Steroid-selective Initiation of Chromatin Remodeling and Transcriptional Activation of the Mouse Mammary Tumor Virus Promoter Is Controlled by the Site of Promoter Integration*

The mouse mammary tumor virus (MMTV) promoter has target sequences recognized by several steroid receptors. We present evidence for a novel mechanism that confers hormone specificity to this promoter. We show that remodeling of MMTV chromatin and the concomitant activation of the MMTV promoter are induced equally by glucocorticoids and progestins in one chromosomal context but are selective for glucocorticoids in another. Furthermore, increased histone acetylation modulates MMTV promoter regulation disparately at the two chromosomal locations. Together, these data indicate that chromosomal architecture commands a crucial role in gene regulation, imposing locus-specific selectivity between regulators with similar sequence recognition.

The mouse mammary tumor virus (MMTV) promoter has target sequences recognized by several steroid receptors. We present evidence for a novel mechanism that confers hormone specificity to this promoter. We show that remodeling of MMTV chromatin and the concomitant activation of the MMTV promoter are induced equally by glucocorticoids and progestins in one chromosomal context but are selective for glucocorticoids in another. Furthermore, increased histone acetylation modulates MMTV promoter regulation disparately at the two chromosomal locations. Together, these data indicate that chromosomal architecture commands a crucial role in gene regulation, imposing locus-specific selectivity between regulators with similar sequence recognition.
Many of the concepts that underlie the current understanding of both steroid hormone action and the role of chromatin in the regulation of gene expression have sprung from studies employing the mouse mammary tumor virus (MMTV) 1 long terminal repeat (LTR) as a model promoter. Target sequences related to a TGTTCT motif direct hormone-induced transcription by glucocorticoid receptors (GR), progesterone receptors (PR), and mineralocorticoid and androgen receptors (1)(2)(3)(4)(5)(6)(7)(8). In stable chromatin, the MMTV LTR also directs the assembly of an ordered nucleosome array that represses loading of basal transcription factors on the promoter. Activation of GR or PR by hormones promotes the remodeling of the MMTV chromatin and the loading of basal transcription factors. In contrast, nucleosomes are randomly arrayed on transiently transfected MMTV templates. Such templates exhibit elevated constitutive levels of transcription factor loading and promoter activity and, consequently, a reduced induction by hormones (9 -11). Thus, the organization of chromatin and its remodeling by steroid hormones are critical to appropriate steroid-mediated regulation.
Although GR and PR recognize the same target elements and each induces certain genes regulated by the other receptor, the two receptors mediate very different actions in tissues such as the mammary epithelium that can express both receptors. GR and PR are expressed at comparable levels in the T47D(A1-2) mammary carcinoma cell line. This cell line also contains 10 copies of a stably transfected MMTV-luciferase (LUC) reporter gene. Surprisingly, luciferase is induced by glucocorticoids but is almost entirely refractory to progestins, whereas a transiently transfected template, MMTV-chloramphenicol acetyltransferase (CAT), is responsive to both hormones. On the MMTV-LUC template, glucocorticoids induce remodeling of chromatin, whereas progestins fail to do so (12).
Selectivity between two regulators that recognize the same target element could be a property inherent to a promoter sequence, a property that cannot be reconstituted on transient chromatin. Alternatively, the chromatin at the site of integration of a transgene may impose a selective mechanism on the integrated promoter. Examples of integration site determining whether a gene is expressed or silenced range from mating type choice in yeast and position effect variegation in Drosophila to recombinant transgenes in mice. However, there is little precedent for more subtle controls such as the imposition of selectivity between two related transcription factors that employ the same target elements. In this report, we address the mechanism that enforces a selective response of MMTV-LUC to glucocorticoids in T47D(A1-2) cells. We show that after stable integration, a second MMTV-driven reporter gene, MMTV-CAT, is responsive to both glucocorticoids and progestins in T47D(A1-2) cells. The responsiveness of MMTV to hormone treatment is reflected in the ability of the GR and PR to remodel MMTV chromatin in an integration site-specific manner. Progestin treatment leads to remodeling of the MMTV-CAT promoter but not MMTV-LUC, whereas glucocorticoids promote chromatin remodeling of both. We also investigated the effect of increased histone acetylation on chromatin remodeling and gene activation of MMTV-LUC and MMTV-CAT in this cell line. Surprisingly, the steroid-dependent induction of the MMTV promoter is potentiated at one chromosomal location but inhibited at another. These studies provide evidence for the imposition of locus-specific constraints in steroid hormone regulation of gene expression.

MATERIALS AND METHODS
Plasmids and Stable Transfection-The plasmids pHHLUC and pHHCAT contain MMTV sequences from HaeIII(Ϫ224) to HpaII(ϩ100) upstream of the coding sequences for the firefly luciferase gene and the bacterial chloramphenicol acetyl transferase gene, respectively.
For stable transfection, 1 ϫ 10 6 T47D(A1-2) cells/10-cm dish were plated and allowed to grow for 24 h before transfection in minimal essential medium plus 200 g/ml G418. One hour before transfection, the culture medium was changed to allow proper pH equilibrium. A calcium phosphate/DNA precipitate was prepared essentially according to Wigler et al. (13). One milliliter of this mixture containing a total of 21 g/ml DNA (20 g/ml pHHCAT plus 1 g/ml pSV2hygro) was added to the culture dishes containing 10 ml of growth medium. After 4 h, the media were removed, and the cells were washed briefly in serum-free minimal essential medium and subjected to a glycerol shock for 3 min. Shock medium is 10% glycerol in HTB buffer (137 mM NaCl, 5 mM Na 2 HPO 4 , 6 mM dextrose, 21 mM HEPES, pH 7.1). Cells were then washed once with serum-free minimal essential medium and returned to growth medium for 24 h. At that time, medium was replaced with a 50:50 mixture of fresh growth medium and conditioned medium containing 400 g/ml hygromycin B. Conditioned growth medium was collected as spent medium from confluent cultures of T47D cells. The selection medium was changed every 3 days. Individual colonies were harvested with cloning rings. Two independent colonies were successfully expanded. These clones contained both pHHCAT and pHHLUC and are referred to as T47D(C&L) cells. Both clones were maintained in growth medium containing 200 g/ml G418 and hygromycin B.
Reporter Gene Assays-For reporter gene assays, cell monolayers were rinsed twice with wash buffer (40 mM Tris-Cl, pH 7.4, 150 mM NaCl, 1 mM EDTA). Cells were lysed by the addition of 1 ml of lysis buffer (20 mM K 2 PO 4 , pH 7.8, 5 mM MgCl 2 , 0.5% Triton X-100). The lysate was transferred to a microfuge tube and centrifuged for 2 min to pellet cell debris. Luciferase and CAT assays were performed exactly as described previously (12). Alkaline phosphatase assays were performed using a Phospha-Light kit (Tropix) according to the manufacturer's instructions, except the 2ϫ buffer contained 0.2 M diethanolamine (pH 9.5), 2 mM MgCl 2 . The protein concentration of each extract was determined by dye binding using a commercial kit (Bio-Rad).
Nuclease Accessibility Assays-Nuclease accessibility assays were performed essentially as described previously (12). Intact nuclei isolated from hormone-treated T47D(C&L) cells were treated with SacI followed by linear amplification of gene-specific primers with Taq polymerase. SacI cleaves a site found within nucleosome B. Nucleosome B spans MMTV DNA from about nucleotide Ϫ221 to Ϫ75 including the steroid receptor binding sites ( Fig. 2A). In untreated cells, nucleosome B prevents transcription factor access to promoter elements required for basal expression from the MMTV promoter (9) and, in isolated nuclei, prevents SacI cleavage. The primers used in the linear polymerase chain reaction amplifications were 5Ј-CCATTTTACCAACAG-TACCG-3Ј for luciferase and 5Ј-CAACGGTGGTATATCCAGTG-3Ј for CAT.
Fluorescence in Situ Hybridization (FISH) Analysis-For FISH analysis, DNA probes were labeled with biotin by nick translation using a BioNick labeling kit (Life Technologies, Inc.) or with digoxigenin using a mixture of dNTPs (including digoxigenin-dUTP) from Boehringer Mannheim. Chromosome preparations were submitted to the following procedure before hybridization: treatment with 70% glacial acetic acid for 1 min, digestion with RNase A (100 g/ml) for 1 h at 37°C, dehydration in an ethanol series, denaturation in 70% formamide/2ϫ SSC (pH 7.0) at 70°C for 2 min followed by dehydration in an ethanol series. The digoxigenin-labeled luciferase probe was denatured at 70°C for 10 min and then incubated at 37°C for 45 min. The biotin-labeled Cot-1 probe was denatured for 10 min at 70°C and placed on ice. Selected sections of the preparation were exposed to 10 l of the hybridization mixture (100 ng of luciferase digoxigenin, 50 ng of Cot-1 biotin in 50% formamide, 10% dextran sulfate, 2ϫ SSC, 1% Tween 20), covered with coverslips, and sealed with rubber cement. Hybridization was allowed to occur in a humidified chamber at 37°C for 48 h. Three posthybrid-ization washes were performed at 45°C for 15 min each with 50% formamide and 2ϫ SSC (pH 7.0), 0.3ϫ SSC (pH 7.0), and then 2ϫ SSC (pH 7.0). After blocking with 5% nonfat milk in TN buffer (0.1 M Tris-HCl, 0.15 M NaCl, 0.05% Tween 20), biotinylated probes were detected by fluorescein isothiocyanate (green), and digoxigenin-labeled probes were detected by rhodamine (red) using an Oncor double detection kit. The selected areas of the slides were then counterstained with DAPI/antifade solution and covered with glass coverslips. Analyses were performed on metaphase spreads of the specimen/probe combination under an epifluorescence microscope (Zeiss AXIOSKOP) using interference filter sets for DAPI, fluorescein isothiocyanate and Texas red, and a triple band pass filter (DAPI/fluorescein isothiocyanate/ Texas red). For documentation, images were captured using an Olympus BX50 system equipped with a similar set of interference filters and a CCD camera (Photometrics). Images were processed in a digital image analysis system using BDS-image software (Oncor). GTG banding analysis was performed using Applied Imaging CytoVision software.
Core Histone Acetylation Analysis-n-Butyric acid (free acid) was obtained from Sigma. Trichostatin A was obtained from Wako (Richmond). For histone analysis, aliquots of nuclei prepared for in vivo chromatin analysis were resuspended in 200 l of histone isolation buffer A (120 mM NaCl, 25 mM EDTA, 10 mM ␤-mercaptoethanol) and placed on ice for 30 min with mixing every 10 min. The samples were centrifuged at maximum speed for 5 min, and supernatants were transferred to a fresh tube. This procedure was repeated using histone isolation buffer B (250 mM NaCl, 10 mM EDTA, 10 mM ␤-mercaptoethanol) followed by histone isolation buffer C (2 M NaCl, 10 mM EDTA, 10 mM ␤-mercaptoethanol). The combined supernatants were brought to 0.28 M HCl and placed on ice for 1 h. Samples were centrifuged at maximum speed for 5 min at 4°C. Supernatants were transferred to a fresh tube and made 20% in trichloroacetic acid and placed on ice for 30 min. Samples were centrifuged at maximum speed for 10 min at 4°C. The pellets were washed once with acetone containing 0.2% HCl followed by two washes with acetone alone. The pellets were dried and resuspended in acid/urea loading buffer (6 M urea, 5% ␤-mercaptoethanol, 5% glacial acetic acid, 0.2% methyl green). Histones were analyzed on acid/urea/Triton gels as described (14).

Steroid-selective Induction of the MMTV Promoter Is Determined by the Locus of Integration of MMTV-To distinguish
whether selective induction of the MMTV-LUC genes in T47D(A1-2) cells is an inherent property of stable versus transient chromatin or is imposed on MMTV by the properties of integration site(s), we stably introduced another MMTV-driven transcription unit (MMTV-CAT) into T47D(A1-2) cells and examined its regulation in response to glucocorticoids or progestins. A clone derived from this transfection, T47D(C&L), expressed comparable levels of GR and PR like the parent T47D(A1-2) line and contained one to two copies of MMTV-CAT (data not shown). In contrast to the selective glucocorticoid induction of LUC (Fig. 1A), both progestins and glucocor- ticoids induced CAT activity to a similar level in T47D(C&L) cells (Fig. 1B). We also examined expression of an endogenous hormone-regulated gene, the liver/bone/kidney form of alkaline phosphatase (ALP). R5020 induces ALP to levels as high or higher than seen with dexamethasone (Fig. 1C). Virtually identical LUC, CAT, and ALP induction results were observed in a second, independent MMTV-CAT-transfected clone (data not shown). It is unlikely that the difference in inducibility of MMTV-CAT and MMTV-LUC is attributable to the reporter gene itself. Stably integrated MMTV-LUC is inducible by both progestins and glucocorticoids in another cell line we have constructed. Also, a recent report describes a cell line with integrated MMTV-CAT genes that are differentially induced by glucocorticoids and mineralocorticoids (15). Together, these data indicate that differential induction is not determined by the structural gene or simply the packaging into stable chromatin but, as we suggest, by the chromosomal context of the transcription unit. Thus, the chromatin environment at the site of MMTV integration can have a profound influence on the nature of its regulation by steroid hormones.
Steroid-selective Chromatin Remodeling of the MMTV Promoter Corresponds with MMTV Promoter Activity-In stable chromatin, an ordered array of nucleosomes is assembled on the MMTV LTR (9). The steroid receptor target elements are present in DNA assembled into nucleosome B (10). These sequences are relatively inaccessible until hormone stimulation induces a remodeling event at nucleosome B. Previous studies using an in vivo nuclease accessibility assay showed that the differential inducibility of the MMTV promoter in T47D(A1-2) cells was associated with the failure of progestins to promote the remodeling of MMTV chromatin (12). We used a similar assay to assess the action of GR and PR at the MMTV promoter in T47D(C&L) cells ( Fig. 2A). Glucocorticoids induced chromatin remodeling (increased SacI sensitivity) of nucleosome B of MMTV from both LUC and CAT loci. Progestins likewise induced chromatin remodeling of MMTV-CAT yet at the same time failed to induce remodeling of the MMTV-LUC transgenes in the same cells (Fig. 2, B and C).
Further evidence that differential hormone induction of MMTV-LUC is functionally linked to rearrangement of nucleosome B in the MMTV-LUC promoter was provided when we used the histone deacetylase inhibitor butyric acid. Hormoneinduced chromatin remodeling of the MMTV-LUC promoter was unchanged or enhanced following butyrate (Fig. 2B) at the same time butyrate consistently suppressed remodeling at the MMTV-CAT promoter (Fig. 2C). These results correlate with the enhancement of GR-mediated transcription of MMTV-LUC and the suppression of GR-and PR-mediated transcription of MMTV-CAT (see below).
The inability of PR to remodel MMTV chromatin and induce LUC expression is unlikely to be due to a failure to bind to the PR target elements. We infer that receptor target sites in MMTV-LUC are available to PR because coadministration of R5020 with dexamethasone inhibits the glucocorticoid induction of MMTV-LUC (12). These data suggest that ligand-bound PR competes for binding to target sites on the MMTV promoter with hormone-activated GR. It is unlikely that this inhibition is attributable to competition for coactivators rather than competition for binding target elements, because transcriptionally inactive PR bound to a ligand antagonist also suppresses the glucocorticoid induction of luciferase (16).
The mechanistic implication of the data is that MMTV chromatin at the LUC loci is structured in a novel fashion under the influence of the surrounding chromatin so as to abrogate the remodeling and concomitant gene induction that would normally occur following PR binding to its target elements. This abrogation may result from the inability of PR to interact with chromatin remodeling proteins that are necessary for steroid receptor-mediated gene induction (17)(18)(19)(20)(21). In contrast, the ability of GR to initiate remodeling of MMTV-LUC chromatin and transcriptional induction is not impaired.
Chromosomal Architecture Surrounding MMTV Luciferase Integration Sites-To explore the nature of the integration site(s) of MMTV-LUC transgenes in T47D(A1-2) cells, FISH analysis was performed with a probe specific for luciferase. Approximately 50 metaphases were analyzed per hybridization site. Four fluorescent bands were detected in the long arm of a   FIG. 2. The ability of steroid receptors to remodel MMTV chromatin presages differential responsiveness to hormonal induction. A, schematic representation of the proximal portion of the MMTV LTR indicating its nucleosomal structure, sites of SacI cleavage, and oligonucleotide primer used for the polymerase chain reaction (PCR) assay. Cylinders labeled A and B indicate nucleosomes A and B, respectively. B, T47D(C&L) cells were treated for 24 h with vehicle, 100 nM dexamethasone (Dex), or 10 nM R5020. For butyrate treatment, cells were incubated with 10 mM butyrate for 24 h followed by an additional 24-h incubation with the indicated hormones. Nuclei were isolated and digested with SacI, which cuts both within nucleosome B and outside nucleosome B. After purification of genomic DNA, 10 g of each sample were analyzed using linear Taq polymerase amplification with a 32 Plabeled single-stranded primer specific for the luciferase gene. Purified extension products were analyzed on 6% polyacrylamide denaturing gels and subjected to autoradiography. The lanes marked control represent extensions out to the nonnucleosomal SacI site and serve as a loading control. Lane 1 (pHHLUC) represents a control extension reaction performed on SacI-digested pHHLUC and serves as a size standard for digestion at the proximal SacI site in nucleosome B. C, an aliquot (20 g) of the same DNA from SacI-digested nuclei used in the analysis of pHHLUC was subjected to linear Taq amplification using an oligonucleotide specific for the CAT gene. Due to the absence of the SacI site outside nucleosome B in pHHCAT, the DNA was digested with BamHI to completion. The lanes marked control represent extension out to a BamHI site in the vector and serve as a loading control. Lane 1 (pHH-CAT) represents a control extension reaction performed on SacI-digested pHHCAT and serves as a size standard for digestion at the SacI site in nucleosome B. single, C size, submetacentric, marker chromosome in all cells analyzed (Fig. 3A). Cot-1 repetitive DNA was used to paint metaphase chromosomes. Cot-1 DNA sequences are extremely repetitive and highly represented in constitutive heterochromatin. The LUC loci were relatively devoid of Cot-1 signal, whereas the flanking regions were intensely highlighted (Fig.  3B). Similar evidence that the MMTV-LUC loci differ distinctly from the flanking chromosomal domains is revealed by GTG banding and DAPI staining. The MMTV-LUC loci corresponded to light G bands flanked by distinctive, dark G bands (Fig. 3E). DAPI staining exhibited a similar pattern whereby the LUC loci were flanked by bright DAPI bands (Fig. 3C). The exclusion of luciferase from the Cot-1 and DAPI staining chromosomal domains is evidenced in Fig. 3D, in which the luciferase, Cot-1, and DAPI signals are overlaid. DAPI has affinity for AT-rich regions of DNA, and dark G-bands are also rich in AT sequences, poor in coding sequences, and rich in L1 type interspersed repeats (23). Summarized schematically in Fig.  3F, the presence of distinctive, defined chromosomal domains flanking each of the MMTV-LUC loci suggest that chromatin structure contributes to the novel, locus-specific, selective reg-ulation of the MMTV promoter.
Differential Effects of Histone Acetylation at the MMTV Promoter Determined by MMTV Integration Sites-Additional evidence that the integration sites of MMTV-LUC and MMTV-CAT endow the promoter with functionally distinct properties comes from experiments in which we used inhibitors of histone deacetylation. Histone acetylation modifies chromatin architecture, and histone acetyltransferases have been demonstrated to interact with nuclear receptors and to serve as transcriptional coactivators (24 -27). Treatment of T47D(C&L) cells with the deacetylase inhibitor butyrate led to a dose-dependent increase in histone acetylation (Fig. 4A). At all doses, butyrate dramatically potentiated the glucocorticoid induction of luciferase (Fig. 4B). While butyrate did not lead to restoration of progesterone induction of luciferase to levels comparable with that seen with glucocorticoid activation, the small induction seen with progestins was also potentiated by butyrate (note scale difference, Fig. 4B). In contrast, the hormonal induction of MMTV-CAT in the same cells was inhibited at all doses of butyrate (Fig. 4C). Hormone-induced chromatin remodeling of the MMTV-LUC promoter was unchanged or enhanced follow- ing butyrate (Fig. 2B) at the same time butyrate consistently suppressed remodeling at the MMTV-CAT promoter (Fig. 2C). A differential action upon the hormone induction of the MMTV-LUC and MMTV-CAT genes was also seen with the more specific deacetylase inhibitor trichostatin A. Trichostatin A potentiated hormone induction of MMTV-LUC activity much like butyrate but had little effect on induction of MMTV-CAT (data not shown). Thus, while a differential effect is still observed with trichostatin A, the data suggest that the butyrate inhibition of CAT induction may be through mechanisms other than deacetylase inhibition or involve deacetylases not inhibited by trichostatin A. These data confirm that the activity of the MMTV promoter can be regulated in a very distinct manner at different chromosomal locations. Moreover, these data provide a likely basis to account for the conflicting findings regard-ing the effect of butyrate on hormone responsiveness of MMTV (28, 29).

DISCUSSION
Specificity is a central question in the regulation of gene expression. Within a family, transcription factors expressed in the same cell can have identical DNA sequence recognition properties yet distinct activities. For example, GR and PR are both expressed in mammary epithelial cells, but the two hormones have very different effects. The mechanisms by which such specificity is generated are not well understood. In this report, we have presented evidence for a novel mechanism that confers specificity so that a promiscuously hormone responsive promoter is enabled to respond selectively to one of the two receptors. The MMTV promoter has been widely used as a model system for studying the role of chromatin structure in the regulation of gene expression by steroid hormone receptors. In stable chromatin, the MMTV LTR directs the assembly of an ordered nucleosome array (9,10). The target sites for steroid receptor binding lie within the region occupied by nucleosome B, and this nucleosome is remodeled upon activation of the MMTV promoter by hormones. This is evidenced by an increased accessibility of the chromatinized DNA to cleavage by restriction enzyme digestion and increased loading of basal factors (9 -11). Beginning with the demonstration that the activity of GR expressed in yeast requires the SWI-SNF chromatin remodeling complex (17), subsequent work has supported a role for chromatin remodeling in steroid hormone action (18 -21). Recently evidence has been presented for a direct interaction of GR with the human SWI-SNF homolog, BRG1 (30). Activation of the MMTV promoter is also accompanied by a partial depletion of histone H1 from the template in vivo (31).
Glucocorticoids and progestins can induce many of the same promoters including MMTV. This is not surprising, since the DNA binding domains of GR and PR have very high sequence identity (32) and target sequence utilization (33,34). Thus, the question of how these two transcriptional activators can generate specific responses in tissues where both may be expressed is an important issue. We previously made the surprising observation that in T47D(A1-2) cells engineered to express comparable levels of GR and PR, a stably integrated MMTV-luciferase template can be induced by glucocorticoids but is almost completely refractory to progestins. Unlike glucocorticoids, progestins fail to initiate remodeling of the MMTV promoter (12). Here we explored the unique possibility that the MMTV integration site determines the ability of the PR to remodel MMTV chromatin and activate transcription. Our approach involved the stable introduction of an additional MMTV-driven reporter gene (MMTV-CAT) into T47D(A1-2) cells. We then examined the induction of this reporter in response to treatment with glucocorticoids and progestins. In contrast to the differential induction of MMTV-LUC, a robust induction of MMTV-CAT was observed after treatment with either the synthetic glucocorticoid dexamethasone or the synthetic progestin R5020. Because the MMTV LTR sequences used in the construction of MMTV-LUC and MMTV-CAT are the same, the observed differential capacity of PR to induce transcription from MMTV-LUC and MMTV-CAT cannot be attributed to differences in MMTV promoter sequences. Additionally, the full complement of factors required for PR activation appears to be present in T47D(C&L) cells as evidenced by PR-mediated induction of MMTV-CAT or the endogenous ALP gene. The specific characteristics of these chromosomal domains, which either allow or prevent PR from activating MMTV, remains unknown. Efforts are currently under way to explore the nature of these MMTV integration sites and their role in the differential activation of MMTV by steroid hormone receptors.
Previously, it has been reported that histone acetylation can have either inhibitory or stimulatory effects on MMTV promoter activity, depending on the experimental system. Hager and colleagues (29) reported that the histone deacetylase inhibitor butyrate inhibits GR activation of MMTV by blocking remodeling of nucleosome B. Beato and colleagues (28) have shown that the response of the MMTV promoter can be modulated by the degree of histone acetylation; moderate histone acetylation led to activation of MMTV, while high levels of histone acetylation led to a reduction of MMTV activity. In this paper, we show that increased histone acetylation is interpreted very differently at different chromosomal sites. At one, the hormonal induction of MMTV promoter activity is potenti-ated, whereas at another it is inhibited or unaffected. These results suggest a mechanism, perhaps a specific chromatin architecture at the integration site, that can impose selectivity on a promiscuously responsive promoter. The ramifications of this suggestion are of enormous importance, implying that such chromosomal regulatory domains are a critical determinant of tissue and hormone specificity of gene responsiveness. Such mechanisms could also play important roles in development and homeostasis by enforcing specificity of action among members of transcription factor families that have similar DNA recognition properties.
There are many examples where chromatin architecture has been implicated in determining whether a gene is expressed. Notable examples under active investigation include position effect variegation in Drosophila and gene silencing near telomeres in yeast. Although mammalian examples are less tractable experimentally, certainly X chromosome inactivation and parental imprinting have received considerable attention. The results of the present work imply that controls more subtle than a simple on or off decision are at work, identifying a novel mechanism for differential regulation of gene expression. It will be important to identify additional examples. One possibility is the recent report of a stably transfected MMTV-CAT gene inducible by glucocorticoids but refractory to mineralocorticoids (15).
Chromatin architecture may also influence the modulation of steroid-mediated induction by cellular signal transduction pathways. The hormonal activation of the MMTV promoter has been reported to be potentiated by cAMP in transient chromatin but inhibited in stable chromatin (35). However, inhibition is not a universal finding. Both GR and PR action on stably integrated MMTV promoters can be potentiated in fibroblasts and mammary carcinoma cell lines, including T47D(A1-2) (36 -39). 2 Although it is difficult to compare results between different cell lines, these data may suggest that the response of the steroid induction to cAMP elevation may differ at different chromosomal locations. Together with the present data, these findings provide support for the concept of chromatin architecture imposing constraints that can confer differential transcriptional responses.