Cellular-specific role of toll-like receptor 4 in hepatic ischemia-reperfusion injury in mice§


  • Potential conflict of interest: Dr. Hackam received grants from Abbott and Vertex.

  • This work was supported by a Howard Hughes Medical Institute Physician-Scientist Award (to A.T.), R01-GM95566 (to A.T.), R01-GM50441(to T.R.B.), an Association of Academic Surgery Foundation Research Fellowship Award (to G.W.N.), an ACS Resident Research Scholarship (to J.R.K.), and an American Heart Association Predoctoral Fellowship (to B.R.R.).


Ischemia-reperfusion (I/R) injury is a process whereby an initial hypoxic insult and subsequent return of blood flow leads to the propagation of innate immune responses and organ injury. The necessity of the pattern recognition receptor, Toll-like receptor (TLR)4, for this innate immune response has been previously shown. However, TLR4 is present on various cell types of the liver, both immune and nonimmune cells. Therefore, we sought to determine the role of TLR4 in individual cell populations, specifically, parenchymal hepatocytes (HCs), myeloid cells, including Kupffer cells, and dendritic cells (DCs) subsequent to hepatic I/R. When HC-specific (Alb-TLR4−/−) and myeloid-cell–specific (Lyz-TLR4−/−) TLR4 knockout (KO) mice were subjected to warm hepatic ischemia, there was significant protection in these mice, compared to wild type (WT). However, the protection afforded in these two strains was significantly less than global TLR4 KO (TLR4−/−) mice. DC-specific TLR4−/− (CD11c-TLR4−/−) mice had significantly increased hepatocellular damage, compared to WT mice. Circulating levels of high-mobility group box 1 (HMGB1) were significantly reduced in Alb-TLR4−/− mice, compared to WT, Lyz-TLR4−/−, CD11c-TLR4−/− mice and equivalent to global TLR4−/− mice, suggesting that TLR4-mediated HMGB1 release from HCs may be a source of HMGB1 after I/R. HCs exposed to hypoxia responded by rapidly phosphorylating the mitogen-activated protein kinases, c-Jun-N-terminal kinase (JNK) and p38, in a TLR4-dependent manner; inhibition of JNK decreased release of HMGB1 after both hypoxia in vitro and I/R in vivo. Conclusion: These results provide insight into the individual cellular response of TLR4. The parenchymal HC is an active participant in sterile inflammatory response after I/R through TLR4-mediated activation of proinflammatory signaling and release of danger signals, such as HMGB1. (HEPATOLOGY 2013)

Thorough understanding of the pathophysiology of hepatic ischemia-reperfusion (I/R) is vital because it is commonly encountered clinically during elective liver surgical procedures, solid organ transplantation, trauma, and hypovolemic shock. The liver exhibits both direct cellular damage as the result of ischemic insult as well as further dysfunction and damage resulting from activation of inflammatory pathways.1 Although the distal events involved in the inflammatory response resulting in liver damage after I/R injury have been well studied,2, 3 proximal events dictating the propagation of the inflammatory response and further tissue damage remain poorly understood.

The role of Toll-like receptor (TLR)4 in the recognition of endotoxin is well established; however, only recently has it become apparent that TLR4 also participates in the response to acute tissue injury.4-6 TLR4 has been found to be an essential constituent of I/R injury, with mice lacking intact TLR4 signaling being significantly protected from injury.5-8 The liver has a complex cellular composition consisting of parenchymal hepatocytes and nonparenchymal cells (NPCs), such as Kupffer cells (KCs), hepatic dendritic cells (DCs), stellate cells, and natural killer cells, with functional TLR4 found ubiquitously.9, 10 Previous work has investigated whether parenchymal hepatocytes (HCs) or NPCs are the TLR4-responsive population in sterile inflammatory response.5, 8, 11 Our studies with TLR4 chimeric mice demonstrate that in the setting of noninfectious I/R-induced injury, bone-marrow (BM)-derived cells are primarily responsible for TLR4-dependent hepatocellular injury.7 In contrast, other studies have also suggested a role for parenchymal/non-BM-derived cells contributing to TLR4-dependent injury.8, 11 Therefore, the role of TLR4 on specific cell types is still unclear.

The aim of our study was to investigate the role of TLR4 on various cell types of the liver, both parenchymal and immune, during hepatic I/R using cellular-specific TLR4 knockout (KO) mice. This is unique from other studies, where the more global effect of TLR4 on the liver has been investigated. In this work, we have generated transgenic (Tg) cell-specific TLR4 KO mice to illustrate the dichotomous role of TLR4 after I/R. We find that TLR4 on DCs contributes primarily a protective role, whereas TLR4 on both HCs and myeloid cells promotes injury. In addition to immune cells, HCs are identified as one of the key cellular constituents in the innate immune response associated with I/R. These findings represent an advance over previous knowledge, given the important cell-specific findings.


Abs, antibodies; BM, bone marrow; cDNA, complementary DNA; DAMP, damage-associated molecular pattern; DC, dendritic cell; ECs, endothelial cells; ELISA, enzyme-linked immunosorbent assay; ERK, extracellular signal-regulated kinase; HCs, hepatocytes; HMGB1, high-mobility box 1; HO-1, heme oxygenase 1; IF, immunofluorescent; IHC, immunohistochemistry; IL, interleukin; I/R, ischemia-reperfusion; IRF-1, interferon regulatory factor 1; JNK, c-Jun-N-terminal kinase; KC, Kupffer cell; KO, knockout; LDH, lactate dehydrogenase; LPS, lipopolysaccharide; MAP, mitogen-activated protein; mRNA, messenger RNA; NF-κB, nuclear factor kappa B; NPC, nonparenchymal cell; PRR, pattern recognition receptor; RT-PCR, reverse-transcriptase polymerase chain reaction; sALT, serum alanine aminotransferase; SD, standard deviation; SEM, standard error of the mean; Tg, transgenic; TLR, Toll-like receptor; TNF-α, tumor necrosis factor alpha; WT, wild type.

Materials and Methods


Male wild-type (WT) (TLR4loxP/loxP) mice, cell-specific, and global TLR4−/− mice were bred at our facility and used at the age of 8-12 weeks. All mice developed were on a C57BL/6 genetic background. Animal protocols were approved by the animal care and use committee of the University of Pittsburgh (Pittsburgh, PA), and experiments were performed in strict adherence to the National Institutes of Health Guidelines for the Use of Laboratory Animals.

Generation of TLR4loxP/loxP and Cellular-Specific TLR4−/− Mice.

In brief, the TLR4loxP allele was created by inserting loxP sites within introns 1 and 2 and flanking exon 2 of TLR4. An overview of this construct is shown in Supporting Fig. 1. Mice homozygous for TLR4loxP were generated by Ozgene (Bentley, WA). TLR4loxP/loxP mice were interbred with stud males (TLR4loxP/−; Alb-cre, TLR4loxP/−; Lyz-cre, or TLR4loxP/−; CD11c-cre) to generate the desired genotype. Mice homozygous for Cre recombinase linked to the albumin (alb), CD11c (cd11c), and lysozyme (lyz) promoter are commercially available from The Jackson Laboratory (Bar Harbor, ME). Tg male mice used for experiments were confirmed to be a desired genotype by standard genotyping techniques. WT mice used in this study were TLR4loxP/loxP mice without the introduction of Cre recombinase. Global TLR4−/− mice were globally lacking the loxP flanked exon 2 (i.e., they were global homozygotes for the same mutation contained within the conditional KO mice). Sodhi et al. have recently provided a detailed description of the novel TLR4−/− mice used in this study.12

Genomic and Functional Characterization of Tg TLR4−/− Mice.

For confirmation of TLR4 messenger RNA (mRNA) expression in WT, Alb-TLR4−/−, and TLR4−/−, we first isolated HCs, NPCs, or tissue using the Qiagen RNeasy Mini Kit (Qiagen, Valencia, CA) to isolate RNA and Clontech Sprint RT Complete-Double PrePrimed (Clontech, Mountain View, CA) to make complementary DNA (cDNA). For confirmation of TLR4 expression in Lyz-TLR4−/−, we first isolated peritoneal macrophages and performed positive selection using F4/80 beads (BD Biosciences, San Jose, CA). Specific primers were as follows: forward 5'-TGCCACCAGTTACAGATCGTC-3' and reverse 5'-GAGTTTCTGATCCATGCATTGG-3' for TLR4, and β-actin primers, as described previously.5 Response of NPCs from WT, Alb-TLR4−/−, or TLR4−/− mice and isolated macrophages from WT, Lyz-TLR4−/−, or TLR4−/− mice was determined by exposing cells to 10 ng/mL of lipopolysaccharide (LPS; Sigma-Aldrich, St. Louis, MO) for 6 hours and using tumor necrosis factor alpha (TNF-α) or interleukin (IL)-6 enzyme-linked immunosorbent assay (ELISA) for quantification (R&D Systems, Minneapolis, MN). Confirmation that KCs in CD11c-TLR4−/− mice retained functional TLR4 was accomplished by isolating KCs from NPCs by performing positive selections using F4/80 beads with subsequent exposure to LPS.

Liver I/R.

A nonlethal model of segmental (70%) hepatic warm ischemia and reperfusion was used as previously described.5 TLR4loxP/loxP mice were used as WT control for all in vivo experiments. The c-Jun-N-terminal kinase (JNK) inhibitor (SP600125; 10 mg/kg; Calbiochem, San Diego, CA) and p38 inhibitor (SB203580; 10 mg/kg; Calbiochem) were administered intraperitoneally 1 hour before ischemia.

Isolation, Culture, and Treatment of HCs and NPCs.

HCs and NPCs were isolated and plated as previously described.7 For experiments involving hypoxia, the medium was replaced with media equilibrated with 1% O2, 5% CO2, and 94% N2 and placed into a modular incubator chamber (Billups-Rothenberg, Inc., Del Mar, CA), which was flushed with the same gas mixture. For experiments using JNK inhibitor (SP600125) or p38 inhibitor (SB203580), 25 μM were added to media 30 minutes before treatment with hypoxia.

Serum Alanine Aminotransferase, High-Mobility Box1, and Cytokine Quantification.

Serum alanine aminotransferase (sALT) levels were measured using the DRI-CHEM 4000 Chemistry Analyzer System (Heska, Des Moines, IA). High-mobility box 1 (HMGB1) was quantified using an ELISA kit (IBL International GmbH, Hamburg, Germany). Serum cytokine quantification was performed using the Cytometric Bead Array Mouse Inflammation Kit (BD Biosciences).


A western blotting assay was performed using whole cell lysates from either liver tissue or HCs, as previously described.13 Membranes were incubated overnight using the following antibodies (Abs): TLR4 (Imgenex Corp., San Diego, CA), HMGB1, and heme oxygenase 1 (HO-1; Abcam, Cambridge, MA); mouse monoclonal HMGB1 Ab and β-actin (Sigma-Aldrich); and phospho-p38, p38, phospho-c-Jun, c-Jun, phospho-JNK, JNK, extracellular signal-regulated kinase (ERK), phospho-ERK, p65, and phospho-p65 (Cell Signaling Technology, Inc., Danvers, MA).

Immunofluorescent and Immunohistochemistry Staining.

Immunofluorescent (IF) staining was performed using HMGB1 Ab (1:1,000; Abcam), as previously described.14 Immunohistochemistry (IHC) for neutrophil infiltration was accomplished using Anti-Neutrophil Ab [7/4] (Abcam).

Preparation and Delivery of Adenoviral Vectors.

An E1- and E3-deleted adenoviral vector carrying AdTLR4 and AdLacZ cDNA was constructed and utilized in vivo as previously described.15

SYBR Green Real-Time Reverse-Transcription Polymerase Chain Reaction.

SYBR green polymerase chain reaction (PCR) was performed as previously described using β-actin as endogenous control.14 Specific primers were as follows: IL-10, forward 5'-TACCTGGTAGAAGTGATGCC-3' and reverse 5'-CATCATGTATGCTTCTATGC-3', and HO-1, which is commercially available from Qiagen.

Statistical Analysis.

Results are expressed as either mean ± standard error of the mean (SEM) or mean ± standard deviation (SD). Group comparisons were performed using analysis of variance and Student t test. A probability value of P ≤ 0.05 was considered statistically significant.


Confirmation of TLR4 KO Mice.

To investigate the role of TLR4 on an individual cellular population, we generated HC-, myeloid-cell–, and DC-specific TLR4 KO (Alb-TLR4−/−, Lyz-TLR4−/−, and CD11c-TLR4−/−, respectively) mice using Cre-loxP technology. Mice with loxP sites flanking exon 2 of TLR4 were interbred with mice that had Cre recombinase linked to the desired promoter. WT mice used had loxP inserted without expression of Cre recombinase, and TLR4−/− mice were globally lacking the loxP flanked exon 2. Both WT and TLR4−/− mice were born healthy and fertile, without any grossly apparent phenotypic differences.

Verification of specificity of TLR4 KO in Alb-TLR4−/− mice was accomplished by isolating both HCs and NPCs as well as analyzing these cells for the presence of TLR4 mRNA transcription using reverse-transcriptase (RT)-PCR with primers specific for exon 2 of TLR4 (Fig. 1A). TLR4 was present in both HCs and NPCs of WT mice, whereas Alb-TLR4−/− mice had TLR4 expressed only in NPCs. Global TLR4−/− had no detectable TLR4 in either cell population. Western blotting analysis was performed to confirm that HCs isolated from Alb-TLR4−/− mice had TLR4 protein levels that were undetectable (Fig. 1B).

Figure 1.

Confirmation of specificity of TLR4 knockout (TLR4−/−). (A) RT-PCR was used to determine mRNA levels of TLR4 within either isolated HCs or NPCs from WT, Alb-TLR4−/−, and TLR4−/− mice. β-actin was used as endogenous control to determine relative expression, compared to WT. (B) TLR4 protein levels were assessed by western blotting analysis. (C) NPCs were isolated and stimulated with LPS for 6 hours, and the supernatant was analyzed by ELISA for either TNF-α or IL-6. (D) RT-PCR analysis of peritoneal macrophages to determine TLR4 mRNA expression in Lyz-TLR4−/−. (E) LPS response was determined in peritoneal macrophages from WT, Lyz-TLR4−/−, and TLR4−/− mice. Data represent mean ± SD. N = 2 samples per group run in duplicate. Figure is representative of two experiments with similar results.

It has previously been shown that the immune cells of the liver play a major role in I/R injury.7, 8, 11 Although the presence of TLR4 was noted in the NPC population of cells in Alb-TLR4−/− mice, we also wanted to confirm that the functional characteristics of TLR4 in these cells was unchanged. Therefore, we studied the response of isolated NPCs to LPS by determining the level of TNF-α and IL-6 in the supernatant using ELISA (Fig. 1C). There was no significant difference between NPCs of WT and Alb-TLR4−/− mice, whereas TLR4−/− NPCs failed to respond to LPS, as expected. Additional confirmation of the functional deletion of TLR4 in Alb-TLR4−/− mice was demonstrated by the lack of Alexa 488–labeled LPS uptake in Alb-TLR4−/− HCs similar to global TLR4−/− (data not shown). It was also noted that hepatic intracellular downstream signaling to TLR4 activation was concordant with the above-described, with p65 (nuclear factor kappa B; NF-κB) activation attenuated in response to LPS in both Alb-TLR4−/− and global TLR4−/− mice (Supporting Fig. 2A).

By linking Cre recombinase expression with the lyz promoter, mice were generated with TLR4−/− specific to the myeloid cell lineage, including KCs.16 We confirmed that peritoneal macrophages lack both TLR4 expression (Fig. 1D) in addition to functional response to LPS (Fig. 1E). Cre recombinase linked to the cd11c promoter was used to generate DC-specific TLR4−/− mice. This has previously been shown, by both others17 and also our lab (unpublished data), to be an effective method for developing DC-specific Tg mice. Additionally, we confirmed that KCs isolated from CD11c-TLR4−/− mice retained functional TLR4 (Supporting Fig. 2B,C).

HC and Myeloid Cell TLR4 Is Required for Maximal Hepatic I/R Injury.

Although our previous studies with TLR4 chimeric mice highlight the importance of BM-derived cells in mediating TLR4-dependent inflammation in response I/R,7 the role of TLR4 on individual cell types can now be investigated with the use of Tg mice. To better clarify whether TLR4 on HCs, myeloid cells, and DCs affects inflammatory response during I/R, Tg mice were subjected to hepatic I/R. In WT control mice, the sALT level was significantly greater than both Alb-TLR4−/− mice and Lyz-TLR4−/− mice; however, global TLR4−/− mice had significant protection, compared to both Alb-TLR4−/− and Lyz-TLR4−/− mice (Fig. 2A). Interestingly, CD11c-TLR4−/− mice had significantly increased hepatocellular injury, compared to WT (Fig. 2A). Degree of liver damage on histologic analysis was concordant with sALT results (Fig. 2B). Sham livers demonstrated no histologic evidence of hepatocellular injury (data not shown). These results demonstrate that lack of TLR4 on both HCs and myeloid cells results in protection from I/R, whereas TLR4 on DCs may have a protective role subsequent to liver I/R.

Figure 2.

Cellular-specific role of TLR4 on hepatocellular injury after I/R. (A) sALT levels were analyzed in WT, cell-specific, and global TLR4−/− mice after either sham laparotomy or 1 hour of ischemia and 6 hours of reperfusion. (B) Hematoxylin and eosin–stained liver sections obtained from mice 6 hours after reperfusion (original magnification: ×100). Images are representative liver sections from six mice per group. Necrotic area is outlined in black dashed line, and percent of area that is necrotic is indicated. (C) Serum cytokine analysis after I/R was performed. (D) Number of neutrophils infiltrating into liver tissue after I/R, as assessed by IHC of neutrophils per high-power field (×200). Data represent mean ± SEM; n ≥ 6 mice per group; *P < 0.05, when compared to WT; #P < 0.05 when compare to global TLR4−/−.

Additionally, serum levels of TNF-α, IL-6, and monocyte chemotactic protein 1 were quantified (Fig. 2C). Cytokines generally reflected sALT levels with intermediate cytokine decrease in cell-specific KO mice, compared to global TLR4−/− mice. Quantification of neutrophil infiltration was also determined (Fig. 2D). Interestingly, the number of neutrophils was significantly decreased in not only global TLR4−/−, but also in Alb-TLR4−/− mice. These results again demonstrate the importance of hepatocyte TLR4 in I/R inflammatory response.

Serum HMGB1 Release After Hepatic I/R Is Dependent on TLR4.

HMGB1 is an evolutionarily conserved protein present in the nucleus of almost all eukaryotic cells, where it functions to stabilize nucleosomes and acts as a transcription factor.18 HMGB1 is also rapidly mobilized and released in the setting of hepatic I/R to act as a key damage-associated molecular pattern (DAMP) molecule.5, 19 TLR4 and HMGB1 are intimately related, with TLR4 both functioning as a receptor for HMGB1 in addition to mediating its nucleocytoplasmic shuttling and subsequent release.7, 19 Thus, we sought to determine the role of cell-specific TLR4−/− in the release of HMGB1 after hepatic I/R. When serum HMGB1 levels after I/R were analyzed, Alb-TLR4−/− Tg mice had significantly lower serum HMGB1 levels, compared to WT (Fig. 3A). Lyz-TLR4−/− also had lower serum HMGB1 levels, but did not reach statistical significance (Fig. 3A). Alb-TLR4−/− and global TLR4−/− mice had HMGB1 levels that were similar and significantly lower than Lyz-TLR4−/− mice (Fig. 3A). On the other hand, CD11c-TLR4−/− mice did not have any significant difference in HMGB1 levels, compared to WT.

Figure 3.

Cellular-specific role of TLR4 on HMGB1 release after I/R. (A) Serum HMGB1 ELISA after 6 hours of reperfusion. *P < 0.05, when compared to WT. (B) Serum HMGB1 ELISA in WT, Alb-TLR4−/−, and TLR4−/− after either 3 or 6 hours of reperfusion. *P < 0.05, when compared to WT. (C) IF of liver tissue after 3 hours of reperfusion. Red, HMGB1; green, F-actin; blue, nuclei (original magnification: ×600). Data represent mean ± SEM; n ≥ 6 mice per group mice per group.

Because TLR4 on HCs appeared to be the main contributor to TLR4-mediated HMGB1 release after I/R, we next further investigated HMGB1 release in Alb-TLR4−/− and global TLR4−/− mice. These mice had decreased levels of circulating HMGB1 after both 3 and 6 hours of reperfusion, when compared to WT mice (Fig. 3B). IF staining of liver sections of these mice confirmed the role that TLR4 plays in the release of HMGB1 after I/R. Both Alb-TLR4−/− and global TLR4−/− mice livers had retained nuclear and decreased cytoplasmic HMGB1, when compared to WT mice (Fig. 3C). Our findings show that TLR4, on parenchymal cells, are the main contributors to circulating HMGB1 release during liver I/R.

IL-10 and HO-1 Expression Is Altered in Lyz-TLR4−/− and CD11c-TLR4−/− Mice.

It has been found previously that decreased expression of hepatoprotective factors HO-1 and IL-10 from KCs and decreased IL-10 from DCs resulted in increased I/R injury.20-22 Therefore, we investigated IL-10 and HO-1 expression in Lyz-TLR4−/− and CD11c-TLR4−/− mice. When compared to WT mice, Lyz-TLR4−/− mice had both IL-10 and HO-1 up-regulated after I/R, possibly leading to the protection noted in these mice (Fig. 4A,C). This expression pattern was confirmed at the protein level as well (Fig. 4B,D). Additionally, expression of IL-10 was decreased in CD11c-TLR4−/− mice after I/R, suggesting a mechanism for the increased hepatocellular injury noted in these mice (Fig. 4C,D). Alb-TLR4−/− did not show any notable differences in either IL-10 or HO-1 expression, when compared to WT (data not shown). Therefore, TLR4-mediated modulation of the expression of IL-10 and HO-1 in myeloid cells and IL-10 in DCs may contribute to the differences noted in hepatic damage.

Figure 4.

HO-1 and IL-10 mRNA and protein expression is altered in Lyz-TLR4−/− and CD11c-TLR4−/− mice after I/R. Real-time PCR was used to determine (A) HO-1 and (C) IL-10 expression in liver tissue after I/R in Lyz-TLR4−/− and CD11c-TLR4−/− mice. Protein quantification was determined using western blotting analysis for HO-1 (B) and cytometric bead array assay for IL-10 (D). β-actin was used as endogenous control to determine relative expression, compared to WT. *P < 0.05, when compared to WT. (A) and (C) are representative from triplicate experiments with pooled RNA from six animals per group. (B) shows two representative animals for each group.

HC TLR4 Mediates Inflammatory Signaling After I/R.

Because the Alb-TLR4−/− mice were significantly protected, we further investigated the mechanisms by which this was taking place. Among the most proximal inflammatory signaling events after I/R is the activation of mitogen-activated protein (MAP) kinases.23, 24 To determine whether HC TLR4 was involved in the activation of MAP kinase signaling, we performed western blotting analysis on liver lysates from WT, Alb-TLR4−/−, and global TLR4−/− mice after I/R. Phosphorylation of the MAP kinases, JNK and ERK, were substantially reduced at 3 hours of reperfusion in both Alb-TLR4−/− and global TLR4−/− mice, when compared to WT mice (Fig. 5). We found no role for HC TLR4 in p38 phosphorylation at this time point; however, p38 phosphorylation occurs very early after reperfusion.23 This may account for the lack of difference noted at 3 hours of reperfusion. Notably, we did not observe major differences in MAP kinase activation at either the 1-hour or 6-hour time points. To confirm that these findings were related to the local effects of I/R and not systemic inflammatory mediators, we demonstrated no increased phosphorylation of these proteins in the nonischemic lobes (Fig. 5). Therefore, HC TLR4 seems to be an important mediator of MAP kinase activation after I/R.

Figure 5.

MAP kinase activation is decreased in Alb-TLR4−/− mice after I/R. Western blotting of phosphorylated or total MAP kinases from protein isolated from either ischemic lobe or nonischemic lobe after 1 hour of ischemia and 3 hours of reperfusion. Figure is representative from three experiments of four animals, each with similar results.

Hypoxia-Induced JNK and p38 Activation in HCs Is Dependent on TLR4.

Our above-described experiments found that HC TLR4 was involved in the activation of JNK signaling in the liver after I/R. JNK is activated by exposure of cells to cytokines and environmental stress and has previously been demonstrated to be activated in HCs by both hypoxia and liver I/R.25, 26 Therefore, we exposed WT HCs to hypoxia and rapidly observed increased phosphorylation of JNK and p38, compared to normoxia (Fig. 6A). When TLR4−/− HCs were exposed to hypoxia, the phosphorylation of JNK, c-Jun (the downstream target of JNK), and p38 was substantially reduced, compared to WT HCs (Fig. 6B). We observed no increase in NF-κB (p65) or ERK phosphorylation with hypoxia exposure (Fig. 6B). To confirm that this response was, in fact, the result of the lack of functional TLR4 and not some other mechanism, HCs from TLR4−/− mice were then transfected with either a control adenoviral vector (AdLacZ) or recombinant adenovirus encoding TLR4 (AdTLR4). TLR4 expression using AdTLR4 was confirmed by western blotting (Fig. 6C). Transfection of TLR4−/− HCs with AdTLR4 restored JNK and p38 phosphorylation in response to hypoxia (Fig. 6D), indicating that this is, in fact, a TLR4-dependent response. Thus, these results demonstrate that HCs respond to hypoxic stress with a rapid activation of JNK and p38 in a TLR4-dependent manner.

Figure 6.

Hypoxia-induced JNK and p38 phosphorylation is TLR4 depedent. (A) Western blotting of p-JNK and p-p38 after WT HCs were exposed to hypoxia for the indicated time period. (B) Western blotting of phosphorylated and total MAP kinase levels during hypoxia with comparison between WT and TLR4−/−. (C) Western blotting analysis of TLR4 expression after TLR4−/− HCs were transfected with AdTLR4 or the control AdLacZ. (D) Western blotting analysis of p-JNK, and p-p38 after WT, TLR4−/−, KO+AdLacZ, or KO plus AdTLR4 HCs (multiplicity of infection [MOI]: 100) were exposed to hypoxia for 30 minutes. Figures are representative of three trials with identical results.

HMGB1 Release From HCs Under Hypoxic Stress Is JNK Dependent.

We next sought to determine whether the release of HMGB1 was mediated by JNK phosphorylation. Therefore, we added the JNK inhibitor (SP600125) to the media of HCs exposed to hypoxia. Phosphorylation of the target of JNK, c-Jun, was inhibited with the addition of the JNK inhibitor (Fig. 7A). With the addition of the JNK inhibitor and 24-hour exposure to hypoxia, there was a substantial reduction of the release of HMGB1 into the hepatocyte supernatant (Fig. 7B). Additionally, we determined that this difference was not the result of decreased hepatocyte death and passive HMGB1 release by determining supernatant levels of lactate dehydrogenase (LDH) and β-actin (Fig. 7B).

Figure 7.

HMGB1 release from HCs under hypoxic stress is JNK dependent. JNK inhibitor (SP600125; 25 μM) or negative control (NC) was added to media of HCs and were either exposed to hypoxia or normoxia for 30 minutes. (A) Western blotting analysis of JNK and c-Jun activation in the absence or presence of JNK inhibitor. (B) Western blotting of HC supernatant for HMGB1 after 24 hours of either normoxia or hypoxia. Western blotting of the supernatant was also performed for β-actin and LDH to estimate cellular death. (C) Western blotting for in vivo confirmation of JNK inhibitor efficacy after 1 hour of ischemia and 3 hours of reperfusion. Representative samples from two mice of each group are shown. (D) ELISA of serum HMGB1 levels after either negative control or JNK inhibitor were given before I/R. Data represent mean ± SEM; n ≥ 6 mice per group; *P < 0.05.

The effect of the JNK inhibitor on HMGB1 release in vivo after I/R was also investigated. Efficacy of the JNK inhibitor was first confirmed by decreased phosphorylation of c-Jun, compared to vehicle control, on western blotting analysis (Fig. 7C). With administration of the inhibitor given before I/R, there was a significant decrease in serum levels of HMGB1 after I/R (Fig. 7D). We, again, confirmed that this decrease in HMGB1 was not solely the result of decreased hepatocellular injury with JNK inhibition by determining that sALT levels were unchanged at 3 hours of reperfusion (Supporting Fig. 3), in addition to histologic analysis (data not shown).

The p38 inhibitor, SB203580, was also studied both in vitro and in vivo similar to the JNK inhibitor. With administration of the p38 inhibitor before hypoxia exposure in vitro and before I/R in vivo, there was no inhibitory effect noted on HMGB1 release (data not shown), suggesting that p38 does not play a major role in TLR4-mediated HMGB1 release. Therefore, it seems that activation of JNK, but not p38, is required for the extracellular release of HGMB1, both after hypoxic stress in vitro and I/R in vivo.


Hepatic I/R is dependent on the pattern recognition receptors (PRRs) to sense and initiate the sterile inflammatory response. Although the central role of the PRR, TLR4, in this process had been previously demonstrated,5, 6 the role of TLR4 on individual cell types, specifically, parenchymal versus NPC, during the sterile inflammatory response was conflicted. Therefore, in this study, we describe the novel use of Cre-loxP technology to knock out TLR4 in HCs, myeloid cells, and DCs and elucidate their individual role in I/R injury. The key and novel findings include the following: (1) Both HC and myeloid cell TLR4 is required for maximal I/R-associated injury; (2) DC TLR4−/− worsens injury after I/R and is associated with decreased IL-10 expression; (3) HCs are a major source of circulating HGMB1 after I/R; (4) HCs respond to hypoxia with increased phosphorylation of MAP kinases (JNK and p38) in a TLR4-dependent fashion; and (5) hypoxia-induced HMGB1 release from HCs is dependent on the function of JNK.

Previous work to define the function of TLR4 on individual cellular populations was limited to the use of chimeras. Although we have shown that there was not a significant difference in hepatic I/R-induced injury with lack of TLR4 on non-BM-derived cells, there was a trend toward an effect and others have subsequently shown that both BM and non-BM-derived populations have a role in mediating I/R injury.7, 11 A substantial weakness of the studies using chimeric mice is that there are many individual cell types affected in each experimental group. For example, with chimeric mice, it is impossible to differentiate the role of TLR4 on HCs versus endothelial cells (ECs) or myeloid cells versus DCs. The use of Cre-loxP technology to generate Tg mice has major advantages in helping to elucidate the precise role of receptors on individual cellular populations. Notably, Cre recombinase linked to lyz is highly expressed in all myeloid-derived cells, including KCs, neutrophils, and monocytes, but not within DCs.16 However, this model is not perfect, and deletion of TLR4 may occur within a small portion of CD11c+ DCs in these mice, though our functional studies suggest that this spillover is negligible. Additionally, whereas the albumin promoter is active in immature cells that can differentiate into either HCs or cholangiocytes, only the HCs continue to express albumin.27-29 Therefore, it may be possible that some cholangiocytes have some deletion of TLR4, but this is likely negligible because it has been shown to take 6 weeks for maximal Alb-Cre-mediated recombination to take place.30 Although other methods exist for targeting HCs specifically, such as the AAV8-Ttr-Cre model,28 this is not useful against the other cell types considered here. Therefore, although this technology is not perfect, it is useful here in that it allows for meaningful comparison between parenchymal and nonparenchymal cell-specific knockouts. Our characterization, along with the previous reports, have demonstrated that Cre expression linked to alb, lyz, and cd11c promoter is an efficient, specific way of developing cellular-specific knockouts.16, 17, 31

Hepatic DCs are thought to primarily be anti-inflammatory. Consistent with this, Loi et al. have previously shown that although hepatic I/R leads to DC maturation, they preferentially produce inhibitory cytokines IL-10 and transforming growth factor beta.32 Interestingly, our results indicate that DC TLR4 plays a protective role with the lack of functional TLR4 in DCs associated with a decrease in IL-10 expression and worsening of hepatocellular injury. Our results mirror the TLR9 results of Bamboat et al.,22 where TLR9 activation by HC DNA led to the production of IL-10 and hepatoprotection from I/R, leading us to hypothesize that DC TLR9 and TLR4 function similarly after I/R, possibly in a redundant fashion. KCs, on the other hand, have traditionally been thought to be a major mediator of I/R-associated injury.1 Our results confirm this finding and further demonstrate this effect to be dependent on TLR4 expression in these cell types. However, other studies in addition to our unpublished data using liposomal clodronate for KC depletion show a decrease in IL-10 and HO-1 expression and increase in hepatocellular injury after I/R, suggesting that KCs may also provide a protective role, in addition to the proinflammatory role driven by TLR4.20, 21 Additionally, the role of TLR4 ligands in the protective response of KCs has been suggested by Ellet et al. using a model of total hepatic I/R with bowel congestion,20, 33 a model with direct TLR4 activation through LPS release. In TLR4−/− mice, KC depletion leads to increased hepatocellular injury and decreased IL-10 response.33

We have shown previously that TLR4 signaling is necessary for hepatic I/R response, and that this response is, in part, mediated by HMGB1.5, 19 Here, we demonstrate the role HCs play in this response, with release of HMGB1 significantly reduced with lack of functional HC TLR4 signaling, approximately equal to mice with global TLR4 deficiency. Furthermore, there was an intermediate decrease in serum HMGB1 with lack of TLR4 in myeloid cells, even though the hepatocellular injury was not significantly different in these mice. These findings suggest that HCs are the primary cell type responsible for TLR4-dependent HMGB1 release after I/R, which is a novel finding and contrary to the current thought that HMGB1 release is primarily dependent on immune cells.34 However, it certainly seems plausible that HCs may be the primary producer of HMGB1 early in I/R, because, in our previous work, we have shown that HCs can actively release HMGB1 in response to oxidative stress in a regulated process.15, 19, 35

There are a number of cellular pathways involved with hypoxia-induced HMGB1 release by HCs, all of which are actively regulated.15, 19, 35, 36 The hyperacetylation of HMGB1, which is largely regulated by histone deacetylases, leads to the shuttling of nuclear HMGB1 into the cytoplasm.35, 36 Additionally, HMGB1 translocation and subsequent extracellular release is also dependent on calcium/calmodulin-dependent kinases and also on functional interferon regulatory factor 1 (IRF-1).15, 19 JNK has recently been shown to be able to induce expression of IRF-1,37 substantiating our hypothesis that JNK is upstream of other known pathways leading to HMGB1 release. Although JNK inhibition has been shown to be protective in I/R, these effects are noted at time points >6 hours, despite JNK activation occurring much earlier. Therefore, we hypothesized that JNK activation may be responsible for the release of a proinflammatory mediator, rather than directly contributing to injury. Here, we provide evidence that the role of JNK includes the facilitation of HMGB1 release from hepatocytes both in vitro and in vivo, thus providing one possible solution.

In summary, we use cellular-specific TLR4−/− Tg mice to establish that TLR4 may either worsen or alleviate hepatocellular injury after I/R, depending on cell type. The role of TLR4 on both myeloid and HCs is primarily proinflammatory, compared to DCs, in which TLR4 plays a primarily anti-inflammatory role (Fig. 8). We are intrigued that parenchymal cells, such as HCs, are not mere passive recipients of injury during I/R, but active participants in the sterile inflammatory process. In addition to the cell types studied here, there are certainly other cellular populations that have influential roles in sterile inflammation, such as ECs, and we are currently investigating these. The novel findings of this study help to dissect out the role of the complex cellular interactions occurring in I/R, with the potential to develop therapeutic interventions to abrogate the sterile inflammatory response.

Figure 8.

Schematic summary highlighting the role TLR4 plays on individual cell populations after hepatic I/R. MPT, mitochondrial permeability transition; ROS, reactive oxygen species.


The authors thank Nicole Hays and Junda Chen for their technical assistance in preparing this manuscript.