Abstract: "Magic shotgun" (multiple drug targets) and "magic bullet" (single drug target) are two contrasting philosophies for drug development. For polygenic, multi-factorial, and complex diseases, the shotgun approach (defined as "selectively non-selective") may work better than the "magic bullet."
Development of novel therapeutic entities with which to treat disorders of the central nervous system (CNS) that are both more effective and more specific, poses significant challenges to the drug discovery industry. The normal focus of drug research is the search for a “magic bullet,” which acts on a specific protein (or receptor), ideally with no other interactions with other proteins. These are termed “clean” drugs, as they have a single action with few side effects. However, most common CNS disorders are highly polygenic in nature, i.e., they are controlled by complex interactions between numerous gene products. As such, these conditions do not exhibit the single gene defect basis that is so attractive for the development of highly-specific drugs largely free of major undesirable side effects (”the magic bullet”). Secondly, the exact nature of the interactions that occur between the numerous gene products typically involved in CNS disorders remain elusive, and the biological mechanisms underlying mental illnesses are poorly understood.
In the absence of a complete understanding of these mechanisms, and without appropriate animal models within which to study them in detail, it seems unlikely that efforts to develop novel drugs that treat mental illness via highly specific interactions with single target molecules will be productive. For this reason, single drug treatments for these disorders may require “dirty” drugs, i.e., those that work via interactions with a number of receptors, also termed “selectively non-selective” drugs (i.e., drugs that are broadly selective for a class of receptors, e.g., for amines, but are non-selective for a particular amine). A good example of this type of drug, are the psychoactive drugs as the rest of this article will attempt to illustrate.
Despite the barriers alluded to above, highly effective psychoactive drugs have been discovered and developed, albeit largely by serendipitous means. For example, the blockbuster drug chlorpromazine (Thorazine) was developed not as an antipsychotic, but as a highly sedative antihistamine, as a direct result of the off-label use of closely-related antihistamines as an aid to anesthesia. Yet another antihistamine that was hoped to have antipsychotic activity, imipramine (Tofranil), turned out to worsen psychotic symptoms in some patients, but was subsequently found to be a powerful antidepressant. Iproniazid, an antibiotic introduced as a treatment for tuberculosis, was also found to be an antidepressant, one that would spawn a new class of antidepressants known as the monoamine oxidase inhibitors.
In their review article, Roth et al., (2004) propose that the most effective new treatments for mental disorders will, like their predecessors, interact with a variety of target molecules (as opposed to single molecular targets), but will do so in a much more defined “selectively non-selective” manner.
A precedent for this idea clearly already exists, in that, of the psychoactive drugs developed thus far, the most effective in treating mental disorders have proven to be not specific. Furthermore, selective serotonin reuptake inhibitors (SSRIs), the most specific drugs currently available for the treatment of depression, are only marginally better treatments than less specific drugs with less well-defined mechanisms of action, and no single SSRI has proven effective for all patients.
As a rule, psychoactive drugs have very complex pharmacological profiles, interacting with a wide variety of biogenic receptors. The “selectively non-selective” approach presented by Roth et al. (2004) argues that, although poorly understood, polygenic disease mechanisms may very well preclude the development of highly-specific single-target drug treatments. This doesn’t rule out the development of drugs, which have only minimal interactions with the receptors that have been identified to generate undesirable side effects.
A case in point is the antipsychotic drug, clozapine, marketed by the name of Clozaril, and the first anti-psychotic drug to be characterized as an atypical antipsychotic. Originally discovered nearly fifty years ago, during a time when the first blockbuster psychoactive drugs were discovered, clozapine remains in use today, and is still considered one of the best atypical antipsychotic medications available to physicians. Clozapine was characterized as “atypical” in that although it could control delusions and hallucinations just as effectively as its “typical” counterparts (for example, chlorpromazine, fluphenazine, and haloperidol), unlike them, clozapine only rarely produced Parkinson’s like symptoms or tardive dyskinesia (the impairment of voluntary movement, normally a side effect of antipsychotic drug treatment), and was less likely to cause apathy and social withdrawal. Its clinical efficacy is still considered superior as a treatment for schizophrenia even today, but serious side effects, which include agranulocytosis (lesions of the throat and other mucous membranes, such as the gastrointestinal tract as well as the skin, due to a marked decrease of circulating white blood cells) and concomitant infections, seizures, weight gain and even diabetes, mean that its use is reserved for only very serious cases of schizophrenia.
Characteristic of antipsychotic drugs, clozapine exhibits a highly complex pharmacology, and when profiles of interactions between different antipsychotic drugs and biogenic receptors are compared, correlations between receptor interactions and efficacy versus side effects emerge. It is this sort of information that will drive the development of selectively non-selective drugs.
Systematic analysis of the pharmacology of both typical and atypical antipsychotic drugs has indicated that the latter are distinguished from the former by their high affinity for 5HT2A receptors (S2) relative to their affinity for dopamine class 2 (D2) receptors. This observation defined the S2/D2 definition of atypicality. As a group, it was also noted that S2/D2 atypical antipsychotic drugs were associated with both weight gain and diabetes. Recent evidence has implicated the histamine H1 receptor, 5HT2C receptor and a1 adrenoreceptors with weight gain. Therefore, in using receptor pharmacology to screen drug libraries for a potentially novel atypical antipsychotic that is not associated with significant weight gain, one would look for drugs with an S2/D2 signature, but without high-affinity histamine H1 receptor, 5HT2C receptor and a1 adrenoreceptor interactions. Indeed, this rationale is strengthened by data indicating that other CNS medications that induce weight gain are also characterized by their high affinity interactions with H1 receptor, 5HT2C receptor and a1 adrenoreceptors.
Going back to clozapine, it was this very same approach that first led to the development of other atypical antipsychotic drugs. Whereas clozapine was shown to bind relatively well to D4 dopamine receptors, typical antipsychotics were found to bind more strongly to D2 receptors. Following the advances in our knowledge of receptor-binding patterns, these patterns in potentially novel antipsychotics were compared to those of clozapine. In turn, pharmacologists found novel drug candidates that shared clozapine’s pattern of binding to both serotonin and dopamine receptors, some of which also showed less toxicity towards white blood cells.
The first atypical antipsychotics to be spawned by this line of research were risperidone (Risperdal) and olanzapine (Zyprexa), both highly successful antipsychotic drugs. Clearly, the same sort of approach could be used to identify potentially useful drugs for other mental disorders, while simultaneously selecting against candidates that exhibit receptor pharmacologies associated with specific undesirable side effects.
The idea that “less dirty” antipsychotic drugs may well be more effective than “clean” ones may also apply to therapeutic agents with which to treat depression. For example, electroconvulsive therapy, a very effective treatment for depression, is known to dramatically alter the dynamics of countless neurotransmitters, neuromodulators, intracellular signal transduction mechanisms, and neuronal mitogens. Recent evidence suggests that the effects of antidepressants upon signal transduction and mitogenesis directly influence the effects of these medications upon mood. Furthermore, antidepressants that have more complex actions have clearly been shown to exhibit superior efficacy. Taken together, these observations indicate that the most effective antidepressants will be those treatments that retain considerable plasticity in their receptor interactions, and thereby overall efficacy, while being largely free of high affinity interactions with receptors that specifically mediate undesirable side effects.
Derivation of these selectively non-selective therapeutic entities poses some interesting challenges to the pharmaceutical industry. Conventional wisdom dictates that the best treatments for diseases with well-characterized mechanisms will be those that interact with a single molecule intimately involved in the pathogenic mechanism. Therefore, current drug discovery procedures are not well suited to the task of discovering therapeutic agents that work most effectively when they interact with multiple targets. However, the rapidly evolving field of genomics, offers an attractive and more effective alternative. Just as profiles of receptor interactions can be indicative of both efficacy and undesirable side effects, it may be possible to identify specific gene expression signatures that paint equally informative pictures of therapeutic candidates. Moreover, patterns of gene expression triggered by effective and well-characterized drugs already in use would greatly assist in annotating which changes in gene expression profiles are associated with efficacy versus undesirable side effects. Agents that have too many undesirable receptor interactions would be rejected on the basis of their receptor pharmacology, and candidates with gene expression signatures indicative of serious side effects.
Put another way, the point of such a selection strategy would be to first select against drug candidates on the basis of their known negative effects, and then positively select for those candidates with gene expression signatures most strongly associated with efficacy, effectively ignoring drug-induced changes in gene expression of unknown significance. This represents a major shift away from the “single drug:target interaction” paradigm so prevalent within the drug discovery industry.
A second approach would involve high throughput behavioral screening, in which libraries of compounds enriched for activity within the CNS, are screened in a semi-automated fashion using broad measures of behavioral response to identify compounds capable of modifying CNS activity at the whole organism level. Such an approach would seem far more relevant to the discovery of behavior and mood-altering drugs than would traditional approaches centered upon the identification of specific interactions between a given drug and a single receptor target.
To summarize, the current (but in many cases still historic) treatments of complex multi-factorial disorders such as schizophrenia have a mode of action based on a “magic shotgun” (selectively non-selective) approach rather than a “magic bullet” (single drug with one mode of action) seemingly favored by the drug industry. This type of approach is likely to prove most effective, particularly if the drugs can be made to reduce the unwanted side effects, but still maintain the wanted “scattergun” effects. An analogy to this type of treatment is the equivalent of the multi-drug regimen for AIDS being condensed into a single drug, which has the same effects.
References and Further Readings
Roth BL, Sheffler DJ, Kroeze WK. Magic shotguns versus magic bullets: selectively non-selective drugs for mood disorders and schizophrenia. Nature Reviews Drug Discovery 3(4):353-359, 2004.
[Discovery Medicine, 4(23):299-302, 2004]