The tropomyosin receptor kinase B (TrkB) is the primary receptor for brain derived neurotrophic factor (BDNF), which is broadly expressed and released by neurons in developing and mature brains (Pandya et al., 2013). This signaling pathway plays a critical role in brain development processes, including, neuronal differentiation and survival, synaptogenesis, synaptic transmission, synaptic plasticity (Kowianski et al., 2018; Enkavi et al., 2024) as well as learning and memory (Minichiello, 2009; Bathina and Das, 2015). Given its central role in the healthy brain, it is not surprising that aberrant disrupted BDNF/TrkB signaling has been implicated in neuropsychiatric, neurodegenerative and neurodevelopmental disorders (Numakawa et al., 2018; Deng and Chen, 2024; Peng et al., 2009; Ferrer et al., 1999; Autry and Monteggia, 2012; Stansfield et al., 2012). Research in preclinical rodent studies as well as previous and ongoing clinical trials in human populations indicate targeting BDNF/TrkB signaling pathways holds therapeutic promise for neurodegenerative and neuropsychiatric diseases by driving synaptic plasticity, promoting neurogenesis, and ameliorating motor and cognitive deficits (Yoshii and Constantine-Paton, 2010; Boulle et al., 2012; Lu et al., 2013; Mitre et al., 2017; Ali et al., 2024; Enkavi et al., 2024; ClinicalTrials, n.d.).

Two main isoforms of TrkB exist in the central nervous system (CNS), the full-length receptor, TrkB.FL and the truncated receptor TrkB.T1. TrkB.FL and TrkB.T1 share 100% sequence homology in the extracellular and membrane spanning domains and thus are thought to have equal affinity for BDNF in both human and mouse (Sasi et al., 2017). TrkB.FL uniquely possesses a tyrosine kinase domain that autophosphorylates with BDNF binding to trigger multiple downstream signaling cascades, including changes in cell morphology, cell survival, neuronal differentiation and neurotransmitter release (Cao et al., 2020; Stoilov et al., 2002; Martin-Rodriguez et al., 2020; Boulanger and Poo, 1999; Huang and Reichardt, 2001). While TrkB.T1 lacks the canonical tyrosine kinase domain and much less is understood regarding the downstream signaling of BDNF binding to TrkB.T1. Previous studies have postulated that TrkB.T1 may behave as a dominant-negative receptor whereby it dimerizes with full-length TrkB to prevent autophosphorylation and canonical BDNF-mediated signaling (Fenner, 2012). Additionally, a limited number of studies have demonstrated that BDNF binding to TrkB.T1 may induce activation of protein kinase C (PKC) cascades activated by G protein signaling (Cheng et al., 2007), stimulation of phospholipase C-γ (PLCg) pathways resulting in calcium release and cytoskeletal changes (Rose et al., 2003) and inhibit Rho GTPase activity, (Ohira et al., 2005), thereby inducing cell morphological changes.

We recently demonstrated in the mouse cortex that TrkB.T1 is largely specific to astrocytes and over 90% of all Ntrk2 expression in astrocytes is attributed to TrkB.T1 (Holt et al., 2019). This is in contrast to TrkB.FL, the predominant TrkB isoform expressed by neurons. Surprisingly, given the critical role of BDNF/TrkB signaling, few studies have attempted to identify molecular mechanisms leading to and regulating the CNS expression of TrkB isoforms (Tessarollo and Yanpallewar, 2022). Previous studies from post-mortem human frontal cortical tissue from both suicide completers and age-matched controls demonstrated lower levels of TrkB.T1 mRNA and protein correlated with altered methylation states of several CpG sites in the promoter region (Ernst et al., 2009) and 3′UTR region (Maussion et al., 2014). These studies focused on small, circumscribed, gene regions using bisulfite sequencing or microarray methodology followed by DNA methylation analysis. As indicated above, TrkB is a large, complex gene and it is unclear if methylation changes in these human studies were specific to the regions analyzed, or if changes were cell type specific.

DNA methylation, both 5mC and 5hmC, are powerful forms of epigenetic transcriptional regulation, the majority of which occurs on cytosines that precede a guanine nucleotide or CpG sites in the promoter region, gene body and intergenic gene regions (Moore et al., 2013). DNA methylation determines the spatiotemporal expression pattern of genes in most eukaryotic cells by altering transcription binding and activity, chromosomal structure, DNA conformation and DNA stability (Wang and Wu, 2018). Changes in the expression level of BDNF, the ligand of TrkB, have been observed to be linked to DNA methylation regulation (Alece Arantes Moreno et al., 2023; Ikegame et al., 2013; Mitchelmore and Gede, 2014). Importantly and relevant for this study, it has been reported that the inclusion or exclusion of nearly 20% of alternative exons in human genes is regulated by DNA methylation (Yearim et al., 2015; Ramanouskaya and Grinev, 2017). DNA methylation has been verified to collaborate with transcription factors (TFs) to regulate exon inclusion or exclusion. For example, CCCTC-binding factor (CTCF) binds to unmethylated DNA and promote the inclusion of weak upstream exons (Shukla et al., 2011) while methyl-CpG binding protein 2 (MeCP2) binds to methylated DNA to pause Pol II and leads to exon inclusion (Maunakea et al., 2013). Although less is understood about the role of 5hmC in gene isoform expression or alternative splicing, recent findings indicate that 5hmC peaks are present at the 5′ splice site of exon-intron boundaries, suggesting that 5hmC may regulate these processes (Wen et al., 2014; Khare et al., 2012).

Recently, we employed Nanopore sequencing to profile 5mC and 5hmC modifications in different CNS cell types in mouse cortex. Nanopore sequencing is a 3rd generation, long read sequencing approach which detects the electric current signatures of nucleic acids and modified nucleic acids as they pass through a nanopore. This approach allows for direct sequencing of native DNA and enables the identification of multiple DNA modifications in the same sample. These advantages make this approach ideal for quantitative assessment of DNA methylation patterns for large genomic regions or complex genes such as Ntrk2. Here we analyzed 5mC and 5hmC modification signatures of the Ntrk2 gene in its entirety in enriched astrocyte and neuron cortical cell populations. Notably, we identified 98.74% of differentially methylated sites (DMSs) are hypermethylated in neurons and all the differentially hydroxymethylated sites (DhMSs) have higher 5hmC levels in astrocytes on the gene Ntrk2. Additionally, 37–38% of the DMSs/DhMSs are located around the unique CDS in TrkB.T1. Together, these data suggest DNA methylation patterns may provide instruction for isoform specific TrkB expression across unique CNS cell types.

Decades of research have implicated BDNF/TrkB signaling in early brain development and neuronal maturation (Castren et al., 1992; Altar et al., 1997; An et al., 2008; Bramham and Messaoudi, 2005; Amaral and Pozzo-Miller, 2007; Wang et al., 1995). Further, adaptations to this signaling are observed in neurodevelopmental and neuropsychiatric conditions (Numakawa et al., 2018; Deng and Chen, 2024; Peng et al., 2009; Ferrer et al., 1999; Autry and Monteggia, 2012; Stansfield et al., 2012). Our recent work indicates that BDNF signaling through the astrocyte expressed truncated TrkB receptor is also critical for astrocyte morphological maturation (Holt et al., 2019), contributing to the impact of this signaling pathway in brain development. Despite its importance, very few studies have attempted to quantify differences in expression of TrkB across CNS cell types. Mining publicly available resources (Karlsson et al., 2021; Saunders et al., 2018; Human Protein Atlas, n.d.), we present data that across species TrkB expression is highest in astrocytes relative to other CNS cell populations. Additionally, the truncated isoform TrkB.T1 predominates in the human brain, corroborating our previous results demonstrating TrkB.T1 is more highly expressed in rodent cortex relative to TrkB.FL (Holt et al., 2019).

As it relates to isoform specific expression of TrkB, few studies have attempted to pinpoint the underlying molecular mechanisms. Early work demonstrated two alternative promoters in the TrkB gene in mouse brain. These promoters, designated P1 and P2, were subsequently demonstrated to play no role in distinct isoform expression, but instead generate mRNAs that are different in their 5’UTR region (Barettino et al., 1999). Interestingly, transcripts initiated for either the P1 or P2 promoter of Ntrk2 generated both full length and truncated isoforms of the TrkB gene in cultured cortical neurons. Further, cAMP activation via CREB binding to a CRE element present in the P2 promoter, approximately 0.5 kb, upstream of the transcription start site of Ntrk2, triggered gene and protein expression of both TrkB isoforms (Deogracias et al., 2004). Although not yet evaluated, these results suggest CREB activation may serve to regulate TrkB.T1 expression in astrocytes. More recently, it was identified in human cortical tissue, collected post-mortem from suicide completers demonstrated lower levels of TrkB.T1 mRNA and protein. This finding correlated with altered methylation states of several CpG sites in the promoter region (Ernst et al., 2009) and 3′UTR region (Maussion et al., 2014), of the NTRK2 gene. These findings indicate DNA methylation as a potential regulator of isoform specific TrkB expression, in line with a decade or more of work implicating DNA methylation as a regulator of exon inclusion or exclusion (Wen et al., 2014; Khare et al., 2012; Maunakea et al., 2013; Lev Maor et al., 2015; Shukla et al., 2011; Lopez Soto and Lipscombe, 2020).

Thus, here we assessed the methylation status of the full Ntrk2 gene in enriched populations of cortical neurons and astrocytes; work that was facilitated by recent publication of Nanopore sequencing of mouse cortical astrocytes and neurons (Wei et al., 2024). Accessing these data we evaluated 5mC and 5hmC methylation across all 330 kb of Ntrk2 in both cell types. This analysis revealed no differences in Ntrk2 promoter methylation yet, thousands of differentially DMSs and DhMSs across the rest of the gene with enrichment of these epigenetic marks flanking CDS12, the TrkB.T1 specific isoform. Notably, 100% of DhMSs were hypermethylated in astrocytes relative to neurons, and correspondingly neurons demonstrated higher 5mC methylation. While correlative, these data suggest DNA methylation patterns may provide instruction for isoform specific TrkB expression across unique CNS cell types. Specifically, the high levels of 5mC and low levels of 5hmC in neurons may lead to the exclusion of CDS12, while the low levels of 5mC and high levels of 5hmC in astrocytes promote the inclusion of CDS12. This finding is consistent with previous publications indicating that DNA methylation levels are negatively correlated with TrkB.T1 expression levels (Ernst et al., 2009; Maussion et al., 2014). We observed that both intron 11 and intron 12, which flank the CDS12 region, exceed 5,000 bp in length. Although DMSs/DhMSs are not significantly overrepresented in these regions when normalized to genomic length, previous studies indicate that the prevalence and extent of alternative splicing are positively correlated with the mean intron size of the genome, suggesting that longer introns may contribute to alternative splicing events (Yang et al., 2021). Additionally, the RNA-binding gene Bruno-3 (Bru-3) in Drosophila has demonstrated that all exon-skipping events in this gene occur in exons flanked by introns of at least 800 nucleotides (Kandul and Noor, 2009). This observation is consistent with our findings that CDS12, is bordered by the long introns 11 and 12 and suggests that DNA methylation differences in these regions may facilitate alternative splicing. Together, data presented here provide additional insight regarding CNS TrkB expression across mice and humans and raise the possibility that DNA methylation may be a key driver of isoform specific expression across two unique cell populations.

While our study provides insights into the role of DNA methylation and hydroxymethylation in the regulation of alternative splicing and TrkB isoform expression, several limitations should be considered. First, our analysis is based on data from the P28 mouse cortex. Given that DNA methylation is dynamic throughout development, the methylation profile for Ntrk2 could differ at other developmental stages. Second, our study utilized CO₂ euthanasia, followed by cell isolation. While there is no definitive evidence indicating that CO₂ euthanasia leads to rapid changes in the brain DNA methylome, it is important to acknowledge that DNA methylation patterns may be altered during the euthanasia and cell isolation processes, and that these changes may be cell type specific. Herein, our study corroborates previous work that DNA methylation may regulate TrkB isoform expression (Ernst et al., 2009, Maussion et al., 2014) and further supports the role of DNA methylation in the general regulation of isoform expression (Yearim et al., 2015; Ramanouskaya and Grinev, 2017; Shukla et al., 2011; Maunakea et al., 2013), yet our study does not establish causality between DNA methylation patterns and alternative splicing events of Ntrk2, necessitating further experimental validation.