Potential conflict of interest: Nothing to report.
Apoptosis of hepatocytes results in the development of liver fibrosis, but the molecular signals mediating this are poorly understood. Degradation and modification of nuclear DNA is a central feature of apoptosis, and DNA from apoptotic mammalian cells is known to activate immune cells via Toll-like receptor 9 (TLR9). We tested if DNA from apoptotic hepatocytes can induce hepatic stellate cell (HSC) differentiation. Our data show that apoptotic hepatocyte DNA and cytidine-phosphate-guanosine oligonucleotides induced up-regulation of transforming growth factor β1 and collagen 1 messenger RNA both in the human HSC line LX-2 and in primary mouse HSCs. These effects were opposed by TLR9 antagonists. We have recently shown that adenosine inhibits HSC chemotaxis, and we now show that apoptotic hepatocyte DNA also inhibits platelet-derived growth factor (PDGF)-mediated HSC chemotaxis. Inhibition of HSC chemotaxis by PDGF was blocked by TLR9 antagonists, and was absent in primary HSCs from mice deficient in TLR9 or the TLR adaptor molecule MyD88. Stimulation of TLR9 on HSCs blocked signaling by the PDGF signaling molecule inositol 1,4,5-triphosphate and reduced PDGF-mediated increase in cytosolic Ca2+. Conclusion: DNA from apoptotic hepatocytes acts as an important mediator of HSC differentiation by (1) providing a stop signal to mobile HSCs when they have reached an area of apoptosing hepatocytes and (2) inducing a stationary phenotype-associated up-regulation of collagen production. (HEPATOLOGY 2007.)
Chronic liver injury and hepatocyte apoptosis is associated with liver fibrosis in many human diseases. In vivo experimental models have also shown that hepatocyte apoptosis results in liver fibrosis, but the mediators from apoptotic hepatocytes that stimulate liver fibrosis are poorly understood.1 Several molecules are generated by apoptosing hepatocytes and are candidates for stimulating liver fibrosis.
Members of the family of Toll-like receptors (TLRs) are present on hepatic stellate cells (HSC), and activation of at least 1 of these family members results in HSC activation.2 The family of TLRs was first proposed on a theoretical basis, and subsequently 10 members have been identified that recognize a diverse range of molecular patterns, including peptidoglycan, flagellin, lipopolysaccharide, DNA, and double-stranded RNA.3 The original hypothesis that TLRs recognize molecular patterns present on pathogens has been amply verified, and many cells of the innate immune system are activated via TLRs.4 There is, however, no theoretical reason why ligands for TLRs have to be nonself, and it has been discovered recently that some TLRs can be stimulated by molecules of mammalian cell origin. A notable example is TLR9, which is activated by unmethylated DNA rich in cytidine-phosphate-guanosine (CpG). This motif is present in bacterial DNA at high frequency, but is also present in DNA from healthy mammalian cells.5, 6 As mammalian cells undergo apoptosis, genomic DNA undergoes significant modification, which includes the well-known caspase-activated DNAase–mediated cleavage but also aberrant methylation and oxidative damage.7–9 Such changes may result in enrichment of CpG sequences, as suggested by the observation that DNA recognized by anti-nuclear and and anti-DNA antibodies in patients with SLE has a CpG content 5-6 times higher than random DNA from the human genome.10, 11 Even without such enrichment, delivery of mammalian DNA into the appropriate cellular compartment improves its ability to reach and engage TLR9.12
In this study, we demonstrate that apoptotic hepatocyte DNA induces differentiation of mouse and human HSCs via TLR9 and inhibits platelet-derived growth factor (PDGF)-mediated HSC chemotaxis. The mechanism of inhibition of chemotaxis is the down-regulation of intracellular responses to the secondary messenger inositol 1,4,5-triphosphate (IP3), with reduction in PDGF mediated increase in cytosolic Ca2+. In addition mice lacking TLR9 had significantly reduced liver fibrosis. We propose that activation of TLR9 by DNA from apoptotic cells provides a stop signal to retain HSC at sites of cellular apoptosis, where they subsequently undergo effector function.
LX-2 cells are immortalized human stellate cells and were cultured in RPMI 1640 media plus 5% fetal bovine serum and 1% penicillin/streptomycin.13 HepG2 cells were cultured in Dulbecco's modified Eagle's medium plus 10% fetal bovine serum and 1% penicillin/streptomycin. Primary HSCs were cultured in M199, 10% fetal bovine serum, 1% penicillin/streptomycin, 2% gentamycin, and fungizone in a 1:1000 density, with a media change 24 hours after isolation. Primary HSCs were not passaged and were used between day 3 and 4 after isolation. ODN M362 (human TLR9 ligand), ODN TTAGGG (human TLR9 antagonist), ODN1826 (murine TLR9 ligand), and ODN 2088 (murine TLR9 antagonist) were purchased from Invivogen (San Diego, CA). TRIZOL reagent, trypsin/ethylene diamine tetraacetic acid, M199, RPMI 1640 medium, and Dulbecco's modified Eagle's medium were purchased from GIBCO/Invitrogen (Grand Island, NY). All reagents were of the highest grade that was commercially available.
Preparation of Apoptotic DNA.
HepG2 cells were cultured in a 60 × 15–mm dish, and when near confluent were exposed to 600 mJ of ultraviolet irradiation using a UV Stratalinker 1800 (Stratagene, La Jolla, CA). Cell apoptosis was evident 6 hours after irradiation via typical morphological changes. At this time, DNA was extracted by standard techniques. Apoptosis was confirmed by running the extracted DNA on an agarose gel to visualize the characteristic laddering of fragmented DNA.
Animals and HSC Isolation.
Tlr9−/− and myd88−/− mice were used for HSC isolation. These mice were back-crossed for 10 successive generations onto a C57BL/6 background, and control C57BL/6 mice were purchased from Jackson Labs (Bar Harbor, ME). All experiments and animal handling were performed according to the Yale University Institutional Animal Care & Use Committee guidelines.
Primary mouse HSCs were isolated from the livers of 6-week-old to 8-week-old mice. The method described previously was modified per Liu et al.14 Mice were anesthetized, and in situ liver perfusion and digestion were performed with pronase E (2.4 mg/mL; Roche Molecular Biochemicals, Chicago, IL) and collagenase B (0.3-0.45 mg/ml; Roche Molecular Biochemicals, Branchburg, NJ), and the resulting liver cell suspension was purified via density gradient centrifugation using 9%-11% Nycodenz (Sigma-Aldrich, St. Louis, MO). HSCs were plated on standard tissue culture plastic dishes in M199 medium with 10% fetal bovine serum at a density of 1 × 106 per 1-cm plate and maintained in M199/10% fetal bovine serum supplemented with glutamine and antibiotics [penicillin, 100 U/mL; streptomycin, 100 U/mL; gentamicin, 0.1 mg/mL; and fungizone, 2.5 μg/mL (Invitrogen, Carlsbad, CA)].
Reverse Transcriptase Polymerase Chain Reaction for Expression of TLR9.
Complementary DNA was prepared from the human LX-2 HSC line and day 3-4 primary HSCs with and without TLR9 agonists and amplified with TLR9 sense primers (human, 5′-ggacctctggtactgcttcca-3′; murine, 5 ′-gacttactgttggaggtgcagacc-3 ′) and antisense primers (human, 5 ′-aagctcgttgtacacccagtct-3′; murine, 5′-gaacaccagcaaggcatcata-3 ′). β-Actin was amplified with a sense primer (human, 5′-atctggcaccacaccttctacaatgagctgcg-3′; murine, 5′-tggaatcctgtggcatccatgaaac-3′) and antisense primer (human, 5′-cgtcatactcctgcttgctgatccacatctgc-3′; murine, 5′- atctggcaccacaccttctacaatgagctgcg-3′). Polymerase chain reaction (PCR) was performed with 1 μL of complementary DNA, 1× buffer, 1 mM MgCl2, 200 μM of each deoxyribonucleotide triphosphate, 2.5 U Taq polymerase, and 0.2 μM of each TLR9- and β-actin–specific primers.
Quantitative Real-Time PCR for Messenger RNA Expression of Transforming Growth Factor β1 and Collagen 1.
Primary mouse HSCs and LX2 cells were cultured in the presence of human CpG (5 μg/mL), genomic DNA from healthy HepG2 cells (50 μg/mL), apoptotic HepG2 cells (50 μg/mL), CpG (5 μg/mL) + TLR9 antagonist (5 μg/mL), and apoptotic HepG2 cells (50 μg/mL) + TLR9 antagonist (5 μg/mL). Twenty-four hours after culture, complementary DNA was prepared from the above and control LX2 cells, and quantitative real-time PCR was performed for transforming growth factor β1 (TGF-β1), collagen 1, interleukin 6, and intercellular adhesion molecule 1 using commercial primer-probe sets (Applied Biosystems, Framingham, MA) and the Applied Biosystem 7500 real-time PCR system. Expression of glyceraldehyde 3-phosphate dehydrogenase was used to standardize the samples, and the results were expressed as a ratio compared with untreated HSCs.
HSC Chemotaxis Assay.
The migration of LX-2 cells and primary mouse HSCs from C57BL/6 strain mice, tlr9−/− mice, and myd88−/− mice was studied using transwell inserts equipped with 8-μm-pore polycarbonate-free filters as described previously.15 For the primary HSCs, 2 mice were used per isolation and 2 × 104 HSCs were plated per well. Cells were treated with the appropriate human and murine TLR9 antagonist 15 minutes before adding either human or murine TLR9 ligand, apoptotic DNA, or healthy DNA. PDGF was added to the lower chamber 2 hours afterward. After 24 hours, the lower surface of the membrane was stained using hematoxylin-eosin, photographed, and analyzed. All experiments were repeated in triplicate. For each experimental group, a total of 40 high-power field images were taken. Cells per high-power field were counted, with results expressed as the average number of cells per high-power field. Statistical analysis was performed using the Student t test; a P value less than 0.05 was considered significant.
Analysis of Cytosolic Ca2+ in Response to PDGF.
The intracellular free Ca2+concentration was measured using Fura-2/AM (Molecular Probes) as described previously.16 Fluorescence was monitored in ratio mode using a fluorometer (Polarstar Galaxy; BMG Labtechnologies, Offenburg, Germany). Collected data were analyzed using using Fluostar Galaxy Software (BMG Labtechnologies). Data are espressed as a 340:380 nm ratio.
Analysis of Intracellular Sensitivity to IP3 via Measurement of Changes in Cytosolic Ca2+ Concentration.
LX-2 cells were cultured with or without apoptotic DNA (50 μg/mL) on glass coverslips for 20 minutes and then incubated with a solution containing 3 μM caged IP3 (Alexis Biochemicals), 5 μM Fluo-4 (Molecular Probes), 19.7 mM 4-(2-hydroxyethyl)-1-piperazine ethanesulfonic acid, 130 mM NaCl, 5 mM KCl, 1 mM MgSO4, and 1.25 mM CaCl2 for 30 minutes. These cells were placed in a perfusion chamber on the stage of a Zeiss S10NLO microscope; IP3 was then photoreleased using a custom-built system that couples a mercury lamp to a 1-mm quartz fiber optic cable through a high-speed shutter and filter wheel while cells were observed using time lapse confocal microscopy.
Experimental Liver Fibrosis.
Liver fibrosis was induced in C57BL/6 mice and tlr9−/− mice via intraperitoneal injection for CCL4 twice a week for 8 weeks (each injection contained a CCl4 dose of 850 μL/kg, diluted 1:4 in corn oil). Control mice received injections of corn oil. After 8 weeks, mice were euthanized and liver protein was isolated for Western blot analysis. Liver samples were fixed for histology.
Western Blotting for α-Smooth Muscle Actin.
Primary mouse HSCs were cultured with and without CpG (5 μg/mL). Twenty-four hours after culture, total protein was prepared. Western blotting was performed for α-smooth muscle actin (α-SMA) (1: 1000; Santa Cruz Biotechnology, Santa Cruz, CA) and β-actin (1:1000; Santa Cruz Biotechnology) using standard techniques. Radiographs were scanned and analyzed using TL100 software from the Total Lab range (Nonlinear Dynamics, Newcastle upon Tyne, United Kingdom).
Sirius Red Staining.
Liver samples were incubated in xylene (J. T. Baker, Phillipsburg, NJ) for 15 minutes and washed with ethanol (95% for 5 minutes, 90% for 5 minutes, 80% for 10 minutes, and 70% for 5 minutes). Samples were placed in 0.2% phosphomolybdic acid for 2 minutes, then incubated with 0.4% sirius red (Polysciences Inc., Warrington, PA) for 110 minutes and washed with 0.01 N HCl and ethanol (95% and 100%, respectively). After washing with CitriSolv 3 times for 3 minutes each time (FISHERbrand, Pittsburgh, PA), samples were covered by mounting media with a coverslip. Imaging was performed using polarized light.
Immunohistochemistry for α-SMA.
Slides were deparraffinized, and immunohistochemistry was performed using standard techniques and anti–α-SMA primary antibody (1:650; Sigma, St. Louis, MO) in PBS with 0.1% bovine serum albumin for 45 minutes at 37°C. Washing was performed with PBS with 0.1% bovine serum albumin 3 times for 3 minutes each time. Goat anti-mouse immunoglobulin G Alexa 488 was added as the secondary antibody (1:200) in PBS with 0.1% bovine serum albumin for 45 minutes at 37°C. Washing was performed again with PBS with 0.1% bovine serum albumin 3 times for 3 minutes each time. Samples were mounted with mounting media (Vector, Burlingame, CA).
Phalloidin Staining for Confocal Microscopy.
LX-2 cells and mice HSCs were washed with PBS and fixed with a 3.7% paraformaldehyde solution in PBS for 10 minutes at room temperature. After 2 washes with PBS, cells were permeabilized with 0.1% TritonX 100 in PBS for 5 minutes. Cells were stained with Alexa Fluor 594 phalloidin (1; 40) (Invitrogen) and TO-PRO-3 iodide (1; 500) (Invitrogen) for 30 minutes, washed with PBS, and placed on the glass slide with mounting media. The cells were imaged using a Zeiss microscope, and Z-sections were obtained.
Enzyme-Linked Immunosorbent Assay.
LX2 cells were cultured in the presence of human CpG (5 μg/mL), or apoptotic hepatocyte DNA (50 μg/mL). Forty-eight hours after culture, supernatant was collected from the above and control LX2 cells, and TGF-β concentrations were checked via enzyme-linked immunosorbent assay. We used biotinylated anti–TGF-β1 antibody as a detection antibody (R&D Systems, Inc, Minneapolis, MN) and monoclonal anti–TGF-β1, anti–TGF-β2, and anti–TGF-β3 antibodies as capture antibodies (R&D Systems Inc., Minneapolis, MN).
For the chemotaxis experiments, 40 high-power fields were counted, and for each experimental group the average number of cells per high-power field was calculated. The Student t test was performed, with P < 0.05 considered significant. For quantitative real-time PCR, data samples were run in groups of 12 each sample, generating a ratio compared with the control sample; mean and standard deviations were generated from these samples, and a Student t test was performed, with P < 0.05 considered significant.
Apoptotic Hepatocyte DNA and CpG Up-regulate TGF-β1 and Collagen-1 in HSCs.
Transcripts for human and mouse TLR9 are present in the HSC line LX-2 and primary mouse HSCs (Fig. 1A) and were up-regulated by stimulation via TLR9 (Fig. 1B). Up-regulation of messenger RNA (mRNA) for TGF-β1 and collagen 1 is an indicator of an increase in the activation state of HSCs. We therefore tested if apoptotic hepatocyte DNA induces up-regulation of the mRNA transcript for TGF-β1 and collagen 1. Twenty-four hours after adding apoptotic hepatocyte DNA and CPG, the levels of TGF-β1 and collagen 1 in LX-2 cells were significantly greater than those of controls (TGF-β1, P < 0.01; collagen 1, P < 0.01) (Fig. 1C) and primary mouse HSCs (TGF-β, P < 0.05; collagen 1, P < 0.05) (Fig. 1D). CpG-DNA is a potent agonist for TLR9 and also induced up-regulation of TGF-β1 and collagen 1 in LX-2 (TGF-β1, P < 0.01; collagen 1, P < 0.01) and primary mouse HSCs (TGF-β1, P < 0.05; collagen 1, P < 0.05). DNA from healthy hepatocytes induced significantly less up-regulation of TGF-β1 and collagen 1 than DNA from apoptotic hepatocyte DNA in both LX-2 (TGF-β1, P < 0.01; collagen 1, P < 0.01) and primary mouse HSCs (TGF-β1, P < 0.01; collagen 1, P < 0.01). There was no significant up-regulation of intercellular adhesion molecule 1 or inerleukin 6 gene expression in primary mouse HSC by apoptototic or CpG-DNA (data not shown). Consistent with up-regulation of the TGF-β transcript, there was up-regulation of TGF-β at the protein level by TLR9 signaling in LX-2 and primary HSCs (Fig. 1E-H).
TLR9 Antagonists Block the Up-regulation of TGF-β1 and Collagen 1 mRNA Induced by Apoptotic Hepatocyte DNA.
To test if the up-regulation of TGF-β1 and collagen 1 by apoptotic hepatocyte DNA was occurring via activation of TLR9, LX-2 and primary mouse HSCs were incubated with a TLR9 antagonist prior to exposure to apoptotic hepatocyte DNA. The human TLR9 inhibitor ODNTTAGGG significantly reduced the ability of apoptotic hepatocyte DNA to induce an increase in TGF-β1 (P < 0.01) and collagen 1 (P < 0.01) in LX-2 cells (Fig. 2A). The mouse TLR 9 inhibitor ODN 2088 also significantly reduced the ability of apoptotic hepatocyte DNA to induce an increase in TGF-β1 (P < 0.05) and collagen 1 (P < 0.05) in primary mouse HSCs (Fig. 2B). As expected, the TLR9 antagonist also reduced the response produced by the TLR agonist CpG-DNA.
Apoptotic Hepatocyte DNA and CpG-DNA Induce TLR9-Dependent Actin Reorganization and Stellation of HSCs.
Activation of HSCs is associated with changes in cell shape, typically into a more stellate morphology. LX-2 cells have a flat, polygonal morphology as shown in Fig. 3A by phase contrast, and Fig. 3D by phalloidin staining and confocal imaging. Twenty-four hours after exposure to apoptotic or CpG-DNA, the change in morphology is evident in phase-contrast as well as the z-sections of phalloidin-stained cells. Apoptotic or CpG-DNA induce identical changes. Primary HSCs also underwent morphological change in response to apoptotic or CpG-DNA, as shown in the phase-contrast and phalloidin-stained images in Fig. 3G-L. The response of primary cells to apoptotic or CpG DNA is indistinguishable.
Apoptotic Hepatocyte DNA and CpG-DNA Inhibit PDGF-Mediated HSC Chemotaxis.
We recently showed that adenosine, which is produced in areas of cellular apoptosis, inhibits HSC chemotaxis induced by PDGF.15 We tested if activation of TLR9 can also inhibit PDGF-mediated chemotaxis. Figure 4A shows Transwell chemotaxis data for LX-2 HSCs. PDGF induces chemotaxis of LX-2 HSCs, and this is significantly inhibited by pretreatment with apoptotic hepatocyte DNA or CpG (P < 0.001 and P < 0.001, respectively). To confirm that the inhibition of PDGF-induced chemotaxis by apoptotic hepatocyte DNA was via TLR9, LX-2 cells were pretreated with a TLR9 antagonist prior to adding apoptotic hepatocyte DNA, and this blocked the ability of apoptotic hepatocyte DNA to inhibit PDGF-induced chemotaxis (P < 0.001) (Fig. 4A). Primary mouse HSCs have a very similar response, with apoptotic hepatocyte DNA and CpG inhibiting PDGF-mediated chemotaxis (P < 0.001 and P < 0.001, respectively) (Fig. 4B). Pretreatment with a TLR9 antagonist blocked the ability of apoptotic hepatocyte DNA and CpG to inhibit PDGF-induced chemotaxis (P < 0.001 and P < 0.001, respectively). In addition, DNA from healthy hepatocytes was not able to inhibit PDGF-induced chemotaxis. The ability of apoptotic hepatocyte DNA to inhibit PDGF-induced chemotaxis was not blocked by the pan-adenosine inhibitor 8-PST (data not shown). This rules out the possibility that the mechanism of the effect was via generation of adenosine and stimulation of A2a receptors as shown previously by our group.15
Apoptotic Hepatocyte DNA and CpG Do Not Inhibit PDGF-Mediated Chemotaxis of Primary Mouse HSCs from tlr9−/− and myd88−/− Mice.
To confirm the findings from the TLR9 antagonist that inhibition of chemotaxis by apoptotic hepatocyte DNA was TLR9-specific, primary HSCs from tlr−/− and myd88−/− mice were used. As seen in Fig. 5A -B, PDGF-induced chemotaxis in primary HSCs from both tlr−/− and myd88−/− mice demonstrated that they are viable and able to respond to PDGF. In contrast to HSCs from wild-type mice, both apoptotic hepatocyte DNA and CpG were unable to inhibit PDGF-induced chemotaxis.
Activation of TLR9 Inhibits Increase in Cytosolic Ca2+-Induced by PDGF in HSCs.
Cellular chemotaxis is known to be dependent on an increase in cytosolic Ca2+, and we wished to test if activation of TLR9 induces changes in cytosolic Ca2+ concentration. Activation of TLR9 in untreated LX-2 cells did not induce any changes in cytosolic Ca2+ (data not shown). Figure 6 A shows duplicate plots of the expected transient increase in cytosolic Ca2+ after exposure to PDGF. Activation of TLR9 prior to exposure to PDGF significantly diminished the amplitude of the peak and plateau of the increase in cytosolic Ca2+ (P < 0.02).
Apoptotic Hepatocyte DNA Inhibits IP3-Mediated Signaling in HSCs.
It was of interest that the kinetics of the PDGF-induced increase in cytosolic Ca2+were very similar in control and TLR9-activated LX-2 HSCs, but there was a significant decrease in amplitude in the TLR9-activated group. This suggested a desensitization of the cytosolic Ca2+ response to PDGF. Activation of the PDGF receptor results in an increase in cytosolic IP3with subsequent activation of endoplasmic reticulum IP3receptors (IP3Rs). We tested if the mechanism of TLR9 inhibition of PDGF-induced increase in cytosolic Ca2+ was by inhibition of IP3-mediated signaling. LX-2 HSCs loaded with the Ca2+ flurophore Fluo-4 and cell-permeant caged IP3ester (C-iso IP3, Alexis Corp.) were imaged during uncaging of the IP3. In control cells, IP3 uncaging results in a rapid increase in cytosolic Ca2+, which persists for approximately 30 seconds (Fig. 6B,D). In LX-2 HSCs pretreated with apoptotic hepatocyte DNA, there is minimal increase in cytosolic Ca2+ in response to uncaging IP3 (Fig. 6B,D). This demonstrates reduced sensitivity of IP3Rs to an increase in cytosolic IP3.
Mice Lacking TLR9 Have Reduced Collagen Deposition and HSC Activation in a Model of Experimental Fibrosis.
As expected, wild-type mice and tlr−/− and myd88−/− mice receiving corn oil had no liver fibrosis (Fig. 7A,B). Wild-type mice receiving CCl4 developed significant fibrosis as shown by a representative sirius red–stained image (Fig. 7C), and the degree of fibrosis was significantly less in tlr−/− and myd88−/− mice (Fig. 7D). Immunostaining for α-SMA showed a significant signal from the livers of wild-type mice given CCl4, consistent with HSC activation, and minimal signal from tlr−/− and myd88−/− mice given CCl4 (Fig. 7G-H). This difference was supported via Western blot analysis (Fig. 7I).
The ability of hepatocyte apoptosis to stimulate matrix deposition and liver fibrosis is an important observation that demonstrates the presence of adaptive repair responses after pathological cellular apoptosis. Very little is known about the mediators of such adaptive responses. Conceptually, they are either newly generated molecules, molecules that are already present but are compartmentalized within healthy cells, or a combination of the two. One example of the former is adenosine, which is generated intracellularly and extracellularly by the degradation of nucleic acids by the uric acid pathway.17 In addition to being an intermediary degradation product, it stimulates a family of receptors and mediates adaptive changes to cell death in many organs, including the liver.18 This example of a degradation product mediating adaptation to cellular apoptosis prompted us to investigate whether DNA from apoptotic mammalian cells can also act as a mediator of adaptive responses to apoptosis. The only known receptor for DNA molecules is TLR9, and this was initially identified as the receptor responsible for the activation of the innate immune system by viral and bacterial unmethylated DNA sequences rich in CpG motifs.5 The identification of TLR9 as the receptor for viral and bacterial DNA categorized it, along with other TLRs, as a detector of nonself molecules functioning to alert the immune system to the presence of pathogens. However, the specificity of TLR9 for viral and bacterial DNA is not absolute, and the concept of apoptotic mammalian DNA activating TLR9 on HSCs has experimental support in immunology where B cells and plasmacytoid dendritic cells are known to be activated by apoptotic DNA via TLR9.19 For B cells, TLR9 activation requires the nuclear material to be presented as immune complexes, which have low affinity for the B cell receptor, likely resulting in internalization and localization to endosomal compartments. One consequence of this is the ability of cellular apoptosis to stimulate the development of autoimmunity, including the generation of auto-antibodies against nuclear components such as histones, which are tightly bound to DNA. It is important to note that the response of TLR9 to DNA sequences is very complex and is dependent not just on the sequence, but also on the methylation state of the cytosine base and the hairpin and tertiary structure of the molecule.5 There is likely biologically significant complexity in TLR9 signaling that has not been fully appreciated by the use of synthetic CpG-DNA.
The expression of TLR9 is limited mostly to cells of the immune system. As a first step in evaluating the role of TLR9 in HSC biology, we detected expression of TLR9 in human LX-2 and primary mouse HSCs (Fig. 1A). The demonstration of an increase in TGF-β1 and collagen 1 mRNA transcript, along with actin reorganization in response to a TLR9 agonist (CpG-DNA), shows that these receptors may have biological significance and supports the hypothesis of TLR9-mediated differentiation of HSCs (Figs. 1-3). The ability of DNA from apoptotic cells to also up-regulate TGF-β1 and collagen 1 and to induce morphological change makes the important point that DNA from apoptotic hepatocytes may also be a biologically active molecule in HSC differentiation. The use of TLR9 antagonists and HSCs from TLR9- and MyD88-deficient mice provides significant evidence that the effects of apoptotic hepatocyte DNA are mediated via TLR9. These data suggest that apoptotic hepatocyte DNA can increase the amount of matrix deposition and induce morphological changes known to be associated with HSC activation via TLR9. The relative inability of DNA from healthy hepatocytes to induce these changes further supports that this is an apoptosis-dependent phenomenon, and is therefore a candidate mechanism by which in vivo hepatocyte apoptosis results in HSC differentiation and fibrosis.
After demonstrating HSC differentiation via TLR9 activation, we chose to pursue a recent finding by our group of adenosine-mediated inhibition of PDGF-induced HSC chemotaxis.15 As can be seen in Fig. 4, apoptotic hepatocyte DNA is also a potent inhibitor of LX-2 and primary mouse HSC chemotaxis induced by PDGF. The replication of the inhibition by CpG-DNA and the ability of TLR9 antagonists to block the inhibitory effect of apoptotic hepatocyte DNA is significant evidence that the inhibitory effect occurs via TLR9. This is further supported by the data using primary HSCs from mice deficient in TLR9 or the adaptor molecule MyD88 (Fig. 5). In the absence of either of these molecules, primary HSCs are able to undergo PDGF-mediated chemotaxis, but apoptotic hepatocyte DNA and CpG-DNA are unable to block such chemotaxis. Collectively, this is substantial evidence for the ability of apoptotic mammalian DNA to block PDGF-induced HSC chemotaxis via TLR9.
The mechanism of the inhibitory effect of TLR9 activation on PDGF-induced chemotaxis is of great interest. The important role of an increase in the free cytosolic Ca2+ concentration [Ca2+]i in chemotaxis prompted us to examine the effect of TLR9 activation on [Ca2+]i. Although activation of TLR9 did not result in any detectable changes in basal [Ca2+]i, it significantly reduced increases in [Ca2+]i in response to PDGF (Fig. 6A). The increase in cytosolic [Ca2+]i in response to PDGF is mediated by the secondary messenger IP3, which activates IP3Rs.20 These receptors are a family of 3 proteins (types 1, 2, and 3) that form tetrameric ion channels in the endoplasmic reticulum membranes.20 On binding of IP3, the channels open and Ca2+ stored within the endoplasmic reticulum flows into the cytoplasm. The results from our experiments using photo-release of caged IP3with and without activation of TLR9 clearly demonstrates inhibition of IP3-mediated signaling in response to activation of TLR9. This finding suggests that TLR9 signaling results in a decrease in IP3R function and has been previously reported in response to a number of membrane receptors, including cholecystokinin and angiotensin II. One mechanism of a reduction in IP3R function is by a reduction in total cellular IP3R content, which can decrease by 90% in neuroblastoma cells in response to phospholipase C (PLC)-linked cell surface receptors.20, 21 Increased ubiquination of IP3Rs in response to cholecystokinin with subsequent degradation by proteosome-dependent and proteosome-independent pathways has also been reported.22 Finally IP3Rs have several phosphorylation sites that can alter the response of IP3Rs to IP3. Mutagenesis studies suggest that phosphorylation by both protein kinase A and G results in enhancement of IP3-mediated Ca2+ release.23 A detailed analysis of the effects of phosphorylation of the known sites on IP3R activity has not been conducted yet, and phosphorylation at some sites may result in a decrease in IP3R activity. Identification of the mechanisms responsible for TLR9-mediated reduction in the response of IP3R to IP3 is of great interest and deserves further study.
The localization of TLR9 in the endosome necessitates internalization of DNA for signaling to occur, but the mechanisms for this process are not well understood. There is, however, detailed descriptive information on the uptake of DNA into phagosomes, with subsequent localization to tubular lysosomes.24 Fusion of these tubular lysosomes with endosomes allows binding of DNA to TLR9. HSCs are known to phagocytose apoptotic hepatocyte bodies, with subsequent up-regulation of TGF-β1 and collagen α1 mRNA.25 Apoptotic bodies contain nuclear material, including DNA degradation products, and after phagocytosis by HSCs are likely to undergo fusion with endosomes containing TLR9.26 B cells and plasmacytoid dendritic cells internalize DNA bound to immune complexes, via the IgM molecule on B cells and the Fc receptor on plasmacytoid dendritic cells.27–29 It is noteworthy that primary rat HSCs have recently been shown to express all 3 forms of the Fc receptor, raising the possibility that HSCs may also use these receptors to take up DNA-immune complexes outside of apoptotic bodies.30
We propose that the inhibition of PDGF-induced chemotaxis provides a stop signal to activated HSCs when they have reached an area of hepatocyte apoptosis. In common with the broader field of chemotaxis, several molecules have been identified that induce HSC chemotaxis and explain movement of HSCs toward their source.31 However, there is little explanation of the mechanism for stopping cells when they arrive at the source of chemokines, and where they need to undergo effector function. Receptor desensitization at very high concentrations of chemokines has been proposed, but as a sole mechanism it would require complex overlapping of maximal chemokine concentrations to provide the very tight localization of cells known to occur in vivo. A chemotaxis stop signal, produced at the site of cellular apoptosis, is an additional attractive mechanism for localizing HSCs to the site requiring matrix remodeling. Naïve HSCs do not express the PDGF receptor, and responsiveness to PDGF is an indicator of at least a minimum level of HSC activation.32 We therefore propose that DNA from apoptotic cells acts as a late differentiation signal inducing a change from very mobile, partially activated HSCs to immobile effector HSCs.
These results suggest that apoptotic hepatocyte DNA is an important mediator in liver fibrosis and have implications for the use of TLR9 agonists as immune modulators in human diseases. Proposed applications of TLR9 agonists include enhancement of the immune system to treat infections as well as an adjuvant to vaccines. There are currently phase 2 trials of TLR9 agonists for treatment of chronic hepatitis C, and also as an adjuvant for vaccinations for influenza and anthrax.33, 34 Enhancement for the immune system as a treatment of melanoma and other cancers is also under investigation in human studies.35 The ability of TLR9 agonists to stimulate a Th1 response and suppress a Th2 response also makes them attractive candidates for therapy in immune diseases that are dependent on a Th2 pathway. Currently there are phase 1 and 2 trials underway for treatment of a wide range of immune mediated conditions such as asthma, allergic rhinitis, and conjunctivitis.36 Based on our results, the application of TLR9 agonists for immunotherapy may be associated with a risk of HSC differentiation and liver fibrosis. This risk is further accentuated by the known concentration of administered TLR9 agonists in the liver.37
In conclusion, we have demonstrated that DNA from apoptotic hepatocytes induces TLR9-dependent changes consistent with the late stages of HSC differentiation, with up-regulation of matrix components, changes in HSC morphology, and inhibition of PDGF-mediated chemotaxis.
We are grateful to Dr. Scott Friedman for expert advice on HSC biology and for providing the LX-2 cell line. We are also grateful to Dr. M. Nathanson for his expert advice on confocal microscopy. Dr. R. A. Flavell is an Investigator of the Howard Hughes Medical Institute.