Fusion proteins of ER-LBD monomers with split-luciferase fragments can be used to measure ligand-induced ER dimerization and also to distinguish the effects of ligands of different biocharacter on ER homo- or heterodimer stability

In our previous work (17), we designed single receptor monomers to which we attached both the N- and C-terminal portions of the split-luciferase enzymes. In those experiments, the split-luciferase complementation signal output was a measure of regional conformational changes near the C-terminus of the LBD that acts as a molecular switch in response to the binding of ligands of differing biocharacter. In this prior intramonomer split-luciferase complementation configuration, there might also have been a small contribution from complementation occurring between monomers within a dimer, as well as from changes in global receptor structural stability, especially when the split proteins were placed near the far C-terminus (F-domain) of the receptor (17).

By contrast, to study receptor dimerization in this report, we designed each monomer to contain only a single portion of the split-luciferase protein but to be used as pairs, each member of which would contribute either the N- or the C-terminal fragment of the split luciferase. The split-luciferase complementation signal in these experiments has three potential sources: 1) Dimerization leading to complementation; 2) global receptor structural stability allowing for more efficient complementation; and 3) regional molecular switch conformational changes near the C-terminal end of the LBD.

To find the optimum pairs, we transiently cotransfected HEK293T cells with all of the possible combinations of either N- or C-terminal split-luciferase ERα and ERβ monomers and assayed them for complementation signal after exposure to ligands E2, genistein (Gen), 4-hydroxytamoxifen (OHT) and raloxifene (Ral), as well as dimethylsulfoxide as solvent control. Ligands were used at 1 μm to ensure that both ER would be fully saturated, despite differences in ligand binding affinities. Significant levels of complemented luciferase signals through heterodimerization (P < 0.01 for all ligands in comparison with solvent control) were seen with all combinations. We selected for further study the combination coexpressing NFluc-ERβ-LBD and ERα-LBD-CFluc; although all eight combinations gave robust signals with some of the ligands, the selected pair gave both high induction signal level through heterodimerization with various ligands, while at the same time maintaining a trend of signal patterns that reflected the varying biocharacter of different ligands (E2: agonist; Gen: ERβ-specific partial agonist/antagonist; OHT and Ral: SERM) (Supplemental Fig. 1 published on The Endocrine Society's Journals Online web site at http://mend.endojournals.org).

The chosen combination of vectors was used for making stable cell lines and in all other cell culture studies, and for subsequent imaging in animals. Of note, the lack of change in the signal trend across ligands of various biocharacter with heterodimer split-luciferase complementation using these monomer constructs is a good outcome, because it indicates that the regional molecular switch conformational changes (mechanism c) plays a minimal role during the production of this cross-dimer complementation signal. Thus, the major contributors to the complementation bioluminescence signal in the current experiments appear to be receptor dimerization, with perhaps some contribution from global receptor structural stability.

We also used our previously reported ERα single-mutation receptor G521T. This mutant has essentially no affinity for E2, but shows minimal loss of affinity for OHT and diethylstilbestrol (DES), and has increased affinity for the pyrazole-based ligands propyl pyrazole triol (PPT) and methyl piperidino pyrazole (MPP) (18). It proved useful in our studies on half-occupied ER dimer stability.

Western blot analysis of HEK293T cells engineered to express ERα/ERβ heterodimerization system identifies no substantial change in the expression levels of different monomers before and after exposure to different ER ligands

HEK293T cells engineered to stably express fusion proteins NFluc-ERβ-LBD and ERα-LBD-CFluc were tested for the expression level of fusion monomers by Western blot analysis, after exposure to different ER ligands [E2, OHT, Ral, MPP, PPT, and diarylpropionitrile (DPN)] by using anti-Fluc antibody. We found no significant change in the expressed levels of monomers in response to the treatment of different ER ligands (Supplemental Fig. 2).

SERM-bound receptors form the strongest homo- and heterodimers

The SERM-bound ERα and ERβ receptors lead to the highest split-luciferase complementation signal within homo- and heterodimers (Fig. 2, A–C). Within the ERα homodimers, we note the highest dimer stability signal from OHT, Ral, MPP-bound receptors (Fig. 2A), and within the ERβ homodimers, the highest dimer stability signal came from OHT and Ral-bound receptors (Fig. 2B). Finally, within the ERα/ERβ heterodimers, we note the highest dimer stability signal from OHT, Ral, and MPP (Fig. 2C). Our previous work using small molecule fluorophore-labeled receptors also demonstrated similar pronounced SERM-bound ERα homodimer stability (19).

Open in a separate window

Fig. 2.

ER ligand-induced dimerizations [ERα/ERα-homodimerization (panel A); ERβ/ERβ-homodimerization (panel B); and ERα/ERβ-heterodimerization (panel C) studied in HEK293T cells cotransfected with respective combination of vectors (ERα/ERα: NFluc-ERα-LBD/ERα-LBD-CFluc; ERα/ERβ: NFluc-ERα-LBD/ERβ-LBD-CFluc; ERβ/ERβ: NFluc-ERβ-LBD/ERβ-LBD-CFluc) after exposure to several ligands including ERα-selective (MPP and PPT), ERβ-selective (DPN, THC, and Gen), and ER subtype nonselective ligands [ERα and ERβ; estrone (E1), E2, estriol (E3), OHT, and Ral]. The results showed significant level of luciferase signal induced by all ligands (P < 0.05). The subtype-specific ligands induced higher level of signals with their respective homodimerization along with the heterodimerization. DMSO, Dimethylsulfoxide.

Split luciferase complementation assay reveals ERα and ERβ homodimerization induced by ER subtype-selective ligands

We next studied the ligand-mediated regulation of ER homodimerization (ERα/ERα and ERβ/ERβ) using HEK293T cells stably expressing the appropriate sets of NFluc-ER-LBD and ER-LBD-CFluc subtypes. The complemented luciferase signal was measured after 24-hr exposure to ligands that are either subtype-selective for ERα [propyl pyrazole triol (PPT), 410-fold selective (20)], methyl piperidino pyrazole [MPP, 220-fold selective (21)], or subtype-selective for ERβ [diarylpropionitrile (DPN), 70-fold selective (22)], Gen (~400-fold selective) (23)], tetrahydrochrysene [THC, 16-fold selective (24)]. We also used a series of subtype-nonselective ligands as controls.

Subtype-nonselective ligands (estrone, E2, estriol, and DES, OHT, and Ral) induced significant (P < 0.02) levels of complemented luciferase signal in both ERα/ERα and ERβ/ERβ homodimers, demonstrating that they could effect homodimerization of both pure subtypes (Fig. 2, A and B). By contrast, subtype-selective ligands were more effective in inducing homodimerization interactions of their cognate ER subtype, to which they bound preferentially. For example, the highly ERα-selective ligands PPT and MPP were more effective in homodimerizing ERα than ERβ, whereas the ERβ-selective ligands, Gen, DPN, and THC, more effectively dimerized ERβ than ERα as homodimers.

A single ligand-occupied monomer can form stable homo- and heterodimers with an apo-monomer

The biology of an ER homo- and heterodimer in which only one monomer component is occupied by a ligand is not well understood, despite the fact that in biological settings with low ligand concentrations, such half-ligand-occupied ER dimers are undoubtedly present and are likely to make important contributions to the observed patterns of ligand-regulated gene expression. To evaluate these biologically significant, yet poorly understood, molecular interactions, we used our split-luciferase complementation assay together with subtype-selective ligands. As shown in Fig. 2C, we found that a receptor monomer bound with a subtype-selective ligand (ERα with PPT or MPP, or ERβ with Gen or DPN) formed stable heterodimers with an apo-ERβ or apo-ERα monomer, respectively.

We also investigated formation of singly liganded ER heterodimers using the ERα_G521T mutant that we previously identified as having no affinity for E2 but with retained affinity for all other nonsteroidal ligands, including DES, PPT, and OHT (18). In transfected cells, one or both of the two ERα fusion proteins in the ERα homodimerization system, and the ERα protein in the ERα/ERβ heterodimerization system, were replaced by the fusion protein containing the mutant form of ERα-LBD (ER_G521T), and then assayed for complemented luciferase activity after exposure to different ligands, including E2 (Fig. 3, A and B).

Open in a separate window

Fig. 3.

A, ERα/ΕRα homodimerization studied in HEK293T cells cotransfected with vectors expressing fusion proteins of split-luciferase fragments with LBD derived from the wild-type ERα (ERα_wt), and E2-nonresponsive mutant-ERα (ERα_G521T). The cells cotransfected to express both wild-type ER-LBD (NFluc-ERα-LBD_wt and ERα-LBD_wt-CFluc), wild type with a mutant ER-LBD (NFluc-ERα-LBD_wt and ERα-LBD_G521T-CFluc), or both mutant ER-LBD (NFluc-ERα-LBD_G521T and ERα-LBD_G521T-CFluc), were assayed for luciferase activity after exposure to E2, OHT, and PPT, along with solvent dimethylsulfoxide as control. The results showed significant (*, P < 0.01; #, P < 0.05) level of inductions from all ligands in all three combinations, except cells that were cotransfected to express fusion proteins with both mutant ERα-LBD, which showed significantly lower level of complementation for ligand E2 (1 and 2 in comparison to 3; P < 0.01). B, ERα/ERβ-heterodimerization was studied with either wild-type ERα-LBD (NFluc-ERα-LBD_wt + ERβ-LBD-CFluc) or mutant ERα-LBD (NFluc-ERα-LBD_G521T and ERβ-LBD-CFluc). The cells cotransfected with the mutant ERα-LBD showed considerably lower luciferase complementation signal than those with wild-type ERα-LBD, especially when exposed to ligand E2. Con, Control.

In the ERα/ERα homodimerization system, we observed high or equivalent dimerizations induced by OHT and PPT when none, one, or both of the ERα-LBD fusion protein partners were replaced with mutant-ERα-LBD (ER_G521T), consistent with the good affinity that these ligands have for both wild-type and G521T mutant ER. By contrast, with E2, good dimerization was observed in the wild-type (ERα_wt/ERα_wt) combination, and the wild type with mutant (ERα_wt/ERα_G521T), but when both ERα partners were mutant (ERα_G521T/ERα_G521T), no dimerization was evident, because neither of the mutant monomers can bind E2 (Fig. 3A). This suggests that ERα homodimers can form even if only one of the monomer partners is occupied by E2, but, as expected, dimers do not form with two apo-ER monomers.

There is a general loss of signal across all ligand-bound ERβ/ERα_G521T heterodimers (Fig. 3B). Because this occurs regardless of whether the monomer components are singly or doubly occupied, it is likely due to a somewhat decreased dimer stability and/or decreased global structural rigidity of the ERα_G521T mutant receptor that leads to less efficient complementation. In this regard, it is notable that when ERα/ERβ heterodimers are singly occupied with subtype-selective ligands (Fig. 2C), there is no evident loss of signal intensity compared with the signals from the corresponding doubly occupied homodimers (e.g., PPT and MPP in Fig. 2, panel C vs. panel A, and DPN in Fig. 2, panel C vs. panel B).

Simultaneous monitoring of ERα/ERα homo- and ERα/ERβ heterodimerization studied using a split-Renilla and split-firefly luciferase dual reporter complementation system

The split-protein complementation technologies allow for simultaneous evaluation of multiple and competing protein-protein interactions within a mixed heterogeneous cellular milieu. Potentially, this technology can be used in a multi-modality imaging approach to monitor specific protein interactions via bioluminescence or positron emission tomography signals in living animals (25).

We used a combination of split-Renilla (Rluc) and split-firefly (Fluc) luciferase-based bioluminescence complementation systems as a proof-of-concept experiment to investigate how ligands regulate the competition between ERα/ERα homodimerization vs. ERα/ERβ heterodimerization. In this more involved protocol, HEK293T cells were cotransfected to express three fusion proteins: NFluc-ERα-LBD-ChRluc, ERβ-LBD-CFluc, and NhRluc-ERα-LBD; dimerization of the first and second proteins would enable complementation of firefly luciferase and produce a signal unique for ERα/ERβ heterodimerization, whereas dimerization of the first and third proteins would complement Renilla luciferase and produce a signal unique for ERα/ERα homodimerization.

The results identified significant (P < 0.01) levels of complemented activities from both luciferases after exposure to different ER ligands, including subtype-selective ones (Fig. 4), indicating that both ERα/ERα homodimers and ERα/ERβ heterodimers were forming (Fig. 4A). Nevertheless, despite the quantitative differences in the magnitude of the Fluc and hRluc signals, these can be readily distinguished because they are measured independently after the addition of different bioluminescence substrates. The Western blot analysis for the expressed level of all three-fusion proteins showed no substantial change in response to the treatment of different ligands, further confirming the specificity of our system (Fig. 4B).

Open in a separate window

Fig. 4.

Simultaneous monitoring of ligand-induced ERα/ERα homo- and ERα/ERβ heterodimerization by dual reporter complementation system. A, The cells cotransfected to express fusion proteins NFluc-ERα-LBD-ChRluc, ERβ-LBD-Cfluc, and NhRluc-ERα-LBD were assayed for ERα/ERα homodimerization (Renilla luciferase) and ERα/ERβ heterodimerization (firefly luciferase) 24 h after exposure to different ER ligands including subtype-selective ligands. The results showed significant levels of αα-homodimerization (P < 0.05) and αβ-heterodimerization (P < 0.01), as measures of complemented luciferases activities. B, Western blot analysis of cell lysates from panel A assayed for the expressed fusion proteins using ERα- and ERβ-specific antibodies. DMSO, Dimethylsulfoxide.

These are early results from a technically challenging but uniquely informative approach for monitoring two competing and simultaneous protein-protein interactions of the ER. The higher signal seen with the heterodimer complexes compared with homodimers may be related to different levels of activation upon Fluc vs. hRluc complementation or to conformational changes within these fusion constructs, and not necessarily to differences in receptor dimer stability. Nevertheless, all of the changes in bioluminescence signals are statistically significant; they again show a dependence on ligand structure that is similar to that seen in the two-component complementation systems (Fig. 2, A and C). Further investigations with specific controls are needed to more fully characterize this NFluc-ERα-LBD-ChRluc double-split fusion ER construct and its influence upon the native receptor homo- and heterodimerization.

Split luciferase complementation provides a uniquely powerful approach to study specific ER dimerization events in cells and in vivo

To begin to unravel the role of the ER subtypes and their homo- and heterodimers as mediators of estrogen biology, we adapted a split-luciferase reporter protein complementation system that could be used to monitor various ligand-regulated attributes of the ER (15, 17, 27). Our previous ER constructs were carefully engineered to contain the two complementary portions of the luciferase protein fused to the N- and C-termini of a single receptor monomer (17). That earlier work focused on ligand regulation of ER regional conformations with respect to the well-evaluated molecular switch at the C-terminal portion of the LBD. Here, we have focused on how ligands regulate interactions of ER proteins, namely the formation of ERα and ERβ homo- and heterodimers. This required that the N- and C-terminal complementary fragments of luciferase be attached separately to different ER monomers, either to two ERα monomers to study ERα homodimerization, two ERβ monomers to study ERβ homodimerization, or one ERα monomer and one ERβ monomer to study ERα/ERβ heterodimerization (18). It is notable that the two complementing luciferase fragments provide unique tags of individual ER subtypes (or of an ERα mutant), and these unique tags then enable the formation of specific dimers to be monitored uniquely through complementation, even in the presence of other untagged ER. In principle, such luciferase complementation can be extended in a dual fashion, using three firefly and Renilla luciferase-complementing fragments, and in this regard we provide results from preliminary experiments that establish the feasibility of this experiment.

We found that the LBD of ERα and ERβ were sufficient to maintain the ligand-mediated homo- and heterodimerization properties of these receptors, and the luciferase complementation assays were more sensitive and more convenient when the ER LBD rather than the full-length ER were used, consistent with the more structured nature of the LBD portion of the ER. We also optimized both the positioning and orientation of the N- and C-terminal luciferase fragments on ERα and ERβ. Importantly, the signal output from the split-luciferase complementation appears to be reflective of dimer stability and not the regional conformational changes within the C-terminal portion of the LBD.

Ligands of different pharmacological activity and ER subtype selectivity have different effects on ER dimer strength and ER subtype dimer selectivity

We found that ligand regulation of ER homo- and heterodimer formation, as sensed by luciferase complementation, demonstrates that SERM-bound ER dimers give the highest signal output, suggesting the formation of the strongest dimer complexes. Earlier, in a purely in vitro system, we also found that SERM binding afforded the greatest stability of ERα/ERα homodimers (19). This increased dynamic stability of SERM-bound ER homo- and heterodimers may reflect mechanistic aspects of partial antagonist-bound receptor biology and may be related as well to the well-known slower cellular turnover of SERM-bound ERα (28).

In this regard, it is of note that the THC compound, which has a mixed ERα and ERβ biocharacter (ERα agonist/ERβ antagonist), demonstrates no significant dimer stabilization within either ERα/ERα or ERβ/ERβ homodimers. Again, we reported this unique lack of dimer stabilization with THC within ERα/ERα homodimers in our prior in vitro study (19). This lack of dimer stabilization may reflect the unique biological mechanism and mixed pharmacological profile of THC, as is seen in cell-based assays and in crystallographic studies (24, 29). Similarly, the homodimer stability of ER subtypes reflect the binding preferences of the ER subtype-selective ligands [ERα-selective PPT (20) and MPP (21), and ERβ-selective Gen (23), DPN (22), and THC (24)].

ER dimers only partially occupied by ligands can still dimerize. We studied whether ER dimers form when only one monomer component is occupied with ligand by using subtype-selective ligands or a mutant ERα (G521T) that does not bind E2 but retains good affinity for several other ERα ligands (e.g. DES, OHT, and PPT). Previously, this ERα mutant was used to show that half-occupied ERα/ERα homodimers could also form when one monomer was unoccupied (18). By both approaches we found that ERα and ERβ homodimers and ERα/ERβ heterodimers form between a single ligand-bound monomer and an apo-monomer.

This cross talk across the dimer interface represents the influence of a single ligand on the stability of a dimer and is suggestive of a novel aspect of ER action that is difficult to study by other means. For example, when ligand concentrations are insufficient to saturate both monomers in a dimer, dimers could still form; it is interesting to consider what sort of activity these half-occupied dimers might have. Also, if a cell that contains both ERα and ERβ is treated with a subtype-selective ligand, capable of binding to only one ER subtype, ERα/ERβ heterodimers in which only one monomer is occupied could form. The formation of ligand-occupied ER/apo-ER dimers is reminiscent of the behavior of type II nuclear hormone receptors, such as thyroid hormone receptor, peroxisomal proliferator-activated receptor, retinoic acid receptor, and vitamin D receptor, that form heterodimers with unoccupied retinoid X receptor (30).

Simultaneous monitoring of ERα/ERα homo- and ERα/ERβ heterodimerization by dual-split reporters

We noted that, in principle, it should be possible to simultaneously monitor two complementation events by suitable deployment of the split fragments of Renilla-luciferase and firefly luciferase in a three-component system. As a proof-of-principle, in this study we have evaluated the technical feasibility of using such a dual reporter system. We used ERα and ERβ expressing different combinations of split Renilla and firefly luciferase fragments, and we successfully demonstrated ligand-induced homo- and heterodimerization in response to subtype-selective and nonselective ligands. Simultaneous measurement of two events, in a same cell, is useful in extracting significantly valuable information executed by the interacting partners at micro levels.

Bioluminescence imaging of ER dimer formation in vivo

Optical methods have previously been used to study ER homo- and heterodimerization in cell-free in vitro systems and in intact cells (24, 29); however, none of these approaches could potentially be used to study the same process in living animals (31). In earlier work, we used fluorescence resonance energy transfer and excimer-emission methods to study the effect of ligand binding on the kinetic and thermodynamic stability of ERα homodimers in vitro and the additional stabilizing effect due to coactivator binding to ER-agonist complexes (19, 32, 33). Studies in cells using bioluminescence resonance energy transfer methods have also revealed that ERα plays a dominant role in ERα/ERβ heterodimerization (34–36). Similarly, we have found a similar dominance of ERα in the selection of ERs for binding to chromatin (8).

We used bioluminescence imaging to monitor formation of ERα/ERβ heterodimers in vivo: a strong emission signal from a xenograft growing in an athymic mouse could be followed with time, indicating the progressive formation of this heterodimer. It is of note that the specific complementing luciferase fragments used to tag ERα and ERβ in this experiment enabled formation of ERα/ERβ heterodimers to be monitored specifically. Tagging the ER subtypes in different manners could enable formation of ER homodimers and potentially half-occupied ER dimers to be monitored specifically, as well as the effects of ER ligands of different structure, SERM, and subtype-selective ligands.

Construction of plasmids

The NH₂-terminal portion (NFluc, amino acids 1–398) of the Fluc gene was PCR amplified using the forward primer designed with NheI and the start codon, and the reverse primer designed with BamHI with and without a peptide linker sequence (GGGGSGGGGS) using pCMV-Fluc plasmid (Promega) as a template. Similarly, the COOH-terminal portion (CFluc, amino acids 394–550) of the Fluc gene was amplified using the forward primer designed with BamHI and the reverse primer designed with XhoI and a stop codon. The restriction enzyme-digested fragments were cloned into a corresponding enzyme-digested vector backbone of pcDNA3.1(+) (Invitrogen), modified to express the puromycin-resistant gene as a eukaryotic selection marker. The PCR product of full-length [ERα (1–595); ERβ (1–530)] or the LBD [ERα (281–549); ERβ (250–505)] of human ERα and ERβ both upstream (NheI-BamHI sites) and downstream (BamHI-XhoI) of N- and C-terminal firefly luciferase fragments were cloned into the corresponding restriction-digested vectors. The nucleotide sequence coding for linker peptide (GGGGS)₃ was introduced in between the ER and the luciferase fragments in all the vectors. Site-directed mutagenesis was used for constructing different vectors containing the G521T mutant ERα. A similar strategy was used for constructing different vectors containing N- and C-terminal Renilla luciferase (hRluc) fragments (NhRluc: amino acids 1–229 and ChRluc: amino acids 230–310). The sequence-confirmed vectors were used for different experiments. See Fig. 1C for the complete list of plasmid vectors constructed and used in this study.

Cell culture

Human 293T embryonic kidney cancer cells (HEK293T) were grown in MEM supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin solutions.

Cell transfection, firefly, and Renilla luciferase assays

Transfections were performed in 80% confluent 24 h-old cultures of HEK293T cells. For transfection, 250 ng/well of pcDNA molar equivalent of each split hRluc/Fluc reporter, and full-length hRluc/Fluc were used in 12-well tissue culture plates, and DNA and Lipofectamine-2000 in 1:3 ratio were used for different transfection experiments, following the manufacturer's instructions. A vector (1 ng) expressing full-length Renilla luciferase (phRluc-CMV; Promega) was cotransfected per well to normalize for transfection efficiency. The cells were assayed for luciferase activity after a 24-h incubation, either with or without ligands, by keeping the culture at 37 C in 5% CO₂. The luminometer assays for Fluc and hRluc activity were performed as previously described (37, 38). In brief, transfected cells were lysed in 200 μl of ice-cold 1× passive lysis buffer by shaking for 10 min. The cell lysates were centrifuged for 5 min at 10,000 rpm at 4 C to remove cell debris. To determine hRluc activity, 20 μl of supernatant were assayed by addition of 1.0 μg of coelenterazine in 100 μl of 0.05 m sodium PBS at pH 7.0, followed by photon counting in the luminometer (model T 20/20, Turner Designs, Sunnyvale, CA) for 10 sec. The Fluc activity was determined as described for hRluc activity, except for the use of 100 μl of LARII substrate from Promega instead of coelenterazine. Protein concentrations from the cell lysates were determined by using the Bradford assay reagent (Bio-Rad Laboratories, Inc., Hercules, CA). Rluc activities were normalized for protein content and for transfection efficiency using Fluc activity, and they were expressed as relative light units/μg protein · min of counting.

To create stable HEK293T cells coexpressing the ERα/β heterodimerization system, the vectors expressing NFluc-ERα-LBD and ERβ-LBD-CFluc were cotransfected and selected by using puromycin. The cells were plated in 96-well plates, and single clones of cells coexpressing both the fusion proteins at approximately equal levels were expanded and used for different in vitro and in vivo experiments.

Western blot analysis

To determine the expression level of different fusion proteins containing the split-Fluc fragments with ERα or ERβ, cell lysates in triplicate wells assayed for luciferase activity by luminometry were then combined, and protein concentrations were determined using the Bradford assay reagent. A 30-μg portion of total protein in 1×Lamelli loading buffer with β-mercaptoethanol (Bio-Rad) was heated at 95 C for 5 min, resolved using a 4–12% SDS-PAGE gradient gel (Invitrogen), and electroblotted onto a nitrocellulose membrane (Schleicher & Schuell, Keene, NH) of 0.2-μm pore size. The membrane was blocked with 5% nonfat dry milk in Tris-buffered saline containing 0.01% Tween 20 (pH 7.6) for 1 h and probed overnight at 4 C on a rotating platform with the appropriate antibodies as follows: goat polyclonal antibody against firefly luciferase (1:1000 dilution, Abcam, Cambridge, MA); secondary antibodies were peroxidase-conjugated donkey antigoat IgG (1:3000 dilution for both, Abcam). Mouse monoclonal antibody against Renilla luciferase (1:1000 dilution, Chemicon, Temecula, CA); secondary antibodies were peroxidase conjugated donkey antimouse IgG (1:5000 dilution for both; Rockland Immunochemicals, Philadelphia, PA). All Western blot membranes were reprobed with a mouse monoclonal antibody against human α-tubulin (1:5000, clone B-5-1-2; Sigma) to control for protein loading. Immunoblots were developed using the LumiGlo enhanced chemiluminescence method (Cell Signaling Technology, Danvers, MA), following manufacturer's instructions using blue films (Midwest Scientific, St. Louis, MO).

Bioluminescence imaging in living mice

All the animal handling was performed in accordance with Stanford University Animal Research Committee guidelines. Mice were gas anesthetized using isoflurane (2% isoflurane in 100% oxygen, 1 liter/min) during all injection and imaging procedures and were imaged using a cooled charge-coupled device camera (Xenogen IVIS-29; Xenogen Corp., Alameda, CA). An aliquot of 5 × 10⁶ HEK293T control cells or HEK293T cells stably coexpressing ERα-LBD-CFluc and NFluc-ERβ-LBD were implanted sc in the bottom left and right flanks, respectively, of 7-wk-old female nude mice (n = 5) (nu/nu, Charles River Laboratories, Inc., Wilminton, MA). The animals were imaged on 1d (before injecting ligand) and daily for 5 d with the injection of 50 μg ligand MPP every day. Similarly, another group of animals (n = 5) was imaged every day, with the injection of ligand MPP for the first 2 d and then discontinuing any further ligand injection for an additional period of 8 d (to see drug withdrawal effect). d-Luciferin (3 mg in PBS) was injected ip, and animals were imaged with the IVIS imaging system with an acquisition time of 1 min each until the signal reached the peak. Images were obtained using Living Image Software (Xenogen Corp.) and analyzed using Igor Image Analysis Software (Wavemetrics, Seattle, WA). To quantify the measured light, regions of interest were drawn over the area of cell implants, and the maximum photons/sec/cm²/steradian (sr) was obtained, as validated previously.

We thank Dr. Benita S. Katzenellenbogen for helpful suggestions.

This work was supported by National Institutes of Health Grants: NC ICMIC P50 CA114747;, NCI SAIRP R24 CA92865;, and R01 CA82214; (to S.S.G.) and R37 DK015556; and P01 AG024387 (to J.A.K.). T.F.M. received partial salary support from the NIHR Cambridge Biomedical Research Centre. R.P. was funded by the department of Radiology, Stanford University.

Disclosure Summary: The authors have nothing to disclose.