In contrast to the methylation that is catalyzed exclusively
In contrast to the methylation that is catalyzed exclusively by DNMTs, there have been suggested several pathways that could induce demethylation (Watt & Molloy, 1988). However, the existence and influencing of enzymes that promote active demethylation remains elusive. Studies aiming to gain more insight into the demethylation mechanism in vertebrates suggested that, under specific conditions, mammalian DNMTs could act as active DNA demethylases by removing the methyl or hydroxymethyl group from 5-methylcytosine (5-mC) or 5-hydroxy-mC (5-hmC), respectively (Chen et al., 2012, Chen et al., 2013). Importantly, the demethylase activity of DNMTs is Ca2+-dependent (Chen, Wang, & Shen, 2013). Considering that increased Ca2+ mediates signal transduction in neuronal populations, it could be speculated that in activated neurons DNMTs promote active demethylation. In our study, the animals were sacrificed thirty minutes after the mnemonic test raising the possibility that the action of DNMTs was shifted towards the demethylation pathway. In that case, administration of RG108 could prevent demethylation rather than methylation explaining the initially counterintuitive finding of increased Bdnf1 methylation after treatment. Additional studies are required in order to shed light into the complex epigenetic changes occurring after DNMT inhibition as well as to determine the exact location of DNMT binding in the promoter region of Bdnf1. Additionally, of outmost important is the location of the different CpG islands in relation to the transcription starting site (TSS) of the promoter as well as their interaction with transcription activators or repressors. It has been shown that methylation at the downstream region of the TSS in the region of exon I of a gene is more related to transcriptional silencing in comparison to methylation upstream of the TSS, in the promoter region (Brenet et al., 2011, Okitsu and Hsieh, 2007). Additionally, a study of Tian et al. demonstrated that activity-dependent VO-Ohpic trihydrate mass remodeling occurring in the promoter of Bdnf1 follows a temporally distinct pattern (Tian et al., 2009). This phenomenon produces “waves” of transcription that cumulative determine the transcriptional outcome of the Bdnf1 mRNA (Tian et al., 2009). The above findings also indicate that the observed increase in Bdnf1 expression in our study possibly represents a net effect of DNA methylation/demethylation in Bdnf1 promoter. This is also depicted on the fact that increased methylation in the CpG islands is not correlated anymore with increased expression levels of Bdnf1. It could be argued that epigenetic changes in the Bdnf1 gene occur due to handling, training and previous exposure to the learning paradigm. Rapid changes in methylation status of genes have been suggested as part of a priming mechanism based on which epigenetic modifications could take place after a certain stimulus and these changes could facilitate a faster response of the genes during the next (same) stimulation. Alternatively, it is suggested that such epigenetic changes are transient and, during the absence of the original stimulus, the methylation levels of the gene return to baseline. Along the same line, persistent methylation due to a stimulus is thought to negatively alter the ability of a neuron to respond to later stimuli, rather than contributing to it (for a review see Baker-Andresen, Ratnu, & Bredy, 2013). Considering that epigenetic changes most likely have a transient nature, we suggest that the observed changes in Bdnf1 expression and methylation are due to treatment with RG108 and that they are not related to other experimental aspects of our study to which the animals have already habituated. Pharmacological agents that improve pattern separation could be beneficial for the treatment of disorders that are related to impairments in pattern separation, like anxiety disorder (Kheirbek et al., 2012) and schizophrenia (Tamminga et al., 2010). In that respect, treatment with DNMT inhibitors could provide a novel approach, especially since current treatment for the above disorders lacks efficacy. Therefore, future studies could investigate the potency of DNMT inhibition in these disorders and further determine whether this effect is related to upregulation of Bdnf expression. Nevertheless, treatment with DNMT could have wide spread effects in the whole gene. This makes it necessary to check other brain regions than the dorsal hippocampus alone, as changes in those regions could also be the underlying cause of the observed behavior. Additionally, it is important to determine whether chronic treatment also improves pattern separation performance without introducing unwanted side effects or showing decreased efficacy (e.g. due to tolerance).