fax: (81) 798-40-5423
Liver Biology and Pathobiology
Contribution of Toll-like receptor/myeloid differentiation factor 88 signaling to murine liver regeneration†
Article first published online: 18 FEB 2005
Copyright © 2005 American Association for the Study of Liver Diseases
Volume 41, Issue 3, pages 443–450, March 2005
How to Cite
Seki, E., Tsutsui, H., Iimuro, Y., Naka, T., Son, G., Akira, S., Kishimoto, T., Nakanishi, K. and Fujimoto, J. (2005), Contribution of Toll-like receptor/myeloid differentiation factor 88 signaling to murine liver regeneration. Hepatology, 41: 443–450. doi: 10.1002/hep.20603
Conflict of interest: Nothing to report.
- Issue published online: 22 FEB 2005
- Article first published online: 18 FEB 2005
- Manuscript Accepted: 23 NOV 2004
- Manuscript Received: 13 JUL 2004
- Ministry of Education, Culture, Sports, Science and Technology of Japan (MEXT)
- Hitec Research Center grant from MEXT
- The Osaka Foundation for Promotion of Clinical Immunology
- Hyogo College of Medicine
- Yamanouchi Foundation for Research on Metabolic Disorders
Toll-like receptors (TLRs) act as innate immune signal sensors and play central roles in host defense. Myeloid differentiation factor (MyD) 88 is a common adaptor molecule required for signaling mediated by TLRs. When the receptors are activated, cells bearing TLRs produce various proinflammatory cytokines in a MyD88-dependent manner. Liver regeneration following partial hepatectomy (PH) requires innate immune responses, particularly interleukin-6 (IL-6) and tumor necrosis factor α (TNF-α) production by Kupffer cells, although the recognition and activation processes are still unknown. We investigated whether TLR/MyD88 signaling is critical for induction of innate immune responses after PH. In Myd88−/− mice after PH, induction of expression of immediate early genes involved in hepatocyte replication and phosphorylation of STAT3 in the liver, and production of TNF-α/IL-6 by and activation of NF-κB in the Kupffer cells were grossly subnormal and were associated with impaired liver regeneration. However, TLR2, 4 and 9, which recognize gram-negative and -positive bacterial products, are not essential for NF-κB activation and IL-6 production after PH, which excludes a possible contribution of TLR2/TLR4 or TLR9 to MyD88-mediated pathways. In conclusion, the TLR/MyD88 pathway is essential for incidental liver restoration, particularly its early phase. (HEPATOLOGY 2005;41:443–450.)
Tissue repair and cell renewal are essential for maintenance of homeostasis in organisms. Although the immune system responds promptly to invading pathogens, the liver has the capacity to regenerate rapidly in response to various stimuli, such as traumatic or infectious injuries. Hepatocytes are quiescent under normal conditions. When more than 70% of the liver is resected in mammals, hepatocyte proliferation progresses until the liver reaches its appropriate size.1, 2 Many researchers have investigated the molecular mechanisms underlying liver regeneration after partial hepatectomy (PH). They have found that adequate hepatocyte replication requires 2 sequential biological processes designated the priming phase and the progression phase; during these phases the actions of various cytokines and growth factors are orchestrated.1–6 The priming phase is characterized by the immediate induction of expression of various early genes that are involved in cell replication and organ regeneration, such as c-myc, c-fos, and members of the Jun family in hepatocytes.1, 2, 7, 8 However, the roles that individual gene products play in the induction of the priming phase and how early gene expression is regulated are unknown. During the progression phase, hepatocytes express various cell cycle genes, including cyclin D1, and various types of growth factor, such as epidermal growth factor and hepatocyte growth factor, which regulate their replication.9–11 Hepatocyte growth factor is important for hepatocyte proliferation as well as protection from hepatic fibrosis during the progression phase.12 Nonparenchymal liver cells, such as Kupffer cells, play an essential role in the priming phase. It is well established that tumor necrosis factor alpha (TNF-α) and interleukin-6 (IL-6), produced by nonparenchymal liver cells, play a critical role in activation of the priming phase.7, 8, 13, 14 Indeed, Tnfr type I−/− or Il-6−/− mice exhibit impairment of liver regeneration after PH.7, 8 Moreover, normal activation of nuclear factor κB (NF-κB) and signal transduction and activation of transcription (STAT) 3, which yield signaling factors TNF-α and IL-6, respectively, are also essential for liver regeneration.15–17 However, the signal pathways involved in the production of these cytokines/factors after PH are still unknown. The molecular basis for the induction of TNF-α/IL-6 and the activation of NF-κB/STAT3 during the priming phase of liver regeneration needs to be elucidated.
A family of Toll-like receptors (TLRs) is an essential component of signaling receptors expressed on the cell surface; it distinguishes pathogen-associated molecular patterns (PAMPs) produced by microorganisms from intact self-molecules.18–20 Individual TLRs recognize different classes of PAMPs. After stimulation with an appropriate ligand, TLRs relay a signal via myeloid differentiation factor (MyD) 88, a common signal adaptor molecule shared by all the members of TLRs except for TLR3,21, 22 to activate NF-κB23; this process subsequently results in the production of various pro-inflammatory cytokines, including IL-6 and TNF-α. Thus, the TLR/MyD88-mediated biological events are reminiscent of those observed in the liver after PH. Moreover, the stress of PH might destroy the blood-liver barrier and allow the liver to be exposed to PAMPs derived from intestinal commensal bacteria.24 Interestingly, recent reports have shown that denatured self-derived molecules, such as heat shock proteins, have the capacity to activate TLRs.25, 26 These observations prompted us to assume that TNF-α/IL-6 production is attributable to activation of TLR/MyD88 pathways. We sought to determine whether TLR/MyD88-mediated pathways, whether activated by PAMPs or self-derived intrinsic TLR ligands, participate in liver regeneration after PH.
Materials and Methods
Mice and Operative Procedure.
Specific pathogen-free male C57BL/6 mice (8-10 weeks old), purchased from Clea Japan (Osaka, Japan), were used. Myd88−/− mice, Tlr2−/−Tlr4−/− mice, and Tlr9−/− mice were backcrossed with C57BL/6 mice; male F8 mice (8-10 weeks old) were used.27–30 Mice underwent 70% hepatectomy according to the Higgins and Anderson method.31 Briefly, the left lateral, left median, and right median lobes were separately isolated and then removed using a single ligature. At the indicated times after operation, serum and liver specimens were obtained. In some experiments, we measured the liver weight and calculated liver/body weight ratio as follows: liver/body weight ratio as % = 100 × (mean liver weight/mean body weight of the experimental group)/(mean liver weight/mean body weight of a control group). In general, more than 5 mice were used for 1 experimental or control group. All animals received appropriate care as outlined in the Guide for the Care and Use of Experimental Animals (Hyogo Medical College Animal Care Committee, and Osaka University Animal Care Committee).
Electrophoretic Mobility Shift Assay (EMSA).
Double-stranded, NF-κB–specific oligonucleotide probe, the consensus sequence (5′-TCG-AGG-GCT-GGG-GAT-TCC-CCA-TCT-C-3′), was labeled with [32P] dCTP by Klenow fragment.30 Whole-liver nuclear extract17 (20 μg) was incubated with 300 fmol probe in a total of 25 μL binding buffer (10 mmol/L HEPES [pH 7.8], 30 mmol/L KCl, 1 mmol/L ethylenediaminetetra-acetic acid, 10% glycerol, and 2 μg poly-dIdC) for 30 minutes at room temperature. After incubation, samples were electrophoresed. The gel was subsequently dried and evaluated by autoradiography.
Western Blot Analysis.
Western blot analyses were undertaken as previously described.17 The primary antibodies (Abs) against phosphotyrosine 705 STAT3 (9131) and total STAT3 (9132) were purchased from Cell Signaling Inc. (Beverly, MA). Anti-cyclin D1 (sc-717) was purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Secondary horseradish peroxidase–conjugated anti-rabbit Abs were used at a dilution of 1:2000 (Amersham Pharmacia Biotech, Piscataway, NJ).
Northern Blot Analysis.
Total RNA was extracted from the liver specimens by using an RNeasy kit (QIAGEN, Hilden, Germany). Northern blotting was undertaken as previously described.30 Briefly, RNA (5 μg) was separated on 1.0% agarose gels containing 2.4% formaldehyde; it was then transferred onto positively charged nylon membranes. After fixation under calibrated ultraviolet irradiation, the membranes were hybridized with digoxigenin-labeled riboprobes for c-Fos, c-jun, JunB, and c-Myc; visualization using alkaline phosphatase-labeled anti-digoxigenin antibody was according to the manufacturer's instructions (Roche Diagnostics GmbH, Mannheim, Germany). Probes for c-Fos, c-jun, and JunB were provided by Dr. Tokuhisa, Chiba University (Chiba, Japan).32
Reverse Transcription–Polymerase Chain Reaction.
Reverse transcription–polymerase chain reaction (RT-PCR) for TNF-α and β-actin were undertaken as previously described.33 The amplifying cycle was 25 for β-actin and 34 for TNF-α.
Histology and Immunohistochemistry.
Liver specimens were fixed in 10% buffered formalin for subsequent histological and immunohistological analyses according to methods described in previous reports.12, 33 For monitoring nuclear translocation of the NF-κB p65 subunit, liver sections were incubated with goat anti–NF-κB p65 Ab (Santa Cruz Biotechnology), as described previously.33 For a positive control, we used liver specimens obtained from WT (wild-type) mice 30 minutes after intraperitoneal injection of 500 ng TNF-α. To assess bromodeoxyuridine (BrdU) staining, 1 hour before harvest of liver specimens animals were injected intraperitoneally with BrdU (Sigma, St. Louis, MO) (50 mg/kg of a 0.2% solution in phosphate-buffered saline). Liver sections were incubated with mouse monoclonal anti-BrdU Ab (DAKO, Kyoto, Japan), as described previously.33 BrdU-positive hepatocytes were counted in three low-power (80×) fields on each slide; the percentage of total hepatocytes that were BrdU positive was calculated. To assess the number of hepatocyte mitoses, liver sections were stained with hematoxylin-eosin. Mitotic hepatocytes were counted in 30 high-power (200×) fields on each slide; counts were expressed as mitotic cells/high-power field.
Assay for Cytokines.
Concentrations of IL-6 were determined by using an enzyme-linked immunosorbent assay kit (Genzyme Techne, Boston, MA).
All data are expressed as means ± SD of triplicate samples. Differences between experimental and control groups were assessed by using the unpaired Student t test. P values less than.05 were considered significant.
Impaired Liver Regeneration in Myd88−/− Mice.
First, we investigated whether the TLR/MyD88-mediated pathways are required for recovery of liver mass after PH. To do this, we compared liver/body weight ratio of WT and Myd88−/− mice after PH. The liver/body weight ratio of Myd88−/− mice was significantly lower than that of WT mice at 48 and 72 hours (Fig. 1A); by 96 hours, the ratios in the two groups were similar (Fig. 1A). Thus, involvement of MyD88-mediated pathways in the early recovery of liver mass after PH appeared to be critical.
Next, we examined whether MyD88 deficiency causes an impairment of hepatocyte replication, by evaluating DNA synthesis by hepatocytes and counting mitotic hepatocytes after PH. In WT mice, an increase in the proportion of hepatocytes that were BrdU positive after PH was seen33; this proprtion reached a peak at 36 hours and had subsequently decreased by 72 hours (Fig. 1B–C). Myd88−/− mice exhibited distinct kinetics of the BrdU binding. Their livers exhibited moderate and modest increases in the proportion of hepatocytes that were BrdU positive at 36 and 48 hours, respectively (Fig. 1B–C); the increases were smaller at 72 hours and later times, when they were comparable to those in WT mice (Fig. 1B–C). We also counted the number of mitotic hepatocytes in Myd88−/− mice at various times after PH and compared them with corresponding counts for WT mice. Consistent with a previous report,8 the kinetics of mitotic hepatocytes were similar to those of BrdU staining, but there was a delay in the onset of mitoses of 12 to 24 hours in both types of mice (Fig. 1C,E). The number of mitotic hepatocytes reached a peak at 48 hours and, thereafter, gradually decreased in WT mice, whereas this number remained low until 48 hours in Myd88−/− mice (Fig. 1D–E). The number of mitotic hepatocytes in the mutant mice was similar to that in WT mice at 72 hours and later times (Fig. 1D–E). Taken together, these findings indicate the importance of TLR/MyD88-mediated pathways in hepatocyte replication during the early phase, but not the late phase, of liver regeneration.
MyD88-Dependent Induction of Immediate Early Gene Expression.
The selective defect of early phase liver regeneration in mutant mice (Fig. 1) prompted us to investigate whether TLR/MyD88-mediated pathways are essential for the induction of the immediate early gene expression. WT mice express c-fos, c-myc, c-jun, and JunB within 1 hour of PH; the duration of expression depended on the individual genes (Fig. 2A). In contrast, Myd88−/− mice exhibited little expression of these genes after PH throughout the period examined (Fig. 2A). These findings clearly indicate that immediate early genes are expressed in a MyD88-dependent manner.
Next, we investigated whether cyclin D1 is induced independent of TLR/MyD88-mediated pathways, because recovery of liver mass eventually occurs in the absence of MyD88 (Fig. 1A). After PH, Myd88−/− mice exhibited induction of cyclin D1 that was comparable to that in WT mice (Fig. 2B). Thus, after PH hepatocytes can enter into the cell cycle in the absence of MyD88.
MyD88-Dependent NF-κB Activation After PH.
Because NF-κB activation is critical for hepatocyte proliferation,17, 34 we investigated whether MyD88 is essential for the activation of NF-κB after PH. To do this, we examined DNA-binding activity of NF-κB in the liver of Myd88−/− mice at specified times. In WT mice, NF-κB DNA binding activity was appreciably increased at 0.5 hours; the increase persisted up to 6 hours, and subsequently it had decreased by 12 hours (Fig. 3A).These observations are consistent with a previous report.7 In contrast, after PH Myd88−/− mice exhibited little increase in DNA-binding activity of NF-κB in the liver (Fig. 3A). Next, to identify the cell types that exhibited NF-κB activation, we employed immunohistochemistry for p65 of NF-κB, RelA. Normal WT liver expressed little NF-κB p65 (Fig. 3B). By 2 hours after PH in WT mice, nuclear localization of p65 was detected primarily in non-parenchymal liver cells (Fig. 3C–D), and infrequently in hepatocytes (data not shown). MyD88 deficiency markedly diminished this nuclear translocation of NF-κB in non-parenchymal liver cells (Fig. 3E–G). Mice treated with TNF-α, another NF-κB activator, exhibited NF-κB activation in both parenchymal and non-parenchymal liver cells (Fig. 3H–I), thereby excluding possible selective artificial binding of this Ab to non-parenchymal liver cells. These findings indicate the importance of MyD88 for NF-κB activation induced by PH, and suggest that nuclear translocation of NF-κB in non-parenchymal liver cells possibly accounts for the prompt early increase in NF-κB DNA-binding activity after PH.
Requirement of MyD88 for the Induction of IL-6 and TNF-α.
Next, we investigated whether MyD88 is required for the induction of IL-6 production after PH. We measured IL-6 concentrations in sera. In WT mice, IL-6 concentrations started to increase at 1 hour, peaked at 3 hours, and had decreased by 6 hours and later times (Fig. 4A). In contrast, IL-6 was undetectable in serum of Myd88−/− mice up to 12 hours (Fig. 4A). This finding indicates the importance of MyD88 in the induction of IL-6 production after PH. We also examined STAT3 activation, as an output of IL-6 signaling, in the liver after PH. In parallel with the kinetics of IL-6, STAT3 was phosphorylated in the livers of WT mice at least from 2 to 6 hours after PH (Fig. 4A–B). Myd88−/− mice exhibited similar kinetics of STAT3 phosphorylation, but the levels were much lower than corresponding levels in WT mice (Fig. 4B). These findings indicate a requirement of MyD88 for IL-6–dependent STAT3 activation.
Because TNF-α can activate NF-κB and directly induce IL-6 production,3, 7, 35 we assessed TNF-α expression in Myd88−/− and WT mice after PH. However, we consistently failed to detect serum levels of TNF-α in either strain of mouse. Accordingly, we undertook RT-PCR for TNF-α mRNA by using total RNA extracted from liver, After PH, TNF-α mRNA was expressed by 2 hours; expression persisted up to 6 hours in WT mice (Fig. 4C). In contrast, TNF-α mRNA was poorly expressed by Myd88−/− mice after PH (Fig. 4C). Taken together, these findings indicate that the MyD88-mediated signal pathway is required for induction of TNF-α and IL-6 after PH, and suggest that impaired activation of NF-κB in the mutant mice may contribute, at least partly, to impaired induction of TNF-α.
TLR2, 4, and 9 Are Not Essential for NF-κB Activation and IL-6 Production After PH.
Finally, we attempted to identify the TLR involved in MyD88-dependent liver regeneration. Intestinal gram-negative and -positive bacteria seem to be candidates for activation of MyD88-mediated pathways. Because their cell wall components and heat shock proteins activate TLR4- or TLR2-mediated signaling,20, 26, 36 we undertook PH in Tlr2−/−Tlr4−/− mice. Tlr2−/−Tlr4−/− mice, however, exhibited elevated serum IL-6 levels, increased NF-κB DNA binding activity in the liver, and kinetics of BrdU binding and mitotic hepatocytes that were similar to corresponding findings in WT mice (Fig. 5A–D). We next investigated TLR9, which plays a role in sensing CpG oligonucleotide derived from bacterial DNA.29 The variables studied increased equally in Tlr9−/− and WT mice (Fig. 5A–D), but increased relatively poorly in Myd88−/− mice (Fig. 5A–D). These findings exclude a possible contribution of TLR2/TLR4 or TLR9 in MyD88-dependent early events during liver regeneration.
This report clearly demonstrates a requirement of TLR/MyD88-mediated pathways in the initiation of the liver regeneration after PH. In the immune system, TLR/MyD88-mediated pathways are essential for eradication of pathological microbes as a consequence of recognition of the appropriate PAMPs at the innate immune phase.18–20, 36 Additionally and importantly, immune responses involving these pathways influence subsequent acquired immunity by preferential induction of type 1 helper T-cell–deviated responses.37 Thus, TLR/MyD88 pathways play an essential role in the relationship between innate and acquired immunity, and this is necessary to ensure efficient host defense. In this study, we observed major nonimmunological roles for the TLR/MyD88 system. After PH, NF-κB is activated, via TLR/MyD88-mediated pathways in non-parenchymal liver cells, to induce IL-6 and TNF-α (Figs. 3, 4); together they may render hepatocytes competent to proliferate in response to subsequent stimuli. In parallel, activated TLR/MyD88-pathways also directly or indirectly induce expression of the immediate early genes associated with cell replication in hepatocytes (Fig. 2A). Hepatocytes primed by IL-6, TNF-α, and unidentified stimuli may lead to progression of the cell cycle (Fig. 2B) and rapid expansion of the response to subsequent potent growth stimuli, such as EGF and HGF, which presumably act independently of MyD88 (Figs. 1, 2B). Potentially, our findings provide new insights into liver regeneration incidentally induced by various stresses, such as microbes, drugs, trauma, or surgical maneuvers.38–40
Intestine-derived lipopolysaccharide (LPS) has been considered to be an initiator of the liver regeneration,41 but our current findings clearly show that LPS is not essential for induction of liver regeneration (Fig. 5). Because it recognizes flagellin, a major constituent of the flagella of bacteria, and transduces signals via MyD88, its action may involve TLR5.42 TLR7, another TLR that requires MyD88 for its signaling, does not seem to be involved because it has been reported to recognize viral single-stranded RNA.43 Because it does not use MyD88 as a signal adaptor molecule,21 the role of TLR3 appears to be negligible. Alternatively, bacteria themselves do not activate a single TLR for initiation of host defense. Tlr2−/−Tlr4−/− and WT mice exhibit comparable clearance of the gram-positive bacterium Listeria monocytogenes at an early phase of infection with this organism, but Myd88−/− mice are much more susceptible to it.44 Thus, it seems plausible that multiple TLRs are activated after PH.
Both IL-1 and IL-18 also use MyD88 as a signal adaptor molecule for their signaling,30, 45 suggesting a possible contribution of IL-1 or IL-18 to MyD88-dependent phenomena. However, Caspase-1−/− mice, which lack the capacity to release IL-1β and IL-18,46–49 exhibit no obvious defects in the biological events involved in liver regeneration after PH, suggesting non-essential roles of IL-1β and IL-18 in PH-induced liver regeneration (data not shown).
IL-6 and TNF-α have been reported to be substantially involved in liver regeneration after PH. Because MyD88-mediated signaling is essential for induction of these cytokines (Fig. 4), we assumed that Myd88−/− mice would exhibit a severe defect in liver regeneration after PH. Tnfr type I−/− mice exhibit impaired liver regeneration, which is associated with decreases in NF-κB activation, IL-6 production, and hepatocyte replication.7 Exogenous IL-6 can restore liver regeneration in Tnfr type I−/− mice without NF-κB being activated,7 suggesting that TNF-α signaling initiates IL-6 induction after PH. Il-6−/− mice exhibit similar degrees of impairment of liver regeneration but have intact nuclear translocation of NF-κB.8 These findings indicate the importance of endogenous IL-6 in normal liver regeneration and suggest that NF-κB activation by TNF-α might be an upstream event relating to IL-6 induction. Interestingly, neither Il-6−/−8 nor Tnfr type I−/−7 mice exhibit such severe impairment of the induction of immediate early gene expression as do Myd88−/− mice (Fig. 2). MyD88 is important in at least two major events during liver regeneration: immediate early gene expression independent of IL-6 or TNF signaling, and the induction of IL-6 and TNF-α. These phenomena might explain Myd88−/− mice exhibiting more severe impairments of liver regeneration than Il-6−/− or Tnfr type I−/− mice. Taken together, these reports and our present observations indicate that, after PH, TLR/MyD88-mediated pathways play a major role in the induction of these two cytokines, and roles in the initiation of biological events that are more important than IL-6 or TNF-α signaling.
We have demonstrated that non-parenchymal liver cells, but not hepatocytes, are the cells that transduce TLR/MyD88-mediated activation of NF-κB after PH. The nuclear localization of NF-κB p65 was observed primarily in non-parenchymal liver cells (Fig 3C–D,F-G). Recently, it has been reported that gene-manipulated mice with selective disruption of NF-κB activation in their hepatocytes exhibit normal nuclear translocation of NF-κB in their non-parenchymal liver cells and normal liver regeneration after PH.50 These findings indicate that NF-κB activation in non-parenchymal liver cells, but not in hepatocytes, contributes to the liver regeneration after PH.
The liver/body weight ratio does not necessarily reflect total liver cell numbers. After PH, MyD88 deficiency causes early (48 hour) inhibition of mitogenic activity and liver/body weight ratio (Fig. 1); the liver/body weight ratio becomes similar to that of WT mice by 96 hours (Fig. 1). Thus, a compensatory mechanism may occur in MyD88 mice. However, both BrdU labeling and mitotic counts did not become normal by 96 hours. This PH finding may reflect hypertrophy of Myd88−/− hepatocytes, a phenomenon reported in Il-6−/− mice.8 Indeed, the numbers of hepatocytes in the mutant mice were approximately 20% less than those in WT mice (data not shown).
Although important for events during the priming phase (Figs. 1, 2A, 3, 4), TLR/MyD88-mediated pathways play a minor role in the development of the late phase of progression of liver regeneration (Fig. 1). Cyclin D1 induction normally occurs in the absence of MyD88 (Fig. 2B). ERK1/2 phosphorylation was almost intact in Myd88−/− mice after PH (data not shown). Therefore, Myd88−/− mice appear to show delayed, but complete, recovery of liver mass after PH.
In conclusion, host innate immune responses, particularly those involving TLR/MyD88-mediated pathways, participate in not only host defense but also tissue renewal as a consequence of prompt induction and activation of inflammation. Some inflammation-associated cytokines can directly activate parenchymal cell proliferation; some may activate phagocytes to eliminate efficiently harmful or injured cells and allow prompt tissue repair/cell renewal. Our present observations may imply that TLR/MyD88 will become a novel target for induction of efficient tissue renewal in response to homeostatic or pathological stimuli.
The authors thank Dr. David A. Brenner, Columbia University, for enthusiastic discussion of this work, and Dr. T. Tokuhisa, Chiba University, for providing us with probes for various genes. We also thank Mss K. Mitani, S. Yumikura-Futatsugi, K. Aizawa, and N. Nakano for excellent technical assistance.
- 31Experimental pathology of liver: restoration of liver of the white rat following partial surgical removal. Arch Pathol 1993; 365: 179–183., .
- 35TNF-a and apoptosis: implications for the pathogenesis and treatment of psoriasis. J Drugs Dermatol 2002; 1: 264–275., .
- 37Toll-like receptors as adjuvant receptors. Biochem Biophys Acta 2002; 2002: 1–13., .