Introduction
Current treatment options for bipolar affective disorder (BP) are unsatisfactory for a substantial proportion of sufferers. Many are symptomatically ill for almost half of their lives (Judd et al. Reference Judd, Akiskal, Schettler, Endicott, Maser, Solomon, Leon, Rice and Keller2002) and others experience frequent relapses (Gitlin et al. Reference Gitlin, Swendsen, Heller and Hammen1995). Fifty per cent of patients fail to respond to lithium, and of those that do, 50–90% are symptomatic within 1 year (Licht, Reference Licht1998). Moreover, many drugs licensed for BP have significant adverse drug reactions (ADRs), with potential negative consequences on long-term physical health that may adversely affect compliance and thereby mental state. Thus, maintaining patients on effective, well-tolerated medication regimes is a clinical priority.
The development of new drugs for the treatment of BP is dependent on the identification and subsequent validation of novel, more effective targets. Although in vitro testing of disease-relevant tissue and the use of animal models can be helpful in the identification stage, validation of drug targets ultimately relies on the time-consuming and expensive drug discovery pipeline that ends with clinical trials. This process leads to a high rate of attrition; that is the drug is identified as unsuitable for clinical use. Indeed, the total cost of bringing a new drug to the market is now estimated at over 800 million US dollars (DiMasi et al. Reference DiMasi, Hansen and Grabowski2003). An attractive alternative is to identify which potential drug targets are linked to the pathophysiology of the disorder before they are trialled in the clinic. Until recently, identification of the aetiological genetic variation associated with BP has represented a significant challenge because, in linkage or association studies, the results have been inconsistent and rarely replicated (Kato, Reference Kato2007).
More consistent findings have been shown in three recently published large genome-wide association studies of bipolar disorder: the Wellcome Trust Case Control Consortium (WTCCC, 2007), Baum et al. (Reference Baum, Akula, Cabanero, Cardona, Corona, Klemens, Schulze, Cichon, Rietschel, Nothen, Georgi, Schumacher, Schwarz, Abou, Hofels, Propping, Satagopan, tera-Wadleigh, Hardy and McMahon2008) and Sklar et al. (Reference Sklar, Smoller, Fan, Ferreira, Perlis, Chambert, Nimgaonkar, McQueen, Faraone, Kirby, de Bakker, Ogdie, Thase, Sachs, Todd-Brown, Gabriel, Sougnez, Gates, Blumenstiel, Defelice, Ardlie, Franklin, Muir, McGhee, MacIntyre, McLean, VanBeck, McQuillin, Bass, Robinson, Lawrence, Anjorin, Curtis, Scolnick, Daly, Blackwood, Gurling and Purcell2008). In our own reanalysis of the publicly available WTCCC data (www.wtccc.org.uk), applying a less stringent significance threshold and seeking to identify independent signals within the same locus, we identified a moderate association between 15 single nucleotide polymorphisms (SNPs) and BP, in intron 2 of the calcium channel, voltage-dependent, L-type, alpha 1C subunit gene CACNA1C. Each SNP had a p value of between 7.38×10−5 and 3.88×10−4 (Table 1). According to HapMap, these association signals were located in three blocks of largely distinct regions of linkage disequilibrium (LD), and may therefore be considered as three relatively independent associations between CACNA1C and BP (Fig. 1).
Table 1. ID, position, genotypic frequency and p value of each of the SNPs identified in the publicly available WTCCC data
SNP, Single nucleotide polymorphisms; WTCCC, Wellcome Trust Case Control Consortium.
In a subsequent study, Sklar et al. (Reference Sklar, Smoller, Fan, Ferreira, Perlis, Chambert, Nimgaonkar, McQueen, Faraone, Kirby, de Bakker, Ogdie, Thase, Sachs, Todd-Brown, Gabriel, Sougnez, Gates, Blumenstiel, Defelice, Ardlie, Franklin, Muir, McGhee, MacIntyre, McLean, VanBeck, McQuillin, Bass, Robinson, Lawrence, Anjorin, Curtis, Scolnick, Daly, Blackwood, Gurling and Purcell2008) reported a moderate association of one of these SNPs (rs1006737) (p=6.96×10−4), and by combining the results from both their own study and those from the WTCCC, they reported a nearly genome-wide significant association between rs1006737 and BP (p=3.15×10−6). This chromosomal region, 12p13.3, has been previously identified as showing moderate linkage in a study of BP (Detera-Wadleigh et al. Reference Detera-Wadleigh, Barden, Craddock, Ewald, Foroud, Kelsoe and McQuillin1999), and an earlier study by Sklar et al. (Reference Sklar, Gabriel, McInnis, Bennett, Lim, Tsan, Schaffner, Kirov, Jones, Owen, Craddock, DePaulo and Lander2002) had reported a significant association of BP with an exonic synonymous SNP in the CACNA1C gene.
Fig. 1. −Log10 of p values given by the Wellcome Trust Case Control Consortium (WTCCC) plotted against chromosomal position and the position of known genes identified using the University of California, Santa Cruz (UCSC) genome browser.
CACNA1C codes for the α1 subunit of a voltage-dependent, L-type calcium ion channel (LTCC) known as Cav1.2. The α1 subunit is part of the pore-forming unit of Cav1.2 and is also the binding site for LTCC modulators such as the dihydropyridine (DHP) group of drugs. Replication of CACNA1C variation in BP is therefore important not only in understanding the aetiology of the disorder, but also potentially in the development of, and prediction of response to, new drugs for the treatment of BP and related disorders.
BP as a calcium channelopathy
There is evidence to suggest that intracellular calcium ion dysregulation is an important aetiological component of BP. Calcium ion concentrations have been consistently found to be elevated in blood platelets and lymphocytes of patients with BP (Dubovsky et al. Reference Dubovsky, Thomas, Hijazi and Murphy1994). In addition, several calcium channelopathies [disorders associated with genetic variation in voltage-gated calcium channel (VGCC) genes, including LTCCs], such as epilepsy and familial hemiplegic migraine (Ophoff et al. Reference Ophoff, Terwindt, Vergouwe, van Eijk, Oefner, Hoffman, Lamerdin, Mohrenweiser, Bulman, Ferrari, Haan, Lindhout, van Ommen, Hofker, Ferrari and Frants1996; Mulley et al. Reference Mulley, Scheffer, Petrou and Berkovic2003), show phenotypic overlap with BP. Moreover, associations between BP and migraine (Radat & Swendsen, Reference Radat and Swendsen2005) and between BP and epilepsy (Ettinger et al. Reference Ettinger, Reed, Goldberg and Hirschfeld2005) are well documented and indicate a shared aetiology (Sheftell & Atlas, Reference Sheftell and Atlas2002). This has led to the suggestion that calcium channel dysfunction may also contribute to the genetic aetiology of other polygenic psychiatric disorders (Gargus, Reference Gargus2006) such as schizophrenia, where an association with the calcium channel gene CACNA1F has been reported (Wei & Hemmings, Reference Wei and Hemmings2006).
Calcium channel agonists and antagonists (CCAs) in the treatment of BP
CACNA1C is expressed in both the heart and the brain and, along with Cav1.3, Cav1.2 is considered the main DHP-sensitive LTCC in the brain. Both DHP-based calcium channel antagonists (CCAs) and agonists have been developed. In mice, the administration of agonists such as BayK8644 (known as ‘BayK’) causes severe dystonic neurobehavioural symptoms such as self-biting and prolongation of duration of immobility in the ‘behavioural despair test’, whereas the administration of DHP CCAs such as nifedipine has been shown to have antidepressant effects, that is reduction of BayK-induced prolongation of immobility (Mogilnicka et al. Reference Mogilnicka, Czyrak and Maj1988). Of note, the effect of BayK8644 is also antagonized by desipramine and imipramine. Sinnegger-Brauns et al. (Reference Sinnegger-Brauns, Hetzenauer, Huber, Renstrom, Wietzorrek, Berjukov, Cavalli, Walter, Koschak, Waldschutz, Hering, Bova, Rorsman, Pongs, Singewald and Striessnig2004) attempted to produce a DHP-resistant mice strain by inducing a mutation in CACNA1C. In these mice, BayK induced a ‘behavioural despair’ phenotype, and attempts at blocking Cav1.2 with nifedipine no longer had an antidepressant effect. This suggests that nifedipine and other DHPs work selectively at the Cav1.2 LTCC and that the sensitivity of Cav1.2 to these drugs is directly related to variation in the CACNA1C gene.
There are three main groups of CCAs: DHPs (e.g. nifedipine), phenylalkylamines (e.g. verapamil) and benzothiapines (e.g. diltiazem). Although these three groups of CCAs function in the same way, there are significant differences in the calcium channels they target. The DHP group are highly selective for cerebrovascular and neuronal voltage-gated LTCCs (Triggle, Reference Triggle and Epstein1992).
Because of their apparent similarity to lithium in their effect on interneuronal calcium ion activity, several CCAs have already been trialled in patients with BP. Verapamil has been shown to be more successful than lithium in the treatment of some cases of mania (Garzatrevino et al. Reference Garzatrevino, Overall and Hollister1992), although evidence regarding its efficacy is not consistent (Levy & Janicak, Reference Levy and Janicak2000). Unlike verapamil, the efficacy of the DHP group relates directly to Cav1.2 functioning. Although most members of the DHP group have generally been prescribed for angina and hypertension, only nifedipine and nimodipine have been trialled in the treatment of BP.
Nifedipine was successfully trialled in a small case study with the CCA diltiazem (De Beaurepaire, Reference De Beaurepaire1992). In two patients, the withdrawal of both drugs was followed by relapse and readmission. However, the readministration of nifedipine in these patients as monotherapy had no effect on their manic symptoms. It is possible that nifedipine works predominantly on the depressive phase of BP, which would be supported by the observation that nifedipine has apparent antidepressant properties in mice.
The second-generation DHPs show greater central nervous system (CNS) penetration and therefore may be more promising agents for BP, particularly nimodipine, which has the greatest affinity for neuronal rather than vascular LTCCs. Indeed, nimodipine has shown some success in both an open (Brunet et al. Reference Brunet, Cerlich, Robert, Dumas, Souetre and Darcourt1990) and a controlled trial as a monotherapy in rapid cycling and refractory recurrent BP (Pazzaglia et al. Reference Pazzaglia, Post, Ketter, George and Marangell1993, Reference Pazzaglia, Post, Ketter, Callahan, Marangell, Frye, George, Kimbrell, Leverich, Cora-Locatelli and Luckenbaugh1998). However, the limited number of studies with small sample sizes means that replications are required before any firm conclusions can be drawn regarding the efficacy of the drug. Nicardipine, isradipine, amlodipine and felodipine have been designed to be more selective for vascular rather than neuronal LTCCs, which may explain why they have not yet been trialled in BP, other than in the above study, in which Pazzaglia et al. (Reference Pazzaglia, Post, Ketter, Callahan, Marangell, Frye, George, Kimbrell, Leverich, Cora-Locatelli and Luckenbaugh1998) substituted nicardipine for nimodipine in those who responded to the drug. Of interest, nicardipine was able to show maintenance of the efficacy.
Finally, most of the above LTCC antagonists have been trialled in migraine. Moreover, there is pharmacological overlap between CCAs and medications with antihistaminergic and antidopaminergic activity. One such agent, flunarizine, blocks calcium channels, including LTCCs, and is among the most effective agents for migraine prophylaxis (Amery, Reference Amery1983; Silberstein & Goadsby, Reference Silberstein and Goadsby2002). This drug has not yet been trialled in BP.
As there is substantial phenotypic overlap with between BP and unipolar depressive disorder (UPD) (McGuffin et al. Reference McGuffin, Rijsdijk, Andrew, Sham, Katz and Cardno2003), and migraine is associated with both disorders, the relevance of CACNA1C may also extend to other affective disorders such as UPD. Although clinical trials of DHPs in UPD have so far been disappointing, these have been relatively few and limited by sample size (Levy & Janicak, Reference Levy and Janicak2000). Should BP and/or UPD indeed be associated with sequence variation in CACNA1C, it is also possible that this could influence the response of individuals with these disorders to agents (such as DHPs) that affect this channel. We suggest that CACNA1C should be considered as a potential candidate gene for exploration in pharmacogenetic studies in both BP and UPD, and in any further relevant emerging genome-wide association studies (GWAS) data.
Conclusion
Given the phenotypic overlap between BP and other known channelopathies, the pharmacological overlap in their treatment, the promising efficacy of some CCAs in the treatment of BP and evidence that suggests calcium is dysregulated in BP, we suggest that VGCCs (particularly LTCCs) are both relevant candidate genes for BP and promising drug targets for its treatment. Although evidence regarding the efficacy of CCAs is overall mixed, DHPs, which target Cav1.2 and may therefore relate directly to variation in CACNA1C, have shown the most promising results. Second-generation DHPs, especially nimodipine, are of particular interest because of their good CNS penetration and we therefore suggest that clinical trials should be conducted within this group. In addition, flunarizine and the first-generation derivatives of antihistamines should be reconsidered and investigated for their effect on ion channels, especially for individuals with both BP and migraine.
Acknowledgements
Robert Keers is funded by a Medical Research Council (MRC) PhD Studentship to the MRC SGDP Centre at the Institute of Psychiatry, King's College London. Dr Aitchison and Professor A. E. Farmer are funded partly by a UK National Institute of Health Research Biomedical Research Centre Grant (BRC-SLAM).
Declaration of Interest
None.
