Abstract
Toxoplasma gondii is an intracellular parasite that persistently infects the CNS and that has genetically distinct strains which provoke different acute immune responses. How differences in the acute immune response affect the CNS immune response is unknown. To address this question, we used two persistent Toxoplasma strains (type II and type III) and examined the CNS immune response at 21 days post infection (dpi). Contrary to acute infection studies, type III-infected mice had higher numbers of total CNS T cells and macrophages/microglia but fewer alternatively activated macrophages (M2s) and regulatory T cells (Tregs) than type II-infected mice. By profiling splenocytes at 5, 10 and 21 dpi, we determined that at 5 dpi type III-infected mice had more M2s while type II-infected mice had more classically activated macrophages (M1s) and these responses flipped over time. To test how these early differences influence the CNS immune response, we engineered the type III strain to lack ROP16 (IIIΔrop16), the polymorphic effector protein that drives the type III-associated M2 response. IIIΔrop16-infected mice showed a type II-like neuroinflammatory response with fewer infiltrating T cells and macrophages/microglia and more M2s and an unexpectedly low CNS parasite burden. At 5 dpi, IIIΔrop16-infected mice showed a mixed inflammatory response with more M1s, M2s, T effector cells, and Tregs, and decreased rates of infection of peritoneal exudative cells (PECs). These data suggested that type III parasites need the early ROP16-associated M2 response to avoid clearance, possibly by the Immunity-Related GTPases (IRGs), IFN-γ dependent proteins essential for murine defenses against Toxoplasma. To test this possibility, we infected IRG-deficient mice and found that IIIΔrop16 parasites now maintained parental levels of PECs infection. Collectively, these studies suggest that, for the type III strain, rop16III plays a key role in parasite persistence and influences the sub-acute CNS immune response.
Author Summary Toxoplasma is a ubiquitous intracellular parasite that establishes an asymptomatic brain infection in immunocompetent individuals. However, in the immunocompromised and the developing fetus, Toxoplasma can cause problems ranging from fever to chorioretinitis to severe toxoplasmic encephalitis. Emerging evidence suggests that the genotype of the infecting Toxoplasma strain may influence these outcomes, possibly through the secretion of Toxoplasma strain-specific polymorphic effector proteins that trigger different host cell signaling pathways. While such strain-specific modulation of host cell signaling has been shown to affect acute immune responses, it is unclear how these differences influence the sub-acute or chronic responses in the CNS, the major organ affected in symptomatic disease. This study shows that genetically distinct strains of Toxoplasma provoke strain-specific CNS immune responses and that, for one strain (type III), the acute and sub-acute immune responses and parasite survival are heavily influenced by a polymorphic parasite gene (rop16III).
Introduction
Toxoplasma gondii is a ubiquitous obligate intracellular parasite that chronically infects the brain, heart, and skeletal muscle of humans (1,2). Up to one third of the world’s population is estimated to be chronically infected with Toxoplasma (3). While most infected people are asymptomatic, in some immunocompromised individuals and developing fetuses Toxoplasma can cause fever, chorioretinitis, toxoplasmic encephalitis, and even death (4). While the host immune status plays a key role in determining disease outcomes, clinical data suggest that the genotype of the infecting Toxoplasma strain may also play a role (5–14). Toxoplasma strains are classified into 15 genetic haplotypes which include the three canonical strains — type I, type II, and type III (now haplotype 1, 2, and 3 and Clade A, D, and C respectively) (15–17). Of the canonical strains, type I and III are relatively genetically similar compared to type II.
Our understanding of how different Toxoplasma strains might cause distinct disease outcomes in mice and potentially humans has greatly expanded in the last decade. We now know that Toxoplasma highly manipulates host cells through the injections and secretion of effector proteins that can be polymorphic among strains. In turn these polymorphisms can profoundly affect the host cell response. For example, during acute in vitro infection of fibroblasts or immune cells only the type I/III allele of ROP16 (ROP16I/III), not the type II allele, causes direct and prolonged phosphorylation of the transcription factors STAT3 and STAT6 (18,19). In macrophages, this prolonged activation of STAT3/6 leads to decreased production of IL-12, a key pro-inflammatory cytokine (19). Conversely, only the type II allele of GRA15 (GRA15II), not the type I/III allele, activates the transcription factor, NFκB, which leads to host cell production of pro-inflammatory cytokines (20,21). In addition, strains that express Gra15II polarize infected macrophages to a classically activated phenotype whereas, strains that express ROP16I/III polarize macrophages to an alternatively activated phenotype (19,21,22). Yet, how these strain-specific modulations of infected cells influence global immune responses, or sub-acute or chronic immune responses in the central nervous system (CNS), remains unknown. The only studies looking at Toxoplasma strain-specific tissue immune responses during chronic infection were done 20 years ago and used histology to define the CNS immune response. While these studies identified strain-specific neuroinflammatory responses, the strains also produced different CNS parasite burdens making it impossible to determine if the immune response differences were simply driven by the differences in parasite burden (23,24).
To address this gap and leverage our increased understanding of strain-specific effects, we infected mice with a representative strain from either of the two canonical, encysting Toxoplasma strains (type II or type III) and then defined the neuroinflammatory response using quantitative immunohistochemistry (IHC), multiplex cytokine analysis, and flow cytometry. At 21 days post infection (dpi), compared to type II-infected mice, type III-infected mice showed a higher number of macrophages, infiltrating T cells, and levels of pro-inflammatory cytokines (e.g IFN-γ) in the CNS, even though type II and type III-infected mice showed the same CNS parasite burden. In addition, our flow cytometry analyses of CNS and splenic mononuclear cells showed that type III-infected mice had fewer alternatively activated macrophages (M2s) and regulatory T cells (Tregs) compared to type II-infected mice, the opposite of what is seen with acute infection in vivo and in vitro (21,22). By examining the peripheral macrophage immune response over time, we determined that, early in infection, type III-infected mice have more M2s compared to type II-infected mice and that this response changes over time, leading to fewer M2s in the spleen and brain of type III-infected mice by 21 dpi. To define if the differences in the early macrophage response influenced the subsequent CNS immune response, we engineered the type III strain to lack the rop16 gene (IIIΔrop16), which, as noted above, is the driver of the early type III-associated M2 response (19,21). Consistent with our hypothesis, compared to the parental type III strain, IIIΔrop16-infected mice showed a more type II-like CNS immune response with fewer macrophages and infiltrating T cells and an increase in M2s in the CNS. Unexpectedly, the IIIΔrop16-infected mice showed a substantial decrease in the CNS parasite burden; a mixed acute inflammatory immune response (i.e. an increase in classically activated macrophages (M1s), M2s, T effector cells, and Tregs); and rapid clearance from the site of inoculation. As the type III strain is sensitive to destruction by Immunity-Related GTPases (IRGs) (25,26), these results suggested that, to persist, the type III strain requires rop16III to dampen the initial immune response, including the IRG response. We tested this possibility by infecting mice that lack the IRG response (27) and found that the IIIΔrop16 strain now maintained parental levels of infection at the site of inoculation. Collectively these data suggest that Toxoplasma strain-specific immune responses persist in the sub-acute phase of disease and that, for the type III strain, rop16III is required for persistence and plays a role in determining acute and sub-acute systemic and CNS immune responses.
Results
Type III-infected mice have an increased CNS T cell and macrophage/microglia response compared to type II-infected mice
To determine if genetically divergent Toxoplasma strains cause strain-specific CNS immune responses, we infected mice with either a type II (Prugniaud) or type III (CEP) strain and analyzed the CNS macrophage and T cell immune response at 21 dpi, which we consider a sub-acute time point of CNS infection. We focused on the macrophages/microglial and T cell responses because prior work has established that these cells are essential for controlling acute and chronic toxoplasmosis (1,28–31). To quantify macrophages/microglia and T cells in the CNS after Toxoplasma infection, we stained brain sections with antibodies against Iba-1, a pan macrophages/microglial marker, or CD3, a pan-T cell surface marker. Stained sections were then analyzed by light microscopy. We found that type III-infected mice had approximately twice the number of CNS macrophages/microglia compared to type II-infected mice (Fig. 1A,B). Type III-infected mice also had a similar increase in the number of CNS T cells (Fig. 1C,D).
To determine how the influx of these immune cells changed the global CNS cytokine/chemokine environment, we isolated and analyzed protein from brain homogenates of control (saline inoculated) or infected mice using a 25-plex cytokine and chemokine LUMINEX assay. As expected, compared to control mice, type II and type III-infected mice showed a >2-fold increase in most of the pro-inflammatory cytokines and chemokines in the panel (Table S1). A subset of these cytokines and chemokines also showed a >2-fold increase in type III-infected mice compared to type II-infected mice (Fig. 1F).
Together, these data show that, at 21 dpi, type III-infected mice have significantly higher numbers of both macrophages/microglia and T cells in the CNS as compared to type II-infected mice. Consistent with this increase in CNS immune cells, type III-infected mice have a stronger CNS pro-inflammatory cytokine and chemokine milieu compared to type II-infected mice.
Differences in the CNS immune response between type II and type III-infected mice are not driven by parasite burden
Given the consistent differences we found in the number of macrophages/microglia and T cells (Fig. 1), we sought to determine if these differences simply reflected disparities in parasite burden. To address this question, we analyzed CNS parasite burden by two methods. First, using DNA isolated from brain homogenates, we performed quantitative PCR (qPCR) for the Toxoplasma-specific gene B1 (32–35). Second, we quantified CNS cyst numbers by performing immunofluorescent assays on brain sections using Dolichos biflorous agglutinin (DBA), a lectin that stains the cyst wall, to identify cysts (36). By both measures, we found that type II and type III-infected mice had equivalent CNS parasite burdens at 21 dpi (Fig. 2A,B).
To ensure that differences in parasite dissemination to the CNS were not driving the immune response differences, we quantified the parasite burden in the spleen, liver, lungs, and CNS at 5, 10, and 21 dpi using B1 qPCR. At these time points, we found that the parasite burden in these different organs did not differ between type II and type III-infected mice (Fig. 2C-F).
These data suggest that the CNS immune response differences we identified in type II and type III infection are not secondary to major differences in parasite dissemination to or persistence in the CNS, at least up to 21 dpi.
Type III-infected mice have fewer alternatively activated macrophages and regulatory T cells in the CNS and spleen at 21 dpi
While our immunohistochemistry (IHC) data suggest that type III infection causes a higher number of macrophages and T cells to infiltrate into the CNS compared to type II infection, they do not address whether infection with type II or type III parasites affects the phenotype of these cells. To address this question, we isolated immune cells from the CNS of Toxoplasma-infected mice and then used flow cytometry to identify the frequency of different immune cell populations, focusing primarily on macrophages/microglia and T cells. Additionally, we performed the same studies on splenocytes from the infected mice to define if the CNS immune response was tissue-specific or merely reflective of differences in the global immune response.
As these studies represented the first studies to use flow cytometry to compare strain-specific CNS macrophage and T cell responses, we sought to profile major classes of cells by using previously identified markers (37–40). To this end, we focused on classically activated macrophages (M1s), alternatively activated macrophages (M2s), effector T cells (Teffs), and regulatory T cells (Tregs). The gating schemes we used are shown in Fig. S1 (macrophages) and Fig. S2 (T cells). In our analyses, we placed CD80/CD86 (M1s) and MMR/CXCR3 (M2s) in the same channels because transcriptional data have shown that M1s consistently co-express CD80 and CD86 (38) and a prior study that examined CNS macrophages in type II-infected mice showed that macrophages that express MMR also express CXCR3 (41). To validate this staining protocol, we verified that we obtained the same results regardless of whether CD80 and CD86 or CXCR3 and MMR are placed in individual channels or in the same channels (Fig S3). Finally, to further confirm the identity of the M1/M2 populations we isolated CD80+/CD86+ or MMR+/CXCR3+ splenocytes and used qPCR to determine the expression of IL-12, Arg-1, and IL-4 (21,22,39,41). As expected, the population we defined as M1s (CD80+/CD86+) expressed IL-12 and not Arg-1 and IL-4, while the M2s (MMR+/CXCR3+) expressed Arg-1 and IL-4 but not IL-12 (Fig. S4A-D). In addition, approximately 35% of our CD80+/CD86+ population was positive for IFN-γ while less than 1% of the MMR+/CXCR3+ population was positive for IFN-γ (Fig. S4E,F). Collectively, these data highly suggest that the macrophage population identified by CD45+, F4/80+, CD11bhi, CD11clo, CD80+/CD86+ staining is consistent with prior descriptions of M1 macrophages and is pro-inflammatory. Similarly, the macrophage population identified by CD45+, F4/80+, CD11bhi, CD11clo, MMR+/CXCR3+ staining is consistent with prior descriptions of M2 macrophages, which are less inflammatory. From here forward, we will simply denote these populations as M1s and M2s.
Based upon these validations, our flow analyses of the CNS immune cells showed that type III-infected mice had approximately half the number and frequency of M2s compared to type II-infected mice (Fig. 3A,B). For M1s, we observed no significant difference in the absolute number or frequency between the groups (Fig. 3C,D). In addition, we found that type III-infected mice had half the number and frequency of Tregs (CD3+, CD4+, FoxP3+) as compared to type II-infected mice (Fig. 3E,F) and no difference in the number or frequency of Teffs (CD3+, CD4+ or CD8+, CD44+) (Table S2).
Consistent with our findings in the CNS we observed that splenocytes from type III-infected mice had approximately half the absolute number and frequency of M2s as compared to splenocytes from type II-infected mice (Fig. 4A,B). There was no significant difference in the absolute number and frequency of M1s in the spleen (Fig. 4C,D). Akin to our CNS data, splenocytes from type III-infected mice had half the number and frequency of Tregs compared to splenocytes from type II-infected mice (Fig. 4E,F).
The numbers of total CD3+ T cells, CD4 T cells (CD3+, CD4+), CD8 T cells (CD3+, CD8+), exhausted T cells (CD3+, CD8+, PD-1+), or macrophages (CD45+, F4/80+, CD11bhi, CD11clo) in either the CNS or the spleen (Table S2) were not statistically different. In addition, to allow us to track infected cells and/or cells injected with parasite effector proteins (42), we infected Cre reporter mice that only express GFP after Cre-mediated recombination (43) with mCherry+ parasite strains that trigger Cre-mediated recombination (42,44). At 21 dpi, we identified no mCherry+ and/or GFP+ cells in the T cell or macrophage populations isolated from the CNS or spleen. The lack of GFP+ immune cells in the CNS at 21 dpi is consistent with our prior work using the same system (45).
These data strongly suggest that in addition to quantitative differences in the CNS immune response, type II and III-infected mice show differences in the phenotype of immune cells infiltrating into the CNS. Type III infection provokes a more pro-inflammatory sub-acute CNS immune response with a relative decrease in the macrophages (M2s) and T cells (Tregs) that suppress the pro-inflammatory response. In addition, as these differences are also seen in splenocytes, these data suggest that, at 21 dpi, the CNS immune response is reflective of the systemic immune response. Finally, the lack of mCherry+ and/or GFP+ immune cells suggests that our findings are not driven by a small population of immune cells that are actively infected or directly manipulated by parasites.
The macrophage phenotype switches over time during type II and type III infection
Our finding that type III infection provokes a stronger pro-inflammatory response at 21 dpi was unexpected because of the prior work showing that macrophages infected with type III parasites are polarized to M2s while macrophages infected with type II parasites are polarized to M1s (21,46). As our work was done in vivo at 21 dpi and the prior work was done in vitro or very early in vivo (1-3 dpi), one explanation for these discrepancies is that the in vivo immune response evolves over time. To test this possibility and as we had found that splenocytes were accurate predictors of the CNS immune response, we phenotyped splenocytes from infected mice at 5, 10, and 21 dpi.
At 5 dpi, we observed that type III-infected mice had an approximately 3-fold higher frequency and 2-fold higher number of splenic M2s as compared to type II-infected mice (Fig. 5A). Conversely, at this time point, type II-infected mice showed an increased frequency and 1.5-fold higher number of splenic M1s as compared to type III-infected mice (Fig. 5B). By 10 dpi, the macrophage compartment from both type II and type III-infected mice had expanded and no difference in macrophage polarization state was seen (Fig. 5C-F). By 21 dpi, the splenic macrophage compartment was contracting and now type III-infected mice had fewer splenic M2s both by absolute number and frequency compared to type II-infected mice (Fig. 5C,D). At 5 and 10 dpi, for macrophages from type II or type III-infected mice, we found 1% or less of the macrophage population was infected or injected with parasite protein (i.e. ≤ 1% of the macrophage population was mCherry+ and/or GFP+). We found no strain-specific differences in the Tregs response at 5 or 10 dpi (Fig. 5G,H).
These data show that early in infection type III-infected mice show a stronger M2 macrophage response than type II-infected mice. However, as the infection progresses, the immune response evolves such that by 21 dpi type III-infected mice now have a significant decrease in these anti-inflammatory macrophages compared to type II-infected mice, even though parasite dissemination to the CNS is equivalent (Fig. 2). Unlike the macrophage response, we did not observe strain-specific differences in Tregs until 21 dpi. Our data also show that these strain-specific differences are primarily found in uninfected macrophages.
ROP16III affects the type III CNS immune response and parasite persistence
Given the evolution of these strain-specific differences from 5 to 21 dpi, we hypothesized that the early macrophage immune response might heavily influence the ensuing sub-acute immune response. We focused on the early macrophage response for several reasons. Tissue resident macrophages are some of the first cells to interface and respond to infecting microbes (47). Consistent with this concept, at 5 dpi, we found strain-specific differences in the macrophage compartment but not the T cell compartment (Fig. 5A-H). Furthermore, a growing body of literature suggests that this early response influences the ensuing T cell response, possibly through the secretion of cytokines (48). Finally, as noted above, prior in vitro and in vivo work has already established that macrophages infected with type II parasites polarize to M1s while macrophage infected with type III parasites polarize to M2s. Importantly, these studies also determined that specific alleles of Toxoplasma effector proteins (GRA15II for M1s and ROP16III for M2s) drive these macrophage phenotypes (19,21,22), giving us a mechanism for altering these responses. Thus, to determine if early macrophage responses influence the subsequent CNS immune response, we used CRISPR/Cas9 (49–53) to engineer a type III strain that lacked rop16 (Fig. S5A). We validated the deletion of rop16 (IIIΔrop16) using locus-specific PCR (Fig. S5B) and a functional assay to show that these parasites no longer induced host cell phosphorylation of STAT6 (Fig. S5C), the transcription factor linked to the rop16III-associated M2 phenotype (19). Given that this strain should lack the early type III-associated M2 response, we predicted that it would provoke a more type II-like sub-acute CNS immune response. To test this prediction, we infected mice with type II, type III, or IIIΔrop16 parasites and, at 21 dpi, analyzed the CNS immune response by IHC and flow cytometry. Consistent with our hypothesis, by quantitative IHC, we found that the CNS immune response in IIIΔrop16-infected mice looked akin to type II-infected mice with fewer infiltrating macrophages/microglia and T cells compared to type III-infected mice (Fig. 6A-E). By flow cytometry, IIIΔrop16-infected mice again looked similar to type II-infected mice in terms of M2s frequency and absolute number (Fig. 7A,B). The frequency and the absolute number of M1s were not statistically different between type II, type III, or IIIΔrop16-infected mice (Fig. 7C,D). Unexpectedly, by both Toxoplasma-specific qPCR and cyst count, the IIIΔrop16-infected mice showed a substantial decrease in the CNS parasite burden compared to type II or type III-infected mice (Fig. 8A,B).
To verify that the lack of rop16III drove the immune response and parasite burden changes we identified, we generated a IIIΔrop16::ROP16III strain that ectopically expresses rop16III. We validated the expression of rop16III using gene-specific PCR and a functional assay to confirm parasite-induced host cell phosphorylation of STAT6 (Fig. S5A-C). We then infected mice with type III, IIIΔrop16, or IIIΔrop16::ROP16III parasites and analyzed the CNS immune response by quantitative IHC and flow cytometry. The IIIΔrop16::ROP16III strain produced a CNS immune response akin to the parental type III strain, and distinct from the IIIΔrop16, in terms of macrophage and T cell numbers (Fig. 9A,B), parasite burden (Fig. 9C,D), and M2 number and frequency (Fig. 9E). The M1 number and frequency was not different between the three strains (Fig. 9F).
In summary, these data highly suggest that, in the context of a type III infection, rop16III influences the CNS immune response and enables parasite persistence.
A lack of rop16III induces a mixed inflammatory response during acute infection
To confirm that the changes in the CNS were downstream of a change in the acute inflammatory response, at 5 dpi we phenotyped splenocytes from mice infected with type II, type III, or IIIΔrop16 parasites. Unexpectedly, mice infected with the IIIΔrop16 strain showed a mixed immune phenotype with an increase in both M2s and M1s (Fig. 10A,B). In addition, and unlike either type II or type III-infected mice, IIIΔrop16-infected mice showed an increase in the number and frequency of Tregs (Fig. 10C) as well as a 3-fold increase in the number of IFN-γ+ CD4 and CD8 T cells (Fig. 10D,E). Using the mean fluorescence intensity of IFN-γ staining to assess IFN-γ production per cell, we found that T cells from IIIΔrop16 and type II-infected mice produced similar amounts of IFN-γ, which was 1.5-fold higher than T cells from type III-infected mice (Fig. 10D,E).
To verify that these acute peripheral immune response changes were driven by rop16III, we infected mice with the type III, IIIΔrop16, or IIIΔrop16::ROP16III parasites and phenotyped the splenocytes at 5 dpi. As expected the IIIΔrop16::ROP16III strain produced a splenocyte immune response akin to the parental type III strain in terms of M2s, M1s, and Treg frequency and absolute number (Fig. 11A-C). In the IFN-γ+ CD4 and CD8 compartment, the IIIΔrop16::ROP16III strain also produced a parental type III response in terms of absolute numbers of CD4 and CD8 IFN-γ+ cells and the level of IFN-γ produced per cell (Fig. 11D,E).
In summary, these data show that type III parasites lacking rop16 produce a mixed inflammatory phenotype at 5 dpi, suggesting that rop16III is an important determinant of the acute immune response to type III parasites.
Type III parasites depend upon rop16 to avoid early clearance by Immunity-Related GTPases (IRGs)
Given this unanticipated acute inflammatory response including the elevated level of T cell IFN-γ production, and decreased CNS parasite burden at 21 dpi, we hypothesized that the IIIΔrop16 parasites were undergoing an increased level of early clearance. This heightened early inflammatory response and the rapid clearance of parasites would then provoke a strong counterbalancing anti-inflammatory response, causing an increase in both M2s and Tregs early in infection. This hypothesis is particularly appealing because type III strains have a very low expression of the virulent allele of rop18, a parasite virulence gene that disables the interferon-γ-inducible Immunity-Related GTPase system (IRGs) which is a major mechanism by which murine host cells kill intracellular parasites in an IFN-γ dependent manner (25,26,54). To test this possibility, we quantified the frequency of infected peritoneal exudate cells (PECs) at 1, 3, and 5 dpi from mice infected with type II, type III, or IIIΔrop16 parasites. At 1 and 3 dpi, all 3 strains showed the same frequency of infected PECs. But, by 5 dpi, IIIΔrop16-infected mice showed a 4-6-fold lower rate of infected PECs compared to type II or parental type III-infected mice (Fig. 12A-C). To confirm that this increase in parasite clearance was secondary to a lack of rop16III, we infected mice with type III, IIIΔrop16, or IIIΔrop16::ROP16III parasites and assayed the frequency of infected PECs at 3 and 5 dpi. As expected, the IIIΔrop16::ROP16III strain maintained parental levels of infected PECs at both time points (Fig. 12D).
Finally, to directly test if the IRGs were the mechanism by which the IIIΔrop16 parasites were being cleared, we assayed the frequency of infected PECs at 3 and 5 dpi in mice that lack both Irgm1 and Irgm3, key mediators of IRG pathway (27,55–58). We reasoned that if the IRGs mediated the rop16III-dependent increase in clearance, then in Irgm1/3 KO mice, the IIIΔrop16 strain should now maintain parental levels of PECs infection at 5dpi, which is what we found (Fig. 12E).
Collectively, these data suggest that, in the context of a type III infection, rop16III is essential to dampen the acute IFN-γ response thereby avoiding rapid parasite clearance by the IRGs.
Discussion
The results presented here show that genetically distinct Toxoplasma gondii strains provoke strain-specific CNS immune responses and that these sub-acute immune responses are likely influenced by the initial systemic immune response. We have shown that, compared to infection with a type II strain, infection with a type III strain induces a more pro-inflammatory, sub-acute CNS immune response in both quality and quantity at the level of infiltrating T cells and macrophages/microglia, and that these strain-specific immune responses are not simply driven by differences in parasite burden. In addition, we have shown that, for the parameters monitored at 21 dpi, the CNS immune response mirrors the systemic immune response as gauged by splenocytes. By temporally profiling the systemic macrophage response, we have shown that this response evolves over time, leading us to hypothesize that the early macrophage response affects the subsequent sub-acute response. This hypothesis is partially supported by our finding that a IIIΔrop16 strain, which induces an early immune response distinct from the parental type III strain, produces a type II-like CNS immune response in quantity and quality, despite having a much lower CNS parasite burden than either the type II or parental type III strain.
Based upon these data and prior work, we propose the following model: early in infection, type II-infected macrophages are polarized to M1s which secrete high levels of IL-12 (21,22). This secreted IL-12 then influences the uninfected macrophages to polarize towards M1s, resulting in a systemic, highly pro-inflammatory early M1 responses (5 dpi), with high levels of IL-12 and IFN-γ production (Fig S4). This early pro-inflammatory response then provokes a counter-balancing anti-inflammatory response that leads to a rise in M2s, which continues as parasites disseminate and enter the brain. This systemic response is then recapitulated in the CNS as parasites establish a chronic CNS infection (Fig. 13A). Conversely, for type III parasites, during acute infection, type III-infected macrophages are polarized to M2s via rop16III leading to increased IL-4 production (Fig S4), which promotes a mixed systemic inflammatory response with more early M2s, resulting in decreased levels of IFN-γ and IL-12 and increased levels of IL-4 (Fig S4) compared to type II-infected mice. This less pro-inflammatory early response enables type III parasites to evade early clearance and avoids provoking the compensatory anti-inflammatory response, so as type III parasites disseminate to the CNS, the immune response that occurs is now strongly pro-inflammatory (Fig. 13B).
While these models fit our data, many unanswered questions remain. What are the molecular and cellular mechanisms by which these early immune responses influence the later responses? How does the type II gra15 allele, which drives an M1 phenotype in infected macrophages, influence the systemic immune response to type II strains? Do these differences persist even in highly established chronic infections (e.g. months to years post infection)? Though these strain-specific immune responses do not appear to be driven by gross differences in systemic dissemination or CNS parasite burden, could strain-specific rates of switching from tachyzoites to bradyzoites play a role in driving immune response differences? We believe we have established a system in which these highly complex host-parasite interactions can systematically be dissected using engineered parasites and mice.
In addition to establishing a tractable model for understanding the evolution of immune responses, several other important points arise from our data. We potentially identified a reason for the retention of the type I/III rop16I/III (rop16I/III) allele. In the original type II x III cross, rop16 was not identified as a virulence gene but rather its strain specificity was detected through strain-specific differences in host cell signaling in human fibroblasts (59). In fact, in a highly lethal type I strain, at 72 hours post infection, the loss of rop16 increased the PECs infection rate, systemic dissemination, and parasite burden in distal organs while also increasing IL-12 production by PECs. These data suggest that the increase in IL-12, which should result in higher IFN-γ production, does not adversely affect the IΔrop16 parasites (19). Conversely, we used a type III strain, which is genetically similar to the type I strain but avirulent in mice because of its’ low expression of rop18, a key protein for blocking murine IFN-γ-dependent cell intrinsic defenses (25,26,54). In the context of low rop18 expression and therefore high susceptibility to the IRGs, the ability of rop16I/III to decrease the early IFN-γ response (Fig. 10E) by decreasing IL-12 and increasing the M2 response, appears crucial for type III strains to avoid rapid clearance during the very earliest part of infection. This proposed mechanism is supported by the data in the Irgm1/3 knockout mice (Fig. 12) as well as prior work showing that IRG-mediated clearance of intracellular parasites is a major murine IFN-γ-dependent mechanism for controlling Toxoplasma (25,26,54). Collectively, these data suggest that, in vivo, rop16I/III is dispensable for type I strains but essential for type III strains, a discrepancy potentially explained by differences in rop18 expression.
We have also shown that the strain-specific polarizations previously identified in infected macrophages (21) also extends to uninfected macrophages found in the spleen. Based upon the work of others showing that type II-infected macrophages secrete more IL-12 while type I/III-infected macrophages secrete less IL-12 and more IL-4 (19,21,22,60), we speculate that the strain-specific differences in secreted cytokines from infected immune cells propagates the immune cell polarization differences to uninfected immune cells (Fig. 13).
Finally, we have shown that the IIIΔrop16 strain is able to elicit a strong brain immune response, including infiltration of T cells and likely monocytes, despite having a very low CNS parasite burden (Fig. 6-8). The finding of a much stronger CNS immune response than parasite burden is consistent with prior work showing that immune cells can and do infiltrate into the CNS in the setting of a strong systemic immune response without brain infection or pathology (61). We suggest that our data add to the growing body of literature that the “immune privileged” status of the CNS is less absolute than previously thought and that we still have much to learn about what governs when and if immune cells infiltrate into the CNS.
Materials and methods
Ethics statement
All mouse studies and breeding were carried out in strict accordance with the Public Health Service Policy on Human Care and Use of Laboratory Animals. The protocol was approved by the University of Arizona Institutional Animal Care and Use Committee (#A-3248-01, protocol #12-391).
Parasite maintenance
All parasite strains were maintained through serial passage in human foreskin fibroblast (gift of John Boothroyd, Stanford University, Stanford, CA) using DMEM, supplemented with 10% fetal bovine serum, 2mM glutagro, and 100 I.U/ml penicillin and 100 µg/ml streptomycin. Unless otherwise mentioned, previously described type II (Pruginaud) and type III (CEP) parasites expressing Cre recombinase and mCherry were used (42,44,45). For experiments with IIIΔrop16 strains, depending upon the IIIΔrop16 clone, we used either the previously mentioned strains or PruΔhpt and CEPΔhpt strains in which the endogenous gene for hypoxanthine xanthine guanine phosphoribosyl transferase (HPT) (gift of John Boothroyd) has been deleted.
Mice
Unless otherwise specified, the mice used in this study are Cre-reporter mice that only express a green fluorescent protein (GFP) after the cells have undergone Cre-mediated recombination (43). Mice were purchased from Jackson Laboratories (stock # 007906) and bred in the University of Arizona Animal Center. We used these mice in combination with our Cre-secreting parasites as a way to identify the immune cells that had been injected with parasite rhoptry proteins (42). In addition, breeding pairs of mice lacking Irgm1 and Irgm3 were provided to us courtesy of Greg Taylor (Duke University, Durham, NC) and subsequently bred in the University of Arizona Animal Center. Mice were inoculated intraperitoneally with 10,000 freshly syringe-lysed parasites diluted in 200 µl of UPS grade PBS.
Tissue preparation for histology and protein/DNA extraction
At appropriate times post infection, mice were anesthetized with a ketamine (24 mg/ml) and xylazine (4.8 mg/ml) cocktail and transcardially perfused with ice cold phosphate buffered saline. As previously described, after harvesting organs, the left half of the mouse brain was fixed in 4% paraformaldehyde in phosphate buffer, kept at 4°C overnight, rinsed in PBS, and then was embedded in 30% sucrose (42,45). Post fixation and sucrose embedding, the brain was sectioned into 40 µm thick sagittal sections using a freezing sliding microtome (Microm HM 430). Sections were stored as free-floating sections in cryoprotective media (0.05 M sodium phosphate buffer containing 30% glycerol and 30% ethylene glycol) until stained and mounted on slides. The right half of the brain was sectioned coronally into 2 halves and stored in separate tubes. These tubes were flash frozen and stored at −80°C until used for protein or DNA extraction.
Immunohistochemistry
As described previously, free-floating tissue sections were stained using a standard protocol (62). Brain sections were stained using the following primary antibodies: polyclonal rabbit anti-Iba-1 (019-19741, Wako Pure Chemical Industries, Ltd., (1:3000); monoclonal hamster anti-mouse CD3ε 500A2 (550277, BD PharmingenTM, (1:300). Following incubation with primary antibody, sections were incubated in appropriate secondary antibodies: biotinylated goat anti-rabbit (BA-1000, Vector Laboratories (1:500) and biotinylated goat anti-hamster (BA-9100, Vector Laboratories, (1:500). Next, sections were incubated in ABC (32020, Thermo Fisher) for 1 hr followed by 3,3’-Diaminobenzidine (DAB) (SK-4100, Vector Laboratories) detection of biotinylated antibodies.
Immune cell quantification
Brain sections stained for anti-CD3 or anti-Iba-1 antibody and detected using DAB were analyzed using light microscopy. For each brain section, twelve fields of view (FOV) throughout the cortex region of the brain beginning with the rostral region and moving caudally were sampled in a stereotyped way and as previously described (35,62). The number of CD3ε cells/FOV was quantified using SimplePCI software (Hamamatsu, Sewickley, PA) on an Olympus IMT-2 inverted light microscope (35). The number of Iba-1+ cells/FOV was quantified by manually counting cells with FIJI software (63). These analyses was performed on 3 brain sections per mouse, after which the resulting numbers were then averaged to obtain the average number of Iba1+ or CD3+ immune cells/brain section/mouse. Investigators quantifying CD3+ and Iba-1+ cells were blinded to the infection status of the mouse until after the data were collected.
Protein Extraction, Quantification and Multiplex LUMINEX Assay
The caudal quarter of the flash frozen brain tissue was homogenized in radioimmunoprecipitation assay buffer (1% Triton-X, 0.1% SDS, 1X PBS) and 1% phosphatase inhibitor cocktail (P5726-1ML, Sigma-Aldrich) and 1% protease inhibitor cocktail (P8340-1ML, Sigma-Aldrich). As previously described, the samples were sonicated on ice in 3s bursts until homogenized and then were centrifuged at 4°C (62). The protein concentration of each sample was measured using Direct Detect® Infrared Spectrometer. Each sample was stored at −80°C until the LUMINEX® assay was performed. Cytokines and chemokines were assessed using the MILLIPLEX-MAP-Mouse-Cytokiine/Chemokine-Magnetic-Bead-Panel (MCYTOMAG-70K, EMD Millipore). This multiplex panel allows the detection of 25 different cytokines/chemokines and includes individual quality controls for each cytokine/chemokine. The samples were plated as duplicates and the plate was analyzed using a LUMINEX MAGPIX xPONENT 1.2 System which uses Milliplex Analyst software and Luminex® technology to detect individual cytokine/chemokine quantities.
Quantitative real time PCR
For quantification of parasite burden, genomic DNA from the rostral quarter of the frozen brain was isolated using DNeasy Blood and Tissue kit (69504, Qiagen) and following the manufacturer’s protocol. The Toxoplasma specific, highly conserved 35-repeat B1 gene was amplified using SYBR Green fluroscence detection with the Eppendorf Mastercycler ep realplex 2.2 system using primers listed in Table S3. GAPDH was used as house-keeping gene to normalize parasite DNA levels. Results were calculated as previously described (35,62). For quantification of Arg-1, IL-4, and IL-12, total RNA from sorted M1 and M2 cells was extracted with TRIzol™ (Life Technologies, Grand Island, NY) and according to the manufacturer’s instructions. First strand cDNA synthesis was performed using a High-Capacity cDNA Reverse Transcription kit (4368814, ThermoFisher) and following the manufacturer’s instructions. Arg-1, IL-4, and IL-12 were amplified using SYBR Green fluorescence detection with the Eppendorf Mastercycler ep realplex 2.2 system using the primers are listed in Table S3 (21,22,41). GAPDH was used as the house-keeping gene for normalization. The reaction condition were as follows: 2 min at 50°C, 2 min at 90°C, followed by 40 cycles of 15 sec at 95°C, 15 sec at 58°C, and 1 min at 72°C, followed by a melting curve analysis.
Cyst counts
Sagittal brain sections were washed and blocked in 3% Goat Serum in 0.3% TritonX-100/TBS for 1 hr. Sections were then incubated with biotinylated Dolichos biflorus agglutinin (DBA) (Vector Laboratories 1031, 1:500), which binds to the cyst wall (64–66). The following secondary was used: 405 Streptavidin (Invitrogen, 1:2000). Sections were mounted as previously described (35). The number of cysts were enumerated using a standard epifluorescent microscope (EVOS microscope). Only objects that expressed mCherry and stained for DBA were quantified as cysts.
Flow Cytometry
At appropriate times post infection, mice were euthanized by CO2 and intracardially perfused with 20 mL ice-cold PBS, after which spleens and brains were harvested. These tissues were then made into single cell suspensions. For brains, mononuclear cells were isolated by mincing the tissue and then passing it serially through an 18-gauge needle and then a 22-gauge needle in complete RPMI (86% RPMI, 10% FBS, 1% penicillin/streptomycin, 1% L-glutamine, 1% NEAA, 1% sodium pyruvate and <0.01% β-mercaptoethanol) as described previously (67). After syringe passage, the cell suspension was passed through a 70 µm strainer and mononuclear cells were isolated using a density gradient that consisted of a 60% Percoll solution in cRPMI overlayed with a 30% Percoll solution in PBS. Brain mononuclear cells were isolated from the interphase. Spleens were made into single cell suspension and passed through a 40 µm strainer (68). Red blood cells were lysed by using ammonium chloride-potassium carbonate (ACK) lysis buffer. The total numbers of viable cells from brain and spleen suspensions were determined by trypan blue exclusion and counting on a hemocytometer. Brain and spleen samples were split in order to stain with either the T cell panel or the macrophage panel. Brain and spleen single cell suspension had Fc receptors blocked with 2.4G2 to prevent non-specific staining. The following directly conjugated antibodies were utilized for flow cytometry analysis of T cells: CD3 APC eFluor® 780 (clone 17A2; eBioscience, 47-0032-80), CD8a PerCP-Cy 5.5 (clone 53-6.7; eBioscience, 45-0081-82), CD4 PECy7 (clone GK1.5; BioLegend, 100422), CD44 Alexa Fluor® 700 (clone IM7; eBioscience, 12-0441-82), CD279 (PD-1) eFluor® 450 (clone J43; eBioscience, 48-9985-82) were used to incubate cells for 30 min protected from light. The following directly conjugated antibodies were utilized for flow cytometry analysis of macrophages/microglia: CD45 PerCP-Cy5.5 (clone 30-F11; eBioscience, 45-0451), F4/80 Alexa Fluor® 700 (clone BM8; BioLegend, 123130), CD11b Pacific Blue™ (clone M1/70; BioLegend, 101224), CD11c PE/Cy7 (clone N418; BioLegend, 117318), CD11c FITC (clone N418; eBioScience, 11-0114-85), CD183 (CXCR3) phycoerythrin (PE) (clone CXCR3-173; BioLegend, 126505), CD183 (CXCR3) PE/Cy7 (clone CXCR3-173; BioLegend, 126516), CD206 (MMR) phycoerythrin (PE) (clone C068C2; BioLegend, 141706), CD80 APC (clone 16-10A1; eBioscience 17-0801-82), CD80 PE/Cy5 (clone 16-10A1; BioLegend, 104712), CD86 APC (clone GL-1; BioLegend, 105012), CD86 PE/Cy5 (clone GL-1; BioLegend, 105016), Ly-6G/Ly-6C (Gr-1) PE/Cy5 (clone RB6-8C5; BioLegend, 108410). Cells were incubated with appropriate antibodies for 30 min, while being protected from light. After surface staining, cells were then stained with live/dead Fixable Yellow Dead Cell Stain Kit (Life Technologies, L34959) for 30 min to distinguish between live and dead cells. For intracellular cytokine staining, while protected from light, samples were washed, permeabilized, and fixed using a permeabilization and fixation kit (eBioscience, 00-5223-56; 00-5123-43; 00-8333-56). An intracellular staining protocol was used to stain for FoxP3 PE (clone FJK-16s; eBioscience, 12-5773-82), and IFN-γ APC (clone XMG1.2; eBioscience 17-7311-82) for 30 min. Samples were washed after each staining step to remove residual unbound antibody. A BD LSR II (BD Biosciences, San Jose, CA; University of Arizona Cancer Center) was used to run the samples and Flowjo (Treestar) was used for all flow cytometry analysis.
For sorting, splenocytes were isolated as described above and stained with the macrophage panel followed by the live/dead staining as denoted above. Samples were washed and resuspended in PBS until sorted, which was done on the same day as isolation. A FACS Aria III (BD Biosciences, San Jose, CA; University of Arizona Cancer Center) was used to sort M1 and M2 populations. After sorting, samples were resuspended in TRIzol™ (Life Technologies, Grand Island, NY) and stored at −80°C until RNA extraction.
Plasmid construction
All the plasmids and primers used to make and validate the IIIΔrop16 strain are listed in Table S3. The rop16-targeting CRISPR plasmids (sg rop16Up and sg rop16Down) were constructed from sgUPRT using a previously described Q5 mutagenesis protocol (51–53). To generate a plasmid for inserting hxgprt into the rop16 locus, upstream (500-bp) and downstream (500-bp) regions directly adjacent to the sgROP16Up and sg rop16Down target sequence were used to flank hxgprt via sequential restriction cloning.
Generation of IIIΔrop16 knockout and IIIΔrop16::ROP16III
To disrupt rop16 in type IIIΔhpt, we transfected parasites with 3 plasmids: the sg rop16Up CRISPR and sg rop16Down CRISPR plasmids and the pTKO (69) plasmid containing rop16 homology regions surrounding a selectable marker (hxgprt) or with the toxofilin-Cre cassette (44) (Fig. S5). Selection by growth for 4 to 8 days in 25 mg/ml mycophenolic acid and 50 mg/ml xanthine (70) was used to obtain stably resistant clones with hxgprt integration. These clones were subsequently screened by PCR to confirm disruption of the rop16 locus (Fig. S5). Clones negative for rop16 and positive for integration of hxgprt were confirmed by western blot to have lost the rop16-dependent phosphorylation of STAT6 (59). In addition, clones with the toxofilin-Cre cassette were confirmed to trigger Cre-mediated recombination as previously described (44).
To complement rop16, IIIΔrop16 parasites were transfected with 50 μgs of linearized plasmid DNA harboring a FLAG-tagged ROP16III gene and a bleomycin resistance marker. Post transfection freshly egressed parasites were resuspended in DMEM supplemented with 50 μg/mL of Zeocin (InvivoGEN, 11006-33-0) for 4 hour and then added to HFF monolayers supplemented with 5μg/mL Zeocin to select for integrants. This process was repeated 3 times prior to plating by limiting dilution to isolate single clones. Single clones were subsequently screened by PCR for rop16 integration (Fig. S5). The IIIΔrop16::ROP16III clones were all derived from the IIIΔrop16 strain that expresses toxofilin-Cre.
Peritoneal Exudate Cells Isolation
Cre-reporter mice or Irgm1/3 knockout mice were infected with type II, type III, IIIΔrop16, or IIIΔrop16::ROP16III. At appropriate times, peritoneal exudate cells (PECs) were collected by peritoneal lavage with 10 ml of cold 1 x PBS. PECs were incubated in FcBlock for 10 min as described above. PECs were subsequently stained with CD45 marker, followed by with live/dead staining as described above, and then analyzed using a BD LSR II (BD Biosciences, San Jose, CA; University of Arizona Cancer Center) (42).
Statistical analysis
Statistical analyses were performed using Prism 7.0 software. To improve distributional characteristics, total numbers of CD3 and Iba-1 cells were log transformed prior to analysis. Unless otherwise specified, two-way analysis of variance (ANOVA) with Fisher’s protected LSD was used, with the cohort as the block factor and parasite strain as the experimental factor. For cytokine levels, the data were analyzed using a one-way ANOVA with Bonferroni’s post-hoc test.
Acknowledgements
The authors would like to thank all members of the Koshy Lab for helpful discussions and critical review of the manuscript. We would like to thank Dean Billheimer (Director of BIO5 Statistical Consulting Group) for his suggestions for data analyses. We would also like to thank the Doyle Lab, especially Vivian Nguyen for the use of their Luminex® MAGPIX for the cytokine assays. Finally, we would like to thank Greg Taylor (Duke University), for providing us with breeding pairs for the Irgm1/3 knockout mice. Research in this manuscript was directly supported by the UA Flow Cytometry Shared Resource funded by the National Cancer Institute (P30CA023074).