Mood disorders are the most common chronic psychiatric disorders in the world and are a leading cause of morbidity. In patients with these disorders, mood can range from elation or mania to deep depression. Patients with bipolar I disorder typically demonstrate at least one major manic episode and usually also a major depressive episode, while those with bipolar II disorder typically show a pattern of depressive symptoms and hypomanic episodes. [1, 2, 3]
Etiology of mood disorders is unclear, although a genetic component has been strongly suggested by family and twin studies. However, the mode of inheritance is complex, with no clear Mendelian pattern. Heritability of mood disorders ranges from 50% in major depression to 80% in bipolar disorder.  Bipolar disorder, in particular, is seen in approximately 1% of the population.
Many chromosomal regions have shown linkage to bipolar disorder, but meta-analyses of microsatellite marker – based linkage studies have not provided consistent findings of susceptibility regions. [5, 6] Nevertheless, mechanisms behind therapeutic agents used in patients with the disorder have lent support to the possible role of a few different genetic pathways and mutations.
A meta-analysis of original data from 11 previous linkage studies in 1067 bipolar disorder families yielded significant findings in chromosomal regions 6q for bipolar I and 8q for bipolar I/II, as well as suggestive findings at chromosomal regions 9p and 20p for bipolar I.  The most significant 6q region harbors the melanin concentrating hormone receptor 2 (MCHR2) gene,  which has been implicated in the phosphoinositol pathway and in intracellular calcium release  as well as in the kynurenine pathway, all of which are thought to play a role in the mood stabilizing effect of lithium. A follow-up study also pointed to the SLC22A16 (organic cation/carnitine transporter) gene at 6q21,  the results which were corroborated by a linkage study with high-density, single-nucleotide polymorphisms. 
Association studies of individual candidate genes have yielded largely mixed findings, [12, 13] but some consistent findings have been reported in a number of genes. For example, the low-functioning variant of a promoter polymorphism (HTTLPR) in the gene coding for the serotonin transporter, [14, 15, 16] the target for serotonin reuptake inhibitor drugs, was associated with bipolar disorder in a meta-analysis.
A number of genes associated with circadian patterns have been found to be associated with bipolar disorder, including aryl hydrocarbon receptor nuclear translocator-like (ARNTL). [17, 18] and circadian locomotor output cycles kaput (CLOCK).  These data support the suggestion that disrupted circadian patterns play a role in bipolar disorder.  In addition, lithium has been shown to influence the circadian rhythm. [21, 22, 23] Indeed, mice carrying a mutant CLOCK gene displayed manialike behaviors that were reversed by lithium treatment.  Recent findings of circadian rhythm – related genes support the continued study of these genes in bipolar disorder and lithium response. 
Markers within the cadherin (FAT) gene. [25, 26] were found to be associated with bipolar disorder in multiple samples, and expression analysis in rodents following administration of lithium or valproate indicated a role for FAT in neurodevelopment signaling pathways. 
Finally, both lithium and valproate have been shown to inhibit GSK3beta.  Neurotransmitters, including serotonin,  as well as neurotrophins, including BDNF, [29, 30, 31] , are known to inhibit GSK3beta,  and the BDNF gene has been possibly implicated in bipolar disorder, [33, 34] particularly with regard to its association with rapid mood cycling.  Most studies on the GSK3B gene in bipolar disorder have been negative, but one study reported an increased number of copy-number variations (deletions or duplications) within the gene in a small sample of bipolar disorder patients.  Moreover, the observation that GSK3beta was required for lithium-associated behavioral effects in rodents  encourages the examination of this gene in lithium response.
Clinical Implications and Genetic Testing
A number of risk genes for bipolar disorder have emerged through genome-wide association studies (GWASs) in large samples of thousands of patients of European ancestry. [1, 38, 39] These include diacylglycerol kinase eta (DGKH), [40, 41] alpha-1 subunit of the L-type voltage-gated calcium channel (CACNA1C), [42, 43, 44, 45] , ankyrin G node of Ranvier (ANK3), [40, 46, 47, 48] nesprin-1 (SYNE1), [43, 49] and teneurin transmembrane protein 4 (ODZ4). [43, 45, 48] However, none has yet been shown to have a strong enough effect to warrant testing in clinical settings.
For example, DGKH phosphorylates diacylglyerol in the phosphoinositol pathway that is sensitive to lithium,  providing a possible candidate for genetic studies of lithium response in bipolar disorder patients. However, even though there was additional support for an association between DGKH and bipolar disorder,  a preliminary study did not support the role of this gene in lithium response.  Also, although ANK3 regulates firing of action potentials through its regulation of voltage-gated sodium channels,  and the ANK3 gene rs10994336 risk marker was associated with poorer performance in a neuropsychological task of sustained attention, more research is required to confirm that the variant has direct functional activity.  The SYNE1 (or candidate plasticity gene 2, CPG2) gene has been shown to mediate the internalization of glutamate receptors in the postsynaptic neuron. [55, 56]
Finally, the rs1006737 risk marker of CACNA1C, which may play a role in the release of hormones and neurotransmitters upon membrane depolarization, has recently been associated with increased CACNA1C gene expression, increased functional MRI hippocampal activation in subjects presented with averse images, and increased prefrontal cortical activation during a working memory task.  In additional studies combining neuroimaging and genetics, bipolar disorder patients carrying the risk allele also had lower volume of the left putamen than did non-carriers.  One of the modes of lithium action could be through the inhibition of inositol 2,3,4-triphosphate receptor,  thus lending support for the role of calcium signaling in the pathophysiology in bipolar disorder.
However, overall, the effect sizes of this and other risk variants are small, suggesting that bipolar disorder is likely genetically complex. In addition, many of the bipolar disorder risk genes discussed above, as well as the D-amino acid oxidase activator (DAOA/G72), [60, 61] disrupted in schizophrenia 1 (DISC1), [62, 63, 64] and zinc finger 804a (ZNF804A) [65, 66, 67] are shared between bipolar disorder and schizophrenia,  suggesting that there are overlapping pathophysiologic mechanisms shared between these two psychiatric disorders.
Breaking down the diagnostic boundaries between schizophrenia and bipolar disorder and investigation of severity of symptom dimensions that exist between these two psychiatric disorders is a potential alternative approach to disentangle their complex genetic and biological mechanisms. [69, 70, 71] Although there seem to be clear signals that bipolar disorder may be associated with changes in circadian rhythm, ion-channel function, neurodevelopment, and/or GSK3 beta signaling, much additional research is needed to understand the potential impact of this knowledge in clinical practice.
By contrast, pharmacogenetics of mood disorders may yield potentially promising findings to help guide clinicians in treatment decisions in the near-term.  For example, Mundo et al found that the short variant of the serotonin transport promoter HTTLPR may predict for patients at risk for the serious adverse effect of antidepressant-induced mania.  This finding has been supported by recent meta-analyses. [74, 75, 76] Candidate gene studies of lithium response have reported mixed findings for HTTLPR, as well as for the GSK3B and BDNF genes, but studies in this area are growing in number, [77, 78, 79, 80, 81] as are pharmacogenetic studies of antipsychotic or antidepressant treatment in patients with bipolar disorder. [82, 83, 84]
For lithium response in particular, a number of GWASs have been conducted in well-characterized bipolar disorder samples. They implicated a number of novel candidate genes, including AMPA glutamate receptor (GRIA2),  syndecan 2 (SDC2),  and neuronal amiloride-sensitive cation channel 1 (ACCN1/ASIC2/MDEG).  A GWAS was conducted on a East Asian sample. The authors reported a highly significant signal in the gene coding for glutamate decarboxylase like protein 1 (GADL1),  with the T allele at the rs17026688 marker associated with better lithium response than the C allele. Investigations of GADL1 gene variants in samples of various ethnicities are warranted. [88, 89] The largest GWAS by the Consortium on Lithium Genetics (ConLiGen) suggests a role of long non-coding RNAs,  and the Pharmacogenomics of Mood Stabilizer Response in Bipolar Disorder (PGBD) prospective multisite trial is currently underway. 
Ultimately, future genetic and pharmacogenetic studies will require uniform, standardized study designs and replications in larger independent samples before firm conclusions can be drawn. For genes that have been associated with bipolar disorder diagnosis, and that have been shown to be regulated by lithium and/or valproate, future pharmacogenetic studies may yield findings that can be translated into better clinical care. [87, 91, 92]