Mechanisms of Antiestrogens

by Krista Bennett

Estrogen, a steroid hormone, exerts its effects in the body by binding to estrogen receptors. The estrogen receptor (ER) is a protein that regulates the differentiation and maintenance of neural, skeletal, cardiovascular, and reproductive tissues. Estrogen receptors are part of the nuclear hormone receptor family of ligand activated transcription factors(9). There are two subtypes of estrogen receptors ERa and the recently discovered ERb . Most ER activity is carried out by ERa (6).

Nuclear hormone receptors, like the estrogen receptor, consist of six domains, A, B, C, D, E, and F. The DNA binding domain (DBD) is located in the central C domain. The ER also contains two activation functions (AF-1 and AF-2) which are transcriptional activators. AF-1 is located in the amino terminal A/B domain and AF-2 is located in the E domain along with the ligand binding domain (LBD)(7).(Figure 1)

Transcriptional response to estrogens occurs in two steps. The first step is the binding of the ligand to the ER to form a dimeric E₂-ER complex. The ER is very flexible in the variety of ligands it can bind although the different ligands will elicit varying degrees of transcriptional response(4). This binding causes the ER to become derepressed (5). It also stimulates the second step, the binding of the complex to specific DNA sequences known as estrogen response elements (ERE). These EREs are located upstream of coding regions of estrogen responsive genes. The ERE has been determined to be a 13 base pair palindromic inverted repeat: 5’-GGRCAnnnTGACC-3’, where n equals any nucleotide in the center spacer region (Figure 2). The liganded-ER-ERE complex is then thought to act with various transcription components(4).

Exposure to estrogens, especially through estrogen replacement therapies, has shown to increase the risk of developing breast cancer. Tamoxifen (TAM) is an antiestrogen drug used to treat breast cancer and is known as a selective estrogen receptor modulator (SERM). SERMs act as estrogen agonists in some tissues and as antagonists in others. Tamoxifen is a nonsteroidal triphenylethylene derivative that is structurally unrelated to the endogenous estrogen, estradiol (E₂) (3,7,9). Tamoxifen competitively binds to the ER with relatively low affinity, but the major metabolite of TAM, 4-Hydroxytamoxifen (OHT), binds with an affinity comparable to that of E₂. The affinity of OHT for the ER, even though it is nonsteroidal, is due to the fact that two rings lie in approximately the same position as the A and D ring of E₂. One of these rings contains a phenolic hydroxyl group at an equivalent position to the C-3 phenol of E2 which greatly contributes to OHT’s affinity for the ER(7). (Figure 3)

Tamoxifen has been used to treat breast cancer in the United States since 1978. Use of TAM on patients with ER positive breast tumors for two years has resulted in a 38% decrease in breast cancer recurrence and a 24% reduction in death from malignancy. Tamoxifen also reduces the risk of contralateral breast cancer by 39%. Tamoxifen has been shown to have very few side effects. Another positive effect of TAM are that it behaves as an estrogen agonist in bone and lipids. This agonistic activity results in maintenance of bone mineral density and a reduction in low density lipoprotein cholesterol(2,3,6). Tamoxifen also shows a small degree of agonistic behavior in endometrial cells. In 1996 the International Agency for Research on Cancer (IARC), concluded that there is a strong association between development of endometrial cancer and use of TAM. The IARC classified TAM as a group 1 carcinogen but treatment is still encouraged by the IARC because the benefits of TAM are far greater than the risk of endometrial cancer(3).

Research is currently being performed to explain the agonistic/antagonistic behavior of antiestrogens in various tissues. Once the mechanisms of action of antiestrogens are understood, new drugs may be developed that will be useful treatments against cancer. This understanding will allow development of drugs without any cell proliferative effects that also display the positive effects of estrogen agonists.

Recent structural studies seem to show that AF-2 activity is regulated by the specific ligand bound to the LBD. The antagonistic activity of antiestrogens can be explained by conformational changes that occur in the LBD when the ER is bound to an antiestrogen(8).

Several proteins that enhance ligand dependent transcriptional activation are classified as transcriptional coactivators. Members of the p160 family of coactivators contain a short sequence, LXXLL (L is leucine and X is any amino acid), called the nuclear receptor(NR) box. TIF2 is a human coactivator that contains three NR box regions. Through site directed mutagenesis it was shown the NR box II binds most tightly to the LBD(8).

When E₂ or a compound that is purely agonistic binds to the LBD of the ER a conformational change occurs that forms a shallow groove. This groove is composed of residues from helix 3, helix 4, the turn between helixes 3 and 4, helix 5 and helix 12 of the LBD. The groove has a hydrophobic floor and sides with the ends being charged. The LBD interacts with the hydrophobic face of an alpha helix in the NR box II peptide. The face of the helix is formed by side chains of Ile-689 and the three leucines (Leu-690,693,and 694) in the LXXLL motif of the NR box. The Leu-690 side chain becomes embedded in the groove and forms van der Waals contacts with Ile-358, Val-376, Leu-379, Glu-380, and Met-543 of the LBD. The Leu-694 side chain forms van der Waals contacts with Ile-358, Val-376, Leu-379, Lys-362, Leu-372, and Gln-375. The Ile-689 and the second NR box leucine, Leu-693 rest against the rim of the groove.(Figure 4) The LBD also stabilizes the conformation of the NR box II peptide by forming capping interaction with both ends of the peptide helix. The g -carboxylate of Glu-542 hydrogen bonds to the amides of the residues of the N-terminal turn of the peptide helix and the e -amino group of Lys-362 hydrogen bonds to the carbonyls of the residues of the C-terminal turn.(Figure 4) When point mutations were induced that changed the hydrophobic nature of the floor and sides of the groove and/or changed the charges at the rim of the groove, the LBD-E₂ complex did not bind coactivators. These experiments confirmed the importance of hydrophobic packing and capping interactions for coactivator binding(8).

When the LBD binds the tamoxifen metabolite, OHT, there is a conformational change. It is different in both secondary and tertiary structure than when E₂ binds to the LBD_. Helices 3, 8, and 11 adopt an extended conformation in the LBD-OHT complex. Also helix 12 has a different composition and orientation. In the E₂ complex, helix 12 consists of residues 538-546; in the LBD-OHT complex, helix 12 consists of residues 536-544. The orientation of the helix has changed so that instead of forming part of the hydrophobic pocket, as in the E₂ complex, it is bound to the coactivator groove. This groove appears to be nearly identical to the groove formed in the E₂ complex except it does not include residues from helix 12.

Helix 12 interacts with the peptide binding groove with a stretch of amino acid residues (540-544) that resemble an NR box. It contains residues LLEML instead of LXXLL as in an NR box peptide. The Leu-540 and Met-543 form van der Waals contacts deep in the hydrophobic groove much like Leu-690 and Leu-693 of the NR box II peptide in the E₂ complex. Interaction of helix 12 with the groove is also stabilized by capping interactions as seen in the E2 complex. In addition to mimicking the binding of an NR box protein, helix 12 also forms van der Waals contacts with an area outside of the LBD protein binding groove.(Figure 4)(9)

The bulky side chain of OHT plays a major role in the orientation of the entire OHT molecule in the LBD complex. The side chain exits the binding pocket between helices 3 and 11. The positioning of the dimethylaminoethyl region of the chain is stabilized by interactions with Thr-347, Ala-350, Trp-383, and by a salt bridge between the diethylamino group and the b -carboxylate of Asp-351. These interactions cause the B ring of OHT to be oriented deeper in the pocket than the equivalent ring of E₂. The shift in the positions of this ring causes the active amino acids in the pocket to adopt conformations distinct from the one they adopt in the E₂ complex. These structural effects also influence residues throughout the pocket, not just the residues that interact with the B ring. These structural adjustments result in the loop between helices 11 and 12 to be longer than in the E₂ complex. This longer loop allows helix 12 to reach the hydrophobic groove. The side chain also affects the orientation of helix 12 more directly. Positioning of helix 12 above the binding pocket, as it is in the E₂ complex, would result in steric clashes between the dimethylaminoethyl region of the side chain and the side chain of Leu-540. These structural changes induced by the presence of the side chain and deeper position of the B ring prevent the OHT-LBD complex from binding a coactivator.(9)

After looking at data from this and previous related studies, it has been proposed that this peptide binding groove formed by residues from helices 3, 4, 5, and 12 is the AF-2 surface of the ER. When an antiestrogen is bound to the LBD, the mechanisms that result in the AF-2 surface being incomplete and blocked by helix 12 can help to explain the antagonistic behavior of antiestrogens. A coactivator cannot bind and activate AF-2 to stimulate transcription of genes normally influenced by estrogen when an antiestrogen is bound(8).

Although these studies can explain one mode of antagonism by antiestrogens, they do not explain why antiestrogens can be agonistic in some tissues. Currently several reasons for the mixed behavior of TAM have been proposed. It is thought that AF-2 may not be required for all promotors(9).

Studies comparing the activities of ERa and ERb have concluded that antiestrogen agonism is ERa AF-1 dependent. The differences between ERa and ERb are that ERb does not respond at all to antiestrogens and that the A/B domains are not highly conserved. This is the domain where AF-1 is located. In these experiments proteins were constructed that contained an ERa A/B domain with the rest of the protein being ERb , and also some with an ERb A/B domain with the rest of the protein being ERa . To produce normal human ERs complementary DNA that codes for the human ERa and human ERb are inserted into plasmids. To construct the proteins that had both ERa and ERb , restriction enzyme sites were created by site directed mutagenesis. The plasmids were then cut by enzymes and joined to PCR generated inserts of the A/B domains. The results showed that antiestrogens only induced transcription when an ERa AF-1 was present(6). The results of this study, as well as several others, indicate that this specific region may be critical in determining whether an antiestrogen will display agonistic behavior(2,6). The current hypothesis is that cofactors interacting with AF-1 could determine the relative agonism of various antiestrogen ligands. The idea is that this cofactor would be present in varying amounts in different cell types, thereby explaining the cell specific actions of antiestrogens(4).

Reasons why AF-1 will sometimes synergize with AF-2, and other times works independently is still unknown. A recent study has shown that AF-1 may also interact with p 160 coactivators. AF-1 and AF-2 can squelch each other’s activity, which suggests they may compete for the same molecules. Experiments were performed where extra GRIP1, a protein that is a member of the p160 family, was added to cells with ER bound to estrogen, ER bound to tamoxifen, and ER with no ligand. With out GRIP1 only the ER bound to estrogen elicited a transcriptional response. When GRIP1 was added, all showed a transcriptional response. To confirm that the increased response was mediated by AF-1, mutations were introduced into ER proteins. Some protein had deleted AF-1 sections, some had AF-2 sections with double point mutations, and some had both. The experiments found that the proteins with deleted AF-1 sections in the presence of GRIP1 showed transcriptional activity with liganded estrogen and with no ligand. The proteins with AF-2 mutations showed response to estrogen and tamoxifen, but not in the absence of ligand. The proteins with both mutations showed no response. This showed that activity in the presence of GRIP1 is AF-2 dependent but ligand independent. GRIP1 can stimulate AF-2 but not AF-1 in the absence of a ligand. It is concluded that when a ligand is bound, the enhanced activity observed in the presence of GRIP1 is due to enhancement of AF-1 activity not enhancement of AF-2 activity.(12)

The A/B domain is divided into three subregions: Box 1 which contains amino acids 41-64, Box 2 which contains amino acids 87-108, and the kinase target region (KTR) which contains the amino acids 108-129 that surround serine residues (amino acids 104,106,and 118) which can be phosphorylated by a kinase. Mutational studies show that all three of these regions are important in AF-1 activity. Deletion of each region leads to stepwise reduction of activity. Box 1 was found to be especially important for AF-1 activity. Alanine substitutions for serines in the KTR region resulted in a slight enhancement of activity, but it is concluded that GRIP1 action at AF-1 is independent of the phosphorylation sites.(12)

Previous experiments have shown that the AF-2 site recognizes and bind to the LXXLL motif known as NR boxes. It has been determined that the AF-1 site does not recognize NR boxes. Studies in which the NR boxes were mutated showed that they bound weakly or did not bind to AF-2, but bound strongly to AF-1. When the C-terminus region of p160s was deleted, it was found that AF-2 bound these proteins tightly while AF-1 did not bind them. The region thought most to contribute to p160 binding to AF-1 is a glutamine rich region (amino acids 1121-1282) that is highly conserved among the p160 family of proteins.(12)

Another hypothesis to explain the varying agonism of TAM is related to the fact that both estrogen and TAM increase the secretion of Transforming Growth Factor-b (TGF-b ). TGF-b is a family of cytokines that regulate many aspects of cellular activity by playing a role in the control of gene expression. This may especially explain the proliferative effects of TAM in endometrial tissues and part of the reason for its repression of breast cancers. TGF-b in early tumor growth inhibits tumor proliferation. During advanced tumor growth TGF-b may actually promote growth and spread of tumors. Researchers do not know what causes this change in behavior. TAM could enhance TGF-b expression in early breast cancers and/or suppress TGF-b expression in late breast cancer. In the uterus the opposite could be occurring to promote endometrial cancer(4).

Studies on mechanisms of estrogens and antiestrogens are being performed to help in cancer assessment, treatment, and prevention. Also these studies are leading to treatments for osteoporosis and new estrogen replacement therapies. Information obtained in a study being conducted on the downstream trancriptional activities of estrogens is now being used to better classify breast cancer cells. Formerly breast cancer cells were classified as either ER- or ER+ cells. About 40% of ER+ cells respond to antiestrogen treatment. It was discovered that ER+ cells can be further divided into hormone dependent or hormone independent. Hormone dependent tumors are the only ones that will respond to antiestrogen treatment. This new assay will allow for more efficient treatment of breast cancer patients who have ER+, hormone independent cells(1).

Other SERMs are being studied and developed for use in breast cancer treatment and prevention, and for hormone replacement therapy, to prevent osteoporosis and Alzheimer’s disease. Toremifene is a tamoxifen analog that has been recently approved for the treatment of breast cancer. Other triphenylethylenes that are in clinical trials are droloxifene, iodoxifene, and TAT-59.(Figure 5) Pure antiestrogens such as ICI 182,780and ICI 167,384 are being explored as treatments for breast cancer after failure of conventional therapy. Raloxifene is a targeted antiestrogen and is a benzothiopene derivative. It was undergoing phase 2 clinical trials at M.D. Anderson Cancer Center in 1986 but did not show effects on breast cancer. Recently it has been prescribed as an alternative to hormone replacement therapy. It does not have an increased risk of cancer as do conventional estrogen replacement therapies; in fact, it seems to have preventative effects(3).

Further research is needed to understand the complex activities involved in estrogen and antiestrogen transcriptional responses. Much is still not known about the activities of AF-1. It is known that the antagonsitic behavior of antiestrogens is AF-2 dependent and agonistic behavior of antiestrogens is AF-1 dependent, although the exact mechanisms are still being determined. These studies have shown that AF-1 will mainly synergize with AF-2 activity, it can weakly mask AF-2 mutations, and that independent AF-1 is strongly dependent on excess of p160 coactivators.(12) Studies on estrogen receptors have led to better methods of breast cancer classification. These studies also have led to the development of drugs for use as cancer treatments as well as hormone replacement therapies. Further elucidation of the events invloved in both estrogen and antiestrogen stimulated transcription needs to be investigated. This information will allow increased future advancement in various women’s health issues.

Copyright © 1999 Krista Bennett and Koni Stone