Targeting Estrogen Receptor-B: A Case Study in Drug Discovery

 Models of estradiol (left) and genistein. For decades, researchers believed that a single estrogen receptor mediated the effects of estrogens in the body. So imagine their surprise when Jan-Åke Gustafsson of the Karolinska Institute in Stockholm announced at a 1996 Keystone symposium the discovery of a second estrogen receptor in the rat prostate. The revelation added unexpected complexity to scientists' understanding of estrogen's biological action. Many attendees scurried back to

By | May 19, 2003

 Models of estradiol (left) and genistein.

For decades, researchers believed that a single estrogen receptor mediated the effects of estrogens in the body. So imagine their surprise when Jan-Åke Gustafsson of the Karolinska Institute in Stockholm announced at a 1996 Keystone symposium the discovery of a second estrogen receptor in the rat prostate. The revelation added unexpected complexity to scientists' understanding of estrogen's biological action. Many attendees scurried back to their laboratories and commenced research to find out more about this new nuclear hormone receptor, called estrogen receptor beta (ER-b).1

A similar case, the discovery of a second cyclooxygenase (COX) receptor in the early 1990s, led to effective new pain medications (see sidebar). Encouraged by this, scientists believed that ER-b held exciting promise based on its relationship to ER-a, the target of numerous successful drugs (e.g., Premarin and tamoxifen). The receptor's tissue distribution fueled their enthusiasm: It was not the dominant estrogen receptor in the uterus, a key organ to avoid stimulating with estrogens in a hormone replacement therapy (HRT) regimen. ER-b's potential as a blockbuster drug target for HRT outweighed uncertainty about its actual therapeutic value. Immediately, a dozen or more pharmaceutical companies launched ER-b programs, each assigning 20 to 30 scientists to investigate the target.

COX-2: A Cautionary Tale
Just a few years prior to the discovery of ER-b, scientists made a similar find in cyclooxygenase-2 (COX-2). The rapid progress from the enzyme's discovery to blockbuster medications (within 10 years two COX-2 inhibitors together made $4 billion (US) per year) made pharmaceutical companies salivate over the new estrogen receptor. But while the drugs continue to be blockbusters for Merck and Pharmacia (now Pfizer), some recent reports suggest that the COX-2 inhibitors may be more of a mixed bag than originally believed, highlighting the difficulty of moving compounds from the lab to the clinic.

For years, researchers had hinted at a second form of COX--the enzyme that aspirin and other anti-inflammatory drugs act upon to block pain--but it wasn't until 1991 that they found it. Scientists speculated that COX-2 inhibition was responsible for desired anti-inflammatory effects, while COX-1 inhibition was responsible for the nasty side effects of stomach irritation and lowered platelet aggregation. Numerous pharmaceutical companies launched COX-2 programs based on the reasoning that the ideal painkiller--an anti-inflammatory agent lacking adverse effects--would target COX-2 only, wrote Rod J. Flower, a biochemical pharmacologist at London's Queen Mary University, in a recent review.1

As a potential target, COX-2 holds many advantages over ER-b. For one thing, COX-2 had a clear physiological link to inflam-mation, unlike ER-b, whose physiology remains unclear. And crystallization of the COX-1 and COX-2 showed that, despite high sequence similarity, a single amino acid change significantly alters the binding domains between the two enzymes. Pharmaceutical companies quickly capitalized on these properties, identifying several compounds with high COX-2 selectivity. Two of these demonstrated effective pain relief without adverse effects in clinical trials, and celecoxib (Celebrex) and rofecoxib (Vioxx) received Food and Drug Administration approval and hit the market.

But the bloom may be coming off COX-2's rose. Some reports have shown that COX-2 inhibitors provide no better protection against stomach ulcers than other nonsteroidal anti-inflammatory drugs, and others link COX-2 with increased risk for heart and kidney complications, demonstrating yet again that what appears simple and elegant on paper and in cell culture can prove frustratingly complex in practice.

1. R.J. Flower, "The development of COX-2 inhibitors," Nature Reviews, 2:179-91, March 2003.

Now, nearly eight years later, ER-b's therapeutic value still remains largely unknown. Frustrated, numerous pharmaceutical companies such as GlaxoSmithKline and Eli Lilly and Co. have shelved their ER-b programs in favor of more promising targets. Other companies such as Wyeth and Merck & Co. have abandoned their initial HRT goals but discovered that ER-b may be exploited for other purposes. As yet, no ER-b-targeted drug has reached clinical trials, and in the more than 1,500 publications on ER-b, the pharmaceutical companies have released only a few clues to its biological significance.

"When ER-b was first disclosed, it was the golden child, the next best prospect for selective HRT, and there was much optimism around it as a drug target. But it's taken longer than expected to pan out, and it's been a very frustrating process," says biologist Heather A. Harris, co-team leader of the ER-b program at Wyeth.

But for Wyeth at least, the effort seems to have panned out at last. The first company to go public with any promising findings, Wyeth is moving forward with an ER-b-targeted anti-inflammatory agent. Potential indications include inflammatory bowel disease, which--though not as large a market as HRT is for menopause--still represents a billion dollar market, according to some drug company estimates.

Indeed, ER-b may turn out to be just as valuable as scientists initially hoped back in 1996. Merck and Schering-Plough also have had some success with ER-b. Gustafsson speculates that ER-b could prove to be a viable target for such maladies as cancer, infertility, and depression.

The long, circuitous route leading to Wyeth's announcement provides insight into how the drug discovery process has both changed and stayed the same. On the one hand, the process is decidedly faster in many respects, with pharmaceutical companies harnessing technological leaps in gene mapping, homology modeling, structure determination, high-speed chemistry, and knockout models. Yet at the same time, it remains reliant on intensive labor, sound intuition, and sometimes, serendipity.

A LEAP OF FAITH When pharmaceutical companies dove into their ER-b projects, they were taking leaps of faith, because they did not yet know the target's actual purpose. They were also changing the landscape of drug development. Traditionally, companies pursued known drug targets based only on their links to particular diseases or disorders. "Who wants to give you 10 to 12 chemists to work on a project when you have no idea what the target does?" Harris asks.

Yet such leaps are becoming more and more common, as companies--buoyed by genomics data--increasingly pursue unvalidated targets based solely on their relationships to validated ones, such as COX-2's relationship to COX-1, or ER-b's relationship to ER-a. "In my department, ER-b was the first genomics-based project," Harris notes. "It's a real shift in philosophy, but ER-b is part of a new era, and we're going to have to get used to doing drug discovery this way."

Adding to scientists' uncertainty over function was their suspicion that ER-b drug development would prove a tough nut to crack. "We had a feeling that it wasn't going to be an easy target," says Brad R. Henke, who led GlaxoSmithKline's medicinal chemistry investigation of ER-b. "The mere fact that ER-a had been known for a long time and no one had known about ER-b clued us in that the proteins are probably pretty similar."

Courtesy of Jan-Ake Gustafsson & Konrad Koehler
 DRUG TARGET: X-ray crystallographic structure of human estrogen receptor (ER)-b complexed with the phytoestrogen genistein.

It didn't take long to validate this hunch. The first steps were simple enough. Using Gustafsson's rat ER-b clone and the human genome sequence then available, researchers obtained the human homolog within weeks--a marked improvement over torturous cloning efforts of years past. They then rushed to obtain the protein's structure. Again, a process that once took a year or more was accomplished in months, thanks to advanced knowledge of nuclear hormone receptor structure.

The resulting data helped the scientists focus their medicinal chemistry efforts, but it also underscored the challenge ahead: In comparing ER-b's crystal structure against ER-a's, they found few differences on which to base their efforts. "Because crystal structures ... revealed there were only two residues within five angstroms of the ligand that were different in the two proteins, we didn't expect many compounds to show selectivity for one receptor over the other," says Henke. "The two mutated residues were very subtle changes. We knew that we didn't have a lot of obvious difference points in the receptor to go after in an attempt to design subtype-selective ligands."

FINDING EFFECTIVE TOOLS Despite these considerable hurdles, the pharmaceutical companies set their sights on developing workable molecular tools. The hope was that ER-b-selective agonists and antagonists might circumvent some of the issues associated with current estrogen therapies, such as uterine stimulation and an increased risk of breast cancer.

To find these chemical tools, scientists at GlaxoSmithKline, Pfizer, and Wyeth, among others, used high-throughput binding assays to screen their companies' entire compound libraries for ER-b binders. (The more desirable approach, functional assays, was less amenable to high throughput in this situation, and in any case, little was known about the receptor's functional responses.)

Over about 10 weeks, Henke's group at GlaxoSmithKline plowed through mountains of 96-well plates containing the hundreds of thousands of compounds in the company's chemical library; finally, they produced a set of reagents that were modestly selective for ER-b over ER-a. (Today, technological advances would allow the team to use higher-density 384- or 1,536-well plates, Henke notes.) They selected the most promising among these for further optimization, considering such parameters as molecular weight, potency, solubility, and toxicology.

Ideally in drug discovery, scientists aim to design ligands that are at least 100-fold selective for their target. But the close similarity between ER-a and ER-b hampered these efforts. Henke's group, for instance, found a set of novel compounds called triazine-based ER modulators that are 30- to 40-fold selective, but could not push these chemicals' selectivity any higher.2

"We found very interesting compounds, but we didn't get ligands that we felt were good enough tools for probing ER-b," Henke says. "If you're going to do whole-animal studies and dose-range studies in an in vivo system, and you're trying to associate a certain kind of pharmacology response to a specific receptor, you really want something that's more selective for the receptor of interest than the compounds that we developed."

Henke speculates that the molecules his team developed were "probably too conformationally flexible" to work effectively, but says that appropriately discriminating compounds are out there; it's simply a matter of finding the right chemical template. As a result of these setbacks, GlaxoSmithKline dropped its ER-b program a few years ago; today it is pursuing estrogen-related research that Henke reports grew out of the ER-b work.

Pfizer also has struggled to develop effective and selective chemical tools. Senior research scientist Richard Chesworth says his team has narrowed its efforts down to a few molecules that it believes are ER-b agonists, but "we still don't know where we're going with our compounds," he says.

AGONIST OR ANTAGONIST? Even if these companies can identify suitable compounds, estrogen receptor drug development is complicated by the fact that one tissue's agonist is sometimes another tissue's antagonist. The cancer therapeutic agent tamoxifen, for example, clearly antagonizes ER-a in breast tissue, but it acts as a partial agonist in bone and the uterus. The osteoporosis drug raloxifene acts as an ER-a agonist in bone, but it is an antagonist in uterine and breast tissue.

Like ER-a, ER-b is found throughout the body. Most organs express both proteins, but differences do exist. ER-a is more prevalent in the liver, for instance, whereas ER-b is more prevalent in specific areas of the central nervous system and gastrointestinal tract. Like the alpha isoform, therefore, ample opportunity exists for confusing pharmacological responses to any beta-directed drug.

"It's very complicated. To say a compound is an agonist, you really need to understand what the cell background or what the tissue is," says Chesworth, who presented Pfizer's research on ER-b-selective compounds at a recent conference. Harris' group at Wyeth has worked to better understand this agonism and antagonism, as well.3

Courtesy of Jan-Ake Gustafsson & Konrad Koehler
 HORMONAL CHALLENGE: An X-ray structure of genistein bound to the ER-b ligand-binding domain demonstrates the difficulty that faced ER-b drug developers: Only two amino acids (shown) differ between the two estrogen receptors.

Click for larger version (71K)

A 'EUREKA' MOMENT Largely thwarted by ER-b's similarity to ER-a, some pharmaceutical companies decided to turn their attention to the new receptor's biological function. An ER-a knockout mouse exhibited various defects in the reproductive system, a known estrogen target. Company researchers looked for developmental defects in ER-b knockout models that would provide similar clues to that receptor's biology.

But again, ER-b stymied their work. Various efforts yielded four or five versions of an ER-b knockout mouse, according to Harris, with only very mild phenotypes, she says. "We spent a lot of time trying to develop an in vivo model, because we needed a positive control to benchmark the activity of our compounds against. We needed an endpoint," Harris says. Lacking a helpful knockout model, her group had to rely on measures of the amount of candidate drug compounds circulating in the blood.

Finally Wyeth caught a break. A colleague in Wyeth's Boston facility had recently tested a nonselective estrogen (estrogens have anti-inflammatory activity) in transgenic rat models of inflammatory bowel disease and discovered that it improved stool character. When Harris heard about the study's results, she asked him to test an ER-b-selective compound. Much to everyone's surprise, the tests showed that the selective compound had strong anti-inflammatory activity. It was definitely a "Eureka!" moment, Harris recalls.

FUTURE PREDICTIONS Today, research into ER-b and its pharmaceutical promise rolls on. The receptor's discoverer, Gustafsson, is now trying to unravel an apparent yin and yang-type relationship between it and ER-a. "It's an interesting way in which nature has constructed estrogen signaling--like two opposite, counter-balancing forces," he says. Whereas ER-a drives aggressive behavior, for example, ER-b seems to promote calmness. In a recent report, he describes studies with knockout models that demonstrate this symbiotic relationship.4

Though ER-b has proven to be a more difficult drug target than originally hoped, Gustafsson has reason to be pleased. "The mechanism of action for ER-b has turned out to be more sophisticated and complicated than we at first thought," he says. "ER-b appears to be involved in many important physiological contexts and very important to the functioning of the body." He cites ongoing investigations into the efficacy of ER-b as a drug target for breast, colon, and prostate cancer; chronic myeloid leukemia; psychological disorders; and infertility, for instance. The receptor may even play a role in the immune system, he says.

Now Gustafsson, as well as many members of the pharmaceutical industry, hopes to hear more stories like Wyeth's, in which the application of new technologies, combined with dogged research and a dose of serendipity, results in a promising new drug.

Jennifer Fisher Wilson ( is a freelance writer in Haddonfield, NJ.

1. G.G. Kuiper et al., "Cloning of a novel estrogen receptor expressed in rat prostate and ovary," Proc Natl Acad Sci, 93:5925-30, 1996.

2. B.R. Henke et al., "A new series of estrogen receptor modulators that display selectivity for estrogen receptor beta," J Med Chem, 45:5492-505, Dec. 5, 2002.

3. H.A. Harris et al., "Characterization of the biological roles of the estrogen receptors, ERa and ERb, in estrogen target tissues in vivo through the use of an ERa-selective ligand," Endocrinology, 143:4172-7, November 2002.

4. M.K. Lindberg et al., "Estrogen receptor (ER)-b reduces ER-a-regulated gene transcription, supporting a "ying-yang" relationship between ER-a and ER-b in mice," Mol Endocrinol, 17:203-8, February 2003.

Scientists have had two main sources of frustration with ER-b: They cannot discern the receptor's physiological role in the body, and they have difficulty finding highly selective ligands. Such troubles have stymied the development of drugs for other promising targets, as well. Now, some researchers and pharmaceutical companies are turning things around: Instead of starting with target proteins, they're starting with the small molecules.

This approach, called chemical genetics, simultaneously pinpoints gene function and drug candidates by screening the compounds against protein targets in cellular assays to study those targets' roles in disease. Traditional drug development comprises progressive steps: target identification, function determination, target validation, assay development, and high-throughput screening, all before identifying a lead molecule. In comparison, chemical genetics appears positively streamlined.

The chemical genetics approach has an interesting historical precedent: Penicillin, cyclosporin, and the glitazones (among others), were all discovered in this way. "Chemical genetics is just a modern term for the use of small molecules to induce phenotypes in cellular systems [analogous to classical genetics' study of DNA mutations]," explains Jeffrey Tong of Infinity Pharmaceuticals, a Boston-based chemical genomics startup. Now private and industry researchers have rediscovered the approach, thanks to advances in genomics and chemical technologies. At least one group has already seen success, describing a novel molecule, PNRI-299, that reduces asthma-associated inflammation.1

Infinity Pharmaceuticals recently completed a platform of small-molecule compounds that will form the basis for its future drug-discovery efforts. Because the company didn't want to limit itself to just those compounds typically found in large libraries, "We had to build a whole new collection from scratch that would afford us a broader access to diversity," says Tong, who is senior director of corporate strategy and development. Now Infinity is moving on to the product-discovery phase, and Tong reports that preliminary data are promising.

The idea of drug discovery that isn't initially tied to a specific protein target runs counter to the way pharmaceutical companies traditionally discover drugs, but chemical genomics is catching on at established companies as well as startups. Aventis, Novartis, Merck & Co., and Schering-Plough, among others, have contracted with chemical genomics companies, according to reports. But chemical genomics is unlikely to completely replace traditional ways of finding new drugs, says Tong; instead it will provide a fresh alternative.

"We're believe that we're starting with a better collection of compounds and interrogating biology on a cellular basis that is more physiologically relevant than has been done in the past," Tong says. "It's a complementary approach to the other multiple paths for drug discovery."

The National Human Genome Research Initiative of the National Institutes of Health evidently agrees. It recently announced plans to give academics access to chemical genomics, so they can investigate pathways and develop new molecular research tools and drugs, too.2

1. C. Nguyen et al., "Chemogenomic identification of Ref-1/AP-1 as a therapeutic target for asthma," Proc Natl Acad Sci, 100:1169-73, Feb. 4, 2003.

2. F.S. Collins et al., "A vision for the future of genomics research: A blueprint for the genomic era," Nature, 422:835-47, April 24, 2003.

Please indicate on a 1 - 5 scale how strongly you would recommend this article to your colleagues?
Not recommended
   Highly recommended
Please register your vote

Popular Now

  1. How Gaining and Losing Weight Affects the Body
    Daily News How Gaining and Losing Weight Affects the Body

    Millions of measurements from 23 people who consumed extra calories every day for a month reveal changes in proteins, metabolites, and gut microbiota that accompany shifts in body mass.

  2. That Other CRISPR Patent Dispute
    Daily News That Other CRISPR Patent Dispute

    The Broad Institute and Rockefeller University disagree over which scientists should be named as inventors on certain patents involving the gene-editing technology.

  3. EPO Revokes Broad’s CRISPR Patent
    The Nutshell EPO Revokes Broad’s CRISPR Patent

    Shortly after ruling out the earliest priority dates on a foundational patent for CRISPR gene-editing technology, the European Patent Office rescinded the patent entirely—and more are likely to follow.

  4. Learning Opens the Genome
    Daily News Learning Opens the Genome

    Researchers map learning-induced chromatin alterations in mouse brain cells, and find that many affect autism-associated genes.