Gadd45β is induced through a CAR-dependent, TNF-independent pathway in murine liver hyperplasia^{† ‡}

Animals.

Eight-week-old B6-129 female mice null for the TNFα receptor 1 (TNFR1^−/−) gene or for TNFα receptors 1 and 2 (TNFR1,2^−/−) and wild-type mice were purchased from Jackson Laboratories (Bar Harbor, ME). In additional experiments, CAR^−/− mice26 were also used. The animals were fed a laboratory chow diet provided by Ditta Mucedola (Settimo Milanese, Italy) and had free access to food and water. All procedures were performed in accordance with the UFAW Handbook on the Care and Management of Laboratory Animals and the guidelines of the animal ethics committee of this university. Experiments were performed in a temperature-controlled room with alternating 12-hours dark/light cycles. Hepatocyte proliferation was induced by a single-gavage treatment with the mitogen TCPOBOP (a gift of Dr. BA Diwan, Frederick Cancer Center, MD), at a dosage of 3 mg/kg body weight, dissolved in dimethylsulfoxide–corn oil solution. Controls received an equivalent amount of the vehicle. A two-third hepatectomy was performed according to Higgins and Anderson.27

For determination of total hepatocyte proliferation, bromodeoxyuridine (BrdU, Sigma Chemical Co., St Louis, MO) was continuously given in drinking water (1 mg/mL) to TNFR1^−/− and TNFR1,2^−/− mice were sacrificed 2 days after treatment with TCPOBOP. In another set of experiments, TCPOBOP-treated wild-type and TNFR1^−/− mice received an intraperitoneal injection of BrdU (50 mg/kg) at 22, 28, 34, and 46 hours after treatment and were sacrificed 2 hours later. Four to five mice were used per group at each time point. Liver segments were for histology or snap-frozen in liquid nitrogen and kept at −80°C until use.

Histology and Immunohistochemistry.

Liver segments were fixed in 10% buffered formalin and processed for staining with hematoxylin-eosin or immunohistochemistry. For determination of hepatocyte proliferation, a mouse monoclonal anti-BrdU antibody was obtained from Becton Dickinson (Becton Dickinson, San Jose, CA), and the peroxidase method was used to stain BrdU-positive hepatocytes, as previously described.28 Labeling index was expressed as the number of BrdU-positive nuclei/100 nuclei. Mitotic index was expressed as the number of mitotic figures/100 hepatocytes. Immunohistochemistry evaluation was done with blinding to the treatment/genotypes Results were expressed as means ± SE of four to five mice per group. At least 2,000 hepatocyte nuclei per liver were scored. Comparisons between groups were performed using Student t test.

RNA Analysis.

Total RNA was prepared from TNFR1^−/−, CAR^−/−, and wild-type mouse livers by using TriPure isolation reagent (Boehringer, Mannheim, Germany). Microarrays analysis was carried out as previously described, using a 9,000-mouse-gene complementary DNA (cDNA) microarray produced in the Microarray Facility of Albert Einstein College of Medicine. Briefly, RNA was further purified by the guanidium isothiocyanate method followed by LiCl precipitation. Fluorescent cDNA was synthesized by using green Cy3- or red Cy5-dUTP. Hybridization and washing were carried out with institutional protocols (http://microarray1k.aecom.yu.edu/). Hybridized arrays were imaged by confocal laser scanning and quantified GenePix software (Axon Instruments Inc., Union City, CA). Hybridization data were further normalized using locally weighted linear regression (LOWESS) analysis (http://microarray1k.aecom.yu.edu/bioinfo_tools/normalization.html).

Northern blots were carried out using 20 to 30 μg heat-denatured total RNA per lane as previously described.21 The following ³²P-labeled probes were used for hybridization: Cyp2b10—a 1.8-kb EcoRI fragment excised from a cDNA cloned in pUC13 (a gift from Dr. Negishi); Gadd45β—a 0.9-kb EcoRI–NotI fragment from cDNA W64388 cloned in pT7T3D-Pac10; and cyclin D1—a 1.3-kb EcoRI fragment from pcBZ054.

Electrophoretic Mobility Shift Assay.

Nuclear extracts were prepared from 200 mg liver tissue according to Schreiber et al.29 in the presence of 10 μg/mL leupeptin, 5 μg/mL antipain and pepstatin, and 1 mmol/L phenylmethyl sulfonyl fluoride (Sigma Chemical Co.). Protein concentration in the nuclear extracts was determined using the Bradford method.30 Ten micrograms nuclear extract was incubated at room temperature for 30 minutes with (2-5 × 10⁴) of the (³²P)-labeled double-stranded oligonucleotide containing the consensus NFκB DNA binding site from the IL-6 gene promoter (5′-GATCATGTGGGATTTTCCCATGT) or the STAT3 DNA binding site from the c-fos promoter (5′-GTCGACATTTCCCGTAAATCG) in 15-μL reaction mixtures containing 20 mmol/L HEPES, pH 7.9, 50 mmol/L KCl, 10% glycerol, 0.5 mmol/L dithiothreitol, 0.1 mmol/L EDTA, 2 μg poly(dI-dC), 1 μg salmon sperm DNA. Products were fractionated on nondenaturing 5% polyacrylamide gels in TBE 0.5× buffer.

CAR Is Required for the Early Response to TCPOBOP.

Initial studies determined whether the entire immediate-early response that we previously characterized22 was mediated by activation of CAR, or whether some of the genes were stimulated by another mechanism. Microarray expression profiles were determined for the CAR-null mouse,26 with and without TCPOBOP treatment, using the same microarray platform as our previous study.21 Although Gadd45β is elevated by 1 hour, a 3-hour treatment interval was studied because it produced a larger set of responsive genes. Untreated and treated CAR-null livers were separately compared with normal mouse liver (Table 1).

After treatment with TCPOBOP, stimulation of Gadd45β and virtually every gene regulation reported in our previous paper was absent in the null mouse. The immediate-early response to TCPOBOP therefore depends on CAR.

In the absence of TCPOBOP treatment, the CAR-null phenotype had limited effects on gene regulation, although a few significant changes were apparent. CyP2b9 and CyP2b10 were both down-regulated, demonstrating that endogenous or dietary ligands act through CAR to establish the basal expression levels of CAR target genes. In contrast, INSIG2 and two endogenous retroviral transcripts were upregulated. In wild-type mice, these transcripts were stimulated by TCPOBOP, so the elevated expression of these highly reactive genes in the null phenotype suggests complex compensatory mechanisms.

Gadd45β/MYD118 Is a Strong Target of Two Different Early Response Pathways.

Our previous study demonstrated the early activation of Gadd45β expression in liver after TCPOBOP, and De Smaele et al.25 and Jin et al.31 have shown that Gadd45β is induced by TNFα via activation of NFκB. The data in Table 1 indicate that CAR regulates Gadd45β but do not rule out an intermediate role for NFκB. Thus, we used the strong regulation of Gadd45β to determine whether PH and TCPOBOP act through independent or overlapping signaling mechanisms.

Gadd45β messenger RNA (mRNA) was analyzed in the livers of TNFR1^−/− mice. Analysis by Northern blot of RNA indicated that Gadd45β mRNA was present at low levels in the livers of wild-type and TNFR1^−/− mice and was strongly induced by TCPOBOP in both strains (Fig. 1). CyP2b10, a well-known CAR target gene, and cyclin D1, an early marker of TCPOBOP-induced proliferation, were also strongly induced in both wild-type and TNFR1^−/− mice. Hence, TNFα-signaling is not required for TCPOBOP-mediated induction of Gadd45β or other CAR target genes. Indeed, the induction of all 3 genes, and the basal level of CyP2b10, was somewhat greater in the null mouse, suggesting that TNFα has a negative effect on these gene regulations in the wild-type.

Many stimuli other than TNFα can activate NFκB,32, 33 so further experiments determined whether induction of Gadd45β could occur through a TNFα-independent NFκB-dependent pathway. Because STAT3 signaling can be activated by TNFα via IL-6 and by several other pathways, we also included a study of this factor. Wild-type and TNFR1^−/− mice were treated with TCPOBOP and sacrificed 30 minutes later. Gel shift analysis (electrophoretic mobility shift assay) studies (Fig. 2) indicated that no NFκB activation occurred in the livers of either wild-type or mutant mice after TCPOBOP administration. Conversely, in agreement with previous studies,7, 12 activation of NFκB and STAT3 occurred in the liver of wild-type mice 30 minutes after PH.

To directly test whether induction of Gadd45β by TCPOBOP occurred through an alternative pathway mediated by ligand-activated CAR, we measured the hepatic levels of Gadd45β mRNA in the livers of CAR^−/− mice (Fig. 3). Treatment of CAR^−/− mice with TCPOBOP did not induce Gadd45β mRNA, confirming that CAR was required for this treatment response. As expected, CAR knockout mice also showed no induction of a classical CAR-target gene, CyP2b10 or cyclin D1, an early marker of TCPOBOP-induced proliferation. These analyses also confirmed the microarray analysis of Table 1.

TCPOBOP-Induced Proliferation in the TNFR1^−/− Mouse.

Expression of Gadd45β, CyP2b10, and cyclin D1 all showed stronger responses in the TNFR1^−/− mouse compared with wild type. We therefore examined whether these changes were associated with a downstream effect on cell proliferation. As an initial evaluation, TCPOBOP-treated mice were continuously labeled with BrdU and sacrificed 48 hours after treatment. TNFR1^−/− mice showed significantly more proliferation than controls (Fig. 4). The higher labeling index, 69% versus 45% in wild type (P = .048; Fig. 4C) correlated with enhanced mitotic activity (2.45% vs. 1.56%) (Fig. 4D).

Because TNFα also acts through a second receptor (TNFR2), similar experiments were performed with mice null for both receptors (TNFR^−/− Fig. 5). As for the single-gene knockout, analysis of BrdU incorporation indicated a higher labeling index in TNFR^−/− compared with wild-type mice (50% vs. 35% in wild type, P = .054). This effect could have resulted from accelerated entry into S phase or higher overall recruitment of hepatocytes into the cell cycle. To evaluate these possibilities, a kinetic analysis was carried out in the TNFR^−/− mouse, using 2-hour BrdU pulse labels at various times after treatment (Fig. 6). This indicated that the increased proliferation was a prolonged effect. There were no significant differences between knockout and wild-type mice at 24 and 36 hours, but at 48 hours, the null mouse showed a 3-fold increase in the rate of proliferation observed in the wild type (9.2% vs 3.3%, P = .042).

Gene Regulation by CAR.

Because the regenerative response is so well characterized in hepatocytes, the existence of a second mitogenic pathway has been controversial. However, although hepatocyte proliferation is often mediated by the injury/regeneration response, in other circumstances it is part of an adaptive response to stress stimuli that are not sufficient to lead to cell death (direct hyperplasia). As a substitute for injury, either TNFα or NFκB might be directly activated by CAR—the activation could be transcriptional or posttranscriptional. As yet another possibility, the experimental process, which includes handling, gavage, and anesthesia, could induce sufficient injury to activate TNFα. The Gadd45β gene provides a critical test of all of these possibilities.

The studies presented here show that CAR is required for the diverse set of early gene regulations that TCPOBOP induces in the liver. These responses, both upregulations and downregulations, fall into two classes. One class is regulated by TCPOBOP but not PH, and includes genes already known to be direct transcriptional targets of CAR, such as CyP2b10. This gene is regulated by a specific “phenobarbital-response element” that contains two strong DR4 binding sites for RXR:CAR heterodimers.34 Although it is not surprising that such transcriptional responses are abolished in the CAR-null mouse, they nevertheless provide a clear demonstration that the knockout is effective.

The second class, genes that are regulated by both PH and TCPOBOP treatment, is more problematic. As for the first gene set, virtually all responses were abolished in the CAR-null mouse, demonstrating that CAR was essential for gene activation. Among these regulated genes, Gadd45β stands out, because its strong early expression has a kinetic pattern that is virtually identical to that of CyP2b10.22 Unlike the latter gene, no obvious DR4 RXR:CAR binding sites have been characterized for Gadd45β. In contrast, the gene has strong NFκB sites.31 Moreover, a causal relationship between NFκB and transcriptional activation has been found in both embryonic fibroblasts and a T-cell hybridoma.25 The requirement for CAR does not rule out critical roles for other factors such as NFκB: (1) CAR could directly activate another factor; (2) CAR could stimulate transcription of a transcription factor that is the direct regulator of Gadd45β; (3) CAR could collaborate with a factor that binds other sites in the Gadd45β promoter; or (4) CAR and other factors could act together in a single peptide complex.

The various studies of null-mice presented here unequivocally establish that the two regulations of Gadd45β are independent. Transcriptional activation occurs in the TNFR1^−/− mouse that lacks the TNFα response, and it occurs without induction of NFκB. The lack of STAT3 induction rules out an additional regulatory possibility. Because the TNFR1^−/− mouse lacks a proliferative response to PH, these differences in Gadd45β regulatory stimulation also confirm that proliferation is activated by two different sets of early gene responses.

Cross-Talk.

Despite the independence of early signals, the findings do suggest an interaction between the two pathways, because CAR-mediated gene inductions and proliferative responses were stronger in the TNFR1^−/− mouse. The effects were moderate but obvious and indicate a degree of negative cross-talk between the two pathways. In contrast to the transcriptional stimulation, increased proliferation was apparent only at 48 hours. The findings suggest that after proliferation is induced in an initial group of hepatocytes (typically 40%-50% of all hepatocytes), an additional group is slowly activated in the knockout. In contrast, an active TNFα response blocks recruitment of these late-proliferating cells.

In conclusion, Gadd45β is a striking marker of the immediate-early phase of hepatocyte cell proliferation. Although the timing and degree of activation are comparable in both PH and TCPOBOP treatment, the gene is activated by distinctive transcription factors in each model. Thus, the two pathways act through different mediators from their earliest stages. The striking early regulation of Gadd45β raises a possibility that this protein has a critical role in both pathways. Recent literature suggests that Gadd45β mediates an antiapoptotic response that is not directly linked to proliferation.25 If so, it is a reminder that both liver regeneration and drug-induced hyperplasia are complex responses that include not only proliferation but also cellular and metabolic adaptations. Resistance to apoptosis could be a valuable adaptive component of both responses. Nevertheless, whether Gadd45β protects hepatocytes from apoptosis is not yet established, and some published studies have indicated other roles for this protein. For example, Gadd45β is proapoptotic and antiproliferative in myeloid leukemia cells.24, 35 These contradictory effects indicate the importance resolving the function of Gadd45β in hepatocytes.