Selective Mutations in Estrogen Receptor α D-domain Alters Nuclear Translocation and Non-estrogen Response Element Gene Regulatory Mechanisms*

The three main mechanisms of ERα action are: 1) nuclear, genomic, direct DNA binding, 2) nuclear, genomic, “tethered”-mediated, protein-protein interactions, and 3) non-nuclear, non-genomic, rapid action responses. Reports suggest the D-domain or hinge region of ERα plays an important role in mechanisms 1 and 2 above. Studies demonstrating the functionality of the ERα hinge region have resected the full D-domain; therefore, site directed mutations were made to attribute precise sequence functionality to this domain. This study focuses on the characterization and properties of three novel site directed ERα- D-domain mutants. The Hinge 1 (H1) ERα mutant has disrupted nuclear localization, can no longer perform tethered mediated responses and has lost interaction with c-Jun, but retains estrogen response element (ERE)-mediated functions as demonstrated by confocal microscopy, reporter assays, endogenous gene expression and co-immunoprecipitation. The H2 ERα mutant is non-nuclear, but translocates to the nucleus with estradiol (E2) treatment and maintains ERE-mediated functionality. The H2+NES ERα mutant does not maintain nuclear translocation with hormone binding, no longer activates ERE-target genes, functions in ERE- or tethered-mediated luciferase assays, but does retain the non-genomic, non-nuclear, rapid action response. These studies reveal the sequence(s) in the ERα hinge region that are involved in tethered-mediated actions as well as nuclear localization and attribute important functionality to this region of the receptor. In addition, the properties of these ERα mutants will allow future studies to further dissect and characterize the three main ERα mechanisms of action and determine the mechanistic role each action has in estrogen hormone regulation.

Many of the biological effects of estrogen are mediated through the estrogen receptors (ERs), 2 ER␣ and ER␤, which belong to the nuclear receptor superfamily (1). Focusing on ligand-dependent activation, to date, there are three main mechanisms of action for ER␣ that include 1) nuclear, genomic, direct DNA binding, 2) nuclear, genomic, "tethered"mediated protein-protein interactions, and 3) non-nuclear, non-genomic, rapid action responses (2)(3)(4)(5)(6)(7)(8). Mechanism 1 involves liganded ER␣ bound to estrogen response elements (EREs) of target genes to mediate changes in gene expression via the "classical" ER␣ DNA binding responses (4,9). The 2nd mechanism involves the recruitment and interaction of ER␣ which "tethers" to other transcription factors, such as c-Jun and Sp1, forming a protein-protein complex that interacts directly with the AP-1 and Sp1 DNA response elements, respectively (7,10,11). Lastly, the 3rd mechanism, involves a small population of non-nuclear ER␣ that mediates rapid signaling events which include cellular calcium mobilization, nitric oxide synthesis, and activation of intracellular signaling cascades such as those involving MAPK/ERK, Src, or Akt (5,12).
ER␣, as with all nuclear receptors, maintains the classical domain demarcations. Each domain can act independently, but for full functionality, proper spatial orientation is necessary for transactivation of target genes (8,(13)(14)(15). The A/B domain of ER␣ harbors the hormone-independent activation function domain (AF-1), the C-domain holds the DNA binding domain (DBD), the D-domain or hinge region contain putative nuclear localization sequences (NLS), and the E/F domain harbors the ligand binding domain (LBD) and the hormone-dependent activation function (AF-2) (15,16). The D-domain originally viewed as a flexible linker between the DBD and the LBD, is demonstrating to be critical for proper control of nuclear receptor activity (17,18). The hinge region of ER␣ is important for proper conformational changes (13), contains the putative NLS (14,19) and is important for tethered-mediated signaling (7, 11, 20 -23).
The classical ER␣-mediated DNA binding responses are well established and are abrogated by specific disruption of two amino acid substitutions (E207A/G208A) in the DNA binding domain (10,23). This ER␣ mutant only performs tethered and rapid action responses, as demonstrated by in vivo microarray of the mouse uterus showing that the tethered response accounts for ϳ25% of the WT transcripts (3). The functions of classical and rapid action-mediated ER␣ responses apart from the tethered-mediated ER␣ responses are not defined. Rapid action, non-nuclear ER␣ responses are attributed to the same ER␣ as the genomic responses (4 -6, 12, 22, 24 -27); however, it is challenging to separate the genomic ER␣ responses from the non-genomic ER␣ responses. Non-nuclear mechanisms are thought to involve activation of membrane-initiated kinase cascades. ER␣ does not have a transmembrane domain, but ER␣ interacts with caveolin-1 (6) and contains a palmitoylation sequence (28), which allows ER␣ to localize to the membrane. Rapid estrogen signaling via membrane-associated ER␣ leads to (a) MAPK, Akt, p21 Ras , Raf, and PKC activation, (b) alterations of potassium channels, (c) increase in intracellular Ca ϩ2 levels, and (d) release of nitric oxide and stimulation of prolactin secretion in various cell and tissue types (6,29). The estrogen dendrimer complexes (EDCs) activate p44/42 MAPK (ERK1/2), Shc, and Src and are ineffective in stimulating endogenous estradiol target genes (30). Rapid action data obtained from the EDCs allow for ER␣ signaling from the membrane but, this method does not define non-nuclear actions alone as endogenous ER␣ is present and able to mediate genomic actions (5).
Previously, studies that have examined the hinge region of ER␣ have completely deleted the D-domain, rendering the receptor without the functionality of this domain (14,31,32). These studies attributed to the understanding of the functionality of the D-domain of ER␣, but the exact sequences in the hinge region of ER␣ critical for nuclear localization and interaction with c-Jun are not identified. Functionality of cytoplasmic/membrane only ER␣ or functionality of ER␣ apart from tethered-mediated activity have not been described. Therefore, with previous literature suggesting that the D-domain is involved in these functions of ER␣, we sought to create specific ER␣ mutants to define these actions. Mutations in the hinge region of ER␣ were generated by site directed mutagenesis rather than deletion mutations. Hinge mutants H1 ER␣, H2 ER␣, and H2ϩNES ER␣ have unique properties that block ER␣ tethered-mediated effects, block nuclear localization without ligand, and block liganded nuclear genomic-mediated actions while maintaining non-nuclear, rapid action responses, respectively. These ER␣ mutants define sequence(s) in the ER␣ D-domain responsible for actions involving nuclear localization and gene selective activation and demonstrate that the D-domain of ER␣ is critical for not only receptor nuclear localization but also in selective transcriptional regulation.
Production of Lentivirus and Stable Cell Lines-All lentivirus was packaged in HEK293T/17 cells according to published protocols (36). Briefly, 293T cells were transiently transfected with pMD2G, psPAX2, and pDEST673 carrying the neomycin resistance gene and the desired ER␣ mutant using Lipofectamine 2000. Supernatant was collected 48 h post transfection and concentrated by centrifugation at 50,000 ϫ g for 2 h. Pellets were resuspended in PBS and used for infection. Titers were determined using quantitative PCR to measure the number of lentiviral particles integrated into the host genome. MOI ranging from ϳ180 to 25 were used for infection of Ishikawa cells. After 3 days of infection, cells selected with Geneticin (1.2 mg/ml, Invitrogen, #11811-031) and a stably pooled population of cells was obtained after 2 weeks. Stable integration of ER␣ was verified by Western blot for ER␣.
Transient Transfection and Luciferase Assay-Cells were seeded in 24 well plates overnight. A total of 0.5 g of DNA, including 0.2 g of expression plasmids (pcDNA/vector, pcDNA/ER␣, pcDNA/H1-ER␣, pcDNA/H2ϩNES-ER␣, or pcDNA/NES-ER␣) 0.2 g of reporter plasmids (3xERE Luc or AP-1 Luc) and 0.1 g of pRLTK plasmids, were transfected overnight using the Effectene transfection reagent (Qiagen, Valencia, CA) according to the manufacturer's protocol. Cells were starved with 10% HyClone Charcoal/Dextran-stripped FBS (sFBS) (Thermo Scientific, Waltham, MA) for 8 h and then were treated as described in the figure legends. For luciferase experiments co-expressing c-Jun, cells were plated and transfected as above with a total of 0.5 g of DNA, including 0.2 g of reporter plasmid, 0.1 g of ER␣ expression plasmids and/or 0.1 g of pRSV/c-Jun and 0.1 g of pRLTK. For experiments coexpressing SRC-2, cells were plated and transfected as above with a total of 0.75 g of DNA, including 0.2 g of reporter plasmid, 0.1 g of ER␣ expression plasmid, 0.4 g of pcDNA3/SRC-2, and 0.05 g of pRLTK. Luciferase assays were performed using the Dual-Luciferase Reporter Activity System (Promega, Madison, WI) according to the manufacturer's protocol. Data are from three independent experiments.
Confocal Microscopy-HeLa cells were grown overnight on Lab-Tek 2 well chamber slides (NUNC, Rochester, NY), transfected with 0.5 g of pEGFP-ER␣ expression plasmids as described above. Cells were starved with 10% sFBS for 8 h and then were treated as described in the figures and figure legends. The cells were then fixed with 4% paraformaldehyde for 1 h. Cells were washed with PBS and coverslipped with ProLong Gold Anti-Fade reagent with DAPI (Invitrogen, #P-36931). The cellular localization of ER␣ was visualized by confocal microscopy (Zeiss LSM-510 equipped with an argon-krypton laser) using a 40 ϫ 1.2 objective lens.
Co-immunoprecipitation (co-IP)-293F cells were plated in 10-cm dishes overnight. Cells were transfected overnight as above with 2 g of pcDNA/ER␣ plasmid (WT and H1 mutant) and/or 3 g of pRSV/c-Jun. Cells were treated for 36 h with vehicle or 10 nM E 2 . Cell lysates were prepared in 400 l of RIPA buffer. HIP buffer ((50 mM Tris-Cl (pH 8.0), 150 mM NaCl, 1% Triton X-100 and protease inhibitor mixture) was added to the cell lysate to a final volume of 4 ml. The cell lysates were precleared with protein A-Sepharose (Amerham, CL-4B, 17-0780-01) for 1 h at 4°C with rocking. Protein lysates (1 ml) were immunoprecipitated by using 5 g of anti-ER␣ (H-184), anti-c-Jun, rabbit IgG, and mouse IgG overnight at 4°C with rocking. The protein antibody complexes were bound to protein A-Sepharose, suspended in the lysis buffer, for 4 h at 4°C with rocking. The bound complexes were washed 6 times with HIP buffer, precipitated, and boiled in SDS sample buffer. The immunoprecipitated proteins were then resolved with the 10% Tris-glycine gel system. Proteins were transferred to nitrocellulose and the presence of ER␣ or c-Jun in the precipitants was detected by anti-ER␣ (MC-20) or c-Jun. The immunoreactive products were detected by the ECL Plus Detection System. RNA Extraction and Real-time PCR-Total RNA was extracted by using TRIzol Reagent (Invitrogen) according to the manufacturer's protocol. First-strand cDNA synthesis was performed using superscript reverse transcriptase according to the manufacturer's protocol (Invitrogen). The mRNA levels of progesterone receptor (PR), pS2, and p21 were measured using SYBR green assays (Applied Biosystems). The sequences of primers used in real-time PCR were as follows: for human PR (NM_000926.4): the forward primer 5Ј-GACGTGGAG-GGCGCATAT-3Ј, reverse primer 5Ј-GCAGTCCGCTGTCC-TTTTCT-3Ј, for human pS2 (NM_003225.2): the forward primer 5Ј-GCCCTCCCAGTCTGCAAATA-3Ј, reverse primer 5Ј-CTGGAGGGACGTCGATGGTA-3Ј, for human p21 (NM_078467.1): the forward primer 5Ј-CCTGTCACTGTCT-TGTACCCT-3Ј, reverse primer 5Ј-GCGTTTGGAGTGGTA-GAAATCT-3Ј. Cycle time values were obtained using the ABI PRISM 7900 Sequence Detection System and analysis software (Applied Biosystems, Foster City, CA). Each sample was quantified against its ␤-actin transcript content: the forward primer 5Ј-GACAGGATGCAGAAGGAGATCAC-3Ј, reverse primer 5Ј-GCTTCATACTCCAGCAGG-3Ј, or its GAPDH transcript content: the forward primer 5Ј-ATGGGGAAGGT-GAAGGTCG-3Ј, reverse primer 5Ј-GGGGTCATTGATGGC-AACAATA-3Ј and then normalized with respect to the control group. Quantification was performed according to the mathematical model described by Pfaffl (37) and as previously described (33). The experiments were repeated three times and results are presented as fold increase ϩ S.D.
MAPK Analysis-Cells were seeded in 60-mm dishes, cultured overnight in phenol red-free DMEM:F12 medium with 10% FBS, and transfected with 2 g of ER␣ expression plasmids as indicated in the figure legends for 8 h following the Effectene manufacturer's protocol. The transfected cells were then starved in phenol red-free DMEM:F12 with 0.1% sFBS for 3 days. Ishikawa stable cell lines were seeded and starved as above. Cells were treated with 100 nM E 2 for 0, 3, 5, 10 min. Cells were placed on ice, washed with cold PBS, and lysed in ice-cold lysis buffer (1% Igepal (Sigma #56741), 0.5% sodium deoxycholate, 0.1% SDS, 1ϫ complete mini-tab (Roche 11-836-170-001), 10 mM sodium fluoride, and 1 mM sodium orthovanadate), for 30 min followed by sonication with a probe sonicator for 15 s on ice (setting 6 on 60 Sonic Dismembrator from Fisher Scientific). The supernatant (2 g) was used for Western blot analysis as described above.
Statistical Analysis-One-way ANOVA with Tukey's posttest was performed using GraphPad Prism version 5.00 for Windows, GraphPad Software, San Diego, CA. Two-way ANOVA with Bonferroni post-test was also performed using GraphPad Prism version 5.00.

Selective Mutations in the D-domain of ER␣ Disrupt Nuclear
Localization-The D-domain of ER␣ is important for nuclear localization and interaction via the tethered mechanism (10,20,21). Computational analysis of ER␣ by LOCtree and Motif Scan suggested putative nuclear localization signals in the C-domain and D-domain (amino acids 234 -240 and 270 -275; amino acids 247-263 and 260 -276, respectively). Amino acids 234 -Amino acids 257-276 are located in the D-domain/hinge region and analysis of the suggested nuclear localization signals in the hinge region, revealed two putative bipartite nuclear localization sequences (NLS-K-(K/R)-X-(K/R)). The putative nuclear localization sequences are 100% conserved between mouse and human ER␣ (Fig. 1A), and in addition, this area is not conserved between ER␣ and ER␤ as seen with alignment analysis. To further characterize the hinge region and to precisely define the sequences involved in ER␣ nuclear localization and ER␣ tethered-mediated responses, site-directed mutagenesis was performed in this area of the hinge region of ER␣ at the sites shown in Fig. 1B.
Confocal microscopy was used to demonstrate ER␣ localization by visualizing enhanced GFP-tagged ER␣ and mutant ER␣ proteins in transiently transfected HeLa cells (Fig. 1C). WT ER␣ is predominately nuclear in the absence or presence of ligand, E 2 . Examination of the H1 mutant, where part of the bipartite NLS is mutated to alanine, demonstrated a disruption of nuclear localization in the absence of ligand; however, in the presence of E 2 , this mutant translocated to the nucleus. The H2 ER␣ mutant, having the full bipartite NLS mutated to alanine, is non-nuclear in the absence of ligand, but in the presence of E 2 , translocation to the nucleus is observed. To force exclusion of the ER␣ from the nucleus the NES ER␣ mutant was engineered to have a nuclear export signal (NES-LXXXLXXLXL (38)) in the same area of the hinge region as the putative NLS. The NES ER␣ mutant translocates to the nucleus in the absence and presence of ligand as is seen with WT ER␣ demonstrating the strength of the ER␣ nuclear localization signal(s) and the nucleophilic nature of ER␣. We then created the H2ϩNES ER␣ mutant which has the full bipartite NLS mutated to alanine with the addition of a nuclear export signal within the putative NLS region. The H2ϩNES ER␣ was not detected in the nucleus under normal conditions (Fig. 1C). Repeating the ER␣ localization experiment in the presence of leptomycin B (LMB), a nuclear export inhibitor, the H2ϩNES ER␣ was detected in the nucleus (supplemental Fig. S1). This suggests that the H2ϩNES ER␣, presumably due to the NES, is rapidly shuttled or pumped out of the nucleus. Additionally, mutating the suggested nuclear localization sequence in the C-domain/DNA binding domain does not disrupt nuclear localization (data not shown). A Western blot for ER␣ protein and all the mutants is shown in Fig. 1D. The H1, H2, H2ϩNES, and NES ER␣ mutant constructs were utilized in this study to assess receptor functionality.
H1 ER␣ Mutant Loses Tethered-mediated Responses but Maintains ERE-mediated Activation-To analyze nuclear responses, the functionality of the H1 ER␣ mutant by promoter activation was examined. Luciferase assays were performed with reporters to observe classical (3ϫ ERE Luc) and tethered (AP-1 Luc)-mediated reporter activation of the ER␣ mutant. Reporter activation of the ERE Luc is seen with both WT ER␣ and the H1 ER␣ mutant when stimulated with E 2 and the response is blocked by the ER␣ antagonist, ICI-182,780 ( Fig.  2A). The AP-1 reporter, as previously published is activated by ICI-182,780 (7,10,39), was used to examine the tethering ability of the H1 ER␣ mutant. Activation of this reporter is abrogated in the H1 ER␣ mutant ( Fig. 2A). Ishikawa, uterine epithelial cells, were stably infected with empty vector, WT ER␣, or H1 ER␣ (supplemental Fig. S2). To validate the reporter assay data, progesterone receptor (PR) (33), a classical ERE-mediated response, and p21 (3), a tethered-mediated response, were examined. WT and the H1 ER␣ mutant activate via the classical ERE-mediated response; however, the H1 ER␣ mutant is no longer activated via the tethered-mediated response (Fig. 2B). In addition, the same loss in tethered activity is observed with the H1 mutant using the Sp1 reporter containing the Sp1 response element (data not shown). These results suggest that the H1 ER␣ mutant retains its ability to translocate to the nucleus and bind to DNA in the presence of E 2 , but has lost its ability to interact with the necessary tethering factors to activate via the AP-1 and Sp1 response elements.
H1 ER␣ Does Not Form a Protein-Protein Complex with c-Jun-ER␣ and the AP-1 family member c-Jun directly interact in the hinge region of ER␣ and this interaction is responsible for AP-1-mediated tethered responses (21). To investigate if the AP-1 tethered response could be rescued by co-expression of c-Jun, cells were co-transfected with the ER␣ plus or minus c-Jun expression plasmids. As evidenced by the AP-1 luciferase assay, enhancement of activity was only observed in cells cotransfected with WT ER␣ and c-Jun. Co-expression of c-Jun did not rescue this response with the H1 ER␣ mutant (Fig. 3A). H1 ER␣ was used to determine if the AP-1-mediated inactivity was due to a loss in protein-protein interaction of H1 ER␣ and c-Jun. Cells co-transfected with ER␣ and c-Jun exhibit proteinprotein interaction as demonstrated by both ER␣ and c-Jun immunoprecipitation (Fig. 3B). This interaction is stronger under E 2 hormone binding conditions; therefore, experiments with the H1 ER␣ mutant were done with E 2 treatment. Coimmunoprecipitation was performed on cells co-transfected with WT ER␣ ϩ c-Jun or H1 ER␣ ϩ c-Jun expression constructs. Immunoprecipitation for ER␣ and c-Jun did not coimmunoprecipitate c-Jun or ER␣, respectively, in cells co-transfected with the H1 ER␣ mutant (Fig. 3C). These data suggest that the loss of tethered-mediated activity is due to a loss in the ability of ER␣ to form the proper protein-protein complex involving this specific sequence in the D-domain of ER␣ necessary for mediating this particular mechanism of ER␣ action.
Co-activation of ER␣ by the steroid receptor co-activators (SRCs) are known to occur by protein-protein interactions in the E/F-domain of ER␣ (40). Therefore, to determine if the H1 ER␣ mutant was still able to interact with co-activators and the tethered-mediated response was specific to this sequence of the hinge region, ERE-and AP-1-mediated luciferase assays with SRC2 co-expression were performed. Co-expression of SRC-2 stimulates the ERE-mediated response of WT and the H1 ER␣ mutant. However, the H1 ER␣ mutants still remains inactive in the AP-1-mediated luciferase assay (Fig. 4, A and B). Similar results were obtained with SRC-1 and SRC-3 co-expression (data not shown). These data suggest that the H1 ER␣ mutant still retains the ability to interact with co-activators that are necessary for the ERE-mediated responses while lacking the tethered-mediated function.

Non-nuclear Cytoplasmic Signaling of H2ϩNES ER␣ Is Maintained while Nuclear Genomic Responses Are Lost-The
H2 ER␣ mutant, with the full bipartite NLS mutated to alanine, is cytoplasmic with vehicle conditions, but translocates to the nucleus with E 2 treatment. To prevent the nucleophilic ER␣ from maintaining nuclear localization, and keeping the focus on the D-domain, a nuclear export signal (NES; LXXXLXX-LXL) was added (Fig. 1, B and C). For the study of non-nuclear, rapid action only events, the ER␣ must not perform genomic-mediated responses (ERE and tethered). Promoter activation of the H2, NES and H2ϩNES ER␣ mutants were examined using the ERE Luc and AP-1 Luc reporters. In the ERE-mediated assay, the H2 and the NES ER␣ mutants responses were comparable to that of the WT ER␣ response (Fig. 5A), but the response is abrogates in the H2ϩNES ER␣ mutant. As expected, the H2 and the H2ϩNES ER␣ mutants no longer activated the AP-1-mediated reporter while the NES ER␣ mutant maintained activity (Fig. 5A). To further examine the ERE-mediated nuclear genomic function of these mutants, endogenous ERE-mediated target gene expression was validated in 293F cells transiently transfected with WT ER␣, H2 ER␣, NES ER␣ mutant, and H2ϩNES ER␣ mutant expression plasmids. Endogenous gene expression of PR and trefoil factor 1 (pS2) (ERE-mediated target genes (33)) were examined by real-time RT-PCR (Fig. 5B). PR and pS2 are induced by E 2 in the WT ER␣, H2 ER␣ mutant and the NES ER␣ mutant transfected cells; however, the H2ϩNES ER␣ mutant is unable to activate nuclear target gene expression. In addition, we confirmed the ERE-mediated target gene expression in stably infected Ishikawa cells (WT and H2ϩNES ER␣) by examining PR expression levels (supplemental Fig. S3A). The lack of ERE-and tethered-mediated responses in combination with the nuclear localization results with the H2ϩNES ER␣ mutant further sug-  gests that the loss in activity is consistent with the inability of the mutant receptor to maintain nuclear translocation.
To establish that a potential loss of genomic function was not due to a loss in DNA binding ability, an ERE transcription factor ELISA assay was performed. The H2ϩNES ER␣ binds DNA comparable to WT ER␣ suggesting that any loss of genomic function is not due to lack of DNA binding (supplemental Fig. S3B).
To verify the functionality of the H2ϩNES ER␣ mutant, the non-nuclear, non-genomic, rapid action response of this mutant was examined. ER␣ deficient, HeLa cells were transiently transfected with empty vector, WT ER␣, or the H2ϩNES ER␣ mutant constructs. Concomitantly, low expressing ER␣ Ishikawa cells, were stably transfected with the ER␣ constructs (supplemental Fig. S2). Both cells were starved for 60 to 70 h to decrease basal phospho-p44/42 MAPK as described under "Experimental Procedures." To stimulate the rapid action response, the transfected HeLa cells were treated for 0, 3, 5, and 10 min with E 2 (100 nM) and the stable Ishikawa cell lines were treated for 0, 5, and 10 min with E 2 . An increase in phospho-p44/42 MAPK was observed in WT ER␣ and H2ϩNES ER␣ mutant expressing cells after 3 and 5 min of E 2 stimulation (Fig. 6). These data show that the H2ϩNES ER␣ mutant, unable to maintain nuclear translocation, retains the properties for mediating nongenomic rapid action responses comparable to WT ER␣.

DISCUSSION
Traditionally, the D-domain ER␣ mutations used to demonstrate the ability to interact with proteins in a tethered-mediated fashion and NLS functionality have been with mutations of ER␣ containing complete deletions of the described D-domain or the suspected NLS (13, 14, 19 -22). Complete deletion of this domain would, presumably, result in dramatic alterations in receptor protein structure. To address the portion of the ER␣ sequence in the hinge region responsible for gene regulation via the tethered mechanism and nuclear localization, we chose to make specific site-directed changes to the ER␣ hinge region as opposed to domain deletions.
Computational analysis (LOCtree and Motif Scan) of ER␣ revealed putative nuclear localization signals in the C-and D-domains (AA 234 -240 and 270 -275; 247-263 and 260 -276, respectively). The hinge region is the least conserved region between mouse and human ER␣, but the nuclear localization sequences are 100% conserved from mouse to human as shown in Fig. 1A. These predicted nuclear localization signals correlated to those suggested by Ponglikitmongkol et al. (32) and Ylikomi et al. (14). The ER␣ deletions made by Ylikomi et al. disrupted nuclear localization, but did not completely exclude ER␣ from the nucleus. To define the precise sequence responsible for nuclear localization in ER␣, we mutated the R/K residues in the predicted area to A. When R/K residues from AA 234 to 256 in C-domain/exon 4 were mutated to A (8 R/K to A substitutions in the 2nd zinc finger of the C-domain), nuclear localization was not disrupted (data not shown). We then chose to focus on the basic amino acids in the D-domain/exon 5 from 260 to 281 that suggested a bipartite nuclear localization signal. When half (5 R/K to A substitutions-H1 ER␣ mutant) of the bipartite nuclear localization signal was mutated, ER␣ was observed to be more cytoplasmic under vehicle conditions than WT ER␣, but fully translocated to the nucleus when treated  with E 2 (Fig. 1C). Mutation of the full putative bipartite nuclear localization signal (9 R/K to A substitutions-H2 ER␣ mutant) was required for nuclear exclusion in the absence of ligand, but with E 2 treatment, partially translocated (CϭN) to the nucleus.
To counter the ligand-dependent nuclear localization of the H2 ER␣ mutant and to keep experimental focus on the functionality of the hinge region, a NES flanking the mutated putative nuclear localization signal sequence in the hinge region was incorporated. H2ϩNES ER␣ is blocked from maintaining nuclear localization even with hormone treatment, while NES ER␣ harboring only the addition of a nuclear export signal, is not excluded from the nucleus (Fig. 1C). The H2 ER␣ mutant must be counteracted by a nuclear export signal to keep the nucleocentric ER␣ out of the nucleus. The glucocorticoid receptor (GR) has a nuclear retention signal that overlaps the nuclear localization function and is necessary for ligand bound transcriptional activation (41). A similar nuclear retention signal or event for ER␣ is possible and the cytoplasmic versus nuclear distribution of the H1 and H2 ER␣ mutants suggest that this region of ER␣ may be responsible for ligand-dependent nuclear retention of ER␣. Further studies are necessary to confirm whether nuclear retention of ER␣ is through phosphorylation or protein-protein-mediated interactions as others have noted (42,43).
Functional analysis of H1 and H2 ER␣ revealed direct DNA binding activities were unaffected. However, the AP-1 mediated luciferase response was lost with H1 and H2 ER␣ (Figs. 2B and 5B). H1 ER␣, having less amino acid substitutions than H2 ER␣ and still maintaining the abrogation in AP-1-mediated activity, demonstrated that this loss in activity is likely due to loss of direct protein-protein interaction between H1 ER␣ and c-Jun (Fig. 3). The loss in AP-1-mediated activity due to a loss in protein-protein interactions (21) is not remarkable, but it is remarkable that the sequence mutated in ER␣ that abrogates this interaction (RMLKHKRQR 3 AMLAHAAQA) is a conserved nuclear localization consensus sequence. The H1 ER␣ Ishikawa stable cells demonstrated the loss in activity of an E 2 AP-1 responsive gene, p21. Great promoter redundancy and cell type specificity is seen with AP-1 and Sp1 target genes (44); therefore, conducting microarray analysis on the H1 ER␣ stable cell line to examine gene expression changes over a time course may uncover unknown E 2 ER␣ tetheredmediated gene responses.
The hinge region of nuclear receptors, initially thought to allow flexibility of the receptor, is demonstrating functionality in transactivation in other nuclear receptors as well. This region is regulated by posttranslational modifications such as methylation (45,46), acetylation (47), and sumoylation (48). The hinge region of androgen receptor (AR) and vitamin D receptor (VDR) regulates nuclear localization, DNA binding, and plays a role in coactivator recruitment and N/C-terminal interactions (17,49). The GR hinge region interacts with HEXIM1 and Bag-1 M which mediates glucocorticoid-mediated transcriptional repression of biological responses (17,50,51). The hinge region of ER␣ is post translationally modified; the sites important in human ER␣ for methylation and acetylation are Lys-302 and Lys-303 which correspond to Lys-306 and Lys-307 in mouse ER␣ (52,53). Sumoylation sites that are involved in the regulation of ER␣ transcriptional activity are human ER␣ sites Lys-299,302 and Arg-303 (mouse ER␣ K303, 306 and R307) with K266 and R268 (mouse ER␣ Lys-270 and Arg-272; located in our H1 ER␣ mutant) demonstrating no change in transcriptional activity (48). Kim et al. (54) show mutant K266Q/R268Q (mouse ER␣ K270/R272) when acetylated, increases ER␣'s DNA binding interaction with p300. In our experiments, we saw no change in DNA binding of the H1 ER␣ mutant, but this is likely due to the constitutive activity of the K266Q/R268Q mutant.
In contrast to H1 and H2 ER␣, H2ϩNES ER␣ demonstrates a loss in the ability to perform nuclear genomic responses involved in the classical and tethered mediated actions of ER␣ while maintaining the ability to bind DNA and perform nonnuclear, non-genomic, rapid action mediated phosphorylation of p44/42 MAPK (Fig. 5). This is in contrast to Zhang et al. (24) who do not see phospho-p44/42 MAPK with WT ER␣ and their NLS deletion mutant, HE241G, (deletion of amino acids 256 -303 of human ER␣, but do see p44/42 MAPK phosphorylation with their NLS myristoylation tagged mutant HE241G-mem. Chambliss et al. (22) demonstrate that with deletion of the NLS-hinge (amino acids 250 -274 of human ER␣) eNOS is no longer regulated via the non-genomic p44/42 MAPK response. The discrepancies presented here may be attributed to the deletion of 47 or 24 amino acids in these mutated ER␣ constructs, respectively, that disrupts proper protein folding compared with our point mutation approach. Many of the rapid action studies which do not use endogenous ER␣, but require transfection of ER␣ into the cells, focus on p44/42 MAPK activity. We find that in HeLa and Ishikawa ER␣ transfected cells, there is no change in activity of Akt (Ser-473), Src (Tyr-416 or Tyr-527), or GSK3␤ (Ser-9) with E 2 treatment. This finding highlights the sensitivity and the cell type specificity of the E 2 -mediated rapid action responses.
Localization and function of ER␣ at the plasma membrane requires Ser-522 of human ER␣ (26). A membrane-targeted ER␣ variant created by attaching myristoylation and palmitoylation sequences to the N and C terminus, respectively, of ER␣ missing the putative nuclear localization signals (human amino acids 256 -303) demonstrate membrane localization in the absence of estradiol and this mutant failed to regulate endogenous estradiol-responsive genes (31). A MOER (membraneonly estrogen receptor ␣) model has been generated, but in contrast to our H2ϩNES ER␣ mutant, the ER␣ receptor in this mouse contains only the E-domain with the addition of multiple palmitoylation sites to direct this form of the receptor to the cell membrane (55). Our H2ϩNES mutant, without the added myristoylation and palmitoylation tags, but retaining the natural sites for this activity, exhibits a similar phenotype by no longer retaining the ability to activate E 2 -mediated target genes, activate via ERE-or AP-1-driven reporter assays, is cytoplasmic without E 2 , and demonstrates rapid action responses via phospho-p44/42 MAPK. Additionally, H2ϩNES ER␣ is unique because the backbone spacing of ER␣ is retained and likely preserves ER␣ folding more adequately than deletion mutants.
The hinge mutants generated here are important tools to tease apart the three modes of action of ER␣ from one another. This functionality compliments published and ongoing studies in our laboratory, that utilize knock-in mouse models to examine the physiological relevance of the various functions of ER␣ (2). The models developed to date nicely attribute phenotypic changes to different functional regions of ER␣ and suggest that the mutations characterized here, will also demonstrate phenotypic changes important to biochemical pathways and estrogen biology.
In conclusion, the ER␣ mutants studied here are specific D-domain site directed mutants that demonstrate unique properties. We find as others that ER␣ harbors multiple nuclear localization sequences that renders this receptor in the absence of hormone to mainly be localized to the nucleus, and this is in contrast to ER␤, which contains only one potential putative NLS and is very often localized in the cytoplasm and translocates to the nucleus slowly over a 6 -12 h time period (56). Our experiments demonstrate the strength and utility of ER␣ nuclear localization signals and genomic mediated activities. Mutation of half of the bipartite NLS (H1 ER␣) only slightly disrupts nuclear localization but abrogates the tethered genomic response while maintaining ERE-genomic responses. Mutating the full bipartite NLS blocks nuclear translocation without ligand, but the addition of hormone translocates this receptor to the nucleus where it is able to activate ERE-mediated responses (H2 ER␣). The addition of a nuclear export signal in the putative NLS region of ER␣ does not disrupt localization or activity of the NES ER␣ mutant. A nuclear export signal must be added in addition to mutation of the nuclear localization sequence(s) to exclude ER␣ from the nucleus and render this form of the receptor unable to perform genomic mediated responses (H2ϩNES ER␣). This study assigns functionality to specific sequences in ER␣ D-domain involved in nuclear localization and tethered-mediated responses, and provides new avenues to fully dissect and use these ER␣ mutants to understand the varied functionality of estrogen and ER␣ in different tissues and to create therapeutic drug targets which could regulate a precise activity of ER␣ apart from the other modes of action.