Potential conflict of interest: Nothing to report.
In this study, we determined the role of the nuclear factor-kappaB (NF-κB) subunit c-Rel in liver injury and regeneration. In response to toxic injury of the liver, c-Rel null (c-rel−/−) mice displayed a defect in the neutrophilic inflammatory response, associated with impaired induction of RANTES (Regulated upon Activation, Normal T-cell Expressed, and Secreted; also known as CCL5). The subsequent fibrogenic/wound-healing response to both chronic carbon tetrachloride and bile duct ligation induced injury was also impaired and this was associated with deficiencies in the expression of fibrogenic genes, collagen I and α-smooth muscle actin, by hepatic stellate cells. We additionally report that c-Rel is required for the normal proliferative regeneration of hepatocytes in response to toxic injury and partial hepatectomy. Absence of c-Rel was associated with blunted and delayed induction of forkhead box M1 (FoxM1) and its downstream targets cyclin B1 and Cdc25C. Furthermore, isolated c-rel−/− hepatocytes expressed reduced levels of FoxM1 and a reduced rate of basal and epidermal growth factor–induced DNA synthesis. Chromatin immunoprecipitation revealed that c-Rel binding to the FoxM1 promoter is induced in the regenerating liver. Conclusion: c-Rel has multiple functions in the control of liver homeostasis and regeneration and is a transcriptional regulator of FoxM1 and compensatory hepatocyte proliferation. (HEPATOLOGY 2010.)
Nuclear factor-kappaB (NF-κB) is a regulator of hepatic inflammation, wound-healing, regeneration, and carcinogenesis.1, 2 These functions reflect the ability of NF-κB to stimulate expression of cytokines, chemokines, growth factors, and regulators of apoptosis and cell proliferation.3 The classic NF-κB activation pathway is induced in response to a variety of stimuli including inflammatory mediators and microbial or host ligands of the Toll-like receptor system. In response to these stimuli the inhibitor of NF-κB (IκB) kinase (IKK) complex (IKK1, IKK2, and NEMO [NF-κB essential modifier]) is activated, leading to phosphorylation of the inhibitor IκBα and subsequent nuclear transport of active NF-κB.1–3 Most studies of hepatic NF-κB have focused on this classic pathway and employed genetic or pharmacological modulation of IKK or IκBα.1, 2, 4–6 These studies did not address the discrete functions of the five NF-κB subunits (RelA, c-Rel, RelB, p50, and p52) that combine to form functionally distinct NF-κB homodimers and heterodimers.7
Gene knockout mice exist for all five NF-κB subunits and reveal nonoverlapping functions.7 For example, relA−/− is embryonic lethal due to profound loss of hepatocyte survival mechanisms during embryogenesis.8 The other subunit knockouts are all viable with no obvious liver dysfunction. However, specific immunological and hematological defects are documented for mice deficient in c-rel, relB, nfkb1 (p50), and nfkb2 (p52), raising the possibility of influences on wound-healing responses in the liver.7, 9–12 In support of this, nfkb1−/− mice are susceptible to hyperinflammatory and fibrogenic responses in the liver.13, 14
NF-κB complexes containing c-Rel are mainly found in hematopoietic cells; however, c-Rel is expressed in a variety of cell types and organs at varying levels.15 Mice deficient in c-rel display multiple immunological abnormalities including proliferative and functional defects in mature B and T cells, as well as aberrant expression of cytokines and cell survival factors.9, 16 c-Rel is essential for dendritic cell (DC) maturation and for their ability to stimulate T cell responses.17 DCs and macrophages lacking c-Rel display defects in expression of interleukin-12, with production of the p35 subunit defective in DCs and expression of p40 defective in macrophages.18, 19 Functions for c-Rel outside of the immune system are emerging with overexpression, amplification, or rearrangement of the human gene reported for solid tumors.20 To date, hepatic functions of c-Rel in either normal or pathological conditions have not been investigated.
In this study, we show that c-Rel is expressed in the adult mouse liver and we report defects in the hepatic inflammatory, wound-healing, and regenerative responses of c-rel−/− mice, thus revealing a previously unrealized function for c-Rel as an orchestrator of the healing response of the damaged liver.
c-rel−/− mice were backcrossed nine times to a pure C57BL/6 background.21 Male mice (25–30 g) were intraperitoneally injected once (acute) or twice weekly for 12 weeks (chronic) with CCl4 at 1 μL/g body weight (CCl4:olive oil at 1:1 [vol/vol] and 1:3 [vol/vol] [chronic]). Partial hepatectomy (PHx) was performed by removal of 70% of the liver, and sham-operated animals were used as controls. PHx mice were injected intraperitoneally with 100 mg bromodeoxyuridine (BrdU)/kg body weight, 2 hours before culling. Bile duct ligation (BDL) was performed by exposing the bile duct and double-ligating it, then cutting through between the ligations. Mice were then allowed to develop cholestatic disease and fibrosis over a period of 21 days. All animals received humane care according to the criteria outlined in the “Guide for Care and Use of Laboratory Animals”.
Western Blotting and Enzyme-Linked Immunosorbent Assay.
Sodium dodecyl sulfate polyacrylamide gel electrophoresis and immunoblotting was performed on 50 or 200 μg whole liver protein as previously described.13 Antibodies tested are listed in Supporting Table 1. RANTES (Regulated upon Activation, Normal T-cell Expressed, and Secreted; also known as CCL5) protein was measured using a mouse RANTES immunoassay (Quantikine; R&D Systems, Minneapolis, MN) and 125 μg of whole liver protein extract.
Immunohistochemistry and Densitometry.
Deparaffinized-sections were incubated in hydrogen peroxide/methanol. Alpha-smooth muscle actin (α-SMA) staining was performed as described.13 For antigen retrieval, sections were incubated at 37°C in 0.0005% trypsin–ethylene diamine tetraacetic acid (proliferating cell nuclear antigen [PCNA], BrdU) or 0.01% pronase (CD68, neutrophil marker [NIMP]) in phosphate-buffered saline for 20–30 minutes. Sections were blocked using avidin/biotin (Vector) and blocking serum (Sigma) for 20 minutes each. Antibodies tested are listed in Supporting Table 1. Sections were hematoxylin counterstained. Liver slides were blinded, and random fields were counted by two researchers in twenty 400× magnification areas. Image analysis was performed using Leica Qwin.
Isolation of Hepatocytes and Hepatic Stellate Cells.
Livers were perfused with Earle's Balanced Salt Solution minus Ca2+/Mg2+ (EBSS; Gibco) with 500 μM ethylene glycol tetraacetic acid, then EBSS without Ca2+/Mg2+ only, and finally EBSS with Ca2+/Mg2+ and 0.05% (wt/vol) collagenase A (all solutions at 37°C). Digested livers were filtered through nybolt and the cells collected by centrifugation (50 g for 3 minutes) and washed twice in EBSS. Hepatocytes were cultured in William's medium E (WME; Gibco) supplemented with 10% fetal bovine serum. Hepatic stellate cells (HSCs) were isolated and cultured as described.13
Isolation of Total RNA and Real-Time Reverse Transcription Polymerase Chain Reaction.
RNA was isolated using the RNeasy Mini Kit (Qiagen) following the manufacturer's instructions. Reverse transcription was performed as previously described.13 Real-time polymerase chain reaction (PCR) was based on SYBR-Green. Primer sequences are included in Supporting Table 2.
Tritiated Thymidine Incorporation Assay.
Freshly isolated hepatocytes were seeded in a 96-well plate at 6 × 104 cells/well in 100 μL WME containing 10% fetal bovine serum. After 5 hours, medium was replaced to serum-free WME. After 24 hours of serum-starvation, cells were pulsed with 0.5 μCi [3H]thymidine (Amersham) and incubated for 18–20 hours. To quantify [3H]thymidine incorporation, cells were harvested and measured by scintillation counter.
Chromatin immunoprecipitation (ChIP) assays were carried out as described in the Supporting Materials and Methods text using the mouse-specific primers forkhead box M1 (FoxM1) cRel Ch1F = 5′-GCC ACG TAA CCG CAA GTC TA-3′ and FoxM1 cRel Ch1R = 5′-TCA GTG GTC GAC TTC CTT CC-3′.
Data are expressed as means ± standard error of the mean (SEM). All P values were calculated using a two-tailed paired or unpaired Student t test. Statistically significant data is represented in figures where *, **, and *** denote P values of < 0.05, < 0.01, and < 0.001, respectively.
Defective Hepatic Inflammatory Responses in Injured c-rel−/− Mice.
Western blot confirmed hepatic expression of c-Rel in adult wild-type (Wt) C57BL/6 male mice and absence of expression in c-rel−/− mice (Fig. 1A). Repeated administration of carbon tetrachloride (CCl4) induces hepatic inflammation and fibrosis which resolve upon removal of injury.22 Wt and c-rel−/− mice injured for 12 weeks with CCl4 were culled at days 1, 3, 7, and 10 following final administration of CCl4 so as to analyze pathology at peak injury (day 1) and during recovery (days 3-10). Serum alanine aminotransferase (ALT) and aspartate aminotransferase (AST) confirmed similar levels of liver injury between Wt and c-rel−/− mice at day 1, which declined to control levels by day 3 (Supporting Fig. 1). Immunohistochemistry for the neutrophil marker NIMP revealed a marked defect in the neutrophilic response of c-rel−/−; at peak injury there were 60% less neutrophils in the injured knockout liver compared with Wt (Fig. 1B,C). Similar numbers of neutrophils were detected in the spleen of Wt and c-Rel–deficient mice, indicating normal neutrophil production in c-rel−/− mice (Fig. 1B,D). We additionally observed a trend toward reduced numbers of macrophages (CD68+) in c-rel−/− livers; however, differences did not reach statistical significance (Supporting Fig. 2).
RANTES has previously been reported to be regulated by c-Rel.23 Mice that overexpress RANTES revealed a preferential role for the chemokine in the recruitment of neutrophils.24 We therefore investigated if deficiency of c-Rel is associated with attenuated induction of RANTES. In Wt mice, peak injury (day 1 following final CCl4 injection) was associated with increased RANTES expression (Fig. 2A). With recovery, there was a gradual decline in RANTES transcript, reaching baseline levels by day 10. RANTES transcript was found at reduced levels in the livers of uninjured (olive oil control) and at a peak in injured in c-rel−/− mice. These RANTES defects were also observed by enzyme-linked immunosorbent assay on whole liver protein extract (Fig. 2B). A single administration of CCl4 provides a model for acute resolving inflammation. With this model, we observed defective neutrophil recruitment at 24 hours after injury in c-rel−/− livers (Fig. 3A,B), which was associated with reduced expression of RANTES transcript compared to Wt (Fig. 3C). Of note, higher serum ALT and AST levels were observed in acute injured c-rel−/− mice, suggesting an increased susceptibility to liver damage that was somehow compensated for in chronic injury (Supporting Fig. 3).
The Fibrogenic Response to Injury Is Attenuated by Deficiency of c-Rel.
Chronic CCl4 injury provokes activation of HSCs which adopt an α-SMA+ myofibroblastic phenotype characterized by expression of type I collagen and the matrix metalloproteinase inhibitor tissue inhibitor of metalloproteinase 1 (TIMP-1).22 Persistence and proliferation of myofibroblasts leads to the formation of tracts of fibrotic matrix that bridge vessels and thicken as the fibrotic reaction matures. Chronic injury of C57BL/6 mice led to advanced fibrosis (fibrosis scoring 3 on Sirius red–stained sections) associated with highly elevated numbers of α-SMA+ HSCs at peak injury (day 1) (Fig. 4A,B). Both the grade of fibrosis (score 2) and numbers of α-SMA–positive cells were reduced for c-rel−/− mice. Quantification of Sirius red staining by densitometric analysis confirmed a significantly reduced level of collagen deposition at peak injury in c-rel−/− compared to Wt livers (Fig. 4B). BDL-induced fibrosis was also attenuated in c-rel−/− mice (Fig. 5A) with reduced collagen deposition again being confirmed by densitometric analysis (Fig. 5B). The observation of reduced fibrosis in both the CCl4 and BDL models indicates an important and previously unreported role for c-Rel in fibrogenesis. Activation of HSCs and fibrogenesis is influenced by the inflammatory response25 and RANTES, which are both defective in c-rel−/− livers and may therefore partly explain the attenuated fibrosis. However, we also observed intrinsic differences in the phenotype of in vitro culture-activated c-rel−/− HSCs, with reduced expression of collagen I (Fig. 5C), suggesting a direct influence of c-Rel on fibrogenesis. We also observed a trend toward reduced expression of α-SMA in c-rel−/− HSCs, although this did not reach statistical significance. Of note, levels of TIMP-1 were similar between Wt and c-Rel–deficient HSCs. These data suggest selective influences of c-Rel on fibrogenic gene expression and that the decreased fibrosis observed in injured c-rel−/− mice may primarily be due to reduced collagen expression, despite similar levels of TIMP-1.
Hepatocyte Regeneration Requires c-Rel.
Toxic injury of the liver is associated with loss of hepatocytes that triggers compensatory hepatocyte proliferation.26 Proliferating hepatocytes were detected in chronic CCl4-injured livers using antibodies recognizing PCNA and by hematoxylin counterstaining to visualize mitotic bodies (Fig. 6A). The proliferative markers were most abundant at day 3 after injury, indicating a burst of replicative activity occurs during the early phase of recovery from fibrotic disease (Fig. 6B). The c-Rel–deficient livers displayed low numbers of PCNA-positive hepatocytes, indicating a defect in hepatocyte DNA synthesis. The “gold standard” model for study of hepatocyte regeneration is partial hepatectomy (PHx) which induces synchronized compensatory hepatocyte proliferation in the remnant liver.27 BrdU and PCNA were widely detected at 36 hours after PHx in Wt hepatocytes and were 14-fold and 4-fold lower, respectively, in c-rel−/− livers at this time point (Fig. 7A). This dramatic effect on hepatocyte DNA synthesis was associated with a modest reduction in recovery of liver mass at 72 hours (Supporting Fig. 4). FoxM1B (FoxM1) is a transcription factor that is essential for normal mitosis in the regenerating liver.28 Regeneration in Wt liver was associated with a sharp peak of FoxM1 transcript expression at 36 hours after PHx (Fig. 7B). FoxM1 expression was blunted in c-rel−/− livers at 36 hours, but slightly elevated compared to Wt at 72 hours, suggesting a requirement of c-Rel for appropriate timing of FoxM1 expression during regeneration. ChIP analysis using chromatin prepared from sham-operated liver revealed an absence of c-Rel binding at the FoxM1 promoter (Fig. 7C). However, induced c-Rel binding at the FoxM1 promoter was observed at both 24 and 36 hours after hepatectomy (Fig. 7C), with a 25-fold induction at the latter time point, which coincided with maximal expression of FoxM1 transcript (Fig. 7B). FoxM1 is therefore a direct target for c-Rel but only in response to injury/regeneration. Subsequent targets for transcriptional stimulation of DNA replication by FoxM1 are cyclin B1 and Cdc25C.29 In Wt livers, cyclin B1 and Cdc25C transcripts were induced at 36 hours after PHx and expression subsequently declined at 72 hours (Fig. 7D). Induction of cyclin B1 and Cdc25C was suppressed by 50% in c-rel−/− livers at 36 hours but displayed a subsequent rise in expression peaking at 72 hours to levels observed in Wt mice at the earlier 36-hour time point. Lower expression of cyclin B1 in c-rel−/− versus Wt livers at 36 hours was confirmed at the protein level by western blot (Fig. 7E). We also detected raised levels of cyclin-dependent kinase inhibitor p21Cip1 (p21) in c-rel−/− livers (Fig. 7E). The sustained expression of p21 in c-rel−/− livers resembles data from PHx studies with Foxm1b−/− mice where sustained expression of p21 resulted in decreased activation of Cdk2 kinase.28 To determine hepatocyte function for c-Rel, we established primary hepatocyte cultures from Wt and c-rel−/− livers and determined their baseline (Fig. 8A) and epidermal growth factor–stimulated (Fig. 8B) rates of hepatocyte DNA synthesis, both of which were reduced in c-rel−/− hepatocytes. Western blot analysis of FoxM1 expression was also consistently lower in cultured c-rel−/− hepatocytes compared to Wt (Fig. 8C). These data suggest a function for c-Rel in the control of hepatocyte FoxM1 expression and in the regulation of hepatocyte DNA synthesis.
Repair and regeneration of the injured liver requires orchestration of immune, wound-healing, and regenerative responses involving interplay between nonparenchymal and parenchymal cells. We have discovered pleiotropic functions for c-Rel in the hepatic wound-healing response. Absence of c-Rel leads to multiple defects including the appropriate production of RANTES, neutrophil recruitment, the fibrogenic response, and hepatocyte proliferation. We conclude that c-Rel is an important regulator of liver homeostasis via multiple functions in parenchymal and nonparenchymal cells.
Neutrophil recruitment is essential for the innate immune response to injury and is controlled by several different chemokines, including RANTES. Transgenic overexpression of RANTES leads to selective enhanced recruitment of neutrophils following tissue injury. RANTES can also directly target HSCs to promote their proliferation and migration, and mice deficient for RANTES or its receptors chemokine (C-C) motif receptor 1 (CCR1) and CCR5 display substantially reduced fibrosis.30 Here, we show that deficiency of c-Rel is associated with substantially reduced baseline and injury-induced expression of RANTES, which may therefore help explain the reduced numbers of recruited neutrophils, lower numbers of α-SMA+ HSCs, and the attenuated fibrogenic response. However, using the culture model of HSC transdifferentiation, we also discovered inherent defects in c-rel−/− HSCs, specifically reduced expression of collagen I and α-SMA transcripts. NF-κB is a regulator of HSC survival and their expression of inflammatory regulators intercellular cell adhesion molecule-1 and interleukin-6.31 Pharmacological blockade of NF-κB can promote HSC apoptosis and regression of liver fibrosis.32, 33 However, the precise contribution of the individual NF-κB subunits toward the fate and function of HSCs has not been investigated. Our previous report that the p50 subunit is a suppressor of the inflammatory properties of HSC-derived myofibroblasts,13 taken together with the potential for c-Rel to regulate expression of collagen I, α-SMA, and RANTES suggests the need for detailed studies of the functions of the NF-κB subunits in HSCs and fibrosis.
Nonparenchymal cells, including HSCs, can influence liver regeneration through paracrine stimulation of hepatocyte proliferation.34 Defective function of the inflammatory and fibrogenic compartments may therefore contribute to the attenuated DNA synthesis and mitosis of hepatocytes observed in injured and PHx livers of c-rel−/− mice. However, we propose that c-Rel also plays a more direct role as a regulator of hepatocyte DNA replication. B cells deficient in c-Rel display deficiencies in cyclin D3 and cyclin E expression, cyclin-dependent kinase activity, Rb phosphorylation, and E2F activity and fail to progress through the cell cycle in response to B cell receptor stimulation.35 Because ChIP analysis confirmed recruitment of c-Rel to the FoxM1 promoter following PHx, we suggest that c-Rel regulates hepatocyte proliferation via transcriptional control of the cell cycle regulator FoxM1, which following PHx, was not induced at the appropriate time or level of expression in c-Rel–deficient livers. FoxM1 regulates proliferation of many cell types and in the developing liver and heart is essential for normal mitosis.36 Expression profiling identified a cluster of FoxM1-regulated genes including G2/M-specific genes such as cyclin B1 and CENP-F (centromere protein F).37 In particular, transcriptional activation of cyclin B1 by FoxM1 is crucial for timely mitosis.37 Induction of cyclin B1 was delayed in the regenerating c-Rel–deficient liver. FoxM1 expression is repressed in quiescent hepatocytes, but is expressed during liver regeneration and is required for DNA synthesis and mitosis.38 Aberrant induction of FoxM1 following PHx and the associated defects in the expression of cell cycle factors (delayed induction of cyclin B1 and Cdc25C combined with sustained expression of p21) found in regenerating c-rel−/− livers resembles the phenotype described in mice with hepatocyte-targeted disruption of foxm1.38 We conclude that c-Rel is required for appropriate timing of the induction of FoxM1 and exerts a regulatory influence on hepatocyte DNA synthesis during the regenerative response.
Recent work provides conflicting data for the effects of hepatocyte-targeted blockade of NF-κB on proliferative responses to injury and PHx.39-41 These studies employed inducible hepatocyte-selective transgenic expression of a degradation-resistant IκBα transgene or hepatocyte-targeted deletion of IKK2, both of which lead to inhibition of the canonical (RelA/p50) NF-κB pathway. However, functions for IKK2 are emerging outside of the NF-κB system, including influences of proteins intimately associated with cell cycle control.42 Further investigation of distinct NF-κB subunit-specific functions may help better define the role of the NF-κB system in liver homeostasis and regeneration.
In summary, c-Rel may now be considered an important regulator of hepatic wound-healing. Moreover, the potential for c-Rel activities to influence pathological features of the chronic injured liver—including hepatitis, fibrosis, and hepatocellular carcinoma—should be explored.