Synapsin II is a candidate gene that was originally identified through a linkage study of Li-responsive BD families. This gene was also shown to be dysregulated in the post-mortem brains of patients with BD as compared to psychiatrically healthy controls in the same study 
. Thus, we were interested to investigate the effect of Li treatment on the expression of this gene. We did so in the genetic context of the disorder by treating with Li monotherapy Epstein-Barr virus-transformed lymphoblastoid cell lines from BD patients characterized as excellent Li-responders or non-responders, as well as healthy controls with no history of psychiatric disorders. We found that the pattern of expression was significantly different in Li-responders compared to both non-responder BD patients as well as controls. However, the direction of change of expression was not uniform across subjects (), resulting in no overall mean differences between groups. These data suggest that Li modulates SYN2 expression in a way that is specific to Li-responders, possibly reflecting significant genetic heterogeneity.
The relevance of SYN2 expression in peripheral cells compared to the central nervous system in BD patients is not clear from our findings, particularly since we saw no mean differences between Li-responders, non-responders, and controls. It is however clear from the literature that the gene is expressed, though at more basal levels, in lymphoblasts as well as many other cell types. Despite their peripheral origin, studying transformed LCLs offers the benefit of performing in vitro assays on cells from patients and studying putative factors in their endogenous expression context. However, results from these experiments should be considered with a level of scepticism, as the relevance of SYN2 expression in this cell type is unclear.
Environmental factors could be involved in Li's regulatory role, which might account for the observed patient-specific effects in Li-responders. To investigate this possibility we computed correlations with a number of environmental factors relating to age of patients, Li therapy, and family history of other psychiatric disorders (for a complete list, refer to ). However, none of the potential covariates correlated with SYN2a or SYN2b expression values, suggesting that the source of variation may be related to genetic or possibly epigenetic differences between patients. For example, variants in CREB genes 
or GSK3B 
, have been shown to associate with Li-treatment response. Similarly, it is possible that epigenetic factors may increase SYN2 expression variance among patients. Though this is of interest, to our knowledge, no studies have investigated the role of Li treatment on epigenetic modifications in the human brain. However, valproate, another widely used mood stabilizer, is well known for its inhibitory effect on histone deacetylases (HDACs) 
and therefore, it is possible that at least part of Li's action may be related to epigenetic regulation. Another epigenetic regulatory level where lithium's effect could be confounded is microRNA-mediated regulation. Studies in LCLs 
and animal models 
have shown the drug's global effect on this class of molecules. For a variety of biological reasons, each patient's LCLs could be enriched in a combination of regulatory factors which could then impact the response to Li treatment.
Since our LCL results do not automatically represent what is occurring in the brain, we sought to determine if Li would have a cell-type-specific effect on SYN2 expression in model cell lines representative of the brain, and showed a significant change in the neuronal cell line SK-N-AS only ( and ). There was an effect at 1.0 and 2.0 mM Li, but not at 0.5 mM, suggesting that this concentration was not high enough to elicit a response. Interestingly, the effect was specific to the SYN2a variant (), as the SYN2b variant remained unchanged between conditions (). Originally, SYN2 had been believed to display neuron-specific expression in the brain; however, further studies demonstrated the gene's expression in other cell types, though at considerably lower concentrations 
. SYN2 is expressed at basal levels in various cell types and thus lithium likely modulates its expression to a certain degree in these cells but perhaps not in a functionally-relevant manner. This is consistent with the fact that synapsins are evolutionarily conserved from humans to very primitive organisms and likely their expression has become more specialized in higher organisms through a loss of the ability to regulate other cellular functions but not necessarily through a complete loss of expression 
According to our results, in neurons, Li treatment significantly increases SYN2 expression perhaps by also recruiting other neuron-specific transcription factors that bind to the gene's promoter such as EGR1 (early growth response 1), which has been suggested to regulate the gene 
, or AP-2alpha, which has been shown to be regulated by lithium 
. Our results from LCLs are seemingly contradictory, as Li has an up-regulating effect on SYN2 in some patients, and a down-regulating effect in others. To interpret these results, one needs to consider that lithium acts as a mood stabilizer in patients who present both manic and depressive episodes. These clinical episodes are characterized by symptoms that are on opposite sides of the mood spectrum. Accordingly, manic patients present mood and neurovegetative activation, while depressed patients are characterized by a decreased mood levels and neurovegetative inhibition. Therefore, in order to be an effective mood stabilizer, Li needs to act by normalizing variance.
One interesting addition to this study would have been direct evidence for the effect of Li on SYN2 expression in the central nervous system of BD patients. An ideal study would investigate the expression of SYN2 variants in the post-mortem brains of BD patients who had been excellent responders to prophylactic Li for an extended period of time, so as to match the criteria used for our LCL samples. However, post-mortem brain donors with a history of BD are most often suicide completers. The literature provides extensive evidence for the anti-suicidal effects of Li prophylaxis through observational studies 
, randomized controlled studies 
and meta-analyses 
. Thus, such a study would be logistically quite challenging.
Another limitation of our study is the lack of protein-level evidence to support our mRNA-level findings. Such validation would be interesting in the pursuit of qualifying SYN2 as a factor of potential pharmacological significance. However, the results presented here mainly point to SYN2 as a new mediator of Li action. Perhaps by further investigating how SYN2 is regulated we will also elucidate lithium's mode of action. There are likely several regulatory levels at play and clarifying them will be instrumental for our understanding of lithium response in BD, but as it stands the pharmacological application of this work is preliminary.
In conclusion, this is, to our knowledge, the first study attempting to determine the effect of Li treatment on mRNA-level expression of SYN2. We found a responder-specific effect of Li in LCLs from BD patients, suggesting that even though the gene is important for BD in general, there are genetic or epigenetic differences in Li responders that make them more susceptible to modulation of SYN2. Additionally, we showed that the effect of long-term treatment with Li is likely cell-type specific. As far as brain expression, our data suggest that the effect of lithium treatment is only significant in neuronal cells and not in astrocytic or glial cells. Support from additional cell types would be important to strengthen the validity of these conclusions. Our distinct findings for the two SYN2 variants as well as the reported homology in sequence and function of the family of synapsin genes opens up the question of whether the other synapsins have a neuron-specific effect, as well as a patient-specific effect. Our study points to a very interesting player in response to Li prophylaxis, but more studies are required to decipher the full pathway of Li action that leads to its stabilizing effect in a large fraction of BD patients.