Chemical entities might be viewed as a primordial soup of nonselective agents from which selective agents can evolve. The primordial pool might
From: The Receptors: The Serotonin Receptors: From Molecular Pharmacology to Human Therapeutics Edited by: B. L. Roth © Humana Press Inc., Totowa, NJ
already contain some selective agents, but appropriate pharmacological techniques and methodologies must be available (or developed) so that they can be "fished out." At the heart of medicinal chemistry is the design and synthesis of agents with a given pharmacological action. Although this no longer remains the problem it once was, one of the most difficult tasks in medicinal chemistry remains the design of selective agents; that is, once an agent with a given biological function has been identified (a relatively straightforward goal), a subsequent and more difficult task is the introduction of selectivity to, for example, reduce undesirable side effects or develop tools that might be useful as selective agonists, antagonists, or radioligands for pharmacological studies. Difficulties encountered in the development of selective ligands are epitomized in the field of serotonin. In addition to interacting with the serotonin transporter (SERT), serotonin binds at seven major families or populations of receptors (5-hydroxytryptamine; 5-HTj to 5-HT7) (1,2). Multiple subpopulations (and species homologs and splice variants) exist for many of these. With this number of targets, development of an agonist or antagonist ligand with selectivity for one serotonin receptor (sub)population over another represents a rather daunting challenge. Several populations of 5-HT receptors are still without selective agonists and/or antagonists. Indeed, even identification of a novel nonselective structure type can sometimes present challenges.
The purpose of this chapter is not to review those agents that are currently thought to be selective for one population of serotonin receptors over another, or even to describe the various agents that have been investigated for seroton-ergic activity. Rather, the intent is to discuss some of the strategies that we have employed in our attempts to develop "selective" serotonergic agents. What is presented are selected case studies that have emanated from our laboratory; the chapter documents some of our efforts (and problems encountered along the way). The studies should be instructive in that they might eventually lead to serotonergic agents with even greater selectivity. In most instances, the approaches are quite general and also should be applicable to nonserotonergic fields of study.
The field of serotonin research exploded over the past two decades with thousands of articles now appearing each year (nearly 3000 articles were published in the first three quarters of 2004 alone). Certainly, our work has not been conducted in a vacuum. Without question, our work has been enormously dependent on, and impacted by, the results of others. Given limitations of space, we focus primarily on our studies, giving just due to certain other investigators whose key results highly influenced the direction of our work. We apologize in advance for not having the space to present a more thorough review of the contributions others have made to this field or even to provide a more comprehensive account of those investigations that directly or indirectly bear on the studies that are presented here. Early work (covering the literature to 1985) on the medicinal chemistry of serotonergic agents has been reviewed (3). For detailed descriptions of the different populations of 5-HT receptors and their ligands, a number of recent reviews are available (2,4-6).
What does selectivity mean? Also, how selective must an agent be to be termed "selective"? Selectivity refers to the ability of a compound to recognize its target without interacting with other, related targets (7). The selectivity that an agent shows for one target over another is sometimes referred to as the selectivity index. A selectivity index of 1000 to 10,000 or more is considered ideal, but in some instances 50-fold to 100-fold can provide sufficient selectivity to derive a good therapeutic index (7). Thus, selectivity is relative. A drug need not be selective to be useful or therapeutically effective—indeed, most drugs are not; some drugs are even quite nonselective (to wit the antipsychotic agent clozapine). Nevertheless, selective agents make good pharmacological tools, and much is learned along the way to achieving selectivity that can be applied to the formulation of structure-affinity relationships (SAFIR) and/or structure-activity relationships (SAR). The latter information can often be applied to drug design. At this point, it might also be noted that an agent can display binding selectivity and/or functional selectivity. The two concepts are not strictly related. Binding selectivity is usually related to affinity. Functional selectivity, which can be achieved without binding selectivity, is the more complex of the two and is complicated by issues of efficacy; it can sometimes be related to the pharmacological assay employed. For example, an agonist ligand might show some selectivity for one receptor population over another (i.e., it binds with higher affinity at one receptor than another); however, if its efficacy is significantly lower at the first than at the second, it might appear functionally selective for the latter. An extreme case is where an agent binds equally well at two populations of receptors but is an agonist at one and an antagonist at the other; such an agent would appear to be a functionally selective agonist in an in vivo assay even though it possesses no binding selectivity between the two receptor populations. Antagonists can also demonstrate functional selectivity. For example, an antagonist binds equally well at two receptor populations but one population is inaccessible; this might be encountered if one population is located only in the periphery and the other only in the brain, and the antagonist is unable to penetrate the blood-brain barrier. The studies described in this chapter deal almost exclusively with binding selectivity. Our work has been predicated on the need to identify an agent that binds in a selective fashion at one serotonin receptor population over the others prior to any investigation of functional selectivity.
In its most general terms, the overall question to be addressed is How does one go about developing a selective agent in the absence of information necessary to achieve selectivity? With specific application to serotonin, how can selectivity be achieved for one 5-HT receptor population over another? With the discovery of some of the newer 5-HT receptors, little more might be known other than that 5-HT binds at the receptor. Certain other agents might have been examined at the receptor, but these are generally standard, nonselective, structurally unrelated agents used to authenticate the novelty of the receptor. Often, the structures of these latter agents are such that they cannot be related to that of 5-HT or to one another, and they are generally not very useful for subsequent design purposes. Some of the agents examined might even be selective for another 5-HT receptor population and are included in an investigation to show that they do not bind to a newly identified receptor. Because it is known that 5-HT binds at a receptor (i.e., 5-HT represents a unique and naturally occurring template structure, as do other neurotransmitters), there is a temptation to explore derivatives of 5-HT to develop selective serotonergic agents. Although this approach is sometimes successful (and examples will be provided later), it should be realized that the closer a structure approaches that of 5-HT, the greater the likelihood that it will not be selective for one 5-HT receptor population over another; after all, 5-HT itself is a nonselective agent. In contrast, if a close structural analog of 5-HT can be made selective for a specific population of 5-HT receptors, there is reasonable likelihood that it will not bind to nonserotonergic receptors because serotonin itself does not display appreciable affinity for such receptors.
Today, several different approaches exist for developing or identifying novel agents. Perhaps the most useful of these is high-throughput screening (HTS), for which hundreds or thousands of compounds can be relatively quickly screened in an applicable pharmacological assay. However, the method is predicated on the a priori availability of a selective agent(s) or radioligand(s) (or a pharmacological preparation that possesses only one specific population of receptors that is responsive to a nonselective ligand). In other words, useful though the method might be, previously generated information is required for this approach to be most successful. Furthermore, once a "lead" structure has been identified, structural modification is very frequently required to optimize its actions and/or selectivity. Other methods include application of pharmacophore models and utilization of quantitative SAR (QSAR) techniques. Here, too, these approaches presuppose the availability of ligands, or assay data, to which these techniques can be applied. A more recent approach—receptor structure-based drug design, or simply structure-based drug design—is the utilization of three-dimensional graphics models of the various serotonin receptors to design agents that will interact with specifically identified amino acid residues thought to be present in a binding pocket. This approach, coupled with site-directed mutagenesis, offers a powerful means for the development of selective agents. To date, however, not every 5-HT receptor type has been modeled. Furthermore, because of the lack of receptor crystal structures and because of the different modeling techniques being utilized, multiple homology (or de novo) models can be described for a given receptor. So, at the moment, this approach is one of trial and error; models are constructed and ligands are designed, synthesized, and evaluated to determine whether they bind. The resulting information is then used to further modify the model; that is, for the most part, this approach is currently focused as much on verification of various receptor models as it is for drug design purposes. Nevertheless, this technique is still quite young and holds great promise for the future.
In the absence of a "lead" structure or "structural template," it seems that the above methods have not brought us any closer to the desired goal: development of selective agents. When developing a selective ligand, it is first necessary to identify a structural template—typically a nonselective structural template—and to then modify its structure in such a manner so as to achieve enhanced selectivity. Thus, this is actually a two-part problem. If it is specifically desired to develop a selective agonist (or partial agonist, or antagonist), this then becomes a more complicated three-part problem, with efficacy forming the third leg of the triad (vide supra). This process becomes yet more complex when attempting to develop a new clinical entity because pharmacokinetic, metabolic, and other properties need to be considered as well; this is usually less of a problem or concern when developing pharmacological tools. We have employed several different means for template identification and selectivity enhancement, and two of the most productive have been the "deconstruction-reconstruction-elaboration" approach and application of the "standard series" concept. The former relies on the availability of a known nonselective agent that shows some affinity for a given receptor type. The latter approach is applied when there is no known lead structure. Both will be described below.
It should be noted that the selectivity of serotonergic agents has been a temporal phenomenon (8); that is, as the number of 5-HT receptor types has continued to grow, agents once thought selective for a particular population might eventually be shown to bind at one or more of the newer 5-HT receptor subpopulations. This has required the constant re-examination of "selective" ligands at new receptor types once they are identified. Of course, earlier pharmacological findings also need to be re-evaluated or reinterpreted once a so-called "selective" agent is found to no longer possess the originally claimed selectivity. Sometimes, this can present enormous problems. For example, if a particular pharmacological effect (e.g., contraction of a particular isolated tissue preparation) has been classified as being mediated via a specific population of receptors on the basis that the effect was produced (or antagonized) by a so-called selective agent and then this pharmacological effect is used to classify newer agents, considerable effort might be required to correct the literature if the agent originally used to classify the effect is subsequently shown to be nonselective or more selective for a different population of receptors. Indeed, the field of serotonin is booby-trapped with pharmacological data that have eventually required re-evaluation as new receptor populations were identified and as newer, more selective agents have been developed. The interested reader is cautioned to rely on the most recent findings to avoid some of the confusion surrounding the older literature.
Was this article helpful?