The causes of bipolar disorder are far from clear, although family, twin and linkage studies have shown that genetic factors play an important part in its development. Reporting in Nature Genetics, Kakiuchi and co-workers now identify XBP1 — a gene that is central to the endoplasmic reticulum (ER) stress response — as contributing to the genetic risk factor for this condition.

The rate of concordance for bipolar disorder in monozygotic twins is considerably higher than that in dizygotic twins, but some monozygotic twins are discordant — that is, only one of the pair of twins is affected. The biological basis of discordance between monozygotic twins is not fully understood, although several mechanisms have been documented.

Kakiuchi et al. carried out a DNA microarray analysis of lymphoblastoid cells — which have altered signal-transduction systems in patients with bipolar disorder — from two pairs of monozygotic twins who were discordant for the condition. They found that XBP1 and HSPA5 were downregulated in both of the affected twins. HSPA5 gene expression is known to be induced by the mood stabilizer valproate, and is regulated by XBP1. The team went on to show that a single-nucleotide polymorphism (−116C→G) in the promoter region of XBP1 that affects a putative binding site for XBP1 itself was more common in Japanese patients with bipolar disorder than in healthy subjects, and was overtransmitted from parents to affected offspring in a collection of samples from people of mixed ethnicity, chiefly of European origin.

Because an XBP1-binding motif seems to be abolished in the risk allele, the group reasoned that the −116C→G polymorphism might alter the positive feedback activity of XBP1. They found that XBP1-dependent transcription of the −116G allele was significantly lower than that of the −116C allele; in addition, in cells with the −116G allele, the induction of XBP1 expression in response to ER stress was reduced.

So, the −116C→G polymorphism in XBP1 seems to cause a defect in its positive feedback system and to increase the risk of developing bipolar disorder. But do valproate and other mood stabilizers affect this feedback loop? Kakiuchi et al. found that valproate, but not lithium or carbamazepine, could rescue the impaired ER stress response by inducing ATF6 , the gene upstream of XBP1.

This study implicates the XBP1 loop of the ER stress response in the pathophysiology of bipolar disorder. The fact that only valproate could reverse the impairment of the XBP1 loop raises the possibility of customizing treatments for patients according to the genetic risk for the condition. Moreover, these findings point to ATF6 as a potential target for the development of mood stabilizers. Interestingly, the authors found that XBP1 was expressed at relatively high levels in the human prefrontal cortex. Further studies of this system could provide insights into the molecular mechanisms of mood.