Dr. Richard A. Glennon
Professor
Department of Medicinal Chemistry
Virginia Commonwealth University
A. INTRODUCTION
Our work with serotonin (5-HT) began in the late 1960s and was largely motivated by the earlier work of Woolley and colleagues who speculated that various aspects of mental illness might be related to 5-HT, and that certain hallucinogenic agents, notably LSD, might produce some of their behavioral effects via serotonergic mechanism. Serotonin had been identified as a putative neurotransmitter in the 1950s, and the late 1950s and early 1960s were considered banner years for the investigation of 5-HT. However, interest in 5-HT began to wane in the late 1960s. Although it was speculated that 5-HT might be involved in numerous physiological functions - both central and peripheral - it was difficult to understand how a single Serotonin (5-HT) neurotransmitter could be involved in functions as diverse as mental illness and hallucinogenic activity, to muscle contraction, to regulation of body temperature. Furthermore, with specific regard to hallucinogenic activity, there was a growing controversy as to whether hallucinogens behaved as 5-HT agonists or as 5-HT antagonists. Few tools (pharmacological techniques and potent novel agents) were available to investigate 5-HT.
Establishing a link between 5-HT and hallucinogenic agents. Having an interest in both hallucinogenic agents and in serotonin, I sought to study under someone with similar interests. Few investigators shared this interest at the time, one of the very few people working in this area was Dr. Peter Gessner (Department of Pharmacology, School of Medicine, State University of New York at Buffalo). What made Dr. Gessner's laboratory particularly attractive was not only had he trained under Dr. Irving Page, one of the original groups at the Cleveland Clinic responsible for the discovery of 5-HT in the 1940s, but that his laboratory was equipped to conduct both pharmacological and synthetic studies. I joined Dr. Gessner shortly after he and Page had synthesized 5-methoxy-N,N,dimethyltryptamine (5-OMe DMT), speculated that it might be a major psychoactive component of certain hallucinogenic crude plant products used by South American Indians, and had demonstrated that it was a potent serotonin agonist. It was also during this time that I met one of Dr. Gessner's ex-graduate students, Dr. Jerry Winter, who had just joined the Pharmacology faculty at Buffalo and was conducting some of the early and pioneering work with what was then a very new technique for investigating the actions of centrally acting agents in animals: drug discrimination.
My investigations with Dr. Gessner were focussed on the concept that hallucinogens might be acting as 5- HT agonists. Funded by a postdoctoral fellowship from what was then ADAMHA (i.e., the Alcohol, Drug Abuse, Mental Health Administration), we set out to determine if there was a relationship between the human hallucinogenic potency of tryptamine-based hallucinogenic agents and their affinity for 5-HT receptors. The procedure employed was the rat fundus preparation of Vane. Interestingly, we soon found a significant relationship between the two actions. Too few hallucinogens were known to allow much confidence in the findings. New agents could be synthesized to continue this work, but the question arose: how do we get hallucinogenic potency data on these new agents? Indeed, how would we know if these agents were even hallucinogenic. Perhaps, rather than searching for relationships between 5-HT receptor affinity and human hallucinogenic activity, we might rely on some other measure of behavioral activity. In late 1974, we initiated an investigation of the tremorgenic activity of hallucinogens, but a job offer precluded the completion of these studies.
In 1974, I was offered a position in the Department of Medicinal Chemistry at Virginia Commonwealth University. Fortunately, VCU was the home of another of the pioneers in drug discrimination: Dr. John Rosecrans. With the help of funding from NIDA, the rat fundus assay was reestablished, synthesis was begun, and Dr. Rosecrans permitted us to use some of his equipment for the initial drug discrimination studies. We spent several years evaluating different agents in order to identify a suitable training drug for the drug discrimination studies. Agents investigated as training drugs included 5-OMe DMT, mescaline, LSD, and others. Eventually, we settled on DOM, or 1-(2,5-dimethoxy-4-methylphenyl)-2-aminopropane. What came of all of this was that there was a significant relationship between rat fundus 5-HT receptor affinity and drug discrimination-derived (i.e., DOM stimulus generalization) potency for a large series of DOM compounds. But, how were the 5-HT receptors of a peripheral tissue preparation, such as rat fundus, related to the behavioral effects of these agents - an action that was almost certainly centrally mediated? So, even with this 5-HT/hallucinogen connection, there was as uneasy feeling that the results might be fortuitous, or worse - meaningless.
Multiple Populations of 5-HT receptors. Contemporaneous with these investigations, Dr. Solomon Snyder and others were developing techniques to examine the binding of various agents at different brain neurotransmitter receptors. Alas, a technique was available to measure the affinity of hallucinogens at central 5-HT receptors. In the later 1970s, several fortunate events occurred.
* During a seminar invitation to Eli Lilly, it was agreed that they would examine the binding of several of our agents at brain 5-HT receptors. Unfortunately, the samples provided to them, including DOM, were eventually found to bind with rather low affinity. However, during this initial visit, Dr. Ray Fuller described some of his work with a novel 5-HT agonist: TFMPP or (1-(3-(trifluoromethyl)phenyl)piperazine and provided us a sample. From our work with the DOM-trained animals, we were developing a hypothesis that hallucinogens might be acting as 5-HT agonists. If this were the case, then the DOM- trained animals might recognize this new 5-HT agonist. However, subsequent tests of stimulus generalization revealed that the DOM stimulus did not generalize to TFMPP. That is, DOM and TFMPP were found to produce dissimilar stimulus effects in the animals.
* Around this same period of time, I was invited to a Medicinal Chemistry symposium in Sweden and met with Dr. Uli Hacksell. Dr. Hacksell had just reported on yet another novel 5-HT agonist: 8-OH DPAT, or 8-hydroxy-2-(di-n-propylamino)tetralin. He provided us with a sample, and we found that the DOM- trained animals failed to recognize this 5-HT agonist, too.
* Dr. John Fozard requested that we examine a new 5-HT antagonist that he had just developed (subsequently termed MDL-72222), and we found that this antagonist failed to antagonize the DOM stimulus. Perhaps hallucinogens don't work via a 5-HT agonist mechanism after all! And yet, certain other 5-HT antagonists (such as, for example, pizotyline or BC-105) were able to effectively antagonize the DOM stimulus.
In 1979, Snyder and Peroutka reported the existence of two different populations of 5-HT receptors in mammalian brain: 5-HT1 and 5-HT2 receptors. Shortly thereafter, we established two collaborations, one with Dr. David Nelson (who had just returned from a postdoctoral position in France where he was studying 5-HT receptors with Dr. M. Hamon and had accepted an academic position at the University of Arizona), and another with Dr. Milt Teitler (who had just begun an academic position in Toronto). Dr. Nelson was interested in 5-HT1 receptors whereas Dr. Teitler was interested in 5-HT2 receptors. Dr. Nelson confirmed that hallucinogens possessed low affinity for 5-HT1 receptors (most likely that same population of receptors examined by Eli Lilly earlier) whereas Dr. Teitler found that several of the hallucinogens displayed significant affinity for 5-HT2 receptors. The germ of an idea was born - hallucinogens bind at 5-HT2 receptors, and it might be this population, not 5-HT1 receptors, that are responsible for their actions. Although our collaboration with Dr. Nelson took a different turn, he ultimately left Arizona for a position at Eli Lilly. Our collaboration with Dr. Teitler (now at Albany Medical College) continues to this day - nearly twenty years after our collaboration began.
Reexamination of some serotonergic agents. With the discovery of two different populations of 5-HT receptors, it was attractive to speculate that hallucinogens might represent 5-HT2 agonists whereas agents such as TFMPP and 8-OH DPAT might represent 5-HT1 agonists. This would explain why the DOM- trained animals failed to recognize TFMPP and 8-OH DPAT.
* In order to better characterize TFMPP, we trained rats to discriminate TFMPP from vehicle. As expected, these animals failed to recognize DOM (or other hallucinogenic agents). But, these animals also failed to recognize 8-OH DPAT!
* Animals were next trained to discriminate 8-OH DPAT from vehicle. Again, as expected, these animals failed to recognize DOM (or other hallucinogenic agents). But, the 8-OH DPAT-trained animals also failed to recognize TFMPP.
All evidence from our drug discrimination studies suggested that the three 5-HT agonists DOM, TFMPP, and 8-OH DPAT produce dissimilar stimulus effects in animals. Initially, this was difficult for us to accept. And then, in 1981, David Nelson reported the existence of two different subpopulations of 5-HT1 receptors: 5-HT1A and 5-HT1B receptors. Shortly thereafter, Fozard and Middlemiss identified 8-OH DPAT as a 5-HT1A agonist, and Hamon and colleagues introduced [3H]8-OH DPAT as a radioligand for labeling 5-HT1A receptors. Thus, we unexpectedly found ourselves with a behavioral assay (i.e., drug discrimination with rats trained to discriminate 8-OH DPAT from vehicle) with which to investigate novel 5-HT1A ligands.
From this point, our investigations took several different directions:
* Formulation of structure-activity relationships (SAR) for 5-HT1A agonist activity.
* Attempts to develop a 5-HT1A antagonist.
* Characterization of the TFMPP stimulus.
* Continued examination of relationships between hallucinogens and 5-HT2 receptors.
These new directions also forged a significant link between our use of classical medicinal chemistry techniques, the drug discrimination paradigm, and radioligand binding methodology, as major tools to investigate serotonergic and hallucinogenic agents.
B. STUDIES WITH SEROTONERGIC AGENTS
Examination of SAR of 5-HT1A ligands using radioligand binding and drug discrimination.
We synthesized and examined a variety of aminotetralins and other 5-HT1A agents. Although Hacksell and colleagues developed 8-OH DPAT, initially they examined 8-OH DPAT analogs only in functional, not in radioligand binding, assays. We thought that certain aminotetralins - those that were inactive in their functional assays - might be good candidates for possible development of 5-HT1A antagonists if they displayed reasonable 5-HT1A affinity. We found that the 8-hydroxy group, although important, was not critical for binding. We also found that introduction of large amine substituents, although resulting in compounds with little or no agonist activity, retained 5-HT1A affinity. In order to explain disparities between some important structure-function differences, we coined the term SAFIR (structure-affinity relationships). That is, SAR studies provide information for functional activity (intrinsic activity + affinity), whereas SAFIR studies provide information only for a subset of SAR - affinity. Often, differences exist between the two. While we were working on the aminotetralins, we also worked with the arylpiperazines. The latter group of compounds began providing us with some rather interesting results, so we switched our focus primarily to the latter.
Formulation of SAR and SAFIR for the serotonergic actions of arylpiperzines.
We undertook SAR and SAFIR studies on the arylpiperazines. Arylpiperazines, a structural moiety once thought to be predictive of 5-HT1B activity, were found almost immediately to be relatively nonselective agents. Certain arylpiperazines were shown to bind with high affinity, and some with considerable selectivity, for 5-HT1A receptors. One major structural feature responsible for high affinity 5-HT1A binding was found to be the nature of the N4-substituent. Agents with no N4-substituent, or a small N4- substituent, were termed simple arylpiperazines, whereas those with more elaborate N4-substituents we termed long-chain arylpiperazines, or LCAPs. Some LCAPs were found to bind at 5HT1A receptors with very high affinity (Ki approx. 0.1 nM). Detailed 5-HT1A, 5-HT1B, and 5HT-2 SAR and SAFIR studies were continued with the arylpiperazines.
Formulation of SAR and SAFIR for the serotonergic actions of aryloxyalkylamines.
The b-adrenergic antagonists propranolol and pindolol were shown by several groups of investigators to behave as 5-HT1A antagonists. These aryloxyalkylamines, then, might represent a new template for the development of 5-HT1A-selective agents. In addition to their affinity for 5-HT1A and b-adrenergic receptors, these agents also display affinity for 5-HT1B receptors. We set out to develop a series of aryloxyalkylamines with reduced 5-HT1B and b-adrenergic affinity. We investigated the SAFIR for the binding of the aryloxyalkylamines at 5-HT1A and 5-HT1B receptors and found that the alkyl -OH group was not required for binding, and that the side chain alkyl group could be shortened from three to two methylene groups without untoward effect on 5-HT1A affinity. Taking advantage of our work with the aminotetralins and arylpiperazines which suggested that 5-HT1A receptors tolerate tertiary amines with bulky substituents, whereas b-adrenergic receptors typically prefer secondary amines bearing an isopropyl group, we were fairly successful in ultimately identifying a series of compounds with fairly good selectivity for 5-HT1A receptors. These compounds, which we termed the MEP series, displayed a broad range of efficacy in an adenylate cyclase functional assay.
Investigation of arylpiperazines as potential anxiolytic agents.
Coincidental with our investigations, the arylpiperazine buspirone was described as a novel anxiolytic agent. We examined buspirone in animals trained to the benzodiazepine anxiolytic agent diazepam and found that diazepam-trained animals failed to recognize buspirone. Buspirone had been investigated by Bristol-Myers in radioloigand binding experiments shortly after the discovery of 5-HT1 and 5-HT2 (but before the discovery of 5-HT1A and 5-HT2A) receptors and had been demonstrated to bind with modest affinity at 5-HT1 receptors. Because buspirone, like TFMPP, is an arylpiperazine, we administered buspirone to the TFMPP-trained animals; again, no stimulus generalization . However, because we had found that certain arylpiperazines display high affinity for 5-HT1A receptors, and were recognized by 8- OH DPAT-trained animals, we examined buspirone in the latter group of animals. Indeed, buspirone was found to be 8-OH DPAT-like. Once it was realized that 8-OH DPAT was a 5-HT1A agonist, it seemed logical to conclude that buspirone might produce at least some of its effects via a 5-HT1A-agonist mechanism. Using rats trained to discriminate buspirone or ipsapirone from vehicle, we later showed that these animals recognized buspirone and 8-OH DPAT, but not TFMPP or diazepam. Taken together with the investigations of many other groups, current evidence suggests that the LCAP anxiolytic agents likely act via a 5-HT1A agonist or partial agonist mechanism.
Development of 5-HT1A antagonists.
A problem that long plagued early 5-HT1A work was the lack of a selective antagonist. Application of the SAR and SAFIR concepts developed in the course of our work with the arylpiperazines ultimately led to the 5-HT1A antagonist: NAN-190. This agent antagonized the 8-OH DPAT stimulus, and the hypothermia produced in rats by 8-OH DPAT. It was also found to be a very low efficacy partial agonist in the adenylate cyclase assay. Unfortunately, NAN-190 was later found to bind with considerable affinity at a- NAN-190 adrenergic receptors, and others showed that NAN-190 possesses some agonist character at presynaptic 5- HT1A receptors. It should be noted that Bristol-Myers developed BMY-7378 as a 5HT-1A antagonist at approximately the same time we introduced NAN-190; BMY-7378 also displayed some agonist actions. Several years later, Wyeth-Ayerst reported on a series of branched LCAPs that were more selective, and that represented the first silent 5-HT1A antagonists.
QSAR studies with 5-HT1A ligands.
Over the course of our studies with 5-HT1A ligands, we synthesized and evaluated the binding properties of a large number of compounds. Some were examined in functional assays (e.g., inhibition of forskolin- stimulated adenylate cyclase). We formulated concepts for the design of 5-HT1A agonists and antagonists. Our major focus was on the LCAPs (figure1). We conducted the first CoMFA study on 5-HT1A binding and eventually arrived at the conclusion that all LCAPs probably do not bind in a common manner.
Figure 1. General structure of long-chain arylpiperazines (LCAPs). The Aryl moiety can be phenyl, substituted phenyl, or heteroaryl; the Spacer can be branched or unbranched with an alkyl chain length of about two to five atoms being optimal; the spacer may contain a heteroatom. The Terminus can be widely varied amongst alkyl, aryl, amide, or imide. The Terminus seems to be associated with a region of bulk tolerance; we have incorporated substituents as large as adamantyl with no detriment to affinity. Depending upon the nature of the Aryl, Spacer, and Terminus groups, the LCAPs may behave as 5HT1A agonists, partial agonists, or antagonists.
Figure 2. LCAPs may bind at 5-HT1A receptors by using the piperazine moiety (X=N) as an anchor. Parallel structural changes either on the aryl portion (i.e., R) or on the terminus, do not necessarily result in parallel changes in receptor affinity. Multiple terminus binding sites may be possible; the length of the spacer, the nature of the terminus moiety, and/or aryl substituents may determine the specific manner of binding.
Examination of the stimulus properties of TFMPP.
Up until this time, only three populations of 5-HT receptors had been identified: 5-HT1A, 5HT1B, and 5HT2 receptors. By default, many investigators including us assumed that TFMPP must be a 5-HT1B agonist. Initially, we thought that animals trained to discriminate TFMPP from vehicle might serve to further characterize 5-HT1B-mediated actions. Using the drug discrimination paradigm with animals trained to discriminate TFMPP, we conducted SAR studies. SAFIR studies were also performed using radioligand binding data. Later, we and others came to the conclusion that TFMPP is a nonselective 5-HT agonist.
Development of 5-HT3 ligands and use of drug discrimination to study a 5-HT3 stimulus.
The next population of 5-HT receptors to be identified was the 5-HT3 receptors. [Interestingly, one of the 5-HTQ first 5-HT3 antagonists was MDL-72222; it now became clear why we were unable to block the DOM stimulus with this agent several years before.] The first 5-HT3 agonist to see widespread application was 2-methyl-5-hydroxytryptamine, or 2-Me 5-HT. Even though it displayed only modest affinity for 5-HT3 receptors (Ki ca. 1000 nM), its affinity was roughly comparable to that of 5-HT itself. We set out to develop a higher affinity 5-HT3-selective agonist. We reasoned that because 5-HT3 receptors were the only population of 5-HT receptors directly coupled to an ion channel, they might accommodate quaternary ligands much in the same way that other ion channel receptors do. We synthesized the quaternary amine derivative of 5-HT, N,N,N-trimethylserotonin, or 5-HTQ, and found it to bind with higher affinity than 5- HT, and to be quite selective for 5-HT3 receptors.
At the same time, we wished to establish a 5-HT3 discriminative stimulus. Choice of an appropriate agonist was limited. 5-HTQ would likely be unable to penetrate the blood-brain barrier. Phenylbiguanide had been just introduced as a 5-HT3 agonist, and we attempted a few trial studies. Eventually, we settled on 2- Me 5-HT as a training drug. It was uncertain whether 2-Me 5-HT would serve as a training drug because it should have been difficult, although not as much as 5-HTQ, in penetrating the blood-brain barrier. We were ultimately successful in training a small group of animals. As expected, the 5-HT3 stimulus generalized to 5-HT3 agonists. The 2-Me 5-HT stimulus was very potently blocked by several 5-HT3 antagonists, but not by a quaternary 5-HT3 antagonist. On the basis of these and other studies, it would seem that these were the first animals to be trained to recognize a 5-HT3 agonist, and that the stimulus was central, not peripheral, in origin.
We were intrigued by the novel 5-HT3 agonist phenylbiguanide. It seemed quite selective for 5-HT3 receptors, but displayed rather low affinity (Ki >1,000 nM). In a prior study with Dr. S. Peroutka, we had investigated the SAFIR of various arylpiperazines at 5-HT3 receptors. Arylpiperazines, as mentioned earlier, are relatively nonselective agents; however, many bind at 5-HT3 receptors with significantly higher affinity that phenylbiguanide. We identified some structural similarities between the arylpiperazines and phenylbiguanide and, in collaboration with Milt Teitler, made a series of hybrid analogs that we hoped would bind with higher affinity than phenylbiguanide. Two such analogs were meta- chlorophenylbiguanide (mCPBG) and 2-naphthylbiguanide (Ki = 10-20 nM); both displayed significantly higher affinity than phenylbiguanide. Although we reported these compounds in abstract form, a full paper MD-354 on mCPBG independently appeared by another group of investigators at the same time. It was not until a few years later that we finally published a full paper on these agents. However, in the course of our studies, we identified a novel class of 5-HT3 agonists: the arylguanides. MD-354, for example, was found to bind at 5-HT3 receptors with high affinity (Ki ca. 35 nM) and to display agonist actions in several assay systems.
Development of 5-HT1D agonists.
For a short period of time in the mid to early 1980s, it seemed that each month would herald the identification of a new population of 5-HT receptors. One of the next populations to some along was 5-HT 1D. We began with the formulation of SAFIR for 5-HT1D binding, and focussed, at least initially, on tryptamine derivatives. We were asked to prepare a review article for Drug News and Perspectives and titled the article: "5-HT1D Receptors: A Serotonin Receptor Population for he 1990s". This turned out to be more prophetic than we had thought. Today, 5-HT1D receptors are an important area of research, especially with respect to the treatment of migraine.
We began a collaboration with Allelix Biopharmaceuticals (Toronto) to develop a new 5-HT1D agonist that lacked affinity for 5-HT1A receptors. At the time, it was thought that the 5-HT 1A actions of sumatriptan might be responsible for some of its cardiovascular side effects. On the basis of our earlier SAFIR work, and using a calf 5-HT1D receptor preparation, we felt that extension of a 5-position substituent on a tryptamine scaffold should ultimately result in the desired selectivity. That is, both receptor populations can accommodate bulk at the tryptamine 5-position but, the two receptors being different, it seemed logical that these two regions of bulk tolerance should be different. Indeed, we were NOT (ALX-1423) successful in identifying NOT, or 5-(nonyloxy)tryptamine. We found that as the length of the 5-position substituent increased, 5-HT1D selectivity began to increase. Unfortunately, affinity at both populations began to decrease as the alkyl chain was extended. A nonyloxy substituent was found to be optimal. NOT was found to bind with high affinity and with 100-fold selectivity for 5-HT1D versus 5-HT1A receptors.
Other 5-position substituents were examined, including arylalkyloxy derivatives, and analogs were prepared that displayed > 400-fold selectivity for 5-HT1D versus 5-HT1A receptors. We also found that the 5- position oxygen atom is not required for 5-HT1D binding and that shorter chains are tolerated in the absence of the oxygen atom.
Aryloxyalkylamines and Benzylimidazolines as 5-HT1D ligands.
Because many 5-HT1D ligands are derivatives of tryptamine, it was of interest to develop some Propranolol, ALX-1355, ALX-1462 nontryptaminergic agents. Aryloxyalkylamines such as propranolol and pindolol had been demonstrated to bind at 5-HT1B receptors, but displayed low affinity (Ki > 10,000 nM) at 5-HT1D receptors. With the appropriate structural modifications, we found that we could significantly enhance the affinity of the aryloxyalkylamines for 5-HT1D receptors. For example, ALX-1355 binds with a Ki of about 30 nM. Because these aryloxyalkylamines are conformationally flexible, we also examined various conformationally-constrained analogs and found that the preferred conformation for 5-HT1D binding is best represented by ALX-1462 (Ki = 10 nM).
We also pursued a lead that was originally described by Dan Hoyer and co-workers. They had reported that the a-adrenoceptor agonist oxymetazoline is a 5-HT1D agonist that displays nanomolar affinity for calf brain 5-HT1D receptors. We thought that with the appropriate structural modifications we might be able to develop a novel class of 5-HT1D ligands based on the structure of oxymetazoline. That is, we wanted to Imidazoline Derivatives remove those structural features that contribute to a-adrenergic binding and retain (or introduce) structural features important for 5-HT1D binding. Our initial results suggested that this could be accomplished to some extent, but all derivatives prepared and examined displayed lower affinity than oxymetazoline itself. As luck would have it, the 5-HT1D receptor was cloned and eventually shown to exist as to subpopulations: 5-HT1Da (now h5-HT1D) and 5-HT1Db (now h5HT-1B). The oxymetazoline analogs were reexamined at the two subpopulations. It was quite discouraging when the initial results indicated that oxymetazoline binds at the two populations with high affinity (Ki approx. 0.4 nM) but with no selectivity. However, it was soon realized that the two populations enjoy distinct binding requirements. That is, certain of the oxymetazoline-related imidazoline derivatives displayed > 100-fold selectivity for h5HT=1D versus h5-HT1B receptors. At the time, these compounds represented the first h5-HT1D versus h5-HT1B - selective agonists described.
