Serendipity and Science

On the origins of chemical species, by Dan Lednicer

The enormous strides that have recently been made in molecular biology hold great promise for speeding the discovery of pharmaceuticals to treat diseases that have so far been recalcitrant to drug therapy. The day may well be in the offing when a majority of important new pharmaceutical products will owe their existence to carefully crafted research programs based on the increasingly detailed understanding of the molecular biology involved in the particular disease that is being addressed. The cost of drug discovery and development has become so high that research management is eagerly looking forward to a more systematic and hopefully predictable process than that which has prevailed in the past. The genealogy of quite recently introduced drugs however provides a good illustration of the role that serendipity, intuition or even pure chance have played in drug discovery up until quite recently.

The chemical structures of these compounds can be traced back to a drug first introduced over half a century ago. Molecular manipulation by organic chemists over the next fifty years led to the discovery of at least three compounds each of whose annual sales are currently in the billion-dollar range. The activity of the first of these, the oncology product, Novaldex (tamoxifen) can be extrapolated from that of the compound that started the research. The biological activity of the other two descendants, the selective NSAIDs Celebrex (celecoxib) and Vioxx (rofecoxib) could hardly have been predicted.

The story begins in the 1930s with the discovery of the estrogens. The state of structural determination was still in its infancy; the various instrumental methods now used routinely to unravel structures had not even been conceived. By dint of hard and very elegant work the structure of principal hormonal agent, estradiol, was found to be a based on the so-called steroid nucleus.

It was known by then that low levels of estrogen were associated with some diseases suffered by women. A promising approach for treating those conditions involved increasing hormone levels by administering pure estradiol. Supplies were however a major problem since the compound occurred at such low levels that it was not considered practical to isolate it from animal sources. (It is interesting, as an aside, to note that exactly such an extract, Premarin, which consists of conjugated estrogens isolated from mare’s urine comprises the enormously successful drug used mainly by postmenopausal women).The methods available in the 1930’s for chemical synthesis were not yet up to the task for preparing by total synthesis what were then considered to be complex structures. The search for purely synthetic structurally simpler compounds that showed estradiol-like activity led to the discovery of the ill-fated drug, diethyl stilbestrol (DES). Subsequent research was to show that the receptor for estradiol is notoriously promiscuous and will interact with a large variety of compounds that only vaguely resemble its endogenous activator. Omitting one of the ethyl groups and replacing the other by a substituted benzene ring led the yet another purely synthetic compound, chlorotrianisene (TACE) that was used as an estrogen in the early 1940s.

Both DES and TACE are poorly soluble in water making it difficult to administer those drugs by injection. The phenolic groups in DES are too weakly acidic to ionize at physiological pH while TACE lacks any ionizable groups whatever. Arguably prompted by this chemists at the Richardson Merell Company tackled the problem in the late 1950s by attaching an amino group that would form salts with acids to one of the benzene ring of a TACE-like molecule by way of a so-called basic ether1. The resulting products were, as hoped, somewhat more soluble in water than the parent compounds. To their surprise however those basic ether derivatives were no longer simple estrogens. Instead they also showed a measure of anti-estrogenic activity. In addition, because of that new activity many also acted as a contraceptive in the rodent model then in use for screening compounds for antifertility activity. Further refinement of the structure led the company to market one of these drugs as clomiphene. This compound, which is a mixture of cis-trans isomers, shows a mixture of estrogenic and antiestrogenic activities.

Reports of this discovery occasioned the start of work in a number of other pharmaceutical laboratories aimed at exploiting this unforeseen lead. Chemist at the then ICI Company (now part of Aventis via Zeneca) replaced the potentially unstable chlorine by an ethyl group. They then investigated the resulting compound, tamoxifen, extensively as an estrogen antagonist in the 1960s. The growth neoplasms such as breast cancers were known even then to be stimulated by circulating estrogens. Heroic measures, such as hysterectomy or administration of androgenic drugs, had been used by oncologists of the day to treat breast cancer.  Clinical trials showed that much the same effect could be accomplished with oral doses of tamoxifen. The concurrent development of tests for estrogen sensitivity of tumors allowed clinicians to select those patients that would profit from drug treatment. Tamoxifen today comprises mainline adjunct post-operative therapy for estrogen receptor positive breast cancer2.

Several other groups on the other hand focused on the antifertility activity displayed by the Merell compounds hoping to find non-steroidal contraceptives. Chemists Upjohn designed an analogue that included a new ring. This would, they thought, more closely resemble a steroid. The new ring was five-membered since this was accessible in a relatively few steps. The fact that these compounds were active in the screen led to a full scale analogue program. The most potent indene of the series,U-11,555A3 went as far as Phase I clinical testing; it failed at this point because it caused photosensitivity.

The ring expanded dihydronaphthalenes that were somewhat more difficult to prepare turned out to be uniformly more potent than corresponding indenes4. One of those, U-11,000A, replaced U-11,555A in the clinic. Though the compound failed as a contraceptive it was tested successfully by the National Cancer Institute as an estrogen antagonist under the USAN name nafoxidine.

Some years later, the series was revisited by scientists at Lilly. The structures of the compounds that they investigated differed from those prepared at Upjohn mainly by the inclusion of a ketone carbon atom (C=O) between the fused ring and that which holds the side chain with the amino group. This will of course subtly alter the shape of the molecules. Increasing sophistication of pharmacological tests had led to the identification of subtypes among receptors, including those that recognized estrogens. These two compounds, trioxifene5 and raloxifene6 were investigated at in some detail because their pattern of activity on the receptor subtypes differed from the classical estrogens and their antagonists. The pharmacological response pattern of these compounds was deemed to be particularly suitable for alleviating menopausal and post-menopausal symptoms. This selective activity has led the Lilly scientists to refer to them as 'designer estrogens'. These compounds, like many estrogens are in addition quite useful in treating or preventing osteoporosis. Raloxifene is currently used extensively under the trade name of Evista.

The activities of the compounds that have been considered up to now result from interaction with a receptor that recognizes estradiol, the agent that prompted the series to begin with.  The second half of the story is more interesting in that it involves a totally unrelated biological target.

Another group at Upjohn was heavily involved in indole chemistry at roughly the same time as the work on the indenes. They prepared a series of indoles that carried two benzene rings in the same position as those on the indene U-11,555. One of these unexpectedly showed very good anti-inflammatory activity. This compound was taken to the clinic under the generic name indoxole. This agent like its indene predecessor also failed Phase I testing; interestingly for the same reasons. It too caused intolerable photosensitivity.

There was great reluctance to walk away from this lead because of the very good activity the compound had shown in animal model. The fact that it was not related to any of the then-known anti-inflammatory agents added reason to try to salvage this lead. The major program aimed at producing non-phototoxic analogues of indoxole gave disappointing results. Most of the modified indoxoles lost anti-inflammatory activity; the few analogues that did retain activity also retained the toxicity. It became quite clear that the phototoxicity in both this molecule and its precursor, U-11,555 was due to the shared structural feature: the presence of two benzene rings on the double bond in the fused five-membered ring7.

A modest series of analogues was prepared to find out whether the indole ring could be replaced by a simpler five heterocyclic ring. The two methoxy-substituted benzene rings were retained since the earlier program had indicated that they were crucial for anti-inflammatory activity. One of those, the dianisyl thiazole below, interestingly, retained anti-inflammatory activity. Though the level of activity was too low to prompt further follow-up it was however good enough to prompt the filing for a patent on this and some of its derivatives8.

The indoxole program dates back to the mid-1960s, just at the time that Upjohn was becoming ever more deeply involved in prostaglandin research. The biochemical role of these hormone-like substances was still far from clear. The discovery that prostaglandins were directly involved in such injurious responses as inflammation and platelet aggregation that leads to thrombosis and strokes was still some years off. By the mid-1970s however Vane and his colleagues had demonstrated that nonsteroid anti-inflammatory drugs (NSAID) worked by inhibiting an enzyme – cyclooxygenase – involved in the biosynthesis of prostaglandins. Inhibitors of that enzyme will also counteract platelet aggregation. A new screen for such inhibitors picked up the old thiazole lead. The ensuing analogue program culminated in the synthesis of a compound in which the hydrogens on the methyl group were replaced by fluorine9. This agent showed enough activity so that it was assigned a generic name, itazigrel, an indication that it was a clinical candidate. Publications as recent as 1994 identify itazigrel as an inhibitor of cyclooxygenase10.

Research on new and improved NSAIDs had pretty much run its course by the mid-1990s. There is room on the market for only so many 'profens'. All these drugs seemed to have the same propensity for contributing to stomach ulcers. By then it had been fairly well established that this side effect was in fact a consequence of the drug’s action on cyclooxygenase. This is due to the fact that the enzyme is also involved in the synthesis of prostaglandins that normally protect the lining of the stomach. . The discovery that the enzyme occurs in two forms led to a new wave of research on NSAIDs. One of those enzymes, COX-2, seemed to be involved in inflammation but not in the chain of events that led to ulcers. In 1998, G.D. Searle, by then part of Monsanto, announced that they had developed a specific COX-2 inhibitor. This drug Celebrex (celecoxib) was heralded as an NSAID with much reduced effects on the stomach. Examination of the structure of this compound shows that it bears an interesting relation to itazigrel and its ancient predecessor. The relation becomes even closer when one considers that one of the methoxyl groups (CH3O) in itazigrel is almost certainly cleaved to a phenolic hydroxyl (OH) in the body. It is well known that sulfonamide groups (H2NO2S) such as that present in celecoxib are often biologically equivalent to phenols.  The structure of newer COX-2 drug, refecoxib, still retains the basic backbone though with many changes in functionality.


1.  R.E. Allen, F.P. Palapoli, E.L. Schumann, M.G. Vancampen, US Patent 2,914,563 (1959).
2.  G.R. Bedford, D.N. Richardson, Nature, 212, 733(1966).
3. D. Lednicer, J.C. Babcock, P.E. Marlatt, S.C. Lyster, G.W. Duncan, J.Med.Chem., 8, 52(1965).
4. D. Lednicer, S.C. Lyster, B.D. Aspergren,  G.W. Duncan, J.Med.Chem., 9, 172(1966).
5. C.D. Jones, T. Suarez, E.H. Massey, L.J. Black. F.C. Tinsley, J.Med.Chem., 22, 962(1979)
6. C.D. Jones, M.D. Jevnikar, A.J. Pike, L.J. Black, A.R. Thompson J.F. Falcone, J.A. Clemens, J.Med.Chem., 27, 1057(1984)
7.  J. Szmuszkovicz, E.M. Glenn, R.V. Heinzelman, J.B. Hester, and G.A. Youngdale, J.Med.Chem., 9, 527(1966).
8.  D. Lednicer, U.S. Patent 3,560,514 (1971).
9. R.H. Rynbrandt, E.E. Nishizawa, D.P. Balgoyen, A.R. Mendoza, K.A. Annis, J.Med.Chem.,24, 1507(1981).
10. A. Tanaka, H. Sakai, Y. Motoyama, T. Ishikawa, H. Takasugi, J Med Chem., 37,1189(1994).

All opinions expressed in this article are those of the author and are not necessarily those held by DBSW, or its associates.