 0001). The proliferative stage, known as the tachyzoite, replicates exponentially by endodyogeny (asexual division whereby two daughter cells form within a single mother cell) inside of host cells with a doubling time of c. 7 h (Radke & White, ). The intracellular tachyzoites, housed in a membrane-bound compartment called the parasitophorous vacuole (PV), can convert into virtually quiescent forms known as bradyzoites, which transform the parasitophorous vacuole membrane (PVM) into a cyst wall in the process. Bradyzoite cysts remain infectious and can form in skeletal, heart, and CNS tissues, granting Toxoplasma the ability to spread to a new host following predation of its former host. Felines serve as the definitive host for Toxoplasma, whose intestinal milieu is the only known environment suitable for the sexual stage of the parasite life cycle. Owing to factors that have yet to be identified, the cat gut induces bradyzoites to differentiate into macrogametes and flagellated microgametes that fuse and lead to the formation of oocysts. Dissemination of the stable, highly infectious oocysts into the environment provides another major route of transmission through the food and water chain (Fig. 0001). While it is not possible to distinguish infection from meat vs. oocysts, it has been proposed that the surge of infection in teenagers and low prevalence in younger children indicate that transmission through cysts in undercooked meat is important (Dubey & Jones, ). As a consequence of these multiple routes of transmission, Toxoplasma has made itself at home in at least one in three people. Seroprevalence varies widely between geographical regions, but estimates suggest that between 30% and 65% of humans worldwide are infected (Tenter et al., ).
The ability to differentiate into bradyzoites and form these impenetrable cysts makes it currently impossible to eradicate Toxoplasma from the host. While several drugs are available which control acute toxoplasmosis, such as pyrimethamine plus sulfadiazine, no short-term treatment exists which can eliminate the cysts, which also appear impervious to the immune response. The presence of latent bradyzoite cysts makes Toxoplasma a chronic infection, and the ability of bradyzoites to reconvert into rapidly growing tachyzoites explains the high frequencies of acute toxoplasmosis often observed in immunocompromised individuals (Wong & Remington, ; Montoya & Liesenfeld, ). Toxoplasmic encephalitis, characterized by intracerebral lesions, is a significant CNS complication in patients with AIDS (Fig. 0002).
Microbial persistence and latency are conserved strategies that numerous pathogens use to their advantage. The ability of Toxoplasma to form latent cysts is an evolutionary trade-off that reduces virulence while increasing transmission – a win–win situation for the parasite. In terms of human and veterinary health, understanding bradyzoite conversion is the key to minimizing transmission and managing chronic infection. In this review, we will discuss bradyzoite development and physiology, with a focus on the molecular mechanisms orchestrating differentiation. We will also highlight the methods used to study bradyzoite cysts in vitro and in vivo and present important questions that require further investigation.
Biology of the bradyzoite cyst
Morphological features of bradyzoites and cysts
Bradyzoite tissue cysts develop and remain inside nucleated cells, exhibiting tropism similar to tachyzoites, but are more frequently found in brain, eye, skeletal muscle, and cardiac tissue. Within the brain, bradyzoite cysts appear to form in a wide variety of regions, although a subtle tropism for the medial and basolateral amygdala has been reported (Vyas et al., ). Mature tissue cysts can be detected within 7–10 days postinfection, but hallmarks of bradyzoite differentiation can be detected within 3 days postinduction in vitro. Recent studies using bioluminescence imaging technology also show that conversion to bradyzoites begins rapidly in vivo, as early as 1 day postinfection (Di Cristina et al., ). Tissue cysts vary in size and shape. Young cysts can be as small as 5 μm in diameter with as few as two bradyzoites, while mature cysts can be up to 70 μm with c. 1000 bradyzoites (Dubey et al., ) (Fig. 0003a). Individual bradyzoites (7 × 1.5 μm) are slender relative to tachyzoites (6 × 2 μm), with a more pronounced bow or crescent shape (Mehlhorn & Frenkel, ). Also different from tachyzoites, the nucleus in mature bradyzoites resides at the parasite's posterior, rhoptry organelles are electron dense, and there is a marked increase in micronemes and amylopectin (glucose storage) granules (Dubey et al., ) (Fig. 0003b).
The bradyzoite PVM/cyst wall is < 0.5 μm thick and elastic to accommodate expansion. The cyst wall is built from chitin and glycoproteins secreted by the parasites (Boothroyd et al., ; Zhang et al., ). The sugars in the cyst wall are capable of binding lectins concanavalin A, wheat germ agglutinin, soy bean agglutinin, and Dolichos biflorus agglutinin (Sethi et al., ), the latter of which is commonly used as a diagnostic tool for cyst wall staining (Fig. 0004). In mature cysts, a matrix material fills the PV space between bradyzoites (Ferguson & Hutchison, ). The number of cyst wall/matrix proteins that have been characterized to date is limited. CST1 is a 116-kDa cyst wall glycoprotein that is the major binding constituent of Dolichos lectin (Zhang et al., ). MAG1 is a 65-kDa protein that localizes to the cyst wall and matrix (Parmley et al., ). At least one dense granule protein, GRA5, has been reported to be highly concentrated in the cyst wall; less intense staining of GRA1, GRA3, and GRA6 can also be observed (Lane et al., ; Ferguson, ). Upon ingestion by a host, the cyst wall is degraded by acid-pepsin digestion, releasing the bradyzoites. Unlike tachyzoites, bradyzoites are resistant to proteolytic enzymes (Jacobs et al., ), a key feature that assures survival in the host stomach. The bradyzoites proceed to invade intestinal epithelial cells to begin infection and dissemination throughout the host.
Studies have demonstrated that mature bradyzoites isolated from the brains of mice infected with Toxoplasma oocysts represent a growth-arrested stage with parasites suspended in G0 with uniform 1N DNA content (Radke et al., ). To provide insights into the dynamics of early stage conversion, an in vitro system using type II (Prugniaud) parasites subjected to CO2 deprivation was developed to maintain bradyzoites for up to 2 weeks. Using this system, it was observed that immature bradyzoites replicate slowly (requiring > 12 h compared to the c. 7 h for tachyzoites) and asynchronously, using a combination of endodyogeny and endopolygeny (asexual replication whereby more than two daughter cells form within the mother cell) (Dzierszinski et al., ). Interestingly, within the context of this system, 10–20% of developing bradyzoite vacuoles contained individual parasites lacking an apicoplast, a nonphotosynthetic plastid-like organelle acquired from an algal endosymbiont (Dzierszinski et al., ; McFadden, ). Despite their reputation for quiescence, the developing bradyzoites generated under these conditions were highly motile and capable of exiting a host cell without destroying it to invade a neighboring cell; bradyzoites within mature cysts purified from mouse brains also exhibit intravacuolar motility (Dzierszinski et al., ). Bradyzoite motility within and between host cells may explain why increases in cyst burden are observed in chronically infected mice despite the low frequency of cyst rupture.
Metabolism in the latent stage
Changes in parasite metabolism accompany the switch to a latent lifestyle. A number of metabolic enzymes have tachyzoite- and bradyzoite-specific isoforms (e.g. ENO2/ENO1 and LDH1/LDH2), suggesting a fine tuning of metabolism between these two life cycle stages. The abundance of polysaccharide in the form of amylopectin granules in the bradyzoites reflects a major shift in carbohydrate metabolism. This idea is supported by biochemical analyses which concluded that bradyzoites lack a functional TCA cycle and respiratory chain, suggesting a predominant role for anaerobic glycolysis during this stage (Denton et al., ). In contrast, tachyzoites likely use both mitochondrial oxidative phosphorylation and glycolysis to generate ATP. Pyruvate kinase and lactate dehydrogenase activities are markedly higher in bradyzoites, suggesting that lactate production is particularly important during latency (Denton et al., ). A bradyzoite-specific isoform of lactate dehydrogenase (LDH2) has been identified which is likely to be specialized for the latent stage (Yang & Parmley, ). LDH2 is resistant to acidic pH and would continue to function during the acidification that would occur during the catabolism of amylopectin to lactate in bradyzoites. Knockdown of LDH2 resulted in parasites that were unable to establish significant cyst burdens in the brains of infected mice (Al-Anouti et al., ).
Two other bradyzoite-specific isoforms of key glycolytic enzymes have been identified: glucose-6-phosphate isomerase (G6-PI) and enolase-1 (ENO1) (Dzierszinski et al., ). The enolase isoforms display distinct enzymatic properties and differ in their stability. The tachyzoite-specific enolase (ENO2) exhibits a similar Km value for the 2-phosphoglycerate substrate but has a threefold higher specific activity at Vmax compared to ENO1 (Dzierszinski et al., ). Considered together, these studies are consistent with the idea that stage-specific enzymes exist which are designed to tune glycolysis to accommodate proliferation or dormancy. Alternatively, or in addition to tailoring glycolysis, the enolases may play a role in transcriptional regulation. ENO1 and ENO2 have been found not only in the cytoplasm but also in the parasite nucleus of bradyzoites and tachyzoites (Ferguson et al., ). An alternatively translated form of human ENO1 also localizes to the nucleus and can repress transcription by binding to the c-myc promoter (Feo et al., ).
A number of enzymes with roles in the metabolism of oxygen radicals appear to be upregulated in bradyzoites, suggesting that the encysted forms are equipped to deal with long-term exposure to reactive metabolites (Manger et al., ). Consistent with this idea, it has been reported that mRNAs encoding various DNA repair enzymes increase in bradyzoites (Manger et al., ; Yahiaoui et al., ).
Triggers of bradyzoite formation
Three predominating strains of Toxoplasma (designated type I, II, and III) have been documented in North America and Europe which differ in growth rate, virulence in mice, and ability to form cysts (Howe & Sibley, ). Genetic polymorphism analyses indicate that these clonal lineages emerged after a single genetic cross, and their swift expansion resulted from the acquisition of direct oral infectivity (Su et al., ). It is estimated that this genetic cross occurred c. 10 000 years ago, coincident with the establishment of agriculture and domestication of animals such as the cat in the Fertile Crescent. Oral infectivity is a critical factor linked with the ability of some strains to efficiently differentiate into bradyzoites (Fux et al., ).
Type I strains grow rapidly and are hypervirulent in mice (LD50 – c. 1 parasite). The commonly used type I laboratory strain is designated RH, after the initials of the boy from which it was isolated in 1939 (Sabin, ). RH strain is generally considered to have lost the ability to develop into mature cysts at high frequency, perhaps as a consequence of prolonged propagation in vitro. To what degree RH strain has lost this developmental competency is a matter of debate and seems to vary between individual laboratory strains. During stress, type I strains including RH will activate gene expression of some bradyzoite marker genes, and in some cases, bradyzoite surface antigens and cyst wall proteins have been detected (Soete et al., ; Bohne & Roos, ; Lescault et al., ). Isolated reports claim RH parasites can form cysts in mice, via treatment with atovaquone and pyrrolidine dithiocarbamate (Djurkovic-Djakovic et al., ) or prior vaccination with soluble tachyzoite antigen and IL-12, although the bradyzoites formed by RH differ ultrastructurally and in their sensitivity to pepsin–HCl (Yap et al., ). In addition, RH strain parasites attenuated in virulence owing to disruption of dense granule gene GRA2 are capable of establishing chronic infection in mice (Mercier et al., ). Other type I strains that have not been subject to extensive in vitro culture, such as GT1, are able to form normal cyst walls in vitro during stress (Khan et al., ). Interestingly, it has been suggested that the phenotypic differences between type I strains are unlikely to be owing to sequence variation (Khan et al., ). Indeed, both epigenetic and translational control mechanisms have been linked with stage conversion.
Type II (e.g. Prugniuad or Pru, ME49) and type III strains (e.g. VEG) have lower replication rates and readily form cysts in vitro and in vivo; consequently, they are hypovirulent in mice (LD50 – c. 104). The most commonly isolated strain from clinical toxoplasmosis is type II (Howe et al., ), although type I strains are typically responsible for acute outbreaks. The genomes of representatives from all three lineages have been sequenced (data are available at ToxoDB.org), but which genes account for the differences reported in these strains have only begun to be resolved.
In terms of studying bradyzoite differentiation, strain choice is a crucial factor. While type I strains, particularly RH, grow faster and are easier to genetically manipulate, they are limiting in the study of cyst formation. However, several groups have made important discoveries regarding bradyzoite development using type I RH strain. There are also more transgenic parasites made in the RH background that facilitate gene tagging/disruption, or conditional gene expression (Meissner et al., ; Fox et al., ; Huynh & Carruthers, ). Type II and III strains form mature cysts readily but grow slowly and are more difficult to manipulate. Consequently, there are presently fewer tools available to study parasite physiology in the hypovirulent strains.
In vitro stresses induce stage conversion
It is well documented that conversion to the latent stage is a stress-mediated response, coupled with a slowing of the parasite cell cycle. One of the most commonly used in vitro methods to prompt bradyzoite differentiation is alkaline pH 8.0–8.2 (Soete et al., ). A wide variety of other stress agents have since been reported, including sodium nitroprusside, which acts as a source of exogenous nitric oxide (NO) and also inhibits proteins involved in the parasite mitochondrial respiratory chain (Bohne et al., ). Similarly, drugs that interfere with the parasite mitochondria also induce differentiation to bradyzoites (Bohne et al., ; Tomavo & Boothroyd, ). Heat shock and treatment with sodium arsenite also trigger the expression of bradyzoite antigens (Soete et al., ). Nutrient deprivation is a potent inducer of bradyzoite formation and can be achieved through arginine starvation (Fox et al., ), axenic incubation (Yahiaoui et al., ), or pyrimidine depletion in uracil phosphoribosyltransferase (UPRT)-deficient parasites subjected to ambient (0.03%) CO2 (Bohne & Roos, ; Dzierszinski et al., ). More recently, insults that cause endoplasmic reticulum (ER) stress or that interfere with calcium-induced egress have also been found to induce bradyzoite cyst formation (Nagamune et al., ; Narasimhan et al., ). Treatment with interferon gamma (IFN-γ) does not induce conversion to bradyzoites cultured in human fibroblasts (Soete et al., ) but will do so in murine macrophages, presumably owing to the stimulated releases of NO (Bohne et al., ). Long-term culturing of cysts was accomplished in vitro using murine astrocytes and intermittent inclusion of IFN-γ in the culture media (Jones et al., ). Methods used to convert tachyzoites to bradyzoites in vitro are summarized in Box .
Acidic or alkaline pH
Heat shock (43 °C)
Pyrimidine deprivation (∆UPRT parasites, 0.03% CO2)
Fluridone (disruption of ABA-mediated calcium signaling)
Whether these stress treatments act on the parasite directly (while they are extracellular), and/or if they act indirectly on intracellular parasites by stressing the host cell, is unclear. Weiss et al. () have shown that when extracellular parasites are exposed to alkaline stress for 1 h and allowed to reinfect host cells, bradyzoite differentiation is observed, albeit at a lower frequency than stressed intracellular parasites. Additionally, extracellular tachyzoites deprived of host cells for 12 h converted to bradyzoites upon reinfection of host cells (Yahiaoui et al., ). These studies suggest that extracellular parasites not only sense their environment but are marked in some way to initiate the bradyzoite development program upon reentry into a host cell. Epigenetic-mediated gene regulation offers a mechanism by which this form of ‘short term memory’ of the parasite's environment could occur.
Conversion to bradyzoite cysts may be driven by physiological factors other than exogenous stresses. Tachyzoites can spontaneously convert into bradyzoites, and low MOIs and/or frequent removal of egressed tachyzoites from cultures of infected cells enriches for cysts (McHugh et al., ). The proclivity toward spontaneous differentiation is influenced by the type of parasite strain and host cell background (Ferreira da Silva et al., ), and cysts are more frequently detected in differentiated host cells that are long-lived (Dubey et al., ). Further evidence that the host cell environment is a determinant in parasite differentiation comes from studies of a trisubstituted pyrrole known as Compound 1 (Donald et al., ). Compound 1 was shown to act directly on human host cells to slow tachyzoite replication and induce bradyzoite-specific gene expression in type II and III strain parasites (Radke et al., ). In these studies, human cell division autoantigen-1 (CDA1) was associated with promoting parasite differentiation, as tachyzoites infecting host cells overexpressing CDA1 underwent bradyzoite conversion (Radke et al., ). Prestressing human foreskin fibroblast host cells prior to infection can also stimulate bradyzoite formation (Radke et al., ). Collectively, these studies argue that tachyzoites are capable of assessing not only the type of host cell they invade, but also whether those host cells are under strain. These studies also suggest that the trigger(s) to differentiate are complex and multifactorial, consisting of both endogenous and exogenous factors.
In vivo factors relevant to the maintenance of bradyzoite cysts
The primary host response controlling Toxoplasma infection is mediated by CD8+ T cells with synergistic action from CD4+ T cells (Gazzinelli et al., , ). While CD8+ T cells have been shown to be cytotoxic to Toxoplasma-infected peritoneal macrophages (Kasper et al., ), their production of IFN-γ likely has a greater impact on controlling infection. IFN-γ is well documented as a key cytokine in the host immune response to Toxoplasma infection that limits intracellular replication of the parasite. Infected mice that are immunodepleted of IFN-γ succumb to toxoplasmosis instead of developing chronic infection (Suzuki et al., ). Similarly, administration of IFN-γ protects against lethal infection (McCabe et al., ). With regard to the mechanism of IFN-γ, it has been reported that in human brain microvascular endothelial cells, IFN-γ induces the tryptophan-degrading enzyme indoleamine 2,3-dioxygenase (Daubener et al., ), which in turn would starve parasites of tryptophan (Pfefferkorn et al., ). Starvation for key amino acids like tryptophan is likely to slow parasite growth and trigger cyst formation. Further support for this idea comes from studies demonstrating that arginine starvation induces bradyzoite differentiation (Fox et al., ). IFN-γ also induces an oxidative burst and the production of NO, which may act as a direct trigger for bradyzoite differentiation (Bohne et al., ). IFN-γ-inducible immunity-related GTPases (IRG proteins/p47 GTPases), such as IGTP and LRG-47, have also been linked to reduced parasite viability in activated macrophages (Butcher et al., ). Most recently, the IRG protein Irga6 (IIGP1), which participates in the disruption of the PVM, is phosphorylated and inactivated by the type I ROP18 kinase as an immune evasion strategy that promotes virulence (Fentress et al., ; Steinfeldt et al., ). In short, numerous mechanisms exist which account for how IFN-γ controls tachyzoite proliferation in vivo, but whether IFN-γ acts directly on parasites to induce or maintain bradyzoites in cyst forms is less clear.
Other cytokines have been linked to promoting bradyzoite differentiation. In murine macrophages, tumor necrosis factor (TNF) α synergizes with IFN-γ to facilitate conversion to bradyzoites (Bohne et al., ). Another study reported that the proinflammatory interleukin IL-6 favors the formation of bradyzoites in vitro in human fibroblasts (Weiss et al., ). Given that heat shock is a potent in vitro trigger for bradyzoite differentiation, it is conceivable that fever could contribute to stage conversion during acute toxoplasmosis in vivo.
As discussed in the preceding section, host cell background appears to be a key factor influencing bradyzoite formation. Bradyzoite cysts have a clear predilection to form in brain (specifically in neurons, astrocytes, and microglia) and muscle tissue rather than lungs, liver, spleen, or kidneys (Dubey et al., ; Luder et al., ). Cysts have been found in several regions of the brain, including the cerebral cortex, superior and inferior colliculus, cerebellum, olfactory bulbs, and medulla oblongata, and have also been found in the spinal cord (Di Cristina et al., ). It remains to be determined whether there is something special about the intracellular environments of these host cells that favor spontaneous conversion to latent cysts in addition to being immunologically privileged sites.
The pathogenesis of Toxoplasma infection also relies in part on host genetics. Studies in humans and mice have revealed varying abilities to control infection and cyst burden (Brown et al., ; Suzuki et al., ; Mack et al., ; Suzuki, ). For example, BALB/c mice are genetically resistant to Toxoplasma infection, harboring fewer brain cysts compared to other infected mouse strains. It was recently determined that BALB/c CD8+ T cells can eliminate brain cysts through their perforin-mediated activity (Suzuki et al., ). This study provides evidence that certain cytotoxic T cells harbor a capacity to recognize and destroy intracellular brain cysts, which have been previously believed to be impervious to host immunity.
Reactivation of acute infection
In vitro generated bradyzoites will quickly revert to proliferative tachyzoites upon removal of the stress agent used to differentiate the parasites. These studies support the idea that cellular stress is a key factor not only in prompting the development of bradyzoites, but also in maintaining the encysted form. In immunocompromised patients, reactivation of Toxoplasma infection is typically seen after the CD4+ T-cell count drops below 100–200 cells per mm3 (Pereira-Chioccola et al., ) (Fig. 0002). Ferguson et al. have reported that on rare occasion, bradyzoite cysts will rupture in immunocompetent mice, leading to rapid recruitment of inflammatory cells to the site (Ferguson et al., ). In the absence of a normal immune response to induce the newly released tachyzoites to convert into bradyzoites, the tachyzoites will continue to replicate and disseminate throughout the host, leading to serious complications and possibly death. Mouse models for reactivated toxoplasmosis have been developed which further underscore the relevance of IFN-γ in controlling latency (Suzuki & Joh, ; Dunay et al., ). The steroid immunosuppressant dexamethasone has also been used to reactivate toxoplasmosis in chronically infected mice (Nicoll et al., ; Djurkovic-Djakovic & Milenkovic, ). It is not currently known whether host immunity directly prevents bradyzoites from reactivating and/or functions to quickly kill or trigger the differentiation of reactivated tachyzoites escaping cysts.
Role of stress signaling in bradyzoite development and maintenance
As described earlier, virtually all types of stress can induce cyst development in vitro. Common denominators between the stresses that induce conversion to bradyzoites include the slowing of parasite proliferation, the induction of heat shock proteins, and translational control mediated by eIF2α phosphorylation (Fig. 0005).
Bradyzoite development and the parasite cell cycle
Cell cycle blocks do not trigger bradyzoite differentiation, demonstrating that cell cycle progression is required for cyst formation (Gubbels et al., ). Virtually all of the stress conditions that promote bradyzoite differentiation mentioned above reduce the proliferation of tachyzoites. Slowing of the parasite cell cycle has been linked to the initiation of the bradyzoite developmental program from a late-S/G2 subpopulation containing 1.8–2N DNA content (Jerome et al., ; Radke et al., ). During bradyzoite differentiation, these parasites proceed through M phase and then arrest in G1/G0 with uniform 1N DNA content. The unique late-S/G2 stage represents a premitotic cell cycle checkpoint for the commitment to bradyzoite formation and growth arrest following mitosis. The identification of cyclic expression of several bradyzoite-specific mRNAs, which exhibit peak expression in the late mitotic period, lends support to this model (Behnke et al., ). Curiously, one of the mRNAs with peak expression in tachyzoite cytokinesis is also significantly increased during bradyzoite differentiation; this mRNA encodes AP2VIIa-1, a likely transcription factor. It is enticing to speculate that AP2VIIa-1 is a master regulator that coordinates changes in the transcriptome that lead to bradyzoite conversion. The nature of the signaling mechanisms that accompany slowed growth and ultimately lead to bradyzoite differentiation has yet to be fully elucidated; the following sections detail what has been determined to date.
Heat shock proteins and signaling pathways
One of the earliest described bradyzoite-specific genes is Hsp30/BAG1, related to the small heat shock proteins found in plants (Bohne et al., ; Parmley et al., ). BAG1 appears c. 2–3 days after bradyzoite induction, but its relevance in differentiation was clouded when disruption of the gene in a type II strain failed to block tissue cyst formation (Bohne et al., ). A similar study by Zhang et al. () confirmed that BAG1 knockouts still form cysts in mice but do so at a significantly lower frequency. Weiss et al. () have demonstrated that Hsp70 is induced during alkaline-mediated bradyzoite differentiation, and that an inhibitor of Hsp90, Hsp70, and Hsp27 synthesis can suppress bradyzoite development in vitro. Another study also implicated a role of Hsp70 during the reactivation of chronic toxoplasmosis in vivo (Silva et al., ). Hsp90, in complex with p23 co-chaperone, has also been implicated in bradyzoite development (Echeverria et al., ) and exhibits different localization patterns, being found in both the nucleus and cytoplasm in bradyzoites rather than just the cytoplasm (Echeverria et al., ). Consistent with these data, serial analysis of gene expression (SAGE) libraries indicate that Hsp90 is detected early in bradyzoite differentiation (Radke et al., ). Intriguingly, geldanamycin, an antibiotic that perturbs the normal function of Hsp90, blocks the conversion of tachyzoites to bradyzoites as well as the reversion of bradyzoites to tachyzoites (Echeverria et al., ). The mitochondrial chaperone Hsp60 exists as two alternatively spliced transcripts, both of which are upregulated in bradyzoites (Toursel et al., ). Upon bradyzoite induction, TgHsp60 localizes to two unknown vesicular bodies distinct from the parasite mitochondrion. Another heat shock protein, DnaK-tetratricopeptide repeat (DnaK-TPR), which also interacts with p23, was recently found to be expressed predominantly in bradyzoites (Ueno et al., ). Three Hsp40/DnaJ family members were also found to be upregulated in alkaline-stressed tachyzoites: TGME49_115690, TGME49_010430, and TGME49_023950 (S. O. Angel and W. J. Sullivan Jr., unpublished data).
Other conventional stress response signaling pathways well characterized in higher eukaryotes may also function in Toxoplasma and possibly contribute to bradyzoite development. Homologues of mitogen-activated protein kinases (MAPKs), which regulate diverse biological processes including cellular stress responses, have been identified in Toxoplasma (Lacey et al., ). During in vitro differentiation to bradyzoites, mRNA levels for TgMAPK-1 increase, suggesting a possible role in stage conversion (Brumlik et al., ). Increasing levels of cGMP and cAMP, as well as inhibiting downstream cGMP- and cAMP-dependent kinases, have also been linked to bradyzoite differentiation (Kirkman et al., ; Eaton et al., ). The precise roles of Toxoplasma HSPs and signaling pathways during stage conversion warrant more detailed investigation.
It was recently discovered that stress conditions also induce the synthesis of the phytohormone abscisic acid (ABA) by the apicoplast (Nagamune et al., ). ABA leads to the production of cyclic ADP ribose (cADPR), which controls the release of intracellular calcium stores in Toxoplasma and induces egress. Pharmacological inhibition of ABA synthesis with fluridone blocked egress and triggered bradyzoite differentiation, suggesting that ABA-mediated calcium signaling is an important factor in whether the parasite is lytic or latent (Nagamune et al., ).
Translational repression appears to operate in both tachyzoites and bradyzoites. Appreciable mRNAs encoding the bradyzoite-specific proteins G6-PI and MAG1 are readily detected in tachyzoites as well, suggesting that such messages are translationally repressed during the tachyzoite stage (Dzierszinski et al., ; Weiss & Kim, ). Similarly, LDH1 protein is only expressed in tachyzoites, yet LDH1 mRNAs are equally detectable in both stages, suggesting that LDH1 is translationally repressed in bradyzoites (Yang & Parmley, ). The Toxoplasma eukaryotic translation initiation factor 4A (eIF4A), which facilitates the binding of capped mRNA to the 40S ribosomal subunit, is downregulated in bradyzoites, possibly reflective of the global reduction in protein synthesis during latency (Gastens & Fischer, ).
Recent studies have expanded and clarified the role of translational control in the parasite stress response and differentiation. Cellular stresses are well-known inducers of protein translation control through phosphorylation of the alpha subunit of eukaryotic translation initiation factor-2 (eIF2α). When phosphorylated, eIF2α dampens global translation initiation, thereby favoring the preferential translation of a subset of mRNAs that encode proteins geared toward alleviating the stress (Wek et al., ). As in other species, Toxoplasma phosphorylates its eIF2α orthologue (TgIF2α) in response to numerous stresses, including those that trigger bradyzoite differentiation (Sullivan et al., ). Alkaline pH, heat shock, sodium nitroprusside, sodium arsenite, and tunicamycin have all been shown to cause phosphorylation of TgIF2α (Sullivan et al., ; Narasimhan et al., ). Moreover, TgIF2α remains phosphorylated in mature bradyzoites generated in vitro (Narasimhan et al., ). We have also found that salubrinal, a specific inhibitor of eIF2α dephosphorylation (Boyce et al., ), induces bradyzoite development in vitro (Narasimhan et al., ). Collectively, these studies suggest a major role of TgIF2α phosphorylation (TgIF2α~P) in the development and maintenance of bradyzoite cysts. This idea has been supported in studies performed in the related apicomplexan parasite Plasmodium. Zhang et al. demonstrated that a Plasmodium eIF2 kinase designated IK2 controls the latency of sporozoites in the mosquito salivary gland (Zhang et al., ).
TgIF2α~P may also facilitate Toxoplasma dissemination to immune-privileged sites where bradyzoites tend to form. A greater capacity to survive outside host cells may help explain the increased ability of type I strains to disseminate throughout the host organism, particularly to the CNS. Two prevailing models describing how tachyzoites traverse biological barriers include the ‘Trojan horse’ mechanism and paracellular transmigration (Elsheikha & Khan, ). Type I parasites are not as efficient as type II and III strains in inducing migratory phenotypes in infected dendritic cells (Lambert et al., ), suggesting that type I strains may rely more heavily on dissemination as extracellular parasites (Barragan & Sibley, ). Being ill-equipped to respond robustly to the stress of an extracellular environment, type II and III strains must remain confined to the intracellular environment, a limitation that is likely to contribute to their decreased virulence. Through the generation of a nonphosphorylatable mutant, we have found that TgIF2α~P promotes survival of extracellular tachyzoites and does so in a strain-dependent manner (Joyce et al., ). Type I RH strain shows rapid and robust TgIF2α~P within hours after egress, whereas the type II strain is slower to phosphorylate TgIF2α (Joyce et al., ). It is tempting to speculate that the translational control induced by TgIF2α reprograms gene expression to protect extracellular parasites, and the increased viability of RH strain may stem from its increased capacity to phosphorylate TgIF2α for cytoprotection.
The relevance of eIF2α~P in Toxoplasma and Plasmodium latent forms underscores the need to characterize the parasite family of eIF2α kinases. Higher eukaryotes possess four eIF2α kinases, each appearing to respond to specific stress arrangements: GCN2 is activated by nutrient deprivation, protein kinase RNA-activated (PKR) is activated by viral infection, PEK/Perk is activated by ER stress, and HRI is activated by heme deficiency and heat shock (Wek et al., ). Toxoplasma also possesses four eIF2α kinases (TgIF2K-A through –D), some of which contain motifs suggestive of functional equivalency to mammalian enzymes. For example, based on the presence of a transmembrane domain, we hypothesized that TgIF2K-A may be equivalent to PEK/Perk. Indeed, TgIF2K-A localizes to the parasite ER and associates with the ER-resident chaperone BiP/GRP in a stress-dependent fashion (Narasimhan et al., ). TgIF2K-D is most similar to GCN2, possessing a histidyl-tRNA synthetase (HisRS)-related domain that stimulates kinase activity by binding to uncharged tRNAs that accumulate during nutrient starvation. We have recently demonstrated that TgIF2K-D enhances the survival of extracellular tachyzoites deprived of host cell resources (Sullivan, unpublished data). The roles of the other parasite eIF2 kinases, and how they become activated during different stress conditions, are important areas for future study.
Reprogramming the genome for bradyzoite development
In other species, stress-induced eIF2α~P reduces global protein synthesis in favor of a subset of mRNAs encoding factors needed to respond to the cellular insult or signal. Master regulator transcription factors such as GCN4/ATF4 are preferentially translated during stress, which subsequently reprogram the genome for stress remedy. Using polyribosome profiling and [35S]-Met/Cys labeling, we have verified that TgIF2α~P significantly curtails protein production (Narasimhan et al., ). These data align with an earlier study showing that eIF4A, a DEAD-box RNA helicase that facilitates binding of capped mRNA to the 40S ribosomal subunit, is strongly downregulated in bradyzoites (Gastens & Fischer, ). Recently, we have used Toxoplasma microarrays to identify mRNAs that are preferentially translated in the polyribosome fraction during ER stress (Sullivan, unpublished data). Among these genes were several so-called AP2 proteins, which contain plant-like DNA-binding domains and represent a major lineage of transcription factors in Apicomplexa (Painter et al., ). In the absence of GCN4/ATF4 homologues in Apicomplexa, it is plausible to speculate that AP2 proteins may be functional equivalents of these well-conserved master regulators.
Transcriptional regulation clearly plays a major role in bradyzoite development as evidenced by numerous studies showing stage-specific gene expression (Manger et al., ; Cleary et al., ; Radke et al., ; Sullivan et al., ). A seminal study generated multiple SAGE libraries to construct progressive ‘snapshots’ of the developmental transition from sporozoite to tachyzoite to bradyzoite in different Toxoplasma strains (Radke et al., ). Results from this study showed that distinct gene expression cascades occur through developmental transitions, underscoring the importance of transcriptional regulation throughout these events. Promoters that regulate BAG1 and a bradyzoite-specific NTPase during bradyzoite development were fine-mapped to a 6- to 8-bp resolution, and these minimal cis-elements were capable of converting a constitutive promoter to one that is induced by bradyzoite conditions (Behnke et al., ). Together, these studies reveal that conventional eukaryotic promoter mechanisms are fundamentally at work to coordinate gene expression driving stage differentiation, but the usage of AP2 proteins as transcriptional regulators is different than higher eukaryotic counterparts. Details on the roles of AP2 factors, 11 of which are induced in bradyzoites, are emerging in ongoing studies.
Recently, the bradyzoite-specific promoter ENO1 was used as ‘bait’ in an affinity-based strategy to isolate DNA-associating factors that may repress this gene in tachyzoites. Thirty-nine putative nuclear factors were found and divided into three categories: (i) 11 proteins with significant similarity to known nuclear factors, including a protein with a RNA-specific DEAD/DEAH box helicase domain, a pinin domain, and an FK506BP homologue, (ii) 7 proteins corresponding to kinases, phosphatases, and HSPs, and (iii) 21 hypothetical proteins, two of which have significant homology to Alba proteins, which are chromatin-associated silencing factors best characterized in Archaea. The FK506BP homologue (TgNF3) was extensively characterized. While not a direct binder of DNA, TgNF3 physically interacts with free core and nucleosome-associated histones to exert a gene silencing function at ENO1 and 18S ribosomal RNA genes, consistent with the yeast homologue known to be a histone chaperone that regulates rDNA silencing. Interestingly, TgNF3 is in the nucleus (enriched in the nucleolus) of tachyzoites but in the cytoplasm of bradyzoites (Olguin-Lamas et al., ), suggesting that differential compartmentalization may influence activation of the bradyzoite ENO1 gene.
Another important factor implicated in contributing to bradyzoite gene expression is histone modification and chromatin remodeling (Bougdour et al., ; Dixon et al., ). The first indication that histone modifications play a role in Toxoplasma differentiation emerged when it was reported that histones at promoter regions of bradyzoite genes are hypoacetylated in tachyzoites and become acetylated after bradyzoite induction. Conversely, tachyzoite-specific genes that are hyperacetylated in the tachyzoite stage display decreased acetylation levels upon induction of differentiation (Saksouk et al., ). The lysine (K) acetyltransferase (KAT) TgGCN5-A was localized to promoters of active developmentally regulated genes, while the lysine deacetylase TgHDAC3 was situated at the promoters of inactive genes (Saksouk et al., ). Further study of a TgGCN5-A knockout parasite has revealed a vital role for this KAT in the expression of bradyzoite-specific genes during differentiation. Although TgGCN5-A appears dispensable in type I tachyzoites during normal culture conditions, the TgGCN5-A knockout parasites fail to upregulate bradyzoite marker genes upon induction of differentiation by alkaline stress (Naguleswaran et al., ). The importance of histone acetylation for the control of differentiation is underscored by the finding that chemical inhibition of TgHDAC3 with low doses of the compound FR235222 caused conversion to bradyzoites. The conversion was accompanied by hyperacetylation of the upstream regions of > 350 genes, one-third of which are specific to bradyzoites (Bougdour et al., ).
While acetylation of histone lysine residues leads to gene activation, methylation of histone residues can result in either gene activation or repression. Monomethylation of lysine 20 of histone H4 (H4K20) by the methyltransferase TgSET8 promotes heterochromatin formation and subsequent gene silencing (Sautel et al., ). Concurrent with the global downregulation of gene expression in the quiescent bradyzoite, monomethylation of H4K20 is significantly enriched (Sautel et al., ), suggesting that TgSET8 is involved in the maintenance of this transcriptionally suppressed state. In contrast, arginine methylation of histones has been linked to gene activation. Chemical inhibition of the arginine methyltransferase TgCARM1 triggers bradyzoite differentiation (Saksouk et al., ). Consistent with this observation, TgCARM1 associates with the promoter regions of tachyzoite-specific genes in tachyzoites but is enriched at bradyzoite-specific genes upon the induction of differentiation with alkaline pH (Saksouk et al., ).
The Toxoplasma genome contains 17 predicted members of the SWI/SNF family of ATP-dependent nucleosome remodeling complexes (Dixon et al., ), a number of which may be involved in stage conversion. An EST analysis of bradyzoites formed in vivo by Manger et al. identified a SNF2-like protein (TGME49_073870) that is upregulated during differentiation (Manger et al., ). Message levels for another SNF2-like homologue, TgSRCAP, were shown to be upregulated during in vitro bradyzoite differentiation (Sullivan et al., ).
Finally, alterations in nucleosome composition are another likely method of gene expression control in bradyzoite differentiation. Toxoplasma possesses novel H2A and H2B histone variants, substitutions of which can modulate gene expression (Sullivan et al., ). The H2AZ and H2Bv variants are enriched at active genes; however, the H2AX variant present at inactive genes is upregulated in the bradyzoite stage (Dalmasso et al., ). Toxoplasma expresses many more chromatin remodelers that have yet to be fully characterized which are candidates contributing to the reprogramming of the genome during stage conversion (Dixon et al., ).
The haploid nature of tachyzoite/bradyzoite stages facilitates the generation of gene knockouts and disruptions for mutational analysis. The power of Toxoplasma as a molecular genetics system has been exploited to determine genes involved in the stage conversion. A number of genes have been linked to having a role in bradyzoite differentiation using this approach, including the aforementioned BAG1 gene, the P-type H+-ATPase PMA1 (Holpert et al., ), and the bradyzoite surface antigen SAG1-related sequence SRS9 (Kim et al., ). Recently, knockouts of dense granule proteins GRA4 and GRA6 were shown to have dramatically reduced cyst burdens at 3 weeks postinfection in C57BL/6 mice, especially as a dual knockout (Fox et al., ). A significant defect in cyst burden was also reported at 5 weeks postinfection for mutants lacking the entire 14-kb ROP4/7 locus (Fox et al., ). It should be mentioned that mutations resulting in decreased cyst numbers in vivo may not directly correlate to effects on tachyzoite to bradyzoite conversion, as it is difficult to rule out the effects on tachyzoite viability and dissemination in vivo.
To discover novel genes involved in bradyzoite development, various groups have generated developmental mutants. Singh et al. () developed a transgenic parasite clone in type II Pru strain that contained GFP fused to the bradyzoite-specific promoter LDH2. After exposing these parasites to a chemical mutagen, variants defective in bradyzoite conversion (GFP negative) could be isolated using FACS following culture in bradyzoite-inducing conditions. Microarray analysis of the mutants following various stresses revealed a hierarchy of gene activation, supporting the idea that multiple induction conditions lead to a common pathway that reprograms for bradyzoite gene expression. It was noted that a 14-3-3 homologue, PITSLRE kinase, and a vacuolar ATPase exhibited decreased expression levels in most of the bradyzoite differentiation mutants (Singh et al., ). A similar approach was followed to develop insertional mutants impaired in bradyzoite gene activation, leading to the discovery that a nucleolar CCHC zinc finger protein (ZFP1) and a pseudouridine synthase homologue (PUS1) are involved in bradyzoite differentiation (Vanchinathan et al., ; Anderson et al., ).
Matrajt et al. () applied insertional mutagenesis on type I RH parasites lacking UPRT, which facilitates differentiation to bradyzoites following CO2 starvation. Using a selectable marker driven by a bradyzoite-specific promoter, insertional mutants were isolated which have deficiencies in BAG1 expression and cyst wall formation. Like the mutants generated in the Singh study, these mutants replicated well under bradyzoite differentiation conditions, and microarray analysis revealed a gene expression profile more aligned with that seen for tachyzoites. Among the genes rescued and implicated as having a role in the differentiation process include a splicing factor, an oocyst wall protein, and AP2XII-6 (Lescault et al., ). AP2XII-6 is particularly intriguing as AP2 domain proteins are potential transcription factors. One mutant contained an insert just upstream of a gene prediction for a DNA replication factor, but the mapped disrupted locus may be a noncoding RNA (Matrajt, ). Moreover, state modeling was used to capture hidden variation between parasite lines to reveal additional genes likely to be involved in bradyzoite differentiation, including transcription elongation factor Spt5, DNA primase subunit, TBC domain protein, Homologous to the E6-AP carboxyl terminus (HECT) type ubiquitin ligase, and six hypothetical genes of unknown function (Lescault et al., ).
The Knoll laboratory has generated and screened over 8000 insertional mutants by immunofluorescence microscopy for defects in bradyzoite cyst formation. Nine mutants were identified as defective in both cyst wall formation and expression of BAG1. One of these mutants contained an insertion in a gene encoding a serine/proline-rich proteophosphoglycan. This proteophosphoglycan is upregulated in bradyzoites and enhances cyst wall component expression and assembly through an unknown mechanism (Craver et al., ). Another insertional mutant indicates that TgRSC8, which has homology to the catalytic component of the SWI/SNF and Remodel the Structure of Chromatin (RSC) complexes found in Saccharomyces cerevisiae, is important for the upregulation of a subset of bradyzoite genes during in vitro differentiation (Rooney et al., ). The latter further highlights the importance of chromatin remodeling during stage conversion.
How these gene products identified from the mutants contribute to bradyzoite conversion awaits further investigation. It is important to note that in each of these studies, the bradyzoite-deficient mutants are leaky, suggesting that control of bradyzoite differentiation is a complex process with redundancies, and no single gene appears to be able to completely ablate bradyzoite formation.
Summary and future outlook
A schematic highlighting recent key discoveries in the mechanisms of stage conversion is presented in Fig. 0005. Despite our advances, there are still many fundamental questions concerning bradyzoite differentiation to address, such as the composition of the cyst wall and matrix, how multiple signals are integrated into a common differentiation response, and which AP2 factors (or other transcription factors) operate as master regulators to coordinate the bradyzoite gene expression program.
As the molecular toolbox for parasite investigation expands, so will our understanding of bradyzoite differentiation. A major impediment to interrogating bradyzoite development is the lack of tools developed specifically for type II strains. The recent generation of Ku80 knockouts in type II Pru strain will greatly facilitate the study of genes linked to bradyzoite conversion (Fox et al., ). Conditional knockout systems also need to be developed in type II backgrounds for the study of essential genes. Like EST and SAGE library predecessors, Toxoplasma microarrays (ToxoGeneChiP) continue to be utilized to resolve the bradyzoite transcriptome further, and ChIP-chip approaches can be applied to elucidate the bradyzoite epigenome; high-throughput sequencing promises to increase the resolution of these datasets even further. Methods that need improvement include techniques to purify tissue cysts from host cells, which would provide a clearer picture of the bradyzoite proteome and metabolome. Another important advance is the application of bioluminescence imaging to study the course of Toxoplasma infection, including the reactivation of chronic infection, in real time in living mice (Saeij et al., ).
Future studies must also begin to focus on targeting bradyzoite formation and/or eliminating tissue cysts. Is it possible to develop avirulent, nonencysting Toxoplasma mutants for potential vaccines, particularly in livestock to reduce zoonotic transmission? The study of cytokines relevant to cyst formation and stability, and the observation that certain CD8+ T cells can directly target cysts (Suzuki et al., ), opens the door for immunomodulation approaches to eliminate cysts. More traditional pharmacological methods may still be useful in eradicating cysts if the drugs are conjugated to arginine oligomers, which have been shown to penetrate cyst walls (Samuel et al., ). D-luciferin, used during bioluminescent imaging of infection, was also noted as being able to access and be metabolically processed by bradyzoites within tissue cysts (Di Cristina et al., ). The latest discoveries illuminating the mechanics of how the bradyzoite gene program unfolds offer a rich repository of potential novel drug targets.
Numerous recent studies suggest that the persistence of Toxoplasma cysts in the brain may have a significant impact on the health of immunocompetent hosts. It is now well documented that infected rats undergo an exquisite change in behavior that would work to enhance transmission of the parasite back to its definitive host (Webster, ). Remarkably, rodents with latent Toxoplasma infection converted their aversion to feline odors into attraction, suggesting that they would be more easily preyed upon in the field (Vyas et al., ). Whether bradyzoite cysts within human brains affect behavior is less clear, and we are far from understanding the mechanisms (Fekadu et al., ; Webster & McConkey, ).
A great deal of progress has been made since Toxoplasma tissue cysts were first observed in 1928, but much work remains to be completed to fully understand the molecular mechanisms driving latency and the impact of chronic infection on host behavior. The significance for continued study of latency in Toxoplasma is underscored by parallels between bradyzoites and other eukaryotic pathogens with dormant stages, exemplified by the discovery that translational control through eIF2α phosphorylation is a critical determinant in latent malarial sporozoites as well as Toxoplasma bradyzoites.
Research in the Sullivan laboratory is supported by grants from the National Institutes of Health (R01 AI077502, R21 AI084031, and R21 AI083732). We thank Drs Louis Weiss (Albert Einstein College of Medicine) and David Ferguson (University of Oxford) for supplying images. Illustrations in Figures 1 and 5 were completed by Christopher M. Brown (Office of Visual Media, Indiana University School of Medicine). We also thank Drs Louis Weiss and Sherry Queener (Indiana University School of Medicine) for critically reading the manuscript and providing helpful suggestions.