CpG DNA-mediated Induction of Acute Liver Injury in d-Galactosamine-sensitized Mice

Unmethylated CpG motifs present in bacterial DNA (CpG DNA) induce innate inflammatory responses, including rapid induction of proinflammatory cytokines. Although innate inflammatory responses induced by CpG DNA and other pathogen-associated molecular patterns are essential for the eradication of infectious microorganisms, excessive activation of innate immunity is detrimental to the host. In this study, we demonstrate that CpG DNA, but not control non-CpG DNA, induces a fulminant liver failure with subsequent shock-mediated death by promoting massive apoptotic death of hepatocytes in d-galactosamine (d-GalN)-sensitized mice. Inhibition of mitochondrial membrane permeability transition pore opening or caspase 9 activity in vivo protects d-GalN-sensitized mice from the CpG DNA-mediated liver injury and death. CpG DNA enhanced production of proinflammatory cytokines in d-GalN-sensitized mice via a TLR9/MyD88-dependent pathway. In addition, CpG DNA failed to induce massive hepatocyte apoptosis and subsequent fulminant liver failure and death in d-GalN-sensitized mice that lack TLR9, MyD88, tumor necrosis factor (TNF)-α, or TNF receptor I but not interleukin-6 or -12p40. Taken together, our results provide direct evidence that CpG DNA induces a severe acute liver injury and shock-mediated death through the mitochondrial apoptotic pathway-dependent death of hepatocytes caused by an enhanced production of TNF-α through a TLR9/MyD88 signaling pathway in d-GalN-sensitized mice.

Unmethylated CpG motifs present in bacterial DNA (CpG DNA) induce innate inflammatory responses, including rapid induction of proinflammatory cytokines. Although innate inflammatory responses induced by CpG DNA and other pathogen-associated molecular patterns are essential for the eradication of infectious microorganisms, excessive activation of innate immunity is detrimental to the host. In this study, we demonstrate that CpG DNA, but not control non-CpG DNA, induces a fulminant liver failure with subsequent shock-mediated death by promoting massive apoptotic death of hepatocytes in D-galactosamine (D-GalN)-sensitized mice. Inhibition of mitochondrial membrane permeability transition pore opening or caspase 9 activity in vivo protects D-GalN-sensitized mice from the CpG DNA-mediated liver injury and death. CpG DNA enhanced production of proinflammatory cytokines in D-GalN-sensitized mice via a TLR9/MyD88-dependent pathway. In addition, CpG DNA failed to induce massive hepatocyte apoptosis and subsequent fulminant liver failure and death in D-GalN-sensitized mice that lack TLR9, MyD88, tumor necrosis factor (TNF)-␣, or TNF receptor I but not interleukin-6 or -12p40. Taken together, our results provide direct evidence that CpG DNA induces a severe acute liver injury and shock-mediated death through the mitochondrial apoptotic pathway-dependent death of hepatocytes caused by an enhanced production of TNF-␣ through a TLR9/MyD88 signaling pathway in D-GalN-sensitized mice.
Innate immune cells, such as monocytes/macrophages and dendritic cells (DCs), 3 play a critical role in modulating the host immune response to infection. The innate immune cells recognize conserved molecular patterns present in microbes (pathogen-associated molecular patterns) through a family of proteins known as toll-like receptors (TLRs) that function as pattern recognition receptors and are thereby activated to exert various effector functions, including secretion of pro-inflammatory cytokines and mediators such as TNF-␣, IL-6, IL-12, and nitric oxide (1). Production of inflammatory mediators and pro-inflammatory cytokines by innate immune cells is indispensable for the efficient control of growth and dissemination of invading pathogens. However, excessive and uncontrolled production of inflammatory mediators and pro-inflammatory cytokines caused by bacterial infection is potentially harmful to the host and may lead to severe systematic inflammatory complications, including microcirculatory dysfunction, tissue damage, septic shock, and death (2)(3)(4)(5).
Vertebrate and bacterial DNA have marked differences in the frequency of CpG dinucleotides and their cytosine methylation (only vertebrate DNA shows extensive under-representation and methylation of CpG dinucleotides, termed CpG suppression) (6). Unmethylated CpG dinucleotides in particular sequence contexts (CpG motifs, GACGTT for murine and GTCGTT for human) in bacterial DNA are capable of activating innate immune cells (7) and are one of the well studied pathogen-associated molecular patterns. The ability of bacterial DNA to activate innate immunity can be mimicked by synthetic oligodeoxynucleotides containing the unmethylated CpG motif. This unmethylated CpG motif containing bacterial DNA and synthetic oligodeoxynucleotides (CpG DNA) are endocytosed and then recognized as a conserved molecular pattern by a pattern recognition receptor TLR9 in an endosomal compartment of antigen-presenting cells in the vertebrate innate immune system (8 -14). Upon recognition of CpG DNA, TLR9 recruits the adaptor molecule, myeloid differentiation factor 88 (MyD88), through interaction between their C-terminal toll/IL-1 receptor (TIR) domains. This recruitment of MyD88 to toll/IL-1 receptor domains of TLR9 initiates a signaling pathway that sequentially involves IL-1R-associated kinase (IRAK) family proteins and tumor necrosis factor-␣ receptorassociated factor (TRAF) 6 and TRAF3 (10,(15)(16)(17)(18). This TLR9/MyD88mediated signaling pathway is essential for the CpG DNA-induced activation of NF-B, interferon regulatory factors 5 and 7, small GTPase family protein Ras, mitogen-activated protein kinases, including c-Jun N-terminal kinase, p38, and extracellular signal-regulated kinase, and subsequent production of proinflammatory cytokines, chemokines, and modulators in innate immune cells (10,15,(17)(18)(19)(20)(21)(22)(23)(24)(25).
The CpG DNA-mediated activation of innate immune cells and its beneficial effects on our immune system, as well as its therapeutic appli-cations, have been extensively studied. However, its adverse effects because of uncontrolled proinflammatory response under certain conditions have not been extensively addressed. Moreover, direct and indirect biologic effects of CpG DNA on nonimmune cell activation and its detrimental effects on other organs in the body are largely unknown. Two different mouse models of the septic shock-like syndrome induced by CpG DNA or bacterial DNA have been established previously (26,27). We have demonstrated that pretreatment with CpG DNA or bacterial DNA dramatically potentiates the toxicity of lipopolysaccharide (LPS) by priming mice with IFN-␥, which leads to excessive release of inflammatory cytokines, such as TNF-␣ and IL-6, that are critical in producing the sepsis-like syndrome and eventual death (26). On the other hand, Wagner and co-workers (27) elegantly demonstrate that CpG DNA or bacterial DNA induces septic shock-like death caused by TNF-␣-mediated hepatic injury in mice sensitized with D-galactosamine (D-GalN). These two in vivo mouse models of the CpG DNAmediated sepsis probably occur through different mechanisms and may represent different diseases. The CpG DNA/LPS model may more closely mimic the sepsis syndrome caused by overwhelming bacterial infections, whereas the CpG DNA/D-GalN model may represent the sepsis seen in certain genetically predisposed individuals or with fulminant hepatitis, such as viral hepatitis. The mechanisms by which CpG DNA induces shock-mediated death in these two different in vivo mouse models need to be elucidated, and a better understanding of how exaggerated proinflammatory responses caused by microbial components such as LPS and CpG DNA lead to organ damage and septic shock-mediated death in the host would provide information useful in preventing and treating septic shock and certain acute inflammatory liver diseases. In this study, we have investigated whether CpG DNA directly alters levels or activities of pro-and anti-apoptotic factors in the livers of D-GalN-sensitized mice, and whether severe acute liver injury and the subsequent shock-mediated death of D-GalN-sensitized mice induced by CpG DNA are mediated through the mitochondrial apoptotic pathway-dependent death of hepatocytes caused by an enhanced production of proinflammatory cytokines through a TLR9/MyD88 signaling pathway.

EXPERIMENTAL PROCEDURES
Mice-BALB/cAnNCr mice and C57BL/6NCr mice at 4 -5 weeks of age were obtained from NCI-Frederick, National Institutes of Health, and were used within 3 weeks. TLR9 gene-deficient mice with a BALB/c background (TLR9KO) were provided by Dr. A. Marshak-Rothstein (Boston University, Boston) with permission from Dr. S. Akira (Osaka University, Osaka, Japan). MyD88 gene-deficient mice (MyD88KO) were provided by Dr. S.-C. Hong (Indiana University, Indianapolis) with permission from Dr. S. Akira (Osaka University, Osaka, Japan). C57BL/ 6-Tnfrsf1a tm1Mak (TNFRI KO), B6;129S6-Tnf tm1Gkl (TNF KO), B6.129S2-Il6 tm1Kopf /J (IL-6 KO), B6.129S1-Il12b tm1Jm /J (IL-12p40 KO), and B6129SF2 mice were purchased from The Jackson Laboratory (Bar Harbor, ME). All animal care and housing requirements set forth by the National Institutes of Health Committee on the Care and Use of Laboratory Animals of the Institute of Laboratory Animal Resources were followed, and animal protocols were reviewed and approved by the University of Tennessee Animal Care and Use Committee.
Cell Line and Culture Conditions-Mouse hepatocyte cell line NMH (provided by Dr. N. Fausto at the University of Washington) was cultured at 37°C in a 5% CO 2 -humidified incubator and maintained in Dulbecco's modified Eagle's medium/F-12 supplemented with 5 g/ml insulin, 5 g/ml transferrin, 5 g/ml selenium, 0.04 g/ml dexamethasone, 0.1% gentamicin, 10 mM nicotinamide, and 20 ng/ml epidermal growth factor. RAW264.7 cells (ATCC, Manassas, VA) were cultured at 37°C in a 5% CO 2 -humidified incubator and maintained in Dulbecco's modified Eagle's medium supplemented with 10% (v/v) heat-inactivated fetal calf serum, 1.5 mM L-glutamine, 100 units/ml penicillin, and 100 g/ml streptomycin. All culture reagents were purchased from Invitrogen and Sigma.
Experimental Protocol-Mice were injected intraperitoneally with PBS (200 l), CpG DNA (20 g), control non-CpG DNA (20 g), or LPS (10 g) in the presence or absence of D-GalN (20 mg). The viability of mice was observed for the indicated time periods. In some experiments, mice were anesthetized at designated time periods after injection; blood samples were obtained to prepare serum. To investigate the role of mitochondrial membrane permeability transition, mice were injected intraperitoneally with ethanol (17% v/v in PBS, 100 l) or cyclosporin A (CsA, 500 g in 100 l of 17% ethanol, Sigma) for 1 h before the CpG DNA/D-GalN challenge. In the experiment investigating the role of caspase 9, mice were injected intraperitoneally with Me 2 SO (5% v/v in PBS, 200 l/injection) or a caspase 9 inhibitor (Z-LEHD-fmk; Calbiochem) 1 h before, at the time of, and 1 h after the CpG DNA/D-GalN challenge. Each mouse received a total of 250 g of Z-LEHD-fmk (100 g/200 l of 5% Me 2 SO/injection in the first two injections and 50 g/100 l of 5% Me 2 SO/injection in the last injection). At the end of the designated time periods, mice were killed, and the livers were removed. Liver lobes were excised and then either fixed by submersion in 10% neutral buffered formalin (Fisher) for histopathological analysis or kept at Ϫ80°C for further analysis. In all experiments, 3-10 mice were used for each condition.
Histopathology and Immunohistochemistry-Liver specimens fixed in 10% neutral buffered formalin were embedded in paraffin and sectioned at a thickness of 5 m prior to staining with hematoxylin and eosin (H & E) for light microscopic examination.
Preparation of Cytosolic Extract and Aspartate-specific Cysteine Protease (Caspase) Assays-Cytosolic extracts from liver tissue were prepared by Dounce homogenization in hypotonic extraction buffer (25 mM HEPES, pH 7.5, 5 mM MgCl 2 , 1 mM EGTA, and 1 g/ml each pepstatin, leupeptin, and aprotinin) and subsequently centrifuged (15 min, 13,000 rpm, 4°C). The protein concentration of supernatant was adjusted to 1 mg/ml with extraction buffer and stored at Ϫ80°C. Caspase 3/7, caspase 8, and caspase 9 activities in cytosolic extracts of liver tissue were measured using a Caspase-Glo assay kit (Promega Co., Madison, WI) and modified protocol. Caspase-Glo assay kit contains a proluminescent-specific caspase substrate. After caspase cleavage of the proluminescent caspase substrate, a substrate for luciferase (aminoluciferin) is released; this results in the luciferase reaction and the production of luminescent signal. The signal generated is proportional to the amount of caspase activity present. To measure caspase 3/7, caspase 8, and caspase 9 activities in liver tissues, an equal volume of reagents containing proluminescent caspase 3/7 substrate (DEVD), caspase 8 substrate (LETD), or caspase 9 substrate (LEHD) and 10 g/ml cytosolic protein were added to a white-walled 96-well plate and incubated at room temperature for 1 h. The luminescence of each sample was measured in an L max microplate luminometer (Molecular Devices, Sunnyvale, CA).
Preparation of Whole Cell Lysates and Western Blot Analysis-Whole cell lysates were prepared from liver cells as described previously (13). To detect the presence of the uncleaved form of Bid and the cleaved forms of caspase 3 and actin, equal amounts (15 g/lane) of whole cell lysates were subjected to electrophoresis on a 14% polyacrylamide gel containing 0.1% SDS, and then Western blots were performed using antibodies specific for the uncleaved form of Bid and the cleaved forms of caspase 3 or actin, as described previously (13). Actin was used as a loading control. Antibody specific for the cleaved caspase 3 was purchased from Cell Signaling Technologies. (Beverly, MA). Antibodies specific for the uncleaved form of Bid and actin were purchased from Santa Cruz Biotechnology (Santa Cruz, CA).
Preparation of RNA-free DNA and DNA Fragmentation Analysis-RNA-free DNA was isolated from each liver sample by using DNAzol (Molecular Research Center Inc., Cincinnati, OH) following the manufacturer's protocol. To detect DNA fragmentation, equal amounts (2 g/lane) of DNA were subjected to electrophoresis on a 2% (w/v) agarose gel containing ethidium bromide (0.5 g/ml; Sigma).
Preparation of DNA-free RNA and RT-PCR-DNA-free total RNA was isolated from mouse liver cells, MNH cells, and RAW264.7 cells by using the RNeasy mini kit (Qiagen Inc., Valencia, CA) following the manufacturer's protocol. To measure the relative amount of selected gene transcripts, isolated RNA (1 g from each sample) were reversetranscribed with oligo(dT) primer using Superscript II reverse transcriptase (Moloney murine leukemia virus reverse transcriptase; Invitrogen). One-tenth of the cDNA product was then amplified with gene-specific primers. Twenty to 40 cycles of PCR were conducted. PCR products were separated by 1% agarose gel electrophoresis and visualized. The sequences of RT-PCR primers are listed in supplement 1. All RT-PCR primers were purchased from Integrated DNA Technologies, Inc. (Coralville, IA).

Sequence-specific Induction of an Acute Liver Injury and Subsequent Shock-mediated Death by CpG DNA in D-GalN-sensitized Mice-Pre-
vious studies have demonstrated that the shock-mediated death in D-GalN-sensitized mice or rats by the TLR4 agonist LPS is because of acute liver injury, including hemorrhagic necrosis and massive apoptosis of liver cells (29 -31). Similar to LPS, bacterial DNA and CpG DNA induce shock-mediated death in D-GalN-sensitized mice (27). However, the mechanisms by which this induction occurs have not been extensively studied. We have investigated whether CpG DNA induces shockmediated death in D-GalN-sensitized mice by causing an acute liver injury, including massive hemorrhage and liver cell death by apoptosis in a sequence-specific manner. BALB/c mice were injected intraperitoneally with PBS, CpG DNA, D-GalN, CpG DNA plus D-GalN (CpG DNA/D-GalN), or control non-CpG DNA plus D-GalN (nCpG DNA/D-GalN). The survival of mice was observed for 24 h after the challenge, and liver injury was evaluated by histological examination. In agreement with previous observations (27), we found that CpG DNA in the pres-ence of D-GalN induced shock-mediated death of mice between 10 and 16 h after the challenge (Table 1). In addition, microscopic examination of the livers from CpG DNA/D-GalN-treated mice uniformly revealed signs of severe liver injury, including steatosis, massive hemorrhage, and massive apoptosis of hepatocytes characterized by shrinkage and hypereosinophilia of the cytoplasm, nuclear condensation, and fragmentation (Fig. 1d). However, sinusoidal lining cells, endothelium, and/or Kupffer cells were not affected. These signs of liver damage in mice were observed as early as 4 h post-treatment with CpG DNA/ D-GalN, and the severity was dramatically increased at later time points ( Fig. 1 and data not shown). In contrast, no evidence of hepatic injury was found in mice treated with PBS, CpG DNA alone, D-GalN alone, or control nCpG DNA/D-GalN (Fig. 1). Furthermore, no deaths occurred in mice that received D-GalN alone, CpG DNA alone,

TABLE 1 CpG DNA, but not control non-CpG DNA, induces a shock-mediated death in D-GalN-sensitized mice
BALB/c mice were injected intraperitoneally with PBS, CpG DNA (20 g), non-CpG DNA (20 g), and/or D-GalN (20 mg). Viability of mice was observed 24 h after injection. Eight to 10 mice/group were used for the experiment.

CpG DNA, in a Sequence-specific Manner, Induces Activation of Initiator and Effector Caspases and DNA Fragmentation in Liver Cells of D-GalN-sensitized
Mice-Because histological examination of livers from CpG DNA/D-GalN-treated mice showed all the hallmarks of apoptotic cell death, we have investigated whether CpG DNA induces activation of initiator caspases 8 and 9 and effector caspase 3 and fragmentation of DNA in liver cells of D-GalN-sensitized mice. BALB/c mice were injected intraperitoneally with PBS, CpG DNA, D-GalN, CpG DNA/D-GalN, or control nCpG DNA/D-GalN. Activities of caspases 8 and 9 in liver cells were measured using a Caspase-Glo assay kit. Cleavage of Bid and the activation status of caspase 3 were assayed by observing the disappearance of full-length Bid and the presence of cleaved forms of caspase 3, respectively, in liver cell extracts by Western blot using an antibody specific for the uncleaved form of Bid or cleaved forms of caspase 3. As shown in Fig. 2, A-E, CpG DNA induced activation of initiator caspase 8, cleavage of Bid, and activation of caspase 9 and effector caspase 3 in liver cells of D-GalN-sensitized mice. Furthermore, CpG DNA in combination with D-GalN treatment substantially induced typical apoptotic DNA fragmentation in liver cells within 8 h after administration, indicating induction of massive hepatocyte apoptosis (Fig. 2, F and G). In contrast, control non-CpG DNA alone or in combination with D-GalN failed to induce activation of caspases 3, 8, and 9, Bid cleavage, and DNA fragmentation in liver cells (Fig. 2, A-C, E, and G, and data not shown). Neither CpG DNA nor D-GalN alone induced detectable levels of caspases 3, 8, and 9 activation, Bid cleavage, and DNA fragmentation in liver cells (Fig. 2). These results demonstrate that CpG DNA, in a sequence-specific manner, induces the process of programmed cell death of hepatocytes by inducing activation of initiator and effector caspases and DNA fragmentation in liver cells of D-GalN-sensitized mice.
The Mitochondrial Apoptotic Pathway Plays a Key Role in the CpG DNA-mediated Acute Liver Injury and Shock-mediated Death in D-GalN-sensitized Mice-Our results demonstrated that CpG DNA induces Bid cleavage in hepatocytes in D-GalN-sensitized mice (Fig. 2B). The truncated form of Bid, tBid, has been known to target mitochondria and cause the release of cytochrome c into cytoplasm, which in turn  (20 mg). A and C, mice were killed at 12 h after injection, and liver specimens were sampled. Liver specimens were taken at the time of death if the mice died less than 12 h post-injection. Equal amounts of cytosolic protein (10 g/ml) were subjected to Caspase-Glo assay. Data are presented in mean relative light units (RLU) Ϯ S.D. of each caspase activity in liver samples from three different mice. B, D, and E, mice were killed at the indicated times or 12 h after injection, and liver specimens were sampled. Equal amounts of whole liver cell lysates (30 g/lane) were subjected to SDS-PAGE, and then Western blots were performed using a specific antibody for the full-length Bid (BID) (B), cleaved forms of caspase 3 (D and E), or actin. Actin was used as loading control. Each line represents a liver sample from an individual mouse. Data represent results obtained from three mice. F and G, mice were killed at the indicated times or at 12 h after injection (G), and liver specimens were sampled. Equal amounts (2 g/lane) of DNA from liver specimens were subjected to electrophoresis on a 2% (w/v) agarose gel. Each line represents a liver sample from an individual mouse. Data represent results obtained from three mice.
leads to the activation of caspase 9 and downstream cascades of apoptosis (32)(33)(34). Therefore, we have investigated whether the mitochondrial apoptotic pathway plays a pivotal role in the CpG DNA-mediated hepatic injury and shock-mediated death in D-GalN-sensitized mice. We have employed CsA, which prevents activation of the mitochondrial apoptotic pathway by inhibiting the opening of the mitochondrial membrane permeability transition pores (PTP) (35). BALB/c mice were injected intraperitoneally with vehicle (17% ethanol) or CsA 1 h before the CpG DNA/D-GalN challenge. As shown in Figs. 3 and 4a and in Table 2, CpG DNA/D-GalN treatment in vehicle-pretreated mice caused the expected apoptosis of hepatocytes (assessed from Bid cleavage, caspases 9 and 3/7 activation, and DNA fragmentation), liver tissue damage, and shock-mediated death of the mice. CsA pretreatment did not prevent induction of Bid cleavage in liver cells by CpG DNA/D-GalN (Fig. 3A). However, CpG DNA/D-GalN-mediated activation of caspases 9 and 3/7 and fragmentation of DNA were substantially inhibited in liver cells isolated from mice pretreated with CsA (Fig. 3). In addition, treatment of mice with CsA 1 h before CpG DNA/D-GalN challenge fully protected mice from liver injury (assessed from H & E staining of liver sections) and shock-mediated death ( Fig. 4b and Table 2). These results indicate that CsA prevents CpG DNA-induced hepatocyte apoptosis in D-GalN-sensitized mice, presumably by inhibiting mitochondrial membrane PTP opening, which is a target for the cleaved Bid (tBid) and causes release of cytochrome c and subsequent activation of caspases 9 and 3.
In the mitochondrial apoptotic pathway, opening of PTP leads to a mitochondrial inner transmembrane potential (⌬⌿m) loss that results in the release of proteins of the mitochondrial intermembrane space, including cytochrome c. Cytochrome c interacts with and activates APAF-1, which oligomerizes to form an apoptosome, which in turn recruits and activates caspase 9. The activated caspase 9 cleaves and activates executioner caspases such as caspase 3 (36,37). Because our results demonstrated that an inhibitor of PTP opening CsA inhibits CpG DNA/D-GalN-mediated activation of caspases 9 and 3/7 and subsequent liver damage and shock-mediated death of mice, we further investigated whether inhibition of caspase 9 can protect D-GalN-sensitized mice from the CpG DNA-mediated liver injury and death. BALB/c mice were injected intraperitoneally with Me 2 SO (5% v/v in PBS) or a caspase 9 inhibitor Z-LEHD-fmk 1 h before, at the time of, and 1 h after the CpG DNA/D-GalN challenge. As expected, CpG DNA/D-GalN challenge in vehicle-treated mice caused apoptosis of hepatocytes (assessed from Bid cleavage, caspases 9 and 3/7 activation, and DNA fragmentation), liver tissue damage (assessed from H & E staining of liver sections), and shock-mediated death of the mice (Fig. 3 and Fig. 4c and Table 2). In contrast, CpG DNA/D-GalN-mediated activation of caspases 9 and 3/7 and fragmentation of DNA were substantially inhibited in liver cells isolated from mice treated with Z-LEHD-fmk (Fig. 3,

inhibitor (Z-LEHD-fmk) in vivo protects D-GalN-sensitized mice from shock-mediated death induced by CpG DNA
To investigate the role of mitochondrial membrane permeability transition, BALB/c mice were injected intraperitoneally with ethanol (17% v/v in PBS, 100 l) or CsA (500 g in 100 l of 17% ethanol) 1 h before the CpG DNA (20 g)/D-GalN (20 mg) challenge. To examine the role of caspase 9, BALB/c mice were injected intraperitoneally with Me 2 SO (5% v/v in PBS, 200 l/injection) or a caspase 9 inhibitor (Z-LEHD-fmk in 200 l of 5% Me 2 SO) 1 h before, at the time of, and 1 h after the CpG DNA (20 g)/D-GalN (20 mg) challenge. Each test mouse received a total of 250 g of Z-LEHD-fmk (100 g/injection in the first two injections and 50 g in the last injection). Viability of mice was observed 25 h after injection. Three to 10 mice/group were used for the experiment.

Pretreatment
Treatment Dead/total In addition, treatment of mice with Z-LEHD-fmk fully protected the mice from liver injury (assessed from H & E staining of liver sections) and the shock-mediated death induced by CpG DNA/D-GalN challenge ( Fig. 4d and Table 2). These results demonstrate that activation of caspase 9 precedes caspase 3 activation, DNA fragmentation, and hepatocyte apoptosis and that CpG DNA-induced liver injury and shockmediated death in D-GalN-sensitized mice are mediated through activation of caspase 9. Collectively, our results indicate that the mitochondrial apoptotic pathway may play a critical role in the induction of CpG DNAmediated severe acute liver injury and shock-mediated death of D-GalNsensitized mice.
CpG DNA Induces Acute Liver Injury and Shock-mediated Death in D-GalN-sensitized Mice through a TLR9/MyD88-dependent Pathway-TLR9 has been identified as a receptor for CpG DNA (10). Upon recognition of CpG DNA, TLR9 recruits the adaptor molecule MyD88, which in turn initiates a signaling cascade in innate immune cells (17). Most, if not all, known biologic effects of CpG DNA are dependent on this TLR9/MyD88-signaling pathway. Therefore, we have investigated whether CpG DNA-induced hepatocyte apoptosis, acute liver injury, and shock-mediated death in D-GalN-sensitized mice are also dependent on TLR9 and MyD88. First, to investigate whether CpG DNA induces death of D-GalN-sensitized mice through a TLR9-and/or MyD88-dependent pathway, TLR9 ϩ/ϩ , TLR9 Ϫ/Ϫ , MyD88 ϩ/ϩ , or MyD88 Ϫ/Ϫ mice were injected intraperitoneally with CpG DNA/D-GalN. Survival of mice was observed at 16 h post-injection. As expected, within 16 h after administration, CpG DNA/D-GalN treatment resulted in the death of all wild type mice (both TLR9 ϩ/ϩ and MyD88 ϩ/ϩ ) ( Table  3). In addition, in the presence of D-GalN, LPS (a TLR4 agonist) induced shock-mediated death of both wild type and TLR9KO mice. In contrast, CpG DNA failed to induce shock-mediated death in D-GalN-sensitized TLR9KO mice and D-GalN-sensitized MyD88KO mice (Table 3).
Second, to investigate whether CpG DNA-induced acute liver injury, including hepatocyte apoptosis in D-GalN-sensitized mice, is dependent on TLR9 and/or MyD88, TLR9 ϩ/ϩ , TLR9 Ϫ/Ϫ , MyD88 ϩ/ϩ , or MyD88 Ϫ/Ϫ mice were injected intraperitoneally with CpG DNA/D-GalN. Liver injury was evaluated by histological examination. Hepatocyte apoptosis was evaluated by analyzing activation of caspases 8 and 9, cleavage of Bid and caspase 3, and fragmentation of DNA in liver cells. As expected, livers from CpG DNA/D-GalN-treated wild type mice showed signs of severe liver injury, including steatosis, massive hemorrhage, massive apoptosis of hepatocytes (characterized by shrinkage and hypereosinophilia of the cytoplasm), and nuclear condensation and fragmentation (supplement 2, a and d). In addition, CpG DNA induced activation of caspases 8 and 9, cleavage of Bid and caspase 3, and typical apoptotic DNA fragmentation in liver cells of D-GalN-sensitized wild type mice, indicating induction of massive hepatocyte apoptosis (Fig. 5). In contrast, CpG DNA/D-GalN treatment failed to induce detectable signs of severe liver injury, activation of caspases 8 and 9, cleavage of Bid and caspase 3, and fragmentation of DNA in liver cells in both TLR9KO mice (Fig. 5, A-E and supplement 2b) and MyD88KO mice (Fig. 5, A-D and F and supplement 2e). Of note, LPS also induced severe liver injury, activation of caspases, Bid cleavage, and typical apoptotic DNA fragmentation in liver cells in both D-GalN-sensitized wild type and TLR9KO mice (Fig. 5, B, D, and E, supplement 2c, and data  not shown). Taken together, these results demonstrate that CpG DNA induces hepatocyte apoptosis, severe acute liver injury, and shock-mediated death in D-GalN-sensitized mice through a TLR9/MyD88-dependent signaling pathway.

Mouse Liver Cells Express TLRs and TLR Downstream Signaling Modulators-Recent studies have demonstrated that expression of
TLRs is not restricted to innate immune cells. Toll-like receptors, including TLR4 and TLR9, have been found in human primary hepatocytes and a hepatocyte cell line (HepG2), and a TLR4 ligand, LPS, has been shown to induce activation of NF-B in mouse hepatocytes (38). These observations suggest that hepatocytes could play an active role in inflammation during sepsis. Therefore, we investigated whether mouse primary liver cells and a mouse hepatocyte cell line, NMH, express TLR9, and if so, whether CpG DNA in the presence of D-GalN directly induces hepatocyte apoptosis. Our RT-PCR results revealed that both NMH cells and primary liver cells express TLR 3, TLR4, and TLR5 (supplement 3). In addition, both NMH cells and mouse primary liver cells expressed common TLR downstream signaling modulators, including MyD88, IL-1R-associated kinase family proteins, and TRAF6. However, very small amounts of TLR2 message and TLR7 message were present in NMH and mouse primary liver cells. Very small amounts of TLR9 message were present in mouse primary liver cells, and almost no TLR9 message was detected in NMH cells (supplement 3). In addition, we could not detect the presence of TLR9 protein in NMH cells and mouse primary liver cells under our experimental conditions (data not shown). We also investigated whether CpG DNA directly induces apoptosis of NMH cells in the presence or absence of D-GalN. Regardless of the presence of D-GalN, CpG DNA did not induce hepatocyte apoptosis (data not shown). While ruling out the possibility that CpG DNA may directly affect survival of hepatocytes, these results suggest that CpG DNA may induce D-GalN-sensitized mouse hepatocyte apo-ptosis indirectly by promoting production of hepatocyte apoptosisinducing mediators in other cells, such as macrophages/monocytes and DCs.
CpG DNA Enhances Cytokine Production in D-GalN-sensitized Mice-Up-regulation of proinflammatory cytokines has been shown to be correlated with the induction of liver injury by LPS in D-GalN-sensitized mice and rats (39,40). Proinflammatory cytokines TNF-␣ and IL-6 also have been shown to induce massive apoptosis of hepatocytes and subsequent fulminant liver failure in D-GalN-sensitized mice and rats (41)(42)(43). In addition, Sparwasser et al. (27) showed that CpG DNA fails to induce shock-mediated death in either anti-TNF-␣-treated or TNFRI gene-deficient D-GalN-sensitized mice. Moreover, our results from in vitro experiments (data not shown) suggested that the soluble factor(s) produced by monocytic cells after CpG DNA stimulation is necessary for the induction of hepatocyte apoptosis in D-GalN-sensitized mice. Although these results indicate that CpG DNA induces fulminant liver injury and shock-mediated death in D-GalN-sensitized mice through a proinflammatory cytokine (especially TNF-␣)-dependent manner, to date it is not known whether CpG DNA actually has the ability to induce proinflammatory cytokines in D-GalN-sensitized mice. BALB/c mice were injected intraperitoneally with PBS, D-GalN, CpG DNA, or CpG DNA/D-GalN. Concentrations of selected cytokines in serum were measured by ELISA. As expected, CpG DNA rapidly induced systemic production of proinflammatory cytokines IL-1␤, TNF-␣, IL-6, and IL-12p40 (Fig. 6). D-GalN alone induced production of relatively small amounts of IL-1␤ and TNF-␣ within 1 h after treatment, but these cytokine levels went down to the basal levels within 4 h after treatment (Fig. 6, A and B). D-GalN also induced production of IL-6; levels peaked at 8 h after the treatment and returned to the basal level by 12 h after treatment (Fig. 6C). In contrast, D-GalN alone did not induce production of detectable amounts of IL-12p40 (Fig. 6D). Interestingly, CpG DNA did not enhance IL-1␤ production in D-GalN-sensitized mice (Fig. 6A). However, CpG DNA enhanced production of TNF-␣, IL-6, and IL-12p40 in D-GalN-sensitized mice (Fig. 6, B-D). During the early time periods, the amounts of cytokines produced after CpG DNA/ D-GalN treatment were much lower than those produced after treatment with CpG DNA alone. In contrast to these early responses, serum concentrations of TNF-␣, IL-6, and IL-12p40 at 12 h after the challenge, the time point when mice start die because of liver failure and subsequent shock, were higher in CpG DNA/D-GalN-treated mice compared with the serum cytokine concentrations in mice treated with CpG DNA alone. These results suggest that similar to the effect of LPS in D-GalNsensitized mice, the enhanced production of proinflammatory cytokines by CpG DNA in D-GalN-sensitized mice may correlate with acute liver injury and subsequent shock-mediated death.

CpG DNA-mediated Enhanced Production of Proinflammatory Cytokines in D-GalN-sensitized Mice Is Dependent on TLR9 and MyD88-It
has been demonstrated that cytokine production in macrophages/ monocytes in response to CpG DNA is mediated through a TLR9 and its adaptor MyD88-dependent signaling pathway (10,17). Our results from experiments using TLR9KO and MyD88KO (Fig. 5, Table 3, and supplement 2) demonstrated that CpG DNA induces fulminant liver failure and death in D-GalN-sensitized mice through a TLR9/MyD88dependent signaling pathway. Therefore, if CpG DNA-induced acute liver injury and shock-mediated death in D-GalN-sensitized mice are correlated with the enhanced production of proinflammatory cytokines by CpG DNA in these mice, we would expect that the CpG DNA-mediated enhanced systemic production of proinflammatory cytokines in D-GalN-sensitized mice is also dependent on TLR9 and MyD88. To test this possibility, BALB/c mice (TLR9 ϩ/ϩ ), TLR9KO (TLR9 Ϫ/Ϫ ), MyD88KO (MyD88 Ϫ/Ϫ ), and their wild type littermates (MyD88 ϩ/ϩ ) were injected intraperitoneally with PBS or CpG DNA/D-GalN. Amounts of selected proinflammatory cytokines in serum were measured by ELISA. As expected, CpG DNA induced systemic produc-tion of proinflammatory cytokines TNF-␣, IL-6, and IL-12p40 in all wild type mice sensitized with D-GalN. In contrast, CpG DNA failed to induce systemic production of these proinflammatory cytokines in D-GalN-sensitized-TLR9KO mice and D-GalN-sensitized-MyD88KO mice (Fig. 7). These results demonstrated that CpG DNA induces production of proinflammatory cytokines in the D-GalN-sensitized mice through a TLR9/MyD88-dependent signaling pathway. Our results also suggest that CpG DNA-mediated liver injury and shock-mediated death of D-GalN-sensitized mice may be due to the TLR9/MyD88-dependent production of proinflammatory cytokines.

CpG DNA Fails to Induce Acute Liver Injury and Shock-mediated Death in D-GalN-sensitized Mice That Lack TNF-␣ or TNFRI-Previous
studies have demonstrated that proinflammatory cytokines TNF-␣ and IL-6 play a critical role in the induction of liver injury in D-GalN-sensitized mice and rats (41,43,44). Therefore, we investigated whether any of the proinflammatory cytokines produced in response to CpG DNA in D-GalN-sensitized mice play a crucial role in the induction of severe acute liver injury and shock-mediated death. First, to test whether TNF-␣, IL-6, and/or IL-12 play pivotal roles, wild type mice or mice that lack TNF-␣, IL-6, or IL-12p40 were injected intraperitoneally with CpG DNA/D-GalN. Survival of mice was observed at 25 h post-injection. As shown in Table 4, over 77% of wild type mice (B6129SF2 and C57BL/6) were dead within 16 h post-administration of CpG DNA/D-GalN. In contrast, no deaths were observed in TNF Ϫ/Ϫ mice and TNFRI Ϫ/Ϫ mice within 25 h after CpG DNA/D-GalN treatment, indicating that CpG DNA failed to induce shock-mediated death in D-GalN-sensitized mice in the absence of TNF-␣ or a TNF-␣ signal transduced through TNFRI. CpG DNA-induced IL-6, but not IL-12p40, production was substantially inhibited in D-GalN-sensitized TNF Ϫ/Ϫ mice and D-GalN-sensitized TNFRI Ϫ/Ϫ mice, indicating CpG DNA-induced IL-6 production is partially dependent on TNF-␣ in D-GalN-sensitized mice (data not shown). Similar to TNF Ϫ/Ϫ mice and TNFRI Ϫ/Ϫ mice, none of the IL-6 Ϫ/Ϫ mice died within 25 h after CpG DNA/D-GalN treatment (Table 4). However, unlike TNF Ϫ/Ϫ mice and TNFRI Ϫ/Ϫ mice that showed no apparent signs of illness, IL-6 Ϫ/Ϫ mice showed signs of illness (rough coat and lack of reaction to cage motion) at the conclusion of the experiment. In contrast to TNF Ϫ/Ϫ mice, TNFRI Ϫ/Ϫ mice, and IL-6 Ϫ/Ϫ mice, 75% of IL-12p40 Ϫ/Ϫ died within 10 h after CpG DNA/ D-GalN treatment (Table 4).
Second, to investigate whether CpG DNA-induced hepatocyte apoptosis and severe liver injury in D-GalN-sensitized mice are also dependent on proinflammatory cytokines produced in response to CpG DNA, wild type, TNF Ϫ/Ϫ , TNFRI Ϫ/Ϫ , IL-6 Ϫ/Ϫ , and IL-12p40 Ϫ/Ϫ mice were injected intraperitoneally with CpG DNA/D-GalN. Liver cell apoptosis was evaluated by analyzing activation of caspases 8 and 9, cleavage Bid and caspase 3, and fragmentation of DNA in liver cells. Liver injury was also determined by histological examination of livers. As expected, CpG

Strain
Treatment The surviving mice at the time of data collection in these groups showed signs of illness (rough coat and a lack of reaction to cage motion) and were euthanized without further observation.
DNA induced activation of caspases 8 and 9, cleavage of Bid and caspases 3, and fragmentation of DNA in liver cells of all D-GalN-sensitized wild type mice (B6129SF2 and C57BL/6), indicating induction of massive hepatocyte apoptosis (Fig. 8). In addition, microscopic examination of livers from all wild type mice treated with CpG DNA/D-GalN uniformly revealed signs of severe liver injury, including infiltration of lipids, massive hemorrhage, and apoptotic death of hepatocytes, characterized by moderate granular condensation of the cytoplasm and a hyperchromatic nuclear membrane (supplement 4, a and c). CpG DNA also induced activation of caspases 8 and 9, cleavage of Bid and caspase 3, and fragmentation of DNA in liver cells and subsequent severe liver damage in IL-6 Ϫ/Ϫ mice and IL-12p40 Ϫ/Ϫ (Fig. 8 and supplement 4, e and f). However, livers from TNF Ϫ/Ϫ or TNFRI Ϫ/Ϫ mice treated with CpG DNA/D-GalN did not show any sign of severe hepatic injury, activation of caspases, Bid cleavage, or DNA fragmentation in liver cells ( Fig. 8 and supplement 4, b and d). These results demonstrate that TNF-␣, but not IL-6 or IL-12p40, plays a critical role in the CpG DNAmediated induction of fulminant liver injury and shock-mediated death in D-GalN-sensitized mice. Taken together, these results provide direct evidence that TNF-␣ produced through a TLR9/MyD88-dependent pathway in response to CpG DNA stimulation and its signal transduced through TNFRI play a pivotal role in the induction of a severe acute liver injury (including caspase activation and the resulting massive hepatic apoptosis) and subsequent shock-mediated death caused by CpG DNA in D-GalN-sensitized mice.

DISCUSSION
The present study attempts to address questions of whether CpG DNA alters expression or activity of pro-and anti-apoptotic factors in livers of D-GalN-sensitized mice, and whether severe acute liver injury and subsequent shock-mediated death of D-GalN-sensitized mice induced by CpG DNA are mediated through the mitochondrial apoptotic pathway-dependent death of hepatocytes caused by an enhanced production of proinflammatory cytokines through a TLR9/MyD88 sig-naling pathway. Regardless of the mouse strain, CpG DNA induces a severe acute liver injury in a sequence-specific manner and subsequent shock-mediated death in mice treated with the hepatocyte-specific transcription inhibitor D-GalN ( Fig. 1 and Table 1). The liver injury and shock-mediated death induced by CpG DNA in D-GalN-sensitized mice are very similar to that induced by LPS or proinflammatory cytokine TNF-␣ in D-GalN-sensitized mice. As assessed from H & E staining of liver sections (Fig. 1), the fulminant liver injury in D-GalN-sensitized mice induced by CpG DNA may be due mainly to the massive hepatocyte apoptosis. No visible damage in sinusoidal lining cells, endothelial cells, or Kupffer cells was observed in livers of CpG DNA/D-GalNtreated mice. In addition, CpG DNA-induced apoptotic cell death occurred almost exclusively in the liver in D-GalN-sensitized mice ( Fig.  1 and data not shown). There was no detectable apoptosis in other organs in the CpG DNA/D-GalN-treated mice under our experimental conditions (data not shown). In contrast, previous studies have shown that LPS causes apoptotic cell death in the kidney, thymus, spleen, and lymph nodes as well as the liver in D-GalN-sensitized mice (31). These findings suggest that the effects of CpG DNA in D-GalN-sensitized mice may be different to some degree than those of LPS.
Similar to TNF-␣/D-GalN (45) or LPS/D-GalN (30) treatment, CpG DNA/D-GalN treatment causes extensive hemorrhage (Fig. 1). The mechanism by which these inflammatory mediators induce extensive hemorrhage remains unexplained. It has been suggested that hepatic hemorrhage caused by TNF-␣/D-GalN may be due to edema of hepatocellular microvilli, widening of sinusoidal endothelial fenestrae, transmigration of erythrocytes and platelets to the space of Disse without exfoliation and necrosis of the sinusoidal endothelial cells, and fibrin deposits in areas adjacent to injured hepatocytes (45). A recent study suggested that hepatic hemorrhage in LPS/D-GalN-treated mice reflects a break in the integrity of microvascular endothelium with attendant extravasation of erythrocytes and formation of intravascular platelet aggregates that is accompanied by a precipitous drop in circulating platelets and increases in the level of fibrin degradation product (a marker of intravascular coagulation) and plasminogen activator inhibitor-1, which promotes vascular thrombosis (30). Long term multiple administrations of CpG DNA cause systemic reductions in cholesterol, platelet counts, erythrocyte counts, and hematocrit while increasing blood urea nitrogen (46). However, it has yet to be explored whether an acute administration of CpG DNA or CpG DNA/D-GalN causes decreases in circulating platelets and/or increases in fibrin degradation product generation and plasminogen activator inhibitor-1 expression. Liver sections stained with H & E from LPS/D-GalN-treated mice versus CpG DNA/D-GalN-treated mice were almost indistinguishable under the light microscope. Further investigation is needed to determine whether TNF-␣, LPS, and CpG DNA induce extensive hepatic hemorrhage in D-GalN-sensitized mice through the same mechanism.
Cell death signals transmitted through Fas or TNFRI lead to activation of a series of intracellular caspases that exist as inactive proenzymes, including caspases 8, 9, and 3/7. The active caspase 8 causes cleavage of Bid. Together with other Bcl 2 family members, Bad and Bax, the truncated form of Bid (tBid) targets mitochondria and causes the release of cytochrome c into cytoplasm. The consequent accumulation of cytochrome c in the cytosol leads to the activation of caspase 9. Caspase 3 is an executioner caspase that can be activated directly by caspase 8 or by caspase 9. This cascade of proteolytic events mediated by caspases leads to nucleosomal DNA fragmentation and chromatin condensation (47,48). Previous studies have demonstrated that in the D-GalN-sensitized mouse model, cytochrome c release and caspase 9 activation are required for the activation of caspase 3 in TNF-␣-induced hepatocyte apoptosis, but caspases 8 and 2 play a minimal role, if any (47). It has also been demonstrated that administration of LPS/D-GalN induces activation of caspases 8, 9, and 3/7, cleavage of Bid, and release of cytochrome c from mitochondria to the cytosol in liver cells (49). In addition, Bernardi and co-workers (35) have demonstrated that CsA inhibits opening of permeability transition pores ex vivo and prevents the hepatotoxic effects of TNF-␣ in LPS/D-GalN rat model by blocking the mitochondrial proapoptotic pathway. Similar to the LPS and TNF-␣ death signal transmitted through the TNFRI (47,49,50), we found that CpG DNA induces a cascade of apoptosis that includes activation of initiator caspases 8 and 9, truncation of Bid, activation of caspase 3, and typical nucleosomal DNA fragmentation (Fig. 2). In addition, our results using a mitochondrial PTP opening inhibitor (CsA) and a caspase 9 inhibitor (Z-LEHD-fmk) (Fig. 3 and 4 and Table 2) indicate that mitochondrial PTP opening precedes caspase 9 activation, and caspase 9 plays a pivotal role in caspase 3 activation in CpG DNA-mediated hepatocyte apoptosis in the D-GalN-sensitized mice. Hence, the mitochondrial proapoptotic pathway plays an indispensable role in this CpG DNA-mediated liver injury and death of D-GalN-sensitized mice. However, it has yet to be revealed whether CpG DNA activates caspase 2 and whether caspase 8 plays a role in the CpG DNA-mediated activation of caspases 3 and 9 in the D-GalN-sensitized mice.
To date, all studied biologic responses induced by CpG DNA are dependent on a TLR9/MyD88 signaling pathway. Our results demonstrated that CpG DNA-induced apoptotic death of hepatocytes, severe liver injury, and shock-mediated death of D-GalN-sensitized mice are also dependent on a TLR9/MyD88 signaling pathway (Fig. 5, Table 3, and supplement 2). A recent study has shown that human primary hepatocytes and a hepatocyte cell line (HepG2) express TLRs, including TLR4 and TLR9, and the TLR4 ligand LPS induces activation of NF-B in mouse hepatocytes (38), suggesting the possibility that hepatocytes could play an active role in inflammation during sepsis. Therefore, although a previous study (27) has shown that TNF-␣ plays a critical role in the CpG DNA-mediated death of D-GalN-sensitized mice, it became necessary to explore whether hepatocytes can directly respond to CpG DNA and whether CpG DNA in the presence of D-GalN directly induces hepatocyte apoptosis. We found that, unlike human hepatocytes, the mouse hepatocyte cell line NMH and primary liver cells express very small amounts, if any, of TLR9 message and protein and that CpG DNA does not directly induce hepatocyte apoptosis regardless of the presence of D-GalN (supplement 3 and data not shown). These results rule out the possibility that CpG DNA directly affects the survival of hepatocytes and suggest that CpG DNA may induce D-GalNsensitized mouse hepatocyte apoptosis indirectly by promoting production of hepatocyte apoptosis-inducing mediators in other cells, such as macrophages/monocytes and DCs.
Several cytokines have been shown to play a critical role in liver injury and shock-mediated death caused by various stimuli. In endotoxin shock, fulminant hepatitis, and virus-associated hemophagocytic syndrome, TNF-␣ has been shown to be a key cytokine in cell injury and host damage (51). TNFs are critical cytokines for inducing T cell activation-associated (ConA-induced) hepatitis, and the induction of hepatitis is almost completely controlled by recombinant IL-6 through multiple mechanisms, including the reduction of TNF production (52). In the LPS/D-GalN model, LPS induces the release of variety of cytokines, including TNF-␣, IFN-␥, IL-6, IL-10, and IL-12 (53,54). Among these cytokines, TNF-␣, IFN-␥, and IL-6 are detrimental mediators of LPS/D-GalN-induced liver failure and lethality, whereas IL-10 is hepatoprotective (39,(55)(56)(57)(58)(59). The detrimental effects of IFN-␥ and protective effects of IL-10 correlate with their ability to regulate TNF-␣ expression positively and negatively, respectively (56,58). However, the detrimental effects of IL-6 in D-GalN-sensitized mice have been demonstrated to be independent on TNF-␣ (43). Although CpG DNA alone induces production of both IFN-␥ and IL-10 in mice (26,60), neither IFN-␥ nor IL-10 was produced at a detectable level in mice treated with CpG DNA/D-GalN under our experimental conditions (data not shown). CpG DNA induces production of proinflammatory cytokines, with similar kinetics to those seen in the LPS/D-GalN-treated rat (49), that precede the histological events of apoptotic changes in liver cells. The TNF-␣ level reaches a maximum at 1 h and is down to normal at 8 h after CpG DNA/D-GalN treatment, whereas IL-6 and IL-12p40 levels reach maximums between 4 and 12 h and stay at the maximal levels (Fig. 6). Approaching the time of death, serum concentrations of IL-6 and IL-12p40 were greatly elevated in mice treated with CpG DNA/D-GalN versus mice treated with D-GalN or CpG DNA alone (Fig. 6, C and D). Similar to its role in the LPS/D-GalN model, we found that TNF-␣ plays a critical role in the CpG DNA/D-GalN-mediated liver injury and shock-mediated death. CpG DNA failed to induce hepatocyte apoptosis, severe acute liver injury, or shock-mediated death in mice that lack either TNF-␣ or TNF-RI (Fig. 8, Table 4, and supplement 4). In addition, like CpG DNA-mediated hepatocyte apoptosis, liver injury, and lethality in D-GalN-sensitized mice ( Fig. 5 and supplement 2), induction of TNF-␣ production by CpG DNA was completely inhibited in TLR9KO and MyD88KO mice (Fig. 7). These results provided direct evidence that the enhanced production of TNF-␣ induced by CpG DNA in D-GalN-sensitized mice is responsible for the acute liver injury and subsequent shock-mediated death. In contrast, IL-12p40 was neither a detrimental nor a protective mediator in the CpG DNA-induced severe liver injury and shock-mediated death in D-GalNsensitized mice (Fig. 8, Table 4, and supplement 4). Although signs of severe liver injury, including hepatocyte apoptosis, were not attenuated, the deaths of CpG DNA/D-GalN-treated IL-6-deficient mice were delayed (Fig. 8, Table 4, and supplement 4). Neither LPS/D-GalN-nor CpG DAN/D-GalN-induced TNF-␣ production is suppressed in IL-6-deficient mice (43), 4 indicating delayed death of CpG DNA/D-GalN-treated IL-6-deficient mice is not because of the attenuated production of TNF-␣. Currently, it is not known how the lack of IL-6 results in delayed death of CpG DNA/D-GalN mice without protecting mice from severe acute liver injury and hepatic apoptosis. IL-6 can regulate systemic production of other pro-inflammatory mediators that contribute to shock-mediated death. Therefore, it is possible that contribution of IL-6 in the CpG DNA/D-GalN-mediated death is through yet to be determined extrahepatic factors. Whether the liver is the only target of IL-6 that mediates lethality of D-GalN-sensitized mice needs to be determined. Whether IFN-␥ plays a pivotal role and IL-10 plays a protective role in CpG DNA-mediated hepatocyte apoptosis in D-GalN-sensitized mice and whether these cytokines play a role in expression of TNF-␣ in CpG DNA/D-GalN-treated mice have yet to be clarified.
By demonstrating the ability of CpG DNA or bacterial DNA to induce high levels of systemic TNF-␣ production in BALB/c mice in the absence of D-GalN sensitization and the inability of CpG DNA to induce death in D-GalN-sensitized mice treated with neutralizing antibody for TNF-␣ or in D-GalN-sensitized mice lacking TNF-RI, Wagner and coworkers (27) suggested that death of D-GalN-sensitized mice after CpG DNA stimulation is because of sepsis caused by an overwhelming systemic proinflammatory response. Death from septic-shock syndrome caused by exaggerated systemic inflammatory response is characterized by damage to multiple organs, including liver, lung, kidney, and intestine, caused by necrosis and apoptosis (55,61,62). In addition, in the liver, DNA fragmentation is not restricted to hepatocytes but also occurs in nonparenchymal cells (55). In contrast, we found that although CpG DNA enhances proinflammatory cytokine production in D-GalN-sensitized mice, the levels of proinflammatory cytokines, including TNF-␣, were rather moderate and not near the levels that can cause shock-mediated death (Fig. 6). In addition, we found that CpG DNA-induced death of D-GalN-sensitized mice actually resulted exclusively from massive apoptotic death of hepatocytes and complete liver destruction, without affecting sinusoidal lining cells, endothelial cells, and Kupffer cells (Fig. 1). These results imply that CpG DNA-induced acute liver injury and subsequent shock-mediated death in D-GalNsensitized mice may be not due to the sepsis mediated by an overwhelming systemic inflammatory response but due to alterations in hepatocytes by D-GalN, including blockade of transcription of yetto-be-identified gene(s) involved in suppression of the TNF-RItransmitted death signal.
In summary, this study demonstrates that CpG DNA-mediated fulminant liver injury and shock-mediated death in D-GalN-sensitized mice is because of the mitochondrial apoptotic pathway-dependent death of hepatocytes caused by an enhanced production of TNF-␣ through a TLR9/MyD88 signaling pathway.