(Received for publication, January 23, 1997, and in revised form, March 25, 1997)
From the Laboratory of Receptor Biology and Gene Expression, National Cancer Institute, National Institutes of Health, Bethesda, Maryland 20892-5055
In vivo, transcription factors interact with promoters having complex nucleoprotein structures. The transiently expressed progesterone receptor (PR) efficiently activates a transfected mouse mammary tumor virus (MMTV) promoter but is a poor activator of the MMTV promoter when it acquires an ordered chromatin structure as an endogenous, replicating gene. We show that the deficiency in PR activity is not due to insufficient expression of either B or A isoforms or competition between the two types of MMTV templates. Rather, this deficiency reflects an inability to induce the chromatin remodeling event that is required for activation of the replicated MMTV template. To determine whether this characteristic is common to transiently expressed steroid receptors or specific to the PR, we examined the activity of transiently expressed glucocorticoid (GR) receptor. Unlike the PR, the transiently expressed GR is an effective activator of both MMTV templates and efficiently induces the necessary chromatin remodeling event at the replicated template. These results indicate that the GR and PR have unique requirements for activation of promoters with ordered chromatin structure. These differences may provide a mechanism for establishing target gene specificity in vivo for steroid receptors that recognize and bind to identical DNA sequences.
In the living cell, transcription and replication take place in the context of chromatin. Factors must interact with promoters and other important DNA regions that have complex nucleoprotein architecture. Inactive genes, particularly those that are tissue-specific, are characterized by chromatin structure which is inaccessible to factors and nucleases (reviewed in Ref. 1). During development, newly expressed transcription factors must interact with these repressive structures to activate their target promoters.
The mouse mammary tumor virus (MMTV)1 promoter is an in vivo model for the role of chromatin structure in transcription. In its stably replicating form, the MMTV LTR exists as an ordered array of nucleosomes (2) that occur along the DNA in a frequency-biased distribution of translational frames (3). All of the glucocorticoid response elements (GREs) in the promoter are associated with the B family of nucleosome frames. Upon activation of the glucocorticoid receptor (GR) this nucleosome region undergoes a transition in structure, which allows the previously excluded factors NF1 and OTF1 to bind to their sites on the promoter (4, 5) and may involve the loss of histone H1 (6). The GR also either recruits and/or stabilizes the interaction of the TFIID complex with the template (4, 7, 8). In contrast, analysis of transiently transfected MMTV-reporter constructs has revealed that these templates do not have an ordered nucleosomal repeat, that they are constitutively accessible to NF1, OTF1, and various nucleases, and that GR does not induce any transition in nucleoprotein structure (7).
These results indicate that the GR acts bimodally on the stably replicating template as follows: first, to derepress it through the structural transition and second, to activate it by participating in the formation of an active transcription initiation complex (7). Since the transient template does not undergo the derepression step, its activity is largely a measure of interactions between soluble factors, and it would not be an adequate model for the interactions of transcription factors with components of ordered chromatin in vivo. Because the two MMTV templates have the same LTR sequences but differ in their nucleoprotein structure, functional differences may represent mechanisms by which ordered chromatin structure participates in regulation of the MMTV promoter. Differences in the behavior of the two templates have been observed in response to butyrate treatment (9), activation of cAMP signaling (8), and in the kinetics of GR-induced activation (5). Similar functional differences have been observed in other systems (10, 11).
The MMTV promoter can also be activated by progesterone, mineralocorticoid, and androgen receptors, all of which bind to the same DNA sequences (12-18). Previously we reported that transiently expressed progesterone receptor (PR) could not efficiently activate the stably replicating template even though it significantly induced transcription from the transient MMTV template (19). In the current study we further characterized the activity of transiently expressed PR using a highly efficient cell sorting method involving antibody-coupled magnetic beads (20) to isolate cells that have taken up exogenous DNA. We find that the deficiency in PR activation of the stably replicating template represents an inability to induce the necessary chromatin remodeling event. We have also asked whether the deficiency in activation of the stably replicating template is specific to the PR or a common feature of transiently expressed receptors by examining the activity of a transiently expressed GR (21). In contrast to the PR, transiently expressed GR can activate both MMTV templates efficiently and induce the chromatin remodeling event. These results indicate that there are intrinsic differences in the way the two receptors interact with the stably replicating MMTV template which are not manifested on the transiently transfected template. These differences will provide insight into how this family of receptors, which recognize the same DNA sequence, are able to specifically regulate different sets of genes in vivo.
Cell line 1471.1 has been described previously (22). It contains multiple copies of stably replicating MMTV-CAT transcription unit in the context of bovine papilloma virus sequences and expresses GR but not PR. Cell line 904.13 contains 200 copies of stably replicating MMTV-ras transcription unit in the context of bovine papilloma virus sequences and also expresses only GR (3, 7). Both cell lines were maintained in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) plus 10% charcoal-stripped serum (Hyclone). Magnetic beads coated with goat anti-mouse IgG were purchased from Dynal and the magnetic plates from BioMag. Various monoclonal antibodies directed against the interleukin 2 receptor (IL2R) were obtained from Amersham Corp., Boehringer Mannheim, and Upstate Biotechnologies, Inc. Previously described plasmids used in this study are as follows: pcPRO (chicken PR expression vector) (23), pLTRluc (full-length MMTV LTR driving luciferase) (24), and pCMVIL2R (IL2R expression vector) (25).
Cloning of Reporters and Receptor Expression VectorsMMTV-reporter construct pMTVbgln consists of the
full-length LTR (1187 to +103 bp) driving the expression of the
rabbit
-globin gene. It was made by inserting a
DraI/BamHI fragment from pM18 (2) containing the
LTR region into pMggnOVEC (kindly provided by S. Rusconi) digested with
SmaI and BamHI to remove MMTV LTR sequences from
a different MMTV strain. The resulting plasmid contained the MMTV LTR
fused to
-globin coding sequence from the second exon. To restore
the first exon and intron of the
-globin gene, a polymerase chain
reaction fragment containing these sequences was inserted. The C656G
expression vector, pCI-nH6HA-C656G, was made by insertion of the C656G
cDNA (21) (kindly provided by S. Simons) into pCI-nH6HA. This
plasmid is derived from pCI (Promega) into which the sequence (sense
strand) 5
GCTAGCGAAGGAGATCCGCCATGGCCCACCATCACCACCATCACGGATATCCATACGACGTGCCAGATTACGCTCAGCTGGAATTC 3
was inserted at the NheI and EcoRI sites. This
sequence contains an initial methionine codon, six histidines, and the
hemagglutinin A (HA) epitope. The cDNA was inserted such that
the receptor would be expressed with the histidine tag and the HA
epitope at its amino terminus. Both the CMV and T7 promoters lie
upstream of the cDNA. The high level expression vector for the PR,
pnH6HA-cPR(B), was made by inserting an
EcoRV/KpnI fragment containing the entire coding
sequence of the chicken PR except for the first methionine into
the PvuII and KpnI sites in the polylinker
of pCI-nH6HAfs2, a frameshift variant of pCI-nH6HA. The
-actin plasmid (pT7B-Actn-F) used to make the S1 probe contains a
91-bp polymerase chain reaction fragment from the second exon of the
mouse
-actin gene in pT7Blue (Novagen).
Cell line 1471.1 was transfected by either calcium phosphate coprecipitation using BES-based buffers (26) or electroporation. Briefly, cells were transfected with a combination of plasmids. In every case (unless noted) transfection was carried out with the IL2R expression vector along with an MMTV-reporter construct (either pLTRluc or pMTVbgln) and a receptor expression vector (pcPRO, pCI-nH6HA-C656G, or pnH6HA-cPR(B)). Amounts of each DNA used are indicated in the figure legends. Electroporation was carried out in a Cell-porator (Life Technologies, Inc.) at 250 V, 800 microfarads with aliquots of 2 × 107 cells in 300 µl of Dulbecco's modified Eagle's medium. Cells were plated after electroporation and were treated and harvested the following day. Transfection by calcium phosphate coprecipitation using BES-based buffers was carried out overnight at 37 °C, 2.9% CO2 on 10-cm dishes (7 × 105 cells), each receiving a total of 20 µg of DNA. The following day cells were re-fed and transferred to 37 °C, 5% CO2. Cells were treated and harvested 2 days after transfection. Magnetic affinity sorting was carried out as described with some modification (20). Prior to sorting goat anti-mouse IgG-coated magnetic beads were mixed with IL2R monoclonal antibody at a ratio of 50 mg bead suspension to 50 µg antibody. Beads were diluted in Medium S (4 mM EGTA, 100 µg/ml chondroitin sulfate, 0.1% gelatin, 10 mM Hepes, pH 8.0, 1 mM MgCl2, 1 mM MgSO4, 8 mg/ml non-fat dry milk, and 100 µg/ml bovine serum albumin, all in PBS without Ca2+ or Mg2+) at a ratio of 15 µl bead suspension (50 mg/ml) to 1 ml Medium S. Cells were washed with PBS and incubated with Medium S/bead mixture for 15 min at 37 °C. The cells were washed again with PBS and harvested by brief trypsinization followed by neutralization with trypsin inhibitor. Flasks containing the cells were placed between magnetic plates to separate the beaded cells from the unbeaded cells. The two resulting cell pools were washed several times with PBS, pelleted, and frozen for later analysis.
Analysis of RNA, Luciferase, and CAT ActivityRNA was
isolated from cell pools and subjected to S1 nuclease digestion (8-10
µg of RNA per sample) as described previously (19), except pMTVbgln
digested with SacI and pT7B-Actn-F digested with either
EcoRI or SpeI were used as templates to make
single-stranded probes by multiple rounds of linear primer extension
with Taq polymerase and short oligonucleotides homologous to
-globin or
-actin coding sequences. S1 digestion products were
separated on 8% denaturing gels that were dried and exposed to
PhosphorImaging screens. All quantitation was carried out using
ImageQuant software (Molecular Dynamics).
Cell extracts used in assays for CAT and luciferase activity were made from sorted cell pools by resuspension of cell pellets in 0.25 M Tris, pH 7.5, and three cycles of freezing and thawing. Cellular debris was pelleted, and the supernatant was used in the assays. Protein concentrations were determined by the method of Bradford using protein assay dye (Bio-Rad). Five µg of cellular extract protein from each sample was used for CAT analysis (27). Visualization of the products was carried out using a Molecular Dynamics PhosphorImager and quantitation with ImageQuant software. Luciferase assays were carried out as described previously (24) except that a Microlumat LB 96 P (EGG Berthold) machine was used to measure the luminescence. All values for luciferase activity were normalized to the amount of cellular protein used for each sample.
Preparation of Cytosolic Extracts and ImmunoblottingCytosolic extracts for immunoblotting of steroid
receptors were made as follows. Cells from sorted pools were washed
with PBS, pelleted, and resuspended in HEGDM (10 mM Hepes,
pH 7.4, 1 mM EDTA, 10 mM sodium molybdate, 2 mM dithiothreitol, 10% glycerol) containing 0.1% Nonidet
P-40 and a protease inhibitor mixture (0.1 mM
phenylmethylsulfonyl fluoride, 0.1 mM benzamidine, 1 µg/ml leupeptin, 5 µg/ml aprotinin). Cell membranes were allowed to lyse for several minutes, and nuclei and cellular debris were pelleted
5 min at 12,000 × g. The supernatants were stored at 80 °C. Proteins were separated by SDS-polyacrylamide gel
electrophoresis (3% stacking gel, 8% separating gel) and transferred
to Immobilon or Hybond nitrocellulose (Amersham Corp.) for 3-4 h in
Tris glycine buffer with 20% methanol at 300 mA. Membranes were
blocked for 1 h at room temperature in either Tris-buffered saline
(TBS) (10 mM Tris, pH 7.5, 0.14 M NaCl, 2.7 mM KCl, 0.7 mM CaCl2, 0.5 mM MgCl2) containing 2% nonfat dry milk (for
anti-GR, -PR antibodies) or 2.5% blocking reagent (Boehringer
Mannheim) in 100 mM maleic acid, pH 7.5, 150 mM
NaCl (for anti-HA antibody). Incubation with primary antibodies was
carried out overnight at 4 °C. Primary antibodies include PA512
(Affinity Bioreagents) against the GR, 12CA5 (kindly provided by W. Dixon and J. Campbell) against the HA epitope, and PR22 (kindly
provided by D. Toft) against both isoforms of the chicken PR. Blots
were washed several times with TBS containing either 0.1% Tween 20 (for anti-GR, -PR antibodies) or 0.4% Tween 20 (for anti-HA antibody)
before exposure to the appropriate secondary antibody for 2 h.
Blots were washed as above and exposed to reagents provided in an ECL
or ECF kit (Amersham Corp.). In the case of detection by enhanced
chemiluminescence, the membranes were exposed to XAR-5 film (Kodak) for
1 min or less. For analysis with ECF, membranes were scanned with a
Molecular Dynamics Storm 860 instrument in the blue fluorescence
mode.
Cells
transfected with the IL2R expression vector and either pcPRO,
pCI-nH6HA-C656G, or pnH6HA-cPR(B) were treated with hormones for 1 h in Medium S. Antibody-coated beads were added 15 min before harvest.
Cells were harvested by brief trypsinization and neutralization with
trypsin inhibitor. After a brief sort, nuclei were isolated from sorted
cell populations as described previously (8). SacI digestion
of nuclei was carried out at 30 °C for 15 min in 50 mM
NaCl, 50 mM Tris, pH 8.0, 1 mM
MgCl2, 1 mM -mercaptoethanol, 2.5% glycerol
at a SacI concentration of 10 units per µg of DNA. The
reaction was terminated by the addition of 5 volumes of 10 mM Tris, pH 7.5, 10 mM EDTA, 0.5% SDS, and 100 µg/ml proteinase K. DNA was purified by phenol/chloroform/isoamyl
alcohol extraction and digested to completion with DpnII. An
end-labeled oligonucleotide containing MMTV-transcribed sequences (+1
to +27 bp) was used in multiple rounds of linear amplification with
Taq polymerase to detect digestion products.
We have designed a system in which the
activity of transiently expressed steroid receptors can be monitored on
two types of MMTV templates. The first stably replicates, being either
episomal or integrated, and consists of the full-length LTR driving
expression of the CAT gene (22). Its chromatin structure is
characterized by an ordered array of non-randomly positioned
nucleosomes (2, 3). The access of general transcription factors such as
NF1 and OTF1 to this template is largely inhibited in the absence of
hormone (4, 5) reflecting a structure that is repressive to
transcription. The activated GR binds to its sites that lie within the
B family of nucleosomes and causes a structural transition that results
in a more open and accessible structure (2, 28). The second template is
a transiently transfected reporter construct consisting of the
full-length LTR driving the expression of the luciferase or -globin
genes. Its nucleoprotein structure is characterized by the lack of an
ordered nucleosome repeat and a constitutively open conformation, which
allows access of transcription factors NF1 and OTF1 in the absence of
hormone (7). Activated GR does not induce any apparent transition in
structure but does cause the increased association of the TFIID complex
with the template. Since these two types of MMTV template have the same
LTR sequence, differences in their function are most likely a result of
their differing structures.
To measure the activity of transfected receptors on stably replicating
templates, it was necessary to sort the transfected cells into a
fraction enriched in those cells that express the exogenous DNA,
because the receptors are expressed in only a fraction of the total
cell population. In our previous study (19) this was accomplished by
fluorescence-activated cell sorting, a lengthy procedure that was
limited for the number of conditions examined. In this study we have
employed a highly efficient magnetic affinity-based cell sorting
procedure (20) (Fig. 1). Cells containing stably replicating MMTV templates were transfected with an MMTV-reporter construct, a receptor expression vector, and an expression vector for
the Tac subunit (29) of interleukin 2 receptor (IL2R).
Monoclonal antibodies that recognize the extracellular region of the
IL2R subunit were used to coat magnetic beads with anti-mouse
immunoglobulins attached. These beads bound only to cells that took up
exogenous DNA and expressed the IL2R. After trypsinization, the beaded
cells were separated from the non-transfected cells with the use of magnets. The cell pools were then used as sources of RNA, DNA, or
cytosolic fractions. The IL2R-expressing population of cells is highly
enriched for the transient template and the transiently expressed
receptor.
Activity of Transiently Expressed PR on Stable and Transient MMTV Templates
Previously we determined that, unlike endogenous GR, transiently expressed PR was not able to significantly activate an integrated MMTV template (19). In contrast, a transient MMTV template was efficiently activated by PR. Because of the length of the FACS-based sorting procedure, we could not, in the same experiment, sort dexamethasone (Dex)-treated cells in addition to non-treated and R5020-treated cells. Therefore, we were unable to directly compare the abilities of the endogenous GR and the transiently expressed PR to activate the transient template. Since its nucleoprotein structure is non-ordered and accessible, activation of the transient template is most likely a reflection of interactions between soluble factors and DNA rather than between soluble factors and chromatin components. Therefore, the activity of the various receptors on this template is a measure of their potential as transactivators at the MMTV promoter in a given cell type. Determination of relative transactivation potentials is important for assessing the abilities of various receptors to activate the stably replicating template and evaluating the impact of nucleoprotein architecture on this process.
Using the magnetic affinity cell sorting procedure, we examined the
activity of the transiently expressed PR relative to the endogenous GR.
Cells were transfected with expression vectors for the IL2R and the
chicken PR as well as an MMTV/-globin reporter construct. After
sorting, RNA was isolated and analyzed by S1 nuclease protection assay
with a probe that differentiates between transcripts generated from the
stably replicating and transient MMTV templates. Also included was a
probe to detect
-actin transcripts that served as an internal
standard.
Fig. 2A shows a representative experiment.
Both the endogenous GR and the transiently expressed PR activated the
transient MMTV template to approximately the same extent. However,
whereas the GR induced a large increase in the level of mRNA
generated from the stably replicating template (40-fold in this
experiment), the PR induced only a weak response (4-fold). Fig.
2B shows the average relative activities of the two
receptors from multiple experiments. The transiently expressed PR is
approximately 7 times less active than the endogenous GR on the stable
template but is just as efficient in activating the transient
template.
One possible explanation for the differential in transactivation is that the two template types compete for the PR, the transient template being more successful, perhaps given its quantity or more accessible structure. Therefore, we carried out the same experiment in the absence of the MMTV-reporter construct. Fig. 2C shows that the PR was still a weak activator (5-fold induction in this experiment) of the stable template. Therefore, the function of the PR at the stably replicating template is independent of the relative amounts of templates present and reflects a true deficiency in activity uniquely associated with some property of the stably replicated template.
To determine whether the concentration of PR in the transfected cells
affected its ability to transactivate the two templates, titration of
the transfected PR expression vector was carried out. Increasing
amounts were transfected into 1471.1 cells along with the IL2R
expression vector, and an MMTV-reporter construct, in which the
full-length LTR directs expression of luciferase. Fig.
3, A and B, shows the titration
curves for the stably replicating and transient MMTV templates. In each
case the induction by Dex remained relatively constant over the range
of PR expression vector transfected. For the transient template,
inductions by R5020 increased (reaching the level of Dex inductions),
peaked, and then declined, probably due to squelching. In the case of
the stable template, R5020 inductions increased initially and then
plateaued at a low level, approximately 20% of the Dex induction,
which correlates well with the RNA data shown in Fig. 2B. An
immunoblot of cytosols from transfected cells shows the accumulation of
PR as the amount of expression vector transfected increased (Fig.
3C). The expression level of PR clearly increased even as
the amount of activity on the stably replicating template remained
level.
The PR is expressed as two isoforms, B and A, the A form being a shorter version of the B form, lacking an amino-terminal region. The B form of the PR has been shown to be more active than the A form on the MMTV promoter (30, 31). In addition, the A form can be a repressor of the B form (31, 32). It was therefore possible that the transiently transfected cells were lacking the B form, or expressed a predominance of the A form. Fig. 3C shows that both A and B forms of the receptor were expressed and that the B form was expressed at a higher level than the A form. In fact, the transiently transfected cells have a predominance of the B form when compared with a similar cell line containing stably transfected PR (data not shown) in which the stably replicating MMTV template is activated by R5020 (19). It is therefore unlikely that the PR is deficient in activating the stably replicating template due to a lack of the B isoform.
Activity of Transiently Expressed GR on Stable and Transient MMTV TemplatesTo determine whether the observed deficiency in ability
to activate the stably replicating MMTV template is specific to the PR
or characteristic of transiently expressed receptors in general, we
sought to test the activity of transiently expressed GR. Our cell lines
that contain the stably replicating template also express GR
endogenously, so we used a GR mutant carrying a cysteine to glycine
substitution in the ligand binding domain (amino acid 656, rat GR)
which causes the receptor to have a higher affinity for ligand (21).
This is reflected in the Dex dose-response curve shown in Fig.
4A, where the C656G receptor was able to
fully activate the transient MMTV template at 1 nM Dex, a
concentration which did not result in activation of the endogenous GR.
Using C656G allowed us to measure the activity of a transiently
expressed GR in the same cells used to test the PR.
Cells were transfected with the IL2R expression vector and the
MMTV/-globin reporter construct in the presence and absence of the
expression vector for C656G and treated as shown in Fig. 4B,
a representative RNA analysis of the cell pools. Quantitation of the
results is shown in Fig. 4C. In cells with only endogenous GR, there was no significant induction in mRNA levels from either template at 1 nM Dex (lanes 2, 5, and
8), whereas at 100 nM Dex there was a strong
response from both (lane 6). In cells expressing the C656G
receptor, treatment with 1 nM Dex led to a robust increase in mRNA levels from both templates (lane 11). In the
case of the stably replicating MMTV template, at 1 nM Dex,
RNA levels were induced to approximately the same levels as those
induced by the endogenous GR in the absence of C656G (compare
lanes 6 and 11, Stable). In addition,
in the presence of C656G, treatment with 100 nM Dex led to
greater accumulation of RNA (compare lanes 11 and
12, Stable), which is due to activation of both
the endogenous GR and C656G. In the case of the transient MMTV
template, C656G was able to induce greater levels of RNA at 1 nM Dex than the endogenous GR at 100 nM Dex in
the absence of C656G (compare lanes 6 and 11,
Transient). At 100 nM Dex there was no further
induction of RNA levels from the transient template, unlike the stable
template (compare lanes 11 and 12).
Titration curves for the C656G expression vector are shown in Fig.
5, A and B, as normalized MMTV RNA
levels. (Basal levels of RNA generated from the transient template were
undetectable, and therefore, fold inductions could not be calculated.)
Over a range of C656G levels, the templates behaved similarly in
response to treatment with 1 nM Dex. The amount of RNA
induced from both templates increased at similar rates with peak levels
achieved at the same point. These results stand in marked contrast to
those obtained with the PR (Figs. 3 and 6) where the
level of activation at the stable template did not respond to
increasing expression of the PR. The disparate behavior of the
transiently expressed PR and GR (C656G) at the stably replicating
template provide strong evidence that the two receptors interact with
this template in distinctly different ways.
The RNA levels induced from the two templates by endogenous GR at 100 nM Dex in the absence of C656G are indicated by the dashed lines. Whereas C656G induced similar levels of RNA at its maximum activity on the stable template, it was able to induce greater levels from the transient template. The reasons for this observation may lie in relative receptor expression levels. To obtain full activation of the stable template, the transfected GR must be expressed at levels 3-5-fold higher than the endogenous GR, as shown in Fig. 6A. The extent of activation at the transient template by transfected C656G may be a function of the greater amount expressed relative to the endogenous GR. At the stable template there may be factors required for activation in the presence of ordered chromatin that, due to limited availability, restrict the overall amount of activation achieved. The endogenous GR may have an advantage in activating the stably replicating template. Its stable presence in the cells prior to expression of C656G may give it greater access to limiting factors necessary for transactivation of genomic templates. Thus C656G must be expressed at higher levels to compete with the endogenous GR for these limiting factors.
If there are limiting factors uniquely necessary for activation of the stably replicating template for which transfected receptors must compete, it is possible that the transfected PR requires the same factor(s) necessary for C656G. To address this possibility, we cloned the chicken PR cDNA into the same expression vector used to express C656G so that both receptors would be tagged with the hemagglutinin (HA) epitope and expressed at similar levels in transfected cells. Fig. 6A shows an immunoblot of cytosolic extracts from sorted cells transfected with the expression vectors for either C656G or PR, in which a primary antibody against the HA epitope was used to detect both receptors. At the amounts of expression vector used, the two receptors are expressed at approximately the same levels in the IL2R+ populations. The activity of the PR at these expression levels is shown in Fig. 6, B and C. Even at levels equivalent to those at which C656G fully activates the stably replicating template, the transfected PR is still a poor activator, achieving only 27% of the activation induced by the endogenous GR (Fig. 6C). The amount of PR generated from this expression vector is several times greater than the amounts generated in the experiments shown in Fig. 2 (data not shown). These results indicate that the GR and PR have different requirements for productive interaction with the MMTV promoter in the context of ordered chromatin.
Chromatin Remodeling Activity of Transiently Expressed PR and GRA hallmark of GR-induced activation of the stably replicating
MMTV template is a structural transition in the B nucleosome region of
the promoter, which is measured by increased nuclease cleavage within
this region (2, 28). In particular, treatment of cells with Dex results
in increased cleavage of a SacI site located within the
proximal GREs. In a variety of cell lines tested, this increase in
cleavage is reproducibly in the range of 5 and 15% relative to the
total number of templates.2 This transition
in nucleoprotein structure is necessary for activation of the stably
replicating template because it allows previously excluded
transcription factors access to their binding sites (7). As shown in
Fig. 7, we examined the ability of the transiently expressed PR and C656G to induce the nucleoprotein transition by
measuring SacI access in nuclei of sorted cells. Since
1471.1 cells characteristically show smaller hormone-induced increases in SacI cleavage at the stably replicating template
(5-10%), we also tested 904.13 cells that show larger increases
(10-15%). Unlike the endogenous GR, the transiently expressed PR did
not cause an increase in cleavage by SacI in the presence of
R5020 relative to the basal level of cleavage in nuclei from untreated cells (Fig. 7A, 1471.1 cells and B and
C, 904.13 cells). In contrast, C656G, when activated in the
presence of 1 nM Dex, caused increased cleavage by
SacI similar to that induced by the endogenous GR (IL2R,
100 nM Dex). Under these conditions similar levels of RNA
were induced from the stable template (see Fig. 4B, 1 µg
of pCI-nH6HA-C656G at 1 nM versus 0 µg at 100 nM). In addition, treatment with 100 nM Dex in
the presence of C656G results in higher levels of fractional cleavage
and higher levels of RNA induced (see Fig. 4B). These
results indicate that the deficiency in PR-induced activation of the
stably replicating MMTV lies in the derepression of the template. This
is entirely consistent with the fact that both receptors can activate
the transient template, since the nucleoprotein transition, and hence
derepression, does not occur at the transient template. Transiently
expressed PR is therefore unable to induce the chromatin remodeling
event necessary for increased transcription at the MMTV promoter in
ordered chromatin.
In mammals progestins, androgens, mineralocorticoids, and glucocorticoids elicit profoundly different responses, even though their receptors recognize the same DNA sequence. The mechanism by which this specificity is achieved has been proposed to be due to receptor expression patterns (33), receptor-specific accessory proteins (34), or receptor-specific interaction with other transcription factors (35), among others. Our results provide evidence that receptor-specific interactions with chromatin components are involved in achieving gene specificity in vivo. We have shown that the GR and PR can be present in the same cells but differentially activate a promoter that contains their binding sites and has an ordered, repressed chromatin structure. Our experimental system has exposed previously unidentified differences in the way the GR and the PR interact with the MMTV promoter and has provided the opportunity to define receptor-specific factors that are specifically necessary for activation of genomic templates.
Unlike endogenous GR, the transiently expressed PR is impaired in its ability to activate the MMTV promoter when it has an ordered and repressed chromatin structure. In the same cells, however, it efficiently activates an MMTV promoter which is non-replicating and has a disorganized, accessible nucleoprotein structure. We have now shown that the inefficient activity of the PR on the stably replicating template is not due to competition between the two templates, insufficient expression of the PR, or lack of the B isoform. Rather, this deficiency in activation is a consequence of the particular nucleoprotein structure of the promoter. This study has also provided insight into the mechanism by which this deficiency occurs. In vivo restriction enzyme access experiments show that the PR is unable to induce the chromatin remodeling event at stably replicating MMTV template which is necessary for transcriptional activation.
An obvious question arising from these results is whether the deficiency in activation of the stably replicating template is specific to the PR or a common feature of transiently expressed receptors. It is possible that newly expressed receptors must undergo some kind of intracellular processing to allow them to productively interact with chromosomal templates. Alternatively, the various steroid receptors may require different cofactors to allow them to function in complex chromatin. To test the activity of a transiently expressed GR in cells that express endogenous GR, we used a GR mutant, C656G (21), having a higher affinity for ligand. This receptor was able to activate both the transient and stably replicating templates efficiently when transiently expressed. At peak amounts of activity, C656G induced levels of RNA from the stably replicating template that were equivalent to those induced by the endogenous GR. In addition, C656G was able to induce the characteristic structural transition in the MMTV promoter to the same extent as the endogenous GR. These results stand in stark contrast to those obtained in experiments with the PR. Whereas both templates showed greater levels of activation as C656G expression increased, only the transient template responded to PR in this fashion. As PR levels increased, activation of the stably replicating template increased only slightly and remained level over a very wide range of expression levels (Figs. 3 and 6).
Relative to the activity of the endogenous GR, however, C656G was more efficient in activating the transient template. This is probably due to the fact that C656G was expressed at higher levels than the endogenous GR at the concentrations required for full activation of the stably replicating template. The endogenous GR, given its stable presence in the cell, may have an advantage over C656G which is expressed for only a short amount of time. Factors uniquely necessary for activation of genomic templates may be sequestered by or preferentially available to the endogenous GR, and therefore, C656G must be overexpressed to successfully compete for them. However, although the results strongly suggest the existence of such factors, they are not the sole cause of the deficiency in PR activity because, even when the PR was expressed at levels equivalent to those of C656G, it did not efficiently activate the stably replicating template.
The differential activity of the transiently expressed PR and GR on the stably replicating template provides strong evidence that the two receptors interact in distinct ways with the MMTV template when it adopts a complex nucleoprotein structure. The inability of the PR to induce the chromatin remodeling event may reflect a deficiency at one of two points in the activation process. First, the PR may be unable to bind to its target sites in the context of positioned nucleosomes. In vitro, PR and GR were shown to bind to nucleosomes reconstituted with the hormone-responsive region of the MMTV promoter (36, 37). However, one must be cautious in translating these results to conditions found in vivo, where histone H1 and non-histone chromatin components may be present. In fact, histone H1 has been shown to be associated with the MMTV promoter region in vivo (6). There is evidence that the GR and PR have different affinities for the various GREs in the MMTV promoter. In vitro the PR has been shown to produce a stronger and more extended footprint on the proximal GRE than the GR (36, 38) and may be more dependent on this element for full activity (16, 38, 39). In addition, various mutations in the proximal element, a complex GRE containing three half-sites, differentially affected the action of the two receptors (38, 39). Thus, nucleoprotein architecture may influence access to either set of GREs, magnifying differences in affinity for either site, and thereby resulting in major differences in the observed transcriptional activity.
Alternatively, the PR may bind its site on the promoter but fail to induce the transition in structure necessary for the binding of other factors and the activation of transcription. Neither the nature of the transition nor the mechanism by which it occurs is well understood. Recent analysis of the stably replicating MMTV promoter has indicated that nucleosomes in the GRE region are not displaced by activated GR (3, 40, 41), although a region of approximately 185 bp becomes hypersensitive to nucleases.2 However, there is an associated loss of histone H1 from the promoter region (6). The transition may therefore be a combination of changes in higher order structure as well as increased binding of factors to the region. How GR affects this transition is unknown but may be facilitated by the SWI·SNF complex (42), which has been shown to be necessary for GR- and ER-mediated gene activation in vivo (43, 44) and to facilitate factor binding to nucleosomes in vitro (45-47). The transiently expressed PR may not efficiently interact with the machinery necessary to remodel chromatin and may require receptor-specific accessory factors that are not expressed in our system. Progestin treatment is known to result in the induction of hypersensitive sites on target genes in other experimental systems (48-50), implying that the PR is not generally deficient in the ability to remodel chromatin.
Our results show that PR is not completely deficient in activation of the stably replicating MMTV template when transiently expressed. However, the amount of activation appears to be saturable (see Fig. 3B) and implies that there are a limited number of templates in the cell population that can respond. This template subpopulation may have a more accessible structure, perhaps due to DNA replication or microheterogeneity in nucleosome positioning (3). We previously reported that upon long term expression, the PR becomes much more efficient at activating the stably replicating template (19). Over time the PR may be able to remodel the template or induce the expression of factors necessary for its function. We cannot yet differentiate between these possibilities, but future studies will explore the roles of DNA replication and receptor processing in the mechanism by which PR becomes competent to function in the context of target genes with complex nucleoprotein structures.
There are multiple examples of endogenous genes in which steroid receptors induce DNase I-hypersensitive sites within the region of their binding sites (48, 49, 51, 52). Steroid receptors may contain domains that can initiate local changes in nucleoprotein structure. The observed differential in ability of the GR and PR to induce the chromatin remodeling event at the MMTV promoter has provided an opportunity to define regions of the GR and PR that are important for the activation of the MMTV promoter in the context of ordered chromatin and may be sites for interaction with receptor-specific accessory factors that provide target gene specificity in vivo.
Functional comparison of two MMTV templates differing in nucleoprotein structure has allowed us to identify layers of transcriptional regulation that involve the interaction of soluble factors with components of ordered chromatin in mammalian cells. Similar comparative studies coupled with the described magnetic affinity cell sorting method should be applicable to examination of other endogenous mammalian promoters and the mechanisms governing their interaction with transcription factors in a complex and physiological structural setting.
We thank members of the Hager lab for helpful discussions. We also thank Dr. Stoney Simons for generously providing the C656G cDNA, Dr. Sandro Rusconi for pMggnOVEC, and Dr. Pierre Chambon for pcPRO. Our thanks are extended also to Dr. David Toft for providing antibody PR22, raised against chicken PR isoforms, and Drs. Wendy Dixon and Judith Campbell for providing affinity purified 12CA5 antibody, directed against the HA epitope.