Reg2 inactivation increases sensitivity to Fas hepatotoxicity and delays liver regeneration post-hepatectomy in mice


  • Potential conflict of interest: Dr. Hunt holds intellectual property rights with A Smith at Edinburgh.


Reg2/RegIIIβ is the murine homologue of the human secreted HIP/PAP C-type lectin. HIP/PAP transgenic mice were protected against acetaminophen-induced acute liver failure and were stimulated to regenerate post-hepatectomy. To assess the role of Reg2, we used Reg2−/− mice in a model of fulminant hepatitis induced by Fas and in the post-hepatectomy regeneration. Within 4 hours of J0-2 treatment (0.5 μg/g), only 50% of the Reg2−/− mice were alive but with an increased sensitivity to Fas-induced oxidative stress and a decreased level of Bcl-xL. In contrast, HIP/PAP transgenic mice were resistant to Fas, with HIP/PAP serving as a sulfhydryl buffer to slow down decreases in glutathione and Bcl-xL. In Reg2−/− mice, liver regeneration was markedly impaired, with 29% mortality and delay of the S-phase and the activation of ERK1/2 and AKT. Activation of STAT3 began on time at 3 hours but persisted strongly up to 72 hours despite significant accumulation of SOCS3. Thus, Reg2 deficiency induced exaggerated IL-6/STAT-3 activation and mito-inhibition. Because the Reg2 gene was activated between 6 and 24 hours after hepatectomy in wild-type mice, Reg2 could mediate the TNF-α/IL-6 priming signaling by exerting a negative feed-back on STAT3/IL-6 activation to allow the hepatocytes to progress through the cell cycle. In conclusion, Reg2 deficiency enhanced liver sensitivity to Fas-induced oxidative stress and delayed liver regeneration with persistent TNF-α/IL6/STAT3 signaling. In contrast, overexpression of human HIP/PAP promoted liver resistance to Fas and accelerated liver regeneration with early activation/deactivation of STAT3. Reg2/HIP/PAP is therefore a critical mitogenic and antiapoptotic factor for the liver. (HEPATOLOGY 2006;44:1452–1464.)

Fas/FasL (APO-1, CD95) interaction is involved in many pathological situations such as graft-versus-host disease, liver transplant and major surgical resection, hepatitis B and C, acute alcoholic hepatitis, drug overdose, and some biliary diseases. Fas activation–induced acute liver failure is characterized by massive hepatic necrosis/apoptosis. Survival of patients depends on the speed with which quiescent liver cells re-enter the cell cycle and proliferate to compensate the functional loss.1, 2 Fas receptor–induced apoptosis of Fas-bearing cells includes hepatocytes. Injection of mice with Fas-specific antibodies (J0-2) leads to fulminant hepatitis and death. Oxidative damage plays a prominent role in hepatic apoptosis mediated via caspase activation.3 Partial hepatectomy was reported to prevent hepatocyte apoptosis in response to Fas engagement, and alternatively Fas engagement accelerated liver regeneration after hepatectomy. These results suggested that factors triggered by the growth response protected mice against the lethal effects of Fas.4 Indeed, hepatocyte growth factor and cytokines (interleukin-6 [1L-6]) promoted hepatic survival by stimulating liver regeneration and provided hepatoprotection in a variety of liver injury models, including Fas.5–8 Insulin-like growth factor binding protein-1,9, 10 amphiregulin,11, 12 and heparin-binding epidermal growth factor13 also displayed potent hepatoprotective or mitogenic effects. Thus, the identification of critical factors for in vivo antiapoptotic and mitogenic signaling pathways is of clinical interest.

Reg2,14 also called RegIIIβ, is the mouse homologue of human hepatocarcinoma-intestine-pancreas/pancreatic-associated protein (HIP/PAP), whose cDNA has been cloned in different species under different names: human HIP/PAP or Reg3A15, 16; rat Reg2, PAPI, or peptide 23; mouse Reg2/RegIIIβ; and hamster islet neogenesis associated protein or INGAP, respectively.17 These 16-kDa secreted proteins, lithostatine/RegIA and the recently characterized RegIV proteins,18 are classified in the group of the Reg family of the C-type lectin gene superfamily19 and present one common cation-dependent carbohydrate recognition domain of 115-130 amino acid residues that is linked to a signal peptide.20 We discovered the human HIP/PAP gene by differential screening of a hepatocarcinoma15 and demonstrated that HIP/PAP promoted hepatocyte adhesion in primary cultures and interacted with the proteins of the extracellular matrix.21 Rat PAPI also promoted the growth of epithelial intestinal cells.22 In neurons, rat Reg2/HIP/PAPI prevented cell death in primary motor neuron cultures23 and stimulated mitogenesis in Schwann cells and nerve regeneration.24 Hamster INGAP improved nerve function and enhanced regeneration in streptozotocin-induced diabetic C57BL/6 mice.17 In human pancreatitis, HIP/PAP functioned as an acute-phase reactant,16 and HIP/PAPI protected pancreatic cells against apoptosis induced by reactive oxygen species.25, 26 In the liver, we showed that HIP/PAP was not expressed in normal hepatocytes but frequently was overexpressed in tumors and regenerative hepatic/ductular diseases.27 These results suggested that HIP/PAP could regulate both viability and proliferation in hepatocytes. We have shown that HIP/PAP exhibited mitogenic and antiapoptotic properties for hepatocytes in primary cultures. We have shown in vivo that the human HIP/PAP C-type lectin was a novel hepatic growth factor that acted through paracrine dual mitogenic and antiapoptotic properties. Indeed, after a single injection, HIP/PAP markedly stimulated liver regeneration and protected against acetaminophen-induced acute liver failure.28

However, the endogenous role of Reg2 in the mouse liver is still unknown, and we generated Reg2−/− mice by deletion of the Reg2 gene.14 Two models relevant to human liver failure resulting from acute clinical conditions and surgery were investigated: fulminant hepatitis generated by J0-2 injection (Fas agonist) and liver regeneration induced by 70% partial hepatectomy. We found that Reg2 deficiency increased liver sensitivity to Fas-induced oxidative stress and led to a deleterious decay in the level of Bcl-xL. In contrast, human HIP/PAP-expressing mice were resistant to Fas-induced oxidative stress and preserved the Bcl-xL level. We also found that Reg2 was required for the normal regenerative response post-hepatectomy. Reg2 deficiency induced 29% mortality, delayes DNA synthesis, in accumulation of cyclin A and p21, and phosphorylation of ERK1/2 and protein kinase B (AKT). We have shown IL-6/signal transducer activator transcription factor 3 (STAT3) signaling was strongly persistent, inducing a persistent inflammatory response in Reg2−/− mice, even though there was accumulation of the suppressor of cytokine signaling 3 (SOCS3). Since we have shown that the Reg2 gene was activated in the liver between 6 and 24 hours post-hepatectomy and that liver regeneration was accelerated in human HIP/PAP-expressing mice, we propose that Reg2 may act as a relay between the tumor necrosis factor alpha (TNF-α)/IL-6 signaling pathways and growth factors in the regenerating liver.


HIP/PAP, hepatocarcinoma-intestine-pancreas/pancreatic-associated protein; BrdU, bromodesoxyuridine; STAT3, signal transducer activator transcription factor 3; SOCS3, suppressor of cytokine signaling 3; IL-6, interleukin-6; TNF-α, tumor necrosis factor alpha; MAPK/ERK, mitogen-activated protein kinase/extracellular-signal-regulated kinase; AKT, protein kinase B; ALT, alanine aminotransferase; AST, aspartate aminotransferase; GSH, reduced glutathione; GSSG, oxidized glutathione; Ref-1, redox-related protein 1.

Materials and Methods


Animals 8-10 weeks of age (20-22 g) were used after habituation for one week to their environment. All studies were performed according to the American National Institute of Health guidelines for animal care. Mice were anesthetized with isofluorane. The HIP/PAP transgenic mice, which were previously described,29 were backcrossed in a C57BL/6J (Janvier, L'Arbresles, France) genetic background. The C57BL/6J mice were termed Reg2+/+2 and were used as wild-type controls of the HIP/PAP transgenic mice.

Generation of Reg2-Deficient Mice by Gene Targeting.

To obtain a Reg2 genomic DNA clone, a mouse λ 2001 129/O1a library was screened with the mouse cDNA as probe. A 7.7-Kb Xhol fragment containing all the Reg2 exons was subcloned into pBluescript and used as a backbone for making the targeting vector. A region of sequence from a position within exon 2 extending to a position in exon 5 was replaced by the reporter/selection cassette IRES-Tau-LacZ/loxP/MC1neopA/loxP, thus deleting entirely exons 3 and 4. The reporter/selection cassette, a modified version to that previously described,30 was designed to permit expression of a Tau-LacZ fusion protein via the IRES element after correct targeting, but also contains a neomycin resistance gene (MC1neopA) independently expressed from its own constitutively active promoter. Two copies of an HSVtk gene cassette (Mc1tk) were added to the end of the left homology arm for negative selection.31 Linearized targeting vector was electroporated into E14-Tg2a (129/O1a strain) ES cells and the cells selected in G418 and ganciclovir. Resistant clones were picked into 96 well plates for freezing and genomic DNA preparation. The targeted clones were identified by Southern blotting of genomic DNA after separate restriction digestion with BamHI and EcoRV and hybridization with flanking 5′ and 3′ probes respectively. Figure 1 shows a graphical representation of homologous recombination of the targeting construction into the Reg2 locus with the predicted sizes of the relevant restriction fragments detected with the flanking probes before and after targeting. Targeting efficiency was 4.9%. Two targeted ES cell clones were injected into C57BL/6 blastocysts and chimeras identified by coat color. One male chimera from each clone gave germ line transmission after test-crossing with C57BL/6 females. Reg2 +/−mice were subsequently intercrossed to generate Reg2−/− mice, which were thus on a C57BL/6 × 129/O1a hybrid background. Reg2+/+ wild type littermates from the intercrosses (henceforth referred to as Reg2+/+ 1 wild-type) were used as controls in the analyses. Genotyping was performed by tail DNA biopsies followed by PCR analysis with primer pairs (Fig. 1A) specific for the Reg2 wild-type gene and for the neomycin resistance (neo) gene: Reg2 sense 5′-gtc ctc cat ggt gaa gag aac-3′ and anti-sense 5′-att ccc atc cac ctc cat tg-3′ (600bp PCR product); neol 5′-aga ggc tat tcg gct atg act-3′ and neo2 5′-cct gat cga caa gac cgg ctt-3′ (400 bp PCR product). Reg2 +/+ mice were therefore identified by the presence of a 600 bp product only, Reg2 −/+ mice by both 600 bp and 400 bp products, and Reg2 −/− mice by a 400 bp product only (Fig. 1B). Complete knockout of the gene was also confirmed in nervous tissue with both immunocytochemistry using Reg2 antibodies: absence of Reg2 mRNA was confirmed using real-time PCR in brain and regenerative liver (data not shown).

Figure 1.

(A) Target disruption of the murine Reg2 gene. Graphical representation of homologous recombination of the targeting construct with the Reg2 locus. The vector deletes a region between positions within exons 2 and 5 (indicated by thin dotted lines between vector and locus) with replacement by an IRES-Tau-LacZ-loxP-MC1neopA-loxP selection/reporter cassette, in which the neo gene is expressed from its own enhancer/promoter. The vector contained two copies of a tk negative selection marker (MC1tk cassette) on the end of one homology arm to enrich for homologous integrations. The positions of BamHI and EcoRV restriction sites are shown, and the sizes of their relevant fragments that are detected with 5′ and 3′ flanking probes and diagnostic of correct targeted vector integration in ES cells. PCR primer pair positions used for genotyping are also shown with size of PCR product. (B) PCR results for confirmation of knockout (−/−), wild-type Reg2 +/+1 and heterozygous (+/−) Reg2 mice as described in the text using oligonucleotide primer pairs specific to the neo gene (giving a 400 bp product) and the exon 4 to 5 interval in the deleted region (giving a 600 bp product). The control is a no DNA sample using H2O.

J0-2 Intoxication.

Fas intoxication was achieved by intraperitoneal injection of 0.5 μg/g or 0.3 μg/g of the Fas agonist mAb J0-2 (Pharmingen, San Diego, CA). Survival curves were established with 0.5 μg/g J0-2. For experimental analysis, the mice were sacrificed 2 or 4 hours after 0.5 μg/g J0-2 (just before the Reg2−/− mice began to die) or 12 hours after 0.3 μg/g J0-2. Blood was collected for measurement of serum aminotransferase activity. Alanine aminotransferase (ALT) and aspartate aminotransferase (AST) were quantified using a standard clinical automatic analyzer (Hitachi, Roche, Meylan, France). For protein analyses, livers were frozen in liquid nitrogen or fixed overnight at 4°C in a 4% acetic formaldehyde solution, then dehydrated and embedded in paraffin. Sections (4 μm) of liver tissue were stained with hematoxylin/eosin and red picrosirius for histological examination.

Partial Hepatectomy.

Two-thirds partial hepatectomy (70%) and overall liver regeneration were evaluated by recovery of liver mass (ratio of liver/body weight expressed as percentage of the initial ratio in quiescent liver), and incorporation of bromodesoxyuridine (BrdU) was performed as previously described.29


Immunochemistry was performed using anti-HIP/PAP antibodies as previously described.29

Western Blot Analysis.

Western blot analysis from whole regenerative liver and nuclear extracts was performed as previously described.29 Primary antibodies were diluted according to the instructions of the manufacturers: total and phospho–(Ser 473) AKT, total and phospho-p44/42 Map-ERK kinases, Bcl-xL, and total and phospho-STAT3 from Cell Signaling Technology (Beverly, MA); and cyclin A (c-19 sc-596), actin (1-19 sc-1616), SOCS-3 (M-20 sc-7009), p21 (F-5 sc-6246), and Ref-1 (sc-9919) from Santa Cruz Biotechnology Inc. (Santa Cruz, CA). Protein concentration was measured using the Bovin Serum Albumin microbiuret assay (Pierce, Bezons, France).

STAT3 Transcription Factor Activation.

STAT3 transcription factor activation on nuclear extracts was assessed with a TramsAM kit (Active Motif, Rixensart, Belgium) according to the manufacturer's instructions.

SybrGreen Real-Time Quantitative Polymerase Chain Reaction.

Total RNA from liver and intestine was extracted using TRIzol Reagent (Invitrogen). Two micrograms of total RNA was reverse-transcribed and analyzed by real-time quantitative PCR using the Light Cycler Fast Start DNA master SybrGreen I (Roche Diagnostic, Meylan, France), and the following primers: actin housekeeping gene: actin sense—5′-tga gag gga aat cgt gcg tga c-3′, actin antisense—5′-tca tgg atg cca cag gat tcc-3′ Reg2 gene, Reg2 sense (104S)—5′-cgc ccc gga ttc atg ctg cct cca aca gcc tgc-3′, Reg2 antisense (105S)—5′-cgc aag ctt tta acc agt aaa ttt gca gac ata c-3′; IL-6 sense—5′-ctt cca tcc agt tgc ctt ctt g-3′, IL-6 antisense—5′-ggt agc atc cat cat ttc ttt gta-3′; TNF-α sense—5′-cct cca gaa aag aca cca-3′, TNF-α antisense—5′-ctt ggt ggt ttg cta cga-3′.

To monitor specificity, final PCR products were analyzed by melting curves and electrophoresis. The absolute amount of Reg2β transcripts was calculated from serial dilutions of intestine random-primed cDNA used as established standard curves, as there was no expression of Reg2β in quiescent liver. IL-6 and TNF-α mRNA were quantified by the 2−ΔΔ method.32 Changes in gene expression are equal to 2−ΔΔCt, where Ct is the threshold cycle and ΔΔCt is (Ctarget − Cactin)time, mice− (Ctarget− Cactin)0h,Reg2+/+.

Caspase-3 Activity.

Caspase-3 activity was measured in liver cytosolic fractions as previously described.33 Results are expressed as pmol · min−1 mg−1 protein.

Determination of GSH and GSSG in Liver Extracts.

Liver tissues were homogenized in 10 mmol/L potassium phosphate, pH 7.8, containing 1 mmol/L EDTA (100 mg/mL w/v) with an Ultra-Turax homogenizer (Janken Kunkel Ika-Werk, Staufen, Germany) and sonicated 30 seconds in ice. Then 10 μL of homogenate was mixed with 390 μL of 10 mmol/L phosphate buffer saline, pH 2.7, and 10 μL of 5% metaphosphoric acid for deproteinization. Reduced glutathione (GSH) and oxidized glutathione (GSSG) were simultaneously determined by high-performance liquid chromatography with electrochemical detection as previously described.34 Briefly, 180 μL of homogenate supernatant was mixed with 20 μL of 12 μmol/L N-acetylcysteine as the internal standard. Separation was performed on a Uptisphere column (ODB, 250 × 4.6 mm, 5 μmol/L; Interchim, Montluçon, France) using 10 mmol/L phosphate buffer saline solution, pH 2.7, containing 2% methanol and delivered at a flow rate of 1 mL/min. Analytical cell potentials were set at 400 mV for E1 and 800 mV for E2. The retention times of GSH, GSSG, and N-acetylcysteine were 6.5, 17, and 15 min, respectively. Total glutathione was calculated as the sum of GSH and GSSG concentrations, the latter expressed as GSH equivalents (1 mole of GSSG corresponds to 2 GSH equivalents).

Statistical Analysis.

The statistical significance of variation in liver weight, incorporation of BrdU, mitotic index, and quantification by densitometry found with RT-PCR was determined by the paired t test or the Mann-Whitney U test. The statistical significance of differences in survival rate following J0-2 intoxication was calculated by the Kaplan-Meier method (SigmaStat 3.1 Systal Software Inc., Les Ulis, France). Quantitative variation was considered significant at P less than .05.


Generation of Reg2−/− Mice.

Two models of transgenic mice were used in the present study. The first model expresses the human HIP/PAP gene under the control of the albumin regulatory region for liver expression (HIP/PAP mice) and has been previously described.29 The second is the Reg2-deficient mouse, in which the endogenous mouse Reg2 gene, homologous to the human HIP/PAP gene, has been inactivated. These mice were homozygous for a targeted disruption of the Reg2 gene locus created in ES cells, in which a region from exons 2 to 5 was deleted and replaced by an IRES-Tau-LacZ-loxP/MC1neopA/loxP reporter/selection cassette thus resulting in a null allele. Figure 1A shows a graphical representation of homologous recombination of the linearized targeting construct into the Reg2 locus of E14-TG2a ES cells and the Southern blot strategy used to identify targeted clones. Targeted ES cell clones were identified at a frequency of 4.9%, and two of these clones were subsequently injected into C57/BL6 blastocysts to generate chimeras. One chimera from each clone gave germ line transmission of the targeted allele in test crosses with C57 B1/6 mice. Test-cross offspring heterozygous for the targeted allele (Reg2 +/−) were then inter-crossed to generate homozygous (Reg2−/−) mice, identified by PCR analysis (Fig. 1B). The absence of Reg2 mRNA in the regenerative livers and neonatal brains of the Reg2−/− mice was confirmed by RT-PCR and immunohistochemistry (data not shown). Mice homozygous for the disrupted Reg2 null allele were phenotypically indistinguishable from their wild-type or heterozygous littermates. No embryonic lethality or significant developmental defects were observed in the Reg2−/− animals. The adult Reg2−/− mice of both sexes were fertile, and litter size was normal. Liver/animal weight ratios were normal, and no spontaneous liver cancer was detected during the life span.

Sensitivity of Reg2−/− and HIP/PAP Mice to J0-2-Induced Fulminant Hepatitis Was Mediated by Resistance to Oxidative Stress.

After fulminant hepatitis induced by J0-2, survival of the Reg2−/− and HIP/PAP transgenic mice was compared to that of the wild-type Reg2+/+1 and Reg2+/+2 mice, respectively (Fig. 2A). Indeed, differences in the genetic background of the control mice induced differences in their susceptibility to Fas, with the C57/B6 Reg2+/+2 mice being highly sensitive.35 After injection of J0-2 0.5 μg/g, 79% of HIP/PAP transgenic mice survived versus 30% of the respective wild-type Reg2+/+2 mice (P < .03). For the same dose, only 12.5% of the Reg2−/− mice survived versus 84% of their wild-type Reg2+/+ mice (P < .03). Four hours after the injection, only 50% of the Reg2−/− mice survived versus 93% of the Reg2+/+1 mice. Serum levels of ALT and AST were higher in the Reg2−/− mice than in the Reg2+/+ mice, reflecting increased Fas hepatotoxicity in the Reg2−/− mice. As early as 2 hours after injection, ALT values were 35 ± 18 versus 20 ± 10 UI/L, and AST values were 170 ± 28 versus 52 ± 3 UI/L (P < .001; n = 4) in Reg2−/− and Reg2+/+1, respectively. After 4 hours, ALT values were 677 ± 413 versus 45 ± 40 UI/L, and AST values were 2,717 ± 1,718 versus 142 ± 120 UI/L in Reg2−/− and Reg2+/+1, respectively. Their livers turned dark red, and histological examination revealed massive hepatic hemorrhage with many apoptotic and necrotic cells. Moreover caspase-3 activity was statistically higher in the Reg2−/− mice than in the Reg2+/+1 mice (434 ± 72 vs. 160 ± 60 pmol · min−1 mg−1 protein, respectively; P < .01, after 4 hours with 0.5 μg/g J0-2). Caspase-3 activity was also lower in HIP/PAP mice than in the Reg2+/+2 mice (122 ± 80 vs. 476 ± 122 pmol · min−1 mg−1 protein; P < .01). Twelve hours after a lower dose of J0-2 (0.3 μg/g), only 55% of the Reg2−/− mice were still alive versus 100% of the Reg2+/+1 mice and 100% of the HIP/PAP transgenic mice versus 75% of the Reg2+/+2 mice. ALT and AST values dramatically rose in the Reg2−/− mice: ALT values were 10,056 ± 4,258 versus 2,305 ± 1,881 UI/L, and AST values were 18,615 ± 4,466 versus 2,504 ± 2112 UI/L in the Reg2−/− and Reg2+/+ mice, respectively (P = .016, n = 4). Caspase-3 activity was 172 ± 38, 108 ± 36, 39 ± 11, and 72 ± 29 pmol · min−1 mg−1 protein in the Reg2 −/−, Reg2+/+1, HIP/PAP, and Reg2+/+2 mice, respectively (Fig. 2B). Collectively, the results showed that the Reg2 −/− mice were more sensitive to J0-2 and died earlier than the Reg2+/+1 mice, whereas the HIP/PAP transgenic mice were more resistant to J0-2 intoxication than were the Reg2+/+2 mice. We then investigated the level of the antiapoptotic protein Bcl-xL in liver tissue after up to 4 hours of J0-2 treatment because the Reg2−/− mice had rapidly died after 4 hours (Fig. 3). We observed a dramatic, 50% decrease of Bcl-xL in the Reg2−/− mice as early as 2 hours after treatment (P < .05). In contrast, Bcl-xL increased by 50% in the Reg2+/+1 mice 4 hours after J0-2 treatment and was significantly higher than that in the Reg2−/− mice (P < .05). Because Fas-induced oxidative stress could control Bcl-xL level, we investigated whether the Reg2−/− mice were more sensitive to the generation of reactive oxygen species than were the Reg2+/+1 mice. We analyzed levels of GSH, GSSG, and redox-related protein 1 (Ref-1), which suppresses reactive oxygen species generation in mouse liver.6 The basal ratio of GSH/total glutathione was lower in the Reg2−/− mice than in the Reg2+/+1 mice and resulted from the level of GSSG being higher (1.5 ± 0.7 μmol/g) in the Reg2−/− mice than in the Reg2+/+1 mice (0.78 ± 0.2 μmol/g, P < .01). After J0-2 treatment, GSH and GSSG levels in the Reg2−/− mice significantly decreased (P < .05), whereas those in the control Reg2+/+1 mice were not modified up to 4 hours after treatment. In basal conditions Ref-1 protein level was lower in the Reg2−/− mice than in the Reg2+/+1 mice (P < .03), and the level did not increase after 4 hours in the Reg2−/− mice. Thus, a high level of GSSG and a low level of Ref-1 reflected the high sensitivity of the Reg2−/− mice to oxidative stress (Fig. 3A). In HIP/PAP transgenic mice, Bcl-xL level increased by 30% as early as 2 hours after treatment and was sustained to 4 hours, whereas decreases of 30% and 50% were observed in the Reg2+/+2 mice 2 and 4 hours after J0-2 injection, respectively. According to our previous study of acetaminophen intoxication,28 basal GSH and GSSG levels were similar in the HIP/PAP transgenic and Reg2+/+2 mice. After JO-2 injection, GSH and total glutathione were higher in the transgenic HIP/PAP mice than in the Reg2+/+2 mice (P < .05 after 2 hours). No difference was detected in Ref-1 protein level. Thus, Reg2/HIP/PAP could play a role in the balance between GSSG and GSH, protecting Bcl-xL protein level during Fas apoptosis. Survival is associated with cellular proliferation, which led us to compare the regeneration induced by a two-thirds hepatectomy in the Reg2−/−, HIP/PAP transgenic, and control mice (Fig. 3B).

Figure 2.

Different sensitivities of Reg2−/− and HIP/PAP mice and their wild-type Reg2+/+ counterparts to treatment with the Fas agonist J0-2. (A) Survival curves and hematoxylin-eosin stained livers microscopic views (25-fold magnification) of Reg2+/+1 (a, e, i) and Reg2−/− (b, f, j). Arrows indicate features of apoptotic bodies organized in “rosettes,” characteristic of Fas apoptosis. Arrowheads indicate infiltrated blood in hepatic tissue of Reg2−/− mice; in HIP/PAP (c, g, k) and Reg2+/+2 (d, h, l) mice. Mice were injected with 0.5 μg/g (a-h) or 0.3 μg/g (i-l) of J0-2. (B) Survival of the different strains of mice was analyzed, and caspase-3 activity (pmol · min−1 mg−1 protein of liver extract) and ALT and AST (UI/L) were measured after injection of 0.5 or 0.3 μg/g of J0-2 (*paired values statistically different at P < .03, * *paired values statistically different at P < .01).

Figure 3.

Sensitivity of Reg2−/− and HIP/PAP transgenic mice to J0-2-induced fulminant hepatitis was mediated by sensitivity to oxidative stress. Bcl-xL and Ref-1 protein levels and GSH and GSSG contents were compared in liver extracts of (A) Reg2−/− versus Reg2+/+1 mice and (B) HIP/PAP versus Reg2+/+2 mice before and after J0-2 injection (0.5 μg/g): a, GSSG higher in Reg2−/− mice than in Reg2+/+1 mice in basal Reg2+/+1 conditions (*P < .01); b, GSH lower in Reg2−/− mice than in Reg2+/+ mice 4 hours after J0-2 injection (*P < .01); c, GSH higher in HIP/PAP transgenic mice than in Reg2+/+2 mice 2 hours after J0-2 injection (*P < .04). After J0-2 injection, total glutathione decreased in Reg2−/− mice but not in Reg2+/+1 mice and was higher in HIP/PAP transgenic than in Reg2+/+2 mice (*at least P < .05, n = 4 mice).

Increased Mortality and Delayed Liver Mass Restoration in Reg2−/− Mice Post-hepatectomy.

We compared the Reg2−/− and HIP/PAP transgenic mice and their respective wild-type counterparts, Reg2+/+1 and Reg2+/+2 mice. Because similar results were obtained with both wild-type mice, liver regeneration data for only one of them are presented. Partial hepatectomy induced less than 5% and 8% mortality in the HIP/PAP and Reg2+/+ mice, respectively. In contrast, a high number (29%) of the Reg2−/− mice died around 36 hours (P = .012, Fig. 4A). Although the initial liver/body ratios were similar in the Reg2−/−, HIP/PAP, and Reg2+/+ mice (data not shown), liver mass recovery did not start until 24 hours post-surgery in the surviving Reg2−/− mice in contrast to the other strains. After a delay of 24 hours, mass recovery began to increase, but the weight curve was always statistically under the wild-type curve from 24 to 120 hours posthepatectomy (Fig. 4B). Thus, the results indicated a defect in an early step of the regenerative process in the Reg2−/− mice posthepatectomy.

Figure 4.

Increased mortality and delayed restoration of liver mass in Reg2−/− mice posthepatectomy. (A) Survival curves of Reg2−/− (▵), Reg2+/+ (□), and HIP/PAP (●) mice. Reg2−/− died close to 36 hours postsurgery (P = .012). (B) In the surviving Reg2−/− mice, liver mass recovery was assessed by calculating the liver/body weight ratio expressed as a percentage of the initial ratio in quiescent liver. There was no difference in the initial ratio between Reg2−/−, HIP/PAP, and Reg2+/+ mice. The average percentage recovery of normal liver weight at various times posthepatectomy showed that mass recovery in Reg2−/− mice was delayed 24 hours compared to that in the other strains. The weight curve was always statistically under the Reg2+/+ curve from 24 to 120 hours posthepatectomy (*P = .023, * *P < .004, ***P < .001; n = 5-7 mice at each time).

Blunted DNA Synthetic Response Following Two-Thirds Partial Hepatectomy in Reg2−/−-Deficient Mice.

The same liver slide was used to determine the number of S-phase cells after BrdU injection 2 hours before sacrifice and to estimate the mitotic index. We found that no BrdU incorporation occurred before and 48 hours after surgery, consistent with the cells being in the quiescent (G0) stage in all the various liver types. After 48 hours, BrdU incorporation was significantly higher in the HIP/PAP mice than in the wild-type and Reg2−/− mice (P < .05), and there was no difference between the wild-type and Reg2−/− mice. For HIP/PAP transgenic and wild-type mice, maximal DNA synthesis was observed after 48 hours, indicating the S-phase traverse was restricted to 12 hours. In the Reg2−/− mice, BrdU incorporation was maximal from 48 to 60 hours after surgery, indicating the S-phase traverse was extended to 24 hours in these mice. The absence of liver mass recovery during the first 24 hours probably induced a compensatory extension of the S-phase in the surviving Reg2−/− mice. The number of mitoses was also markedly lower in the Reg2−/− mice than in the other strains after 60 hours (P < .001). This result suggested that the regenerative process was desynchronized in the Reg2−/− mice. To confirm that hepatocyte DNA synthesis and mitotic index were delayed in the Reg2 −/− livers, reflecting impaired progression of hepatocytes across the G1/S transition, we measured the steady-state levels of several cell-cycle-associated gene products (Fig. 5B). Cyclin A is a reliable marker of the S-phase entry, and its expression is inhibited by p21. In the Reg2+/+ mice, cyclin A peaked and p21 was downregulated 60 hours after partial hepatectomy. In contrast, in the Reg2−/− mice, cyclin A peaked only at 72 hours, when the p21 level became low. In the HIP/PAP transgenic mice, cyclin A peaked 48 hours postsurgery, earlier than in the other strains. The P21 level in the HIP/PAP transgenic mice was low at surgery and upregulated after 36 hours. Thus, the regeneration process was delayed in most of the surviving Reg2−/− mice, whereas it was accelerated in the HIP/PAP transgenic mice. To further investigate the molecular mechanism responsible for this delay in the Reg2−/− mice, signaling pathways known to be regulated during liver regeneration were investigated in the Reg2−/−, Reg2+/+, and HIP/PAP transgenic mice.

Figure 5.

Blunted DNA synthetic response following two-thirds partial hepatectomy in Reg2−/− mice. (A) Immunohistochemical detection of S-phase and mitotic hepatocytes in liver sections of Reg2−/− (▴), HIP/PAP (●), and wild-type mice (□) are indicated as percentages. Statistically different paired values were compared to those of Reg2+/+ mice with at least *P < .04. (B) Western blot analysis of hepatic cyclin A and p21 levels in regenerative whole-liver extracts prepared from of Reg2−/−, Reg2+/+, and HIP/PAP mice during the time course of liver regeneration. Cyclin A and p21/actin ratios were calculated from pools of 3 mice at each time and represent 2 independent experiments.

Delays of ERK and AKT Signaling Pathways and Persistence of STAT3 Activation in Reg2−/− Mice During Liver Regeneration.

Activation of the mitogen-activated protein kinase/extracellular-signal-regulated kinase (MAPK/ERK) cascade is a key signaling pathway involved in the regulation of the progression of the G1 phase during liver regeneration.36 Crosstalk is known to occur between HIP/PAP and the MAPK pathway in the rat pancreas,25 and we therefore investigated ERK1 (p44 MAPK) and ERK2 (p42 MAPK) activation in the Reg2−/−, HIP/PAP transgenic, and Reg2+/+ mice (Fig. 6B). Total levels of ERK1 and ERK2 proteins were similar in all the strains. We observed a delay in the accumulation of phospho-ERK1 and ERK2, as accumulation peaked after 48 hours in the Reg2−/− mice versus after 24 and 12-24 hours in the Reg2+/+ and HIP/PAP transgenic mice, respectively. Reg2 has been reported to stimulate AKT phosphorylation in rat motoneurons.23 In the regenerating livers, maximal phospho-AKT was delayed from 6 to 12 hours in the Reg2+/+ mice versus in the Reg2−/− mice. In HIP/PAP-expressing regenerative liver, phospho-AKT peaked after 6 hours, but the levels were higher than that in the Reg2+/+ mice 1 and 3 hours posthepatectomy (Fig. 6B). Overall, Reg2/HIP/PAP mediated ERK and AKT activation during liver regeneration.

Figure 6.

Persistence of STAT3 and SOCS3 activation and delay in the activation of ERK1/2 and AKT signaling cascades in regenerating livers of Reg2−/−, HIP/PAP transgenic, and Reg2+/+ mice. (A) STAT3 activation was monitored in liver nuclear extracts of Reg2−/− (▵), Reg2+/+ (□), and HIP/PAP (●) mice using the TramsAM™ kit during the indicated regeneration time course. The results are expressed as percentages of the positive control provided in the kit (nuclear extract prepared from HepG2 cells treated by IL-6) and correspond to means obtained with 4 mice (*P < .01). Phospho-STAT3 antibodies were used in a Western blot analysis on liver nuclear extracts, and normalization was visualized by nonspecific coloration of the membrane with Red Ponceau dye. SOCS3 antibodies were used in a Western blot analysis of total liver extracts, and normalization was visualized by actin hybridization. Quantification of IL-6 and TNF-α mRNA levels before and 24 hours after hepatectomy by quantitative real-time PCR (* *P < .05, n = 3). (B) Phospho-ERK1/2 and phospho-AKT antibodies were used in a Western blot analysis of total liver extracts, and normalization was visualized by total ERK1/2 or total AKT antibody hybridization. Activation of ERK1/2 and AKT was delayed in Reg2−/− mice versus in Reg2+/+ and HIP/PAP mice.

We also investigated phospho-STAT3 DNA-binding level by ELISA and nuclear accumulation of phospho-STAT3 by Western blot analysis (Fig. 6A). We previously reported earlier nuclear phospho-STAT3 activation/deactivation in HIP/PAP transgenic mice than in Reg2+/+ mice.28 Although STAT3 phosphorylation appeared 3 hours postsurgery in the Reg2−/− and Reg2+/+ mice, a strong and obvious persistence of nuclear STAT3 activation was observed up to 72 hours posthepatectomy in the Reg2−/− mice. In contrast, STAT3 deactivation was complete 12 and 24 hours post-hepatectomy in the HIP/PAP transgenic and Reg2+/+ mice, respectively. SOCS3, a known downregulator of STAT3 activation, was also strongly induced in the Reg2−/− mice from 3 to 72 hours posthepatectomy (Fig. 6A). SOCS3 level was lower in the HIP/PAP transgenic and Reg2+/+ mice than in the Reg2−/− mice, and its expression followed STAT3 activation/deactivation. Overall, data indicated that SOCS3 was not able to repress STAT3 activation in the Reg2−/− mice and that Reg2 should act downstream of the priming STAT3/IL-6 signaling. To investigate how overactivation of STAT3 might contribute to delayed liver regeneration, we quantified IL6 and TNF-α mRNA levels before and 24 hours after hepatectomy (Fig. 6A). IL-6 and TNF-α mRNA levels were higher in the livers of the Reg2−/− mice than in the control and HIP/PAP transgenic mice 24 hours after hepatectomy, when STAT3 was still activated in the Reg2−/− mice and deactivated in the HIP/PAP and control mice. IL-6 secretion is stimulated by TNF-α and is a major signal for the stimulation of acute-phase protein synthesis by hepatocytes. However, high levels of IL6 can promote deleterious inflammation and delay liver regeneration.37 Thus, the persistence of TNF-α/IL-6/STAT3 signaling detected in the Reg2−/− mice might have induced a deleterious and persistently inflammatory situation that probably led to the mortality observed around 36 hours and delayed the liver regeneration in the surviving mice. To investigate the role of Reg2 in persistent STAT3 activation, we then examined the activation of the Reg2 gene during liver regeneration.

Activation of Reg2 Gene During Liver Mouse Regeneration Posthepatectomy in Reg2+/+ Mice.

Reg2 gene expression was investigated with real-time PCR during the course of liver regeneration in the Reg2+/+ mice. The Reg2 gene was not expressed in quiescent liver; its expression began 6 hours after hepatectomy. Thus, activation of the Reg2 gene started after STAT3 and ended after deactivation of STAT3 in the Reg2+/+ mice. This indicated that Reg2 as a plausible target of STAT3 was expressed between 6 and 24 hours after surgery (Fig. 7A). Immunohistochemical analysis with HIP/PAP antibodies revealed hepatocytes that were positive for regeneration, which was predominantly localized around the portal tracts (Fig. 7B).

Figure 7.

Reg2 expression in the liver of Reg2+/+ mice during the time course of regeneration posthepatectomy. (A) Expression of Reg2 transcripts shown relative to expression of control gene, β-actin. No expression was detected in quiescent liver. (B) Immunohistochemical staining of liver sections with anti-HIP/PAP antibodies (25× original magnification) 12 hours postsurgery.

Collectively, we have demonstrated that liver regeneration was impaired in Reg2−/− mice, with mortality of 29% almost 36 hours posthepatectomy. The surviving Reg2−/− mice showed a 24-hour delay in liver mass restoration, DNA synthesis, and cycle gene accumulation. AKT and ERK1/2 were also delayed. STAT3 was activated on time but persisted abnormally after its induction, indicating a defect downstream of the IL-6/STAT3 signaling pathway.38 This was consistent with the critical role of Reg2 as a relay between the TNF-α/IL-6 signaling pathway and growth factors in the regenerative liver.


The identification of endogenous protective mechanisms triggered by liver injury is important not only for the study of the physiopathology of this organ but also for the design of more effective therapies to enhance natural defense responses. In the present study, we investigated the critical role of the mouse homologue Reg2 of the human HIP/PAP gene in fulminant hepatitis and in liver regeneration. We made use of Reg2−/− mice to gain further insight into the role of Reg2 during liver damage.

Sensitivity to J0-2-Induced Fulminant Hepatitis.

We have shown that the Reg2−/− mice were more sensitive and the HIP/PAP transgenic mice were more resistant to J0-2-mediated acute liver failure than their wild-type Reg2+/+ counterparts. Sensitivity to Fas-induced fulminant hepatitis depends on genetic background, with the C57Bl/6, Reg2+/+2 mice highly sensitive.35 However, survival indicated that the Reg2−/− mice were the most sensitive strain investigated in this study. We observed that caspase-3 activity was upregulated and that Bcl-xL level (expressed in hepatocytes) decreased dramatically in the Reg2 −/− mice following J0-2 injection, indicating apoptosis was accelerated in these mice. In contrast, caspase-3 was not activated and Bcl-xL level increased in the resistant HIP/PAP transgenic mice. FasL/Fas interaction induced the death of hepatocytes, predominantly by generation of reactive oxygen species and mitochondrial disruption.3 The primary role of mitochondria during Fas-mediated apoptosis has been demonstrated in Bcl-2 and Bcl-xL transgenic mice. Bcl-xL or Bcl-2 overexpression blocked programmed cell death by inhibiting cytochrome c release from mitochondria and prevented Fas-mediated apoptosis of hepatocytes.39–41 Therefore, a resistance to reactive oxygen species produced during Fas intoxication could likely control the Bcl-xL level and consequently the susceptibility of the Reg2−/− mice to Fas-induced apoptosis. We thus investigated whether the Reg2−/− mice were more sensitive to reactive oxygen species. High-performance liquid chromatography methodology allowed us to directly quantify the GSH and GSSG of each liver extract in the same run. The GSH/total glutathione ratio was low in the Reg2−/− mice, indicating these mice had basal susceptibility to oxidative stress. After J0-2 injection, GSH and GSSG levels quickly decreased in the Reg2−/− mice, whereas they were not modified in the control Reg2+/+1 mice, which were resistant to J0-2 treatment. The resistance of the HIP/PAP transgenic mice was also associated with the slowdown in change of the glutathione level. These results indicate that Reg2/HIP/PAP as a sulfhydryl buffer could play a role in the balance between GSSG and GSH, maintaining the level of glutathione in its reduced GSH state. These results are in accordance with those of our previous work, in which human HIP/PAP exhibited superoxide dismutase–like and glutathione reductase–like activities, which explained its protective effect against reactive oxygen species–reducing mitochondrial damage, subsequent cytochrome c release, and caspase-3 activation following acetaminophen overdose.28 We have also studied the redox-related protein Ref-1, which is transcriptionally up-regulated in response to oxidative stresses. Previous work showed that the adenoviral overexpression of Ref-1 in mouse liver successfully improved post–ischemic liver injury through suppression of generation of reactive oxygen species and apoptosis in hepatic tissue.42 Interestingly, Ref-1 was proposed to reduce apoptosis in liver following J0-2 injection.6 Ref-1 was always lowest in the Reg2−/− strain, which might have contributed to its high sensitivity to Fas-mediated apoptosis.

Experiments conducted in isolated mouse hepatocytes have shown that HIP/PAP directly stimulated PKA-dependent Bad phosphorylation, leading to inactivation of proapoptotic molecules.29 Reg2/HIP/PAP was able to stimulate the phosphorylation of AKT in rat motoneurons, thus contributing to AKT-dependent phosphorylation of Bad.23 We also detected a higher level of phosphorylated AKT in the regenerating livers of the HIP/PAP transgenic than in the wild-type mice. Bad phosphorylated by AKT and PKA bound 14-3-3 proteins and then did not bind Bcl-2 or Bcl-xL, thus exerting their antiapoptotic effects.43 We cannot exclude other mechanisms implicated in the sensitivity of the Reg2−/− mice such as direct activation of caspase 8 (FLICE) by Fas/FasL interaction.6 However, we did not observe differences in the activation of phospho-STAT3 or in the cleavage of FLICE and the level of its inhibitor FLIP after J0-2 treatment in all the types of mice investigated (data not shown). This may indicate that Reg2/HIP/PAP suppresses Fas-induced liver injury by mitochondrial redox-dependent mechanisms rather than by direct inhibition of FLICE cleavage.

Liver Regeneration.

Promotion of survival is often correlated with induction of cellular proliferation. After partial hepatectomy, quiescent liver cells are stimulated to reenter the cell cycle and proliferate to restore the original liver mass.2, 44 The biological significance of Reg2 in liver regeneration was investigated using the Reg2−/− mice. Although the mice demonstrated normal development, they showed delayed liver regeneration and had increased mortality (29%) after partial hepatectomy. Members of the Reg family were thus important in the regenerating liver. Because no deaths were detected after 36 hours, Reg2 may control an early critical step in the regenerative process. We have further characterized the deregulation of the proteins that control the cell-cycle progression in the Reg2−/− mice. DNA synthesis was delayed 12 hours in the Reg2−/− mice versus the Reg2+/+ and HIP/PAP transgenic mice. Moreover, the cell cycle could be desynchronized because maximal mitosis appeared 12 hours after maximal DNA synthesis in the Reg2−/− mice. Cyclin A accumulated 12 hours later in the Reg2−/− mice than in the control mice, whereas cyclin A accumulated 12 hours earlier in the HIP/PAP transgenic than in the control mice. Collectively, the HIP/PAP transgenic had advanced, whereas the Reg2−/− mice were late compared to the control mice. Thus, Reg2/HIP/PAP participates in the timing of DNA synthesis, allowing hepatocytes to progress through the cell cycle.

TNF-α and IL-6 are two known regulators of the initial phase of liver regeneration and prime hepatocytes to enter the G1 phase of the cell,38 when they become fully responsive to growth factors. Levels of these cytokines rise after partial hepatectomy, with the peak of circulating TNF-α preceding that of IL-6. However, the persistence of TNF-α/IL6 signaling pathways is deleterious for progression through the cell cycle.37, 45

We recently showed that HIP/PAP stimulated DNA synthesis in isolated hepatocytes and during liver regeneration, probably by accelerating the accumulation/degradation of nuclear phospho-STAT3 levels.28 We further investigated whether the deletion of Reg2 controlled the timing of activation/deactivation of STAT3. Activation of STAT3 in Reg2−/− and Reg2+/+ mice occurred after 3 hours, indicating Reg2 is downstream of the TNF → IL-6 signaling pathways. However, STAT3 activation clearly was strong and persistent up to 72 hours in Reg2−/− mice, whereas deactivation was complete after 12 and 24 hours in HIP/PAP and Reg2+/+ mice, respectively. IL-6 is known to induce STAT3 activation, and we found that TNF-α and IL-6 expression was higher in the Reg2−/− mice than in the other strains 24 hours posthepatectomy. SOCS3 directly interacted with the phosphorylated JAK kinases and furthermore inhibited the activation of STAT3 in the Reg2+/+ and HIP/PAP mice. Even though SOCS3 was highly activated in the Reg2−/− mice, SOCS3 was unable to repress STAT3/IL-6 signaling. In the Reg2+/+ mice, a marked upregulation of Reg2 mRNA was detected between 6 and 24 hours posthepatectomy, thus occurring after STAT3 activation. Indeed, the rat HIP/PAP-I or Reg2 promoter contains STAT3 binding sites, indicating that HIP/PAP/Reg2 gene expression may be upregulated by IL-6/STAT3 in the liver.46 The expression of Reg2 between 6 and 24 hours is consistent with it having a role as a relay between the TNF-α/IL-6 signaling and growth factors in the liver. Thus, Reg2 might regulate STAT3 activation through a negative feedback mechanism, allowing the hepatocytes in the Reg2+/+ mice to progress in the cell cycle and terminating the IL6 signaling. Interestingly, Reg2 as an antioxidant molecule could control excess reactive oxygen species generated by TNF-α posthepatectomy and thus might determine whether the effect of TNF-α on hepatocytes is mitogenic or apoptotic. The delay observed in the regenerative process of the Reg2−/− mice could be mediated by the persistent expression of TNF-α/IL-6 and the activation of STAT3, leading to a deleterious inflammatory context. Furthermore, STAT3 overexpression has been linked to p21 upregulation in the impaired regeneration of fatty livers.47, 48 Accordingly, we have noted a striking inverse correlation between nuclear accumulation of phospho-STAT3 and DNA synthesis/mitotic index. We have also noted that STAT3 activation was correlated with p21 level: deactivation of STAT3 and reaccumulation of p21 were only observed in the Reg2−/− mice after 72 hours. Thus, these results linked exaggerated STAT-3 activation with impaired hepatocyte proliferation, promoting mitotic inhibition by an upregulating mechanism that impedes cell-cycle progression.

MAPK/ERK1/2 and AKT are known to transmit mitogenic signals in the Ras pathway and are activated by phosphorylation in the regenerating liver under the control of Met/HGF.36 These signaling pathways are involved in regulating the progression of the G1 phase. We have shown that phosphorylation of ERK1/2 kinases and phosphorylation of AKT were both delayed in the Reg2−/− regenerative livers. Previously studies reported that Reg2 stimulated AKT in motoneurons23 and that ERK1/2 induced expression of Reg2/HIP/PAPI in the pancreas.25 In the present study, we have shown that Reg2 mediated ERK1/2 and AKT activation in regenerating liver. Liver-specific STAT3 knockout mice showed normal activation of the ERK1/2 pathway after partial hepatectomy, indicating that the effects of IL-6 on ERK1/2 are not mediated only by STAT3.45 Therefore, the analyses of signaling cascades in the Reg2−/− mice indicated an intricate interplay between Reg2 and other signaling systems during liver regeneration.

In conclusion, this study demonstrates the critical involvement of Reg2 in the regulation of antiapoptotic and mitotic signaling pathways in the liver. An appropriate level of Reg2 is required to protect against stress oxidative-inducing liver failure and to properly drive hepatocytes into the cell cycle during liver regeneration. Our results are consistent with Reg2 having a role downstream of STAT3 activation during liver regeneration. Interestingly, in conditional STAT3 knockout motoneurons, expression of Reg2 and Bcl-xL was reduced after axotomy.49 Thus, it should be interesting to investigate the regulation of the Reg2 gene in liver-specific STAT3 knockout mice after partial hepatectomy. Finally, these results point to the potential clinical value of the Reg2/HIP/PAP molecule in human liver failure that results from chronic and acute clinical conditions and surgery.


We thank Evelyne Souil for helping with the analysis of the histopathological study (from the anatomopathological platform of Université Paris 5).