Nuclear Receptor CAR Suppresses GADD45B-p38 MAPK Signaling to Promote Phenobarbital-induced Proliferation in Mouse Liver

This article is featured in Highlights of This Issue, p. 1207

Hepatocellular carcinoma (HCC) is the third leading cause of cancer-related death in the world (1). Phenobarbital promotes HCC development in rodents. A two-stage mouse model, initiation by a genotoxic carcinogen diethylnitrosamine (DEN) and subsequent promotion by chronic treatment with phenobarbital, has long been utilized for mechanistic studies of drug-induced development of HCC (2). Following the discovery that nuclear receptor constitutive active/androstane receptor (CAR; NR1I3) is essential for phenobarbital-promoted HCC (3), the molecular mechanism of the HCC promotion has been intensively investigated. Our previous studies demonstrated that activated CAR represses phosphorylation of JNK1 in mouse primary hepatocytes (4). CAR directly bound to growth arrest and DNA-damage-inducible 45 beta (GADD45B) and accelerated the ability of GADD45B to inhibit MAPK kinase 7 (MKK7) that phosphorylates JNK1. Moreover, TNFα/actinomycin D (ActD)–induced cell death was attenuated by a CAR activator, and the attenuation was not observed in GADD45B knockout (KO) mice as well as CAR KO mice. Thus, GADD45B appears to be a crucial factor that regulates CAR-mediated repression of JNK1 signaling and of cell growth in mouse primary hepatocytes. However, the function of GADD45B and its molecular mechanisms have not been fully elucidated.

GADD45B is a member of the GADD45 family with the other members GADD45A and GADD45G. GADD45 proteins are stress inducible signal scaffolds that interact with various protein kinases and phosphatases to regulate diverse cellular functions (5). In addition to JNK1, GADD45B is also known to regulate p38 MAPK. GADD45B indirectly activates p38 MAPK through a GADD45B-MTK1-MKK6-p38 MAPK pathway (6), while it directly represses JNK1 activation. Upon activation, p38 MAPK activates a tumor suppressor p53 and cell-cycle checkpoint-related proteins such as MAPK-activated protein kinase 2 (MAPKAPK2) and induces apoptosis and cell-cycle arrest (7–9). Moreover, p38 MAPK prevents accumulation of reactive oxygen species (ROS; refs. 10, 11). Through these regulations, p38 MAPK functions as a tumor suppressor (12, 13). In fact, hepatocyte-specific KO of p38 MAPK increased susceptibility to the genotoxin DEN, predisposing mice to HCC (10, 14). Furthermore, restoring p38 MAPK activity by ablation of the BH3-only protein (BID) significantly delayed tumor development in mice (15). Therefore, these findings prompted us to investigate whether phenobarbital and CAR regulate p38 MAPK activity and, if they do, its molecular mechanism should help us to understand the phenobarbital-promoted/CAR-regulated HCC development in mouse livers.

In this study, we first utilized C57BL/6, C3H/HeNCrlBR, CAR KO, and GADD45B KO mice to demonstrate that phenobarbital induces dephosphorylation of p38 MAPK in mouse livers and that this dephosphorylation requires both CAR and GADD45B. Subsequently, the molecular mechanism of p38 MAPK dephosphorylation was examined in cell-based assays. GADD45B, acting as a scaffold protein, interacted with p38 MAPK and potentiated phosphorylation of p38 MAPK by upstream MAPK kinase 6 (MKK6). CAR was shown to interact with GADD45B, preventing GADD45B to form a complex with p38 MAPK and MKK6. Finally, GADD45B KO mice were treated with phenobarbital, and BrdUrd-positive hepatocytes were counted in the liver of these mice. Here, with these experimental observations, we will discuss the hypothesis that GADD45B is a cornerstone that enables CAR to repress p38 MAPK signaling and to promote hepatocyte proliferation, implicating phenobarbital-induced promotion of HCC development.

Phenobarbital sodium salt, 1,4-bis[2-(3,5-dichloropyridyloxy)]benzene (TCPOBOP), anisomycin, DEN, anti-FLAG M2 affinity gel, an anti-FLAG M2- horseradish peroxidase (HRP) antibody, 3x FLAG peptides, and 5-Bromo-2′-deoxyuridine (BrdUrd) were purchased from Sigma-Aldrich; an anti-GFP (HRP) antibody from Abcam; antibodies against phospho-p38 MAPK (p-p38; Thr180/Tyr182; #4511), p38 MAPK (#8690), phospho-MAPKAPK2 (#3007S), and MAPKAPK2 (#3042) from Cell Signaling Technology; an anti-V5 antibody from Invitrogen; Lipofectamine 2000 from Life Technologies; TaqMan Gene Expression Assays (probe and primer sets) for CYP2B10 (AssayID: Mm00456591_m1), GADD45A (Mm00432802_m1), GADD45B (Mm00435123_m1), GADD45G (Mm00442225_m1), and mouse GAPDH (FAM) from Applied Biosystems; cOmplete mini protease inhibitor cocktail tablets from Roche Diagnostics Corp.; a Vectastain Elite ABC kit from Vector Laboratories.

CAR cDNA (GenBank accession no. NM_009803.5) was previously cloned into pCR3 vector (Invitrogen) or pcDNA V5-His vector (Invitrogen). The FLAG tag was inserted within the 5′-flanking region of CAR/pCR3 (16). p38α cDNA (NM_011951.3) was cloned into pcDNA3.1 vector. GADD45B cDNA (NM_008655.1) was cloned into pcDNA3.1 vector harboring a 3xFLAG tag or pEYFP-c1 vector. The p38α cDNA or GADD45B cDNA was cloned into pGEX-4T-3 vector (GE Healthcare) for GST-tagged proteins. FLAG-MKK6 active mutant (Ser207Glu and Thr211Glu)/pcDNA vector (Addgene ID: 13518) was purchased from Addgene. A GADD45B 1-92 mutant/pEYFP and a GADD45B 93-160/pEYFP were constructed by a deletion method.

C3H/HeNCrlBR and C57BL/6 mice were obtained from Charles River Laboratories and Jackson Laboratories, respectively. CAR KO and CAR wild-type (WT) in the C3H/HeNCrlBR background (3) and in the C57BL/6 background and also GADD45B KO and GADD45B WT in the C57BL/6 background (4, 17) were maintained at the National Institute of Environmental Health Sciences (NIEHS). Phenobarbital in PBS (100 mg/kg body weight), TCPOBOP in DMSO/corn oil (3 mg/kg body weight), or a control solution was intraperitoneally injected. For a two-stage model of HCC development, 5-week-old C57BL/6 male mice were intraperitoneally injected with DEN (90 mg/kg body weight) and housed for two weeks, followed by drinking water containing phenobarbital (500 ppm) for additional two weeks. Animal experiments were conducted according to the protocols approved by the Animal ethics committee at NIEHS/NIH.

Starting 48 hours before phenobarbital treatment, mice received BrdUrd (0.5 mg/mL) in their drinking water to analyze hepatocyte proliferation. Phenobarbital in PBS (100 mg/kg body weight) was intraperitoneally injected 48 hours later, and 4 mice from each of the four groups (C57BL/6 male and female and GADD45B KO male and female; around 9 weeks of age) were then sacrificed at 0, 24, 36, and 48 hours. A section of liver was taken at necropsy from each animal, fixed in 10% neutral buffered formalin, processed, sectioned at 5 μm, and stained with hematoxylin and eosin. Sections were also stained by an anti-BrdUrd antibody (1:500) to assess cell proliferation. Counting of hepatocytes stained for BrdUrd was performed with an Olympus BX51 light microscope, using a 40× objective and an ocular grid as per the method of Ton and colleagues (18). BrdUrd-stained hepatocytes were identified as cells located within the hepatocyte plates and having rounded nuclei with abundant cytoplasm. Binucleate hepatocytes were often seen and were counted singly. The labeling index for each liver was calculated by dividing the number of BrdUrd-stained hepatocytes in 20 fields by the total number of BrdUrd-positive and -negative hepatocytes in the same 20 fields, expressed as a percentage.

Human hepatoma cell line Huh-7 cells were obtained from Japan Collection of Research Bioresources Cell Bank. Huh-7 cells were authenticated by short tandem repeat analysis at ATCC in January 2018. Cells were cultured in minimum essential media (Invitrogen) supplemented with 10% FBS and penicillin/streptomycin at 37°C with 5% CO₂. Plasmids were transfected into Huh-7 cells with a Lipofectamine 2000 reagent according to the manufacturers' instructions.

Huh-7 cells were lysed in a lysis buffer [20 mmol/L Tris-HCl (pH 7.5), 0.5 mmol/L EDTA, 100 mmol/L NaCl, 1% Triton X-100, 10% glycerol, a protease inhibitor cocktail, and phosphatase inhibitor cocktails 2 and 3], sonicated, and centrifuged. Obtained supernatants were used as extracts for immunoprecipitation assays. Protein extracts were incubated with anti-FLAG M2 affinity gels or anti-GFP antibody–conjugated agarose beads. These gels or beads were washed three times with TBS [25 mmol/L Tris-HCl (pH 7.4), 140 mmol/L NaCl, and 2.7 mmol/L KCl]. Washed gels or beads were heat-treated in an SDS sample buffer (2×) [157 mmol/L Tris-HCl (pH 6.8), 4% SDS, 25% glycerol, and 0.01% bromophenol blue] and subjected to Western blot analyses.

For double immunoprecipitation assays, immunoprecipitation was sequentially repeated with anti-FLAG M2 affinity gels at the 1st step and anti-GFP antibody–conjugated agarose beads at the 2nd step; each step of the immunoprecipitation was performed as described previously. Proteins bound at the 1st step were eluted by incubation with FLAG peptides (100 μg/mL) in TBS for 2–4 hours at 4°C and subjected to the 2nd step of immunoprecipitation with anti-GFP antibodies and to subsequent Western blot analyses.

FLAG-MKK6 was expressed in Huh-7 cells, followed by treatment with 200 nmol/L anisomycin for 15–30 minutes to activate MKK6. Cells were lysed in the abovementioned lysis buffer containing 2.5 mmol/L Na₄P₂O₇ and 1 mmol/L Na₃VO₄. FALG-MKK6 was bound to anti-FLAG M2 affinity gels for 3 hours at 4°C, washed by TBS and utilized as an active MKK6 enzyme. GST-tagged p38 MAPK was expressed in and purified from Escherichia coli BL21 (DE3; Agilent Technologies) using GSH Sepharose 4B (GE Healthcare). After the GST tag was removed by thrombin digestion, p38 MAPK was used as a substrate. Recombinant GADD45B was obtained by the same way that p38 MAPK was purified.

In in vitro phosphorylation assays, recombinant p38 MAPK was phosphorylated by active MKK6 with or without the presence of recombinant GADD45B in a kinase buffer [25 mmol/L Tris-HCl (pH 7.5), 2.5 mmol/L Na₄P₂O₇, 1 mmol/L Na₃VO₄, 10 mmol/L MgCl₂, and 1 mmol/L dithiothreitol] with ATP (130–200 μmol/L). Recombinant GST protein was used as a control for GADD45B. Kinase reaction was continued for 20 minutes at 30°C. The reaction was stopped by adding 1× SDS sample buffer. Phosphorylation levels of p38 MAPK were determined by Western blot analysis.

Western blot analysis (16) and IHC were performed as described previously (19).

The CAR ligand-binding domain (LBD) and GADD45G were docked using ZDOCK (3.0.2; ref. 20). As the crystal structure of GADD45B is unavailable, GADD45G (PDB ID: 3CG6) was used to dock with CAR LBD (PDB ID: 1XNX). Stability was calculated using a protein interfaces, surfaces, and assemblies' service PISA at the European Bioinformatics Institute (Hinxton, Cambridgeshire; ref. 21).

mRNA expression levels were analyzed by Student t test. A value of P < 0.05 was considered statistically significant.

Liver extracts were prepared from CAR WT and CAR KO mice in the C3H/NeCrIBR background after phenobarbital treatment and subjected to Western blot analysis to examine phosphorylation levels of p38 MAPK. Hepatic p38 MAPK was phosphorylated and remained phosphorylated 12 hours after phenobarbital treatment (Fig. 1A). However, phosphorylation levels of p38 MAPK was greatly decreased in CAR WT but not in CAR KO mice 24 + 4 hours after phenobarbital treatment (Fig. 1B). This phenobarbital-induced dephosphorylation was equally observed in both male and female mice (Fig. 1C). Although, C3H/NeCrIBR mice are susceptible in developing HCC but C57BL/6 mice are resistant in it, p38 MAPK was equally phosphorylated in livers of both strains (Fig. 1D). Phenobarbital-induced dephosphorylation was also observed in C57BL/6 mice and was dependent on CAR (Fig. 1E). This CAR-mediated dephosphorylation was not specific to phenobarbital and also observed after treatment with TCPOBOP, a mouse CAR ligand (Fig. 1F). These results indicate that phenobarbital-induced dephosphorylation of p38 MAPK is regulated by CAR in strain- and sex-independent manners. IHC provided further evidence that phosphorylated p38 MAPK is present in the nuclei of hepatocytes and decreased after phenobarbital treatment in a CAR-dependent manner (Fig. 1G).

To understand how phenobarbital induces dephosphorylation, p38 MAPK phosphorylation by MKK6 was first investigated. p38 MAPK, which was barely phosphorylated in Huh-7 cells, became more phosphorylated as GADD45B was increased in the cells (Fig. 2A). A purified active MKK6 phosphorylated p38 MAPK in in vitro kinase assays (Supplementary Fig. S1A) and recombinant GADD45B protein greatly increased this phosphorylation (Fig. 2B). Thus, GADD45B was capable of directly enhancing MKK6 to phosphorylate p38 MAPK. A FLAG-tagged active MKK6 mutant and an untagged p38 MAPK were expressed with or without EYFP-tagged GADD45B in Huh-7 cells. An anti-FLAG antibody coprecipitated p38 MAPK with MKK6, and this coimmunoprecipitation was dramatically increased in the presence of GADD45B (Fig. 2C, lane 4). These three proteins were confirmed to interact with each other (Supplementary Fig. S1B–S1D). Subsequently, double coimmunoprecipitation assays were employed to examine a complex consisting of GADD45B, p38 MAPK, and MKK6. To this end, EYFP-GADD45B, the FALG-MKK6 mutant, and untagged p38 MAPK were coexpressed in Huh-7 cells from which extracts were prepared for sequential immunoprecipitation with an anti-FLAG antibody and then an anti-GFP antibody. p38 MAPK was coprecipitated with MKK6 and GADD45B (Fig. 2D). These results indicated that GADD45B forms a complex with p38 MAPK and MKK6 to stimulate phosphorylation of p38 MAPK.

Coimmunoprecipitation assays were performed to examine whether CAR alters interactions between p38 MAPK and MKK6. As demonstrated in Fig. 2C (lane 4), coexpressed GADD45B increased coprecipitation of p38 MAPK with MKK6 (Fig. 3A, lane 6), and ectopically expressed CAR nearly abolished this coimmunoprecipitation (Fig. 3A, lane 8). Ectopic expression of CAR is known to enable the cells to contain active CAR proteins (22, 23). Ectopically expressed CAR was confirmed to bind to GADD45B (Supplementary Fig. S2A). Subsequent double coimmunoprecipitation assays provided evidence that a complex formed by GADD45B, MKK6, and p38 MAPK was dissociating in the presence of coexpressed CAR (Fig. 3B). Under these conditions in which CAR dissociated the complex, CAR interacted with GADD45B and p38 MAPK, forming another complex consisting of GADD45B, p38 MAPK, and CAR (Fig. 3C). In addition, GADD45B can also form a complex with CAR and MKK6 (Supplementary Fig. S2B). These results indicated that CAR binds to GADD45B and prevents interactions between p38 MAPK and MKK6 by forming a complex with p38 MAPK or MKK6.

Given the finding that CAR disrupted a GADD45B-schaffoled p38 MAPK-MKK6 interaction, the molecular basis of this CAR-GADD45B interaction was examined by coexpressed CAR and a GADD45B fragment in Huh-7 cells, from which the N-terminal half (1/92) of GADD45B, but not the C-terminal half (93/160), was coprecipitated with CAR (Fig. 4A). An internal deletion mutant GADD45B deletion (Δ41/92) was not coprecipitated with CAR (Fig. 4B). The 2nd and 3rd helices constitute this N-terminal region and a CAR-binding site. Similar coimmunoprecipitation assays revealed that p38 MAPK or MKK6 bound the same region of GADD45B (Supplementary Fig. S2C and S2D). Numerous deletion mutants were constructed with CAR (Supplementary Fig. S3A). Coimmunoprecipitation assays showed a preferential binding of GADD45B to CAR LBD over the CAR DBD (Supplementary Fig. S3B). Furthermore, GADD45B was coprecipitated with CAR when its mutants included helix 3, helix 7, or both (Fig. 4D). Having these potential binding regions in CAR and GADD45B, ZDOCK docking program and PISA program were utilized to visualize a heterodimer between CAR (PDB ID: 1XNX) and GADD45G (PDB ID: 3CG6). As a GADD45B structure has not been determined, the GADD45G was used instead. The most stable heterodimer interface was formed between the helix 7 of CAR and helices 2 to 3 of GADD45G (Supplementary Fig. S4).

GADD45B WT and GADD45B KO male mice were treated with phenobarbital as CAR WT and CAR KO males were treated in experiments shown in Fig. 1B. Liver extracts from these GADD45B mice were subjected to Western blot analyses. Phenobarbital-induced dephosphorylation of p38 MAPK was significantly attenuated in GADD45B KO mice over that in GADD45B WT mice (Fig. 5A), which was reminiscent of CAR-dependent dephosphorylation of p38 MAPK in experiments with CAR KO mice (Fig. 1B).

It was decided to explore whether phenobarbital induces p38 MAPK dephosphorylation in livers after HCC initiation. C57BL/6 males were pretreated with a single injection of DEN for two weeks, prior to drinking phenobarbital for additional two weeks. Western blot analysis of liver extracts from these mice revealed that phenobarbital was capable of inducing dephosphorylation in the conditions of two-stage model of HCC development (Fig. 5B). In addition to p38 MAPK, its downstream kinase MAPKAPK2 (MK2) was also dephosphorylated.

Baseline BrdUrd hepatocyte labeling indices (LI), prior to phenobarbital treatment (0 hour), were 0.43% and 0.39% in C57BL/6 males and females, respectively, and 1.41% and 1.44% in GADD45B KO males and females, respectively (Table 1). These findings indicate a three times higher resting proliferation rate in the GADD45B KO mice, probably reflecting the function of GADD45B as a tumor suppressor. Following phenobarbital treatment for 24 hours, LIs of both strains and both sexes increased. The C57BL/6 males and females increased the LI values a 1.53- and 9.72-fold over their values at 0 hour, respectively. These increases were 1.72- and 6.03-fold in the GADD45B KO mice. By 36 hours following phenobarbital injection, the LI of both the males and females of each strain continued to rise. At this point, the increase was particularly notable in the C57BL/6 males with a 9.35-fold over the 24-hour values, while this increase was only 3.23-fold in the GADD45B KO males. On the other hand, the females declined these folds to 2.05 and 1.28 in the C57BL/6 and GADD45B KO mice, respectively, indicating that the females peaked the increase of LI values by this time. Thus, proliferation had increased most significantly in the C57BL/6 males during this period of phenobarbital treatment. However, the LIs of the GADD45B KO males and females exceeded the LI of their C57BL/6 counterparts, but only slightly. By 48 hours, the average LI of the C57BL/6 males continued to rise by 1.22-fold from 6.17% at 36 hours to 7.54%. Conversely, the LI of the GADD45B KO males fell 2.13-fold from 7.82% at 36 hours to 3.01% at 48 hours. As a result, the LI values of C57BL/6 males exceeded that of GADD45B KO males. These observations suggested that phenobarbital-induced hepatocyte proliferation depends on GADD45B and that these GADD45B-dependent increases continue in males but not in females.

Dephosphorylation of p38 MAPK was observed from 24 hours after phenobarbital treatment (Fig. 1A; Supplementary Fig. S6A and S6B). This p38 MAPK dephosphorylation at late stages should negatively affect phenobarbital-activated transcription of the Cyp2b10 gene, a representative target gene of CAR, since phenobarbital/CAR requires phosphorylated p38 MAPK for transcription of the Cyp2b10 gene (16). In fact, expression levels of CYP2B10 mRNA peak 12–18 hours after the 1st injection of phenobarbital and then begin to decline. The 2nd phenobarbital injection at 24 hours after the 1st injection was unable to restore the induction, even though nuclear levels of CAR increased after this 2nd injection (Supplementary Fig. S6C), probably because p38 MAPK activity became low levels. During the time when CAR/GADD45B induces p38 MAPK dephosphorylation, CAR/GADD45B–dependent hepatocyte proliferation continued to rise in male mice (Table 1). The molecular mechanism that regulates these sequential processes is an important subject in future research.

Phosphorylated p38 MAPK can be a tumor suppressor that represses hepatic proliferation. Phenobarbital, a nongenotoxic carcinogen, repressed p38 MAPK by inducing dephosphorylation in mouse livers. CAR regulated this dephosphorylation by binding to GADD45B that scaffolds p38 MAPK and MKK6 (Fig. 6). While phenobarbital induced hepatocyte proliferation through both GADD45B-depedent and -independent mechanisms, GADD45B-dependent proliferation continued in only male mice. Phenobarbital may utilize this GADD45B-mediated p38 MAPK pathway as an essential cell signal to promote HCC development.

Livers to develop HCC require chronic phenobarbital exposures with genotoxic mutations of β-catenin gene (24, 25). In a two-stage HCC model, DEN mutates the Ctnnb1 gene, which sustains the activity of β-catenin. Consequently, phenobarbital selectively proliferates hepatocytes that bear these mutations of the Ctnnb1 gene (26, 27). Notably, β-catenin seemed to manifest hepatocyte proliferation in a sex-dependent manner, as indicated by the fact that hepatic β-catenin KO mice attenuated proliferation only in males (24). p38 MAPK signaling can suppress downstream factors of β-catenin proliferation signaling. Thus, phenobarbital treatment may attenuate the p38 MAPK-mediated suppression of those factors.

As hepatocyte-specific ablation of either p38 MAPK or JNK1 signaling promotes DEN-initiated HCC (10, 28), CAR/GADD45B–mediated ablation of these signaling can be a causing factor of phenobarbital-induced HCC promotion, although with the caveat that interactions of CAR-GADD45B-p38 MAPK or JNK1 remain to be demonstrated with endogenous proteins in livers in future investigations. Phenobarbital is known to promote HCC development in both mouse strain- and sex-dependent manners in C3H over C57BL/6 mice and males over females. However, the GADD45B-mediated repression of p38 MAPK signaling was equally observed in C3H and C57BL/6 and males and females. On the other hand, phenobarbital-induced and GADD45B-dependent BrdUrd-positive hepatocytes continued to rise in only male mice. Therefore, sex-related factors appear to render the GADD45B-dependent proliferation male-predominant. If the GADD45B-dependent proliferation is sustained, it may become a dominant proliferation signal, since GADD45B-independent proliferation was transient in male mice. GADD45B-mediated regulation of p38 MAPK and JNK1 signaling should be expanded to be an essential determinant that promotes phenobarbital-induced HCC development in future investigations. Phenobarbital appears to prototypically alter MAPK signaling (e.g., ERK1/2, JNK1, and p38 MAPK signaling). To regulate ERK1/2, phenobarbital binds and represses the EGF receptor and/or the insulin receptor (29, 30). GADD45B is positioned as a key factor to regulate JNK1 and p38 MAPK signaling through its interaction with CAR in response to phenobarbital. In addition to the stress-activated protein kinases, it is possible that other MAPKs can also be regulated by CAR/GADD45B. Furthermore, GADD45B is involved in regulation of numerous biological processes including DNA methylation and cell migration. Phenobarbital may also affect GADD45B on these additional regulations that become critical factors in HCC development.

In conclusion, GADD45B is characterized as a target of CAR to repress p38 MAPK activity and to promote cell proliferation in mouse livers, providing us with an experimental basis for understanding the molecular mechanism of phenobarbital-promoted HCC development.

No potential conflicts of interest were disclosed.

Conception and design: T. Hori

Development of methodology: T. Hori, K. Saito, R. Moore, G.P. Flake, M. Negishi

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): T. Hori, K. Saito

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): T. Hori, R. Moore, G.P. Flake, M. Negishi

Writing, review, and/or revision of the manuscript: T. Hori, K. Saito, G.P. Flake, M. Negishi

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): T. Hori, K. Saito, G.P. Flake

The authors thank the DNA Sequencing and Histology Core Facilities (NIEHS/NIH) and the Protein Expression Core Facility (NIEHS/NIH) for anti-GFP antibody agarose beads. This work was supported by the Intramural Research Program of the NIH and National Institute of Environmental Health Sciences (grant numbers Z01ES71005-01).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.