l-DOPA remains the most effective treatment for Parkinson's disease (PD). However, long-term administration of l-DOPA is compromised by complications, particularly dyskinesia. Serotonergic type 1A (5-HT1A) receptor agonists and serotonergic type 2A (5-HT2A) receptor antagonists were, until recently, considered to be promising therapies against dyskinesia. However, there have been some recent high-profile failures in clinical trials, notably with sarizotan, and it seems that these classes of drugs may also impair l-DOPA antiparkinsonian efficacy. A simple explanation for the loss of antiparkinsonian benefit might be lack of good selectivity of these compounds for their respective targets, particularly with respect to off-target actions on dopaminergic receptors or poor dose selection in clinical studies. However, such explanations do not hold broadly when considering the actions of all compounds studied to date, whether in animal models or clinical trials. Here, we review 5-HT1A and 5-HT2A receptor function in PD and provide an anatomically based rationale as to why in some instances 5-HT1A- and 5-HT2A-modulating drugs might worsen parkinsonism, in addition to reducing dyskinesia. We propose that, in addition to selectivity for specific receptor subtypes, to target 5-HT1A and 5-HT2A receptors to alleviate dyskinesia, without worsening parkinsonism, it will be necessary to develop compounds that display anatomical selectivity, targeting corticostriatal transmission, while avoiding 5-HT receptors on ascending serotonergic and dopaminergic inputs from the raphe and substantia nigra, respectively.
Parkinson's disease (PD) is caused by a deficiency of dopamine in the striatum, and dopamine replacement therapy using l-DOPA remains, even after more than 40 years of use, the mainstay of PD treatment (Fahn, 2008). However, long-term dopaminergic treatment is complicated by the emergence of motor complications such as dyskinesia (Hely et al., 2005).
Serotonergic type 1A (5-HT1A) receptor agonists have been studied as potential antidyskinetic agents in both preclinical models and clinical trials (Table 1). Their antidyskinetic efficacy has been consistently demonstrated across studies, but was sometimes marred by a reduction of the antiparkinsonian benefit of l-DOPA. Thus, the 5-HT1A partial agonist buspirone effectively alleviated abnormal involuntary movements (AIMs) in the 6-hydroxydopamine-lesioned rat (Dekundy et al., 2007) and dyskinesia in patients with PD (Bonifati et al., 1994). The 5-HT1A agonist R-(+)-8-OH-DPAT reduced dyskinesia in the 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP)-lesioned primate, although there was some indication of loss of antiparkinsonian benefit (Iravani et al., 2006). The 5-HT1A partial agonist tandospirone alleviated dyskinesia in humans, but at the expense of worsening parkinsonism (Kannari et al., 2002). Sarizotan, a full, but nonselective, 5-HT1A receptor agonist also reduced dyskinesia in MPTP-lesioned, parkinsonian macaques, but compromised l-DOPA antiparkinsonian action at higher doses (Grégoire et al., 2009). In humans, the antidyskinetic efficacy of sarizotan was variable across studies and, although sarizotan was able to reduce dyskinesia consistently in phase II studies, it impaired l-DOPA antiparkinsonian action at higher doses (Olanow et al., 2004; Goetz et al., 2007, 2008).
From the above discussion, it is clear that 5-HT1A agonists can have antidyskinetic actions when administered in combination with l-DOPA. However, these are sometimes compromised, although sometimes not, by the loss of the antiparkinsonian benefit of l-DOPA. Why might this be? The most widely accepted hypothesis is that the compounds used to date are not as selective for their targets as was initially claimed. For instance, sarizotan has equipotent partial agonist actions at D2-like receptors and 5-HT1A receptors (Bartoszyk et al., 2004). In a positron emission tomography (PET) study, sarizotan at 20 to 50 mg daily led to similar 5-HT1A and D2-like receptor occupancy (Rabiner et al., 2002). Because sarizotan has relatively low efficacy as a D2 partial agonist (Bruins Slot et al., 2006), its predominant effect may be to oppose D2 activation mediated by l-DOPA-derived dopamine. Such attenuation of D2 stimulation would, of course, be expected to reduce the antiparkinsonian actions of l-DOPA. Thus, loss of efficacy in some studies may reflect a narrow therapeutic window and difficulty keeping within effective dose ranges when moving from preclinical toward more advanced clinical development. However, such an explanation based on nonselectivity for 5-HT targets is not consistent with other data. Thus, in the MPTP-lesioned, parkinsonian primate, although R(+)-8-OH-DPAT alleviates dyskinesia, it seems impossible to separate the dose-response curve for this effect from reducing antiparkinsonian benefit, despite this compound being 400-fold selective over dopaminergic D2-like receptors (Lejeune et al., 1997). The same applies to tandospirone that reduced dyskinesia, but at the expense of worsening parkinsonism (Kannari et al., 2002). Tandospirone exhibits more than 50-fold selectivity at 5-HT1A over D2 receptors (Hamik et al., 1990). These data suggest that the problem is not solely the selectivity of drugs but may reside within the target itself.
Like 5-HT1A agonists, 5-HT2A receptor antagonists have been studied as potential antidyskinetic agents. The nonselective 5-HT2A antagonist clozapine reduced l-DOPA-induced dyskinesia in MPTP-lesioned, parkinsonian nonhuman primates (Visanji et al., 2006) and patients with PD (Durif et al., 2004). However, clozapine may exacerbate parkinsonism (Wolters et al., 1990). The nonselective 5-HT2A antagonist quetiapine also alleviated dyskinesia in the MPTP-lesioned, parkinsonian monkey (Visanji et al., 2006). In humans, quetiapine alleviated dyskinesia (Baron and Dalton, 2003), but also had a deleterious effect on l-DOPA antiparkinsonian action (Reddy et al., 2002).
From the above discussion, it seems clear that 5-HT2A antagonists can alleviate dyskinesia; the question is really whether such action is caused by their activity at 5-HT2A receptors. Indeed, because most of the aforementioned compounds are nonselective 5-HT2A antagonists that bind to other 5-HT and non-5-HT receptors, the specific contribution of antagonizing 5-HT2A receptors in dyskinesia reduction remains unclear. Compounds such as clozapine and quetiapine interact at multiple additional sites, including dopaminergic and adrenergic receptors (Newman-Tancredi and Kleven, 2011). They also exhibit modest agonist properties at 5-HT1A receptors (Newman-Tancredi et al., 1998), so interpretation of their effects is complex. Any of these non-5-HT2A actions could be responsible for the loss of antiparkinsonian efficacy, and it could be argued that even the antidyskinetic actions of these compounds have little to do with 5-HT2A antagonism.
However, a PET study performed in schizophrenic patients demonstrated that at doses used to treat PD motor and nonmotor complications clozapine leads to greater 5-HT2 than D2-like receptor occupancy (Nordström et al., 1995). Although that study did not rule out a contribution of antagonizing D2-like receptors in clozapine effects on dyskinesia and parkinsonism, it certainly identified 5-HT2A receptors as an important determinant of its actions. Likewise, ritanserin, which exhibits a 70-fold selectivity at 5-HT2A over D2 receptors (Leysen et al., 1985), effectively alleviated dyskinesia, but at the expense of worsening parkinsonism (Maertens de Noordhout and Delwaide, 1986; Meco et al., 1988). A PET study performed in schizophrenic patients demonstrated that at the average dose used in the latter study (Meco et al., 1988) D2 receptor occupancy after ritanserin administration was minimal (Wiesel et al., 1994).
In agreement with an importance of selectively antagonizing 5-HT2A receptors over off-target receptors, pimavanserin (ACP-103) effectively alleviated l-DOPA-induced dyskinesia in the MPTP-lesioned, parkinsonian monkey (Vanover et al., 2008) and patients with idiopathic PD (Roberts, 2006; Mills et al., 2008), without impairing l-DOPA antiparkinsonian action. In contrast to the atypical antipsychotics, pimavanserin is a relatively selective 5-HT2A inverse agonist, although it also exhibits affinity at 5-HT2C receptors (Vanover et al., 2006). However, in the case of pimavanserin, it is unclear whether the lack of an effect on l-DOPA antiparkinsonian action is caused by 5-HT2A selectivity or an antagonist action at 5-HT2C receptors, because antagonizing 5-HT2C receptors enhances the rotational behavior induced by the D2 agonist quinpirole in the 6-hydroxydopamine-lesioned rat (Fox and Brotchie, 1996) and might therefore exert an antiparkinsonian effect. The effects of the most selective 5-HT2A antagonists studied so far in PD and PD animal models on dyskinesia and parkinsonism are presented in Table 2.
Understanding the role of 5-HT1A and 5-HT2A receptors in parkinsonism and dyskinesia may identify routes that could leverage the obvious antidyskinetic actions into a useful therapeutic. Recent advances in both pharmacology of 5-HT receptors and understanding of the role these play in the pathophysiology of parkinsonism and dyskinesia may help identify such a route.
There is a lot of literature on the role of 5-HT1A and 5-HT2A receptors in PD and dyskinesia. At the moment, it is generally believed that 5-HT1A receptor agonists alleviate dyskinesia by reducing the nonphysiological release of dopamine, synthesized from l-DOPA, from serotonergic, raphestriatal terminals (Carta et al., 2007). However, reducing corticostriatal glutamate release could underlie an additional antidyskinetic mechanism (Dupre et al., 2011; Ostock et al., 2011). The mechanism by which 5-HT2A receptor antagonists reduce dyskinesia remains to be demonstrated, but 5-HT2A-mediated modulation of nigrostriatal dopamine release (Lucas and Spampinato, 2000) and corticostriatal glutamatergic transmission (Ferguson et al., 2010) could be important mechanisms.
Here, we review two recent anatomical studies (Huot et al., 2010a,b) in MPTP-lesioned, parkinsonian macaques that broadly support these ideas but also provide insight that might explain the mitigated success of 5-HT1A- and 5-HT2A-modulating drugs. Moreover, we propose a hypothesis that, to alleviate l-DOPA-induced dyskinesia with 5-HT1A agonists or 5-HT2A antagonists, without compromising the antiparkinsonian benefit of l-DOPA, it might be necessary to stimulate certain anatomically distinct subpopulations of these receptors, thus raising the need for pharmacological compounds displaying anatomical selectivity.
The Role of 5-HT1A Receptors in Parkinsonism and Dyskinesia
The first study Huot et al. (2010a) examined 5-HT1A receptor levels throughout the forebrain in monkeys that were either normal, MPTP-lesioned, parkinsonian but untreated or MPTP-lesioned, parkinsonian and treated with l-DOPA, either acutely, where antiparkinsonian benefit was not associated with dyskinesia, or chronically, when animals did express dyskinesia. The findings of the study are summarized in Fig. 1, but, in brief, 5-HT1A receptor levels were decreased in the superficial layers of the premotor cortex and increased in the middle layers of both the premotor and motor cortex of all MPTP-lesioned, parkinsonian animals, regardless of their exposure to l-DOPA; increased in the striosomes of the striatum of MPTP-lesioned, parkinsonian monkeys; and increased in the matrix of the caudate nucleus of dyskinetic monkeys compared with MPTP-lesioned, parkinsonian monkeys that had received just a single treatment with l-DOPA and never expressed dyskinesia.
Cortical 5-HT1A receptors are localized both presynaptically and postsynaptically (Barnes and Sharp, 1999). Cortical presynaptic 5-HT1A heteroreceptors decrease the release of neurotransmitters, such as glutamate and dopamine (Calcagno et al., 2006; Carta et al., 2008), whereas postsynaptic 5-HT1A receptors exert an inhibitory, hyperpolarizing effect (Sprouse and Aghajanian, 1988). In the middle layers of both the premotor and motor cortex, 5-HT1A receptors are located postsynaptically on cell bodies of pyramidal neurons (Cruz et al., 2004). In the superficial layers of the premotor cortex, 5-HT1A receptors are located presynaptically on thalamocortical terminals (Palchaudhuri and Flügge, 2005). In the striatum, 5-HT1A mRNA levels are very low (Mengod et al., 1996); 5-HT1A receptors of the striatal matrix are thus localized on the terminals of afferent fibers, i.e., corticostriatal and raphestriatal inputs.
5-HT1A agonists, by stimulating 5-HT1A receptors of the superficial cortical layers, will reduce thalamocortical glutamate release, which, according to the classic basal ganglia model (DeLong and Wichmann, 2007), is likely to reduce dyskinesia, although this also carries the risk of worsening parkinsonism. Stimulating 5-HT1A receptors of the middle cortical layers will result in a reduction of corticostriatal glutamate release (Dupre et al., 2011) and reduce dyskinesia (Ostock et al., 2011). While acting along the direct pathway, this will exert an antidyskinetic effect; along the indirect pathway, this might even enhance antiparkinsonian benefits. Stimulation of 5-HT1A receptors in the matrix could also exert antidyskinetic actions, by reducing glutamate release from corticostriatal terminals. By reducing the release of dopamine produced from l-DOPA in raphestriatal serotonergic terminals, 5-HT1A agonists might also reduce dyskinesia (Carta et al., 2007, 2008), although it is possible, but yet unproven, that the reduced availability of dopamine might compromise the antiparkinsonian benefit.
Thus, to alleviate dyskinesia without worsening parkinsonism, it may be necessary to stimulate selectively 5-HT1A receptors on corticostriatal projections, while avoiding thalamocortical and raphestriatal terminals.
The Role of 5-HT2A Receptors in Parkinsonism and Dyskinesia
A second study by Huot et al. (2010b) examined 5-HT2A receptors levels in a similar manner. The findings of that study are summarized in Fig. 1 and showed that 5-HT2A receptor levels were increased in the middle layers of the motor cortex of l-DOPA-treated MPTP-lesioned, parkinsonian macaques, whether they were treated acutely or chronically, and increased in the caudate nucleus and putamen of dyskinetic animals.
Like 5-HT1A receptors, 5-HT2A receptors are localized both presynaptically and postsynaptically. Upon stimulation, presynaptic 5-HT2A receptors increase the release of glutamate (Aghajanian and Marek, 1999), whereas postsynaptic 5-HT2A receptors enhance N-methyl-d-aspartate receptor-mediated neurotransmission (Neuman and Rahman, 1996). Within the middle layers of the cerebral cortex, 5-HT2A receptors are widely expressed on pyramidal neurons forming the corticostriatal pathway (Willins et al., 1997). Indeed, substantial colocalization of 5-HT2A and 5-HT1A receptors in glutamatergic and GABAergic neurons is observed in rat cortical layers (Santana et al., 2004). The majority of 5-HT2A-positive structures in the striatum are terminals of neurons projecting from either the cortex or the globus pallidus, although a minority are terminals of afferent fibers from the substantia nigra pars compacta (SNc) (Bubser et al., 2001). The elevated 5-HT2A receptor levels in the middle layers of cortex and striatum observed in MPTP-lesioned, parkinsonian macaques with dyskinesia could thus contribute to the overactivity of the corticostriatal pathway and lead to dyskinesia. In addition, 5-HT2A receptors are moderately abundant within the SNc (Pompeiano et al., 1994), and stimulating these receptors increases dopamine outflow in the striatum (Lucas and Spampinato, 2000; Pehek et al., 2006). A trend toward increased 5-HT2A receptor levels in the SNc of MPTP-lesioned, parkinsonian macaques chronically treated with l-DOPA, i.e., dyskinetic, was described previously (Huot et al., 2010b), which could suggest abnormal 5-HT2A-mediated nigrostriatal dopaminergic transmission. This could lead to an exacerbation of pulsatile stimulation of striatal dopaminergic receptors, an important mechanism in the pathophysiology of dyskinesia (Smith et al., 2003).
5-HT2A antagonism would attenuate enhanced corticostriatal glutamatergic neurotransmission and, as discussed above, would reduce dyskinesia, without compromising l-DOPA antiparkinsonian benefit. 5-HT2A antagonism would also reduce the release of dopamine produced from l-DOPA in surviving nigrostriatal terminals, which would also reduce dyskinesia, although this action might impair l-DOPA antiparkinsonian action.
Thus, to alleviate dyskinesia without worsening parkinsonism, it may be necessary to selectively antagonize corticostriatal 5-HT2A receptors, while avoiding those located on nigrostriatal terminals.
Anatomically Selective Targeting of 5-HT1A and 5-HT2A Transmission
The above considerations suggest that it would be desirable to preferentially target 5-HT1A and 5-HT2A receptors in those brain regions that can provide improved therapeutic properties while avoiding interaction with receptors located in other brain regions. Until recently, such brain region-specific targeting seemed unfeasible for widely expressed receptors such as 5-HT1A and 5-HT2A. However, recent advances in the pharmacology of 5-HT receptors have suggested that it might be possible to develop therapeutics that have anatomical selectivity in addition to selectivity for specific receptor subtypes. The molecular basis for this assertion is the observation that receptors such as 5-HT1A are differentially coupled to different intracellular signal transduction mechanisms in different brain regions. For example, 5-HT1A receptors preferentially couple to Gαi3 G-protein subunits in the raphe nuclei, whereas they couple preferentially to Gαo subunits in hippocampus and a combination of G proteins in cortex and hypothalamus (Mannoury la Cour et al., 2006). At the second messenger level, 5-HT1A receptors are coupled to inhibition of adenylyl cyclase in hippocampus, but not in the raphe (Clarke et al., 1996). In contrast, 5-HT1A receptors couple to inhibition of inositol phosphate synthesis in raphe but not in hippocampus (Johnson et al., 1997). In the case of extracellular signal-regulated kinase 1/2, 5-HT1A receptors mediate phosphorylation of this kinase in the cerebral cortex but inhibit it in hippocampus (Chen et al., 2002; Cowen et al., 2005; Buritova et al., 2009).
It is important to note that some agonists preferentially activate specific signaling cascades even if multiple cascades are available to the receptor, a phenomenon known as “biased agonism” or “functional selectivity” (Berg and Clarke, 2006; Kenakin, 2011). It can be postulated that such agonists will display accentuated effects in those brain regions that possess the appropriate “signaling machinery” (Newman-Tancredi, 2011). If the latter alleviates parkinsonism or dyskinesia, then an improved “therapeutic window” could be obtained by the use of such agonists.
A well known example of a compound that exhibits preferential interaction with a subpopulation of 5-HT1A receptors is the partial agonist pindolol, which preferentially occupies somato-dendritic 5-HT1A receptors in the raphe nucleus, as demonstrated by PET brain imaging studies (Martinez et al., 2001). However, pindolol also acts as a β-adrenoreceptor antagonist and therefore only possesses limited receptor selectivity. In contrast, a recently reported and efficacious 5-HT1A receptor agonist, 3-chloro-4-fluorophenyl-[4-fluoro-4-[[(5-methylpyrimidin-2-ylmethyl)amino]methyl]piperidin-1-yl]methanone (F15599), exhibits more than 1000-fold selectivity for 5-HT1A receptors versus all other sites examined (Newman-Tancredi et al., 2009) and displays a distinctive preferential agonist activity at cortical 5-HT1A sites. Indeed, F15599 preferentially activates immediate early gene (c-fos) expression in rat frontal cortex, while eliciting little or no activation in hippocampus or medial or dorsal raphe (Newman-Tancredi et al., 2009). In addition, F15599 stimulated the electrical activity of pyramidal neurons in frontal cortex at low doses (0.2 μg/kg i.v.), whereas much higher doses (8.2 μg/kg i.v.) were necessary to inhibit the electrical activity of dorsal raphe neurons (Lladó-Pelfort et al., 2010). Furthermore, F15599 increased dopamine release in medial prefrontal cortex of freely moving rats at low doses (ED50 0.03 mg/kg i.p.), an effect mediated by postsynaptic 5-HT1A receptors. In contrast, higher doses of F15599 were necessary to inhibit hippocampal 5-HT release (ED50 0.24 mg/kg i.p.), an effect controlled by somatodendritic 5-HT1A receptors (Lladó-Pelfort et al., 2010). The preferential cortical 5-HT1A activation by F15599 is likely related to its biased agonism whereby it possesses a distinctive “signaling fingerprint,” preferentially activating Gαi3 G proteins and potently eliciting extracellular signal-regulated kinase 1/2 phosphorylation (Newman-Tancredi et al., 2009). Taken together, such observations suggest that it is possible to target anatomically distinct and therapeutically relevant subpopulations of cortical 5-HT1A receptors that should mediate antidyskinetic and antiparkinsonian properties while avoiding the activation of 5-HT1A receptors controlling raphestriatal serotonergic projections, which, as discussed above could compromise antiparkinsonian benefit. Nevertheless, it should be pointed out that direct experimental evidence for this hypothesis has not yet been generated, and the precise signaling pathways associated with parkinsonism or dyskinesia remain to be established.
Examples of biased agonism have also been reported at 5-HT2A receptors, where 5-HT, but not N-methyltryptamines, signals via a β-arrestin2/Src/Akt complex for control of 5-hydroxy-l-tryptophan-induced head twitches in mice (Schmid et al., 2008; Schmid and Bohn, 2010). Chemically diverse agonists also distinguish between 5-HT2A receptor-mediated activation of Gq/11 G proteins and calcium mobilization in vitro (Cussac et al., 2008). Such distinct signaling patterns may underlie the propensity of some 5-HT2A receptor agonists to elicit psychotic symptoms and hallucinations by preferentially activating those receptors in specific brain regions. However, although evidence is available for functionally selective 5-HT2A agonists, anatomically specific actions by 5-HT2A receptor antagonists have not yet been reported, so it is unclear whether it is possible to pharmacologically block corticostriatal 5-HT2A receptors, while not interfering with the signaling of those located on nigrostriatal terminals.
5-HT1A- and 5-HT2A-mediated neurotransmission throughout the cortico-basal ganglia-thalamo-cortical circuit is complex. Thus, stimulating 5-HT1A receptors and antagonizing 5-HT2A receptors without anatomical selectivity has potential to both reduce dyskinesia and increase parkinsonism. With respect to 5-HT1A receptors, to alleviate dyskinesia without worsening parkinsonism, it may be necessary to stimulate 5-HT1A receptors located on corticostriatal neurons while avoiding 5-HT1A receptors on thalamocortical neurons and raphestriatal neurons. With respect to 5-HT2A receptors, to alleviate dyskinesia without worsening parkinsonism, it may similarly be necessary to antagonize 5-HT2A receptors located on corticostriatal neurons while, in this case, avoiding nigrostriatal neurons. Targeting receptors located on specific neuronal populations with precision, while avoiding others, could be achieved by using compounds that display an anatomical selectivity. However, such compounds are not currently available to the clinic. Maintaining drug levels within a specific therapeutic window or administering compounds displaying an appropriate level of partial agonist activity might be ways, albeit limited, to achieve anatomical selectivity in clinical settings while molecules displaying the appropriate anatomical selectivity are commercialized.
Wrote or contributed to the writing of the manuscript: Huot, Fox, Newman-Tancredi, and Brotchie.
This work was supported by The Cure Parkinson's Trust and Krembil Neuroscience Fund. PH was supported by fellowships from the Edmond J Safra Philanthropic Foundation, the Parkinson Society Canada, and the Canadian Institutes of Health Research.
S.H.F. has received consultancy and speaker fees from Acadia, Asubio, Merz, Novartis, Teva, and Biovail. J.M.B. has received consultancy fees from, and holds an equity position in, Atuka Ltd. A.N.-T. holds an equity position with Neurolixis Inc. and has received consultancy and speaker fees from BMS, Sunovion, BioWin, and the ESC Group.
Article, publication date, and citation information can be found at http://jpet.aspetjournals.org.
- Parkinson's disease
- serotonergic type 1A
- serotonergic type 2A
- abnormal involuntary movement
- positron emission tomography
- substantia nigra pars compacta
- globus pallidus
- globus pallidus pars externa
- globus pallidus pars interna
- Received May 16, 2011.
- Accepted July 21, 2011.
- Copyright © 2011 by The American Society for Pharmacology and Experimental Therapeutics