Perspectives

Home , Atypical antipsychotic, Fananserin, Panamesine, Zotepine

PERSPECTIVES

the use of genetically engineered mice to OPINION provide validated models for drug discovery efforts precarious6. Another fundamental difficulty with Magic shotguns versus magic bullets: developing novel CNS therapeutics is the appreciation that the most widely prescribed selectively non-selective drugs for CNS medications, especially those for mood disorders (for example, lithium, anticonvul- sants and antidepressants7) and schizophrenia8 mood disorders and schizophrenia have complex and ill-defined mechanisms of action. As will be summarized below, the dis- Bryan L. Roth, Douglas J. Sheffler and Wesley K. Kroeze covery that the most clinically effective CNS drugs are pharmacologically complex, with Most common central nervous system disorders and schizophrenia — has become pleiotypic actions (that is, they act as ‘magic disorders — such as depression, bipolar one of the most profitable sectors of the phar- shotguns’), has made the development of disorder and schizophrenia — seem to be maceutical market. CNS drugs account for 11 ‘magic bullets’ (that is, drugs selective for a polygenic in origin, and the most effective of the top 25 drugs on the US market, with single molecular target) less likely. medications have exceedingly complex annual US sales in excess of US $17 billion pharmacologies. Attempts to develop (2002 sales figures). Additionally, it is now Why ‘dirty’ drugs might be better more effective treatments for diseases widely accepted that moderately effective Even though clozapine was discovered nearly such as schizophrenia and depression by treatments exist for most of the common 50 years ago3, it remains the ‘gold standard’ discovering drugs selective for single CNS diseases, including schizophrenia, atypical antipsychotic drug because of the molecular targets (that is, ‘magic bullets’) depression, anxiety disorders, insomnia, absence of debilitating extra-pyramidal side- have, not surprisingly, been largely migraine headaches, chronic pain and seizure effects and demonstrated clinical superiority unsuccessful. Here we propose that disorders. This article will focus on the two in treating schizophrenia9 and in reducing designing selectively non-selective drugs largest sectors of the CNS drug market: suicidality10.However, clozapine is also (that is, ‘magic shotguns’) that interact with atypical antipsychotic drugs for schizophrenia associated with severe and potentially life- several molecular targets will lead to new and related disorders, and antidepressants for threatening side effects, including an increased and more effective medications for a variety depression and anxiety disorders. risk of agranulocytosis, seizures, weight gain of central nervous system disorders. Despite the acknowledged potential of and diabetes, and is therefore typically pre- CNS therapeutic agents, few drugs with truly scribed only for individuals with ‘treatment- Despite their enormous potential to alleviate novel mechanisms of action have been intro- resistant’ schizophrenia. Clozapine has a highly human suffering, before the introduction of duced in the past several decades. Indeed, the complex pharmacological profile, with high fluoxetine (Prozac; Eli Lilly) in the late 1980s most widely prescribed drugs — the serotonin- affinity for a number of serotonin (5-HT2A, central nervous system (CNS) therapeutics selective reuptake inhibitors (SSRIs) and 5-HT2C,5-HT6,5-HT7), dopamine (D4), mus- α were not widely embraced as either highly atypical antipsychotic drugs — represent carinic (M1,M2,M3,M4,M5), adrenergic ( 1- α reliable or profitable. This is despite the fact only marginal advances over the prototypes and 2-subtypes) and other biogenic amine that from the 1960s to the 1970s selected areas zimelidine (discovered in 1971 (REF. 2)) and receptors (REF.8 and references cited therein). of CNS drug discovery yielded profitable clozapine (discovered in 1958 (REF. 3)). FIGURE 1 shows the distribution of some of drugs (for example, Valium, Milltown and Because the aetiology of depression and the receptors targeted by clozapine in relation Haldol). This was due, in part, to the lack of schizophrenia is unknown, choosing the to various molecular targets implicated by suitable animal models, disagreements appropriate molecular target for drug dis- genetic studies of schizophrenia (see REFS 4,5 regarding the biological basis of many discovery is especially risky. In terms of aetiology, for details). As can be seen in FIG. 1,many of orders, uncertainty regarding the ultimate it is now widely accepted that the major the genes implicated in the aetiology of schizo- mechanism(s) of action and the clinical mental illnesses are polygenic4,5,with substan- phrenia are found in anatomical loci where ineffectiveness of many CNS medications1. tial environmental and, perhaps, epigenetic they could, directly or indirectly, modulate During the past two decades, CNS drug dis- components. The polygenic and non-genetic glutamatergic and dopaminergic neurotrans- covery — particularly in the areas of mood components of major CNS diseases makes mission in the frontal cortex. Clozapine is

Dopaminergic GABAergic Glutamatergic

α 1

D1 5-HT2A

5-HT2A

α 2

D1 mGluR3 GABA D2 5-HT Glu 2A Glu

D 2 CN NMDA Glia

D3 5-HT1A

Thalmocortical

VTA

Glu D2

GABA D2 D2 CN CN GABA

5-HT2A Striatum

GABA

Figure 1 | Neuronal circuits implicated in schizophrenia aetiology and treatment. Shown is a schematic diagram of the wiring of the frontal cortex, emphasizing inputs into cortical glutamatergic pyramidal neurons. Some of the various molecular targets implicated as risk factors for schizophrenia are γ shown, including calcineurin (CN) -subunit, the D1- and D3-dopamine receptors (D1, D3) and metabotrophic glutamate receptor 3 (mGluR3). Other molecular targets not shown include catechol-O-methyltransferase, reelin, dysbindin, regulator of G-protein signalling-4 and neuregulin (see REFS 4,5 and references cited therein). Receptors at which clozapine is an antagonist are depicted in red and those at which clozapine is a partial agonist are shown in green (see REF. α α 8 and references cited therein) and include the following: 5-HT1A- and 5-HT2A-serotonin, D1-, D2- and D3-dopamine, 1- and 2-adrenoceptors. For the purposes of clarity, not all of the various molecular targets implicated as risk factors for schizophrenia or those occupied by clozapine are shown. For the sake of simplicity, many other neuronal and biochemical interactions are omitted. GABA, γ-aminobutyric acid; GLU, glutamate; mGluR3, metabotrophic glutamate receptor 3; NMDA, N-methyl-D-aspartate; VTA, ventral tegmental dopamine neuron.

thought to normalize glutamatergic and 5-HT1A 5-HT1B dopaminergic neurotransmission in schizo- 5-HT1D phrenia, thereby ameliorating symptoms, via 5-HT1E Atypicality 5-HT2A 5-HT2A complex interactions with a large number of Weight gain 5-HT2C 5-HT2C 5-HT3 molecular targets (FIGS 1,2).These pleiotypic 5-HT5 actions of clozapine are probably responsible 5-HT6 5-HT7 for its exceptionally beneficial actions in D1 Antipsychotic D 9,10 2 D2 schizophrenia and related disorders . efficacy D3 D4 A graphical representation of the relative D α5 affinity values of clozapine and a number of α2A α2B other atypical antipsychotic drugs (aripipra- Adrenergic α2C zole, ziprasidone, zotepine, quetiapine, olanza- α1A Side effects 1B pine, risperidone) and typical antipsychotic M1-muscarinic M2-muscarinic drugs (haloperidol, chlorpromazine) at a por- Muscarinic M3-muscarinic M4-muscarinic tion of the receptorome (that is, that portion of M5-muscarinic γ-A the proteome comprising receptors) is shown γ-B FIG. 2 BZP in .As can be seen, most of the presently NMDA approved atypical antipsychotic drugs have a PCP SERT complex pharmacology, with appreciable NET DAT affinities for a variety of biogenic amine Weight gain H1 H1 receptors. Given the huge potential market for H2 H3 atypical antipsychotic drugs (~US $10 billion H4 EP-3 annually), great effort has been devoted to EP-4 CB-1 uncovering the receptors responsible for effec- Ca2+ channel tiveness, for atypicality and for side effects. The NAR a2/b2 NAR a2/b4 idea has been that if one could design drugs NAR a3/b2 NAR a3/b4 that targeted the appropriate receptors, one NAR a/b2 NAR a7 could develop atypical antipsychotic drugs β 1 1-AR that are more effective than clozapine and β2-AR 10 I1-imidazoline K 100 have fewer side effects. These efforts will be i mGlu1 1,000 mGlu furthered in the future by the precise delin- 10,000 2 mGlu4 eation of the areas of the brain in which the le e e e e e e e e e l e e o n in n in in in n in in o in in z o p o p p p e z p rid z z drug exerts its beneficial effects, as well as char- ra id te rid a a ix a a e na a ip s o e etia nz z th rid x p e m ra Z p u lo io Lo lo h ro rip ip is la C h hio a p acterization of the intracellular biochemical A Z R Q O T T H lu rp F hlo pathways contributing to both effectiveness C and the development of side effects. Figure 2 | Screening the receptorome reveals multiple molecular targets implicated in

antipsychotic drug actions. The affinity (Ki) values for clozapine and a large number of other biologically active compounds at various receptors can be found at the PDSP K Database (see Further Information); ‘S /D ’ drugs: not quite clozapine i 2 2 the database is part of the National Institute of Mental Health Psychoactive Drug Screening Program (see The first ‘non-clozapine’ atypical to be mar- Further Information) and represents the largest database of its kind in the public domain. At present, the keted was risperidone, which potently blocks PDSP Ki database has >26,000 Ki values for more than 300 receptors. the effects of lysergic acid diethylamide (LSD) in laboratory animals by virtue of its high 11 16 affinity for 5-HT2A receptors .A systematic hypercholesterolaemia .In this regard, recent In addition to improving the core symp- analysis of receptor pharmacology of a studies have implicated the histamine receptor toms of schizophrenia, such as hallucinations α number of typical and atypical antipsychotic H1,the 5-HT2C receptor and 1-adrenoceptors and delusions, atypical antipsychotic drugs as drugs led Meltzer12 and others13 to propose — sites for which many atypical antipsychotic a class modestly improve cognition18, though that the single distinguishing feature of an drugs have high affinity — for causing weight it is unknown how these cognition-enhancing atypical antipsychotic drug was a relatively gain and associated metabolic side effects17. actions are mediated. Current hypotheses high affinity for 5-HT2A relative to D2 receptors S2/D2 atypicals that have relatively low affinity suggest that the cognition-enhancing actions α (the ‘S2/D2 hypothesis of atypicality’). Sub- for H1 receptor and 1-adrenoceptors (for of atypicals may be due to interactions with 19 20 sequently, several atypical antipsychotic drugs example, ziprasidone) are less likely to induce 5-HT2A and 5-HT6 receptors, and to their were introduced that fulfilled the ‘S2/D2’cri- weight gain. Because other CNS medications abilities to enhance prefrontal cortical terion, including olanzapine, ziprasidone, that induce weight gain, such as amitryptiline, dopamine release21 (FIG. 1). zotepine and quetiapine. Although these mirtazepine and imipramine (see PDSP Ki Many attempts, largely unsuccessful, have drugs represent an advance in the treatment of Database, Further Information), also have been made to develop drugs that target the α schizophrenia, none of the presently approved high H1,5-HT2C- and 1-adrenoceptor ‘magic receptor’ responsible for clozapine’s atypical antipsychotic drugs is better than affinities, these data strongly imply that salutary effects in schizophrenia and related 14,15 clozapine for schizophrenia .As a class, the antipsychotic drugs that lack affinities at H1, disorders and, thereby, yield novel atypical α ‘S2/D2 atypicals’ are not without serious side 5-HT2C and 1 receptors will be less likely to antipsychotic drugs. For instance, D4-selective effects, including weight gain and the asso- induce the metabolic side effects of many of compounds22, as well as compounds with 23 ciated metabolic sequelae of diabetes and the presently marketed drugs. 5-HT2A/D4 antagonism ,are ineffective in

postsynaptic antagonist and would therefore Table 1 | Non-D2 dopamine antagonists in schizophrenia are inferior ‘stabilize’ dopaminergic neurotransmission. Receptor Representative Results versus Results versus References compound placebo comparator This notion of ‘anatomical selectivity’ is diffi- cult to reconcile with later findings that 5-HT2A M100907 Better Worse aripiprazole is a D2 partial agonist at post- 5-HT2A/2C SR46349B Better Similar 25 synaptic pituitary D2 receptors in vitro and 5-HT SB271046 In Phase I NA *see footnote 33 6 in vivo . Likewise, the D2 partial agonism

5-HT2A/D4 Fananserin Potentially worse or NA 23 has been inconsistently replicated, with some similar to placebo groups reporting that aripiprazole is a high- 34 D2-dopamine (–)PPP Worse with NA 28 affinity partial agonist in vitro ,and others partial agonist moderate efficacy reporting that aripiprazole is a D2 antagonist 35 36 D3-dopamine (+)-UH232 Worse NA 29 in vivo and in vitro .By contrast, Lawler et al. partial agonist proposed that aripiprazole was ‘functionally D -dopamine SB-277011 In trials NA 52 3 selective’ and that the agonist properties of

D4-dopamine L-745,870 Potentially worse or NA 53 aripiprazole were entirely dependent on the similar to placebo cellular milieu in which it was studied. As a 36 CB1 SR-141716 Worse Worse 25 result, it was proposed that aripiprazole cannabinoid might function as a D2 antagonist, agonist or Sigma BMY-14802 Similar to placebo NA 54 partial agonist depending on the precise Sigma Panamesine Better (small trial) NA 55 complement of D2 receptors and G-proteins in a particular cell. This notion is similar to NK3 SR142801 Better Same 25 the idea of agonist-directed trafficking of NT1 SR48692 Worse Worse 25 receptors (see REF. 37 for a recent review), an Glutamate D-cycloserine Similar to placebo with NA 56,57 38 marginal effect on negative idea originally proposed many years ago . symptoms; may augment This idea proposes that receptors can couple actions of atypical to several signal transducing molecules and antipsychotic drugs suggests that “selective agonists and antago- * See http://science.gsk.com/pipeline/pipeline2002oct.pdf for details. 5-HT, serotonin; NK, neurokinin; NT, neurotensin. nists might be developed which have spe- cific effects on a particular receptor-linked effector system.”38 treating schizophrenia. Likewise, the 5-HT2A schizophrenia, it was found to be effective for When the various competing hypotheses selective compound M100907 (REF.24) failed a short period (~one week), after which clinical of aripiprazole’s actions were tested, we to reduce symptoms to the same extent as efficacy was lost — presumably due to receptor found39, in support of the hypothesis of haloperidol (a typical antipsychotic drug desensitization28.Because (–)PPP has sub- Lawler et al.36, that the actions of aripiprazole comparator) in a multi-centre clinical trial, stantial affinities for non-D2 receptors (for were entirely dependent on the cellular σ α α 39 whereas the 5-HT2A/2C antagonist SR46349B example, -, 2B- or 2C-adrenoceptors; see milieu. Interestingly, we also discovered fared better in comparison with haloperidol25 PDSP Database, Further Information), it is that aripiprazole had a robust pharmacologi- (TABLE 1).There have also been suggestions conceivable that the ineffectiveness of (–) cal profile with partial agonism at several 5-HT 26 that compounds with muscarinic or adreno- PPP was due to unforeseen interaction with (5-HT1A, 5-HT2A, 5-HT2C, 5-HT7) and dop- α α ceptors (both 2- and 1-adrenoceptors) non-D2 receptors. Intriguingly, (+)-UH232 amine (D2,D3,D4) receptors. So although actions might be effective antipsychotic drugs actually worsened psychotic symptoms, per- aripiprazole is clearly a functionally selective but, so far, none have shown efficacy in clinical haps via a combination of D3 and 5-HT2A partial agonist, its complex pharmacology 29 trials (see REF.8 and references cited therein). agonism ,because 5-HT2A agonism is known precludes us from concluding that its bene- TABLE 1 lists a number of other ‘magic bullets’ to exacerbate psychosis62. ficial actions in schizophrenia are due solely which, with few exceptions, were shown to On the basis of these early trials, it was to partial agonism of D2 receptors. It is more lack efficacy in schizophrenia. unclear whether (–)PPP’s lack of efficacy likely that the balance of partial agonism and beyond one week was due to either inadvertent antagonism at a multiplicity of receptors is Partial agonists for schizophrenia interaction with psychotomimetic receptors responsible for its efficacy in schizophrenia σ Some years ago, Carlsson proposed that (–)- (for example, 1-adrenoceptor) or the rela- and related disorders. Taken together, these 3-(3-hydroxyphenyl)-N-n-propylpiperidine tively high intrinsic activity of (–)PPP leading findings have profound implications for

((–)PPP), by virtue of its autoreceptor agonist to desensitization. Several other D2 partial CNS drug discovery, because they imply that properties, might represent a prototype for a agonists, including terguride, OPC-4392, simply developing selective low-efficacy D2 new family of atypical antipsychotic drugs27. pramipexole and SDZ HDC 912, have now partial agonists will not yield effective

Carlsson’s notion was that a partial agonist been tested in schizophrenia, mainly unsuc- antipsychotic drugs, but that D2 partial ago- would normalize or ‘stabilize’ dopaminergic cessfully30.To date, only aripiprazole (OPC nists that functionally interact with various 31 neurotransmission in a way that would be 14597), a weak D2 partial agonist , has shown 5-HT and dopamine receptors might be salutary for both the positive and negative efficacy for schizophrenia32, although con- effective. Therefore functionally non-selective symptoms of schizophrenia. Two potential siderable controversy exists regarding its dopamine agonists might represent a new compounds were subsequently tested: (–)PPP mechanism of action (BOX 1). generation of atypical antipsychotic drugs, 31 and (+)-UH232 (a D3-preferring agonist with Kikuchi et al. originally proposed that with aripiprazole being the first member of 5-HT2A agonism). When (–)PPP was tested in aripiprazole was a presynaptic agonist and a this class.

Non-selective antidepressants? Conceptually, at least two non-conventional that the large-scale, automated and random If ‘dirty’ drugs are better for treating schizo- approaches for discovering ‘magic shotguns’ screening of libraries of compounds enriched phrenia than selective (‘clean’) ones, what can be envisioned: behaviour-based screening for activity at CNS targets, using mainly about other CNS disorders? TABLE 2 lists the and genomic approaches. The first, which behavioural assays, will yield compounds with current classes of antidepressants, ranking has been dubbed ‘HTS’-based behavioural novel and, possibly improved, efficacies for a them by their relative effectiveness in treating screening’45,46,relies on the semi-automated variety of CNS diseases. These approaches depression. The most effective treatment for screening of candidate drugs in broad-based carry with them the advantage of examining depression — electroconvulsive therapy (ECT) behavioural assays. At least two novel anti- responses to drugs at the level of entire organ- — alters the dynamics of a vast number of depressants — YKP10A and INN 00835 — isms, and therefore in the context of their neurotransmitters and neuromodulators, and were discovered using this approach. Neither biological functioning, rather than in overly has profound effects on intracellular signalling drug seems to have appreciable affinity for any simplified experimental systems, for example, pathways related to signal transduction and known antidepressant drug target, including as in isolated in vitro binding studies. mitogenesis7.Indeed, very recent studies imply various biogenic amine receptors and trans- Another approach is a genomic one in that the pleiotypic actions of antidepressants porters47, and both have demonstrated effec- which compounds are screened solely on on signal transduction and neuronal mitogen- tiveness in early-phase trials48,47.It is probable the basis of their abilities to modify the esis are required for the beneficial actions of 40 antidepressants on mood .Likewise, so-called Box 1 | Partial functionally selective agonists for schizophrenia? ‘dual-action’ antidepressants, which inhibit the reuptake of both 5-HT and other biogenic The figure shows three different proposed models of aripiprazole action. In these various amines (for example, dopamine and nor- models, receptors coloured green are activated by aripiprazole and those coloured red are adrenaline), have been shown to be more effec- antagonized; receptors colourd yellow can be partially activated by aripiprazole, depending on tive than ‘single-action’ antidepressants such as the assay conditions. In panel a,the anatomical specificity model is highlighted. This model the SSRIs41,42.As an added benefit, the ‘dual- originally proposed that partial agonists which are effective in treating schizophrenia have differential actions at pre- and postsynaptic D dopamine receptors31.This model proposed that action’ antidepressants also seem to effectively 2 aripiprazole, for instance, is a presynaptic agonist and a postsynaptic antagonist; these dual treat chronic pain43.Finally, recent genetic actions thereby ‘normalize’ dopaminergic signalling. In panel b,the density-dependent partial studies have implied that depression, like agonist model34 is shown schematically. This model, which relies on classical receptor theory, schizophrenia, is a complex disorder with proposes that aripiprazole is a partial agonist whose actions depend solely on the relative density 44 linkage to several genes , including many of D -dopamine receptors. As such, in brain regions in which relatively high concentrations of that converge on the transcription factor 2 D2-dopamine receptors exist, drugs like aripiprazole would be partial agonists, whereas in brain cAMP-responsive element binding protein-1 regions with relatively low concentrations of D2 receptors exist aripiprazole would function as an (CREB1). These results imply that improved antagonist. In panel c,we show our conceptualization of the functional selectivity model. This treatments for mood disorders are likely to arise model proposes that aripiprazole’s partial agonist actions at a variety of G-protein-coupled from drugs with several mechanisms of action, receptors (GPCRs) are dependent on the precise cellular complement of receptors and G-proteins36 and that studies that connect new knowledge and relies on our current understanding of GPCR actions. This model predicts that partial of genetic linkage of these diseases in humans agonists, such as aripiprazole, will have a multiplicity of actions, functioning as agonists, partial with those of changes in gene regulation in dis- agonists or antagonists. Indeed, recent studies indicate that the functional selectivity model best 39 ease, as well as after treatment, are likely to be explains aripiprazole’s actions . helpful in the development of new therapies. a Anatomical b Density-dependent c Functionally selective specificity partial agonism non-selectivity Implications for CNS drug discovery γ Given that selectively non-selective drugs are α 5-HT1A β likely to be more beneficial than single- γ D α action agents in many CNS disorders, how 2 β 5-HT2A best to develop them? Clearly, conventional approaches relying on high-throughput γ D D 5-HT2A α screening (HTS) of cloned human molecular 2 2 β targets and the subsequent optimization of these ‘single-target agents’ is not likely to D D2 D2 2 yield selectively non-selective agents, except, D D2 2 γ α D2 perhaps, by chance. Structure-based drug β γ D α design approaches in which ligands are 2 D2 β D2 γ designed to interact with the correct subset D α of molecular targets are also not likely to be 2 β successful. This is because many of the molecular targets selected have a high degree of structural similarity and designing drugs to target a subset of them is not likely to be successful (see REF. 63 for discussion). The implication of these findings is that the screening of small molecules by non- conventional approaches should be considered.

NATURE REVIEWS | DRUG DISCOVERY VOLUME 3 | APRIL 2004 | 357 © 2004 Nature Publishing Group PERSPECTIVES coordinated expression of gene families. In this gene K+ channel for arrhythmias50 and the drugs — ‘magic shotguns’ — could be dis- approach, compounds with known beneficial 5-HT2B receptor for fenfluramine-like valvular covered by combining behavioural and actions and pleiotypic actions (for example, heart disease51). In addition, it might prove genomics-based screening. Once leads are lithium, clozapine) are screened in vivo and possible to use combinations of compounds to discovered, potential toxicities could be rela- in vitro for their effects on coordinated gene ‘fine-tune’ these gene regulatory signatures. tively easily ‘designed out’ by counter-screening expression. Once gene ‘signatures’ are dis- approaches combined with straight-forward covered, compound libraries are subsequently The end of serendipity? medicinal chemistry approaches. Such magic screened to discover small molecules which, Historically, serendipity has been the driving shotguns, or selectively non-selective drugs, when administered in vitro and in vivo,yield force in the discovery of novel and highly are likely to represent highly effective and similar signatures; such an approach is now effective drugs for CNS disorders. Not sur- novel treatments for major CNS disorders. being undertaken by Psychiatric Genomics, prisingly, the most clinically effective treat- The rational discovery, optimization and Inc.49.Lead compounds can then be optimized ments for depression and schizophrenia, eventual marketing of selectively non-selective to eliminate interactions with potentially toxic and perhaps other disorders, continue to be drugs will end our reliance on serendipity as molecular targets (for example, H1 receptor for the ones with the most nonspecific actions. the driving force for effective drug discovery weight gain17,human ether-a-go-go-related It is likely that selectively non-selective for CNS disorders. On the other hand, in a

Table 2 | Antidepressants with complex modes of action are superior to single-action antidepressants Prototypical Class Mode* Molecular Phase of testing Efficacy vs SSRI Company drug target(s) Electro- Somatic therapy C Undefined In use for decades Greater efficacy58 None convulsive therapy Imipramine Tricyclic C NET, SERT, In use for decades Slight advantage41 Generic antidepressant 5-HT2A, 5-HT2C, α 5-HT6, 1-adrenergic, muscarinic Fluoxetine Serotonin-selective S SERT In use for >10 years N/A Eli Lilly reuptake inhibitor Venlafaxine Dual serotonin/ C NET; SERT In use 10 years Slight advantage41,42,59 Wyeth norepinephrine reuptake inhibitor

Pindolol 5-HT1A partial C 5-HT1A Several double-blind, Combination > than Generic agonist/SSRI placebo controlled clinical SSRI alone in combination trials completed; both uncomplicated drugs approved for use depression60 but not in refractory or chronic depression61 Duloxetine Dual serotonin/ C NET; SERT NDA submitted Unknown; predicted to Eli Lilly norepinephrine be >than SSRIs reuptake inhibitor

Gepirone 5-HT1A partial S 5-HT1A NDA submitted; several Unknown; Glaxo Smith agonist 5-HT1A partial agonists predicted to Kline have been previously be < or = to SSRIs tested as single agent treatments for depression with modest success Memantine NMDA antagonist S NR2B NDA submitted for Unknown; predicted Forrest Alzheimer’s; in use in Europe to be < or = to SSRIs for decades; preclinical research suggests efficacy Aprepitant NK1 antagonist S NK-1 Phase III favourable Discontinued; Merck marginally effective

Agomelatine Melatonin/5-HT C MT-1; 5-HT2B; Phase III favourable Unknown; could be Servier receptor antagonist 5-HT2C superior to SSRIs Nemifitide Unknown C‡ Unknown Phase II/III favourable Unknown; dropped InnaPharma (INN 00835) from further testing due to toxicity SB 723620 CRF1 S CRF1 Phase I Unknown; predicted to Glaxo Smith be < or = to SSRIs Kline SNAP-7941 MCH-1 S MCH-1 Preclinical Unknown; predicted to Synaptic be < or = to SSRIs Pharmaceuticals YKP-10A Unknown C‡ Unknown Phase II favourable Unknown SK Corporation

*C, complex mode of action; S, simple mode of action. ‡Presumed.

358 | APRIL 2004 | VOLUME 3 www.nature.com/reviews/drugdisc © 2004 Nature Publishing Group PERSPECTIVES recent colloquium sponsored by the Collegium 19. Williams, G. V., Rao, S. G. & Goldman-Rakic, P. S. 44. Zubenko, G. et al.Genome-wide linkage survey for The physiological role of 5-HT2A receptors in working genetic loci that influence the development of depressive Internationale Neuro-Psychopharmaco- memory. J. Neurosci. 22, 2843–2854 (2002). disorders in families with recurrent, early-onset, major logicum, Arvid Carlsson has argued that the 20. Woolley, M. L., Bentley, J. C., Sleight, A. J., Marsden, C. A. depression. Am. J. Med Genet. 123B, 1–18 (2003). & Fone, K. C. A role for 5-HT6 receptors in retention of 45. Brunner, D., Nestler, E. & Leahy, E. In need of high- field of CNS drug discovery is likely to con- spatial learning in the Morris water maze. throughput behavioral systems. Drug Discov. Today 7, tinue to depend very much on serendipity. Neuropharmacology 41, 210–219 (2001). S107–S112 (2002). 21. Castner, S. A., Williams, G. V. & Goldman-Rakic, P. S. 46. Bohannon, J. Animal models. Can a mouse be The approaches he suggested to facilitate these Reversal of antipsychotic-induced working memory standardized? Science 298, 2320–2321 (2002). deficits by short-term dopamine D receptor stimulation. 47. Amsterdam, J. D., Brunswick, D. J. & Hundert, M. serendipitous discoveries are very similar to 1 Science 287, 2020–2022 (2000). A single-site, double-blind, placebo-controlled, dose- those outlined above, although we would 22. Bristow, L. J. et al. Schizophrenia and L-745,870, a ranging study of YKP10A — a putative, new novel dopamine D receptor antagonist. Trends antidepressant. Prog. Neuropsychopharmacol. Biol. argue that the results of such searches will not 4 Pharmacol. Sci. 18, 186 (1997). Psychiatry 26, 1333–1338 (2002). be serendipitous at all, but rather the result of 23. Truffinet, P. et al.Placebo-controlled study of the 48. Feighner, J. P. et al. Double-blind, placebo-controlled D /5-HT antagonist fananserin in the treatment of study of INN 00835 (netamiftide) in the treatment of looking for the right thing (magic shotguns) 4 2A schizophrenia. Am. J. Psychiatry 156, 419–425 (1999). outpatients with major depression. Int. Clin. in the right place and at the right time. 24. de Paulis, T. M-100907 (Aventis). Curr. Opin. Investig. Psychopharmacol. 16, 345–352 (2001). Drugs 2, 123–132 (2001). 49. Palfreyman, M. G. et al. Novel directions in antipsychotic Bryan L. Roth, Douglas J. Sheffler and 25. Meltzer, H., Arvantis, L., Bauer, D., Rein, W. & Group, M. S. target identification using gene arrays. Curr. Drug Target Wesley K. Kroeze are in the Department of A placebo-controlled evaluation of four novel CNS Neurol. Disord. 1, 227–238 (2002). compounds for the treatment of schizophrenia and 50. Suessbrich, H., Waldegger, S., Lang, F. & Busch, A. E. Biochemistry, School of Medicine, Case Western schizoaffective disorder. Arch. Gen. Psychiatry (in the Blockade of HERG channels expressed in Xenopus Reserve University, 2109 Adelbert Road, press). oocytes by the histamine receptor antagonists terfenadine Cleveland, Ohio 44106, USA. 26. Rasmussen, T. et al. The muscarinic receptor agonist and astemizole. FEBS Lett. 385, 77–80 (1996). Correspondence to B.L.R. BuTAC, a novel potential antipsychotic, does not impair 51. Rothman, R. B. et al. Evidence for possible involvement learning and memory in mouse passive avoidance. of 5-HT(2B) receptors in the cardiac valvulopathy e-mail: [email protected] Schizophr. Res. 49, 193–201 (2001). associated with fenfluramine and other serotonergic doi:10.1038/nrd1346 27. Clark, D. et al.Is 3-PPP a potential antip sychotic medications. Circulation 102, 2836–2841 (2000). agent? Evidence from animal behavioral studies. 52. Remington, G. & Kapur, S. SB-277011 GlaxoSmithKline. Eur. J. Pharmacol. 83, 131–134 (1982). Curr. Opin. Investig. Drugs 2, 946–949 (2001). 1. Shorter, E. Looking backwards: a possible new path for 28. Lahti, A. C. et al. Antipsychotic properties of the partial 53. Kramer, M. S., Last, B., Getson, A. & Reines, S. A. drug discovery in psychopharmacology. Nature Rev. dopamine agonist (–)-3-(3-hydroxyphenyl)-N-n- The effects of a selective D4 dopamine receptor Drug Discov. 1, 1003–1006 (2002). propylpiperidine(preclamol) in schizophrenia. Biol. antagonist (L-745,870) in acutely psychotic inpatients 2. Carlsson, A. & Wong, D. T. Correction: a note on the Psychiatry 43, 2–11 (1998). with schizophrenia. D4 Dopamine Antagonist Group. discovery of selective serotonin reuptake inhibitors. 29. Lahti, A. C., Weiler, M., Carlsson, A. & Tamminga, C. A. Arch. Gen. Psychiatry 54, 567 (1997). Life Sci. 61, 1203 (1997). σ Effects of the D3 and autoreceptor-preferring dopamine 54. Gewirtz, G. R. et al.BMY14802, a receptor ligand for 3. Hippius, H. A historical perspective of clozapine. J. Clin. antagonist (+)-UH232 in schizophrenia. J. Neural Transm. the treatment of schizophrenia. Psychiatry 60 (Suppl. 12), 22–23 (1999). 105, 719–734 (1998). Neuropsychopharmacology 10, 37–40 (1994). 4. Harrison, P. J. & Owen, M. J. Genes for schizophrenia? 30. Benkert, O., Muller-Siecheneder, F. & Wetzel, H. 55. Muller, M. J. et al. Antipsychotic effects and tolerability of Recent findings and their pathophysiological implications. Dopamine agonists in schizophrenia: a review. Eur. the sigma ligand EMD 57445 (panamesine) and its Lancet 361, 417–419 (2003). Neuropsychopharmacol. 5, S43–S53 (1995). metabolites in acute schizophrenia: an open clinical trial. 5. Lewis, C. M. et al.Genome scan meta-analysis of 31. Kikuchi, T. et al.7-(4-[4-(2,3-Dichlorophenyl)-1- Psychiatry Res. 89, 275–280 (1999). schizophrenia and bipolar disorder, part II: Schizophrenia. piperazinyl]butyloxy)-3,4-dihydro-2(1H)-qui nolinone 56. van Berckel, B. et al.Efficacy and tolerance of Am. J. Hum. Genet. 73, 34–48 (2003). (OPC-14597), a new putative antipsychotic drug with D-cycloserine in drug-free schizophrenic patients. Biol. 6. Setola, V. & Roth, B. L. Why mice are neither miniature both presynaptic dopamine autoreceptor agonistic Psychiatry 40, 1298 (1996). humans nor small rats: a cautionary tale involving activity and postsynaptic D receptor antagonistic 57. Evins, A. E., Amico, E., Posever, T. A., Toker, R. & 5-hydroxytryptamine-6 serotonin receptor species 2 activity. J. Pharmacol. Exp. Ther. 274, 329–336 (1995). Goff, D. C. D-Cycloserine added to risperidone in variants. Mol. Pharmacol. 64, 1277–1278 (2003). 32. Kane, J. M. et al.Efficacy and safety of aripiprazole and patients with primary negative symptoms of 7. Coyle, J. T. & Duman, R. S. Finding the intracellular haloperidol versus placebo in patients with schizophrenia schizophrenia. Schizophr. Res. 56, 19–23 (2002). signaling pathways affected by mood disorder and schizoaffective disorder. J Clin. Psychiatry 63, 58. Efficacy and safety of electroconvulsive therapy in treatments. Neuron 38, 157–160 (2003). 763–771 (2002). depressive disorders: a systematic review and meta- 8. Roth, B., Sheffler, D. J. & Potkin, S. Atypical 33. Inoue, T., Domae, M., Yamada, K. & Furukawa, T. analysis. Lancet 361, 799–808 (2003). antipsychotic drugs: unitary or multiple mechanisms Effects of the novel antipsychotic agent 7-(4-[4-(2,3- 59. Stahl, S. M., Entsuah, R. & Rudolph, R. L. Comparative for ‘atypicality’. Clin. Neurosci. Res. (in the press). dichlorophenyl)-1-piperazinyl]butyloxy)-3,4-dihydro- efficacy between venlafaxine and SSRIs: a pooled 9. Kane, J., Honigfield, G., Singer, J., Meltzer, H. Y. 2(1H)-quinolinone (OPC-14597) on prolactin release from analysis of patients with depression. Biol. Psychiatry 52, Clozapine for the treatment-resistant schizophrenic: the rat anterior pituitary gland. J. Pharmacol. Exp. Ther. 1166–1174 (2002). a double-blind comparison with chlorpromazine. 277, 137–143 (1996). 60. Perez, V., Gilaberte, I., Faries, D., Alvarez, E. & Artigas, F. Arch. Gen. Psychiatry 45, 789–796 (1988). 34. Burris, K. D. et al. Aripiprazole, a novel antipsychotic, is Randomised, double-blind, placebo-controlled trial of 10. Meltzer, H. Y. et al.Clozapine treatment for suicidality in a high-affinity partial agonist at human dopamine D2 pindolol in combination with fluoxetine antidepressant schizophrenia: International Suicide Prevention Trial receptors. J. Pharmacol. Exp. Ther. 302, 381–389 treatment. Lancet 349, 1594–1597 (1997). (InterSePT). Arch. Gen. Psychiatry 60, 82–91 (2003). (2002). 61. Berman, R. M. et al. The use of pindolol with fluoxetine in 11. Colpaert, F. C. Discovering risperidone: the LSD model 35. Hemrick–Luecke, S. et al.Pharmacological activity of the treatment of major depression: final results from a of psychopathology. Nature Rev. Drug Discov. 2, aripiprazole at presynaptic and postsynaptic dopamine double-blind, placebo-controlled trial. Biol. Psychiatry 315–320 (2003). receptors and in antipsychotic models. Proc. Soc. 45, 1170–1177 (1999). 12. Meltzer, H. Y., Matsubara, S. & Lee, J.-C. Classification Neurosci. 894.9 (2002). 62. Nichols, D. E. Hallucinogens. Parmacol. Ther. 101, of typical and atypical antipsychotic drugs on the basis 36. Lawler, C. P. et al.Interactions of the novel antipsychotic 131–181 (2004).

of dopamine D-1, D-2 and serotonin2 pKi values. aripiprazole (OPC-14597) with dopamine and serotonin 63. Kroeze, W. K., Sheffler, D. J. & Roth, B. L. G-protein- J. Pharmacol. Exp. Ther. 251, 238–246 (1989). receptor subtypes. Neuropsychopharmacology 20, coupled receptors at a glance. J. Cell Sci. 116, 13. Altar, C. A., Wasley, A. M., Neale, R. F. & Stone, G. A. 612–627 (1999). 4867–4869 (2003).

Typical and atypical antipsychotic occupancy of D2 and 37. Kenakin, T. Efficacy at G-protein-coupled receptors.

S2 receptors: an autoradiographic analysis in rat brain. Nature Rev. Drug Discov. 1, 103–110 (2002). Acknowledgements Brain Res. Bull. 16, 517–525 (1986). 38. Roth, B. L. & Chuang, D. M. Multiple mechanisms of Work in the authors’ laboratories is supported by the National 14. Leucht, S., Wahlbeck, K., Hamann, J. & Kissling, W. serotonergic signal transduction. Life Sci. 41, Institutes of Health. New generation antipsychotics versus low-potency 1051–1064 (1987). conventional antipsychotics: a systematic review and 39. Shapiro, D. A. et al. Aripiprazole, a novel atypical Competing interest statement meta-analysis. Lancet 361, 1581–1589 (2003). antipsychotic drug with a unique and robust The authors declare competing financial interests; see Web version 15. Tuunainen, A., Wahlbeck, K. & Gilbody, S. Newer atypical pharmacology. Neuropsychopharmacology 28, for details. antipsychotic medication in comparison to clozapine: 1400–1411 (2003). a systematic review of randomized trials. Schizophr. Res. 40. Santarelli, L. et al. Requirement of hippocampal Online links 56, 1–10 (2002). neurogenesis for the behavioral effects of 16. Fontaine, K. R. et al. Estimating the consequences of antidepressants. Science 301, 805–809 (2003). DATABASES anti-psychotic induced weight gain on health and 41. Anderson, I. M. Meta-analytical studies on new The following terms in this article are linked online to: mortality rate. Psychiatry Res. 101, 277–288 (2001). antidepressants. Br. Med. Bull. 57, 161–178 (2001). LocusLink: http://www.ncbi.nlm.nih.gov/LocusLink/ 17. Kroeze, W. K. et al.H-histamine receptor affinity 1 42. Thase, M. E., Entsuah, A. R. & Rudolph, R. L. Remission CREB1 predicts short-term weight gain for typical and atypical rates during treatment with venlafaxine or selective antipsychotic drugs. Neuropsychopharmacology 28, serotonin reuptake inhibitors. Br. J. Psychiatry 178, FURTHER INFORMATION 519–526 (2003). 234–241 (2001). National Institutes of Mental Health Psychoactive Drug 18. Meltzer, H. Y. & McGurk, S. R. The effects of clozapine, 43. Briley, M. New hope in the treatment of painful Screening Program: http://pdsp.cwru.edu/

risperidone, and olanzapine on cognitive function in symptoms in depression. Curr. Opin. Investig. Drugs 4, PDSP Ki Database: http://kidb.cwru.edu/ schizophrenia. Schizophr. Bull. 25, 233–255 (1999). 42–45 (2003). Access to this interactive links box is free online.