Neuroscience 2008—D2 Partial Agonist Antipsychotics and Beyond
In his previous SfN meeting reports, our correspondent Tony Altar assessed cholinergic approaches to schizophrenia and non-cholinergic approaches to treating cognition in the disorder. In this report, he returns to the well-trod ground of dopamine receptor pharmacology.
15 December 2008. The "Wall of D2 Partial Agonists" was a prominent feature of antipsychotic presentations at SfN 2008. The success of aripiprazole (Abilify) in treating schizophrenia, in the maintenance of bipolar disease, and as an adjunct to antidepressant therapy has produced the expected rush of "me-too" approaches based on this prototypical success story. [Disclosure: the author co-led the Otsuka/Bristol Myers-Squibb preclinical development team for aripiprazole (Abilify)]. Some of these new partial D2 agonists, and new D2 antagonists, share other receptor mechanisms, notably partial 5-HT1A agonisms, that define aripiprazole, while others combine other high-affinity mechanisms, including 5-HT6 and α2A adrenergic antagonism and selective serotonin reuptake inhibition (SSRI), with their D2 interaction. Each is likely to have a unique side effect profile, and a different therapeutic window of efficacy versus side effects, but hopefully the drugs will be more diverse than those available today (see Leucht et al., 2008). Even moderate diversity would provide the clinicians with more pharmacological options for schizophrenia, bipolar disease, and possibly depression and anxiety. Notable entries include the following.
Taking a cue from the approval for olanzapine and fluoxetine in treating bipolar depression, and the successes of the dopamine-serotonin stabilizer aripiprazole in bipolar maintenance, Chris Moore and colleagues at Pfizer reported on PF-03800130 (abstract 155.2). It has a 35 percent, intrinsic D2 agonist activity in the mouse gamma butyrolactone model of dopamine autoreceptor activation, similar to that of aripiprazole (43 percent), numbers we are happy with and can relate to. This is probably an optimal range as more efficacious agonists (~>50 percent) will be expected to exacerbate positive symptoms of schizophrenia. Both PF-03800130 and aripiprazole inhibited spontaneous locomotor activity, their model for efficacy against manic symptoms. It would be useful to evaluate this compound for its reversal of chlordiazepoxide-amphetamine hyperactivity, a test that appears to be more selective for bipolar drugs, and to compare these effects with a test that is not bipolar drug-selective, spontaneous behavior. PF-03800130 and aripiprazole inhibited spontaneous behavior with a similar minimum-effective-dose (of ~3 mg/kg, PO, 57 percent and 40 percent inhibition at that dose, for each). Catalepsy induction, an extrapyramidal symptom (EPS)-predictive side effect test, was marginal at 30 mg/kg and significant at higher doses. The 100-fold ratio between lowering activity and inducing catalepsy may bode well for PF-03800130, as a high ratio was an excellent predictor for the favorable EPS profile for aripiprazole (Nakai et al., 2003). PF-03800130 blocked the loss of hippocampal serotonin induced by the SERT-requiring neurotoxin, para-choloroamphetamine, but to a small degree at 10 mg/kg, PO, with complete blockade obtained at 30 mg/kg, PO, supporting a weak SSRI action. Similarly, aripiprazole was orally active at 10 and 30 mg/kg, PO, with complete blockade at 100 mg/kg. This fits with aripiprazole's much lesser affinity for SERT than D2 or 5-HT1A receptors, and suggests that a threefold D2/SERT potency ratio may obtain in vivo for PF-03800130. Thus, unlike aripiprazole, PF-03800130 may show SSRI-like action in vivo at similar, or higher doses, required for its antipsychotic/anti-mania profile in patients. In the forced swim model evaluated by water wheel running, those pathetic and depressed Wistar Kyoto rats ran more, as shown in two studies, at >3 mg/kg of PF-03800130, but the effect was only one-quarter as big as that of the clinical gold standard, and rapidly acting for rats drug, imipramine. Aripiprazole was inactive up to 100 mg/kg, IP. This is suggestive of modest therapeutic efficacy to treat depressive symptoms, but broader behavioral studies of chronic PF-03800130 administration to those sad Kyoto rats seems important at this point.
Larry Fitzgerald and colleagues from Pfizer provided the juicy binding and in vitro efficacy data for their D2 partial agonist/serotonin reuptake inhibitor PF-03800130 (abstract 155.5; see also 155.2). PF-03800130 has properties very similar to the prototypical D2 partial agonist drug, aripiprazole, even down to the numerical similarity: high affinity for D2 (Ki = 6 nM, intrinsic activities = 22 percent, although they obtained 35 percent efficacy based on the in vivo GBL model; see abstract 155.2). Aripiprazole showed 39 percent intrinsic efficacy in vitro and 43 percent in vivo. PF-03800130 binds 200-fold more potently at SERT (Ki = 2 nM) than aripiprazole (400 nM), but more like aripiprazole, it is 50-fold less potent for SERT in functional inhibition assays (IC50 =106 nM vs. 736 nM for aripiprazole). These are the data for other receptors of note: 5-HT2A (Ki = 3 nM), 5-HT1A (Ki = 0.9 nM), and 5-HT2B (Ki =35 nM) with 14 percent, 75 percent, and 8 percent intrinsic activities, respectively, versus a full serotonin agonist.
Pfizer again, with Brian Campbell presenting, showed a low propensity for their D2 partial agonist/SSRI drug, PF-03800130, to produce metabolic syndrome or type 2 diabetes (abstract 155.3). Mice were treated acutely with olanzapine, PF-03800130, or aripiprazole (though it seems like chronic treatment would be more relevant here). PF-03800130 elevated circulating glucose or insulin at only doses of 10 or 100 mg/kg or higher, respectively, which was 33X and 330X higher than the minimally effective dose in mouse locomotion (0.3 mg/kg). Aripiprazole produced a significant increase in glucose with no effect on insulin at 100 mg/kg, which is 1,000X higher than doses that lower mouse locomotion. Not surprising by now, olanzapine produced dose-dependent increases in glucose at both 5 and 10 mg/kg with no effect on insulin; this range is similar to the 3 mg/kg dose that lowers locomotion. These data are consistent with the olanzapine propensity for metabolic syndrome and type 2 diabetes, and the rarity of these effects for aripiprazole, and hopefully, PF-03800130, which may have taken "best of show" at the unofficial "Wall of D2 Partial Agonists" competition.
Zheng Li and colleagues, also from Pfizer, conducted a thorough evaluation of in vivo receptor occupancy by PF-03800130 and aripiprazole, determined with ex vivo binding assay methods, and related these findings to plasma drug levels (abstract 155.6). That's a great approach! In agreement with the above in vitro and in vivo studies, they showed with ex vivo rat striatal and cortical binding assays, that orally dosed PF-03800130 occupied D2 receptors labeled by [3H]-raclopride (2 nM) and SERT labeled with [3H]-citalopram (1 nM) at identical doses and at estimated EC50 concentrations of 16 and 12 nM, respectively, whereas oral aripiprazole occupied D2 at 7 nM and occupied only 53 percent of the SERT sites at the supra-physiological dose of 30 mg/kg. Their elegant calculation of efficacious human plasma concentrations of PF-03800130 at ~250 ng/mL based on D2 EC85 and SERT EC80 levels lead them to predict "a more balanced effect on manic and depressive symptoms" of PF-03800130 versus aripiprazole. Whether PF-03800130 will prove to be more favorable in the balance of mania and depression in bipolar disease may depend on 1) whether the SERT component of this drug will precipitate mania as reported for SSRIs in some bipolar patients, and 2) how much its potent 5-HT1A partial agonism (75 percent efficacy) plays into this relationship, which is another property shared in affinity and efficacy with aripiprazole (Jordan et al., 2002), approved recently for bipolar maintenance. These issues will hopefully be clarified in clinical trials which look promising for PF-03800130.
Julie Brennan and C. M. Pulicicchio of Wyeth, reported on WS-50030, another fairly potent and partial D2 receptor agonist (30 percent intrinsic agonism in the [35S]GTP assay) with almost as potent, selective serotonin reuptake inhibitor (it didn't touch the norepinephrine transporter, the dopamine transporter, or the α2A adrenergic receptor) (abstracts 155.18, and 155.19). They, too, modeled discovery efforts on aripiprazole, which we at Otsuka and others have shown to be a dopamine D2 partial agonist that lessens excessive dopamine activity post-synaptically, while providing some level of presynaptic autoreceptor agonism. WS-50030 looks like aripiprazole in animal models, with a potent inhibition of apomorphine climbing, and is inactive (>10 mg/kg) at inducing catalepsy. It was also active in the olfactory bulb lesion model of hyperactivity in the open field, and produced moderate but significant increases in cortical serotonin release, equal to that of olanzapine. The additional finding of potent SSRI activity, and activity in depression models, suggests to the Wyeth researchers, and to me as well, that it may be a treatment for positive and negative symptoms of schizophrenia, as well as comorbid depression. Its weak (17 and 42 percent) inhibition of conditioned avoidance responding at 3 and 10 mg/kg doses may be a concern, as was the 15 nM affinity at that bad penny of psychopharmacology, α1 adrenergic receptors. Collaboration between Wyeth and Solvay on this series of compounds has been stopped, and work at least for now appears to be limited to the preclinical area.
Adrian Newman-Tancredi of Pierre Fabre fame reported on F16615, a prototype benzamide and putative antipsychotic (abstracts 155.8 and 155.9). It blocks D2/3/4 and human 5-HT2B receptors, and is a potent partial agonist at human 5-HT1A. The role for 5-HT2B antagonism in antipsychotic drugs draws support from the activity at that receptor by many atypical antipsychotics, and is also supported by the finding that the 5-HT2B-/- homozygous deletion mouse shows a major reduction in MDMA (aka, ecstasy) hyperactivity. Tancredi and colleagues report that F16615 blocks MDMA hyperactivity in wild-type mice. The 5-HT2B receptor is an autoreceptor on raphe serotonergic neurons, where it reduces serotonin release. F16615 elevates 5HT release by blocking this site. The compound is active in the CAR test, blocks methylphenidate behavioral responses apomorphine climbing and MDMA hyperactivity, and reduces marble burying behavior, an odd yet useful and highly inexpensive means to predict anxiolytic action. F16615 is seen at Fabre as a promising template for new drugs since potent effects of F16615 are obtained in most of these assays at ~<1 mg/kg doses. F16615 also reverses scopolamine-induced social recognition deficits in rats, purportedly a model of short-term memory, and reduces PCP-induced working and reference spatial memory deficits in the hole-board test. Its lack of the dreaded serotonin syndrome, catalepsy, or effects on glucose levels bode well for the normal concerns about metabolic syndrome, diabetes, or EPS. The absence of effects on corticosterone was a good sign, too, although F16615 elevated serum prolactin at a 0.4 mg/kg PO dose. Presumably this is due to potent D2 antagonism. The compound's failure to induce catalepsy may be a result of the partial 5-HT1A agonism, but shouldn't that attenuate prolactin, too? Maybe it does. The F16615 approach looks novel and promising for the treatment of schizophrenia and possibly related disorders like anxiety and bipolar disease. F16615 appears to be a valuable complement to the selective D3 antagonist approaches that are also in development at Pierre Fabre.
In one of the Pierre Fabre abstracts (155.9) Didier Cussac neatly summarized the new generation of antipsychotics as follows:
"First generation" typical antipsychotics like haloperidol and chlorpromazine potently bind D2/D3/D4 receptors and α1 adrenergic receptors, with little interaction with 5-HT receptors, and, for chlorpromazine and thioridazine, potent affinity for muscarinic M1 and histamine H1 receptors.
"Second generation" atypicals antipsychotics (clozapine, risperidone, olanzapine and ziprasidone) bind D2 and, with similar or somewhat higher affinity, bind 5-HT2A and 5-HT2C; some also bind H1 (clozapine, risperidone and ziprasidone) and muscarinic M1 and M3 receptors (olanzapine, clozapine, and N-desmethyl-clozapine).
"Third generation" partial agonist antipsychotics, as I would define them, applies to partial agonism at dopamine and serotonin receptors, shown by the prototype drug of this class, aripiprazole, and bifeprunox, RGH188 (cariprazine) both of which have increased (low nM) affinity at 5-HT1A receptors and similar affinity at D2 where they are partial agonists.
Sharon Mexal of the nascent Cenomed in Irvine, California, reported on three compounds, CM-2,303, CM-2,232, and CM-2,236 (abstract 155.11). Based on an impressive battery of tests, subcutaneous delivery of the first two of these compounds reverses PCP deficits in PPI, and PCP-induced hyperlocomotion in a 0.3-3 mg/kg range, and, like aripiprazole, disrupts the CAR and passive avoidance tests. But CAR was considered by these presentations to reflect cognitive disruption, which may be so, but a large literature (e.g., Smith et al., 2004) shows CAR to be predictive of antipsychotic efficacy, at least for drugs that antagonize D2 signaling. So if CAR disruption is really undesirable, then these data are favorable, but if it is predictive of efficacy, these CM compounds may face greater clinical hurdles as this activity was obtained at doses above those needed to block PCP effects. CM-2,303 and CM-2,232 were less potent in the PCP tests than in tests for their disruption of open field behavior (a sedation measure) or rotarod performance. None of these compounds induced catalepsy at up to 10 mg/kg, but again, higher doses, particularly for CM 2,232, would be worth evaluating.
As reported by Sarah Holguin, in another Cenomed poster (155.12), the CM compounds have agonistic activity at D2 and D3 receptors, but, similar to aripiprazole, they are also 5-HT1A agonists. The question is how "strong" and how potent are the compounds at these three sites? Too much agonism at D2 can exacerbate positive symptoms of schizophrenia, while too much agonism at 5-HT1A can invoke the serotonin syndrome. The CM compounds were all potent (1-14 nM) 5-HT1A agonists, but had quite varying affinities, and agonism/antagonism profiles, at D2s, D2 long form, D3, D4, and α1 adrenergic receptors. Their lead compound, CM-2,303, is a partial but high (64 percent) efficacy D2 agonist, but weak at that site (230 nM Ki), and is far more potent as a D4 and α1 antagonist (5 nM), and a full D3 agonist (Ki = 58 nM). It also antagonized H1 receptors. The clinical consequences of such a mixed D2/3/4 profile with full 5-HT1A agonism and α1 + H1 blockade (sedation) is far from clear. Compared to the subcutaneous route (see abstract 155.11), the oral delivery of these compounds seemed to increase the rat therapeutic ratio to five- or 30-fold for disrupting activities (CAR, hypo-locomotion, or rotorod) versus enhancing activities (reversal of PPI deficits).
Stacey Rizzo of Wyeth showed that, like PF-16615, SLV-313 is a "dual acting" D2 receptor antagonist/full 5-HT1A agonist (abstract 155.16). Their behavioral studies showed that it improves glutamatergic-induced memory impairment, and anxiolytic- and antidepressant-like effects in rodent models. Its full agonism and 1 nM affinity at 5-HT1A receptors, and induction of serotonin syndrome in rats at 1-10 mg/kg (McCreary et al., 2007), however, is a concern. Another poster from Wyeth and their Solvay associates, presented by Shendi Lu of Wyeth, showed that a 10 mg/kg dose of SLV-313 was needed to occupy 70 percent of 5-HT1A receptors in rat hippocampus and frontal cortex (abstract 155.17), two places one would want a partial 5-HT1A agonist to bind, based on a wealth of behavioral studies. Essentially, all D2 sites in striatum were occupied and with greater potency (1 mg/kg), however. While it is true in rodent models that 5-HT1A receptor stimulation can modulate D2 receptor blockade to reduce EPS liability, it was unclear whether one could expect the 5-HT1A full agonism of SLV-313 to minimize EPS or provide antidepressant activity in treatment-resistant depression or in the treatment of negative symptoms associated with schizophrenia.
Cariprazine (RGH-188) is a purported D3 receptor-preferring dopamine D3/D2 receptor partial agonist, and is under development by the Hungarian company Gedeon Richter for schizophrenia and acute mania, based on prior work showing its properties as a dopamine stabilizer, antipsychotic efficacy in rodent models, without EPS liability, and with antidepressant-like and anxiolytic-like properties. Presenter Istvan Gyertyan (abstract 156.4) built upon the anxiolytic aspects, by showing that it has rapid (one hour) antidepressant activity against chronic mild stress relative to reference anxiolytics, based on partial efficacy in the shock suppressed water drinking, and equivalent efficacy relative to anxiolytics in suppressing ultrasonic vocalizations.—C. Anthony Altar.