Hepatocyte-specific high-mobility group box 1 deletion worsens the injury in liver ischemia/reperfusion: A role for intracellular high-mobility group box 1 in cellular protection

Authors


  • Potential conflict of interest: Nothing to report.

  • This work was supported by a Howard Hughes Medical Institute Physician-Scientist Award (to A.T.), an Association of Academic Surgery Foundation Research Fellowship Award (to G.N.), NIH Grants R01-GM95566 (to A.T.), and R01-GM50441 (to T.B.).

Abstract

High-mobility group box 1 (HMGB1) is an abundant chromatin-associated nuclear protein and released into the extracellular milieu during liver ischemia-reperfusion (I/R), signaling activation of proinflammatory cascades. Because the intracellular function of HMGB1 during sterile inflammation of I/R is currently unknown, we sought to determine the role of intracellular HMGB1 in hepatocytes after liver I/R. When hepatocyte-specific HMGB1 knockout (HMGB1-HC-KO) and control mice were subjected to a nonlethal warm liver I/R, it was found that HMGB1-HC-KO mice had significantly greater hepatocellular injury after I/R, compared to control mice. Additionally, there was significantly greater DNA damage and decreased chromatin accessibility to repair with lack of HMGB1. Furthermore, lack of hepatocyte HMGB1 led to excessive poly(ADP-ribose)polymerase 1 activation, exhausting nicotinamide adenine dinucleotide and adenosine triphosphate stores, exacerbating mitochondrial instability and damage, and, consequently, leading to increased cell death. We found that this was also associated with significantly more oxidative stress (OS) in HMGB1-HC-KO mice, compared to control. Increased nuclear instability led to a resultant increase in the release of histones with subsequently more inflammatory cytokine production and organ damage through activation of Toll-like receptor 9. Conclusion: The lack of HMGB1 within hepatocytes leads to increased susceptibility to cellular death after OS conditions. (Hepatology 2014;59:1984–1997)

Abbreviations
Abs

antibodies

ALT

alanine aminotransferase

APC

allophycocyanin

ATP

adenosine triphosphate

BER

base excision repair

Cy7

cyanine 7

DAMP

damage-associated molecular pattern

DCF-DA

dihydrodichlorofluorescein diacetate

ERK

extracellular signal-related kinase

FACS

fluorescence-activated cell sorting

FCM

flow cytometry

FITC

fluorescein isothiocyanate

H&E

hematoxylin and eosin

HMGB1

high-mobility group box 1

4-HNE

4-hydroxy-2-nonenal

IF

immunofluorescence

IL

interleukin

I/R

ischemia-reperfusion

JNK

c-Jun N-terminal kinase

KO

knockout

LDH

lactate dehydrogenase

LPO

lipid peroxidation

MAPK

mitogen-activated protein kinase

MMP

mitochondrial membrane potential

mRNA

messenger RNA

NAD+

nicotinamide adenine dinucleotide

NER

nucleotide excision repair

NIH

National Institutes of Health

NF-κB

nuclear factor kappa B

NK

natural killer

NPCs

nonparenchymal cells

OS

oxidative stress

PAR

poly(ADP-ribose)

PARP-1

poly(ADP-ribose)polymerase 1

PBS

phosphate-buffered saline

PE

phycoerythrin

PRR

pattern recognition receptor

ROS

reactive oxygen species

RT-PCR

reverse-transcriptase polymerase chain reaction

TEM

transmission electron microscopy

Tg

transgenic

TLR

Toll-like receptor

TNF-α

tumor necrosis factor alpha

UV

ultraviolet

WT

wild type

Ischemia-reperfusion (I/R) injury is a process whereby an initial hypoxic insult and subsequent return of blood flow leads to the propagation of an innate immune response and ensuing organ injury. Although the liver may initially exhibit direct cellular damage as the result of ischemic insult, reperfusion further propagates dysfunction and damage resulting from activation of inflammatory pathways.[1] Central to the propagation of this sterile inflammatory response is the recognition of damage-associated molecular pattern molecules (DAMPs) by pattern recognition receptors (PRRs). Several DAMPs, including high-mobility group box 1 (HMGB1), histones,[2] and DNA as well as the PRRs, Toll-like receptor (TLR)4 and TLR9, have been shown to be involved in the damage and inflammation induced by warm liver I/R. We have shown previously that TLR4 signaling is crucial for hepatic I/R response, and that this response is mediated by HMGB1.3,4

HMGB1 is an evolutionarily conserved protein present in the nucleus of almost all eukaryotic cells.[5] The function of HMGB1 is diverse and compartment specific. Outside the cell, HMGB1 exhibits a variety of activities that are dependent on the redox status of the protein. When mobilized to the cytoplasm, HMGB1 can regulate autophagic flux.[6] Additionally, within the nucleus, HMGB1 facilitates gene transcription and DNA repair response.[7] HMGB1 has been identified as possibly playing a role in all four major DNA repair pathways: nucleotide excision repair (NER); mismatch repair; base excision repair (BER); and DNA double-strand break repair.[8] Furthermore, we have previously shown that HMGB1 is rapidly mobilized to the cytoplasm and the extracellular space in hepatocytes after ischemia in vivo or hypoxia in vitro.[9] The function of HMGB1 in specific cell types during injury response has been difficult to fully elucidate secondary to the fact that mice lacking functional HMGB1 die shortly after birth. Therefore, to further investigate the role of HMGB1 within hepatic I/R response, we used novel transgenic (Tg) cell-specific knockout (KO) mice, generated using Cre-loxP technology, in which hepatocytes are deficient in HMGB1. KO of HMGB1 specifically within hepatocytes is ideally suited to further investigate the role of HMGB1 in hepatic I/R because we have recently shown that not only do hepatocytes a play a key role in the inflammatory response associated with I/R, but also that hepatocytes are a major cell type responsible for TLR4-dependent HMGB1 release after I/R.[10]

In this study, we found that deletion of HMGB1 from hepatocytes resulted in greater liver injury after I/R. This is in striking contrast to the proinflammatory role of extracellular HMGB1; these studies reveal a dominant role for intracellular HMGB1 in stabilizing nuclear response to oxidative stress (OS).

Materials and Methods

Animals

Male wild-type (WT; HMGB1loxP/loxP) mice and hepatocyte-specific HMGB1−/− 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, and experiments were performed in strict adherence to the National Institutes of Health (NIH; Bethesda, MD) Guidelines for the Use of Laboratory Animals.

Generation of HMGB1loxP/loxP and Hepatocyte-Specific HMGB1/ Mice

In brief, the HMGB1loxP allele was created by inserting loxP sites within intron 1 and intron 2 flanking exon 2 of HMGB1. An overview of this construct is shown in the Supporting Information (Supporting Fig. 1A). Mice homozygous for HMGB1loxP were generated by Ozgene (Bentley, WA). HMGB1loxP/loxP mice were interbred with stud males (HMGB1loxP/−; Alb-cre) to generate the desired genotype. Mice homozygous for Cre recombinase linked to the albumin (alb) promoter are commercially available from The Jackson Laboratory (Bar Harbor, ME). Tg male mice used for experiments were confirmed to be the desired genotype by standard genotyping techniques. Control mice used in this study were HMGB1loxP/loxP mice without the introduction of Cre recombinase.

Figure 1.

Confirmation of specificity of HMGB1 KO (HMGB1-HC-KO). (A) RT-PCR was used to determine mRNA levels of HMGB1 within isolated hepatocytes from control and HMGB1-HC-KO mice. (B) HMGB1 protein levels within isolated hepatocytes or NPCs from control and HMGB1-HC-KO mice were assessed by western blotting analysis. Figure is representative of three experiments with similar results. (C) IF stain of HMGB1 within cultured hepatocytes from control and HMGB1-HC-KO mice (magnification, 400×). Images are representative of three experiments with similar results. Green, HMGB1; blue, nuclei; red, F-actin. (D) Serum HMGB1 ELISA after 1 or 6 hours of reperfusion. *P < 0.05, when compared against control. (E) IF stain of HMGB1 from sections of normal liver and liver 6 hours after I/R in control and HMGB1-HC-KO mice (magnification, 400×). Images are representative liver sections from 6 mice per group. Red, HMGB1; blue, nuclei; green, F-actin.

Confirmation and Characterization of Tg HMGB1/ Mice

Isolation and determination of HMGB1 messenger RNA (mRNA) expression was performed as previously described[10] using specific primers as follows: forward 5′-CACAGCCATTGCA GTACATTGA-3′ and reverse 5′-TGCTTGTCATC TGCTGCAGTGT-3′ for HMGB1 and β-actin primers as described previously.[4] University of Pittsburgh Genomics and Proteomics Core Laboratories performed mRNA microarray expression analysis using the Illumina bead-array platform with Mouse Refseq[8] BeadChip. Unstimulated hepatocytes were isolated from either KO or control mice, allowed to stabilized overnight in culture, and mRNA was isolated using the RNeasy Mini Kit (Qiagen, Valencia, CA). The experiment was performed using 2 mice from each strain with duplicate samples of each, and expression was normalized using the cubic spline method.

Liver I/R

A nonlethal model of segmental (70%) hepatic warm I/R was used as previously described.[11] Mice received the TLR9 antagonist (ODN2088; 100 µg per mouse; InvivoGen, San Diego, CA) or poly(ADP-ribose)polymerase 1 (PARP-1) inhibitor (3-AB, 20 mg/kg; PJ-34, 10 mg/kg; Sigma-Aldrich, St. Louis, MO) or respective control intraperitoneally 1 hour before ischemia. Sham animals underwent anesthesia, laparotomy, and exposure of the portal triad without hepatic ischemia.

Isolation, Culture, and Treatment of Hepatocytes and Nonparenchymal Cells

Hepatocytes and nonparenchymal cells (NPCs) were isolated, plated, and exposed to hypoxia, as previously described.[11] Supernatants from hypoxic hepatocytes were harvested after a 12-hour hypoxic period and were used as conditioned media in subsequent coculture assays.

Liver Damage Assessment

Serum alanine aminotransferase (ALT) levels, serum aspartate transaminase, and baseline serum chemistry values were measured using the DRI-CHEM 4000 Chemistry Analyzer System (Heska, Des Moines, IA). Extent of parenchymal necrosis in ischemic lobes was evaluated using hematoxylin and eosin (H&E)-stained histological sections at 40× magnification.[12] The necrotic area was quantitatively assessed by using ImageJ software (NIH). Results are presented as the mean of percentage of necrotic area (mm2) with respect to the entire area of one capture (mm2).

Enzyme-Linked Immunosorbent Aassay

Serum tumor necrosis factor alpha (TNF-α) and interleukin (IL)−6 levels in mice were detected by an enzyme-linked immunosorbent assay (ELISA) kit (R&D Systems, Minneapolis, MN). HMGB1 was quantified using an ELISA kit (IBL International Corp., Toronto, Ontario, Canada). Serum histone quantification was performed using the Cell Death Kit (Roche Diagnostics, Indianapolis, IN).

Immunoblotting

Western blotting assay was performed using whole-cell lysates from either liver tissue or hepatocytes, or media, as previously described.[4] Membranes were incubated overnight using the following antibodies (Abs): HMGB1 (Abcam) and β-actin (Sigma-Aldrich); phospho-p38, p38, phospho-JNK (c-Jun N-terminal kinase), JNK, extracellular signal-related kinase (ERK), phospho-ERK, p65, phospho-p65, PARP-1, acetyl-histone H3, acetyl-histone H4, and phospho-histone H2A.X (Cell Signaling Technology, Danvers, MA); and poly(ADP-pibose) (PAR; BD Biosciences, San Jose, CA).

Immunofluorescent Staining

For immunofluorescence (IF) staining, liver sections were fixed, stained, and imaged using confocal microscopy, as previously described.[13] Level of lipid peroxidation (LPO) was determined by measuring a major aldehyde product of lipid oxidation, 4-hydroxy-2-nonenal (4-HNE) Michael adducts, with HNE Ab (1:200; Calbiochem, Albuquerque, NM). Liver tissue or hepatocytes were incubated with the specific primary Abs for HMGB1 (1:1,000; Abcam, Cambridge, MA) and histone H3 (1:500; Abcam), as previously described.[2]

Quantitation of Confocal IF

All images were quantitated for 4-HNE or histone H3 using MetaMorph software (Molecular Devices, Downingtown, PA), as previously described.[14]

SYBR Green Quantitative Reverse-Transcriptase Polymerase Chain Reaction

Total RNA was extracted from liver tissue or NPCs using the RNeasy Mini Kit (Qiagen). mRNA for TNF-α, IL-6, intracellular adhesion molecule 1, monocyte chemoattractant protein 1, C-X-C motif chemokine 10, IL-1β, IL-12p35, interferon-beta, and β-actin was quantified in triplicate by SYBR Green quantitative reverse-transcriptase polymerase chain reaction (RT-PCR). PCR reaction mixture was prepared using SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA), using above-described primers.[11]

Detection of Cellular Reactive Oxygen Species Production

After hypoxia, hepatocytes were stained by using a dihydrodichlorofluorescein diacetate (DCF-DA) cellular reactive oxygen species (ROS) detection assay kit (Abcam), according to the manufacturer's instructions, then read in a fluorescence spectrophotometer (SpectraMAX Gemini XS; MDS Analytical Technologies Inc., Sunnyvale, CA), as described previously.[3]

Adenosine Triphosphate, Nicotinamide Adenine Dinucleotide, and Lactate Dehydrogenase Quantification

Quantification of nicotinamide adenine dinucleotide (NAD+), adenosine triphosphate (ATP) and lactate dehydrogenase (LDH) levels from hepatocytes after hypoxia was performed by using a NAD/NADH assay kit, a ATP assay kit, and an LDH assay kit (Abcam).

Transmission Electron Microscopy

At the end of 6-hour reperfusion, liver tissue was harvested after perfusion with phosphate-buffered saline (PBS), then fixed and processed for transmission electron microscopy (TEM), as described previously.[15] After dehydration, thin sections (70 nm) were stained with uranyl acetate and lead citrate for observation under a JEM 1011CX electron microscope (JEOL, Peabody, MA). Images were randomly selected from a pool of 18 fields from within five grid squares under each condition.

Flow Cytometry Analysis

Ischemic liver lobes were aseptically harvested from HMGB1-HC-KO and control mice after 1 hour of ischemia and 6 hours of reperfusion and prepared as a single-cell suspension. Samples were prepared and analyzed with flow cytometry (FCM) for innate immune cell populations using the following Abs, as previously described16: neutrophils (CD11b+Ly6G+); inflammatory monocytes (CD11b+Ly6Chi); and natural killer (NK) cells (NK1.1+CD11C+). Abs were purchased from eBioscience (San Diego, CA): phycoerythrin (PE) anti-NK1.1 PK136 and PE-Cy7 (cyanine 7) anti-CD11b M1/70; BD Biosciences: allophycocyanin (APC) anti-CD11c HL3 and FITC anti-Ly6C AL-21; or BioLegend (San Diego, CA): fluorescein isothiocyanate (FITC) anti-I-Ab AF6-120.1, APC anti-CD11b M1/70, and PE-Cy7 anti-Ly6G 1A8.

For measurement of mitochondria-associated ROS production, hepatocytes were stained with MitoSOX (Molecular Probes/Invitrogen, Carlsbad, CA), following the manufacturer's protocol for fluorescence-activated cell sorting (FACS) analysis. For measurement of mitochondrial polarization, a tetramethylrhodamine ethyl ester/mitochondrial membrane potential (MMP) assay kit (Abcam) was used, according to the manufacturer's instructions. For measurement of damaged mitochondria, hepatocytes were stained with MitoTracker Green and MitoTracker Deep Red for FACS analysis.

Data were acquired with a BD FACS LSR Fortessa flow cytometer (BD Biosciences) and analyzed with FlowJo analytical software (TreeStar, Ashland, OR). Each experiment was repeated a minimum of three times.

Statistical Analysis

Results are expressed as either standard error of the mean or mean standard deviation. Group comparisons were performed using analysis of variance and the Student t test. A P < 0.05 was considered statistically significant.

Results

Confirmation of HMGB1 KO Mice

To investigate the roles of HMGB1 in hepatocytes, we generated HMGB1-HC-KO mice using Cre-loxP technology. Both control and HMGB1-HC-KO mice were born healthy and fertile, without any grossly apparent phenotypic differences or abnormalities in liver function tests (Supporting Fig. 1B). Additionally, baseline differences in mRNA expression in unstimulated hepatocytes were determined using microarray analysis. Surprisingly, a paucity of significant differences was found in hepatocyte mRNA expression of HMGB1-HC-KO mice, compared against control, at baseline (Supporting Fig. 1C).

Verification of specificity of HMGB1 KO in HMGB1-HC-KO mice was demonstrated by isolating hepatocytes and analyzing these cells for the presence of HMGB1 mRNA expression using RT-PCR with primers specific for exon 2 of HMGB1 (Fig. 1A). HMGB1 was confirmed by western blotting that was present in both hepatocytes and NPCs of control mice, whereas HMGB1-HC-KO mice had HMGB1 expressed only in NPCs (Fig. 1B). IF staining in hepatocytes isolated from HMGB1-HC-KO mice had undetectable HMGB1, compared to HMGB1-positive hepatocytes from control mice (Fig. 1C).

Serum HMGB1 Release After Hepatic I/R Is Dependent on Hepatocyte HMGB1

HMGB1 is rapidly mobilized and released in the setting of hepatic I/R and, when released into the extracellular space, acts as a DAMP.[3, 4] Because hepatocytes are a major source of HMGB1 after hepatic I/R,[10] we anticipated that there would be substantially less circulating HMGB1 after I/R in KO mice and indeed HMGB1-HC-KO mice had significantly lower serum and intrahepatic HMGB1 levels, compared to control mice (Fig. 1D and Supporting Fig. 2A). Consistent with our previous results,[4] HMGB1 localized to the nucleus of hepatocytes in control sham-treated animals. After I/R, HMGB1-positive staining was observed in the cytoplasm of hepatocytes. However, no HMGB1-positive hepatocytes were found in sham- or I/R-treated HMGB1-HC-KO mice, again confirming that these mice do not express HMGB1 even after the stress of I/R (Fig. 1E).

Figure 2.

Cellular specific role of HMGB1 on hepatocellular injury after I/R. (A) Serum ALT levels were analyzed in control and HMGB1-HC-KO mice after either sham laparotomy or 1 hour of ischemia and 1 or 6 hours of reperfusion. Data represent the mean ± standard error (n = 6 mice per group). *P < 0.05 versus HMGB1 control. (B) Quantification of necrotic hepatocytes in H&E-stained liver sections (Supporting Fig. 2B) from HMGB1-HC-KO and control mice 6 hours after reperfusion. Graph is representative of liver sections from 6 mice per group. *P < 0.05 versus control. (C) Serum levels of TNF-α and IL-6 obtained from HMGB1-HC-KO and control mice at 6 hours after reperfusion were measured by ELISA and compared to the sham group. Data represent the mean ± standard error (n = 6 mice per group). *P < 0.05 versus control.

Genetic Deletion of HMGB1 in Hepatocytes Exacerbates Hepatic I/R Injury

To determine the effect of hepatocyte HMGB1 deletion on injury and inflammatory response induced by I/R, HMGB1-HC-KO mice were subjected to hepatic I/R. In HMGB1-HC-KO mice, the serum ALT levels were significantly greater than control mice (Fig. 2A). Degree of liver damage on histologic analysis was concordant with serum ALT results (Fig. 2B and Supporting Fig. 2B). Additionally, serum levels of TNF-α and IL-6 were significantly higher in HMGB1-HC-KO mice, compared to the control mice (Fig. 2C). These results demonstrate that lack of HMGB1 in hepatocytes results in exacerbation of I/R injury and an exaggerated inflammatory response, a surprising finding given the defined role of HMGB1 as a proinflammatory DAMP.

Deletion of HMGB1 From Hepatocytes Results in Enhanced Inflammatory Signaling After I/R

To further investigate how lack of HMGB1 in hepatocytes might mediate the inflammatory response to hepatic I/R injury, we examined the mitogen-activated protein kinases (MAPKs)- and nuclear factor kappa B (NF-κB)-signaling pathways. After 1 hour of hepatic I/R, phosphorylation of JNK, p38, ERK and NF-κB (p65 subunit) increased, compared to sham-treated mice. When HMGB1-HC-KO mice were subjected to I/R, there was a much greater increase in phosphorylation of all three MAPKs and NF-κB, compared to control mice (Fig. 3A), suggesting that these mice have an increased inflammatory response after I/R.

Figure 3.

Lack of HMGB1 in hepatocytes mediates inflammatory signaling and regulates innate immune cells in liver I/R. (A) MAPK activation and phosphorylation at serine 536 of the p65 subunit of NF-κB were determined by western blotting and quantitative densitometry analysis of protein expressions in sham-treated mice and mice that underwent ischemia and 1 hour of reperfusion. Hepatic protein lysates from ischemic lobes were obtained; each lane represents a separate animal. Blots shown are representative of three experiments with similar results. *P < 0.05 versus control. (B) FCM analysis with a quantitative evaluation of NPCs in homogenized ischemia liver lobes in HMGB1-HC-KO and control mice. Fold change of cell numbers of neutrophils, inflammatory monocytes cells, and NK cells. Data represent the mean ± standard error (n = 6 mice per group). *P < 0.05, compared to sham mice. Each experiment was repeated a minimum of three times.

The role of hepatocyte HMGB1 in modulating recruitment of innate immune cells in the liver after I/R was determined by FCM. As expected, liver I/R injury in control mice led to significantly greater recruitment of neutrophils, inflammatory monocytes, and NK cells, compared to sham control mice (Fig. 3B and Supporting Fig. 3). However, ablation of HMGB1 in hepatocytes led to even greater recruitment of innate immune cells after liver I/R, compared to control. Additionally, quantification of chemokine expression revealed that HMGB1-HC-KO mice had increased expression of chemokines after I/R (Supplemental Fig.4). These data demonstrate that elimination of HMGB1 in hepatocytes up-regulates innate immune response by increasing chemokine expression and subsequent influx of inflammatory cells in ischemic lobes after liver I/R.

Figure 4.

Lack of HMGB1 in hepatocytes leads to less DNA repair and more DNA damage, with, subsequently, more cell death and histone release. (A) Acetylation of histone H3 and H4 as well as phosphorylation of H2A.X were determined by western blotting and quantitative densitometry analysis in sham-treated mice and mice that underwent ischemia and 6 hours of reperfusion. Hepatic protein lysates from ischemic lobes were obtained; each lane represents a separate animal. Blots shown are representative of three experiments with similar results. *P < 0.05 versus control. (B) Serum histone levels were assessed after I/R using ELISA. Data represent the mean ± standard error (n = 6 mice per group). *P < 0.05, compared to control. (C) Translocation of histone H3 in cultured hepatocytes from HMGB1-HC-KO or control mice that were stimulated with either hypoxia or normoxic for 12 hours was visualized and observed under a confocal microscope (magnification, 400×). Green, actin; blue, nuclei; red, histone H3. Quantitation of nucleic or cytoplasmic histone H3 in cultured hepatocytes from HMGB1-HC-KO or control mice was measured using the analytical software, MetaMorph and normalized to nuclei within the sample field. *P < 0.05, hypoxia versus normoxia control; **P < 0.05, HMGB1-HC-KO versus control. (D) Cultured hepatocytes from HMGB1-HC-KO or control mice were exposed to hypoxia. Media were subjected to western blotting and quantitative densitometry analysis of histone H1, H2, H3, and H4. Blots shown are representative of three experiments with similar results. *P < 0.05 versus control.

Loss of HMGB1 in Hepatocytes Leads to Nuclear Instability With Increased DNA Damage and Histone Release

Knowing that HMGB1 has an important role in DNA repair,[8, 17] we next sought to determine whether lack of HMGB1 in hepatocytes leads to increased oxidative DNA damage after the oxidative stress (OS) of I/R. Histone acetylation has previously been shown to be a sensitive marker for chromatin accessibility, with acetylated histones allowing for DNA access to repair enzymes.[18] Acetylation of both histone H3 and H4 was markedly increased in liver I/R-treated control mice, compared with sham control (Fig. 4A). However, acetylation of histone H3 and H4 was significantly lower after liver I/R injury in HMGB1-HC-KO mice, suggesting that there is decreased accessibility of chromatin for repair without HMGB1. In addition, histone H2A.X has been shown to be phosphorylated in response to DNA damage, serving as a marker of DNA damage.[18] We found that loss of HMGB1 within hepatocytes resulted in a higher level of H2A.X phosphorylation after liver I/R injury, compared to control mice (Fig. 4A).

Consequential to our findings of increased DNA damage after OS in livers of HMGB1-HC-KO mice, we investigated whether HMGB1 deletion resulted in enhanced release of nucleosome components, such as histones. Histones are released from damaged hepatocytes both in vitro after hypoxia and in vivo after liver I/R injury and function as DAMPs to propagate injury.[2] Although serum histone levels increased in both HMGB1-HC-KO and control mice after liver I/R, levels in HMGB1-HC-KO mice were significantly higher, compared to control mice, at 6 hours after I/R injury (Fig. 4B). Additionally, using IF, it was shown that histone H3 translocation from nucleus to cytoplasm occurred in both HMGB1 KO and control hepatocytes after hypoxia, but significantly more histone-positive staining was observed in the cytoplasm of HMGB1 KO hepatocytes (Fig. 4C). Concordant results were observed in media of hepatocytes under hypoxia by using western blotting analysis (Fig. 4D). Thus, we conclude that loss of HMGB1 results in increased DNA damage and release of nucleosome components, leading to greater injury after OS.

PARP-1 Overactivation in Cells Lacking HMGB1 Leads to Mitochondria Damage by Exhausting NAD+ and ATP Stores

Our above-described results suggest that the lack of HMGB1 in hepatocytes leads to excessive nuclear damage after OS. Upon DNA damage, the nuclear enzyme, PARP-1, consumes NAD+ to form branched polymers of ADP-ribose (PAR) on target proteins, which include histones and PARP-1 itself. Poly(ADP-ribosylation) facilitates DNA repair by preventing chromatid exchange and by loosening histone wrapping.[19] However, PARP-1 overactivation leads to cell death, presumably through mechanisms linked to NAD+ depletion.[20] Therefore, we sought to determine whether HMGB1 deletion in hepatocytes undergoing OS would result in excessive activation of PARP-1. We found that both PAR formation and PAR-PARP-1 were increased in livers of both HMGB1-HC-KO and control mice after I/R injury. However, PARP-1 activity in livers of HMGB1-HC-KO mice was significantly higher than in control mice after I/R (Fig. 5A), suggesting that lack of HMGB1 leads to increased activation of PARP-1.

Figure 5.

Lack of HMGB1 in hepatocytes induces PARP-1 overactivation and damages mitochondria by exhausting NAD+ and ATP. (A) Total PAR formation and poly(ADP-ribosylation) of PARP-1 was determined by western blotting and quantitative densitometry analysis of protein expressions in sham-treated mice and mice that underwent ischemia and 6 hours of reperfusion. Hepatic protein lysates from ischemic lobes were obtained; each lane represents a separate animal. Blots shown are representative of three experiments with similar results. *P < 0.05 versus control. (B) Time course of NAD+ level in cultured hepatocytes from HMGB1-HC-KO or control mice during hypoxia (n = 4-6 for each point). *P < 0.05 versus control. (C) Time course of ATP level in cultured hepatocytes from HMGB1-HC-KO or control mice during hypoxia (n = 4-6 for each point). *P < 0.05 versus control. (D) Time course of mitochondrial potential in cultured hepatocytes from HMGB1-HC-KO or control mice during hypoxia (n = 4-6 for each point). *P < 0.05 versus control. (E) HMGB1 KO or control hepatocytes were cultures under normoxia or 12-hour hypoxia. Cells were then stained with 200 nM of MitoTracker Green and MitoTracker Deep Red (Invitrogen, Carlsbad, CA) for 45 minutes at 37°C, then analyzed using FCM for mitochondrial damage. *P < 0.05 versus control. (F) Mitochondrial ultrastructure from livers of HMGB1-HC-KO or control mice after sham or 1-hour reperfusion were imaged by a transmission electron microscope (magnification, 50,000×). *P < 0.05; results are representative of three separate independent experiments. White arrows point to damaged mitochondrial cristae.

We next sought to confirm that differences in PARP-1 activation led to altered NAD+ and ATP levels in hepatocytes under hypoxic OS. In fact, there was a significantly greater reduction of NAD+ and ATP levels as early as 1 hour during hypoxia in HMGB1 KO hepatocytes, compared to control hepatocytes (Fig. 5B,C). Exhaustion of NAD+ and ATP stores by hyperactivation of PARP-1 has previously been shown to lead to mitochondrial-mediated cell death.[21] Consequently, we next explored whether this reduced energy preservation in HMGB1 KO hepatocytes would alter mitochondrial function under OS. We demonstrated that a significantly greater loss of MMP was found in HMGB1 KO hepatocytes, compared to control hepatocytes, after hypoxia (Fig. 5D). Additionally, HMGB1 KO hepatocytes had a significantly higher percentage of damaged mitochondria when exposed to hypoxia (Fig. 5E). This increased mitochondrial damage was further demonstrated morphologically by TEM, with differences in mitochondrial ultrastructure and different stages of mitochondrial breakdown, suggesting inadequate removal of damaged mitochondria in HMGB1 KO hepatocytes not observed in control cells (Fig. 5F). These findings lead us to conclude that increased cellular injury with OS observed with lack of HMGB1 may be secondary to mitochondrial damage and dysfunction.

Hepatocytes Lacking HMGB1 Exhibit Increased ROS Production and Cell Death After OS

Mitochondrial dysfunction can perpetuate the production of ROS during OS.[22] So, we sought to discern whether lack of HMGB1 in hepatocytes influenced ROS production during OS. Interestingly, we found that there were higher levels of mitochondrial ROS in hepatocytes lacking HMGB1, compared to control hepatocytes exposed to hypoxia (Fig. 6A). Additionally, HMGB1 KO hepatocytes also displayed a significant increase in total cellular ROS production after stimulation with hypoxia (Fig. 6B).

Figure 6.

Lack of HMGB1 in hepatocytes mediates mitochondrial and cellular ROS production. (A) FCM of cultured hepatocytes from HMGB1-HC-KO or control mice that were stimulated with hypoxia and stained with MitoSOX to determine mitochondrial ROS production. Bar graph represents pooled data from three experiments. Zero hours (0h) represents normoxia. *P < 0.05, compared to PBS treatment. (B) Cellular ROS production as detected by DCF-DA assay in cultured whole hepatocytes, which were obtained from either HMGB1-HC-KO or HMGB1 control mice. Zero hours (0h) represents normoxia. *P < 0.05, compared to control. (C) Representative 4-HNE staining (green, actin; blue, nuclei; red, 4-HNE) and (D) quantification of 4-HNE adducts (4-HNE normalized per nuclei in the field) in ischemic liver of HMGB1-HC-KO or control mice. *P < 0.05, compared to control. (E) Time course of LDH activity in cultured hepatocytes from HMGB1-HC-KO or control mice during hypoxia (n = 4-6 for each point). *P < 0.05 versus control.

To examine whether HMGB1-HC-KO mice exhibit more oxidative damage after liver I/R, we measured 4-HNE Michael adducts, a marker of OS.[23] We found a marked increase in formation of 4-HNE after hepatic I/R in control mice, compared to sham control mice (Fig. 6C). However, the increase in 4-HNE observed in liver of HMGB1-HC-KO mice subjected to I/R was much greater, compared to control I/R livers (Fig. 6D). Altogether, these results suggest that loss of HMGB1 in hepatocytes results in more OS in the liver after I/R injury. Furthermore, greater cell death in HMGB1 KO hepatocytes was confirmed by measuring LDH release from hepatocytes in culture supernatants after hypoxia exposure (Fig. 6E).

Histone-Mediated Inflammation Through TLR9 Is Enhanced in Mice Lacking Hepatocyte HMGB1

We have recently shown that histone release after liver I/R is a major contributor to inflammation and organ damage through a TLR9-dependent mechanism.[2] Our above-mentioned findings demonstrate that loss of HMGB1 in hepatocytes may lead to nuclear instability with histone release. To determine whether increased histone release contributed to worsened liver injury after I/R in HMGB1-HC-KO mice, we used supernatants from hypoxic hepatocytes to stimulate hepatic NPCs. TNF-α, IL-6, and IL-1β mRNA levels were significantly increased in NPCs after treatment with media from hypoxic HMGB1 KO hepatocytes, compared to media from hypoxic control hepatocytes (Fig. 7A and Supporting Fig. 5A). Interestingly, this effect was reduced by treatment with the TLR9 inhibitor (ODN2088) (Fig. 7A). We also treated both HMGB1-HC-KO and control mice with a TLR9 inhibitor during liver I/R, and found that the increased liver damage and cytokine production observed in mice lacking hepatocyte HMGB1 was significantly reduced (Fig. 7B and Supporting Fig. 5B). Thus, these results suggest that lack of HMGB1 may also increase hepatocellular injury by enhancing histone-dependent TLR9 activity.

Figure 7.

Addition of TLR9 antagonists and PARP-1 inhibitors both protect HMGB1-HC-KO mice from liver I/R injury. (A) TNF-α, IL-6, and IL-1β mRNA expression was determined in NPCs cultured overnight with media from hypoxic HMGB1 KO or control hepatocytes. NPCs were treated with TLR9 antagonist, ODN2088, or ODN control. Results are expressed as relative increase of mRNA expression compared with PBS treatment. Data represent the mean ± standard error and are representative of three experiments with similar results. *P < 0.05, compared to ODN control-treated HMGB1-HC-KO group; **P < 0.05, compared to ODN control-treated HMGB1 control group. (B) Serum ALT levels in HMGB1-HC-KO or control mice after 6 hours of reperfusion that were treated with ODN2088 or ODN control. Data represent the mean ± standard error (n = 6 mice per group). *P < 0.05, compared to ODN control-treated HMGB1-HC-KO group; **P < 0.05, compared to ODN control-treated HMGB1 control group. (C) TNF-α, IL-6, and IL-1β mRNA expression was determined in NPCs cultured overnight with media from hypoxic HMGB1 KO or control hepatocytes. Hepatocytes were pretreated with the PARP-1 inhibitor, 3-AB or PJ-34, or negative control, PBS. Results are expressed as the relative increase of mRNA expression compared with PBS treatment. Data represent the mean ± standard error and are representative of three experiments with similar results. *P < 0.05, compared to PBS-treated HMGB1-HC-KO group; **P < 0.05, compared to PBS-treated HMGB1 control group. (D) Serum ALT levels in HMGB1-HC-KO or control mice after 6 hours of reperfusion that were treated with the PARP-1 inhibitor, 3-AB or PJ-34, or PBS. Data represent the mean ± standard error (n = 6 mice per group). *P < 0.05, compared to PBS-treated HMGB1-HC-KO group; **P < 0.05, compared to PBS-treated HMGB1 control group.

Inflammation and Organ Damage Associated With Deletion of HMGB1 Was Ameliorated by PARP-1 Inhibition

Our above-mentioned findings demonstrate that PARP-1 overactivation in cells lacking HMGB1 may lead to cellular damage. Interestingly, inhibition of PARP-1 has been shown to protect the liver and heart from I/R induced injury.[21] To determine whether increased hepatocyte damage from loss of HMGB1 could be prevented with PARP-1 inhibition, we treated hypoxic hepatocytes from either HMGB1-HC-KO or control mice with the PARP-1 inhibitors, PJ-34 or 3-AB. Supernatants from these hepatocytes were added to hepatic NPCs, and production of TNF-α, IL-6, and IL-1β was measured. PARP-1 inhibitor treatment in hypoxic hepatocytes blocked cytokine production from hypoxic media-treated NPCs (Fig. 7C and Supporting Fig. 5C). To further confirm the role of PARP-1 in contributing to enhanced injury in mice lacking HMGB1, we also treated mice with PARP-1 inhibitors during liver I/R. PJ-34 or 3-AB significantly reduced hepatic injury in both HMGB1-HC-KO and control mice after I/R; however, the effect was more pronounced in HMGB1-HC-KO mice (Fig. 7D and Supporting Fig. 5D). These results further solidify the importance of PARP-1 in the injurious OS response of HMGB1-deficient hepatocytes.

Discussion

This study was undertaken to elucidate the cell-specific roles of HMGB1 in the setting of liver I/R injury. Previously, the consequences of lack of HMGB1 have been difficult to define secondary to mice deficient in HMGB1 dying shortly after birth.[9] Therefore, we have developed conditional KO mice to investigate the roles of HMGB1 within hepatocytes. We had postulated that hepatocyte HMGB1 deletion would reduce liver damage during I/R because of the known role for extracellular HMGB1 in driving inflammation-associated injury in I/R.[24] Instead, we found that HMGB1-HC-KO mice are much more sensitive to damage in liver I/R, revealing dominant intracellular roles of HMGB1 during redox stress. In this study, we have illustrated that lack of HMGB1 leads to several detrimental cellular changes during OS, including (1) increased DNA damage and decreased chromatin accessibility to repair, (2) excessive nuclear instability and release of nucleosome components leading to activation of the PRR, TLR9, and (3) activation of PARP-1 is increased, leading to decreased energy stores, with subsequent mitochondrial injury and ROS production.

Generation of ROS after I/R is one of the primary early events with many cellular consequences, including membrane LPO, oxidative changes in protein structure or function, and oxidative damage to DNA.[21] DNA bases are prone to oxidation, leading to detrimental effects in replication and transcription if not repaired by either NER or BER, depending on the lesion.[17] In addition to many other roles, HMGB1 also plays critical roles in NER and BER.[8] Previous work with mouse embryonic fibroblasts found that cells lacking HGMB1 are hypersensitive to DNA damage by ultraviolet (UV) irradiation, leading to increased mutagenesis, chromosomal instability, and decreased cell survival.[25, 26] Normally, acetylation of histones H3 and H4 is increased after DNA damage, allowing for chromatin accessibility and subsequent repair.[27] However, we found that, within HMGB1-HC-KO mice there, was decreased acetylation of histones H3 and H4, suggesting that HMGB1 in hepatocytes is essential for providing chromatin accessibility induced by DNA damage after liver I/R injury. Similar to our finding, Lange et al. have previously shown that mouse embryonic fibroblasts lacking HMGB1 have decreased histone acetylation after UVC irradiation.[26] Increased DNA damage in HMGB1-HC-KO mice was further confirmed with increased phosphorylation of γ-H2AX, a marker of DNA damage.

Oxidative damage to DNA subsequently leads to the activation of several nuclear repair enzymes, such as PARP-1. Ordinarily, with low levels of DNA damage, PARP-1 alters chromatin structure and facilitates DNA repair.[28] However, with excessive DNA damage, PARP-1 may lead to cell death.[29] PARP-1, activated by DNA damage, uses NAD+ as a substrate for PAR formation on target proteins, facilitating nuclear DNA repair.[20] However, excessive DNA damage leads to overactivation of PARP-1, which may lead to NAD+ depletion, exhausting ATP stores.[20, 21] NAD+ depletion may subsequently lead to failure of NAD+-dependent processes, in addition to the opening of mitochondrial permeability transition pores, uncoupling of oxidative phosphorylation, and mitochondrial failure.[30] PARP-1 inhibition or PARP-1 deficiency has been demonstrated to prevent organ damage from liver I/R injury.[31-33] Additionally, there is evidence that PARP-1 activation inhibits recovery of damaged mitochondria.[21] Sustained mitochondrial depolarization leads to further oxidant stress, perpetuating cellular injury. Indeed, in this study, we have found that, with lack of HMGB1, there is increased PARP-1 activation and reduced NAD+, which leads to increased mitochondrial injury and subsequently increases both total cellular and mitochondrial ROS generation—thus providing one mechanism for the increased hepatocellular injury observed within HMGB1-HC-KO mice after hepatic I/R.

Downstream from cellular injury, there is release of nucleosome components, including DNA and histones, which may then activate the TLR9-signaling pathway, propagating sterile inflammatory response. We have previously demonstrated that protective effects of blocking extracellular histones and detrimental effects of exogenous histones in hepatic I/R are dependent on TLR9 signaling.[2] In this study, we have further supported that activation of TLR9 signaling exacerbates injury associated with lack of hepatocyte HMGB1. Inhibition of TLR9 did indeed abrogate the increased injury observed in HMGB1-HC-KO mice subjected to hepatic I/R.

In summary, we used novel hepatocyte-specific Tg HMGB1 KO mice to investigate the role of HMGB1 within hepatocytes subjected to OS, either by I/R or hypoxia exposure. The increased injury and inflammatory response observed in these HMGB1-HC-KO mice is, in part, related to increased DNA damage and nuclear instability with resultant increased release of histones. Diagrammatic summarization of the key findings of this study is shown (Fig. 8). Interestingly, we have found that HMGB1 may serve dichotomous roles after a sterile inflammatory insult, having both a beneficial intracellular role and injurious extracellular role. The novel findings of this study help to dissect the essential role of HMGB1 during cellular stress, suggesting that the intracellular role of HMGB1 during sterile inflammatory response may even outweigh its function as an extracellular signaling molecule and DAMP.

Figure 8.

Schematic representation of the intracellular role of HMGB1 during liver I/R injury. During liver I/R injury, HMGB1 deletion in hepatocytes leads to excessive DNA damage, which initiates overactivation of PARP-1, subsequently exhausting both NAD+ and ATP reserves, damaging mitochondria by depolarization of the mitochondrial membrane. Damaged mitochondria are associated with more mitochondrial and cellular ROS production, eventually leading to more cell stress and death. Excessive cell death further propagates inflammatory response and infiltration of innate immune cells in liver I/R by released histones.

Acknowledgment

The authors thank Xinghuang Liao, Nicole Hays, and Junda Chen for their technical assistance in preparing the manuscript.

Ancillary

Advertisement