The Transcriptional Activity of Estrogen Receptor-α Is Dependent on Ca2+/Calmodulin*

Estrogen binds to estrogen receptors in cells, thereby activating the receptors and eliciting biological effects. One of the best characterized effects of estrogen receptor-α (ERα) is transcriptional activation that regulates selected target genes in the nucleus. Work from several laboratories has documented a Ca2+-dependent interaction between ERα and calmodulin. In addition, we previously showed that antagonism of calmodulin function in cells prevented estradiol from inducing ERα transcriptional activity, suggesting that association of ERα with calmodulin participates in ERα function. In this study we adopted a multifaceted approach to directly address this hypothesis. The calmodulin binding domain on ERα was identified and several mutant ERα constructs unable to bind calmodulin were generated. Elimination of calmodulin binding prevented estradiol from stimulating ERα transcriptional activation. Essentially identical results were obtained when intracellular Ca2+ was chelated with the cell-permeable chelator 1,2-bis(o-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid tetra(acetoxymethyl)ester (BAPTA-AM). Moreover, CaM1234, a calmodulin mutant that is unable to bind Ca2+, functioned as a dominant negative construct. Transfection of cells with CaM1234 reduced estradiol-stimulated ERα transcriptional activity. These data indicate that binding to calmodulin is required for normal transcriptional function of ERα.

Estrogen receptors function as ligand-activated transcription factors that regulate the expression of target genes to affect processes as diverse as reproduction, development, and general metabolism (1,2). Estrogen receptor-␣ (ER␣) 1 shares structural organization with other members of the nuclear receptor family. It contains an N-terminal region that harbors a ligand-independent transcriptional activation function (AF-1); a core DNA binding domain, containing two highly conserved zinc finger motifs that target the receptor to specific DNA sequences known as estrogen response elements (ERE); a hinge region that permits protein flexibility to allow for simultaneous receptor dimerization and DNA binding; and a large C-terminal region that encompasses the ligand binding domain, dimerization interface, and a ligand-dependent activation function (AF-2) (1). Upon ligand binding, ER␣ undergoes a conformational change that coordinately dissociates corepressors and facilitates recruitment of coactivator proteins to enable transcriptional activation (3). Protein-protein interaction screens have revealed a large group of proteins classified as coactivators on the basis of their ability to enhance ER␣ action when overexpressed in target cells (3). Some of these proteins have an important role in ER␣ action and provide functional and physical links between the receptor and the transcriptional apparatus. The precise roles of most of these proteins remain to be determined.
Recent studies have shown that regulation of transcription by nuclear ER␣ is more complicated than the classical paradigm would predict (reviewed in Refs. 4 -6.). For example, in some cells both AF-1 and AF-2 are required for maximal transcriptional activities, whereas in others only one is required (7,8). Calmodulin, a ubiquitous Ca 2ϩ sensor protein, appears to participate in ER␣ transcriptional activity. For example, the cell-permeable calmodulin antagonists CGS9343B and trifluoperizine prevented 17-␤-estradiol (E 2 ) from stimulating ER␣ transcription (9,10). Similarly, we documented that E 2 failed to enhance transcriptional activity by ER␣ in cells transiently transfected with a peptide that specifically inhibits calmodulin function in the nucleus (9). Moreover, interaction with calmodulin seems to be required for ER␣ to bind the ERE (11,12) and activate an ER␣-responsive promoter (12). However, the strategies employed in these prior studies were nonspecific or indirect, and recent major reviews continue to ignore the possible role of calmodulin in ER␣ function (1,13,14). Therefore, in this work we employed a multifaceted approach to unequivocally document that a direct interaction of ER␣ with calmodulin is necessary for E 2 to stimulate ER␣ transcriptional activation. We developed a point mutant ER␣ construct (termed ER␣⌬CaM) that is unable to bind calmodulin. Despite binding E 2 with an affinity 3.5-fold greater than that of wild-type ER␣, transcriptional activity of ER␣⌬CaM was not augmented by E 2 . Congruent with our prior documentation that ER␣ binds directly to calmodulin in a Ca 2ϩ -dependent manner (15), chelation of intracellular free Ca 2ϩ ([Ca 2ϩ ] i ) abrogated E 2 -stimulated transcriptional activation of ER␣. Importantly, a point mutant calmodulin that is unable to bind Ca 2ϩ functioned as a dominant negative construct, reducing both basal and E 2 -stimulated transcriptional activity of ER␣. Taken together, these data indicate that calmodulin has an essential role in ER␣ transcriptional activity.
Antibodies-Anti-ER␣ monoclonal (Ab-15) and polyclonal antibodies were from Neomarkers and Santa Cruz Biotechnology, respectively. Anti-hemagglutinin antibody was purchased from Roche Applied Science. The anti-calmodulin antibody has been characterized in detail (16).
Plasmid Construction-The truncated ER␣ cDNAs used in the TNT assays (see Fig. 1C for complete list and specific sequences) were amplified by PCR from pcDNA3-ER␣. All 5Ј-primers contained a BamH1 site and all 3Ј-primers contained an ApaI site. Each PCR fragment was digested with BamH1 and ApaI, then inserted into pcDNA3-myc at the same restriction sites.
Site-directed mutagenesis was performed with the QuikChange sitedirected mutagenesis kit (Stratagene). Plasmid pcDNA3-myc-ER␣ was used as template. The mutant cDNA was amplified with Pfu turbo DNA polymerase using the oligonucleotide 5Ј-CCAAGCCCGCTCATGGAC-GAACGCTCTAAGAAGA-3Ј (mutated residues are underlined). These changes result in replacement of Ile-298 with Glu and Lys-299 with Asp. The construct is termed ER⌬CaM. The sequence of all constructs was confirmed by DNA sequencing.
Cell Culture and Transfection-MCF-7 and COS-7 cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% (v/v) FBS. Cells were plated in 100-mm dishes for Western blots and quantitative PCR, 96-well plates for measurement of transcription, or 12-well dishes for E 2 binding assays. DNA was transiently introduced into cells 24 h after plating using FuGENE 6 according to the manufacturer's instructions. When measuring transcriptional activity, transfections were performed in triplicate with 200 ng of total DNA per well, comprising 10 ng of pRL-TK (which encodes Renilla reniformis luciferase, used as an internal control for transfection efficiency), 40 ng of ERE3-TK-Luc reporter, and 150 ng of ER␣, mutant ER␣, mutant calmodulin (CaM1234) (17) or vector. Six hours after transfection the medium was replaced with phenol red-free medium containing 10% charcoal-treated FBS. Twenty-four hours later, E 2 or an equal amount of vehicle was added to the cells. Cells were incubated for the times indicated in the figure legends, lysed, and processed as described as below.
Luciferase Reporter Assay-Equal numbers of cells were lysed in 50 l (for 96-well plates) or 200 l (for 12-well plates) Passive Lysis Buffer (Promega) and luciferase activity was measured using the Dual Luciferase Reporter Assay (Promega), essentially as described previously (9,18). Briefly, light emission due to firefly luciferase activity was measured using a 300 -650 nm range photomultiplier tube in a Turner Design 20/20 DLReady luminometer for 12 s. Stop & Glo reagent was added to quench the firefly luciferase, and Renilla (control) luciferase activity in the same sample tube was then measured for an additional 12 s. Firefly luciferase activities were normalized for transfection efficiency to the Renilla luciferase internal control. Where indicated, cells were incubated with E 2 or an equal volume of ethanol/dimethyl sulfoxide (vehicle). The concentration and incubation times are indicated in the figure legends. Note that the concentration of BAPTA-AM used did not reduce the viability of the cell lines used in this study (data not shown). In addition, E 2 did not enhance the signal in cells transfected with the TK-Luc plasmid lacking ERE (data not shown).
Western Blot Analysis-Equal amounts of protein lysate were re-solved directly by SDS-PAGE and transferred to PVDF membrane. Membranes were blocked with 5% nonfat powdered milk in TBS-T buffer (25 mM Tris, pH 8.0, 140 mM NaCl, 2.5 mM KCl, and 0.05% Tween 20) and probed with anti-ER␣, anti-hemagglutinin, or anti-calmodulin antibodies. Complexes were visualized with the appropriate horseradish peroxidase conjugated secondary antibody and developed by enhanced chemiluminescence (ECL).
In Vitro Transcription and Translation-[ 35 S]Methionine-labeled ER␣s were produced with the TNT quick-coupled transcription/translation system according to the manufacturer's instructions. As described previously (19), 1 g of DNA was mixed with 40 l of TNT quick master mix and 10 Ci of [ 35 S]methionine (10 mCi/ml, PerkinElmer Life Sciences), then incubated at 30°C for 90 min. Products were confirmed by SDS-PAGE and autoradiography.
Calmodulin Binding-In vitro binding assays were performed by incubating 10 l of reticulocyte lysate containing [ 35 S]methionine-labeled ER␣ (wild-type ER␣ and the ER␣ mutants indicated in the figure legend) with calmodulin-Sepharose beads in 1 ml of Buffer A (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 0.1% Triton X-100, 0.1% protease inhibitor mixture (Sigma), and 1 mM phenylmethylsulfonyl fluoride) in the presence or absence of 1 mM CaCl 2 . Sepharose without calmodulin was used as control. After 4 h at 4°C, beads were washed four times with Buffer A containing 1 mM CaCl 2 or 1 mM EGTA. Proteins were analyzed by SDS-PAGE, and [ 35 S]methionine-labeled ER␣ was detected by autoradiography.
pcDNA3 containing the indicated ER␣ cDNA plasmids, namely ER␣-(1-595), ER␣-(1-270), ER␣-(300 -595), ER␣-(180 -595), ER␣-(180 -353), ER␣-(180 -253), ER␣-(248 -317), or ER␣-(180 -317) (numbers in parentheses correspond to the amino acid residues of ER␣), were transiently transfected into COS-7 cells. Forty-eight hours after transfection, cells were lysed in Buffer A containing 1 mM CaCl 2 and sonicated, followed by centrifugation. After preclearing with Sepharose beads for 1 h at 4°C, equal amounts of protein lysate were incubated with 40 l calmodulin-Sepharose (or Sepharose without calmodulin as control) on a rotator at 4°C for 4 h. Beads were washed four times with Buffer A containing 1 mM CaCl 2 and resuspended in SDS-PAGE sample buffer (20 mM Tris-HCl, pH 7.5, 2% (w/v) sodium dodecyl sulfate, 2% (v/v) ␤-mercaptoethanol, 0.01% (w/v) bromphenol blue, 0.25 M sucrose, and 2 mM EDTA). Samples were heated at 100°C for 5 min and processed by immunoblotting as described above. E 2 Binding Assay-A whole cell ligand binding assay was used to measure the estrogen binding capacity of ER␣ (20). COS-7 cells, plated in 24-well culture dishes, were transfected with wild-type or mutant ER␣. After 24 h, the culture medium was replaced with phenol red-free medium for another 24 h, followed by 2 h incubation with 10 nM [ 3 H]estradiol (PerkinElmer Life Sciences). A 200-fold excess of diethylstilbestrol was added to one set of cells to distinguish between specific and nonspecific binding. The cells were washed once with phosphatebuffered saline supplemented with 1 mg/ml bovine serum albumin and once with phosphate-buffered saline alone. The cells were lysed in buffer containing 10 mM Tris, pH 7.5, 1.5 mM EDTA, 5 mM sodium molybdate, 0.4 M KCl, 1 mM monoglycerol, and 2 mM leupeptin, and disrupted by freeze-thawing three times. The amount of bound radioactivity was quantified by liquid scintillation spectrophotometry. Radioactivity specifically bound to wild-type ER␣ (150,000 to 300,000 cpm) was 10-fold greater than nonspecific binding. A set of cells processed in parallel was used to verify by Western blotting that cells expressed equivalent levels of ER␣.
Quantitative Real-time RT-PCR-Quantitative RT-PCR was performed on the iCycler IQ real-time PCR detection system (Bio-Rad) to examine the effect of E 2 on the estrogen responsive gene pS2. MCF-7 cells transfected with pcDNA3 or CaM1234, and COS-7 cells transfected with wild-type ER␣ or ER␣⌬CaM was incubated with or without E 2 as described above. Total RNA was extracted by TRIzol (Invitrogen) according to the manufacturer's protocol. After quantification in a spectrophotometer, 2 g of RNA was reverse-transcribed to cDNA in a total volume of 40 l using the iScript TM cDNA Synthesis Kit (Bio-Rad) according to the manufacturer's instructions. The resulting cDNA was used in subsequent RT-PCR reactions, performed in 1ϫ iQ SYBR Green Supermix (Bio-Rad) with 5 pmol forward and reverse primers (21). The primers used were, for pS2 hnRNA: forward primer, 5-TTGGAGAAG-GAAGCTGGATGG-3 (start position 3997, within the intron); reverse primer, 5-ACCACAATTCTGTCTTTCACGG-3 (start position 4126, within the second exon); and for ␤-actin: forward primer, 5-TGCGTGA-CATTAAGGAGAAG-3; and reverse primer, 5-GCTCGTAGCTCT-TCTCCA-3. RT-PCR was performed in 96-well optical plates (Bio-Rad) using an iCycler System (Bio-Rad) for 40 cycles (94°C for 10 s, 60°C for 40 s), after an initial 3 min denaturation at 94°C. Serial dilutions of a cDNA sample prepared from MCF-7 cells or COS-7 cells were used to construct standards curves for the pS2 hnRNA amplifications. In all cases, concurrent analysis was performed with ␤-actin as an internal control. All samples and standards were run in triplicate. Results were analyzed using the relative standard curve method as described in Relative Quantitation of Gene Expression (Applied Biosystems User Bulletin 2). pS2 results were expressed relative to ␤-actin as internal control.
Miscellaneous-Statistical analysis was performed by Students's t test, using InStat software (GraphPad Software, Inc.). Protein concentrations were determined with the DC Protein Assay (Bio-Rad).

Develop ER␣ Mutants That Do Not Bind to Calmodulin-We
previously reported that ER␣ interacted directly with calmodulin and that antagonism of calmodulin attenuated ER␣ transcription (9,15). To further elucidate the role of calmodulin in ER␣ function, we set out to develop ER␣ mutant constructs that would not bind calmodulin. The first step was to identify the calmodulin binding region in ER␣. The strategy adopted was to generate radiolabeled fragments of ER␣ and examine their ability to bind to calmodulin-Sepharose (Fig. 1). [ 35 S]Methionine-labeled full-length ER␣ (ER␣-1-595) bound to calmodulin-Sepharose (Fig. 1B). Specificity of binding to calmodulin is evident by the absence of an interaction with Sepharose alone. Binding was markedly reduced when Ca 2ϩ was chelated with EGTA (data not shown). Analysis of fragments of ER␣ revealed that the middle portion of the molecule was required for maximal binding (Fig. 1). Longer exposure of the autoradiogram in Fig. 1B revealed a weak band in the lane containing ER␣ (1-270) (data not shown), implying a possible low affinity binding site in this region of the molecule. By generating multiple fragments of different sizes, residues 248 -317 were observed to be necessary for maximal binding (Fig. 1C). Inspection with a helical wheel projection of the sequence of ER␣ in this region revealed a predicted calmodulin binding motif between amino acids 298 and 317 (IKRSKKNSLALSLTADQMVS).
The effect of deletion of residues 298 -317 of ER␣ on its ability to bind calmodulin was examined next. COS-7 cells were transfected with ER␣, and lysates were incubated with calmodulin-Sepharose in the presence of Ca 2ϩ . Full-length ER␣ bound to calmodulin ( Fig. 2A). Binding to calmodulin was specific as no ER␣ was detected in samples incubated with Sepharose alone. Deletion of residues 298 -317 of ER␣ eliminated its ability to bind calmodulin ( Fig. 2A). Similarly, deletion of smaller portions of ER␣, namely amino acids 298 -310 and 298 -303, abrogated binding to calmodulin. Based on our knowledge of the interaction of calmodulin with target proteins (22), we hypothesized that Ile-298 and Lys-299 were necessary for the interaction of ER␣ with calmodulin. Consistent with our prediction, an ER␣ point mutant in which Ile-298 and Lys-299 were replaced, termed ER␣⌬CaM, failed to bind to calmodulin ( Fig. 2A) To verify the findings, calmodulin binding assays were also performed with [ 35 S]methionine-labeled ER␣. Wild-type ER␣ bound to calmodulin in the presence of Ca 2ϩ ; minimal ER␣ was detected when Ca 2ϩ was chelated with EGTA ( Fig. 2B). As was observed with transfected cells, neither radiolabeled ER␣⌬298 -317 nor ER␣⌬CaM was able to bind calmodulin. Note that equal amounts of [ 35 S]methionine-labeled ER␣ were used (Fig. 2B, input).
Comparison of Transcriptional Activities of Wild-type and Mutant ER␣-The effects of calmodulin on ER␣ transcriptional activation were examined by comparing wild-type and mutant ER␣ in a well characterized assay that measures the activity of an ER␣-responsive reporter (9). To eliminate interference from endogenous ER␣, we used COS-7 cells that lack ER␣. Incubation with E 2 enhanced ER␣ transcriptional activity 5.2-fold in COS-7 cells transiently transfected with wild-type ER␣ (Fig.   3A). In contrast, transcriptional activity of ER␣⌬298 -317, which lacks amino acids 298 -317, was not significantly stimulated by E 2 . It is possible that the removal of 20 amino acids may produce a large conformational change in ER␣, rendering it transcriptionally inactive. Therefore, analysis was repeated using ER␣⌬298 -303, which lacks only six amino acids. Although basal transcriptional activity of ER␣⌬298 -303 was ϳ2fold higher than that of wild-type ER␣, E 2 was unable to augment transcription (Fig. 3A). Similarly, E 2 had no effect on the transcriptional activity of the point mutant construct ER␣⌬CaM. Absent detailed structural information, it is impossible to completely exclude that the conformation of ER␣⌬CaM is altered. However, it is most unlikely that changing only two amino acids would sufficiently perturb the tertiary structure to prevent ER␣ binding to transcription factors. These data con- firm that E 2 cannot stimulate transcriptional activation of ER␣, which lack the ability to bind calmodulin.
It is conceivable that lack of E 2 stimulation of transcription of the mutant ER␣ could be due to altered receptor expression. Therefore, an aliquot of the cell lysate was removed prior to the transcription assay and the amount of ER␣ was evaluated by Western blotting. Mutant ER␣ was expressed to the same level among all the samples (Fig. 3B). Consistent with published observations in breast epithelial cells by this group (15) and others (23,24), E 2 reduced the amount of wild-type ER␣ in COS-7 cells (Fig. 3B). Interestingly incubation with E 2 did not substantially alter the amount of mutant ER␣. The mechanism responsible for this finding is not known. E 2 Binding-The calmodulin binding region overlaps the proximal portion of the ligand binding domain (see Fig. 1A). It is therefore feasible that the inability of E 2 to stimulate transcriptional activation of the mutant receptors may be due to an inability of these receptors to bind to E 2 . This possibility was addressed by examining binding of [ 3 H]estradiol to ER␣ with a whole cell ligand binding assay. Transfecting COS-7 cells with wild-type ER␣ yielded robust binding of [ 3 H]estradiol (Fig. 4). (Specific binding was 10-fold greater than nonspecific binding.) In contrast, ER␣⌬298 -317 completely failed to bind E 2 . The findings with ER␣⌬298 -303 and ER␣⌬CaM were completely different; the E 2 binding capacity of these ER␣ mutants was 3.5-fold greater than that of wild-type ER␣ (Fig. 4). These data validate that ER␣298 -303 and ER␣⌬CaM can bind E 2 , and the lack of transcriptional activity is not because of an inability to bind E 2 .
Ca 2ϩ Is Required for E 2 -stimulated Transcriptional Activity-We (15) and others (25,26) have documented that Ca 2ϩ is necessary for the interaction between calmodulin and ER␣. Therefore, on the basis of the data in Fig. 3, one would anticipate that chelation of [Ca 2ϩ ] i would reduce the ability of E 2 to stimulate transcription by ER␣. This hypothesis was tested by incubating MCF-7 cells with the cell-permeable Ca 2ϩ chelator BAPTA-AM. Chelation of intracellular Ca 2ϩ abrogated E 2stimulated ER␣ transcriptional activity (Fig. 5A). BAPTA, which did not significantly alter cell viability under our assay conditions, had no significant effect on basal ER␣ transcription. In addition, Western blotting demonstrated that BAPTA did not substantially change the amount of ER␣ in the cells (Fig.  5B). These data are consistent with a requirement of Ca 2ϩ / calmodulin binding for ER␣ transcriptional activity.
Effect on ER␣ Transcriptional Activity of a Mutant Calmodulin Unable to Bind Ca 2ϩ -Calmodulin contains four Ca 2ϩ binding domains, two in the N-terminal-half and two in the C-terminal-half of the molecule (27). The identification of the specific amino acids in calmodulin that bind Ca 2ϩ has permitted the development of point mutant calmodulin constructs that are unable to bind Ca 2ϩ (17, 28 -30). These mutants have been shown to function as dominant negative calmodulin constructs in some situations (30,31). To examine a possible effect on ER␣ transcription, CaM1234 was transfected into MCF-7 cells and ER␣ transcriptional activity was measured. Transient transfection of CaM1234 reduced by 50% basal ER␣ transcriptional activity (Fig. 6A). In addition, E 2 was unable to significantly enhance ER␣ transcriptional activity in cells transfected with CaM1234 (Fig. 6). Consistent with our prior documenta- tion that a mutant calmodulin unable to bind Ca 2ϩ has no effect on the number of ER␣ in MCF-7 cells (15), CaM1234 did not alter the level of endogenous ER␣ (data not shown). (The expression of CaM1234 was verified by Western blotting which revealed an increase in total cellular calmodulin levels (data not shown).) These findings indicate that the reduction of ER␣ transcriptional activity produced by CaM1234 is not because of a change in the level of ER␣.
Effects of ER⌬CaM, BAPTA, and CaM1234 on E 2 -induced Transcriptional Activity of the Estrogen Responsive Gene pS2-All of the ER␣ transcriptional activity data presented above were derived with a transfected ERE reporter. In order to demonstrate physiological relevance, we examined the possible effects of calmodulin on the ability of ER␣ to activate an endogenous estrogen responsive gene. E 2 -induced pS2 gene expression was measured by quantitative RT-PCR. pS2 expression is accepted as a reliable measure of transcriptional activity of ER␣ (21,32,33).
When COS-7 cells were transfected with wild-type ER␣, E 2 significantly increased the transcriptional activity of the pS2 gene (Fig. 7A). In contrast, in cells transfected with an equivalent amount of ER␣⌬CaM the ability of E 2 to increase pS2 hnRNA was dramatically reduced. Thus, the elimination of calmodulin binding prevented ER␣ from increasing transcription of two independent reporters in response to E 2 .
The effects of chelating [Ca 2ϩ ] i were evaluated in the same assay. Analogous to the results with ERE, BAPTA eliminated E 2 -induced transcriptional activation of pS2 in MCF-7 cells (Fig. 7B). Finally, transcription of pS2 was analyzed in the presence of CaM1234. Transfection of MCF-7 cells with native calmodulin did not substantially alter E 2 -induced expression of pS2 hnRNA (Fig. 7C). In contrast, CaM1234 reduced by 33% the ability of E 2 to promote transcriptional activation of pS2 hnRNA by endogenous ER␣ (Fig. 7C). The magnitude of the reduction of pS2 hnRNA activation produced by CaM1234 was very similar to the magnitude of the inhibition CaM1234 produced on E 2 -stimulated ERE transcription (see Fig. 6). Collectively, the data obtained with pS2 using three complementary strategies, namely ER⌬CaM, BAPTA and CaM1234, validate the findings generated with ERE that calmodulin is important for E 2 -stimulated transcriptional activity of ER␣. DISCUSSION A fairly large body of published literature implicates Ca 2ϩ and calmodulin in ER␣ function (Ref. 15 and references therein). Notwithstanding this evidence, relatively little attention is focused on the role of calmodulin in ER␣ signal transduction. Several factors are likely to account for this situation. Probably one of the most important reasons is that the majority of evidence implicating Ca 2ϩ and calmodulin in ER␣ function has been derived with chemical inhibitors of calmodulin and is therefore indirect. In order to address this deficiency, we combined the use of mutant and dominant negative constructs to validate that calmodulin is necessary for normal transcriptional function of ER␣.
The initial strategy was to generate mutant ER␣ that is unable to bind calmodulin. Although lacking sequence homology, calmodulin binding domains on target proteins generally fall into one of two main groups: (i) short regions (ϳ14 -26 amino acid residues) that form basic amphiphilic ␣-helices (34) and (ii) the IQ motif, which has a consensus sequence

FIG. 5. Chelation of [Ca 2؉ ] i inhibited E 2 -induced ER␣ transcriptional activity.
A, MCF-7 cells were transiently co-transfected with ERE3-TK-Luc and pRL-TK. Vehicle (DMSO) or 20 M BAPTA was added, followed immediately with vehicle (clear bars) or 10 nM E 2 (shaded bars). After incubating for 16 h, samples were assayed for luciferase activity. Results are expressed relative to cells treated with vehicle alone, which was set as 1.0. *, significantly different from vehicle (p Ͻ 0.01); †, not significant (p Ͼ 0.05). Data are the means of three separate experiments, each performed in duplicate. Means Ϯ S.E. are shown. B, equal amounts of protein lysate from the cells treated as described in A were removed and processed by SDS-PAGE. After transfer to PVDF, blots were probed with anti-ER␣ antibody, and antigenantibody complexes were identified by ECL. A representative experiment is shown.

FIG. 4. E 2 binding capacity of full-length and mutant ER␣.
COS-7 cells were transiently transfected with full-length ER␣ (WT), ER␣⌬298 -317, ER␣⌬298 -303, or ER␣⌬CaM. Twenty-four hours later the culture medium was replaced with phenol red-free medium for another 24 h, followed by 2 h of incubation with 10 nM [ 3 H]estradiol. Cells were lysed, and the amount of bound radioactivity was quantified by liquid scintillation spectrophotometry. Results are expressed relative to cells transfected with wild-type ER␣ alone, which was set at 1.0. *, significantly different (p Ͻ 0.05); **, significantly different (p Ͻ 0.001). Data are the means of three separate experiments, each performed in duplicate. Means Ϯ S.E. are shown. The absence of error bars from ER␣⌬298 -317 indicates that the range is too small to be visible. IQXXXRGXXXR (where X is any amino acid) (22,35). No IQ motif was detected in ER␣ and the sequence (corresponding to amino acids 298 -317 in human ER␣) to which calmodulin binds does not correspond to a canonical target motif (35). These findings are not surprising as recent evidence reveals considerable diversity and variability in the interaction of calmodulin with its target molecules (36). Our data indicate that amino acid residues 298 -303 of ER␣ are necessary for calmodulin binding, with Ile-298 and Lys-299 appearing to be essential.
The calmodulin binding domain in ER␣ is close to the middle of the receptor, located at the N-terminal portion of the hor-mone binding domain and extends partially into the hinge region. This location has potentially important consequences for ER␣ function. In fact, analysis of transcriptional function revealed that all the ER␣ mutant constructs that cannot bind calmodulin failed to increase transcription in response to E 2 . In order to minimize the possible disruption of ER␣ structure produced by deletion of amino acids, we generated a point mutant construct that is unable to bind calmodulin. Changing only two amino acids is highly unlikely to produce a dramatic change in conformation. Like the deletion mutants, ER␣⌬CaM was unable to increase transcription when stimulated with E 2 . These data provide direct evidence to validate our prior observations (obtained with calmodulin inhibitors) (9) that calmodulin has a role in E 2 -stimulated transcriptional activity of ER␣. Nevertheless, some caveats of our work should be borne in mind. Binding of ligand to ER␣ induces conformational changes in ER␣, resulting in enhanced transcription. Calmodulin binds in the hinge region of ER␣, immediately proximal to the ligand binding domain. Therefore, it is conceivable that the two amino acid substitutions could modify secondary or tertiary structure of ER␣. Should this occur, ER␣ dimerization, DNA binding and/or interaction with other targets might be altered independent of calmodulin binding. This premise can be eliminated only by solving the structure of the mutant ER␣.
Our data differ from a prior report describing a mutant ER␣ deficient in calmodulin binding (10). In that work, a point mutant ER␣, termed ER␣ (K302G, K303G) (the lysines at 302 and 303 were replaced with glycine), had the same basal and E 2 -stimulated transcriptional activity as wild-type ER␣ (10). Several factors might contribute to the disparate results between the studies. Although exhibiting reduced affinity for calmodulin, ER␣ (K302G, K303G) is able to bind ϳ20% as much calmodulin as wild-type ER␣. In contrast, ER␣⌬CaM used in our study bound no detectable calmodulin. In addition, we used different reporter constructs and cell lines to those employed in the prior work. Another potentially important factor to account for the findings obtained with ER␣ (K302G, K303G) by Ramos's group is that Lys-302 and Lys-303 of ER␣ are targets for acetylation by p300 (37). In fact, Wang et al. (37) observed that ER␣ mutated at Lys-302 and Lys-303 exhibited E 2 -stimulated transcriptional activity that was 2-4-fold higher than that of wild-type ER␣. Consistent with the latter study, other groups observed that mutation of Lys-303 resulted in hypersensitivity to E 2 in ER␣ transactivation assays (38,39). The reason for the discrepancy between the reports that examined the Lys-303 ER␣ mutants is not clear. These factors, particularly the retention of some calmodulin binding and elimination of ER␣ acetylation, confound interpretation of the role of calmodulin in ER␣ function in the study by Ramos's group. Importantly, the point mutations introduced into ER␣⌬CaM used in our study abrogated calmodulin binding. Moreover, to the best of our knowledge, no post-translational modifications have been identified for the residues we mutated, namely Ile-298 or Lys-299. Therefore, we believe that our investigation for the first time directly assesses the role of calmodulin in ER␣ transcription.
Unexpectedly, some ER␣ constructs deficient in calmodulin binding exhibited enhanced basal transcriptional activity. While the cause of this effect has not been identified, a possible clue is provided by the E 2 binding analyses: ER␣⌬CaM bound 3.5-fold more E 2 than wild-type ER␣. Although the molecular mechanism responsible for the enhanced binding is not known, these findings are consistent with the observation that calmodulin mediates a decrease in E 2 binding to ER␣ (11). The calmodulin binding domain is immediately proximal to (and may even partially overlap) the E 2 binding site, suggesting that FIG. 6. A mutant calmodulin unable to bind Ca 2؉ functions as a dominant negative construct in ER␣ transcriptional activity. MCF-7 cells were transiently transfected with ERE3-TK-Luc, pRL-TK, and pcDNA3 vector (V) or pcDNA3-CaM1234 (CaM1234). Twenty-four hours later, cells were treated with vehicle (EtOH, clear bars) or 10 nM E 2 (shaded bars) for 16 h. After lysis, luciferase activity was determined by luminometry. Results are expressed in two different ways to facilitate interpretation: A, relative to cells transfected with vector and incubated with ethanol, which was set as 1.0 (to permit analysis of the effects of CaM1234). B, relative to ethanol alone, which was set as 1.0 (to permit interpretation of the effects of E 2 ). *, significantly different (p Ͻ 0.001); † , not significant (p Ͼ 0.05). Data are derived from four separate experiments, each performed in triplicate. Means Ϯ S.E. are shown. calmodulin may sterically hinder E 2 binding. As we previously hypothesized (15), it is likely that direct binding to calmodulin alters the tertiary conformation of ER␣. Absent calmodulin binding, the receptor may adopt a conformation that permits increased E 2 binding. Solving the structure of ER␣⌬CaM is necessary to provide a definite answer.
Two independent approaches bolster our observations on transcription derived with the mutant ER␣ constructs. [Ca 2ϩ ] i was chelated with BAPTA (40). Cells were loaded with BAPTA-AM, the acetoxymethyl ester of BAPTA. This derivative is non-polar and permeates into cells where endogenous esterases remove the ester groups, leaving the membrane-impermeable chelator trapped in the cell. BAPTA has been widely used to investigate intracellular Ca 2ϩ function (40 -42). However, caution should always be exercised when interpreting results obtained in cells with chemical compounds. For example, chelation of [Ca 2ϩ ] i can produce effects in the cell that are independent of calmodulin. Notwithstanding this caveat, Ca 2ϩ is necessary for the interaction between calmodulin and ER␣ (15,25,26) and Fig. 2B, this study). Therefore, one would anticipate that if calmodulin binding is necessary for E 2 -stimulated transcription, removing [Ca 2ϩ ] i would eliminate E 2stimulated transcription. That is exactly what we observed, further supporting our hypothesis.
The last strategy employed CaM1234, a mutant calmodulin incapable of binding Ca 2ϩ (17,30). CaM1234 has a dominant negative effect in some cellular processes (30,31). Congruent with these observations, CaM1234 was a dominant negative in ER␣ transcription. CaM1234 reduced E 2 -stimulated ER␣ transcriptional activity for both ERE and the endogenous estrogen responsive gene pS2. Because the Ca 2ϩ -deficient CaM1234 is not expected to bind ER␣, the dominant negative effect is presumably mediated by the Ca 2ϩ -independent binding of CaM1234 to (an)other molecule(s) that is (are) a component of FIG. 7. Effects of ER⌬CaM, BAPTA, and CaM1234 on E 2 -induced transcriptional activity of the estrogen responsive gene pS2. A, COS-7 cells were transiently transfected with wild-type ER␣ (WT) or ER␣⌬CaM (⌬CaM). Twenty-four hours later the culture medium was replaced with phenol red-free medium for 24 h, followed by a 6-h incubation with vehicle (EtOH, clear bars) or 100 nM E 2 (shaded bars). Total RNA was extracted and subjected to quantitative RT-PCR analysis to determine the expression levels of pS2 hnRNA. For all RT-PCR assays, the relative levels of mRNA were normalized with ␤-actin mRNA. Values obtained in vehicle-treated cells transfected with wild-type ER␣ were set to 1. Data represent the means Ϯ S.E. of three independent experiments, each performed in triplicate. B, MCF-7 cells were grown in phenol red-free medium for 24 h. Cells were incubated with vehicle (DMSO) or 20 M BAPTA, followed immediately with vehicle (clear bars) or 100 nM E 2 (shaded bars) for 6 h. Total RNA was prepared and subjected to RT-PCR analysis as described for A. Data are representative of two independent experimental determinations. Error bars depict S.E. derived from triplicate analysis. C, MCF-7 cells were transiently transfected with pcDNA3 vector (V), CaM1234 or native calmodulin (CaM). Twenty-four hours later, cells were incubated with vehicle (EtOH) or 100 nM E 2 for 6 h. RT-PCR was performed as described for A. Data are representative of two independent experimental determinations. Error bars depict S.E. derived from triplicate analysis. the ER␣ transcriptional apparatus. Regardless of the mechanism, our findings, obtained using diverse, but complementary analytical strategies, reveal that E 2 -stimulated transcriptional activity of ER␣ is dependent on Ca 2ϩ /calmodulin. The specific role played by calmodulin in E 2 -stimulated ER␣ transcription is not known. We previously documented that calmodulin binds to ER␣ in a Ca 2ϩ -dependent manner, thereby reducing ER␣ degradation (15). However, the stabilizing function of calmodulin appears to be independent of its transcriptional effect (9), and thus does not explain the requirement of calmodulin for transcription. The interaction of calmodulin with its targets frequently induces a conformational change in the target protein (43). Therefore, a more probable explanation for the effect of calmodulin on ER␣ transcription is that the binding of calmodulin alters the tertiary conformation of ER␣, modulating the interaction of the receptor with (an)other protein(s). The ability of calmodulin to alter ER␣ conformation can be inferred from the protection afforded by calmodulin to in vitro proteolysis of ER␣ (15). Considerable effort over the past few years has been directed toward the identification and characterization of adaptor proteins that modulate ER␣ function. These targets include coactivators and corepressors (3). Recent evidence reveals that calmodulin is an integral component of an ER␣-ERE complex (12). Because calmodulin was necessary, but not sufficient, for complex formation, auxiliary proteins are believed to participate (12). Therefore, calmodulin is likely to modulate ER␣ function by altering the binding of ER␣ to transcriptional coactivators or corepressors. Collectively, these data suggest that the molecular interaction between calmodulin and ER␣ could be the target for therapeutic intervention in patients with breast cancer.