MyD88-Dependent Signaling Contributes to Host Defense against Ehrlichial Infection

The ehrlichiae are small Gram-negative obligate intracellular bacteria in the family Anaplasmataceae. Ehrlichial infection in an accidental host may result in fatal diseases such as human monocytotropic ehrlichiosis, an emerging, tick-borne disease. Although the role of adaptive immune responses in the protection against ehrlichiosis has been well studied, the mechanism by which the innate immune system is activated is not fully understood. Using Ehrlichia muris as a model organism, we show here that MyD88-dependent signaling pathways play a pivotal role in the host defense against ehrlichial infection. Upon E. muris infection, MyD88-deficient mice had significantly impaired clearance of E. muris, as well as decreased inflammation, characterized by reduced splenomegaly and recruitment of macrophages and neutrophils. Furthermore, MyD88-deficient mice produced markedly lower levels of IL-12, which correlated well with an impaired Th1 immune response. In vitro, dendritic cells, but not macrophages, efficiently produced IL-12 upon E. muris infection through a MyD88-dependent mechanism. Therefore, MyD88-dependent signaling is required for controlling ehrlichial infection by playing an essential role in the immediate activation of the innate immune system and inflammatory cytokine production, as well as in the activation of the adaptive immune system at a later stage by providing for optimal Th1 immune responses.

Human monocytotropic ehrlichiosis (HME) is an emerging, tick-borne disease caused by Ehrlichia chaffeensis, which is an NIAID Category C priority pathogen [20,21]. E. chaffeensis infection in human with delayed or ineffective treatment may be fatal, particularly in immunocompromised hosts [22,23]. The ehrlichiae are small, Gram-negative, and obligate intracellular bacteria in the family Anaplasmataceae and reside in endosomal compartments, primarily in macrophages or monocytes. Ehrlichiae frequently cause persistent infection in their natural hosts, which appear to utilize immune defense mechanisms against species-specific ehrlichia strains. Some examples include E. canis in dogs, E. chaffeensis in deer, E. muris in mice, and E. ruminantium in sheep, goats and cattle [20,24]. However, ehrlichiae infection in accidental hosts may result in severe fatal disease, similar to toxic shock-like disease, as seen in in E. chaffeensis infection in humans or infection of the IOE (Ixodes ovatus) Ehrlichia strain in mice [25,26,27].
E. muris, a natural pathogen in mice, is closely related to E. chaffeensis and is frequently used as an Ehrlichia infection model in mice [28]. Infection of E. muris in a laboratory strain of mice such as BALB/c, AKR or C57BL/6 results in splenomegaly, anemia, and splenic follicular hyperplasia followed by the development of a low-level persistent infection [28,29]. Both CD4 and CD8 T cells play important roles in the immune response against Ehrlichia, including IFN-c production [30,31]. Outer membrane proteins (OMPs) of the Ehrlichia membrane play important roles in both pathogenesis and host defense, as they are the major targets of protective antibodies and Ehrlichia-specific T cells [32,33]. NKT cells are also activated during E. muris infection, primarily through direct recognition of microbial lipids, which takes place in a MyD88-independent manner [34].
Although E. muris epitopes for B and T cell activation have been characterized, neither the E. muris ligand(s), nor the corresponding host receptor(s), that activate the innate immune system are known [32,33]. E. muris can induce inflammation in mice, but it appears that E. muris lacks known TLR and NLR ligands, with the exception, perhaps, of bacterial nucleic acids. Ehrlichiae lack the genes required for the biosynthesis of LPS, peptidoglycan and flagellin, which can elicit an innate immune response [35,36] In this study, we have found that MyD88-dependent signaling is essential for protection against E. muris infection. Upon E. muris infection, dendritic cells (DCs), but not macrophages, produce large amounts of IL-12, which is important in driving the Th1 immune responses that are crucial for controlling E. muris infection.

MyD88-dependent protection against E. muris
Although ehrlichiae infect and cause inflammation in humans, mice, and many other mammals, its recognition mechanism remains unknown. We used MyD88-deficient mice to investigate whether TLRs play a role in the detection of and protection against E. muris. We infected MyD88-deficient mice with E. muris by i.p. inoculation and assessed bacterial load in blood and tissues by quantitative realtime PCR (qPCR) using primers specific for the E. muris dsb gene. In a preliminary study, we found that the E. muris load did not significantly increase by day 6, but increased and peaked on day 10. Therefore, we designated the first 10 days post infection as the acute phase, and the following days up to day 14 as the convalescent phase of infection, which corresponds with previous reports [29,37]. MyD88-deficient mice demonstrated a higher E. muris bacterial load in the blood in both phases ( Figure 1A). In wild-type mice, E. muris was cleared gradually from the spleen during the convalescent phase. In contrast, MyD88-deficient mice had significantly increased bacterial loads at both day 10 and 14 post infection ( Figure 1A). We measured the cytokine production in the serum of infected mice. Strikingly, there was a high level of IL-12 p40 and IL-12 p70 in the serum of wild-type mice infected with E. muris, whereas MyD88-deficient mice produced a significantly lower level of these cytokines ( Figure 1B). We did not observe statistically significant differences in TNF-a and IL-6 levels in the serum of wild-type and MyD88-deficient mice ( Figure 1B).
E. muris infection in wild-type mice caused severe splenomegaly that progressively became apparent after infection (Figure 2A and B). MyD88-deficient mice, however, showed a milder degree of splenomegaly (Figure 2A and B). Macroscopically, lymphoid follicles in wild-type mice were smaller and markedly irregular in shape compared to MyD88-deficient mice after infection. In contrast, lymphoid follicles in MyD88-deficient mice were larger, more uniform and distinct in shape. Additionally, the red pulp of the spleen significantly expanded in wild-type mice but only to a mild degree in MyD88-deficient mice ( Figure 2C). Microscopically, this expansion was accompanied by an increased infiltration of neutrophils and macrophages into the marginal zone and the red pulp of the spleen with numerous apoptotic cells ( Figure S1). Larger numbers of apoptotic cells and lymphoblasts were observed in the lymphoid follicles in wild-type mice than in MyD88deficient mice ( Figure S1). These overall histopathological findings in wild-type mice subjected to E. muris infection were consistent with a previous report [29]. In order to quantify the increased number of neutrophils and macrophages, we analyzed Gr-1 + neutrophils and F4/80 + macrophages by flow cytometry. The number of neutrophils and macrophages increased significantly in wild-type mice 10 days after E. muris infection, whereas the number of those cells increased only mildly in MyD88-deficient mice ( Figure 2D). In summary, these results indicate that MyD88dependent signaling plays an important role in the protection against E. muris infection by inducing inflammatory responses accompanied by IL-12 production.
Reduced Th1 response in MyD88-deficient mice during E. muris infection Since IL-12 is an important cytokine in the development of effector Th1 cells from helper CD4 T cells, reduced production of IL-12 in the serum of MyD88-deficient mice suggests that they may have an insufficient Th1 response [38]. Earlier reports have shown that a Th1 immune response is required for efficient ehrlichia clearance in mice [31,39]. Therefore, we isolated splenocytes before and 10 days after infection of mice with E. muris, and IFN-c production in gated splenic CD4 T cells was analyzed by flow cytometry. There was no significant change in the ratio of CD4 to CD8 splenic T cells during E. muris infection between wild-type and MyD88-deficient mice ( Figure S2). However, while the percentage of IFN-c producing CD4 T cells increased 10 days after infection in wild-type mice, the increase of IFN-c producing cells was considerably less in MyD88-deficient mice ( Figure 3A). In order to assess Th1 differentiation ex vivo, we purified CD4 T cells from spleens 10 days after infection and restimulated them with plate-bound antibodies against CD3 or a combination of CD3 and CD28 at different concentrations. As shown in Figure 3B, the production of IFN-c was prominent in CD4 T cells from wild-type mice while CD4 T cells from MyD88-deficient mice produced significantly less IFN-c. This cannot be attributed to insufficient IL-2 production in MyD88-deficient T cells, as wild-type and MyD88-deficient CD4 T cells produced similar levels of IL-2 after TCR ligation ( Figure 3B), suggesting that MyD88-deficient T cells retain the intrinsic ability to produce cytokines. Indeed, MyD88-deficient CD4 T cells produced more IL-4 than wildtype CD4 T cells upon TCR ligation ( Figure 3B). Taken together, these data suggest that MyD88 is required for Th1 responses during E. muris infection.

DCs but not macrophages produce IL-12 efficiently upon E. muris infection
In order to find the major source of IL-12 production during E. muris infection, we generated bone marrow-derived DCs and macrophages from C57BL/6 mice and challenged them with E.  Figure S3A). Furthermore, E. muris proliferated efficiently in both DCs and macrophages, as quantified by qPCR ( Figure S3B). Interestingly, while DCs could produce IL-12 p40 efficiently upon E. muris infection, macrophages could produce only a small amount of IL-12 p40 ( Figure S3C). Both cell types produced only small amounts of TNF-a and IL-6 upon E. muris infection ( Figure S3C). Collectively, these data suggest that DCs, but not macrophages, can serve as a major source of IL-12 production during E. muris infection.
The Nod-Rip2 pathway does not play a major role in the recognition of E. muris in DCs Since ehrlichiae are obligate intracellular bacteria, we tested whether members of the cytoplasmic NLR protein family have the ability to detect E. muris. Caspase recruitment domain (CARD) containing NLR proteins (NLRC proteins) such as Nod1 and Nod2 require the kinase Rip2 for downstream signaling, which involves NF-kB activation and the induction of proinflammatory cytokines [10,40]. Therefore, in order to reveal the role of Nod1 and Nod2 in the detection of E. muris, we generated bone marrow-derived DCs from Rip2-deficient mice. Cells were infected with E. muris, and cytokine production and bacterial loads were assessed. Levels of proinflammatory cytokines such as IL-12 p40, TNF-a, and IL-6 in wild-type DCs were comparable to that in Rip2-deficient DCs ( Figure  S4A). E. muris infection induced similar numbers of bacteriacontaining vacuoles in wild-type and Rip2-deficient DCs ( Figure  S4B). There was also no significant difference in E. muris bacterial load between wild-type and Rip2-deficient DCs ( Figure S4C). Collectively, these data indicate no or little involvement of Rip2 in E. muris detection, suggesting that the NLR family proteins Nod1 and Nod2 do not play major roles in the recognition of E. muris in DCs.
NLR proteins such as Nalp3 (Nlrp3) or Ipaf (Nlrc4) activate caspase-1, resulting in cleavage and maturation of IL-1b [14,15,16,17,18]. Therefore, we sought to determine whether E. muris infection can induce IL-1b secretion in bone marrow-derived macrophages from wild-type, Ipaf-deficient, Nalp3-deficient and Asc-deficient mice. These cells were infected with E. muris in vitro and IL-1b production was assessed by ELISA. We could not detect IL-1b production in macrophages from any strain of mice, suggesting that caspase-1 activating NLR proteins may not play a major role in the recognition of E. muris ( Figure S5).
MyD88 is required for efficient production of IL-12 from DCs upon E. muris infection Since DCs but not macrophages produce IL-12 p40 efficiently, we analyzed IL-12 p40 production in MyD88-deficient dendritic cells. Both wild-type and MyD88-deficient DCs were infected with E. muris and cytokine production was assessed by ELISA. As shown in Figure 4A, wild-type DCs produced a significantly greater amount of IL-12 p40 than MyD88-deficient DCs upon E. muris infection. Both TNF-a and IL-6 production decreased in MyD88-deficient DCs as compared to wild-type cells, although even wild-type DCs produced relatively low levels of these proinflammatory cytokines ( Figure 4A). The decreased production of IL-12 p40 in MyD88-deficient DCs was not attributable to the difference in number of bacteria in the cells since the formation of bacterial vacuoles and the bacterial load in wild-type and MyD88deficient DCs were comparable ( Figure 4B and C).
Although less able than DCs, macrophages can produce measurable amounts of IL-12 p40 during E. muris infection ( Figure  S3C). Therefore, we challenged wild-type and MyD88-deficient macrophages with E. muris and assessed the production of IL-12 p40. The production of IL-12 p40 was dependent on MyD88 as MyD88-deficient macrophages failed to produce IL-12 p40 ( Figure S6A). Wild-type macrophages could also produce TNF-a and IL-6 while MyD88-deficient macrophages produced these cytokines at a significantly lower level ( Figure S6A). Again, this decrease was not due to failure of E. muris growth in MyD88deficient macrophages since we observed similar numbers of bacteria-containing vacuoles and bacterial loads in wild-type and MyD88-deficient macrophages ( Figure S6B and C).

TLR-dependent recognition of E. muris
MyD88 is required for not only TLR, but also IL-1 and IL-18 receptor signaling [41]. Caspase-1 is required for the cleavage and maturation of IL-1 and IL-18 into their biologically active forms [42,43,44]. In order to define the role of TLRs in the recognition of E. muris, we used caspase-1-deficient cells as a control against IL-1/IL-18 function. Caspase-1-deficient DCs infected with E. muris as well as stimulated with TLR ligands produced levels of IL-12 p40 comparable to those of wild-type DCs ( Figure 5A), indicating that the IL-1/IL-18 family does not play a role in E. muris recognition. It has been reported that endosomal TLRs such as TLR3, 7, 8 and 9 require UNC93B1 for their endolysosomal translocation from the endoplasmic reticulum [45,46]. Since E. muris grows in vacuoles in the cytoplasm, we tested whether E. muris recognition is dependent on UNC93B1, using triple defective (3d) mice in which UNC93B1 is mutated and non-functional, resulting in defective TLR3, 7 and 9 signaling [45]. We found that 3d DCs produced similar levels of IL-12 p40 as compared to wild-type DCs, although 3d DCs did not respond to ligands recognized by TLR9 (CpG DNA) or TLR7 (R-848) ( Figure 5B). Taken together, these data indicate that E. muris recognition by DCs is in a caspase-1-independent and MyD88dependent manner, suggesting that UNC93B1-independent TLR(s) may play a major role in E. muris recognition.

Known TLRs do not appear to play a role in the recognition of E. muris in DCs
To identify which TLR is involved in the recognition of E. muris, we generated DCs from Tlr2, 3, 4, 5, 7, 9 and 11-deficient mice. Those cells were infected with E. muris and analyzed for IL-12 p40 production. Surprisingly, DCs deficient for any TLR showed a normal response to E. muris infection ( Figure S7A and B). It has been shown that TLR1 and TLR6 cooperates with TLR2 for ligand recognition [47]. Therefore, these data indicate that TLR1, 2, 3, 4, 5, 6, 7, 9 and 11 may not play a major role in the recognition and host defense against E. muris infection.

Discussion
Although adaptive immune responses during ehrlichiosis have been well characterized, the mechanism by which the bacteria activate an innate immune response has remained unexplored. Ehrlichiae are small obligate intracellular bacteria and have naturally evolved to hide from the immune system in order to survive in host cells. Intracellular bacteria grow in either in the cytoplasm or in specialized organelles that do not fuse with lysosomes, and must have strategies to escape from innate immune surveillance. Such mechanisms include the deletion or conversion of ligands into undetectable forms, rendering the organisms ''invisible'' from the innate immune system of host cells. For example, Coxiella burnetii evades TLR4 by expressing a nonstimulatory lipid A and also lacks flagellin, a ligand for TLR5 and Ipaf [48,49]. Similarly, ehrlichiae do not possess LPS, flagellin and peptidoglycan, but are still able to induce inflammation, indicating that the mammalian innate immune system can sense ehrlichiae. In this study, we showed that MyD88 is required for host protection against E. muris infection. MyD88-deficient mice suffered from increased loads of E. muris in both blood and the spleen, which was accompanied by a decreased inflammatory response in the spleen, as well as reduced IL-12 production (Figure 1 and 2). Splenomegaly and recruitment of macrophages/ neutrophils were less prominent in MyD88 deficient mice, suggesting that impaired inflammatory responses in MyD88 deficient mice correlate with less efficient bacterial clearance ( Figure 2). MyD88 is also important for the induction of a proper Th1 response during E. muris infection, as shown in both in vivo and ex vivo assays wherein CD4 T cells from MyD88-deficient mice produced less IFN-c ( Figure 3). Therefore, since it is likely that a Th1 response, in addition to the recruitment of myeloid cells, is important for host protection against ehrlichial infection, our data suggests that MyD88-mediated recognition of ehrlichiae is important for optimal induction of Th1 immune responses.
IL-12 is an important cytokine that can induce the differentiation of Th1 effector cells from activated CD4 T cells [38]. Our data suggest that during E. muris infection, DCs may be a major source of IL-12. Bone marrow-derived DCs produce a large amount of IL-12 p40 in a MyD88-dependent manner, although bone marrow-derived macrophages produced only a small amount of IL-12 p40 in in vitro ehrlichial infection (Figure 4 and Figure S3). It has been reported recently that antigen presentation by DCs plays an important role in the immune response against ehrlichial infection [50]. Our findings indicate that in addition to ehrlichial antigen presentation, DCs may also contribute to a strong host defense against ehrlichia infection by driving strong Th1 immune responses.
Our data suggest that TLR(s) play a role in the recognition of E. muris since MyD88, but not caspase-1, is required for a response against E. muris infection in DCs (Figure 4 and 5). Recently it was reported that caspase-1 cleaves Mal/Tirap to produce its functional form. TLR2 and TLR4 responses in DCs may not require caspase-1 because a non-cleaved Mal/Tirap may still be functional in DCs, as suggested previously [51]. The identity of the TLR responsible for the recognition of ehrlichiae remains a pressing and intriguing question. Excluding bacterial DNA/RNA, ehrlichiae do not possess known TLR ligands. However, Tlr3, Tlr7 and Tlr9-deficient dendritic cells respond normally to E. muris infection. 3d DCs which cannot respond to ligands for TLR3, 7, 8, and 9, also produced a normal response to E. muris infection, suggesting that E. muris is not detected by any of these endosomal TLRs ( Figure 5B and Figure S7B). Our data also showed that DCs from Tlr2, 4, 5 or 11-deficient mice responded to E. muris infection with levels of IL-12 p40 production comparable to DCs from wildtype mice ( Figure S7). As for the remaining known TLRs, it has been shown that TLR1 and TLR6 heterodimerize with TLR2 to facilitate ligand recognition, and there is no murine homolog of TLR10 [47]. These indicate that the eleven TLRs that have been previously characterized may not be involved in E. muris recognition. It is possible that multiple TLRs contribute partially, and the sum of those effects is equivalent to the loss of functionality observed in MyD88-deficient cells. An alternate possibility is that an uncharacterized TLR accounts for the recognition of E. muris. Unlike humans, mice and other lower vertebrates have the TLR11 subfamily, which includes TLR12 and TLR13, whose ligands and functions are unidentified [2]. The TLR11 gene exists in humans but is not functional due to the presence of a stop codon in the leucine rich repeats [52]. It will be interesting to test if these uncharacterized TLRs are responsible for E. muris recognition in the future.
In addition to TLRs, there have been numerous reports of detection of intracellular bacteria by NLR proteins. For example, Listeria monocytogenesis infection activates Nalp3 and Nod2, Legionella pneumophillae infection activates Nod2, Birc1e and Ipaf, and Salmonella typhi infection activates Nalp3 and Ipaf [9,17,18,53,54,55,56,57,58,59]. Interestingly, our data indicate that E. muris seems to be unable to activate known NLRs. E. muris infection in macrophages did not induce IL-1b secretion, indicating that caspase-1 activating NLR proteins such as Nalp3 or Ipaf are not involved in E. muris recognition ( Figure S5). Also, E. muris infection in Rip2-deficient DCs resulted in IL-12 p40 production comparable with wild-type cells, indicating that Nod1 and Nod2 do not play major roles in E. muris recognition ( Figure  S4). Indeed, E. muris possess neither peptidoglycan, which contain the active moieties that are ligands for Nod1 and Nod2, nor flagellin, a ligand for Ipaf.
In conclusion, this study has demonstrated the importance of MyD88-mediated inflammatory and Th1 responses to the host defense against ehrlichial infection. Furthermore, we have identified DCs as a major source of IL-12 during E. muris infection. Although the specific TLR that recognizes E. muris remains unknown, our data suggests that obligate intracellular bacteria can elicit strong innate immune responses via TLR signaling. To our knowledge, this is the first report describing the role of the MyD88 signaling pathway in the recognition of intracellular bacteria belonging to the genus Ehrlichia.

Materials and Methods
Reagents LPS from Salmonella minnesota was from Alexis. Phosphorothioate-modified CpG oligo DNA was synthesized by Integrated DNA Technology. Poly(I:C) was from Amersham Pharmacia Biotech. Flagellin from S. typhimurium and Pam 3 CSK 4 were from Invivo-Gen. R-848 was from GLSynthesis Inc.

Bacterial stocks and in vitro Ehrlichia infection of cells
E. muris, monocytotropic ehrlichial strain (ATCC VR-1411) was used in this study. E. muris was cultivated in DH82 cells with 5% FBS-supplemented DMEM at 37uC. Ehrlichiae were harvested when 90-100% of the cells were infected, and stored 280uC in sucrose-phosphate-glutamate buffer (0.218 M sucrose, 3.8 mM KH 2 PO 4 , 7.2 mM K 2 HPO 4 , and 4.9 mM monosodium glutamic acid, pH 7.0), and used for in vitro infection. The cells were cultured without antibiotics and E. muris bacterial stock was added to the cells. Infected cells were stained by Diff-Quick (Imeb Inc) or Hoechst 33342 (Invitrogen) staining. In order to produce virulent infectious stocks for in vivo infections, live mice were used as described previously [29,39]. Briefly, C57BL/6 mice were inoculated i.p. with 0.5 ml of a 10 21 dilution of the frozen stock. On day 9 after inoculation, the mice were sacrificed, the spleens were harvested, and the homogenate was suspended in sucrosephosphate-glutamate buffer. Large particles of debris were removed by centrifugation at 500 g for 5 min, and the supernatant was then aliquoted and stored at 280uC as a 10 21 stock of E. muris.

Bone marrow-derived macrophages
Bone marrow-derived macrophages were prepared as described [9]. Briefly, bone marrow from tibia and femur was obtained by flushing with DMEM (Invitrogen). The medium was DMEM supplemented with 10% heat-inactivated FBS, glutamine (both from Invitrogen) and 30% L929 cell supernatant containing M-CSF. Bone marrow cells were cultured in 10 ml at an initial density of 1610 6 cells/ml in 100 mm petri dish (Becton Dickinson) at 37uC in humidified 10% CO 2 for 6 days. Cells were harvested with cold PBS (Invitrogen), washed, resuspended in DMEM supplemented with 10% FBS and used at a density of 2610 5 cells/ml for experiments unless mentioned otherwise.

Bone marrow-derived DCs
DCs were grown from wild-type and various knockout mice as described previously [67]. Briefly, bone marrow from tibia and femur was obtained as described above, and bone marrow cells were cultured in RPMI 1640 medium containing 10% heatinactivated FBS, 50 mM of 2-ME, and 2 mM of glutamine supplemented with 3% J558L hybridoma cell culture supernatant containing GM-CSF. The culture medium containing GM-CSF was replaced every other day. At day 6 of culture, nonadherent cells and loosely adherent DC aggregates were harvested, washed, resuspended in RPMI 1640 supplemented with 5% FBS and used at a density of 2610 5 cells/ml for experiments unless mentioned otherwise.

Ehrlichia challenge of mice in vivo
In vivo E. muris infection method was established previously using E. muris stock prepared from the spleen of the infected mice [29,39]. Briefly, age (6-8 weeks old) and sex matched groups of mice were infected i.p. with E. muris with 250 ml of a 10 21 dilution of the E. muris stock. On the indicated days of infection, mice were sacrificed, and immune responses were assessed. Selected organs were harvested for histology, and bacterial loads were determined by qPCR. Samples of spleen were fixed with 10% formalin and stained by hematoxylin and eosin (H&E) for histopathological examination.

CD4 T cells
CD4 T cells were isolated from spleen using a L3T4 CD4 microbeads (Mitenyi Biotec.) and stimulated with anti-CD3 and/ or anti-CD28 (eBioscience) antibody for 24 hours. Culture supernatant was harvested for the measurement of IFN-c, IL-4, and IL-2.

Determination of Ehrlichia bacterial load in tissues and cells
Since E. muris is an obligatory intracellular bacterium that does not grow outside of the cells, quantification method for the Ehrlichia load was established previously using qPCR [29,39]. Briefly, the ehrlichial load was determined by qPCR (with SYBR Green) of the Ehrlichia dsb gene (GenBank accession number:AY236484), which encodes a thiodisulfide oxidoreductase or disulphide bond formation protein of E. muris. Primer sequences are as follow: E. muris forward, TTCTTTGACTATTCCTGTGGCTAC; E. muris reverse, AAT-GACGCCTCACCGAGTAT; mouse GAPDH forward, CAAC-TACATGGTCTACATGTTC; and GAPDH reverse, CTCGC-TCCTGGAAGATG. The substrate for amplification was DNA purified from samples using the DNeasy Tissue kit (Qiagen, Valencia, CA). qPCR was performed using the 7300 real time PCR system from Applied Biosystems. Results were normalized to GAPDH levels in the same sample and expressed as relative Ehrlichia dsb amounts per GAPDH amount.

Flow cytometric analysis
All antibodies used in this study were: NK1.1, IFN-c, IL-4, CD4, CD8, CD80, CD86, CD11b, CD11c, CD69, CD3, B220, F4/80, Gr-1. Intracellular staining was performed after staining for surface markers. Cells were then fixed and permeabilized using permeabilization/fixation buffer (eBioscience) for 30 min at 4uC, and then stained with IFN-c or IL-4 antibodies diluted in permeabilization buffer (eBioscience). For analysis of cell surface molecule expression, stimulated cells were washed with ice cold PBS, fixed, and stained with FITC-, PE-and APC-labeled antibodies in the presence of anti-CD16 antibody for 30 min on ice. Stained cells were washed, resuspended in PBS/1% FBS/ 0.05% NaN 3 , and analyzed by FACSCalibur followed by analysis using CELLQuest software (Becton Dickinson). All antibodies were obtained from eBioscience.

Statistical analysis
All experiments were performed at least three times. All data are expressed as Mean 6 SD. Data were subjected to Student's t test for analysis of statistical significance, and a P value of ,0.05 was considered to be significant.  Figure S4 Rip2-independent E. muris recognition in DCs. Bone marrow-derived DCs from wild-type and Rip2-deficient mice were infected with E. muris for the indicated time period. (A) Cytokine production from infected cells was assessed by ELISA. LPS (10 ng/ml) stimulation was used as a control. (B) Infected cells (5 days post infection) were stained by Diff-Quick staining. Scale bar, 10 mm. (C) Bacterial loads in infected cells were quantified by qPCR using primers specific for the E. muris dsb gene. Data were normalized by qPCR data for the GAPDH gene in host genomic DNA. Error bars represent SD of triplicate samples. Data are representative of three independent experiments in A-C. Found at: doi:10.1371/journal.pone.0011758.s004 (8.29 MB TIF) Figure S5 IL-1b production by Asc-, Nalp3-, or Ipaf-deficient macrophages infected with E. muris. Bone marrow-derived macrophages from both wild-type and Asc-, Nalp3-or Ipafdeficient mice were treated with LPS (10 ng/ml) for 16 hours and then infected with E. muris in vitro for the indicated time period or stimulated with 2.5 mM ATP for 3 hours. IL-1b secretion in the culture supernatants was assessed by ELISA. Error bars represent SD of triplicate samples. Data are representative of three independent experiments. Found at: doi:10.1371/journal.pone.0011758.s005 (0.18 MB TIF) Figure S6 MyD88-dependent E. muris recognition in macrophages. Bone marrow-derived macrophages from wild-type and MyD88-deficient mice were infected with E. muris in vitro for the indicated time period. (A) Cytokine production from infected cells was assessed by ELISA. LPS (10 ng/ml) stimulation was used as a control. (B) Infected cells (5 days postinfection) were stained by Diff-Quick staining. Scale bar, 10 mm. (C) Bacterial loads in infected cells were quantified by qPCR using primers specific for the E. muris dsb gene. Data were normalized by qPCR data for the GAPDH gene in host genomic DNA. Error bars represent SD of triplicate samples. The p-values were determined by Student's ttest (MyD88+/+ vs. MyD88-/-). *P,0.05. **P,0.01. Data are representative of at least three independent experiments in A-C. N.D., not detected. Found at: doi:10.1371/journal.pone.0011758.s006 (9.43 MB TIF) Figure S7 TLR2/3/4/5/7/9/11 do not play major roles in E. muris recognition. Bone marrow-derived DCs from wild-type, TLR2/4/5/11-deficient (A), or TLR3/7/9-deficient (B) mice were infected with E. muris for the indicated time period, and IL-12 p40 production from infected cells was assessed by ELISA. LPS (10 ng/ml), CpG (1 mM), R-848 (1 mM) and Pam3CSK4 (1 mg/ ml) stimulations were used as controls. Error bars represent SD of triplicate samples. Data are representative of three independent experiments in A and B. N.D., not detected. Found at: doi:10.1371/journal.pone.0011758.s007 (1.00 MB TIF)