This study was conducted according to the principles of the Declaration of Helsinki and under ethical guidelines from the University of Pittsburgh and the University of Illinois at Chicago, Institutional Review Boards. All animal studies and procedures were conducted in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health’s Office of Laboratory Animal Welfare (Assurance Number A3187-01). The Divisions of Laboratory Animal Resources at the University of Pittsburgh and the University of Illinois at Chicago are accredited by the American Association for the Assessment and Accreditation of Lab Animal Care (AAALAC). The Institutional Animal Care and Use Committees at the University of Pittsburgh and the University of Illinois at Chicago approved the protocol.

Female 7- to 8-week-old C57BL/6 mice deficient in MyD88 (MyD88^-/-) and wild type (WT) mice were obtained from Jackson Laboratory (Bar Harbor, ME). All mice were maintained in a pathogen-free environment and were observed daily for signs of illness and survival. The highly virulent monocytic E. japonica strains or mildly virulent Ehrlichia muris (EM) were used in this study and were provided by Dr. Yasuko Rikihisa (Ohio State University, Columbus, OH).

Purified, cell-free E. japonica or EM organisms were prepared from spleens and livers of E. japonica- or E. muris-infected mice harvested on day 7 post-infection (p.i.). following a series of sonication and centrifugation steps. Purified bacteria were suspended in sucrose phosphate buffer as described previously (22, 23).

HCs were isolated from naive or infected WT and MyD88^-/- mice by perfusion technique as described (9). The purity of the isolated primary HCs was evaluated as described previously (24). Isolated primary HCs (4×10⁵ cells/ml) were then plated on 0.2% gelatin-coated 6-well culture plates. In certain experiments, we used the murine HC cell line AML2 (CRL-5564, ATCC).

E. japonica was added to cultured primary murine HCs or AML2 HCs at a multiplicity of infection (MOI) of 5 before infection, cells were pretreated for 1 hr. with the MyD88 inhibitor ST 2825 (APExBIO Technology LLC, Houston, TX) at a 15 μM to block MyD88 signaling. ST 2825 is a specific inhibitor of MyD88 dimerization that inhibits the homodimerization of the MyD88 Toll-interleukin receptor (TIR) domains without affecting the homodimerization of the death domains. TIR domain inhibition interferes with the recruitment of interleukin receptor-associated kinase 1 (IRAK1) and IRAK4 by MyD88. In our initial standardization experiments, we used several doses of MyD88 inhibitor, as suggested by other studies (25, 26) and examined the efficacy of inhibition on MyD88 expression as well as viability of the cells. In all subsequent experiments, we treated cells with 15 μM of the inhibitor, as this dose was able to inhibit MyD88 signal without affecting the viability of the cells ( Supplementary Figure 1 ). The effect of MyD88 inhibitor is specific as treatment of cells with other inhibitors such as CASP1 inhibitor did not result in similar effect seen by the MyD88 inhibitor. To block CASP1 or CASP11 activation, infected cells were pretreated with 30 μg/mL CASP1 (Ac-YVAD-cmk, invivoGene) or 40 μM CASP11 inhibitors (Wedelolactone; Santa Cruz). To block mTORC1 activation, infected cells were pretreated with 10 µM mTORC1 inhibitor rapamycin (InvivoGen, San Diego, CA) for 30 min followed by infection or fresh media control. Uninfected control cells were cultured with mock antigen (antigens prepared from lysed uninfected splenocytes or liver cells). All cells were collected 24 hr p.i. for further analysis by immunoblot, and supernatants were collected and stored at minus 80°C for cytokine analysis.

The MTT assay (ab211091, Abcam, Cambridge, UK) was employed to assess cell viability following treatment with the MyD88 inhibitor ST 2825 (15 μM) at various time points. HCs were pre-treated with the MyD88 inhibitor ST 2825 for different durations (2, 4, 12, and 24 hours) and subsequently plated into 96-well plates. After treatment, each well received 50 μL of serum-free media and 50 μL of MTT reagent, and the plate was incubated at 37°C for 3 hours. Following incubation, 150 μL of MTT solvent was added, and the plate was incubated for 15 minutes. The absorbance was measured at 590 nm, and the percentage of cytotoxicity in treated cells was determined in accordance with the manufacturer’s instructions.

The number of E. japonica in infected hepatocytes was determined at 24 hr p.i. by quantitative reverse transcription PCR (RT-qPCR) of the Ehrlichia dsb gene as described previously (11). Primers and probes were used ( Supplementary Table 1 ). The absolute copy number of Ehrlichia was determined based on serial dilutions of plasmids containing the Ehrlichia dsb gene. Results were expressed as Ehrlichia copies per ng of DNA. The number of live replicating bacteria was determined by RT-qPCR analysis of the 16s rRNA as described previously (11).

HCs were lysed in RIPA buffer (Thermo Fisher Scientific, Waltham, MA) supplemented with protease inhibitors and 1 mM phenylmethyl sulphonyl fluoride (PMSF). The total protein content of the lysates was measured using a Bicinchoninic Acid Assay Kit (Pierce Biotechnology, Waltham, MA). According to standard protocols, Membranes were processed and probed with the following primary antibodies: anti– CASP1 (1:100; EMD Millipore, Billerica, MA), anti- CASP11 (1:100, Cell Signaling), anti- CASP3 (1:100, Cell Signaling), anti-LC3-II (1:1000; Sigma-Aldrich, St. Louis, MO) and anti-p62 (SQSTMI) (1:1000; Cell Signaling). Anti-phospho S6 was from Cell Signaling Technology, Inc. (Danvers, MA) and used at a 1:1000 dilution. Blots were washed in TBST and probed with peroxidase-conjugated bovine anti-rabbit secondary antibodies (1:10000; Santa Cruz Biotechnology, Santa Cruz, CA). Specific signals were developed using the ECL-Plus system (GE Healthcare, Chicago, IL). Blots were stripped with Restore Western Blot Stripping Buffer (Pierce) and re-probed with anti-GAPDH (1:1000, Cell Signaling Technology, Inc.) or anti-β-actin (1:1000, Cell Signaling) as loading controls. The density of bands in the blots was determined using ImageJ software version 1.51W (NIH, Bethesda, MD). LC3II was determined by normalization to GAPDH.

Total RNA from HCs was isolated using Trizol^™ Reagent (Invitrogen Life Technologies, Carlsbad, CA). For cDNA synthesis, 2 μg of total RNA was reverse transcribed in a final volume of 20 μl using the iScript™ reverse transcription supermix kit (Bio-Rad, Hercules, CA), as specified by the manufacturer. In a Bio-Rad T100 PCR/Thermal Cycler, cDNA was combined with specific primer sets ( Table S1 ) using SsoAdvanced™ Universal SYBR^® Green Supermix (Bio-Rad).

The mouse cytokine array C1000 (RayBiotech, Peachtree Corners, GA) was used to detect multiple cytokines in the HCs culture supernatant. HCs (5x10⁵ cells) were cultured in 6-well plates at 37°C for 48 hr. and 5% CO_2. According to the manufacturer’s recommendations, culture supernatants were collected from naive and infected cells at 24 hr p.i. and processed for use in the array. Briefly, membranes were incubated in the blocking buffer at room temperature for 30 min, followed by washing, adding culture supernatant, and incubating at 4°C overnight. The membranes were then washed three times using a wash buffer and then incubated with biotin-conjugated antibodies at room temperature for 2 hr. Finally, membranes were washed and incubated with horseradish peroxidase-conjugated streptavidin at room temperature for 2 hr. Secreted proteins were detected using a luminescence detector (Bio-Rad), and the spots were quantified by Image J. Background staining was subtracted, and the signal was normalized to positive controls on each membrane to obtain relative protein concentrations, represented as fold change compared to naïve (uninfected) and untreated HCs. The concentration of HMGB1 released into the culture supernatant was determined using the mouse HMG1/HMGB1 ELISA Kit (lot # 116578, LifeSpan Biosciences).

Staining HMGB1 aggregates, and quantification by confocal microscopy were performed as previously described (7). Briefly, HCs were cultured on coverslips infected with E. japonica at MOI of 5, washed three times with PBS, fixed with 2% paraformaldehyde for 20 min, and permeabilized with 0.1% Triton X-100 in PBS for 30 min. After blocking with 5% BSA (A2153, Sigma-Aldrich) for 60 min, the following primary antibodies were added for 2 hr. at room temperature: rabbit anti-HMGB1 (1:100, Cell Signaling Technology, Inc.) and anti-p62 (Cell Signaling 1:100). Next, cells were washed and then incubated with fluorescently labeled anti-rabbit secondary antibody DyLight (1:500, VectaFluor, Vector Labs, Burlingame, CA) for 1 hr. Finally, nuclei were stained with DAPI, and cells were analyzed by confocal microscopy (Olympus Fluoview 1000).

The two-tailed t-test was used for comparison of mean values for two experimental groups, and a two-way ANOVA (analysis of variance) followed by post hoc testing using Tukey’s Honestly Significant Difference (HSD) method for comparisons of multiple experimental groups. Data were expressed as mean ± SD. P ≤ 0.05, P ≤ 0.01 and P ≤ 0.001.

Innate immune responses to intracellular pathogens are triggered by TLR signaling via the adaptor molecules MyD88 or TRIF (27, 28). To examine how innate immune response in HCs senses Ehrlichia, we analyzed the differential expression of mRNA of several TLRs, MyD88, and TRIF in an HC cell line infected with either mildly virulent EM or highly virulent E. japonica. Our data show that E. japonica infection of HCs induced significant upregulation of MyD88, TLR9 and TRIF (*P<0.05, **P<0.01, ***P<0.001, respectively) when compared to EM infection of HCs ( Supplementary Figures 2A–C ). Although there was upregulation of TLR2 and TLR7 in both E. muris and E. japonica-infected primary HCs compared to naïve cells, we did not detect a significant difference in mRNA expression of TLR2 and TLR7 (Supplementary Figures 2D, E ). These data are consistent with our previous data showing that the TLR9-MyD88 axis is likely the major pathway that regulates the innate immune response to Ehrlichia infection in macrophages, the primary target cells for Ehrlichia (5).

To investigate the role of MyD88 in the pathogenesis of Ehrlichia-induced liver injury, we infected WT and MyD88^-/- primary HCs with E. japonica at an MOI of 5 and measured the number of intracellular bacteria by PCR. We measured the absolute copy number of E. japonica organisms in primary HCs at 12 hr. and 24 hr p.i. via amplification of dsb gene. We found that the number of E. japonica in both WT and MyD88^-/- HCs was significantly higher 24 hr p.i. (*P<0.05) compared to 12 hr. p.i. However, the numbers of Ehrlichia were dramatically higher in MyD88^-/- HCs at 24 hr p.i. than WT-HCs at the same time point ( Figure 1A ). To determine whether the intracellular bacteria measured by dsb-based PCR are measuring live, but not dead bacteria, we also analyzed the number of intracellular bacteria by 16S rRNA as described under M&M. As shown in previous studies by us and other investigators (5, 29), there is a strong correlation between the bacterial 16S rRNA and viability of Ehrlichia. To determine if the difference in the bacterial number was due to a difference in survival and/or replication of live Ehrlichia, we measured the expression level of Ehrlichia 16S rRNA by RT-qPCR at two time points 12 hr and 24 hr p.i. To this end, cells were infected with E. japonica. At 4hr post infection, the extracellular bacteria were washed and infected cells were further incubated up to 24 hr. At 12 hr p.i., there were significantly higher numbers of intracellular Ehrlichia within WT- and MyD88^-/- HCs compared to uninfected cells; however, there were no significant differences between the two infected groups. On the other hand, although bacterial number was higher in both WT- and MyD88^-/- HCs at 24 hr compared to 12 hr, we detected a significant (*P<0.05) increase in the number of intracellular Ehrlichia in MyD88^-/- HCs compared to infected WT-HC, suggesting a higher replication of live bacteria in HCs lacking MyD88 signals ( Figure 1B ). To confirm the data in primary HC from KO mice, which may have compensatory mechanisms, we also measured the number of intracellular bacteria in HC cell line treated with MyD88 inhibitor. The findings in primary HCs were consistent with data from E. japonica-infected HC cell line (AML2 HCs) treated with MyD88 inhibitor. Lack of MyD88 signaling in infected HCs treated with MyD88 inhibitor, as described under M&M, resulted in a higher number of intracellular live Ehrlichia compared to similarly infected but untreated AML2 cells at 24 hr p.i. ( Figure 1C ). The consistency between data obtained from MyD88^-/- HCs as well as HC cell line treated with MyD88 inhibitor suggest specificity of the MyD88 inhibition. Additionally, our standardization experiments using other inhibitors such as CASP1 inhibitor did not recapitulate the effect of MyD88^-/- HCs as shown below. Together, these data suggest that MyD88 signaling plays a host-protective mechanism by controlling intracellular bacterial survival and/or replication in infected HCs.

Our previous studies indicated that MyD88 signaling triggers activation of canonical (CASP1-mediated) and non-canonical (CASP11-mediated) inflammasomes in macrophages in a host-pathogenic mechanism that mediates liver injury (5). We examined whether CASP1 is activated in HCs following E. japonica infection, and the role of MyD88 in regulation of HC-specific Casp1. E. japonica infection of primary WT HCs did not trigger significant upregulation of Casp1 or IL-1β mRNA compared to uninfected WT HCs. However, MyD88 deficiency in primary E. japonica-infected MyD88^-/- HCs resulted in upregulation of both Casp1 and IL-1β mRNA compared to uninfected primary MyD88^-/-HCs and infected WT HCs ( Figures 2A–D ). Consistent with mRNA data, immunoblot analysis showed an elevated quantity of cleaved CASPl in E. japonica-infected primary MyD88^-/- HCs compared to E. japonica-infected primary WT HCs and uninfected cells ( Figures 2A, B ). These data suggest that MyD88 inhibits CASPl activation in HCs following E. japonica infection. Notably, activation of CASPl was inversely correlated with activation of CASP3 (apoptotic marker), such that E. japonica-infected primary MyD88^-/- HCs showed lower levels of active CASP3 compared to E. japonica-infected primary WT HCs ( Figures 3A, B ). These data suggest that MyD88 signaling inhibits CASPl activation, while promoting CASP3 activation in HCs following E. japonica infection.

Given our observation of an inverse relationship between CASP1 and CASP3 activation, we hypothesized that active CASP1 in HCs may play a hepatoprotective role via negative regulation of CASP3 activation. To address this possibility, we examined CASP3 activation in E. japonica-infected MyD88^-/- HCs and control-infected WT-HCs (showing a reduced amount of active CASP1) in the presence of a CASP1 inhibitor. Our findings revealed that the CASP1 inhibitor had no discernible impact on the extent of CASP3 activation in infected WT HCs. However, notably, inhibition of CASP1 resulted in a significant increase in the level of active CASP3 in MyD88^-/- HCs ( Figures 3A, B ). This suggests that CASP3 activation in infected HCs is negatively regulated by active CASP1 in the absence of MyD88. In other words, MyD88 promotes CASP3 activation via inhibition of CASP1.

Our prior studies have shown that E. japonica infection triggers CASP11 activation in macrophages, which is associated with fatal E. japonica infection in mice (9). We, therefore, examined the contribution of MyD88 to CASP11 activation following infection of HCs with E. japonica. Infection of WT primary HCs with E. japonica increased the mRNA levels of Casp11 ( Figure 3C ). At the protein level, our investigation revealed that E. japonica-infected primary MyD88^-/- hepatocytes (HCs) exhibited a decreased quantity of active CASP11 compared to E. japonica-infected primary WT HCs ( Figures 3D, E ). This data suggests that MyD88 promotes activation of CASP11. Notably, activation of CASP11 in infected HC cell line correlated with host cell pyroptosis and inflammatory host cell death measured by an increased level of LDH release compared to uninfected cells. On the other hand, treatment of E. japonica-infected HC cell line with MyD88 inhibitor attenuated LDH release compared to E. japonica infected but untreated HCs ( Figure 3H ). HC cell line cultured with or without MyD88 inhibitor has the same low level of LDH, suggesting that MyD88 inhibitor alone does not have cytotoxic effect ( Figure 3H ).

Both active CASP1 and CASP11 play a role in inducing cell death through Gasdermin D cleavage in response to inflammasome activation (30). Given that active CASP1 is higher, and active CASP11 is lower in MyD88^-/-, we examined Gasdermin D cleavage as a potential mechanism that induces pyroptotic cell death and LDH release. To this end, we infected WT and MyD88^-/- HCs with E. japonica and examined Gasdermin D cleavage. Our data demonstrate that infection of WT-HCs leads to Gasdermin D cleavage compared to uninfected cells at 24 hr p.i. In contrast, consistent with CASP11 activation, E. japonica infection of MyD88^-/- HCs did not result in Gasdermin D cleavage at same time point when compared to uninfected MyD88^-/- HCs and infected WT HCs. These findings suggest that E. japonica induced, MyD88-mediated pyroptosis occurs via Gasdermin D cleavage. To determine whether Gasdermin D cleavage is regulated by CASP1 and/or CASP11, we infected WT- and MyD88^-/- HCs in the presence or absence of CASP1 and CASP11 inhibitors. Our data showed that Ehrlichia-mediated cleavage of Gasdermin D in WT-HCs is abrogated upon treatment of cells with CASP11, but not CASP1 inhibitor, suggesting that Ehrlichia-induced, MyD88-mediated cleavage of Gasdermin D in HCs is CASP11 dependent ( Figures 3F, G ). Taken together, these data further suggest that MyD88 triggers death of HCs via activation of apoptotic CASP3 pathway and non-canonical CASP11-mediated inflammasome pathway, respectively, while inhibiting CASP1 activation.

HMGB1 is a nuclear protein that acts as a DAMP when translocated to the cytosol. Our recently published study revealed that HMGB1 is secreted by HCs following E. japonica infection and that secretion is CASP11-dependent (7). Extracellular HMGB1 is secreted actively or passively following pyroptotic inflammatory cell death. To determine the contribution of MyD88 to HMGB1 secretion, we first measured HMGB1 cytosolic translocation in primary WT and MyD88^-/- HCs at 24 hr after E. japonica infection using immunofluorescence staining. Our data showed that HMGB1 was detected in the nucleus and cytosol of E. japonica-infected WT-HCs with a higher nuclear than cytoplasmic expression or localization compared to uninfected WT HCs. In contrast, MyD88^-/- HCs exhibited significant cytosolic translocation of HMGB1 ( Figures 4A, B ) with minimal nuclear localization compared to uninfected MyD88^-/- HCs as well as infected WT-HCs. The difference in HMGB1 localization between WT and MyD88^-/- HCs was associated with a significantly higher protein level of HMGB1 in primary WT-HCs compared to similarly infected MyD88^-/- HCs at 24 hr p.i. ( Figures 4C, D ).

Since the total HMGB1 was increased in infected primary WT-HCs, we hypothesized that the small number of cytosolic HMGB1 in primary WT compared to MyD88^-/- HCs is due to the secretion of HMGB1 secondary to the activation of CASP11 and pyroptosis. To this end, we measured HMGB1 in cells or the culture supernatant from WT HCs and MyD88^-/- HCs 24 hr after E. japonica infection. Compared to uninfected WT HCs, E. japonica-infected HCs secreted a significantly higher amount of HMGB1 in culture supernatant. Notably, the lack of MyD88 signaling in primary E. japonica-infected HCs decreased HMGB1 secretion ( Figure 4E ), despite enhanced cytosolic translocation as shown before. These results could be attributed to decreased CASP11-mediated Gasdermin D cleavage and pyroptotic cell death ( Figures 3D, F, H ) in MyD88^-/-HCs compared to WT-HCs, which limit or decrease secretion of cytosolic HMGB1.

Studies have shown that extracellular HMGB1 produced by immune or parenchymal cells binds in a paracrine manner to RAGE (the receptor for advanced glycation end products) expressed on adjacent cells (31). Such binding results in arrays of signaling pathways including activation of NF-κb, JAK/STAT and PI3K/AKT among others, which promote inflammatory responses (32) We thus examined whether the excessive dysregulated inflammation seen in the murine model of fatal E. japonica-induced liver injury and sepsis is linked to MyD88-mediated expression of RAGE receptor and downstream signals in HCs following binding to extracellular HMGB1. Indeed, E. japonica infection of WT-HCs triggered significant upregulation of RAGE mRNA compared to uninfected cells. In contrast, blocking MyD88 signaling in HC cell line significantly abrogated RAGE mRNA expression when compared to infected/untreated HCs ( Figure 4F ). Together, these data suggest that RAGE expression on HCs following Ehrlichia infection is MyD88 dependent. Additionally, since MyD88 deficient cells fail to adequately secrete HMGB1, these results indicate a possible positive feedback loop between HMGB1 secretion and upregulation of RAGE.

Prior studies by our lab and other investigators have shown that inflammasome activation in macrophages is negatively regulated by autophagy (5, 33). Therefore, to examine whether MyD88 mediates activation of the non-canonical inflammasome pathway in HCs via regulation of autophagy, we analyzed autophagosome formation. The first step in autophagy involves recruitment of cytosolic LC3 to the phagophore (34). Recruitment of non-lipidated LC3 (LC3I) to autophagosomes involves proteolytic cleavage and lipidation, resulting in LC3II. To investigate this, we infected primary WT HCs and MyD88^-/- HCs with E. japonica at MOI of 5 and measured autophagy induction by immunoblotting at 24 hr p.i. Immunoblotting analysis demonstrated a significant increase in LC3II quantity, normalized to GAPDH, in E. japonica-infected WT HCs compared to uninfected WT HCs, with a 6-fold higher amount of LC3II in WT HCs infected with E. japonica. This indicated a higher LC3II quantity and autophagosome formation after E. japonica infection in MyD88 sufficient primary HCs. However, the amount of LC3II in E. japonica-infected MyD88^-/- was lowered compared to infected MyD88 sufficient primary HCs cells. ( Figure 5A ). To validate above results, we analyzed LC3 II/LC3 I ratio and our results demonstrated a significant increase in LC3 II/LC3 I ratio, in E. japonica-infected WT HCs compared to uninfected WT HCs, with a 3.2-fold higher ratio in WT HCs infected with E. japonica. This indicated a higher LC3 I conversion and autophagosome formation after E. japonica infection in MyD88 sufficient primary HCs. Our data also showed that this conversion in E. japonica-infected MyD88^-/- was lowered compared to infected MyD88 sufficient primary HCs cells. ( Figure 5B ). The elevated level of LC3II in E. japonica-infected HCs could be due to blocking autophagy flux. To evaluate autophagy flux in primary HCs, we analyzed the level of p62, a selective autophagy adaptor that binds to ubiquitinated proteins and damaged organelles to traffic them to autophagosome-lysosomal compartments for degradation. The level of p62 is inversely correlated with the autophagic flux and activity. Our results showed a significant accumulation of P62 in E. japonica-infected primary WT HCs, suggesting a block of autophagy flux. Notably, lack of MYD88 resulted in decreased accumulation of P62 as evidenced by decreased level of P62 in E. japonica-infected MyD88-/- primary HCs compared to E. japonica-infected WT HCs ( Figure 5C ). Together, these data suggest that MyD88 signaling in HCs promotes autophagy induction, while blocking autophagy flux during E. japonica infection.

mTORC1, a mammalian target of rapamycin complex 1, is a protein complex that functions as a nutrient/energy/redox sensor and controls protein synthesis. We previously reported that MyD88 mTORC1 activation negatively regulates autophagy in macrophages during E. japonica infection (5). Therefore, we examined the hypothesis that MyD88 promotes autophagy via inhibition of mTORC1 activation in E. japonica-infected HCs. We examined mTORC1 activation by measuring phosphorylation of its downstream target ribosomal protein S6 (pS6). Surprisingly, E. japonica infection of untreated HC cell line induced activation of mTORC1 as evidenced by a higher level of quantity of phosphorylated S6 (PS6) downstream target of mTORC1 in these cells compared to uninfected cells. E. japonica-infected HC cell line treated with MyD88 inhibitor had a lower level of pS6 at 24 hr p.i. compared to untreated but infected cells or uninfected HCs ( Figure 6A ), suggesting that MyD88 signaling induces mTORC1 activation in HCs following E. japonica infection. This data is surprising based on the known function of mTORC1 as negative regulator of autophagy. To directly examine the link between MyD88-mTORC1-autophagy axis in E. japonica-infected HCs, we measured LC3II protein in E. japonica-infected WT and MyD88 inhibitor treated HC cell line upon further addition of rapamycin (mTORC1 inhibitor) at 24 hr p.i. Rapamycin pre-treatment of uninfected or E. japonica infected HC cell line that are either treated with MyD88 inhibitor (i.e. MyD88 deficient cells) or left untreated resulted in blocking mTORC1 activation ( Figures 6B, C ). Consistent with findings in Figure 5A , lack of MyD88 signals in untreated but E. japonica-infected cells attenuate LC3II level. Paradoxically, inhibition of mTORC1 activation in untreated E. japonica-infected HCs, with intact MyD88 signaling, decreased autophagy induction as measured by decreased LC3II level ( Figure 6C ). These findings were not due to cell death since inhibition of mTORC1 activation, in presence of MyD88 inhibitor that block MyD88 signaling in E. japonica-infected HCs enhanced induction of autophagy ( Figure 6C ). Thus, enhanced autophagy upon blocking mTORC1 in infected HCs is opposed by MyD88 signaling. Together, these data suggest that MyD88, not only promotes autophagy induction in HCs following E. japonica infection, but also reverse or counter regulate the inhibition of autophagy induction by mTORC1 in HCs.

Our murine model of fatal ehrlichiosis demonstrated a strong link between MyD88 signaling and cytokine and chemokine storm in a murine model of fatal ehrlichiosis caused by systemic E. japonica infection. To determine the contribution of MyD88 signaling to E. japonica-induced inflammation in hepatocytes, we examined the secretome of E. japonica-infected HC cell line cultured with or without MyD88 inhibitor. E. japonica infection of untreated HCs significantly increased the levels of IL-6, which was significantly attenuated in infected cells treated with an inhibitor, suggesting that IL-6 secretion is MyD88-dependent ( Supplementary Figure 3A ). Similarly, E. japonica infection led to a higher secretion of GM-CSF ( Supplementary Figure 3C ) by untreated HCs compared to uninfected controls. However, this effect was MyD88-independent. GM-CSF recruit neutrophils to the liver, where they mediate tissue injury during fatal E. japonica infection in mice (7, 35). Notably, E. japonica infection inhibited secretion of several pro-inflammatory cytokines and chemokines such as IL-12p70, IFN gamma, IL-12p40, TNF receptor II (TNFRII), IL-17A, and leptin ( Supplementary Figures 3D–I ). However, there is no change in M-CSF ( Supplementary Figures 3A, B ). Also, molecules such as TIMP-1, KC (CXCL1), and MIP-1 alpha (CCL3) ( Supplementary Figure 3M ) showed inhibition upon E. japonica- infection in MyD88-independent manner. The levels of MIP-1 gamma showed an increase upon E. japonica-infection compared to uninfected controls. Although, this effect was MyD88 independent ( Supplementary Figure 3M ). Levels of other molecules such as MCP-1 (CCL2), eotaxin-2 (CCL24), TECK (CCL25), BLC (CXCL13), CXCL1, and Fractalkine (TNFSF8) ( Supplementary Figures 3L–P ) showed minimal change or no change. Other inflammatory chemokines and cytokines did not exhibit any statistically significant differences between untreated E. japonica-infected HCs or pretreated with MyD88 inhibitor ( Supplementary Figures 3A–T ).