Identification of a novel cis-acting negative regulatory element affecting expression of the CYP1A1 gene in rat epidermal cells.

Polycyclic aromatic hydrocarbons such as 3-methylcholanthrene are toxic to rat epidermal cells in low passages (3 to 6), but cultures of high passage (>/=15) are resistant. Since such compounds can be metabolically activated by cytochrome P4501A1, we have examined the regulation of this gene in low and high passage cells. Consistent with this difference, little or no 2,3,7, 8-tetrachlorodibenzo-p-dioxin (TCDD)-inducible P4501A1 mRNA or enzyme activity was observed in high passage as compared to low passage cultures. Similarly, transfection of a luciferase reporter construct containing -1317 to +256 base pairs of the 5'-flanking region of the murine CYP1A1 gene was TCDD-inducible in low but not high passage cells. Ligand binding and transfection experiments demonstrated the presence of functional Ah receptor complexes in both high and low passage cells. Deletion analysis identified a 26-base pair negative regulatory DNA (NeRD) element contained within the upstream regulatory region of the CYP1A1 gene responsible for this effect. Nuclear extracts from both low and high passage cells contain a protein which specifically binds to NeRD-containing DNA. Thus, the loss of polycyclic aromatic hydrocarbon sensitivity in high passage rat epidermal cells appears to be due to decreased expression of CYP1A1, and this effect may be mediated by an altered NeRD binding factor(s) present in these cells.

Polycyclic aromatic hydrocarbons (PAHs) 1 are ubiquitous environmental contaminants, some of which have been identified as rodent (1) and human (2) skin carcinogens and are thus a threat to human health. Since the epidermis is the primary extrapulmonary barrier to toxic agents, understanding the mechanism by which PAHs are metabolically transformed by keratinocytes is important for assessing their potential risk. Elucidating the response of normal and neoplastic cells to these compounds will assist in evaluating the applicability of the rodent model to human pathology.
Numerous biochemical studies have demonstrated that PAHs induce cytochrome P4501A1 and are metabolized by this enzyme to DNA-damaging electrophiles that lead to mutations and cellular transformation (3). Mechanistically, the induction of cytochrome P4501A1 by PAHs and related chemicals such as 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD, dioxin) is mediated by the Ah receptor (AhR), a ligand dependent, DNA-regulatory protein to which these chemicals bind with high affinity (4 -8).
After ligand binding, two molecules of hsp90 dissociate from the ligand-AhR complex (9 -11), the receptor translocates into the nucleus (12,13), dimerizes with the Ah receptor nuclear translocator (ARNT) protein, and is converted into its high affinity DNA binding form (8,(13)(14)(15)(16). The binding of the heteromeric ligand-AhR complex to its specific DNA sequence, the dioxin responsive element (DRE), leads to DNA bending, chromatin disruption, increased promoter accessibility, and increased CYP1A1 gene transcription (17)(18)(19)(20)(21)(22). In addition to positive regulation, the CYP1A1 gene is also under negative control in which inducible and constitutive expression of the promoter can be affected (23)(24)(25). Primary cultures of epidermal cells are known to respond to PAHs and TCDD with the induction of P4501A1 and increase in PAH metabolism. Previous work has demonstrated that rat epidermal cells lose PAH responsiveness as they are passaged and become immortalized (26,27). In addition, the cells lose their 3T3 feeder layer dependence, become less stratified, dramatically increase their growth rate and colony forming efficiency and gradually lose their differentiated characteristics (27). 2 Previous studies of murine and human epidermal keratinocytes have demonstrated that PAH responsiveness (CYP1A1 inducibility) is dependent upon the differentiated state of the cells, with those committed to terminal differentiation being most responsive (28 -31). However, the mechanistic basis for the differentiation-dependent change in PAH responsiveness is currently unknown. Although it is not clear whether the spontaneous loss of PAH responsiveness in high passage rat keratinocytes is related to their loss of differentiated phenotype, the availability of a variety of molecular probes to the CYP1A1 system allows us to examine the mechanism of the spontaneous loss of PAH sensitivity in rat epidermal cells in detail. Here we demonstrate that resistance of high passage (HP) rat epidermal keratinocytes to PAH-dependent toxicity is due to reduced levels of inducible P4501A1 activity. Transfection and deletion analysis have identified a novel cis-acting negative regulatory DNA (NeRD) element present upstream of the CYP1A1 promoter that acts to negatively regulate TCDDinducible expression of the CYP1A1 gene in HP cells. Plasmid Constructs-The TCDD-inducible firefly luciferase plasmid construct pGudLuc1.1 was prepared as described previously (32). This construct contains the mouse mammary tumor virus (MMTV) long terminal repeat, inclusive of the viral promoter but lacking glucocorticoid responsive elements (33), as well as a 499-bp region (from Ϫ1317 to Ϫ818 bp upstream of the transcriptional start site of the mouse CYP1A1 gene) which contains four DREs. This DRE-containing fragment is directly upstream of the MMTV viral promoter and confers TCDD responsiveness upon the MMTV promoter and the luciferase gene (18,19,32,34). The construct pGudLuc4.1 contains a 1573-bp HindIII fragment isolated from the plasmid D16 (following conversion of the AflII site to a HindIII site). The fragment contains Ϫ1317 to ϩ256 bp of the mouse CYP1A1 gene which was then inserted into the HindIII site of the plasmid PGL 2 Basic immediately upstream of the firefly luciferase reporter gene. This fragment contains the CYP1A1 promoter, the putative negative regulatory domain, and four DREs (6,35). The construct pGudLuc4.2 contains a region from Ϫ248 to ϩ256 of the CYP1A1 regulatory region and includes the CYP1A1 promoter. The PGL 2 Basic vector was modified by the elimination of the BamHI and BglII sites to assist in making the pGudLuc5.0 deletion constructs. This involved cleavage of the respective sites by restriction digest, blunt ending with Klenow and religation. pGudLuc5.0 contains the same 1573-bp fragment present in pGudLuc4.1. The internal deletion constructs pGudLuc5.1, 5.2, 5.3, and 5.4 were prepared by restriction digest of pGudLuc5.0, blunting with Klenow and religating. The specific areas of the mouse 5Ј-flanking region deleted in each construct were as follows: pGudLuc5.1 (Ϫ643 to Ϫ248 bp deleted), pGudLuc5.2 (Ϫ818 to Ϫ448 bp deleted), pGudLuc5.3 (Ϫ643 to Ϫ448 bp deleted), pGudLuc5.4 (Ϫ818 to Ϫ643 bp deleted) of the mouse CYP1A1 5Ј-flanking region. Finer internal deletion analysis mutants of pGudLuc5.0 were prepared by PCR and the resultant constructs and the specific regions deleted are as follows: pGudLuc5.01 (Ϫ816 to Ϫ803 bp deleted), pGudLuc5.015 (Ϫ816 to Ϫ791 bp deleted), pGudLuc5.02 (Ϫ816 to Ϫ753 bp deleted), and pGudLuc5.03 (Ϫ816 to Ϫ704 bp deleted). The upstream primers were as follows: 5Ј-GCGGGATCCTCCTCCACCATACTTAGATAGC-3Ј (5.01), 5Ј-GCTGGATCCACTTAGATAGCTCTGCACCCG-3Ј (5.015), 5Ј-GCGGGATCCATGCCTGCTTCAGTATGTATGC-3Ј (5.02), 5Ј-GCGG-GATCCAGGGAGCTCGAAGCGGGGACT-3Ј (5.03), and the downstream primer was 5Ј-AAGAAAAAGAATAAAGGTTGGGG-3Ј. A BamHI site was added to each upstream primer, and the PCR products were digested with NcoI (internal in each PCR product) and BamHI as was the construct pGudLuc5.0. The PCR products were then ligated into the BamHI and NcoI sites of pGudLuc5.0 to recreate the upstream regulatory region containing the desired deletions.

Chemicals-TCDD
Cell Culture-Rat epidermal keratinocyte cultures of passages 2-30 were supported with an irradiated 3T3 feeder layer, while those of passages 30 -60 were propagated in the absence of 3T3. Previous experiments have demonstrated no difference in cell growth or PAH responsiveness (e.g. 3-MC induced toxicity or AHH inducibility) between high passage cells grown in the presence or absence of the 3T3 feeder layer (data not shown). Cell growth experiments were performed with feeder layer support. In experiments using low passage cells, the feeder layer was removed either by rinsing with 0.5 mM EDTA in PBS or by the advancing keratinocytes colonies as they reached confluence and therefore did not complicate interpretation of the results. Cells were grown in a 3:1 mixture of Dulbecco-Vogt Eagle's and Ham's F-12 media supplemented with 5% fetal bovine serum, hydrocortisone (0.4 g/ml), epidermal growth factor (10 ng/ml), cholera toxin (10 ng/ml), transferrin (5 g/ml), insulin (5 g/ml), adenine (0.18 mM), triiodothyronine (20 pM), and antibiotics (36). Mouse hepatoma (Hepa1c1c7) cells were grown as described previously (18,19).
Aryl Hydrocarbon Hydroxylase (AHH) Assay-AHH activity was determined as described previously (26,37). Near confluent cultures were incubated for 12 h with 5 nM TCDD or solvent alone (0.05% Me 2 SO), rinsed twice, and then incubated in 5 ml of serum-free medium containing 3.5 M [ 14 C]benzo(a)pyrene (30 Ci/mol). After 4 h the cultures were solubilized in 0.6 ml of 1 N KOH followed by the addition of 3.4 ml of Me 2 SO. Unmetabolized [ 14 C]benzo(a)pyrene was removed by two extractions with hexane, and the concentration of labeled metabolites in the aqueous layer determined by scintillation counting after neutralization with 1 N HCl. The protein concentration was determined by Coomassie G-250 dye binding (38) using bovine serum albumin as the standard.
Ah Receptor Assay-The sucrose density centrifugation assay was performed using cytosol from newly confluent cell cultures as described previously (39,40). Briefly, cytosol (Ϸ3 mg protein/ml) was incubated at 4°C with 10 nM [ 3 H]TCDD in the absence or presence of 1 M unlabeled TCDF competitor for 2 h, followed by removal of free and loosely bound [ 3 H]TCDD with dextran-coated charcoal (0.1 mg of charcoal/mg of cytosolic protein). Sample aliquots (300 l) were layered over 10 -30% sucrose density gradients with [ 14 C]formaldehyde-labeled serum albumin (4.4 S) and catalase (11.3 S) included as internal sedimentation markers (41). After centrifugation, fractions were collected and the radioactivity in each fraction determined by liquid scintillation.
Transfections and Luciferase Assay-Promega Profection and Luciferase Assay kits were used. Briefly, cells were grown to 50% confluency, residual 3T3 cells (if any) were removed with EDTA (45) and fresh medium was added 1 h before addition of precipitated DNA. Plasmid DNA, precipitated with 2 M calcium phosphate and 2 ϫ Hepes-buffered saline for 30 min, was added to cultures at 15 g/6-cm dish. Cultures were incubated for 24 h, washed once with serum-free medium and fresh medium with 5 nM TCDD or Me 2 SO was added for an additional 24 h. Cultures were harvested by scraping cells in lysis buffer (Promega), lysed by sonicating for 10 s, centrifuged, and luciferase activity of the cleared supernatants measured using a Turner luminometer. Luciferase activity was normalized to protein and the activity expressed as relative light units per mg of protein. Triplicate dishes were used for each condition; the values were averaged and expressed as a percentage of the control in each experiment. The controls in the first set of transfections were lysates from Me 2 SO-treated cells transfected with pGudLuc4.1. Since pGudLuc1.1 contained a different promoter (MMTV), TCDD-induced luciferase activity was compared to that of Me 2 SO-treated cells. For the remaining transfections the control was pGudLuc5.0, not treated with Me 2 SO. Each set of transfection experiments was performed a minimum of three times. The values graphed were expressed as the means (normalized to control values) and standard deviation among experiments.
Gel Retardation Analysis-Complementary pairs of synthetic oligonucleotides containing the sequences 5Ј-GATCTGGCTCTTCTCACG-CAACTCCG-3Ј and 5Ј-GATCCGGAGTTGCGTGAGAAGAGCCA-3Ј (corresponding to the AhR binding site of DRE3 (19) and designated as the DRE oligonucleotide) and 5Ј-GATCTCTTAAGGATCCTTCCTC-CACCCTTTCCTCCACCATACTTG-3Ј and 5Ј-GATCCAAGTATGGTG-GAGGAAAGGGTGGAGGAAGGATCCTTAAGA-3Ј (containing the two direct repeats within the NeRD element and referred to as the NeRD oligonucleotide), were synthesized, purified, and annealed, the NeRD oligonucleotide was radiolabeled with [␥-32 P]ATP using T 4 -polynucleotide kinase as described (46). Gel retardation analysis of nuclear extracts prepared from low and high passage rat keratinocytes and mouse hepatoma (Hepa1c1c7) cells were determined as described (18,19,46).

RESULTS
Rat epidermal cells lose their sensitivity to 3-MC inhibition of colony expansion between passages 3 and 15 (Fig. 1). Cultures were plated at low cell density and treated with 2 g/ml 3-MC for 14 days and then fixed and stained. A loss of sensitivity was noted beginning at passage 6 and was complete by passage 15. Colony forming efficiencies of the rat epidermal cells increased from 3% at passage 3 to 12% at passage 15 (data not shown) indicative of spontaneous immortalization (27). Later passages displayed the same results as passage 15 and were used subsequently for convenience.
The loss of 3-MC sensitivity in high passage cells was correlated with a decrease in constitutive and inducible AHH activity (Fig. 2). AHH activity was induced by TCDD treatment by 3.5-fold in low passage cultures and was not induced in high passage cultures. To determine whether reduction in inducible AHH was due to a reduction in mRNA levels of P4501A1 and/or another TCDD-inducible P450, CYP1B1, Northern blot analysis of message levels in the high and low passage cultures were examined. As is evident in Fig. 3, TCDD induced P4501A1 and 1B1 mRNAs in low passage cultures by a factor of 6-and 10-fold, respectively, while in high passage cultures the induction was only 1.2-and 0.9-fold, respectively. Thus, inducible expression of both CYP1A1 and 1B1 is suppressed in high passage cells.
To determine whether the reduction in inducible P4501A1/ 1B1 expression in high passage cells was due to a loss of the AhR, we measured AhR ligand binding using the sucrose density centrifugation assay (Fig. 4). These experiments demonstrated the presence of an 8.5-9.0 S [ 3 H]TCDD:AhR binding peak in both low and high passage cells. Moreover, quantitation of [ 3 H]TCDD specific binding revealed that low and high passage cells contained comparable levels of [ 3 H]TCDD specific binding (Ϸ37 fmol/mg of protein).
The above experiments measure only the ability of the AhR to bind ligand and do not address its ability to activate transcription in these cells. To address this, we carried out transient transfection experiments using a heterologous TCDDinducible reporter gene construct. This vector, pGudLuc1.1 (Fig. 5A), contains the firefly luciferase gene under control of the MMTV viral promoter and four DREs upstream of the CYP1A1 promoter, which confer TCDD/AhR responsiveness (32,33). Transient transfections of this vector into low and high passage cells revealed that TCDD induced luciferase expression by 4 -5-fold (Fig. 5B). These results not only confirm the presence of fully functional AhR complexes in both low and high passage cells, but they indicate that factors necessary for AhR-DRE interaction and transcriptional activity are also present in cells from both passages. These data also indicate that loss of 3-MC responsiveness in high passage cells is AhR-ARNT independent.
Although the transfection experiments with pGudLuc1.1 demonstrate functionality of the AhR-DREs, they do not address whether the loss of TCDD responsiveness is due to other cis-acting elements contained within the 5Ј-flanking region of the CYP1A1 gene (since pGudLuc1.1 contains only 499 bp of the upstream region) or whether it results from some alteration in normal chromatin structure within the regulatory region of the CYP1A1 gene. To address this uncertainty we carried out transfection experiments using a luciferase construct (pGud-Luc4.1) which contains 1573 bp of the upstream region of the murine CYP1A1 gene, inclusive of the CYP1A1 promoter, the putative negative regulatory region and the same four DREs present in the pGudLuc1.1 construct (Fig. 5A). Although the upstream region used in our studies is from the mouse instead of the rat CYP1A1 gene, sequence comparison indicates that the first 1800 bp upstream of the CYP1A1 gene in the rat (47) and mouse (33) are Ϸ95% identical. Thus, we would expect to observe similar results with the murine regulatory sequences if the repression of inducible activity is due to a conserved negative regulatory element. TCDD-inducible luciferase activity was observed in the low passage cells transfected with pGudLuc4.1 but not in those of high passage (Fig. 5B), mimicking the results obtained with the endogenous CYP1A1 gene. These results argue against alteration in chromatin structure as the cause of the loss of responsiveness in the high passage cells and suggest that the CYP1A1 5Ј-flanking region contains a cis-acting element responsible for the decreased responsiveness. Alternatively, the reduced level of expression could result from the lack of a factor necessary for normal promoter activity in high passage cells. To explore this possibility, low and high passage cells were transfected with a construct (pGudLuc4.2, Fig. 5A) containing only the CYP1A1 promoter (inclusive of the first 248 bp upstream of the transcriptional start site). Transient transfections with this construct resulted in high levels of constitutive promoter activity in both the low and high passage cells (Fig. 5B), confirming that loss of promoter function was not responsible for the loss of responsiveness in late passage cells.
The above results support the existence of a NeRD element present in the upstream region at a position which lies between the promoter and the DREs (within Ϫ248 to Ϫ818 upstream of the start site of transcription). To identify the specific region responsible for this effect, we prepared a series of constructs containing internal deletions within the remaining upstream region (Fig. 6A). TCDD-inducible luciferase activity was observed in low and high passage cells transfected with constructs pGudLuc5.2 and 5.4, while those transfected with 5.1 and 5.3 failed to induce (Fig. 6B). It is not clear why pGud-Luc5.3 exhibits such high constitutive activity in the low passage cells. A likely possibility is that since the promoter already appears to be very highly active, little or no additional increase in luciferase can be observed following TCDD treatment. Overall, the deletion results indicate the high passage-specific negative regulatory region is located in the region between Ϫ816 and Ϫ643.
To define the negative region more precisely, small scale deletion analyses in the Ϫ818 to Ϫ643 bp region were generated (Fig. 7A). Transfection of high passage cultures with pGudLuc5.01 (containing a deletion of the region between Ϫ818 and Ϫ803 bp (Fig. 7B), revealed low levels of inducibility relative to that of pGudLuc5.0. Full TCDD inducible activity was restored when the region between Ϫ816 and Ϫ791 bp was deleted (pGudLuc5.015). No significant increase in inducibility was observed with additional deletions. These results indicate that the negative regulatory effect in the high passage cells is modulated by a cis-acting NeRD element(s) contained within this 26-bp region. Moreover, inspection of this region suggests the presence of two NeRD elements, as shown in the alignment using upstream sequences from the mouse (32), rat (47), and human (48) CYP1A1 genes (Fig. 8). Sequence comparisons reveal a highly conserved direct repeat of CCTCCAC(C/T)(A/ C)T in the rat and mouse upstream regions; limited homology was observed with the human sequence. The observation that some TCDD inducible reporter gene activity was observed with pGudLuc5.01 (where only one of the direct repeat sequences was deleted) and full inducible activity was observed with pGudLuc5.015 (where both sequences were deleted) suggest that each of these NeRD elements confers some negative regulatory activity.
The above results indicate that the cis-acting NeRD element is critical for the observed negative regulation of the CYP1A1 gene, but the mechanism and factors involved in this effect remain to be resolved. The simplest explanation of the loss of PAH responsiveness in rat epidermal keratinocytes is that the presence of a NeRD-binding protein(s) in high but not low passage cells is responsible for the negative regulatory activity. To test this hypothesis, gel retardation analysis was carried out to examine the binding of nuclear proteins from low and high passage cells to a 45-bp oligonucleotide containing the direct NeRD repeat (Fig. 9). Although several protein-DNA complexes were revealed using extracts from both passages, no unique high passage protein-DNA complex was observed. In fact, only one protein-DNA complex was demonstrated to be specific for the NeRD-containing DNA oligonucleotide (i.e. it could be com- peted away by an excess of unlabeled NeRD-containing oligomer but not by an excess of DRE-containing oligomer) and this specific complex was present in both low and high passage nuclear extracts (Fig. 9). Additional experiments revealed no difference in protein-DNA complex formation using cytosol or nuclear extracts from Me 2 SO-or TCDD-treated low or high passage cells (data not shown). To determine whether the specific protein-NeRD DNA complex was more widely distributed or present only in rat epidermal keratinocytes, competitive gel retardation analysis was carried out using nuclear extracts from other cell lines. A protein-DNA complex which migrated to the same position and exhibited the same binding specificity as that observed using extracts from the rat keratinocytes was observed using mouse hepatoma (Hepa1c1c7) cells (Fig. 9). A similarly migrating protein-DNA complex was observed in all other lines tested (mouse hepatoma (MLE/BV) and fibroblast (3T3L1), guinea pig intestinal adenocarcinoma (GPC16), and human hepatoma (HepG2) and ovarian carcinoma (BG1) cells) although the quantity of the specific complex varied (data not shown). Thus, the protein(s) which binds to the NeRD-containing oligonucleotide appears to be similar in electrophoretic properties and fairly widely distributed among species and cell types. Whether it is actually the same or a related protein and has a role in the negative regulatory activity of the NeRD element remains to be determined and is the focus of ongoing studies.

DISCUSSION
The loss of AHH activity in high passage rat epidermal cells appears to be due to a dramatic reduction in inducible cyto-chrome P-450 3 and implies that the reduction in PAH metabolism is responsible for the reduction in cytotoxicity. The loss of PAH-inducible CYP1A1 expression in high passage cells is due to an alteration of one or more novel trans-acting factors that negatively regulate CYP1A1 gene expression in high passage cells and involves two novel cis-acting NeRD elements located between Ϫ816 and Ϫ791 bp upstream of the CYP1A1 gene.
In normal Ah-responsive cells the binding of the transformed AhR complex to the DRE results in derepression of the CYP1A1 gene promoter and activation of transcription. Previous studies have demonstrated several mechanisms by which the CYP1A1 gene is negatively regulated (23-25, 35, 49 -51). Putative negative regulatory elements which repress constitutive promoter activity have been identified by deletion studies (23,25,50,51). In the resting cell, little or no expression of CYP1A1 is observed. However, following TCDD treatment the transformed TCDD-AhR complex accumulates within the nucleus and the binding to DREs contained within the upstream regulatory region overcomes the inhibitory effect of the negative regulatory elements, resulting in high levels of CYP1A1 promoter activity (6,35). This inhibitory domain resides between Ϫ780 and Ϫ680 bp upstream of the transcriptional start site of the murine CYP1A1 gene and within the upstream regulatory region of the rat (Ϫ843 to Ϫ746) and human (Ϫ833 to Ϫ558) CYP1A1 genes and appears to consist of multiple negative regulatory elements (23,25). A second level of negative regu- lation of CYP1A1 has been described in which inducible expression of the gene fails to occur (24,52) because access to the DRE is blocked by a constitutively bound protein factor(s). The inability of transformed AhR complex to bind to the DRE prevents inducible expression of the CYP1A1 gene.
The negative regulatory activity identified in our studies is distinct from those described above in at least four ways: 1) our NeRD element is located further upstream than the previously identified cis-acting negative regulatory elements 2) the negative regulatory elements identified by other investigators function only to repress constitutive promoter activity and are overcome by TCDD treatment, while the region defined in our studies represses TCDD-inducible expression as well; 3) unlike the DRE-dependent mechanism of negative regulation, the region defined in our studies does not involve the DREs nor any putative DRE repressor proteins; and 4) the inhibitory effect described in our studies has been detected only in high passage rat keratinocytes, unlike previously described negative regulatory elements which function in a variety of cell types.
It is not yet clear how the NeRD elements contained within the Ϫ816 to Ϫ791 bp region exert their negative regulatory effect on the inducibility of the CYP1A1 gene, but deletion and transfection analysis has demonstrated that each of the NeRD elements exhibits some inhibitory activity on its own; full inhibitory activity required both elements. Studies in other biological systems suggest many scenarios that could explain the negative regulatory effect observed in our cells (53), the most likely of which is the silencing of expression through a direct interaction of a negative regulatory factor(s) with the NeRD element and subsequent interference with the formation or activity of the transcription machinery. Although it is possible that the NeRD-binding proteins could block or quench the functional activity of DREs themselves, they would have to negatively regulate at least 4 separate DREs which are distributed within 300 bp upstream of the NeRD element, a less likely scenario. Our gel retardation analyses, however, do not support the simplest possibility, the existence of a novel NeRD binding factor present only in high passage cells. Although it remains to be determined whether the NeRD-specific DNA binding factor identified here actually plays a role in the negative regulation of CYP1A1 gene expression, one can envision that an alteration in this factor (e.g. by phosphorylation) and/or a protein(s) to which it is associated occurs in high passage cells and is responsible for the inhibitory action. These changes would not likely be detected by the gel retardation assay used here, but would require more sophisticated analysis.
In addition to the specific factor which interacts with the NeRD element, it is not yet clear what specific nucleotides are involved in its regulatory activity. Sequence alignment of the NeRD elements reveals that they are highly conserved C-T rich sequences (Fig. 8) which show homology to C-T-rich negative regulatory elements previously identified in the upstream regions of a number of other genes, including: insulin (54), CYP2C11 (55), glutathione S-transferase P (56) and collagen II, ␤-interferon, c-myc, and lysozyme (57). The specific relationship, if any, between the NeRD element identified in this report and the above elements remains to be determined. The presence of two similar C-T-rich silencer elements in the upstream region of the CYP2C11 gene (55) could provide a possible explanation for the observed decrease in constitutive AHH activ- ity we observe in high passage cells, since this enzyme has been shown to be a major contributor to this activity (58). One can envision that increased activity/potency of negative regulatory factors in late passage cells could result in greater repression of this gene and a decrease in constitutive AHH activity, although this remains to be confirmed.
In addition to CYP1A1, we have observed that TCDD-inducible expression of the CYP1B1 gene is also negatively regulated in high passage cells (Fig. 3B). These results combined with the observed decrease in constitutive AHH activity suggest that this negative regulatory mechanism is not specific for CYP1A1 but may be a more general response in these cells. Although it remains to be determined whether other members of the Ah gene battery are also negatively regulated, analysis of the upstream regulatory sequences of the rat glutathione S-transferase Ya (59), quinone reductase (60), and aldehyde dehydrogenase-3 (61) have revealed several C-T-rich regions with homology to the NeRD element. Comparison of the inducible expression of these genes in low and high passage cultures will provide insight into the specificity/generality of this negative regulatory response.
Coincident with the loss of PAH sensitivity in high passage rat epidermal keratinocytes are a variety of changes in gene expression associated with cellular transformation, such as increased colony forming efficiency and decreased feeder layer dependence (27). A linkage of the loss of PAH sensitivity to immortalization and gradual loss of differentiation characteristics seems possible. In fact, studies in mouse and human epidermal keratinocytes (29,31) demonstrated that inducible expression of CYP1A1 is dependent upon the cellular differentiation state of the cells, with those committed to terminal differentiation as being responsive to the inducing effects of TCDD. Thus, the negative regulatory effect observed in high passage rat keratinocytes could reflect a change in the differentiated characteristics of these cells and raises the possibility that the NeRD element and/or its associated binding protein(s) may have a role in regulating expression of other genes affected by changes in the cellular differentiation state. Determining the involvement of NeRD elements in the differentiation induced changes in CYP1A1 in mouse and human keratinocytes could readily be accomplished using the reporter constructs and approaches described here. Such experiments may even help identify factors involved in differentiation-induced changes in expression of other genes. Thus, characterizing the function of the NeRD element and its binding factor(s) may provide insight into the mechanism of negative regulation not only of CYP1A1 but possibly a battery of other genes as well. FIG. 9. Binding of nuclear proteins from low and high passage rat epidermal keratinocytes and mouse hepatoma (Hepa1c1c7) cells to a NeRD-containing DNA oligonucleotide. Nuclear extracts from the indicated cells were incubated with a 32 P-labeled NeRD-containing oligonucleotide (100,000 cpm, Ϸ1 ng) in the absence or presence of a 100-fold molar excess of unlabeled NeRD-or DRE-containing oligonucleotide and incubations subjected to gel retardation analysis as described under "Materials and Methods." The arrow indicates the position of the specifically bound protein-DNA complex.