Interferon Type I Regulates Inflammasome Activation and High Mobility Group Box 1 Translocation in Hepatocytes During Ehrlichia-Induced Acute Liver Injury

Mice and Ehrlichia Species

Female, 6-8-week-old, C57BL/6 wild-type (WT) mice were obtained from Jackson Laboratory (Bar Harbor, ME). These mice were used to isolate primary HCs. All mice were maintained in a specific pathogen-free environment at the University of Pittsburgh Animal Care Facility. Two Ehrlichia species were used in this study: a highly virulent IOE and mildly virulent Ehrlichia muris (EM). These strains were provided by Dr. Yasuko Rikihisa (Ohio State University, Columbus, OH). IOE and EM cause fatal and mild ehrlichiosis in C57BL/6 mice, respectively.⁽¹⁸⁾

Isolation of Primary Murine HCs

Primary HCs were isolated using a two-step perfusion method as described.⁽²²⁾ Isolated HCs were seeded on collagen-coated culture dishes and used for experiments on the following day.

In Vitro Ehrlichia Infection

IOE and EM organisms were added to the HC cultures at a multiplicity of infection (MOI) of 5. Uninfected or infected cells were cultured in the absence or presence of recombinant IFNβ (rIFNβ; 500 IU/mL) and a caspase-1 inhibitor (10 µm/mL) (Enzo Life Sciences, Ann Arbor, MI). All cells were collected at 24 hours postinfection for further analysis by western blot. The cell-culture media was collected and stored at −80°C for cytokine analysis.

Bacterial Burden Measurement Using Quantitative Real-Time PCR

The number of intracellular Ehrlichia (IOE or EM) in infected HCs was determined at 24 hours and 48 hours postinfection by quantitative real-time polymerase chain reaction (qPCR) amplifying the conserved Ehrlichia disulfide bond formation protein (dsb) gene as described.⁽²³⁾ The primers used for Ehrlichia dsb are listed in Table 1. The relative copy number of intracellular Ehrlichia harvested from in vitro-infected cells was determined by normalization of the Ehrlichia copy number to glyceraldehyde 3-phosphate dehydrogenase (GAPDH). In some experiments, the number of live replicating bacteria was determined by quantitative reverse-transcription PCR (qRT-PCR) analysis of the 16S ribosomal DNA (rDNA) as described.⁽²³⁾ PCR primer sets are listed in Table 1. The GAPDH gene was amplified for normalization of the complementary DNA (cDNA) amount used in qRT-PCR. Reactions were performed in triplicate, and the data were analyzed using the 2^−ΔΔCt method as described.⁽²⁴⁾

Measurement of Cytokines and Chemokines by Enzyme-Linked Immunosorbent Assay

Levels of proinflammatory cytokines and chemokines were measured by multiplex enzyme-linked immunosorbent assay (ELISA) (Bio-Plex Pro TM Mouse Cytokine 23 plex #MD60009RDPD; Biorad, Hercules, CA) following the manufacturer’s recommendation. The concentration of HMGB1 was determined using the mouse HMGB1 ELISA Kit (LifeSpan Bioscience, Seattle, WI). The concentration of IL-1α, IL-1β, and IFNβ was measured with a commercial ELISA kit (R&D Systems, Minneapolis, MN).

Western Blot

HCs were lysed in radio immunoprecipitation assay buffer (Thermo Fisher Scientific, Waltham, MA) supplemented with protease inhibitors and 1 mM phenyl methyl sulphonyl fluoride. Membranes were processed and probed with the following antibodies according to standard protocols: anti-caspase-1 (1:100) (EMD Millipore, Billerica, MA), anti-caspase-11 (1:100), and anti-LC3AB (Sigma-Aldrich, St. Louis, MO). Antibodies against sequestosome 1 (p62/SQSTM1), protein S6 (S6), and phosphorylated S6 (pS6) were from Cell Signaling Inc. Blots were stripped and reprobed with anti-GAPDH (1:5,000) (Sigma-Aldrich) or anti-beta actin (1:2,500) (Abcam, Cambridge, MA) as loading controls. The density of bands in western blots was determined using ImageJ software version 1.48 (National Institutes of Health [NIH], Bethesda, MD). The ratio of LC3II:LC3I was determined by normalization of LC3II and LC3I to GAPDH.

Immunofluorescence Staining and Confocal Microscopy

Staining of LC3 puncta, LysoTracker staining of acidified lysosomes, Ehrlichia, and HMGB1 aggregates as well as quantification by confocal microscopy were performed as described.^{(7, 8)}

Additional information on the Confocal Staining protocols are being provided in the Supplementary Methods Section.

Image Analysis

Analysis of HMGB1 and LC3B puncta as well as colocalization of LC3B with LysoTracker were performed using the NIH ImageJ software package. We analyzed 40-50 cells per group from three independent experiments. The colocalization images were analyzed using PCI software by counting the number of yellow aggregates per cell. Results were presented as average HMGB1 or LC3 puncta per cell or the number of colocalized LC3 puncta with lysosome per cell. At least 10 fields of view for each sample were quantified for each experimental group across at least three independent experiments. Data were analyzed as average ± SD.

Statistical Analysis

All data presented are representative of at least three independent experiments that yielded similar results. The two-tailed t test was used for comparison of mean values for two experimental groups, and a one-way analysis of variance was used for comparisons of multiple experimental groups. Data are expressed as means ± SD.

Ehrlichia Invade and Replicate Intracellularly in HCs

Our previous studies have shown that intraperitoneal infection with IOE causes acute fatal toxic shock-like syndrome, characterized by extensive foci of hepatic apoptosis and necrosis followed by multiorgan failure.⁽⁹⁾ In contrast, infection with mildly virulent EM causes mild and self-limited disease with minimal hepatic pathology.⁽²⁴⁾ To determine how HCs contribute to the pathogenesis of Ehrlichia-induced liver damage, we first examined the HC response to IOE infection compared to EM infection. Primary HCs were isolated from naive WT mice and infected in vitro with IOE or EM at an MOI of 5. Analysis of total number of intracellular Ehrlichia by qPCR indicated that both IOE and EM enter and survive in HCs at 12 and 24 hours postinfection, although the number of EM but not IOE organisms significantly declined at 24 hours compared to 12 hours postinfection (Fig. 1A). Consistent with the qPCR of bacterial DNA, analysis of the number of live and replicating bacteria by 16S rDNA demonstrated a significantly higher number of replicating bacteria in IOE-infected HCs at 24 hours when compared to 12 hours (P < 0.01) while HCs were able to control the EM infection at 24 hours (Fig. 1B). The difference between the number of intracellular IOE and EM in HCs at 24 hours is less likely due to prolonged internalization of IOE from the extracellular environment because we did not detect extracellular EM or IOE organisms in the HC culture supernatant at 12 and 24 hours after infection (data not shown). Immunofluorescence staining of cells with polyclonal cross-reactive anti-Ehrilchia antibodies demonstrated internalization of both EM and IOE organisms in HCs at 24 hours postinfection (Fig. 1C). These data suggest that IOE are able to evade host response and replicate within HCs compared to EM.

HCs are a Cellular Source of IFNβ and Chemokines During Fatal IOE Infection

Recent studies by us and other investigators have defined a central role for IFN-I in severe and fatal ehrlichiosis in mice infected with IOE.^{(14, 19, 25)} To determine whether HCs contribute to IFNβ production and dysregulated production of cytokines and chemokines, we compared responses among uninfected, IOE-infected, and EM-infected HCs at 24 hours postinfection. Compared to uninfected and EM-infected HCs, we detected significant secretion of IFNβ by IOE-infected HCs at the protein level as measured by ELISA (P < 0.001)(Fig. 2A) and at the messenger RNA (mRNA) level as measured by qRT-PCR (P < 0.001 and <0.01, respectively) (Fig. 2B). mRNA quantification data indicated that infection of HCs with IOE significantly up-regulated IFNARs in HCs compared to uninfected and EM-infected HCs (P < 0.05) (Fig. 2C).

Based on these data and the findings that IFNAR signaling on nonhematopoietic cells plays a role in bacterial infection and the development of Ehrlichia-induced sepsis,^{(14, 19, 26, 27)} we hypothesized that binding of myeloid cell-derived IFN-I cytokines to IFNARs on HCs could mediate dysregulated inflammation and immunopathology. To test this hypothesis, HCs were infected with virulent IOE in the presence or absence of IFNβ and measured HC responses. IOE infection alone significantly increased expression of IFN regulatory factor 3 (IRF3) (P < 0.05) but not IRF7 (Fig. 2D,E) or IRF9 (data not shown). Exogenous IFNβ stimulation did not increase IRF3 further compared to infection alone. On the other hand, rIFNβ stimulation of uninfected or IOE-infected HCs significantly triggered up-regulation of IRF7 (P < 0.01) compared to unstimulated controls (Fig. 2E), suggesting that IOE infection alone does not induce the expression of IRF7.

We next examined whether HCs contribute to excessive production of proinflammatory cytokines and chemokines, which is a major pathognomonic finding during fatal infection in mice and humans. Our data showed HCs infected with either IOE or EM did not produce significant levels of inflammatory cytokines tumor necrosis factor alpha, IL-12, IL-4, IL-5, IL-6, or IL-10 compared to uninfected HCs (data not shown). On the other hand, IOE-infected HCs produced significantly higher levels of granulocyte-macrophage colony-stimulating factor (GM-CSF; P < 0.001), macrophage inflammatory protein 1 alpha (MIP1α; P < 0.01), IFNγ-induced protein 10/chemokine (C-X-C motif) ligand 10 (CXCL10), monokine induced by gamma (MIG)/CXCL9, keratinocyte-derived chemokine (KC)/CXCL1, regulated upon activation, normal T-cell expressed, and secreted (RANTES)/chemokine (C-C motif) ligand 5 (CCL5) (P < 0.001), vascular endothelial growth factor (P < 0.05), and monocyte chemoattractant protein 1 (MCP1)/CCL2 compared to uninfected and EM-infected HCs (Fig 3A-H). Notably, rIFNβ stimulation of IOE-infected HCs resulted in a significant increase in the production of GM-CSF (neutrophil chemoattractant and growth factor) and MIP1α (macrophage chemoattractant) (P < 0.0001 and P < 0.001, respectively) when compared to unstimulated infected HCs (Fig. 4A,B). We did not observe a significant increase in the secretion of other chemokines by IOE-infected HCs following IFNβ stimulation (Fig. 4C-E). Taken together, these data suggest that HCs are a major source of several chemokines associated with fatal infection.

IFN-I Signaling Induces Activation of Canonical and Noncanonical Inflammasome Pathways in IOE-Infected HCs

We next examined whether IOE infection triggers inflammasome activation in HCs and if this response was regulated by IFNAR signaling and measured markers of canonical and noncanonical inflammasome pathways. IOE infection significantly enhanced mRNA expression of caspase-1, IL-1β, and caspase-11 in HCs compared to uninfected cells (Fig. 5A-C). Additional rIFNβ stimulation of HCs significantly up-regulated mRNA expression of caspase-1 (P < 0.01), IL-1β (P < 0.01), and caspase-11 (P < 0.00.1) in both uninfected and IOE-infected HCs compared to unstimulated control cells with or without infection (Fig. 5A-C). To validate mRNA data at the protein level, we measured expression of active caspase-1 and caspase-11 by immunoblot. Infection of HCs induced activation of active caspase-1 and caspase-11 when compared to uninfected cells. Although stimulation of uninfected HCs with rIFNβ was higher than unstimulated and uninfected controls, IOE infection of IFNβ-stimulated cells further increased expression of active caspase-1 and caspase-11 rIFNβ (Fig. 5D,E). This synergistic effect of infection and paracrine IFNAR signaling also resulted in significant secretion of IL-1β (Fig. 5F) as well as pyroptotic cell death marked by significant increase in lactate dehydrogenase (LDH) release when compared to IOE infection alone, rIFNβ alone, or uninfected/unstimulated cells (Fig. 5G). Together, these data suggest that IOE infection triggers significant canonical and noncanonical inflammasome activation marked by cleavage of caspase-1, caspase-11, IL-1β secretion, and pyroptosis following paracrine IFNAR signaling.

Ehrlichia Infection Induced HMGB1 Release Is Mediated by IFNAR Signaling

Activation of caspase-11 during gram-negative bacterial infection induces secretion of HMGB1.⁽²⁸⁾ HMGB1 is a nuclear protein that acts as a DAMP when translocated to the cytosol and is secreted actively or passively following necrotic cell death during many inflammatory diseases. To determine the contribution of HMGB1 to Ehrlichia-induced sepsis, we first measured the serum levels of HMGB1 in IOE and EM-infected WT mice on day 7 postinfection. Compared to uninfected and EM-infected mice (mild disease), IOE-infected mice (fatal disease) had significantly higher serum levels of HMGB1 (P < 0.01) (Fig. 6A), suggesting that systemic HMGB1 production is associated with a fatal ehrlichiosis. Next, we examined the contribution of hepatic IFNAR signaling to HMGB1 production by analyzing HMGB1 cytosolic translocation and secretion in HCs following in vitro infection with IOE in the presence or absence of rIFNβ. We detected significant cytosolic translocation of HMGB1 in IOE-infected HCs following stimulation with IFNβ compared to IOE alone or uninfected cells (P < 0.01) (Fig. 6B). Quantitative analysis of nuclear HMGB1 by confocal microscopy demonstrated a significantly higher level of HMGB1 in the nucleus of IOE-infected cells and uninfected cells stimulated with IFNβ compared to the level of nuclear HMGB1 in uninfected HCs or infected cells stimulated with IFNβ (Fig. 6C). This result further confirmed cytosolic translocation in the former conditions. Increased cytosolic translocation of HMGB1 in infected cells following paracrine IFNAR signaling was consistent with higher mRNA expression of HMGB1 compared to controls (Fig. 6D).

We next examined whether paracrine IFNARs promote HMGB1 secretion and whether IFNAR-mediated HMGB1 secretion is dependent on caspase-11. To this end, we measured HMGB1 secretion from IOE-infected WT and caspase-11-deficient (caspase-11^−/−) primary HCs cultured with or without exogenous IFNβ. We found that IFNβ stimulation of IOE-infected WT HCs induced significantly higher secretion of HMGB1 and IL-1α compared to unstimulated but infected cells (Fig. 6E,F). Notably, caspase-11 deficiency significantly decreased HMGB1 and IL-1α secretion from infected HCs stimulated with IFNβ compared to WT controls (Fig. 6E,F). Inhibition of caspase-1 activation in WT HCs also resulted in an attenuated extracellular level of HMGB1 and IL-1α, although the effect was less remarkable compared to the impact of caspase-11 deficiency. Combined inhibition of caspase-11 and caspase-1 pathways in IOE-infected HCs stimulated with IFNβ resulted in complete abrogation of HMGB1 secretion (Fig. 6E,F). Together, these data suggest that IFNAR-mediated HMGB1 secretion is mainly dependent on caspase-11.

IFNβ Signaling Promotes Bacterial Replication in HCs During Ehrlichia Infection

We have previously shown that the increased resistance of IFNAR-deficient mice is associated with decreased bacterial burden in the liver.⁽¹⁹⁾ Thus, we hypothesized that IFNAR signaling could promote bacterial replication in HCs. To test this hypothesis, we measured the total number of live intracellular IOE in primary HCs by qPCR. In vitro stimulation of IOE-infected HCs with rIFNβ significantly enhanced bacterial survival and replication as measured by qPCR of the 16S rDNA at 24 hours postinfection compared to unstimulated uninfected HCs (P < 0.01). This effect was abrogated when infected and stimulated cells were treated with anti-IFNAR blocking antibodies (Fig. 7A) (P < 0.05), confirming that indeed IFNAR signaling is responsible for the increased number of intracellular Ehrlichia.

IFNβ Signaling Increases Autophagy Induction and Autophagic Flux in HCs During Ehrlichia Infection

Because Ehrlichia uses autophagy proteins for their own survival and replication, we hypothesized that IFNAR may enhance bacterial replication in IOE-infected HCs by inducing autophagy. To test this hypothesis, we measured the number of LC3 puncta (as a marker of autophagosome formation) by confocal microscopy as well as conversion of LC3I to LC3II by immunoblot. Confocal microscopy analysis showed that IOE infection of HCs significantly increased the number of LC3 puncta/cells when compared to uninfected cells, with a further increase in the number of LC3 puncta/cells following exogenous stimulation with rIFNβ (Fig. 7B,C). Furthermore, mRNA expression of autophagy-related 12 (ATG12), a major autophagy protein that initiates autophagosome formation, was significantly higher in IOE-infected WT HCs following rIFNβ stimulation compared to unstimulated uninfected controls at 24 hours postinfection (Fig. 7D) (P < 0.05). Consistent with the immunofluorescence data, we detected a significant increase in LC3II expression and higher LC3II/I ratio by immunoblot in IOE-infected HCs following IFNβ stimulation when compared to uninfected or infected HCs cultured without rIFNβ (Fig. 8A). The accumulation of LC3II in IOE-infected and IFNβ stimulated HCs could be due to enhanced autophagosome formation/induction or a reduced degradation of autophagic cargo. To distinguish between these events, we analyzed autophagic flux by measuring the level of p62/SQSTM1, a selective autophagy adaptor/receptor that binds to ubiquitinylated proteins and damaged organelles to target them to autophagosome-lysosomal compartments for degradation. The total cellular p62/SQSTM1 expression levels inversely correlated with autophagic flux and activity. Compared to uninfected HCs, there was slightly less accumulation of p62 in IOE-infected HCs cultured with or without IFNβ, although less accumulation was detected in infected cells following IFNβ stimulation (Fig. 8A). These data suggest that autophagy flux is enhanced in both conditions. Similarly, immunofluorescence staining of LC3B showed that IOE-infected HCs had increased LC3B colocalization with lysosomes (stained by LysoTracker Red) following IFNβ stimulation when compared to uninfected and unstimulated IOE-infected HCs (Fig. 8B,C). These data further confirm that increased LC3II in infected HCs following paracrine IFNAR signaling is due to enhanced autophagy induction but not to blocked flux.

Recently, we demonstrated that induction of autophagy in macrophages is negatively regulated by mTORC1 following infection with IOE.⁽¹⁰⁾ To examine if similar regulation occurs in HCs, we measured mTORC1 activation by immunoblotting. We found that IOE infection induced mTORC1 activation following stimulation with rIFNβ, as marked by pS6, when compared to the uninfected unstimulated cells (Fig. 8D). Together, these results suggest that autophagy regulation in HCs infected with virulent Ehrlichia is mTORC1 independent.

Hepatocyte-Specific Caspase-11 and Caspase-1 Activation In Vivo Correlate With Fatal Outcome

We and others have demonstrated substantial IFNβ production in the liver and spleen of IOE-infected mice that correlates with liver injury, multiorgan failure, sepsis, and death on days 8-10 postinfection.^{(19, 20)} In contrast, mild and self-limited disease in EM-infected mice correlate with lack of IFN-I response. To validate our in vitro data, we isolated primary HCs from the livers of IOE- and EM-infected mice on day 7 postinfection. Our immunoblot data showed that primary HCs from IOE-infected mice exhibit activation of mTORC1, as marked by expression of pS6, as well as cleavage/activation of caspase-11 and caspase-1 compared to HCs from uninfected and EM-infected mice (Fig. 9A-C). Together, these data indicate that hepatic IFNAR signaling mediates caspase-1/caspase-11 activation, which in turn causes release of HMGB1 and inflammatory cell death, which cause liver injury and sepsis following lethal infection.