TLR Adaptor MyD88 Is Essential for Pathogen Control during Oral Toxoplasma gondii Infection but Not Adaptive Immunity Induced by a Vaccine Strain of the Parasite¹

The TLRs have emerged as a major family of pattern recognition molecules involved in sensing infectious non-self molecules and possibly endogenous molecules of the host. There are 11–13 TLRs in mice and humans that recognize diverse molecules including bacterial LPS, lipopeptides, unmethylated CpG oligodinucleotides, as well as ssRNA and dsRNA (1, 2). Signaling through TLR is complex, but all TLR (with the exception of TLR3) use the adaptor molecule MyD88 to initiate intracellular signal transduction (3, 4, 5). Most prominent among these signaling pathways are NF-κB and MAPK cascades leading to induction of proinflammatory cytokines such as IL-12 and TNF-α. MyD88 is also used to relay signals emanating from IL-1R and IL-18R. As such, MyD88 represents a bottleneck through which most TLR-initiated signals, as well as IL-1R/IL-18R-initiated signals, must pass.

The importance of MyD88 in infectious disease models has been firmly established using genetically engineered MyD88 knockout (KO)⁴ mice (MyD88^−/− mice). Animals lacking MyD88 display decreased resistance to bacteria such as Mycobacterium tuberculosis, Staphylococcus aureus, Listeria monocytogenes, and Brucella abortus (6, 7, 8, 9). It is also becoming clear that TLR and MyD88 are important in the innate immune response to parasitic protozoans (10). Thus, MyD88 KO mice are increased in susceptibility to infections with Toxoplasma gondii, Plasmodium berghei, Leishmania major, and Trypanosoma cruzi (11, 12, 13, 14, 15). TLR ligands have been identified for several protozoan species, including T. gondii, Plasmodium, and T. cruzi (16, 17, 18). Nevertheless, although KO of MyD88 may have dramatic effects on resistance to bacterial and protozoan infection, inactivation of genes encoding individual TLR often has little or no impact. This result has led to the view that the response to any given pathogen is likely to involve multiple TLR that act in concert to provide optimal resistance to infection (19, 20, 21).

There is evidence that TLR-MyD88 signaling is important, not only in innate immune responses, but also in initiation of acquired immunity (22). For example, MyD88-transduced signaling leads to up-regulation of costimulatory molecules on dendritic cells (DC) that is required to activate naive T cells, and T cell Ag-specific responses are optimized when microbial peptides are presented within the context of TLR ligands (23, 24). Nevertheless, the role of TLR-MyD88 in initiation of adaptive immunity during infection with complex pathogens is less certain. In some cases MyD88 is not necessary, such as in the generation of L. monocytogenes-specific CD8⁺ T cell responses and during acquired immunity to Borrelia burgdorferi (25, 26, 27). In other cases, MyD88 seems to play a prominent role in adaptive immunity, for example in the humoral immune response to polyoma virus infection and the antiviral CD8⁺ response to lymphocytic choriomeningitis virus (28, 29).

In this study we focus on infection with the intracellular protozoan T. gondii, a parasite known for its ability to elicit strong protective Th1 responses, but causes serious disease in immunocompromised individuals (30, 31). It has previously been shown that MyD88^−/− animals rapidly succumb to Toxoplasma in a model involving i.p. injection of parasites (14). TLR signaling is strongly implicated in this situation because mice defective in IL-1β-converting enzyme display normal resistance to infection despite an inability to produce functional IL-1 or IL-18 (15). Tachyzoite profilin and GPI lipid moieties have recently been identified as parasite ligands of TLR11 and TLR2/4, respectively (32, 33, 34). Furthermore, for the case of profilin, there is evidence that signaling through TLR11 promotes an immunodominant T cell response to peptides derived from this molecule (35).

In this study, we show that MyD88 is required to survive oral infection with Toxoplasma. Early during infection MyD88^−/− animals displayed defects in neutrophil recruitment to mucosal tissues and induction of p47 inducibly expressed GTPase IGTP (Irgm3), a molecule required to survive acute infection. Surprisingly, T cell-derived IFN-γ production, a key event in adaptive immunity to the parasite, reached normal levels later during acute infection, although initial production was delayed. To test the ability of MyD88 KO mice to generate protective immunity to T. gondii we used a genetically engineered avirulent parasite strain that invades but cannot replicate in host cells. Vaccination with this mutant elicited fully functional Th1 cell differentiation and immunity to challenge infection. Our results show that during T. gondii infection, MyD88 signaling is essential for host microbicidal function, but that it is dispensable for the adaptive immune response to this intracellular pathogen.

Female mice (6–12 wks of age) were used throughout these studies. IL-12p40^−/− mice and C57BL/6 controls were purchased from The Jackson Laboratory. MyD88^+/+ and MyD88^−/− mice generated by S. Akira (Osaka University, Osaka, Japan) and provided by Dr. E. Pearlman (Case Western Reserve University, Cleveland, OH) were generated by crosses of heterozygous mice. To genotype F₁ litters, tail snips were lysed in the Lysis Reagent with Proteinase K according to the manufacturer’s instructions (Viagen Biotech). DNA in digests were subjected to RT-PCR analysis to determine the mouse genotype using a protocol supplied by Dr. E. Pearlman (Case Western Reserve University, Cleveland, OH). Primers used for sequence amplification were MyD88 (forward) 5′-TGG-CAT-GCC-TCC-ATC-ATA-GTT-AAC-C-3′, (reverse) 5′-GTC-AGA-AAC-AAC-CAC-CAC-CAT-GC-3′, and (neo) 5′-ATC-GCC-TTC-TAT-CGC-CTT-CTT-GAC-G-3′. Wild-type (WT) and KO status of the animals was confirmed after the mice were euthanized in experiments. Littermate female age-matched mice between 6 and 12 wk of age were used for experiments. The mice were housed under specific pathogen-free conditions at the Transgenic Mouse Facility, College of Veterinary Medicine, Cornell University, overseen by the Institutional Animal Care and Use Committee.

Cysts of the low virulence ME49 strain were obtained from brains of chronically infected Swiss Webster mice. Infections were conducted by oral gavage with a blunt-ended needle. Except where noted, 20 cysts in a volume of 0.2 ml of PBS were administered. Tachyzoites of the RH and cps1-1 strain were maintained in vitro on human foreskin fibroblasts. For the case of cps1-1, 0.3 mM uracil was added to the growth medium. Vaccination with cps1-1 was accomplished by i.p. injections of 2 × 10⁴, 2 × 10⁵, and 2 × 10⁵ tachyzoites at biweekly intervals. Two weeks after the final injection, animals were challenged by s.c. injection of 2000 RH strain tachyzoites.

Tissues were collected, preserved in 10% neutral-buffered formalin, and submitted to the Histology Unit, College of Veterinary Medicine (Cornell University, Ithaca, NY) to be processed into paraffin-embedded blocks for H&E staining. Lesions were scored by blind manner and graded based on severity of inflammation (minimal, mild, moderate, severe), number of inflammatory foci, distribution of inflammation, type of inflammatory reaction, and severity of necrosis.

Sections of formalin-fixed paraffin-embedded tissues were stained with a rabbit anti-Toxoplasma antiserum by the Histology Unit, College of Veterinary Medicine (Cornell University, Ithaca, NY). Other sections were stained with rabbit anti-myeloperoxidase or rabbit anti-IGTP (36) using a standard immunoperoxidase staining protocol. Briefly, sections were deparaffinized and rehydrated by serial immersion in xylene and graded alcohols, and finally water. Exogenous peroxidase activity was quenched by incubation in 0.5% H₂O₂ followed by microwave treatment in citrate buffer (0.1 M citric acid in PBS, pH 6.0 to unmask Ags. Sections were blocked in 10% normal serum in casein-blocking reagent (20 min, 20°C; Vector Laboratories). Primary Ab was diluted in PBS with casein-blocking reagent, and applied to the tissue (2 h, 37°C). Sections were washed four times in PBS with 0.4% Brij (Sigma-Aldrich). A biotinylated secondary Ab was diluted in PBS and incubated with the tissue (20 min, 20°C). Sections were washed three times (PBS with 0.4% Brij). Biotin was detected using streptavidin-conjugated peroxidase (20 min, 20°C). Slides were subsequently washed three times (PBS with 0.4% Brij) and visualized with 3-amino-9-ethylcarbazole (10 min, 20°C, AEC chromagen; Vector Laboratories). Sections were rinsed in water, counterstained in Gill’s Hematoxylin (Vector Laboratories), and mounted and examined by brightfield microscopy. The specificity of IGTP (Irgm3) staining was confirmed by examination of tissues from Irgm3/Igtp^−/− mice.

For phenotypic analysis of splenocytes and mesenteric lymph node (MLN) cells, cell surface staining was accomplished as previously described (37) using the following fluorochrome-conjugated mAbs: anti-CD11c-PE, anti-Gr-1-PerCP-Cy5.5, anti-B220-PE, anti-CD8-PE-Cy7, anti-CD4-FITC, anti-CD44-allophycocyanin (all from BD Pharmingen), anti-F480-allophycocyanin, anti-CD3-allophycocyanin (all from Invitrogen), anti-CD62 ligand (CD62L)-PE, anti-CD25-PE, anti-CD69-PE (all from eBioscience). Intracellular cytokine staining was accomplished by incubating single cell suspensions in complete DMEM (DMEM supplemented with 10% bovine growth serum, 1 mM sodium pyruvate, 0.1 mM nonessential amino acids, 0.05 mM 2-ME, 100 U/ml penicillin, 100 μg/ml streptomycin, and 30 mM HEPES) with or without soluble tachyzoite Ag (STAg) or cps1-1. GolgiPlug (BD Biosciences) was added for the last 4 h of culture. Cells were washed and surface stained for CD4 and CD8 (20 min, 4°C). Cells were then fixed and permeabilized in Cytofix/Cytoperm solution (BD Biosciences) according to the manufacturer’s instructions. Cells were washed in Perm/Wash (BD Biosciences) and resuspended in intracellular staining buffer containing 10% normal mouse serum, 10% rat IgG, and optimal concentrations of fluorochrome-tagged anti-cytokine Abs in Perm Wash (60 min, 4°C). Cells were washed twice in Perm/Wash, resuspended in FACS buffer and immediately analyzed on a FACSCalibur flow cytometer (BD Biosciences) collecting at least 50,000 events per sample. Data were subsequently analyzed using FlowJo software (Tree Star).

Single cell suspensions were prepared, and RBC were removed by using RBC lysis buffer according to the manufacturer’s protocol (Sigma-Aldrich). Cell debris was removed by filtering through a 40-μm cell strainer. Cells were washed with PBS and resuspended in complete DMEM. Viable cell number was determined by trypan blue exclusion. For cytokine protein measurement, cells were plated in a 96-well plate (10⁷/ml). The STAg or cps1-1 tachyzoites were added and cells were incubated at 37°C in humidified 5% CO₂ for 48 and 72 h. Supernatants were recovered for ELISA.

Bone marrow-derived macrophages were prepared as described (38). Cells were plated for 2 h and incubated with or without 100 ng/ml recombinant IFN-γ (PeproTech) in complete DMEM overnight. Cells were then infected with type II PTG strain T. gondii tachyzoites at a multiplicity of infection from 0.5 to 1. After 12 h, cells were harvested and resuspended in 10% normal mouse serum. Surface staining was conducted using anti-F4/80-allophycocyanin (Invitrogen). Cells were fixed and permeabilized with Cytofix/Cytoperm solution (BD Biosciences) according to the manufacturer’s protocol. After washing, anti-p30-FITC Ab (Argene) was used to stain T. gondii. Cells were washed twice and resuspended for flow cytometric analysis.

Mice were administered 0.5 mg of rat IgG (Jackson ImmunoResearch Laboratories) or anti-IL-12 C17.8 Ab by i.p. injection on days −1, 0, 2, 4, and 6 after oral infection with 20 ME49 cysts. Serum and spleens were collected on days 4 and 7 postinfection. Splenocytes were cultured with 3 μg/ml STAg for 72 h.

Cytokine ELISA for IFN-γ and IL-12/23p40 were conducted as previously described (39).

Student’s t test was used to analyze statistical differences between groups. Values for p < 0.05 were considered significant. Survival curves were analyzed using a log-rank test. All experiments were repeated a minimum of two to three times.

It was previously shown that genetic inactivation of Toll/IL-1R adaptor molecule MyD88 rendered mice highly susceptible to i.p. infection with Toxoplasma (14, 15). We sought to determine whether lack of MyD88 also resulted in susceptibility during infection initiated by the natural route of infection, namely, through oral inoculation. As shown in Fig. 1,A, MyD88^−/− mice, but not MyD88^+/+ littermates, uniformly succumbed to infection after oral administration of low virulence ME49 cysts. Interestingly, in parallel groups of mice i.p. infected with the same cyst number, MyD88^−/− mice succumbed at a slightly, but significantly (p = 0.0015) accelerated rate (Fig. 1 B). In H&E staining of gut and Peyer’s patches we failed to detect major differences in pathology of WT and KO mice, but in livers we found MyD88-dependent inflammatory infiltrates (data not shown).

Death during i.p. infection of MyD88^−/− animals is associated with uncontrolled tachyzoite replication and dissemination (14, 15). Likewise, we found greater numbers of parasites during low-dose oral infection of MyD88^−/− relative to MyD88^+/+ littermates. In WT mice, low numbers of tachyzoites were present in lamina propria (Fig. 2,A) and Peyer’s patches (Fig. 2,C) of the small intestine. In contrast, high parasite numbers were present at these locations in KO animals (Fig. 2, B and D, respectively). Regions of lymphocyte depletion were also apparent in Peyer’s patches of KO mice (Fig. 2,D) and these areas were markedly less extensive in WT animals (Fig. 2 C).

We assessed in vitro infection in the presence and absence of MyD88 using IFN-γ-activated bone marrow-derived macrophages from WT and KO strains. As shown in Fig. 3, cells from MyD88 KO mice were equally able to limit infection relative to WT macrophages. Thus, in WT cells there was a 74% reduction in infection, compared with 69% in MyD88 KO macrophages after overnight incubation in the presence of IFN-γ.

We next examined MLN and spleen cell populations of WT and KO mice undergoing oral T. gondii infection. In MLNs of day 4-infected animals, there was a striking decrease in Gr-1⁺ neutrophil recruitment in KO relative to WT animals, but populations of macrophages, DCs, as well as B and T lymphocytes were normal (Fig. 4,A). Nevertheless, neutrophil numbers in MLN cells recovered in MyD88^−/− mice by day 7 postinfection (Fig. 4,B). Similarly, neutrophil numbers in spleens of Day 4-infected KO animals were less than in WT littermates (Fig. 4,C), but this difference was minimal at day 7 postinfection (Fig. 4,D). We next stained sections of small intestine with Ab to myeloperoxidase, an enzyme that is expressed at high level in PMN. We found an influx of myeloperoxidase-positive cells into the lamina propria of WT (Fig. 4,E) but not MyD88 KO animals (Fig. 4F) at day 4 postinfection. In day 7-infected mice, lamina propria neutrophils were recruited in greater numbers in WT mice relative to day 4 (Fig. 4,G), and this response did not require functional MyD88 (Fig. 4,H). We, and other researchers, previously reported rapid recruitment of Gr-1⁺ neutrophils into the peritoneal cavity early during i.p. infection with high virulence RH strain tachyzoites (40, 41). In Fig. 4 I, we show that neutrophil recruitment in this model is also highly MyD88-dependent.

During in vitro parasite Ag stimulation of splenocytes (Fig. 5, A–C) and MLN cells (Fig. 5, D–F), production of IL-12/23p40 was highly MyD88-dependent using as a cell source noninfected (Fig. 5, A and D), day 4 (Fig. 5, B and E), and day 7 (Fig. 5, C and F) orally infected mice. Nevertheless, in splenocytes at all time points low but detectable levels of MyD88-independent p40 could be detected, and at day 7 postinfection residual IL-12 was apparent in MLN cultures from MyD88^−/− animals.

Next, we measured production of IFN-γ, a cytokine long recognized as the major mediator of resistance to T. gondii (42). As shown in Fig. 6, Ag-induced IFN-γ release by splenocytes (Fig. 6,A) and MLN cells (Fig. 6,B) was strictly dependent upon MyD88 at day 4 postinfection. This result was confirmed and extended by intracellular cytokine staining of STAg-stimulated cells (Fig. 6,C). Thus, in the spleen, a small proportion of CD4⁺ T cells produced IFN-γ with dependence on MyD88 (0.4 vs 0.03% for WT and KO, respectively). Interestingly, there was a strong MyD88-dependent IFN-γ response by splenic CD8⁺ T cells under the same conditions (10.0 vs 0.57% for WT and KO, respectively). However, in MLN cultures, the STAg-induced IFN-γ response was relatively minor, and most of the MyD88-dependent IFN-γ derived from CD4⁺ T lymphocytes. It is of further interest to note evidence for a default to the Th2 pathway in MyD88^−/− MLN CD4⁺ T cells (0.09 vs 0.56% IL-4⁺ in WT and KO, respectively) that was not apparent in splenic cultures (Fig. 6 C). Thus, as previously reported in an i.p. model of Toxoplasma soluble Ag injection (43), there is an increase in the Th2 response in the absence of MyD88, but in this experiment we show this response is weak and is restricted to MLNs during oral infection.

When we examined IFN-γ responses at day 7 after oral infection, strikingly different results were obtained compared with the day 4 response. Thus, at day 7, splenocytes (Fig. 7,A) and MLN cells (Fig. 7,B) released high levels of IFN-γ independently of MyD88 and without the need for further in vitro Ag stimulation. In the presence of parasite Ag, these responses were only modestly increased. In Fig. 7,C, we determined the source of IFN-γ by ex vivo staining splenocytes and MLN cells without further Ag stimulation. In this case, there was a vigorous MyD88-independent splenic CD4⁺ T cell response (6.56 and 6.87% for WT and KO, respectively) (Fig. 7 C). In MLNs, there was also a strong CD4⁺ T lymphocyte IFN-γ response, although this was partially MyD88-dependent (7.49 vs 3.07% for WT and KO, respectively). In both spleen and MLNs, the IFN-γ response by CD8⁺ T cells was extremely weak (data not shown).

To determine whether the IFN-γ responses we measured at days 4 and 7 depended upon endogenous IL-12 production, we administered depleting IL-12p40 Ab and examined IFN-γ production over the course of infection. As shown in Fig. 8, the MyD88-dependent day 4 response was abrogated by IL-12 depletion. In marked contrast, the MyD88-independent day 7 response was normal in the presence of Ab. The latter result suggests that emergence of IFN-γ-producing Th1 cells at day 7 does not require MyD88 or IL-12.

We also examined activation markers on CD4⁺ and CD8⁺ T cells during infection. There were no significant differences in expression of CD69 (Fig. 9) or CD25 (data not shown) over the course of infection.

Previously it was found that IFN-γ-dependent expression of p47 GTPase family member IGTP (Irgm3) is necessary to survive Toxoplasma infection (44). In this experiment, we determined the expression pattern of this molecule in small intestines of orally inoculated MyD88^−/− and MyD88^+/+ mice. Interestingly, at day 4 postinfection we detected expression of Irgm3 in endothelial cells in basement membrane-submucosal regions of WT mice (Fig. 10, A and C) and this response was absent in mice lacking MyD88 (Fig. 10, B and D). However, at day 7 after infection, small intestine epithelial cells of WT mice were strongly positive for Irgm3, and this response was identical in tissues from MyD88 KO animals (Fig. 10, E and F). To summarize, during early (day 4) oral infection, IL-12/23p40 and T cell IFN-γ expression was dependent upon MyD88. Similarly, neutrophil recruitment and Irgm3 expression, two responses previously implicated in resistance to Toxoplasma (44, 45), also relied on functional MyD88. Nevertheless, with the exception of IL-12/23p40 expression, each of these immune responses reached normal levels without the need for MyD88 by day 7 after infection.

Although the delayed T cell IFN-γ response did not require MyD88 during oral infection (Fig. 7), and although CD4⁺ and CD8⁺ T lymphocyte expression of activation marker CD69 was identical in the presence and absence of MyD88 (Fig. 9), KO mice were unable to control infection and the animals succumbed to Toxoplasma (Fig. 1). This precluded us from rigorously testing whether MyD88^−/− mice developed a protective immune response to challenge infection, which is known to depend upon T cell production of IFN-γ in normal mice (30). To circumvent this limitation, we used Toxoplasma mutant cps1-1. This RH-derived parasite strain is a uracil auxotroph that invades but does not replicate in host cells. Infection of mice with cps1-1 results in nonpersistent infection that fails to cause pathology associated with WT T. gondii infection (46). Importantly, cps1-1 vaccination has been shown to induce protective immunity to challenge infection (46, 47). As shown in Fig. 11,A, both MyD88^+/+ and MyD88^−/− mice vaccinated with cps1-1 developed protective immunity to challenge with RH strain parasites. In contrast, nonvaccinated mice succumbed within 2 wk of infection with this highly virulent parasite strain. We also vaccinated IL-12/23 p40^−/− mice with cps1-1 (Fig. 11,B). In this case, the animals failed to develop a protective immunity to RH challenge. Finally, in Fig. 11 C, we orally challenged cps1-1-vaccinated mice and found too that MyD88 was not required for resistance to lethal ME49 infection.

We examined naive (CD62L^highCD44^low), effector/effector memory (CD62L^lowCD44^high), and central memory (CD62L^highCD44^high) splenic T cell populations after cps1-1 vaccination of MyD88^+/+ and MyD88^−/− animals. As shown in Fig. 12,A, there were no major differences between WT and KO populations among CD4⁺ (A and B) or CD8⁺ (C and D) T cell populations. To distinguish between effector and effector memory populations that both possess a CD62L^lowCD44^high surface phenotype, we assessed CCR7 that is expressed at a high level on effector but not effector memory subpopulations. Again, we found no difference in these populations between WT and KO T cells (Fig. 12 B).

In splenocyte cultures from cps1-1 vaccinated mice, production of IL-12/23p40 was barely detectable in cells from MyD88 KO compared with the response of WT cells (Fig. 12,C). In contrast, the IFN-γ recall response of MyD88^−/− cells was indistinguishable from that of cells from WT littermates (Fig. 12,C). We further probed the source of IFN-γ using intracellular cytokine staining, and found that both CD4⁺ and CD8⁺ T cells from KO and WT mice contributed to IFN-γ production during in vitro recall with cps1-1 parasites (Fig. 12 D). There was no significant difference between responses of WT compared with KO when responses of several mice were compared. Thus, cps1-1 vaccination of MyD88^−/− mice results in protective adaptive immunity that, by the parameters measured, appears identical with that induced in MyD88^+/+ animals.

TLR-MyD88 signaling is required for resistance to several protozoan pathogens, and parasite ligands that possess TLR activating capability are increasingly being identified (10). For the case of Toxoplasma, parasite profilin and tachyzoite surface GPI anchors are recognized by TLR11 and TLR2/4, respectively (18, 32, 33). In vitro studies using models of i.p. infection or i.p. injection of STAg lysate have shown that MyD88 is required to control infection, and that the immune response to the parasite deviates from a Th1 to a Th2 response in the absence of this TLR adaptor (14, 43).

In this study, we examined how lack of MyD88 signaling impacts host defense in animals undergoing oral infection with T. gondii. As during i.p. infection, orally infected MyD88^−/− mice displayed early death associated with uncontrolled parasite replication. Like other studies using the i.p. infection model, the IL-12 response was severely curtailed during oral infection of MyD88 KO mice. Early during oral infection of MyD88^−/− animals, we found that neutrophil recruitment to mucosal tissues was defective, as was production of T cell IFN-γ and induction of IGTP, an IFN-γ-inducible p47 GTPase required to survive acute T. gondii infection (44). Importantly, later during acute infection, MyD88-independent Th1 responses emerged. The appearance of an apparently intact, albeit delayed, Th1 response in the absence of MyD88 prompted us to examine whether we could elicit protective immunity in the absence of this signaling adaptor. Accordingly, we vaccinated mice with the avirulent cps1-1 Toxoplasma strain. This treatment stimulated protective immunity and strong IFN-γ production independently of MyD88. Taken together, our data suggest MyD88 signaling exerts anti-Toxoplasma activity at a critical early time in infection, but that it is not required to induce strong Th1-based immunity to this parasitic pathogen.

Our results were unexpected because it has been reported that MyD88 signaling plays a role in shaping adaptive immunity to T. gondii. For example, during Toxoplasma infection i.p., defective T cell IFN-γ production was observed during acute infection (14). Following i.p. injection of STAg lysate, the immune response in MyD88^−/− animals deviates from a Th1 to Th2 cytokine response (43). In addition, i.p. injection of parasite profilin has been shown to trigger a TLR11/MyD88-dependent Th1 response to the profilin molecule itself (35). Recently, it was found that MyD88 expression in T cells was required for IFN-γ production and survival during i.p. T. gondii infection (48). We do not completely understand why MyD88^−/− animals undergoing oral T. gondii infection develop strong, albeit delayed, Th1 responses, when other routes of infection and Ag delivery elicit MyD88-dependent Th1 responses. It seems possible that animals succumb with insufficient time to generate a Th1 response during i.p. infection. Another possibility is that DC populations in mucosal tissues, or possibly signals elicited by gut flora in combination with Toxoplasma, play a role in triggering MyD88-independent Th1 immunity during oral infection (49, 50). For the case of protective immunity induced by cps1-1 vaccination, it might be that multiple exposure to low doses of nonvirulent tachyzoites induces MyD88-independent adaptive immunity that is not elicited by i.p. infection with a single dose of parasites.

Our results suggest that the major role for MyD88 during oral Toxoplasma infection is in antimicrobial effector activity, either by mediating recruitment of microbicidal effector cells or by direct antimicrobial function. This suggestion is in line with several other groups. For example, T cell-dependent Ab responses induced by a panel of adjuvants proceeds normally in the absence of MyD88 and TRIF signaling (51). Similarly, MyD88-deficient mice mount normal protective Ab responses and inflammatory arthritis during B. burgdorferi infection (27). During infection with the fungal pathogen Aspergillus fumigatus Th1 responses in infected airways do not require MyD88, although T-bet induction was enhanced by MyD88 signaling in the draining lymph nodes (52). It was also found that Listeria infection induces a protective CD8⁺ T cell response in the absence of MyD88 signaling (25, 26). During aerosol infections with M. tuberculosis, there is evidence both for and against a role for MyD88 in adaptive immunity, depending on the study (9, 53, 54, 55, 56). Regardless, in most of these cases, and as during oral Toxoplasma infection, whereas acquired immunity could proceed without MyD88, this adaptor molecule was clearly required for pathogen control during innate immunity.

Of relevance to the present work is a recent study that examined responses to oral Toxoplasma infection in the absence of TLR9, a receptor whose signaling pathway depends upon MyD88. In that study, mice were more susceptible to infection in the absence of TLR9, and this was associated with decreased T cell IFN-γ responses (57). Although IL-12 was not examined, it is interesting to note that in the previous study, whereas IFN-γ responses were lower in the absence of TLR9, substantial levels of this cytokine were nevertheless still produced.

At present we do not know why MyD88^−/− mice are unable to control T. gondii infection. Recruitment of neutrophils to mucosal tissues was defective in the KO mice during early stages of infection, and this defect possibly underlies the susceptibility of these animals. Others have also found that neutrophil recruitment during infection relies upon MyD88-dependent signaling (58, 59). Previously, we reported that depletion of neutrophils with an anti-Gr-1 Ab resulted in susceptibility to oral T. gondii infection (45). Therefore, it is possible that defective neutrophil recruitment during early infection allows increased parasite survival in the intestine, ultimately leading to death of the animals. Nevertheless, we interpret results obtained using anti-Gr-1 mAb with caution because the Gr-1 epitope is also expressed by other types of myeloid cells (60).

Our data for the first time show tissue localization of IFN-γ-inducible IGTP, a molecule required to survive acute Toxoplasma infection (44). Interestingly, IGTP was expressed in endothelial cells of the intestinal submucosa in an MyD88-dependent manner during early infection. Later, this p47 GTPase was highly expressed in epithelial cells of the small intestine in both MyD88^+/+ and MyD88^−/− mice. It is difficult to understand how early MyD88-dependent IGTP expression confined to endothelial cells could contribute to resistance to T. gondii infection, which is uniformly spread throughout mucosal tissues. Although the function of IGTP (Irgm3) and other p47 GTPases is not well understood, these molecules have been implicated in autophagic destruction of the parasitophorous vacuole in tachyzoite-infected macrophages and astrocytes (61, 62). However, we have so far not detected differences in IFN-γ-mediated macrophage killing of tachyzoites in the presence and absence of MyD88.

Strong MyD88-independent T cell IFN-γ responses were elicited during both oral ME49 infection and cps1-1 vaccination. Inasmuch as production of IL-12 in both cases was minimal, this raises the question of how T cell IFN-γ responses are generated in these cases. Two related possibilities are that residual MyD88-independent IL-12 is sufficient to drive Th1 differentiation, or that more robust MyD88-independent IL-12 responses occur at sites other than the draining lymph nodes and spleen that were examined in this study. It is also possible that Th1 responses are triggered independently of IL-12 in the absence of MyD88. Indeed, when we administered depleting anti-IL-12 Ab, whereas the day 4 MyD88-dependent IFN-γ response was severely curtailed, there was no effect on the day 7 MyD88-independent response. Nevertheless, at least for the case of cps1-1-induced immunity, absence of IL-12/23p40 resulted in lack of a protective response, a result that argues for IL-12 as a mediator of protection in cps1-1 vaccinated MyD88^−/− animals. In this regard, we have found that bone marrow-derived macrophages infected with the Toxoplasma RH strain produce IL-12 independently of signaling through MyD88 (63), yet the same parasite strain clearly triggers MyD88-dependent IL-12 in splenic DCs (our unpublished results) (14, 18). Further work will be required to determine host mediators involved in generating MyD88-independent Th1 responses during Toxoplasma infection.

In summary, our results argue that MyD88-dependent signaling is crucial in antimicrobial effector function against T. gondii. Nevertheless, the view that this adaptor molecule plays an essential role in bridging innate and adaptive immunity to Toxoplasma may require revision. Other signaling molecules and recognition systems are also likely to be important determinants of adaptive immunity to this parasite. Identifying such molecules will provide important insight into immune recognition of Toxoplasma and other microbial pathogens.

We are grateful to S. Akira for supplying us with MyD88^−/− mice, and P. Fisher for assistance with immunohistochemistry.

The authors have no financial conflict of interest.