Specific Deactivation of the Mouse Mammary Tumor Virus Long Terminal Repeat Promoter upon Continuous Hormone Treatment*

We have studied the transcriptional behavior of the mouse mammary tumor virus long repeat (MMTV-LTR) promoter during a prolonged exposure to glucocorticoids. When integrated into XC-derived cells, MMTV-LTR expression reached its maximum during the first day of dexamethasone treatment, but longer exposure to the hormone resulted in the deactivation of the promoter. In contrast, glucocorticoid-responsive resident genes or MMTV-based transiently transfected plasmids maintained or even increased their mRNA levels during the same period of hormone treatment. An integrated chimeric construct containing the hormone-responsive elements from MMTV-LTR but in different sequence context became also deactivated after a prolonged hormone treatment but with a deactivation kinetics significantly slower than constructs containing the entire, chromatin-positioning MMTV-LTR sequence. The decrease on MMTV-LTR-driven transcription was concomitant with a parallel closure of the MMTV-LTR chromatin and with a decrease in glucocorticoid receptor (GR) concentration in the cell. We concluded that the chromatin-organized MMTV-LTR promoter is particularly sensitive to any decrease on GR levels. We propose that chromatin structure may contribute decisively to the differential expression of MMTV-LTR by two mechanisms: limiting MMTV-LTR accessibility to activating transcription factors and accelerating its shutting down upon a decrease on GR levels.

The mouse mammary tumor virus long terminal repeat (MMTV-LTR1) has proved to be one of the most fruitful models for steroid hormone-regulated promoters since the beginning of the 1980s, when several groups defined the concept of hormone-responsive element (HRE). 1 The MMTV-LTR promoter can be activated by glucocorticoids and progestins through the interaction of the hormone-receptor complex with HREs. The MMTV-LTR HREs are termed GRE/PREs (glucocorticoid/progestin response elements) because they bind either glucocorticoid or progestin receptors (GR or PR). GRE/PREs contain two TGTTCT motifs, separated by three nucleotides and arranged in a palindromic structure. In the MMTV-LTR, hormone inducibility depends on the occurrence of one palindromic GRE/PRE and three hemi-palindromes. These GRE/PREs are necessary and sufficient to confer hormone inducibility to homologous and heterologous promoters when placed in their vicinity (1,2). Each of these GREs is recognized and bound by a homodimeric hormone-loaded GR or PR, subsequently triggering or enhancing transcription. Two other steroid hormone receptors, the androgen receptor and the mineralocorticoid receptor, have been found to bind to the same GREs and subsequently activate MMTV promoter (see Ref. 2 for a recent review). GRE sequences are found in regulatory regions of many glucocorticoid-and progestin-responsive genes, like the chicken or the frog ovoalbumin gene, the chicken lysozyme gene, the rat tyrosine aminotransferase gene, or metallothionein (MT) genes from different mammalian tissues (1).
An unsuspected twist was given to the study of MMTV with the discovery of very specific positioning of nucleosomes over a long stretch of the MMTV-LTR sequence (3). One of these positioned nucleosomes, termed nucleosome B, covers the four GREs present in the MMTV-LTR; it coincides in position with a DNase I-hypersensitive site that appears upon hormone induction (3,4). Nucleosome B reconstituted in vitro allows binding of both GR and PR while excluding other transcription factors that also bind to the same DNA sequence when free from histones (5-7). One possible role of the steroid hormone receptors may be the removal or structural alteration of nucleosome B to allow transcription factor binding and therefore enhance transcription (6 -9).
One of the most compelling arguments in favor of the role of chromatin structure in transcriptional activation is the different behavior of transiently transfected hormone-regulated promoters when compared with their chromosomal counterparts (10). Recent studies show that a proper chromatin structure is necessary for the correct expression pattern of the MMTV-LTR promoter both in mammalian (11) and yeast cells (12). In the case we present here, MMTV-based expression constructs transfected into XC cells showed a loss of hormone response upon continuous treatment with hormone. This deactivation was not observed in resident hormone-responsive genes or when the same MMTV constructs were assayed in transient transfection experiments. Although this effect could be due to the presence of a specific repressor in hormone-treated cells, our data favor a mechanism in which deactivation correlates with an exhaustion of GR receptor upon continuous hormone administration. We propose that the response of the different genes and reporters to the decline of the receptor may be related to the ability of each promoter to shut itself down when receptor levels decay under a certain minimum. This mechanism is likely affected by the chromatin structure over the relevant sequences of the promoter. We also propose that such a mechanism may be important for the differential expression of the different hormone-regulated genes in the cell.

Plasmids and Cell Lines
Plasmids-pMS-H5 (13), pAGE5MMTVLu (14), pMMTV-CAT (15), and pTATT-CAT (16) have been previously described; a brief description is given in the text when necessary. pRSVLUC is a construct where the Rous sarcoma virus promoter (17) directs the expression of luciferase. pTATT-AGE5Lu is a derivative of pAGE5MMTVLu where the Sma/PvuII fragment was replaced by the HindIII fragment of pTATT-CAT treated with DNA polymerase I (Klenow fragment). This construct has the same characteristics of pAGE5MMTVLu, except that the luciferase gene is driven by the thymidine kinase promoter and made hormone-responsive by two copies of a synthetic, pseudopalindromic GRE. pMS-HSP27 is an unpublished construct identical to construct pMS-H5, except that the histone H5 sequences were replaced by hamster hsp27 coding sequences; it was a generous gift from Prof. J. Landry (Université Laval, Québec, Canada).
Cell Lines-Two cell lines derived from epithelial-like XC rat sarcoma cells were used. XC-10 cells (13) contain about 300 copies of the plasmid pMS-H5 (a contribution from Prof. A. Ruiz-Carrillo, Laval Université, Québec, Canada). XC-14 cells contain a single integrated copy of pMS-HSP27. 2 XC-8 is a nontransfected XC clone used as a control (13). Cell lines H12 and B11 were generated from stable transfection of XC-8 cells with plasmids pAGE5MMTVLu and pTATT-AGE5Lu, respectively (see below).

Cell Culture and Transfections
Cell Culture-All cell lines were grown in Dulbecco's modified Eagle's medium containing 1 mg/ml glucose, 100 IU/ml penicillin, 100 g/ml streptomycin and supplemented with 2 mM L-glutamine and 10% (v/v) of heat-inactivated fetal calf serum. Cultures were maintained at 37°C in a humidified atmosphere containing 5% CO 2 . Except for the nontransfected XC-8 cells, stock cultures were maintained in the presence of 350 g/ml G418 (Sigma); cells split out for experiments were cultured without antibiotic.
Transient Transfection Experiments-XC-8 cells were transiently transfected by the DEAE-dextran method (18,19). One million cells were incubated at room temperature for 30 min with a mixture containing 5 g of plasmid DNA and 0.25 mg DEAE-dextran in 0.5 ml of 1 ϫ TBS (25 mM Tris-HCl, pH 7.4, 137 mM NaCl, 5 mM KCl, 0.7 mM CaCl 2 , 0.5 mM MgCl 2 ). After incubation, the cells were treated for 2 min at room temperature with culture medium containing 15% Me 2 SO, washed in TBS, further incubated for 2.5-3 h at 37°C, 5% CO 2 in medium containing 0.1 mM chloroquine diphosphate, and finally returned to standard culture medium. Hormone was added to the medium 48 h after transfection, and incubation was continued for 3-4 days before harvesting the cells to assay luciferase and chloramphenicol acetyltransferase (CAT) activity. All transfections and treatments were performed in duplicate.
Establishment of Stable Cell Lines-XC-8 derivatives containing the MMTV-luciferase construct were obtained by calcium phosphate precipitation (20) with 30 g of the plasmids pAGE5MMTVLu or pTATT-AGE5Lu per 9-cm plate (approximately 2.5 million cells). For the selection of transfected cells, the medium was supplemented with 350 g/ml G418. Stable G418-resistant clone pools were seeded at limit dilution, and several single clones were isolated and characterized.

Primer Extension Analysis
Total RNA (20 g) was hybridized in 50 mM Tris-HCl, pH 8.3, 75 mM KCl, 3 mM MgCl 2 , 10 mM dithiothreitol with 0.2 pmol of 32 P-labeled 5Ј end-labeled primer for 4 h at 37°C. The primer was extended in this reaction mixture after the addition of 0.5 mM of each deoxyribonucleotide, 50 ng/ml of actinomycin D, 25 ng/ml of bovine serum albumin, and 200 units of MMTV-reverse transcriptase (Life Technologies, Inc.) for 1 h at 37°C. The reaction was stopped by the addition of 12.5 mM EDTA and 25 ng/ml of pancreatic RNase A. Samples were extracted once with phenol/chloroform/isoamyl alcohol (25:24:1), ethanol precipitated and analyzed on 8% acrylamide-urea sequencing gels. The primers used were: rat actin, 5Ј-GGGACGGAGGAGCTGCAAG-3Ј, and hamster hsp27, 5Ј-GCAGCAGCGAGAAGGGCAC-3Ј. The 5Ј end of the rat actin probe is 55 bases downstream from the rat actin mRNA start site; the hamster hsp27 probe is 321 bases downstream from the origin of transcription driven by MMTV in construct pMS-HSP27. Relative quantitation of the bands was performed using a transmittance scanning densitometer (Molecular Dynamics).

CAT and Luciferase Assays
Cell monolayers were rinsed twice in phosphate-buffered saline and lysed by the addition of 0.6 ml of lysis buffer (25 mM Glycyl-glycine, pH 7.8, 1 mM EDTA, 1 mM dithiothreitol, 1% Triton X-100, 15% glycerol). The lysate was transferred to a microfuge tube and centrifuged for 3 min. 50 l of the supernatants was added to 0.35 ml of luciferase assay buffer (25 mM glycylglycine, pH 7.8, 15 mM MgSO 4 , 5 mM ATP). Luciferase-mediated light output was determined on a Lumat LB 9501 luminometer (Berthold) by injection of 100 l of 1 mM luciferin and integration of the peak of light emission for 30 s. CAT assays were performed with a CAT-enzyme-linked immunosorbent assay kit (Boehringer Mannheim), following the manufacturer's instructions. Protein concentrations of extracts were determined according to Bradford using a commercial kit (Bio-Rad).

DNase I Hypersensitivity Assay
Approximately 200 million cells of the appropriate cell lines were incubated with 1 mM dexamethasone for the indicated periods of time and scraped into phosphate-buffered saline, and the nuclei were isolated as described previously (23). The reaction was initiated by the addition of DNase I (Promega, 1 units/ml) and the tubes were incubated at 30°C. After 10 min the reaction was stopped by the addition of 5 volumes of stop solution (100 mM Tris-HCl, pH 7.4, 10 mM EDTA, 0.5 mg/ml proteinase K, 1% N-lauryl-sarcosine solution) and incubated for 3-16 h at 37°C. DNA was purified by two phenol/chloroform/isoamyl alcohol extractions and two chloroform extractions, ethanol precipitated, and resuspended in TE (10 mM Tris, pH 7.4, 1 mM EDTA) buffer. For Southern blot analysis, 40 g of DNA from each sample was digested to completion with the appropriate restriction enzymes and separated electrophoretically through 0.8% agarose gels in TEA buffer (40 mM Tris acetate, 1 mM EDTA). DNA was transferred to a nylon membrane (Hybond N, Amersham Corp.) in 20 ϫ SSC and cross-linked by UV treatment (Stratalinker, Stratagene). Membranes were hybridized with randomly primed 32 P-labeled DNA probes either from a Hin-dIII/XhoI fragment from construct pMS-HSP27 encompassing the MMTV-LTR sequences (XC-14 cells) or from a AvaI/EcoRI fragment from the luciferase gene (H12 cells).

GR Analysis by Western Blot
Western blots for the GR were performed according to Ref. 24. XC-14 or H12 cells were treated with 1 M Dex for 0, 1, and 4 days. At each time, cell monolayers were washed with phosphate-buffered saline and suspended in hypotonic buffer (10 mM Hepes, pH 7.9, at 4°C, 1.5 mM MgCl 2 , 10 mM KCl, 0.2 mM phenylmethylsulfonyl fluoride, 0.5 mM dithiothreitol). The swollen cells were homogenized with a Dounce pestle B (20 strokes) and centrifuged. Pelleted nuclei were resuspended in 100 l of 20 mM Hepes, pH 7.9, at 4°C, 25% glycerol, 1.5 mM MgCl 2 , 20 mM KCl, 0.2 mM EDTA, 0.2 mM phenylmethylsulfonyl fluoride, 0.5 2 J. Landry, unpublished data. mM dithiothreitol. Nuclear proteins were extracted by adding 100 l of the same buffer containing 0.45 M KCl. The S-100 fraction was prepared by adding 0.11 volumes of 10 ϫ cytoplasmic extract buffer (0.3 M Hepes, pH 7.9, at 4°C, 1.4 M KCl, 30 mM MgCl 2 ) to the cytoplasmic fraction and centrifugation at 100,000 ϫ g for 1 h. The protein concentration in the extracts was determined by the Bradford assay (Bio-Rad). 70 g of protein from each extract was separated on an 8% SDS-polyacrylamide gel and transferred onto a nitrocellulose membrane. The membrane was blocked with 5% nonfat dry milk in 50 mM Tris, pH 7.6, 0.2 M NaCl, 0.05% Tween 20 and incubated with a rat liver GR polyclonal antibody (kindly provided by Pr. O. Wrä nge, Uppsala, Sweden) at 1:200 dilution for 2 h at room temperature. The membrane was washed extensively and incubated with a horseradish peroxidase-conjugated rabbit anti IgG (Promega) at 1:10,000 dilution for 1 h. The membrane was washed again and GR was detected with the Amersham Corp. ECL Western blotting Detection kit. Relative quantitation of the bands was performed using a transmittance scanning densitometer (Molecular Dynamics).

Kinetics of MMTV-LTR Deactivation in XC Cells-Inte-
grated constructs containing MMTV driving either the chicken histone H5 or the hamster hsp27 genes showed a tight transcriptional regulation by glucocorticoids, with no detectable basal expression without hormone (Fig. 1, A and B, ϪDEX  lines). Administration of the synthetic glucocorticoid Dex resulted in a dramatic induction of the chicken histone H5 mRNA in XC-10 cells (Fig. 1A, top panel) and of the hamster hsp27 mRNA in XC-14 cells (Fig. 1B, top panel). In addition, Northern blot analysis of hormone-induced and noninduced XC-14 cells probed with the hamster hsp27 cDNA revealed an endogenous mRNA that was also increased by glucocorticoids (Fig. 1B, band labeled X). We tentatively identified this transcript as the ␣B-crystallin mRNA, based on its size of less than 1 kb (25), its glucocorticoid responsiveness (26), its cross-hybridization with the hamster hsp27 cDNA, and the epithelial-like character of XC cells. This was confirmed by rehybridizing the same blots with an ␣B-crystallin-specific oligonucleotide complementary to the 3Ј end of the rat transcript (see "Experimental Procedures"). As shown in Fig. 1B (second panel), this probe revealed a clear increase in the amount of the ␣B-crystallin transcript after hormone administration. We have also detected Dex induction of the rat metallothionein-I gene (MT-I), which is a well known glucocorticoid-responsive gene (27)(28)(29)(30). As shown in Fig.  2B, its mRNA levels increased two or three times after Dex administration in the XC-14 cells (Fig. 1A, third panel, and Fig.  2B). Simultaneous addition of Dex and cycloheximide did not prevent hormonal activation of the ␣B-crystallin or MT-I genes (not shown), indicating that their response to Dex did not require protein synthesis.
XC-10 and XC-14 cells incubated with Dex for several days showed a continuous fading of transcripts from both histone H5 and hamster hsp27, respectively. By day 5, levels of both transcripts were reduced to 10 -20% of their maximal value (Fig. 1,  A and B). In contrast, ␣B-crystallin mRNA levels did not fade after at least 5 days of continuous hormone administration (Fig. 1B, top and second panels). Likewise, MT-I mRNA concentration showed an increase, rather than a decrease, after 5 days of continuous hormone treatment (Figs. 1B and 2B). These results suggest a specific deactivation of the MMTV promoter not affecting other GR-regulated genes. It is noteworthy that the two MMTV-driven transcripts (histone H5 and hamster hsp27) decayed at an essentially identical rate, despite the different phenotypical effects of Dex administration on each cell line. In the continuous presence of Dex, XC-14 cells continued to divide at their normal rate (without Dex) of about one division every 18 h, similar to the growth rate of the parental XC-8 cells. In contrast, Dex induction of the expression of chicken histone H5 in XC-10 cells led to the arrest of cell growth (not shown). This effect may be a consequence of the incorporation of histone H5 into the host cell chromatin and the subsequent changes on chromatin structure (13). XC-10 cells reinitiated cell division on day 4 or 5, when histone H5 protein levels are down to 10 -20% of the amount on day 1 (13). 3 The behavior of MT-I and ␣B-crystallin mRNAs in XC-10 and XC-14 cells was essentially the same; results from both cell lines are included on the data of Fig. 2B.
To demonstrate that the presence of the hormone-receptor complex was required not only to start transcription but also to maintain the transcription of the Dex-responsive genes, we analyzed the behavior of the different activated mRNAs upon Dex removal. Removal of Dex 24 h after its administration resulted in a relatively rapid decay of the mRNAs from all analyzed GR-activated genes. Transcription of both the MMTV-derivatives and the resident MT-I returned to basal levels by 12 h after Dex removal, whereas the ␣B-crystallin gene decayed at a somewhat slower ratio (Fig. 1C). The apparent mRNA half-lives of the different genes were relatively short: less than 5 h for the hamster hsp27 mRNA, about 12 h for the ␣B-crystallin mRNA, and less than 6 h for the MT-I mRNA. These half-lives are clearly shorter than the time window we are considering for our kinetics experiments (3 or 4 days); therefore, it is likely that mRNA stability has only minor effects on the rates of the observed decays for MT-I, ␣B-crystallin, or hamster hsp27 transcripts.
We ensured the correct initiation of the transcripts over the period examined by primer extension analysis of MMTV-driven hamster hsp27 transcript in XC-14 cells was performed. As seen in Fig. 2A, the amount of correctly initiated transcript increased after hormone administration and decreased upon continuous treatment closely following the pattern of total mRNA decay observed in Northern blots. Fig. 2B shows a quantitation of the correctly initiated MMTV-driven hamster hsp27 compared with the amount of total mRNA as observed in the blots of Fig. 1. The decrease of the correctly initiated hamster hsp27 transcript (line) closely followed the disappearance of the total hsp27 mRNA in XC-14 cells or H5 mRNA in XC-10 cells (bars). On the contrary, neither MT-I nor the ␣B-crystallin transcripts showed any decrease in their levels during the examined period.
Construction and Hormonal Response of Cell Lines Containing Single-copy, Integrated MMTV Reporters-We produced a series of stable cell lines derived from the parental XC-8 cells containing the hormone-responsive, luciferase reporter plasmids pAGE5MMTVLu and pTATT-AGE5Lu. pAGE5MMTVLu contains a 1.4-kb fragment from MMTV-LTR that includes the complete set of HREs. pTATT-AGE5Lu contains two copies of a pseudo-palindromic HRE (5Ј-GGTTACAAACTGTTCT-3Ј), which confer hormonal response to the adjacent thymidine kinase promoter. These plasmids also include a neomycin resistance gene and a scaffold-attachment region of the human interferon ␤ domain (14). The presence of a scaffold-attachment region helps plasmids to integrate in the cell genome mainly as single, nonrearranged copies, with very little position dependence (9,14). Two neomycin-resistant clones showing clear glucocorticoid induction were selected for analysis of their behavior upon continuous hormone treatment: H12 cells, transfected with pAGE5MMTVLu, and B11 cells, transfected with pTATT-AGE5Lu.
Our results showed clear differences among the deactivation kinetics of the two integrated reporters. As seen in Fig. 3, the luciferase activity was strongly induced in H12 cells 8 h after Dex administration (about 40-fold), but it faded upon continuous hormone treatment. Although not identical, this deactivation kinetics resembled the integrated MMTV constructs in XC-10 or XC-14 cells; by day 4, luciferase activity was only 20 -25% of its value at day 1. B11 cells showed clearly lower activation upon hormone administration (4 -5-fold), but although some decrease was observed, the luciferase activity remained at 80% of its maximal value until day 2, decreasing slowly thereafter. Therefore, the response of the pTATT- AGE5Lu construct in B11 cells to the continuous presence of Dex was clearly different from the very closely related pAGE5MMTVLu construct in H12 cells. pAGE5MMTVLu has been reported to show the typical MMTV nucleosome positioning over its 1.4 kb of the MMTV-LTR sequences when integrated in the genome (9). On the contrary, plasmid pTATT-AGE5Lu contains no MMTV sequences except for two pseudopalindromic HREs, identical in sequence to the distal HRE of the MMTV-LTR.
Hormonal Response of Transiently Transfected Constructs upon Continued Hormone Administration-Plasmids transiently transfected by the dextran-sulfate method are kept in the receiving cells for several days (31). As seen in Fig. 4A, a transiently transfected construct based on the Rous sarcoma virus promoter showed a robust transcriptional activity for 5 or 6 days after transfection. However, the amount of transcription decreased very rapidly during the first 2 or 3 days, stabilizing afterward. Between days 3 and 6 after transfection there is only a moderate loss of transcription, on a rate of about 15% decrease per day. We have used dextran-sulfate transiently transfected plasmids in this time window to study the effects of continuous administration of hormone on their transcription. In addition to the relative stability of transfected DNA, the use of this time window has the advantage that by day 3 the amount of total DNA in the cells has decreased considerably (31), reducing the effects of the presence of an excess of GR binding sites into the cells.
We transiently transfected XC8 parental cells with two different hormone-responsive promoters, and we analyzed the response to continuous Dex administration. pMMTV-CAT and pTATT-CAT have identical responsive elements, pAGE5M-MTVLu and pTATT-AGE5Lu, respectively, but drive the CAT gene instead of the luciferase gene. As seen in Fig. 4B, both plasmids produced a continuous increase in the levels of CAT protein through all the examined period, up to 4 days of continuous hormone administration. Removal of the hormone after 2 days of induction resulted on a steady decrease on the levels of CAT protein; control cells continuously exposed to Dex showed a further increase of CAT levels during the next 2 days (Fig. 4C). From these results, we concluded that the increasing levels of CAT protein upon the prolonged hormone exposure reflected a continuous transcription of the gene all over the studied period. This behavior was more similar to the response of the resident ␣B-crystallin and MT-I genes than to the integrated MMTV-driven constructs.
In addition to a difference in transcription levels, a possible explanation for the differences we observed between the stable and transiently transfected constructs could come from protein or mRNA stability. To check these possibilities, we transiently transfected XC8 cells with pMS-HSP27, the same construct used to generate XC-14 cells. Fig. 4D shows the levels of hamster hsp27 mRNA from this construct analyzed by Southern blot after 1 and 4 days of continuous hormone administration. The data showed an accumulation, rather than a decrease, of the transcript from transiently transfected cells from day 1 to day 4. This is a sharp contrast with the behavior of the very same construct integrated in XC-14 cells, where it fades after day 1 of hormone administration (see Fig. 1B, top panel). Thus, whatever the mechanism responsible for the decay of chromosome-integrated MMTV-driven transcription upon continued hormone administration might be, it seems not to apply to MMTV constructs transiently transfected into XC cells.
Chromatin Structural Changes in the MMTV-LTR during MMTV Deactivation-MMTV-LTR chromatin displays an array of very precisely positioned nucleosomes over at least 1.2 kb (3). Upon hormone administration, a DNase I-hypersensitive site appears between positions Ϫ200 and Ϫ100 relative to the transcription initiation site. This has been explained as a destabilization of one of the positioned nucleosomes, the so-called nucleosome B (3,4,9). This DNase I-hypersensitive site seemed to parallel transcriptional activation; therefore, we investigated its appearance on the integrated MMTV-hsp27 construct during the process of induction and deactivation.
A clear DNase I-hypersensitive site appeared upon Dex administration in H12 cells (Fig. 5A), mapping, as expected, at nucleosome B (3). Upon continued hormone treatment, the hypersensitive site became clearly weaker. Fig. 5B shows a quantitative analysis of the appearance and fading of the hypersensitive site in H12 cells. Maximal DNase I cutting was observed 1-4 h after Dex administration, decreasing steadily from then on (30% of maximal cutting at 24 h and 10% of maximal cutting at 96 h; Fig. 5B, bars). The appearance and fading of the DNase I-hypersensitive site preceded by several hours the activation and posterior fading of luciferase transcription (Fig. 5B). Comparable results were obtained in XC-14 cells (not shown). In this case, the partial fading of the hypersensitive site on nucleosome B coincided temporally with the decrease of MMTV-driven hamster hsp27 transcription. We interpret the reduction in hypersensitivity as a refolding of nucleosome B. This nucleosome seems to be necessary and sufficient to avoid binding of key transcription factors to the MMTV-LTR promoter (2,6,7,11,12); therefore, its refolding should result in the shutting down of the promoter.
Effect of Continuous Hormone Treatment on GR Levels-GR expression is down-regulated by the GR-hormone complex (32,33). Our results are consistent with a loss of GR upon continuous hormone exposure, resulting in a simultaneous decay in transcriptional activation and a loss of the structural changes on chromatin occurring upon GR binding. To investigate this hypothesis, we have followed the GR distribution in XC-14 and in H12 cells upon continuous treatment with Dex. The results of a Western blotting of nuclear and cytoplasmatic extracts of both cell lines are presented in Fig. 6. In both cases, the vast majority of GR was present in the cytosol in untreated cells, being readily translocated into the nucleus shortly after Dex administration. After 4 days of continuous hormone administration, GR was located both in the nucleus and in the cytosol, and its amount was reduced by 70 -80%. A quantitative analysis of these changes is shown in Fig. 6B.
The similar behavior of the levels of GR (Fig. 6B), the intensity of the DNase I-hypersensitive site (Fig. 5B), and the integrated MMTV-LTR transcription rate (Fig. 2B) as a response to a prolonged Dex administration is noteworthy. This similarity is lost in the case of transiently transfected reporters or of resident, GR-responsive genes (Fig. 2B). We concluded that the MMTV-LTR transcription rates followed the actual levels of GR more closely than other genes or reporters analyzed. This may indicate that different promoters could respond differently to the loss of GR, suggesting that the amount of hormoneloaded GR required to sustain hormone-activated transcription may vary from one gene to another. DISCUSSION We have analyzed the behavior of different glucocorticoidresponsive promoters after up to 5 days of continuous presence of hormone. We found a clear response to glucocorticoids for different MMTV-LTR-derived constructs as well as for two resident genes, the rat MT-I gene and an mRNA we identified as corresponding to the ␣B-crystallin gene. To our knowledge, this is the first report of ␣B-crystallin expression in XC cells.
The results presented here showed a deactivation of MMTVdriven promoters upon continued hormone administration in XC-derived cell lines. This deactivation did not affect either resident glucocorticoid-responsive genes or transiently transfected reporters. This phenomenon may be related to the reported transient response to Dex of MMTV constructs on BPVbased episomes, whereas transiently transfected plasmids FIG. 4. Hormonal response of transiently transfected plasmids upon prolonged hormone treatment. A, XC-8 cells were transiently transfected with the pRSV-LUC plasmid, and the luciferase activity (RLU/mg protein) was measured at different times after transfection. B, pMMTV-CAT (dark shading) and pTATT-CAT (light shading) were transiently transfected into XC-14 cells, and the CAT amount was measured by enzyme-linked immunosorbent assay 1 and 4 days after Dex administration. Ordinates indicate the amount of CAT protein relative to noninduced cells (day 0). C, XC-14 cells transiently transfected with pMMTV-CAT were treated with Dex for 2 days (black circles). After 2 days, Dex was removed from half of the plates, and the cells were incubated further 2 days (white circles). Control cells were continuously exposed to hormone (black circles). The ordinates indicate amount of CAT protein in arbitrary units. D, XC-8 cells were transiently transfected with pMS-HSP27, and hamster hsp27 mRNA was detected on a Northern blot as described after 1 and 4 days of continuous hormone treatment. Lane labeled as 0 corresponds to untreated cells. Given our protocol of hormone treatment of transiently transfected cells, day 1 of hormone treatment (B-D) corresponds to day 3 after transfection (A). remained actively transcribed (10). We propose that the differential kinetics of activation and deactivation of different hormone-responsive genes may be a general phenomenon, and we believe that this should have important physiological implications.
Our data showed that the deactivation kinetics of the MMTV promoter did not depend on the transcription product. Transcripts of three very different genes, such as luciferase, hamster hsp27, and chicken histone H5, faded at the same rate. Furthermore, we have observed a similar specific MMTV deactivation in completely different cell lines. 4 In addition, the deactivation of the MMTV promoters was independent from the cell cycle, because XC-10 cells stop dividing upon Dex administration, whereas XC-14 and XC-8 cells keep on dividing at the normal rate (13). In turn, deactivation of the transcription of a MMTV-driven reporter depended on whether the construct was integrated into the host cell genome, suggesting a critical role of MMTV-chromatin in the process. This model is reinforced by the differential kinetics of deactivation observed on H12 and B11 cells. These two cell lines contain integrated reporters very similar in structure, except for their hormoneresponsive regions. pAGE5MMTVLu (H12 cells) contains a long fragment from MMTV that has been shown to promote nucleosome positioning (9). pTATT-AGE5Lu (B11 cells) contains only the very same HRE found in the MMTV-LTR. In this case, the fact that the transcription from the former construct decreased much faster than that from the latter can neither be explained by differential stability of the gene products nor by differences on the HRE sequences.
Chromatin structural rearrangements of the MMTV-LTR upon glucocorticoid induction are revealed by the appearance of a DNase I-hypersensitive site (3,4,9). We have observed the appearance of this hypersensitive site upon Dex administration in our integrated MMTV-LTR constructs. Furthermore, the decline of MMTV-driven transcription upon continued hormone administration temporally coincided with a proportional loss of this hormone-induced DNase I-hypersensitive site. Our results showed a temporal relationship between the intensity of the DNase I-hypersensitive site and the transcriptional ac-4 S. Boronat, H. Richard-Foy, and B. Piñ a, manuscript in preparation. tivity of the reporter gene in both XC-14 and H12 cells. We observed a 4 -8-h lag period between the formation of the hypersensitive site and the peak of luciferase activity in H12 cells. We interpret this delay as the time required for the synthesis of luciferase. It is generally accepted that nucleosome B prevents key transcription factors from binding to their cognate sequences onto the MMTV-LTR (5)(6)(7). We propose that the observed fading of the DNase I-hypersensitive site corresponded to the refolding of nucleosome B. Under this point of view, the decay on MMTV transcription would be explained by the inability of the GR to maintain chromatin in an open state during long periods of exposure to the hormone.
Our results showed that transcription driven by integrated chromatin-organized MMTV-LTR displayed an exquisite dependence on the actual levels of hormone-loaded receptor. A loss of response to hormones upon continuous administration has been associated with a down-regulation of the receptor in different cell systems (32)(33)(34)(35). This down-regulation has been related both to a transcriptional repression of the GR gene and to a destabilization of GR mRNA by the hormone-loaded GR (32,33). Our data showed that GR was mainly cytoplasmic in untreated XC cells, entering the nucleus shortly after hormone administration. After 4 days of continuous hormone administration, GR levels decreased significantly to 20 -30% of the initial levels. This decrease correlated with the decay in MMTV-driven transcription but opened the question of how other glucocorticoid-responsive genes maintained their transcriptional levels high, despite the decrease in GR levels.
Several models can account for the observed specificity of MMTV-LTR deactivation. One possibility is the presence of a specific repressor affecting MMTV-LTR but not MT-I or ␣Bcrystallin promoters. We do not favor this model, in part because it does not explain the temporal correlation between the decay in GR concentration, the closing of the MMTV-LTR chromatin, and the decrease in MMTV-LTR-driven transcription. In addition, this model is difficult to reconcile with the maintenance of transcription of transiently transfected MMTV-LTR plasmids. We rather propose that the minimal amount of hormone-loaded GR required for avoiding a shut down of transcription varies from one promoter to another, depending on the chromatin structure over the GREs. In integrated MMTV-LTR constructs, chromatin appears to refold rapidly when the levels of occupied GR decrease, restraining the binding of factors required for sustaining transcription. Therefore, these promoters are especially sensitive to any decrease in hormoneloaded GR levels. In contrast, transiently transfected plasmids seem to accommodate some kind of loose chromatin package, which is very accessible to enzymes or transcription factors (36). This probably makes transiently transfected MMTV-LTR far less sensitive to the decline of GR than their integrated counterparts. This may well also be the case for the transfected HRE-thymidine kinase promoter on B11 cells, in principle devoid of nucleosome-positioning sequences. On the other hand, both the rat MT-I (28 -30) and the ␣B-crystallin (25, 37) promoters contain several different enhancers that respond to different effectors. This probably results in a constitutive opening of the chromatin structure. This is specially true for the rat MT-I gene, which showed a high basal transcription in our culture conditions and displayed a constitutive DNase I-hypersensitive site over its promoter (30).
It is likely that this type of chromatin-based regulatory mechanism applies to many hormone-responsive genes. Specific nucleosome positioning has been shown for different hormoneregulated genes (38 -40). It is conceivable that chromatin organization, among other factors, may determine whether a given gene responds to a given amount of hormone in a given cell type.
It may also determine what concentrations of hormone-loaded receptor are required for maintaining or for shutting down transcription. As chromatin structure changes in many different ways depending on the cell differentiation stage and cell metabolism, this may represent a finely tuned mechanism of modulating hormone response of many different genes.