Peptide Antagonists of the Human Estrogen Receptor

See allHide authors and affiliations

Science  30 Jul 1999:
Vol. 285, Issue 5428, pp. 744-746
DOI: 10.1126/science.285.5428.744


Estrogen receptor α transcriptional activity is regulated by distinct conformational states that are the result of ligand binding. Phage display was used to identify peptides that interact specifically with either estradiol- or tamoxifen-activated estrogen receptor α. When these peptides were coexpressed with estrogen receptor α in cells, they functioned as ligand-specific antagonists, indicating that estradiol-agonist and tamoxifen–partial agonist activities do not occur by the same mechanism. The ability to regulate estrogen receptor α transcriptional activity by targeting sites outside of the ligand-binding pocket has implications for the development of estrogen receptor α antagonists for the treatment of tamoxifen-refractory breast cancers.

About 50% of all breast cancers express the estrogen receptor α (ERα) protein and recognize estrogen as a mitogen (1). In a subpopulation of these tumors, antiestrogens, compounds that bind ER and block estrogen action, effectively inhibit cell growth. In this regard, the antiestrogen tamoxifen has been widely used to treat ER-positive breast cancers (2). Although antiestrogen therapy is initially successful, most tumors become refractory to the antiproliferative effects of tamoxifen within 2 to 5 years. The mechanism by which resistance occurs is controversial; however, it does not appear to result as a consequence of ER mutations or altered drug metabolism (3). It may relate instead to the observation that tamoxifen is a selective estrogen receptor modulator (SERM), functioning as an ER agonist in some cells and as an antagonist in others (4). Consequently, the ability of tumors to switch from recognizing tamoxifen as an antagonist to recognizing it as an agonist has emerged as the most likely cause of resistance. Upon binding ER, both estradiol and tamoxifen induce distinct conformational changes within the ligand-binding domain (5). The tamoxifen-induced conformational change may expose surfaces on the receptor that allow it to engage the general transcription machinery. We used phage display to identify specific peptides that interacted with the estradiol- and tamoxifen-ER complexes and used these peptides to show that estradiol and tamoxifen manifest agonist activity by different mechanisms.

Affinity selection of phage-displayed peptide libraries was performed to identify peptides that could interact specifically with the agonist [17β-estradiol (estradiol) or 4-OH tamoxifen (tamoxifen)], activated ERα, or ERβ (6). Representative peptides from each of four classes presented in this study are shown inFig. 1A. Several peptides that were isolated with estradiol-activated ERα (represented by α/β I) contained the Leu-X-X-Leu-Leu motif found in nuclear receptor coactivators (7). α II was isolated with either estradiol- or tamoxifen-activated ERα. Two classes of peptides, α/β III and α/β V, that interact specifically with tamoxifen-activated ERα and ERβ, respectively, were identified. The α/β V peptide was subsequently shown to interact with tamoxifen-activated ERα (6). Several additional peptides homologous to α/β V were identified. A BLAST search of the National Center for Biotechnology Information database with the derived consensus of the α/β V peptide class revealed that the yeast protein RSP5 and its human homolog, receptor potentiating factor (RPF1), both contain sequences homologous to α/β V. These proteins were previously shown to be coactivators of progesterone receptor B (PRB) transcriptional activity (8).

Figure 1

Isolation of ERα-interacting peptides. (A) ERα-interacting peptides were isolated by phage display (6). Eighteen libraries were screened, each containing a complexity of about 1.5 × 109 phage. Several Leu-X-X-Leu-Leu (boxed)–containing peptides were isolated, of which α/β I is shown. One peptide each was isolated for the α II and α/β III peptide classes. Six peptides were isolated, including α/β V, that contained a conserved motif (boxed). Two proteins, RSP5 and RPF1, containing sequence homology to α/β V are shown. Single-letter abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; X, any amino acid; and Y, Tyr. (B) TRF was used in competition mode to determine if ERα/tamoxifen-interacting peptides recognize a common site on ERα (9). The peptide conjugate used for detection is indicated in each graph with the competing peptides as follows: ▴, no competitor; ○, α II; •, α/β III; and ▪, α/β V.

Peptide-peptide competition studies were performed with time-resolved fluorescence (TRF) to determine if the α II, α/β III, and α/β V peptides were binding the same or distinct “pockets” on the tamoxifen-ERα complex (9). The α/β III and α/β V peptides cross compete, and at equimolar peptide concentrations, 50% inhibition is observed (Fig. 1B). This result indicates that these two peptides bind to the same or overlapping sites on ERα. We believe that the α II peptide binds to a unique site as its binding was not competed by α/β V and only 50% inhibited by a 10-fold excess of the α/β III peptide.

We next assessed whether the peptides interacted with ERα in vivo using the mammalian two-hybrid system (10). The α/β I peptide interacted with ERα in the presence of the agonist estradiol but not the SERMs tamoxifen, raloxifene, GW7604, idoxifene, and nafoxidine or the pure antagonist ICI 182,780 (Fig. 2). The failure of antiestrogen-activated ERα to interact with the α/β I peptide is consistent with previous studies that predict that the molecular mechanism of antagonism results from a structural change in the receptor ligand-binding domain that prevents coactivators from binding (5). α II interacted with the receptor in the presence of all modulators tested, with the unliganded (vehicle) and ICI 182,780–bound receptors showing the least binding activity. α/β III and α/β V interacted almost exclusively with the tamoxifen-bound ERα. ERα did not interact with the Gal4 DNA-binding domain (DBD) (control) alone in the presence of any modulators tested. Further studies indicated that binding of α II, α/β III, and α/β V occurs within the hormone-binding domain between amino acids 282 and 535 (11) and, unlike binding of α/β I, does not require a functional activation function 2 (AF-2) ( These data indicate that SERMs induce different conformational changes in ERα within the cell and firmly establish a relation between the structure of an ERα-ligand complex and function.

Figure 2

ERα-peptide interactions in mammalian cells. The coding sequence of a peptide representative from each class identified was fused to the DBD of the yeast transcription factor Gal4. HepG2 cells were transiently transfected with expression vectors for ERα-VP16 and the peptide-Gal4 fusion proteins. In addition, a luciferase reporter construct under the control of five copies of a Gal4 upstream enhancer element was also transfected along with a pCMV–β-galactosidase (β-Gal) vector to normalize for transfection efficiency. Transfection of the Gal4 DBD alone is included as control. Cells were then treated with various ligands (100 nM) as indicated and assayed for luciferase and β-Gal activity. Normalized response was obtained by dividing the luciferase activity by the β-Gal activity. Transfections were performed in triplicate, and error bars represent standard error of the mean (SEM). Triplicate transfections contained 1000 ng of ERα-VP16, 1000 ng of 5× Gal4-tata-Luc, 1000 ng of peptide-Gal4 fusion construct, and 100 ng of pCMV–β-Gal (10).

When we examined the specificity of interaction between the peptides and heterologous nuclear receptors, we found, as expected, that the α/β I peptide interacted with ERβ, PRB, and the glucocorticoid receptor (GR) when bound by the agonists estradiol, progesterone, and dexamethasone, respectively (Fig. 3, A, B, and C). The α/β V peptide interacted with tamoxifen-bound ERβ and unexpectedly with PRB in the presence of the antagonists RU 486 or ZK 98299 (Fig. 3, A and B). The α/β V peptide, however, did not interact with the GR when bound by RU 486 or ZK 98299. α II and α/β III peptides failed to interact with ERβ, PRB, or GR.

Figure 3

Specificity of nuclear receptor–peptide interactions. Two-hybrid experiments were performed as in Fig. 2between peptide-Gal4 fusion proteins and either (A) ERβ-VP16, (B) PRB-VP16, or (C) GR-VP16 (15). RU 486 and ZK 98299 are pan-antagonists of PRB and GR.

We next tested the ability of the peptide-Gal4 fusion proteins to inhibit ERα transcriptional activity. Tamoxifen displayed partial agonist activity when analyzed with the ER-responsive complement 3 (C3) promoter in HepG2 cells (Fig. 4A). This activity can reach 35% of that exhibited by estrogen and is mediated by three nonconsensus estrogen response elements (EREs) located in the C3 promoter (12). When expressed in this system, the α/β I and α II peptides inhibited the ability of estradiol to activate transcription up to 50% and 30%, respectively (Fig. 4B). Two copies of the Leu-X-X-Leu-Leu sequence found in α/β I enhanced the inhibitory effect of this peptide and blocked estradiol-mediated transcription by about 90% (13). The inability of α/β III and α/β V to block estradiol-mediated transcription correlates well with their inability to bind the receptor when bound by agonist. Expression of α II, α/β III, and α/β V peptides blocked the partial agonist activity of tamoxifen (Fig. 4C). α II and α/β V were the most efficient disrupters of tamoxifen-mediated transcription, inhibiting this activity by about 90%. All peptide-Gal4 fusion proteins were expressed at similar levels, indicating that the relative differences in inhibition are not due to peptide stability (11). We also demonstrated that receptor stability and DNA binding are not affected by peptide expression (11). As expected, α/β I was unable to inhibit tamoxifen-mediated transcription. These findings are in agreement with the binding characteristics of these peptides and suggest that the pocket or pockets recognized by α II, α/β III, and α/β V are required for tamoxifen partial agonist activity. Although α/β V was shown to interact with PRB when bound by RU 486 (Fig. 3B), it was unable to block the partial agonist activity mediated by PRB/RU 486 (11). This result suggests that ERα/tamoxifen and PRB/RU 486 partial agonist activities are manifested differently. However, because α/β V was selected against ERα, this peptide may not bind PRB with high enough affinity to permit it to be useful as a PRB peptide antagonist.

Figure 4

Disruption of ERα-mediated transcriptional activity. (A) HepG2 cells were transfected with the estrogen-responsive C3-Luc reporter gene (12) along with expression vectors for ERα (16) and β-Gal and normalized as in Fig. 2. Cells were induced with either estradiol or tamoxifen as indicated and analyzed for luciferase and β-Gal activity. NH, no hormone. (B) HepG2 cells were transfected as in (A) except that expression vectors for peptide-Gal4 fusions were included as indicated. Control represents the transcriptional activity of estradiol (10 nM)–activated ERα in the presence of the Gal-4 DBD alone and is set at 100% activity. Increasing amounts of input plasmid for each Gal4-peptide fusion are also shown (▴) with the resulting transcriptional activity presented as percentage of activation of control. Data are averaged from three independent experiments (each performed in triplicate) with error bars representing SEM. Triplicate transfections contained 1000 ng of C3-Luc, 1000 ng of ERα expression vector, 100 ng of pCMV–β-Gal, and either 100, 500, or 1000 ng of peptide-Gal4 fusion construct. (C) Same as in (B) except that 4-OH tamoxifen (10 nM) was used to activate the receptor. (D) HepG2 cells were transfected with the AP-1–responsive collagenase reporter gene construct (pCOL-Luc) (12) and expression vectors for ERα and β-Gal. Cells were then induced with either estradiol or tamoxifen as indicated. (E) Same as (D), except that peptide-Gal4 fusion constructs were also transfected as indicated. Control represents the transcriptional activity of either estradiol- or tamoxifen (100 nM)–activated ER in the presence of the Gal4 DBD alone and is set at 100% activity. The transcriptional activity of estradiol and tamoxifen is shown in the presence of each Gal4-peptide fusion with the resulting transcriptional activity presented as percentage of activation of control. Triplicate transfections contained 1000 ng of pCOL-Luc, 1000 ng of ERα expression vector, 1000 ng of peptide-Gal4 fusion construct, and 100 ng of pCMV–β-Gal. Data are presented as in (B) and (C). (F) HeLa cells were transfected with the 1X-ERE-tata-Luc reporter gene along with expression vectors for ERα, β-Gal, and either RPF1 (pCDNA3-RPF1) or control vector [pcDNA3 (Invitrogen, Carlsbad, CA)]. Cells were induced with ligand (10 nM) as indicated. Data are presented as fold induction, which represents the ratio of ligand induced versus vehicle for each transfection.

Finally, we examined the ability of these peptides to inhibit ER transcriptional activity mediated through AP-1–responsive genes. This pathway has been proposed to account for some of the cell-specific agonist activity of tamoxifen (14). Both estradiol and tamoxifen activated transcription from the AP-1–responsive collagenase reporter gene, pCOL-Luc (Fig. 4D). This activity is manifest in the absence of an ERE and is believed to occur through a mechanism involving an interaction between ERα and the promoter-bound AP-1 complex (14). Regardless of the mechanism, each peptide was able to inhibit ERα-mediated transcriptional activity in a manner that reflected its ability to interact with the receptor in a ligand-dependent manner (Fig. 4E).

The mechanism by which tamoxifen manifests SERM activity is not yet known. Evidence presented in this study suggests that the tamoxifen-bound receptor exposes a binding site that is occupied by a coactivating protein not primarily used by the estradiol-activated receptor. The α II peptide, which interacts with both estradiol- and tamoxifen-bound receptors, inhibits the partial agonist activity of tamoxifen efficiently, while minimally affecting estradiol-mediated transcription. This result suggests that this site, although crucial for tamoxifen-mediated transcription, is dispensable for estrogen action. In addition, the ability of α/β III and α/β V to bind tamoxifen-specific surfaces and inhibit tamoxifen-mediated partial agonist activity suggests that these peptides may potentially recognize a protein contact site on ER that is critical for this activity. In this regard, we can demonstrate that, similar to α/β V, overexpression of RPF1 specifically represses tamoxifen-mediated partial agonist activity (Fig. 4F). However, the physiological importance of this activity remains to be determined. In summary, we have identified a series of peptide antagonists of ERα and hence validated additional target sites other than the ligand-binding pocket for drug discovery.

  • * To whom correspondence should be addressed. E-mail: mcdon016{at}


Stay Connected to Science

Navigate This Article