INTRODUCTION

Polycyclic aromatic hydrocarbons such as B[a]P¹, 3-methylcholanthrene, and TCDD are environmental pollutants that elicit a variety of toxic, teratogenic, and carcinogenic responses in exposed animals (1-7). In addition, these substances are potent inducers of certain biotransformation reactions that are catalyzed by the cytochrome P-450-dependent monooxygenases, a super family of isozymic hemoproteins comprising more than 12 gene groups (8). These isozymes display broad substrate specificity and metabolize both endogenous substrates and xenobiotics to electrophilic derivatives, some of which can interact with DNA, thereby potentially activating protooncogenes or inactivating tumor suppressor genes.

A principal cytochrome P-450 induced by B[a]P, 3-methylcholanthrene, or TCDD is CYP1A1, which is in part responsible for the activation of PAHs to carcinogenic intermediates. This induction is mediated by receptors, such as the Ah receptor (or dioxin receptor, 8 S protein). The Ah receptor-mediated mechanism is the best characterized and most widely accepted model for PAH-induced expression of CYP1A1 (9, 10). Upon ligand binding, the Ah receptor undergoes a temperature-dependent transformation accompanied by translocation into the nucleus where the ligand receptor complex binds to specific cis-elements (also known as XREs and DREs) present in the 5-regulatory regions of CYP1A1 and of several other genes in the Ah gene battery. The result of such interaction is the transcriptional activation of Ah responsive genes (3, 5, 6, 11). The Ah receptor contains a 95-kDa helix-loop-helix motif, characteristic of several transcriptional activators (12, 13). The mechanism as outlined is more complicated because functional Ah receptor represents a heterodimer that contains, in addition to the ligand binding moiety, a second helix-loop-helix 84-kDa protein component, known as ARNT, which is needed for the nuclear translocation of the ligand-receptor complex (14-16). In addition, heat shock protein 90 plays an important role in the intracellular localization of the Ah receptor (see review in Ref. 42).

Evidence has been published suggesting that the induction of rat CYP1A1 after treatment with PAHs, such as B[a]P or B[e]P, may require an Ah receptor-independent pathway mediated by another cytosolic receptor protein, the 4 S PAH-binding protein (17-23). Both the Ah receptor and 4 S PAH-binding protein exhibit distinct ligand binding specificities, with 3-methylcholanthrene interacting with both. B[a]P and B[e]P have been demonstrated to act solely as ligands for the 4 S protein in rat liver and rat hepatoma cells (17, 24-28). Like the Ah receptor, the 4 S protein undergoes nuclear translocation upon interaction with the PAH (17, 29-31) and complexes to cis-response elements in different regulatory regions of CYP1A1 (21, 32, 33).

The 4 S protein has been purified to homogeneity using a series of chromatographic steps, involving ion exchange, gel permeation, hydrophobic interaction, and affinity chromatographies (34). Partial sequencing of the 33-kDa 4 S protein indicated its identity as GNMT (S-adenosylmethionine:glycine N-methyltransferase, EC 2.1.1.20). Based upon a number of criteria, GNMT and the 4 S PAH-binding protein were shown to be one and the same protein (34). It has been estimated that this enzyme may be present in high concentrations in rat and human livers (35-38). Despite the high level of this protein in human tissues, the physiological role of GNMT is not well understood.

GNMT was first found in an extract of guinea pig liver (39) and postulated to be involved in the oxidation of the methyl carbon of methionine, although this pathway accounts for only 20% of the total methionine methyl metabolism (40). Later, Kerr (35) suggested that the enzyme may play a role in the regulation of the relative levels of S-adenosylmethionine and S-adenosylhomocysteine in the cell. In a subsequent study by Cook and Wagner (41), GNMT was shown to act as a folate-binding protein in rat liver cytosol. Rat GNMT, in its enzymatic role, is a homotetramer composed of 33-kDa subunits (36); independent binding sites exist for S-adenosylmethionine, glycine, and folate (37, 41, 43) as well as for B[a]P (34). The enzymatic form of GNMT, i.e. the homotetramer, is inactive as a B[a]P-binding protein, and the monomer is unable to function either enzymatically or as a B[a]P binder. Preliminary evidence indicates the homodimer as the functional B[a]P-binding unit.²

The cDNA for GNMT has been isolated and sequenced (43). Because the cDNA sequence for GNMT is known, it was possible to more fully explore the role of the protein in the B[a]P-mediated induction of CYP1A1 in a biological model that was devoid of both Ah receptor and 4 S PAH binding activities. In the present report, we show that the stable transformation of CHO cells with the GNMT gene allows the induction of CYP1A1 by B[a]P, B[e]P, and 3-methylcholanthrene. We conclude that GNMT is involved in the transcriptional activation of CYP1A1, thus providing an alternate, Ah receptor-independent pathway for the modulation of CYP1A1 expression by certain PAHs, such as B[a]P, B[e]P, and 3-methylcholanthrene.

MATERIALS AND METHODS

Tissue culture medium, -minimal essential medium, fetal bovine serum, gentamycin, geneticin (G418), and Lipofectin were purchased from Life Technologies, Inc. The sources of the other materials were: [-³²P]dATP from ICN Biochemicals (Irvine, CA); [³H]B[a]P (60 Ci/mmol) from Amersham Corp.; [³H]TCDD (41 Ci/mmol) from Chem-Syn Science Labs (Lenexa, KS); Immobilon P from Millipore (Bedford, MA); S&S transfer membranes from Schleicher & Schuell (Keene, NH); BM Chemiluminescence Western blotting kit from Biochemica Boehringer Mannheim Corp. (Indianapolis, IN); Tris, TEMED, Tween 20, B[a]P, B[e]P, 3-methylcholanthrene, TCDD, isocitrate dehydrogenase, nicotinamide, ethoxyresorufin, and resorufin from Sigma. Affinity-purified antibodies against GNMT and the Ah receptor were generous gifts from Dr. Conrad Wagner (Vanderbilt University) and Dr. Bill Greenlee (University of Massachusetts Medical Center), respectively. The Ah receptor cDNA-containing plasmid was kindly provided by Dr. Chris Bradfield (University of Wisconsin).

The CHO D422 cells, which are partially deficient in adenine phosphoribosyl transferase, were originally isolated after mutagenesis by Sib selection on the basis of partial resistance to ADP (44). This cell line has been used by us for a number of years in transfection and DNA repair studies. The stock cultures were maintained at 37 °C in an atmosphere of 95% air and 5% CO₂ in minimal essential medium, supplemented with 40 mg/l proline containing 50 µg/ml gentamycin and 10% fetal bovine calf serum. The mouse hepatoma cell line, Hepa-1, which was used as a positive control in the XRE binding experiments, was cultured and maintained as described (45).

The GNMT cDNA was synthesized by RT-PCR methodology from a rat liver poly(A)⁺ RNA preparation. The GNMT-specific forward and reverse primers were 5-GAGCCAGCTAGCGTCAGGATGGTGGAC and 5-TGGGAGCTCGAGCCAGGCTCAGCCTGT, respectively. The subsequent product was purified by agarose gel electophoresis, and the sequence was shown to be identical to published GNMT cDNA (43). Two cloning sites for NheI and XhoI were incorporated into the noncoding regions of the GNMT cDNA by the PCR methodology. The purified double-stranded DNA product was ligated into the NheI/XhoI-digested pMAMneo (CLONTECH, Palo Alto, CA). The plasmid construct, pMAMneo/GNMT, contains the GNMT insert of the proper size in the sense orientation as determined by restriction digestion and sequence determination. pMAMneo/GNMT was transfected into the CHO cells by the Lipofectin method. Briefly, on day 0, 10 µg of DNA in 100 µl of Opti-MEM I was mixed with 15 µl of Lipofectin reagent, overlaid onto approximately 2 × 10⁵ cells in 2 ml of serum-free growth medium, and incubated for 24 h at 37 °C in a CO₂ incubator. Control transfectants were established using DNA from the parent pMAMneo plasmid. On day 1, the DNA-containing medium was replaced by standard serum-included growth medium, and incubation was continued for an additional 48 h. On day 4, the cells were placed in a selection medium containing geneticin (0.4 mg/ml). Single cell clones were picked, grown in the selection medium, and assayed for GNMT expression by Western blotting and B[a]P binding activity.

The stably transfected clones were grown in minimal essential medium containing 0.4 mg/ml G418, 10% dialyzed fetal bovine serum, and 1 µM dexamethasone. When confluent, the cells were washed with ice-cold phosphate-buffered saline, resuspended in TEDGP (40 mM Tris, pH 7.4, 1 mM disodium EDTA, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, 10% glycerol, 20 × SSPE, 3.6 M NaCl, 0.2 M sodium phosphate, pH 7.7, 20 mM EDTA), and homogenized in a Dounce apparatus. The cytoplasm was separated by centrifugation at 10,000 × g at 4 °C. Total cellular protein was subjected to SDS-8% polyacrylamide gel electophoresis (46) under reducing conditions, and the separated proteins were transferred to Immobilon P membranes using a mini-blotting apparatus (Bio-Rad). The membranes were blocked for 1 h in 5% nonfat dry milk in Tris-saline, pH 7.5, and incubated overnight with appropriate primary antibodies, followed by incubation for 1 h with secondary antibody linked to horseradish peroxidase (Boehringer Mannheim). After washing, the Western blot was developed with the chemiluminescence detection reagent according to the manfacturer's instructions.

Specific binding activity was assessed by sucrose gradient analysis as described previously (25). Briefly, the 100,000 × g supernatant fraction (0.5-1 mg protein) was incubated with 10 nM [³H]B[a]P or [³H]TCDD for 1 h at 4 °C and 2 h at room temperature, respectively, with or without a 200-fold excess of unlabeled ligand. Control cultures contained an equivalent concentration of vehicle, 95% ethanol. A 300-µl fraction was layered onto linear 5-20% sucrose density gradients and centrifuged for 2 h at 372,000 × g in a Beckman VTI-65 rotor. 10-drop fractions were collected and assayed for radioactivity in a liquid scintillation spectrometer.

RNA was isolated from stably transfected clones by the ULTRA SPEC RNA isolation system from Biotecx. Total RNA samples (~40 µg) were denatured by heating at 65 °C for 15 min in 50% (v/v) formamide, 2.2 M formaldehyde, 0.5 × running buffer and fractionated by electrophoresis through a formaldehyde gel containing 1.5% agarose (47). The RNA samples were transferred overnight to Nytran membranes (Schleicher & Schuell) in 10 × SSPE, prehybridized, and hybridized as described (48) using various DNA probes that were labeled to a specific activity of >10⁸ cpm/µg DNA with [-³²P]dATP using the random primer method (49). Following hybridization, the membranes were washed (48) and exposed to x-ray film (XAR-5, Eastman Kodak Co.) at 80 °C overnight.

CHO cells grown to 80-90% confluency were challenged with 4 µM B[a]P for 4 h. The cells were washed with ice-cold phosphate-buffered saline, resuspended in 20 mM Tris, pH 7.8, and homogenized in a glass homogenizer. Supernatants were collected by centrifugation at 1500 rpm, and after the addition of cofactor mix, the EROD assay was performed according to a modification (50) of a previously established method (51); the fluorescence of resorufin was determined as described, and the quantity of resorufin was estimated from a standard curve.

Cellular cytosolic extracts for GMSA were prepared by the method of Shapiro et al. (52), followed by treatment with 10 nM TCDD for 3 h at room temperature in vitro. Nuclear extracts were prepared from cells that had been treated with 2 nM TCDD for 90 min at 37 °C essentially as described by Probst et al. (53). Briefly, the cells were suspended in 10 mM HEPES, pH 7.5, for 15 min on ice, centrifuged, resuspended in 1 cell pellet volume of HED (25 mM HEPES, pH 7.5, 2 mM EDTA, 1 mM dithiothreitol) and disrupted in a loose-fitting Dounce homogenizer. After centrifugation in the cold for 10 min at 14,000 × g, the nuclear pellet was resuspended in 1 cell pellet volume of HED, 0.4 M KCl and was rocked at 4 °C for 30 min. After homogenization in a tight-fitting Dounce apparatus, glycerol was added to a final concentration of 20% (v/v). The nuclear fraction was obtained by centrifugation at 180,000 × g for 30 min at 4 °C, and the supernatants were dialyzed against HED.

The ³²P-labeled double-stranded synthetic oligonucleotide, which contained the XRE1 from 1003 to 977 base pairs of the mouse CYP1A1, was prepared essentially as described (54). GMSA was conducted as described (54).

RESULTS

The GNMT cDNA was inserted downstream from a retroviral enhancer and promoter elements and upstream from SV40 splicing and polyadenylation sites. CHO cells were chosen in part as a target for transformation because of the lack of endogenous GNMT gene expression and the efficiency of transfection with pMAMneo/GNMT using the Lipofectin technique. 35 neomycin-resistant clones were randomly selected, isolated, and grown into mass cultures for further characterization. Initial screening was conducted by RT-PCR analyses of the G418-resistant clones for the 4 S PAH-binding protein and -actin gene expression. The data with Clone 6 of CHO-GNMT (lane 5), parental (lane 2), and CHO-neo (lane 3) are presented in Fig. 1. A positive control, i.e. RNA from rat liver, is shown in lane 1. Only the CHO-GNMT cells (in addition to the liver RNA) exhibited the correct truncated 4 S transcript (lane 5). No signals were observed with either the CHO (lane 2) or CHO-neo cells (lane 3). Furthermore, no positive signal was observed in the RNA obtained from the CHO-GNMT cells, when the reactions were performed in the absence of reverse transcriptase (lane 4).

RT-PCR analysis of 4 S and -actin transcripts from CHO cells.

-A

To determine directly if CHO-GNMT cells produce GNMT protein, total cellular protein was analyzed by denaturating gel electrophoresis, and the proteins were transferred to nitrocellulose membranes and probed with affinity-purified polyclonal antibodies to GNMT. As is shown in Fig. 2 (lane b), CHO-GNMT cells express significant levels of GNMT when compared with the undetectable levels existent in parental (lane d) or CHO-neo (lane c) cells.

Specific B[a]P and TCDD binding activities were determined in the various CHO cells (Fig. 3, A and B). Significant B[a]P binding activity was observed with extracts from the CHO-GNMT cells (lane 3, Fig. 3A). Essentially no binding activity was apparent with extracts from the CHO (lane 1) or CHO-neo (lane 2) cells. The K_D for B[a]P interaction, as determined by a Scatchard analysis (data not shown), was 2.4 nM with a binding capacity of 590 fmol/mg of protein. These values are in close agreement with 2.54 nM and 530 fmol/mg protein observed with rat liver cytosol (30).

Specific TCDD binding activity was determined in cytosolic extracts from the positive control, Hepa-1 hepatoma cells (lane 1, Fig. 3B). Essentially no TCDD binding activity was observed in extracts from either the CHO-GNMT (lane 2), CHO-neo (data not shown), or parental CHO (data not shown) cells.

The CHO-GNMT, CHO-neo, and CHO cells were treated with B[a]P, B[e]P, 3-methylcholanthrene, or TCDD for 8 h, after which time total RNA was prepared and analyzed on denaturing formaldehyde agarose gels. RNA that had been transferred to nitrocellulose membranes was simultaneously probed with mouse CYP1A1 and -actin cDNAs. (Fig. 4). Only B[a]P, B[e]P, and 3-methylcholanthrene elicited a positive CYP1A1 mRNA response (lanes 2, 4, and 6, respectively). TCDD, which represents the prototypic ligand for the Ah receptor, was not effective in inducing CYP1A1 in the CHO-GNMT cells. -Napthoflavone was also ineffective in inducing CYP1A1 in these cells (data not shown).

A

Whole cell lysates were prepared from B[a]P-induced, parental, CHO-neo, and CHO-GNMT cells and were analyzed on SDS-denaturing gels for CYP1A1 protein (Fig. 5). The electrolytically transferred protein was probed with polyclonal antibodies against CYP1A1 protein. Only CHO-GNMT cells expressed CYP1A1 protein in response to B[a]P treatment (lane 6). The level of CYP1A1 in the livers from control (lane 2) and B[a]P-treated (lane 1) rats is indicated as a positive control.

The effect of B[a]P treatment on the activity of a CYP1A1-dependent monooxygenase, EROD, was determined in CHO, CHO-neo, and CHO-GNMT cells; these data are presented in Table I. Treatment with the PAH for 6 h caused a 10-15-fold induction of EROD in CHO-GNMT cells compared with negligible amounts in the parental and CHO-neo cells.

Table I. Effect of B[a]P treatment on EROD (CYP1A1) activity in CHO cells The cells were grown to 90% confluency and treated with 4 µM B[a]P in 95% ethanol for 6 h at 37 °C. The values represent the means ± S.E. from three independent experiments. Sample Treatment EROD activity pmol/mg/min Parental Vehicle alone 12  ± 0.2 Parental B[a]P 13  ± 0.4 CHO-neo Vehicle alone 15  ± 0.3 CHO-neo B[a]P 14  ± 0.2 CHO-GNMT Vehicle alone 16  ± 0.5 CHO-GNMT B[a]P 219  ± 8.2

The Ah receptor is a well characterized protein that participates in the dioxin-mediated induction of CYP1A1 by binding to XREs associated with CYP1A1. The presence or the absence of the Ah receptor in the CHO, CHO-neo, and CHO-GNMT cells was determined by both Northern and Western analyses. No detectable amounts of the protein were found in any of these cells (data not shown).

The lack of Ah receptor activity was confirmed by a functional assay involving nuclear translocation and XRE binding experiments with the CHO-GNMT cells; Hepa-1 mouse hepatoma cells were used as the positive control (Fig. 6). The formation of the XRE-Ah receptor complex was demonstrated with cytosolic extracts from Hepa-1 cells that had been treated in vitro with TCDD (lane 2). No such complex was demonstrable with extracts from CHO-GNMT cells that had been treated similarly (lane 6).

The formation of a XRE-Ah receptor complex was also demonstrable with nuclear extracts prepared from Hepa-1 cells that had been treated in vivo with TCDD (lane 4). Again, no such complex was observed with TCDD-treated CHO-GNMT nuclear extracts (lane 8).

These results clearly indicate that CHO-GNMT cells lack any endogenous expression of Ah receptor and that the PAH-induced expression CYP1A1 occurs through a Ah receptor-independent pathway.

DISCUSSION

The 4 S PAH-binding protein was identified as GNMT on the basis of several criteria including purification, sequencing, immunoprecipitation of PAH binding activity by polyclonal antibodies to GNMT, and copurification of the two proteins in various cell lines and tissues (34). The present study was designed to introduce the GNMT gene into cells that lack the ability to express this protein as well as the Ah receptor. These cells would then provide a suitable biological model to test the induction by PAHs such as B[a]P and 3-methylcholanthrene. The results of this study provide direct information on the functional role of GNMT as a PAH-binding protein that may mediate the induction of CYP1A1.

In the present investigation, GNMT was stably introduced into CHO (D422) cells, which lack the endogenous expression of this protein as well as the Ah receptor, by means of a plasmid that contained a Moloney virus long terminal repeat as the promoter and exhibited neomycin resistance as the selection marker. If the hypothesis of GNMT, i.e. the 4 S PAH-binding protein, as a mediator of PAH-induced expression of CYP1A1 holds, the present system should respond to B[a]P in the appropriate manner. Our results demonstrate that the introduction of GNMT expression into these CHO cells does in fact cause the PAH-induced expression of CYP1A1. Furthermore, this interpretation is reinforced by the absence of detectable levels of Ah receptor in this cell line. We have also been unable to detect any mRNA for the Ah receptor in the parental, vector-transformed, or pMAMneo/GNMT-transformed cells (data not shown), nor have we detected any TCDD binding and XRE binding activities demonstrable in these cells.

Previously, B[e]P had been demonstrated to be a selective ligand for the 4 S PAH-binding protein (17-23). In the present study, we have shown that this PAH is also able to induce the expression of CYP1A1 in the GNMT-transfected CHO cells, whereas TCDD, a prototypic ligand for the Ah receptor, failed to do so. These results clearly demonstrate the role of GNMT as a mediator of CYP1A1 expression induced by various PAHs such as B[a]P, B[e]P, and 3-methylcholanthrene through an Ah receptor-independent pathway.

GNMT has a nucleotide binding region (43) and has been localized in rat liver nuclei by various immunohistochemical techniques (56). GNMT has also been proposed as a modulater of gene expression by methylation of substrates that have not yet been identified (43). It is conceivable that GNMT may act indirectly in modulating B[a]P-induced expression of CYP1A1 by methylating unknown substrates that could influence this gene, because hypermethylation has been shown by many laboratories to affect gene activity (57-62).

The present results indicate that GNMT exhibits diverse functions, one as an enzyme and the other as a transcriptional activator. This aspect of molecular economy is not unique to GNMT but is shared by several other proteins that have functions as both enzymes and activators. Several of the enzymes in the glycolytic pathway have been shown to be DNA-binding proteins (63). For instance, the 37-kDa subunit of glyceraldehyde-3-phosphate dehydrogenase also serves in a totally unrelated DNA repair function as uracil DNA glycosylase (64). One of the two primer recognition proteins that stimulate the activity of DNA polymerase- in replication has been identified as 3-phosphoglycerate kinase (42).

How and by what mechanism GNMT is able to fulfill its two unrelated functions in living systems need further investigation. GNMT constitutes about 1% of total cellular protein in rat and rabbit liver (35-37). Enzymatically, GNMT must be present as a homotetramer, which represents the major form (41). On the other hand, it is the dimeric form that serves as a PAH-binding unit and a transcriptional activitor (34). We have also recently reported (31) that the phosphorylation state of GNMT affects the binding of B[a]P and its nuclear translocation. It is conceivable that the extent of dimerization and subcellular localization of GNMT are governed by post-translational modification events as has been reported for other trans-acting factors (64). Our unpublished studies suggest that phosphorylation is involved in stabilizing the PAH-binding form (dimer) of the protein. Consequently, the rate-limiting factors in determining the amount of functional transcriptional activator are the phosphorylation state of the dimeric form of GNMT and the intracellular concentration of the PAH. The amount of homotetramer, i.e. the major form of the GNMT, is irrelevant.

In conclusion we have demonstrated that GNMT is a PAH-binding protein that mediates the induction of CYP1A1 by an Ah receptor-independent pathway. Additional investigations are required to better understand the factors that regulate the amount of dimeric GNMT and therefore control the expression of CYP1A1.