BDNF exon IV promoter activity is regulated by methylation in a CpG dependent manner in the absence of neuronal activity (8). BDNF exon IV promoter methylation has been associated with multiple psychiatric disorders along with MDD (19) as well as treatment response to ADs (24) and other available therapy regiments (25). Previous work from our lab has confirmed the CpG-87 hypomethylation as a biomarker for unsuccessful AD treatment in two different studies (21, 22). This work aimed to explore the link between AD action and CpG-87 methylation mechanistically.
To understand the effect of ADs, methylation and their interaction on the transcriptional activity of BDNF exon IV, we cloned BDNF exon IV promoter region in a CpG-free luciferase vector (26) and transfected it into a neuroblastoma cell line SH-SY5Y and treated them with ADs 24 h after transfection for 48 h. The ADs decreased BDNF exon IV promoter activity and this effect was nullified by methylation arguing in favor of a possible link between antidepressant mediated BDNF increase in serum of patients harbouring methylation at CpG-87 locus. These findings corroborated with our previous study where we used the luciferase vector (Promega) containing CpGs demonstrating that the reduction in the promoter activity is due to the CpG methylation of exon IV promoter and not the vector itself. It must be noted that we methylated the promoter fragment using bacterial methylase MsssI, thereby methylating every CpG in the promoter to approximately 90% rendering our system far from physiological.
Linear effect of pharmacological treatments on BDNF promoter methylation has not been clearly elucidated, mainly because of lack of pre-treatment methylation data from the patients and lack of drug naïve cohorts (19).
The clinical data from our lab shows that antidepressants do not change the methylation over 6 weeks of treatment (21). This patient cohort did not contain any drug naïve patients and therefore we could have missed potential AD mediated changes in methylation. To address this in vitro, we incubated SH-SY5Y cells with ADs and analyzed the CpG methylation of BDNF promoter. Both Venlafaxine and Fluoxetine increased the mean promoter methylation in a concentration dependent manner. We did not observe any change in the methylation of CpG-87 upon treatment, suggesting that the disease does not directly modulate CpG-87 or treatment itself. CpG-147, CpG-66, CpG-39 and CpG-35 (suppl. Figure 1) drove the differences in the mean methylation. Our data justify the scope for further studies investigating the molecular mechanisms underlying the antidepressant moderated epigenetic changes.
Further, to differentiate the regulatory effect of methylation of individual CpGs from the whole promoter, we mutated the CpGs of interest to methylated CpGs in the BDNFIV-pCpGL basic construct and performed luciferase assays with or without ADs. We included CpG-87 to model the clinical findings and CpG-35 and − 39 due to their proximity with CRE (CREB binding element). CREB has been shown to regulate BDNF protein (27–29) and is involved in antidepressant response (30). We observed no difference in the BDNF exon IV promoter activity upon methylating CpG-35, and − 87 in comparison to the unmethylated controls, but CpG-39 methylation increased the transcriptional activity of the promoter. Further plasmid harboring methylated CpG-39 showed a significant increase in promoter activity in the Venlafaxine and Fluoxetine treated cells, suggesting yet again possible drug-methylation interaction. The role of methyl CpG binding protein (MeCP2) has been implicated in depression and more so as a BDNF repressor in resting neurons. MeCP2 can bind to both methylated cytosine (5mC) and hydroxylated methyl cytosine (5hmC), thereby acting as both transcriptional repressor and activator (31). Calcium influx induced neuronal depolarization phosphorylates MeCP2 at specific serine and threonine residues, facilitating its multidirectional role as a regulator (32).
MeCP2 physically and functionally interact with CREB facilitating to the interaction with CRE on the active promoters (33). Therefore, we speculate that this increase in transcriptional activity can be due to enhanced MECP2-CREB complex recruitment to the promoter. We conclude that CpG-87 methylation does not directly increase BDNFIV transcription. A unique population of cells might contribute to the association of positive treatment response to the methylation (1–5%) at this position (measured in blood), which can be identified by emerging techniques like single cell epigenomics (34). Due to technical limitation, we could only methylate single CpG at a time, so we did not account for permutation and combination of methylation patterns that probably exist in the patients.. We will be sequencing a larger cohort of MDD patients using Oxford Nanopore Technologies that allows single strand methylation analysis that can dissect complex epi-allelic methylation methylation patterns.
DNA methyltransferases specifically methylate regulatory CpGs and modulate gene transcription. A suggestive link exists between DNMTs expression and MDD in the literature (35). To address whether the AD dependent BDNF increase in serum/plasma levels of patients and animals is associated with in vitro methylation changes, we overexpressed DNMTs and treated the cells with ADs. Of the methyltransferases, only DNMT3a increased exon IV activity in control and Venlafaxine treated cells. Interestingly, of the two denovo methyltransferases, DNMT3a is expressed in adult brain, whereas DNMT3b is only found in neural progenitor cells (17). When we bisulfite-sequenced the DNMT overexpressed cells, we did not observe any change in the CpG methylation at BDNF exon IV promoter. Taken together, our findings suggest an indirect effect of DNMT3a overexpression on increase in BDNF IV promoter activity. It will be very interesting to get a deeper insight into the expression pattern of DNMTs in the responders and non-responders.
Our system relies on artificially curated methylation marks that do not reflect the complex scenario in vitro. Therefore, it will be interesting to use Cas9 mediated methylation editing by recruiting methyl modifiers (DNMT/TET) to study the interaction between BDNFIV methylation and antidepressant action on its regulation. Another limitation of the study is that we used a neuroblastoma cell line, thus lacking the communication between different cell types that is essential for antidepressant action. Therefore, we plan to replicate the findings in a mixed cell culture model consisting neurons, astrocytes and myelinating oligodendrocytes.