Introduction

BDNF is a member of the neurotrophic family of secreted proteins and has well-established roles in the nervous system, e.g., in neuronal survival and differentiation, synapse formation and maturation, and development of neural circuits (Park and Poo, 2013; Wang et al., 2022). Moreover, BDNF is associated with numerous neuropsychiatric disorders, such as depression, bipolar disorder, schizophrenia, addiction, and various neurodevelopmental conditions (Autry and Monteggia, 2012; Wang et al., 2022). In recent years the functions of BDNF have been demonstrated in various non-neural tissues, including the heart (Donovan et al., 2000; Fulgenzi et al., 2015; Li et al., 2022), lung (Paris et al., 2020), skin (Rutlin et al., 2014), mammary gland (Liu et al., 2012; Sar Shalom et al., 2019), and skeletal muscle (Delezie et al., 2019; Ahuja et al., 2021). Furthermore, BDNF regulates blood insulin levels (Yang et al., 2019; Fulgenzi et al., 2020), and BDNF signaling is important in kidney development (García-Suárez et al., 2006; Endlich et al., 2018) and serves as a potential marker for chronic kidney disease (Afsar and Afsar, 2022). In humans, BDNF also contributes to platelet aggregation (Boukhatem et al., 2021). Taken together, BDNF signaling is important in both the nervous system and in non-neural tissues, highlighting the importance of understanding its developmental regulation.

BDNF protein is synthesized as a prepro-precursor protein into the endoplasmic reticulum, with the pre-region cleaved co-translationally, resulting in proBDNF. proBDNF is processed to mature BDNF protein either in the Golgi apparatus (Mowla et al., 1999, 2001) or extracellularly (Lee et al., 2001). Both proBDNF and mature BDNF form non-covalently associated dimers and bind two types of receptors, TrkB and p75NTR, and therefore affect cells in a diverse way (Lu et al., 2005). The TrkB receptor is a member of the tropomyosin-related kinase family of tyrosine kinases and binds either mature BDNF or NT-4 (Reichardt, 2006). Binding of neurotrophins causes TrkB dimerization and activation of Ras-MAPK, PI3K-Akt, and PLCγ1-IP3/DAG signaling pathways (Reichardt, 2006). The TrkB protein-encoding gene NTRK2 encodes several TrkB isoforms generated by alternative splicing, and the main isoforms in humans are the full-length TrkB isoform (hereafter TrkB-FL), and the C-terminally truncated isoforms TrkB-T1 and TrkB-Shc, both of which lack the tyrosine kinase domain, though the latter retains the Shc-binding site (Stoilov et al., 2002; Luberg et al., 2010). While the TrkB-T1 isoform sequesters both BDNF and TrkB-FL (Haapasalo et al., 2002) and therefore functions as a dominant-negative TrkB isoform, it can also elicit intracellular signals on its own, e.g., to regulate calcium influx in glial cells and cardiomyocytes (Tessarollo and Yanpallewar, 2022). Similarly, the TrkB-Shc isoform acts as a dominant-negative regulator of TrkB signaling as the TrkB-FL does not phosphorylate TrkB-Shc after dimerization (Stoilov et al., 2002). The p75NTR receptor is a member of tumor necrosis factor receptor superfamily and binds all neurotrophins – NGF (Sutter et al., 1979), BDNF (Rodriguez-Tebar et al., 1990), NT-3 (Rodríguez-Tébar et al., 1992), and NT-4 (Rydén et al., 1995). The p75NTR receptor preferentially binds proneurotrophins, whereas TrkB preferentially binds mature neurotrophins (Reichardt, 2006). Following proneurotrophin binding, p75NTR activates NF-kB, RhoA, and Jun kinase pathways which promote neuronal survival, inhibit neurite growth, or induce apoptosis, respectively (Reichardt, 2006). During development p75NTR signaling induces normal cell death, while in adults it triggers apoptosis after injury (Kraemer et al., 2014). In summary, the different effects of BDNF on target cells and tissues depend on the expression of specific TrkB isoforms and the p75NTR receptor.

The genes encoding neurotrophins are highly conserved throughout vertebrates, and they have been proposed to originate from a single ancestral gene (Hallböök et al., 1991; Hallböök, 1999). In murines the Bdnf gene contains eight 5′ non-coding exons that are controlled by distinct promoter regions and are spliced to the common protein-coding 3′ exon (Timmusk et al., 1993b; Aid et al., 2007). The activity-dependent subcellular localization of Bdnf transcripts in neurons has been thoroughly studied (Tongiorgi et al., 1997; Pattabiraman et al., 2005; Chiaruttini et al., 2008, 2009; Baj et al., 2013, 2016; Singer et al., 2018; Colliva and Tongiorgi, 2021). Moreover, studies in mice have shown that different Bdnf transcripts regulate dendrite complexity (Maynard et al., 2017) and impairing the expression of specific Bdnf transcripts has distinct functional consequences. For example, loss of Bdnf exon I-containing transcripts results in increased body weight, reduced thermogenesis (You et al., 2020), heightened aggression in male mice (Maynard et al., 2016), and impaired maternal care (Maynard et al., 2018). In contrast, loss of Bdnf exon IV transcripts affects the development of inhibitory synapses (Hong et al., 2008; Sakata et al., 2009), disrupts sleep and sensory information processing, and impairs fear memory retrieval (Hill et al., 2016).

Here, we aimed to evaluate the potential role of BDNF throughout mammalian development by revisiting the expression of BDNF using bioinformatical analysis of published RNA sequencing datasets. First, we analyzed the spatiotemporal expression of BDNF mRNA, including transcripts with different 5′ exons and 3′ untranslated regions (UTRs), in neural and non-neural tissues of different mammals. We also examined the expression of BDNF receptors TrkB and p75NTR. We then focused on the cell type-specific BDNF expression in adult mice and humans using available single-cell sequencing data. Finally, we complemented the bioinformatics data on BDNF expression at the protein level in the nervous system and non-neural tissues during development of three widely used murine animal models: BALB/c and C57BL/6 J mice and Wistar rats. Altogether, the comprehensive description of the expression patterns of BDNF and its receptors across different tissues and developmental stages provides a highly valuable resource to the neurotrophin research community.

Materials and methods Bioinformatical analysis

Raw data in fastq format from previously published RNA sequencing datasets (Keane et al., 2011; ENCODE Project Consortium, 2012; Merkin et al., 2012; Fagerberg et al., 2014; Yu et al., 2014; Vied et al., 2016; Li et al., 2017; Söllner et al., 2017; Tabula Muris Consortium, Overall coordination, Logistical coordination, Organ collection and processing, Library preparation and sequencing, Computational data analysis, 2018; Cardoso-Moreira et al., 2019; Luo et al., 2020; Shafik et al., 2021; Rayan et al., 2022), were obtained from EMBL-EBI European Nucleotide Archive database using www.sra-explorer.info. Accession numbers for all datasets are shown in Supplementary Table 1.

Adapter and quality trimming were performed using BBDuk (part of BBMap version 38.90) with the following parameters: ktrim = r k = 23 mink = 11 hdist = 1 tbo qtrim = lr trimq = 10 maq = 10 minlen = 25. Mouse sequencing reads were mapped to mm10 (primary assembly and annotation obtained from GENCODE, release M25, GRCm38), rat sequencing reads were mapped to rn6 (primary assembly and annotation obtained from Ensembl, release 104, RGSC 6.0/Rnor_6.0), human sequencing reads were mapped to hg19 (primary assembly and annotation obtained from GENCODE, release 37, GRCh37), rhesus macaque sequencing reads were mapped to Mmul_10 (primary assembly and annotation obtained from Ensembl, release 108, Mmul_10), rabbit sequencing reads were mapped to OryCun2.0 (primary assembly and annotation obtained from Ensembl, release 108, OryCun2.0) and opossum sequencing reads were mapped to ASM229v1 genome (primary assembly and annotation obtained from Ensembl, release 108, ASM229v1) using STAR aligner (version 2.7.4a) with default parameters. To increase sensitivity for unannotated splice junctions, splice junctions obtained from the 1st pass were combined per dataset and filtered as follows: junctions on mitochondrial DNA and non-canonical intron motifs were removed; only junctions detected in at least 10% of samples (rounded up to the nearest integer) in the whole dataset were kept. Filtered junctions were added to the 2nd pass mapping using STAR aligner. RNA sequencing reads were assigned to features using FeatureCounts (version 2.0.1). The following parameters were used for paired-end data: -p –B –C –J; and single-end data: -J. To count 3’ UTR/exon sequencing reads for BDNF and NTRK2, a custom SAF file was used (Supplementary Table 2 Custom SAF files). To characterize the total BDNF, 5′ exon and 3’ UTR levels throughout development, NTRK2 3′ exons and NGFR levels, counts per million (CPM) of the assigned RNA sequencing reads were calculated, and normalized with the length of the feature in kb where indicated. To analyze changes in BDNF 3′ long UTR proportions during development, CPM of the long 3’ UTR region was length-normalized and divided with the length-normalized CPM of BDNF coding sequence (CDS), therefore showing the levels of transcripts with long 3’ UTR from the total BDNF. The UCSC Liftover tool was used to convert the coordinates of BDNF coding sequence, 5′ exons and 3’ UTRs from data of human to rhesus macaque, and from mouse to rabbit and opossum genomes.

Meta-analysis of both mouse and rat data was performed to study Bdnf, Ntrk2 and Ngfr expression during development. For analyzing the expression of unique Bdnf 5′ exons, length-normalized 5′ exon CPM values were further normalized with length-normalized Bdnf CDS CPMs for the corresponding samples. To visualize the distribution of transcripts with Bdnf 5′ exons, sums of the normalized 5′ exon values were calculated in each group and divided by the total sum of all normalized 5′ exon values in the group, showing the composition in percentages. The results were visualized using ggplot2 (version 3.3.5) in R (version 4.1.2).

Data mining and visualization was also performed on human Genotype-Tissue Expression project (GTEx) portal gene and exon datasets (dbGaP Accession phs000424.v8.p2), human developmental transcriptome data from BrainSpan (RNA-Seq Gencode v10 summarized to genes and RNA-Seq Gencode v10 summarized to exons), and The Human Protein Atlas mouse transcript (24. RNA isoform data) and brain subregion gene data (16. RNA mouse brain subregion sample gene data). The human GTEx data used for the analyses were obtained from the GTEx Portal on 12/01/2021, the human BrainSpan data were obtained from the BrainSpan Atlas of the Developing Human Brain on 12/01/2021 and the mouse data from The Human Protein Atlas (see footnote 1) were obtained on 13/09/2021. For the pre-analyzed datasets, the level of each 5′ exon was normalized using the sum of all annotated 5′ exons or transcripts containing the respective 5′ exons in the respective sample to calculate the 5′ exon ratios. Then the distribution of 5′ exons usage in BDNF transcripts was visualized as described previously.

Mimotope variation analysis

To determine the epitope of 3C11 anti-BDNF monoclonal antibody (Icosagen, catalogue #327–100), mimotope-variation analysis was used as described previously (Sadam et al., 2018, 2021). Raw peptide counts in each sample were normalized with total peptide count. The 12-mer peptides obtained from the MVA workflow or from sequencing of the input E. coli M13 phage library were then aligned to different mouse, rat and human neurotrophin protein sequences (obtained from Uniprot) matching in at least 6 amino acid positions. Alignment loads were calculated as sum of normalized peptide counts aligning to each of the amino acid positions, with each amino acid position counted only once per unique peptide. To reduce the effect of spurious alignment of high-count peptides, the peptide with the highest normalized count was removed from the analysis for each amino acid position. Relative antibody binding at each amino acid position was calculated as fold over the input library alignment load.

Protein lysates

Animal work was performed as published in Sirp et al. (2022). All experiments concerning animals were performed in agreement with the local ethics committee and European Directive 2010/63/EU. Briefly, BALB/c mouse strain (Envigo), C57BL/6 J (Envigo), and Wistar rats (RccHan:WI, Envigo) were housed in conventional polycarbonate or H-TEMP polysulfone cages (2–4 animals per cage) with ad libitum access to clean water and food pellets (ssniff Spezialdiäten, GmbH) under a 12-h light/dark cycle in humidity and temperature-controlled room (temperature 22 ± 1°C and humidity 50 ± 10%).

The female mouse estrous cycle was monitored by visual observations (Champlin et al., 1973) followed by breeding in the evening. The presence of vaginal post-coitum protein plug was confirmed the next morning (no later than 12 h from breeding) and determined as embryonic 0.5 gestational stage. To obtain samples from embryonic stages, the pregnant mothers were euthanized by carbon dioxide inhalation, pups were collected, and embryonic brains were dissected in ice-cold 1× phosphate-buffered saline (PBS) solution. Postnatal (P) 0 stage was determined as the day of the birth and all other ages followed accordingly. Samples were collected from both female and male animals. To collect postnatal brain samples, rats and mice were killed by cervical dislocation and decapitated with a guillotine.

Protein lysates were prepared as described in Sirp et al. (2022). Briefly, tissue samples were dissected in ice-cold 1× phosphate-buffered saline (PBS) and stored at −80°C until further use. For each investigated time point, tissue samples from 2 to 3 different animals were combined and homogenized in ice-cold RIPA buffer [50 mM Tris–HCl (pH 8.0), 150 mM NaCl, 1% NP-40, 0.5% Na-deoxycholate, 0.5% sodium dodecyl sulfate (SDS), 1× cOmplete™ Protease Inhibitor Cocktail (Roche)]. For brain and non-neural tissues 7 and 10 μL RIPA per 1 mg of tissue, respectively, was used. All samples were homogenized using tissue grinder PELLET PESTLE® Cordless Motor (Kimble-Chase, DWK Life Sciences), sonicated 15 s with Torbeo Ultrasonic probe sonicator (36810-series, Cole Parmer), and centrifuged at 4°C at 16,000 g for 20 min. Soluble fraction was kept as protein lysate and protein concentration was measured using Pierce™ BCA Protein Assay Kit (Thermo Scientific).

Western blot

Fifty micrograms of total protein and different amounts of recombinant mature BDNF protein (Icosagen, cat. no P-105-100) were separated on 15% SDS-PAGE gel and transferred to a PVDF membrane using Trans-Blot Turbo Transfer system (Bio-Rad, Mixed MW program [1,3A, 25 V (const), 7 min)]. The membrane was blocked for 1 h at room temperature in 5% skimmed milk in TBST buffer (1× Tris-buffered saline (pH 7.4) and 0.1% Tween-20), incubated overnight at 4°C with primary anti-BDNF antibody (Icosagen, cat. no 327–100, clone 3C11, 1 mg/mL, 1:1000) in 2% milk-TBST, and overnight at 4°C with secondary antibody anti-mouse IgG conjugated with horseradish peroxidase (Thermo Fisher Scientific, 1:5000) in 2% milk-TBST. After both incubations, the membrane was washed 3 times with TBST for 5 min at room temperature.

Chemiluminescence signal was produced with SuperSignal™ West Femto or Atto Maximum Sensitivity Substrate (Thermo Fisher Scientific) and measured using imaging system ImageQuant Las 4,000 (GE Healthcare Life Sciences). For loading control, the membrane was stained with Coomassie solution (0.1% Coomassie Brilliant Blue R-250 Dye, 25% ethanol, 7% acetic acid), followed by washes with destaining solution (30% ethanol, 10% acetic acid) and rinsing with tap water. The membrane was imaged using ImageQuant Las 4,000 (GE Healthcare Life Sciences).

BDNF protein levels were quantified using densitometric analysis on ImageQuant TL software (GE Healthcare Life Sciences). As indicated in the figure legend, the amounts of BDNF protein in the tissue lysates were calculated based on the calibration curve of recombinant BDNF protein.

CRISPR interference-mediated silencing of Bdnf expression

The preparation and growing of Sprague Dawley rat primary cortical neurons was performed as described in Esvald et al. (2022). The used guide RNA (gRNA) sequences targeting either Bdnf promoters I or IV (pI or pIV), or control gRNA were as follows: Bdnf pI gRNA 1 5’-GTCACGTAACTGGCTCAGAG-3′, Bdnf pI gRNA 2 5’-GCCCTAGCCTGACAAGGCGA-3′, Bdnf pIV gRNA 1 5’-GCACTAGAGTGTCTATTTCG-3′, Bdnf pIV gRNA 2 5’-GATTTCATGCTAGCTCGCCG-3′, control gRNA 5’-GCTGATCTATCGCGGTCGTC-3′. Lentiviruses encoding the gRNAs or dCas9-KRAB were generated as described in Esvald et al. (2022). Neurons were infected with the indicated lentiviruses on the day of plating, half of culture media was changed at 2 days in vitro (DIV) and 5 DIV. A final concentration of 10 μM FDU (Sigma) was added to the media from 2 DIV. At 7 DIV, spontaneous neuronal activity was suppressed by adding 1 μM tetrodotoxin (Tocris Bioscience), and the cells were treated with 25 mM KCl (with 5 μM D-APV (Cayman Chemical Company) to reduce excitotoxicity) for 6 h at 8 DIV. RNA extraction, cDNA synthesis and qPCR, and protein extraction were conducted as described previously in Esvald et al. (2022). Protein concentration was measured using Pierce BCA Protein Assay kit (Thermo Scientific) and 15 μg of total protein was loaded on the gel along with bacterially expressed recombinant proBDNF (Alomone labs, cat. no B-257) and recombinant mature BDNF (Icosagen).

Overexpression of BDNF-V5 in HEK293 cells

Rat Bdnf coding sequence was cloned into pcDNA3.1/V5-His-TOPO expression vector (Life Technologies) where the CMV promoter was replaced with EF1α promoter. HEK293 cells were grown in Minimum Essential Medium (MEM, Corning, cat. no 10-010-CV) supplemented with 10% fetal bovine serum (PAN Biotech), 100 U/mL penicillin, and 100 μg/mL streptomycin (Gibco). The cells were transfected on 12-well plate with PEI (Sigma) with 1.5 μg DNA per well using 1:2 DNA:PEI ratio. Cells were lysed 24 h post transfection directly into 1x Laemmli buffer (containing 5% β-mercaptoethanol) and proteins were subjected to SDS-PAGE and Western blot using anti-V5 antibody (Thermo Fisher Scientific, #R960-25, 1:5000) or anti-BDNF 3C11 antibody.

Results Bdnf mRNA levels in murine brain during development

As BDNF is widely studied in murines and it plays an important role in the developing central nervous system, we set out to first describe the spatiotemporal expression pattern of Bdnf in mouse and rat development using various publicly available RNA sequencing datasets (listed in Supplementary Table 3). Our analysis shows that total Bdnf mRNA levels increase remarkably during development in the whole brain of both mouse (Figure 1A) and rat (Figure 1B). Of the analyzed tissues, the highest levels of Bdnf mRNA were detected in the adolescent and adult mouse hippocampus (Figure 1A) and in the adolescent rat cerebral cortex (Figure 1B).

Figure 1. The expression levels of mRNAs encoding BDNF and its receptors TrkB (Ntrk2 mRNA) and p75NTR (Ngfr mRNAs) during murine development in different brain regions. Meta-analysis of Bdnf, Ntrk2 (encoding TrkB) and Ngfr (encoding p75NTR) expression levels in mouse (A,C,E) and rat (B,D,F) brain regions throughout development. Levels in embryo (E10.5-E14.5/E11-E15 mouse/rat), late embryo (E15.5-E18.5/E17-E20 mouse/rat), postnatal (P0-P14), adolescent (P22-56) and adult (P62+) animals are shown. (A,B) Total Bdnf expression measured by levels of Bdnf coding sequence (CDS, upper panel) and distribution of Bdnf 5′ exons (lower panel) are shown as depicted on the schematics of murine gene structure with different colors. The exons indicated with gray color were not included in the analysis of 5′ exons. (C,D) The mRNA levels of different TrkB isoforms based on the levels of unique 3′ exons of the Ntrk2 gene (rat or mouse counterpart of human NTRK2 exons 16, 19, 24). The 3′ exons specific for TrkB isoforms TrkB-T1, TrkB-Shc, and TrkB-FL were measured as shown with colors on the schematics [adapted from Luberg et al. (2010)]. Asterisks on the schematics mark stop-codons. (E,F) Total Ngfr mRNA expression levels. Data from individual animals are shown as small dots, circles indicate mean values and error bars represent standard error of the mean (SEM). All used datasets and underlying data are shown in Supplementary Table 3. WB – whole brain, CTX – cerebral cortex, CB – cerebellum, HC – hippocampus, HTH – hypothalamus, MB – midbrain, NT – neural tube, CDS – coding sequence, CPM – counts per million.

We then analyzed the developmental expression of Bdnf transcripts which differ in their 5′ exons. We detected all main Bdnf transcripts, with most transcripts containing 5′ exons IV and VI, and a lower proportion of transcripts containing exons I, IIc, and III (Figures 1A,B). Bdnf exon I and IIc-containing mRNAs contribute more to total Bdnf mRNA levels in the mouse cerebral cortex and hypothalamus and less in the hippocampus (Figure 1A), which is in agreement with previously published contribution of different Bdnf transcripts to BDNF protein levels in these brain regions (Maynard et al., 2016). Notably, the relative proportion of Bdnf exon I and IIc transcripts generally increases and proportion of Bdnf exon VI transcripts slightly decreases during mouse and rat development (Figures 1A,B). In contrast to other brain regions, the proportion of Bdnf exon I-containing transcripts decreases during postnatal cerebellar development in both mouse and rat (Figures 1A,B). Taken together, our results show developmental upregulation of total Bdnf mRNA levels and specific regulation of Bdnf transcripts in different parts of the central nervous system.

Bdnf mRNA levels in adult brain in different mouse strains

Next, we aimed to determine whether Bdnf levels are similar in different laboratory mouse strains, and we found that the levels of Bdnf in the whole brain and hippocampi of adult mice are relatively stable between the strains (Supplementary Figure S1A, Supplementary Table 4). The proportions of different Bdnf transcripts are roughly consistent between different mouse strains, although the hippocampus of PWD/Ph strain exhibits higher proportion of exon I and IIc-containing transcripts than the other analyzed strains (Supplementary Figure S1A). Despite some fluctuations, total Bdnf mRNA levels and proportions of transcripts are similar between mouse strains.

Bdnf mRNA levels in different brain regions of adult mouse and rat

A comprehensive analysis of different brain regions in adult mouse shows that total Bdnf mRNA levels vary substantially, with the highest levels of Bdnf mRNA found in the hippocampus and different cortical regions, and the lowest in the striatum (caudate putamen; Supplementary Figure S2A; Supplementary Table 5). Bdnf exon I-, IIc-, IV- and VI-containing transcripts are the major Bdnf transcripts and are expressed in different brain regions, with Bdnf exon I and IV generally accounting for over half of the Bdnf transcripts (Supplementary Figure S2A). Interestingly, the proportion of exon I and IIc-containing transcripts is the highest in the septum, amygdala, thalamus, hypothalamus, midbrain, and pons and medulla, where these transcripts form approximately half of the total Bdnf mRNA pool (Supplementary Figure S2A). In contrast, retina, pituitary gland, and cerebellum almost completely lack Bdnf exon I-containing transcripts (Supplementary Figure S2A).

Similarly, in the adult rat central nervous system the highest total Bdnf mRNA levels are in the hippocampus and different cortical regions (Supplementary Figure S3A; Supplementary Table 6). Major Bdnf transcripts contain exons I, IIc, IV, or VI, however, exon III also remarkably contributes to the total Bdnf pool in many brain regions (Supplementary Figure S3A). It is worth noting that in this dataset, Bdnf exon VI-containing transcripts do not contribute significantly to total Bdnf mRNA levels in different cortical regions (Supplementary Figure S3A). Altogether, our results show that total Bdnf mRNA levels and proportion of transcripts vary between brain regions in both adult mouse and rat.

TrkB mRNA expression levels in mouse and rat brain during development

Next, we aimed to revisit the expression of different Ntrk2 mRNAs encoding different TrkB protein isoforms (hereafter referred to as TrkB mRNAs). Our analysis shows that both full-length TrkB (hereafter TrkB-FL) and truncated TrkB-T1 mRNA levels increase remarkably in both mouse (Figure 1C) and rat (Figure 1D) whole brain during development. In mouse the mRNA levels of TrkB-T1 and TrkB-FL increase and are mostly similar throughout development, except in the hypothalamus and adult mouse cerebellum where the TrkB-T1 mRNA levels are much higher than the levels of TrkB-FL (Figure 1C). Interestingly, in rat the expression levels of TrkB-T1 isoform reach higher levels than the full-length isoform in both cerebral cortex and cerebellum during development (Figure 1D). Notably, the mRNA levels of TrkB-Shc isoform are almost undetectable in the studied mouse and rat brain regions (Figures 1C,D). In different adult mouse and rat brain regions, TrkB-T1 mRNA levels are consistently higher than full-length TrkB isoforms (Supplementary Figures S2B, S3B). Collectively, our results show upregulation of TrkB isoform mRNA levels during murine development, with higher levels of TrkB-T1 than TrkB-FL.

The mRNA levels of different TrkB isoforms vary in different mouse strains. For example, the mRNA levels of TrkB-T1 and TrkB-FL isoforms are roughly equal in the whole brain of adult BALB/cJ and C57BL6/6NJ mice, whereas TrkB-T1 is expressed at a higher level than TrkB-FL in adult NZO, NOD/Ltj, and 129S1 mice (Supplementary Figure S1B). The mRNA levels of different TrkB isoforms are more stable in the hippocampus of different mouse strains (Supplementary Figure S1B). In summary, our results reveal that BDNF signaling may differ between mouse strains due to the varying expression of TrkB isoforms.

p75NTR mRNA expression levels in mouse and rat brain during development

Next, we analyzed the expression levels of the other BDNF receptor, p75NTR. Our results show that the expression of the Ngfr gene (encoding p75NTR, hereafter referred to as p75NTR mRNA) mainly decreases in both mouse (Figure 1E) and rat (Figure 1F) development, implying that TrkB, whose expression shows the opposite trend, is the main receptor of BDNF in the adult brain. The mRNA levels of p75NTR are low and slightly differ between mouse strains in the whole brain but are more similar in the hippocampi (Supplementary Figure S1C). Interestingly, in the adult mouse brain the levels of p75NTR are generally very low, with the highest expression seen in the retina (Supplementary Figure S2C), where TrkB isoform mRNAs are expressed at very low levels (Supplementary Figure S2B). In adult rat brain regions, the levels of p75NTR are also low, with the highest levels detected in the medial preoptic area (Supplementary Figure S3C). Taken together, our results show a developmental decrease in the levels of p75NTR in the murine central nervous system.

Expression levels of total Bdnf mRNA and different transcripts in mouse and rat non-neural tissues during development

We next focused on describing the expression of Bdnf and its receptors in non-neural tissues. Our analysis shows that during murine development the total Bdnf mRNA levels increase in mouse heart and decrease in testis (Figure 2A; Supplementary Table 7), whereas in rat total Bdnf mRNA levels strongly increase in the heart, but decrease in the liver, kidney, testis, and ovary to almost undetectable levels by adulthood (Figure 2B). In contrast to mouse, Bdnf mRNA levels in rat kidney decrease during development (Figure 2B).

Figure 2. The expression levels of mRNAs encoding BDNF and its receptors TrkB (Ntrk2 mRNAs) and p75NTR (Ngfr mRNAs) during development in murine non-neural tissues. Meta-analysis of Bdnf, Ntrk2 (encoding TrkB) and Ngfr (encoding p75NTR) expression in mouse (A,C,E) and rat (B,D,F) non-neural tissues during development. The expression levels in embryo (E10.5-E14.5/E11-E15 mouse/rat), late embryo (E15.5-E18.5/E17-E20 mouse/rat), postnatal (P0-P14), adolescent (P22-56) and adult (P62+) animals are shown. (A,B) Total Bdnf expression measured by levels of Bdnf coding sequence (CDS, upper panel) and distribution of levels of Bdnf 5′ exons (lower panel) are shown as depicted on the schematics of murine gene structure. The exons indicated with gray color were not included in the analysis of 5′ exons. White box in the proportion of Bdnf 5′ exons indicates that the expression of Bdnf was too low for this calculation. (C,D) The mRNA levels of different TrkB isoforms based on the levels of unique 3′ exons of the Ntrk2 gene (rat or mouse counterpart of human exon 16, 19, 24). The 3′ exons specific for TrkB isoforms TrkB-T1, TrkB-Shc, and TrkB-FL are shown on the schematics (adapted from Luberg et al. (2010)). Asterisks on the schematics mark stop-codons. (E,F) Total Ngfr mRNA expression levels. Data from individual animals are shown as small dots, circles indicate mean values and error bars represent standard error of the mean (SEM). All used datasets and underlying data are shown in Supplementary Table 7. CDS – coding sequence, CPM – counts per million.

We next focused on the expression of different Bdnf 5′ transcripts and determined that Bdnf exon VI-containing transcripts are the major Bdnf transcripts in murine non-neural tissues, contributing to over 75% of total Bdnf transcripts throughout development (Figures 2A,B). Interestingly, Bdnf exon IV transcripts are expressed, although at low levels, in both mouse and rat heart (Figures 2A,B). Also, Bdnf exon I and IIc-containing transcripts are present in the testis during early development in both mouse and rat (Figures 2A,B). Collectively, our results show developmentally regulated levels of Bdnf mRNA in certain murine non-neural tissues where the majority of Bdnf transcripts contain exon VI.

mRNA levels of TrkB isoforms and p75NTR in mouse and rat non-neural tissues during development

We next investigated mRNA levels of different TrkB isoforms during the development of mouse and rat in non-neural tissues. The major TrkB isoform in murine non-neural tissues is TrkB-T1, while TrkB-FL and TrkB-Shc isoforms are almost undetectable (Figures 2C,D). In the kidney, testis, and ovary, the expression of TrkB-T1 isoform is high in early developmental stages and decreases during development (Figures 2C,D). Our analysis also shows that TrkB-T1 isoform is expressed at very low levels in both mouse and rat heart and liver (Figures 2C,D). Similar expression patterns are observed for p75NTR mRNA levels (Figures 2E,F). Altogether, our results show developmental downregulation of TrkB-T1 isoform and p75NTR in murine non-neural tissues.

Bdnf, TrkB, and p75NTR mRNA levels in various tissues in adult murines

We next conducted a meta-analysis of adult mouse and rat neural and non-neural tissues, combining data from various sources (see Supplementary Table 8 for details). Compared to non-neural tissues, Bdnf expression levels are consistently higher in the murine brain (Supplementary Figures S4A,B; Supplementary Table 8). Among the non-neural tissues that significantly express Bdnf, the heart and lung show similar Bdnf mRNA levels to those found in the brain. Other non-neural tissues that express Bdnf include the stomach, kidney, and heart in mouse (Supplementary Figure S4A) and the esophagus in rat (Supplementary Figure S4B). While the dominant Bdnf transcripts in both mouse and rat brain contain exons I, IIc, IV and VI, in most non-neural tissues the major Bdnf transcripts contain exon VI (Supplementary Figures S4A,B). Transcripts containing Bdnf exons I, IIc, and IV also contribute to total Bdnf mRNA levels in adult mouse skin, bone, and adipose tissue (Supplementary Figure S4A). In summary, our results show that some adult murine non-neural tissues express total Bdnf mRNA at comparable levels to those in the brain and the majority of Bdnf transcripts contain exon VI.

Interestingly, while both TrkB-FL and TrkB-T1 isoforms are expressed in adult murine brain, only TrkB-T1 mRNA is expressed in the non-neural tissues – in mouse lung, skin, adipose tissues, esophagus, ovary, and adrenal gland (Supplementary Figure S4C) and in rat lung, esophagus, ileum, and spleen (Supplementary Figure S4D). p75NTR mRNA levels are expressed highest in mouse in skin, mesenteric adipose tissue, spleen, ovary, and testis, and the levels in other tissues show similar low expression levels as in the brain (Supplementary Figure S4E). In rat the p75NTR mRNA levels are highest in the cerebellum, thymus, spleen, and various parts of the digestive system (Supplementary Figure S4F). Overall, our results show TrkB-T1 and p75NTR expression in some adult murine non-neural tissues.

Bdnf, TrkB, and p75NTR mRNA levels in various mouse tissues during aging

Finally, we analyzed Bdnf, TrkB, and p75NTR mRNA levels in aging mouse. In the whole brain and most studied mouse non-neural tissues, total Bdnf mRNA levels are relatively stable throughout adult life, except for gonadal adipose tissue and lung, where the total Bdnf mRNA levels increase during aging (Supplementary Figure S5A; Supplementary Table 9). In non-neural tissues, Bdnf transcripts mainly contain exon VI, albeit in heart a very consistent expression of both Bdnf exon IV- and VI-containing transcripts is observed (Supplementary Figure S5A). During mouse aging the proportion of Bdnf exon I- and IIc-containing transcripts increases in mesenteric adipose tissue and decrease in gonadal adipose tissues (Supplementary Figure S5A). In the brain, TrkB-T1 and TrkB-FL isoforms are both stably expressed throughout life (Supplementary Figure S5B). Of the analyzed non-neural tissues, skin, adipose tissues, and lung show expression of TrkB-T1 isoform, where no major changes in the expression during aging are seen (Supplementary Figure S5B). p75NTR is expressed in the skin, different adipose tissues, liver, and spleen, with the expression levels increasing in the skin and spleen and decreasing in the mesenteric adipose tissues during aging (Supplementary Figure S5C). Collectively, Bdnf and TrkB mRNA levels do not change notably within one tissue during mouse aging.

BDNF, TRKB, and P75NTR mRNA expression levels in human brain during development

Next, we aimed to determine the expression pattern of BDNF and its receptors in humans. Compared to the murine BDNF gene structure, the human BDNF gene is more complex and has two additional 5′ exons (Vh and VIIIh) (Pruunsild et al., 2007). Overall, BDNF transcripts are generated similarly in humans and murines, however, in rare cases multiple 5′ exons are used in humans (e.g., BDNF exons VIII and VIIIh can be used as internal exons) (Pruunsild et al., 2007). Our analysis shows that total BDNF mRNA levels rise in different brain regions during human development (Figure 3A; Supplementary Table 10), reaching the highest levels in the adult hippocampus and infant thalamus (Figure 3A). In the striatum almost no BDNF mRNA can be detected (Figure 3A). Taken together, our results show developmental upregulation of BDNF mRNA levels in the human brain.

Figure 3. The expression levels of mRNAs encoding BDNF and its receptors TRKB (NTRK2 mRNAs) and p75NTR (NGFR mRNAs) in human brain regions throughout development. Visualization of BDNF, NTRK2 (encoding TRKB) and NGFR (encoding P75NTR) expression data from Cardoso-Moreira et al. (2019) (left panels) and human developmental transcriptome data from BrainSpan project (right panels). The expression levels in early embryo [0–9 postcoital week (PCW)], embryo (10–19 PCW), late embryo (20–39 PCW), infant (younger than 12 months), toddler (1–4 years old), school age (7–8 years old), adolescent (10–17 years old), young adult (18–29.99 years old), adult (30–39.99 years old), and middle-aged (40–58 years old) humans are shown. (A) Total BDNF expression measured by levels of BDNF coding sequence (CDS, upper panel) and distribution of levels of BDNF 5′ exons (lower panel) are shown as depicted on the schematics of human gene structure. The exons indicated with gray color were not included in the analysis of 5′ exons. (B) The mRNA levels of different TrkB isoforms based on the levels of unique 3′ exons of NTRK2 gene (exon 16, 19, 24). The 3′ exons specific for TRKB isoforms TRKB-T1, TRKB-SHC, and TRKB-FL are shown on the scheme (adapted from Luberg et al. (2010)). Asterisks on the schematics mark stop-codons. (C) Total NGFR mRNA expression levels. Data from individual animals are shown as small dots, circles indicate mean values and error bars represent standard error of the mean (SEM). All used datasets and underlying data are shown in Supplementary Table 10. AMY – amygdaloid complex, CB – cerebellar cortex, DFC – dorsolateral prefrontal cortex, HC – hippocampus (hippocampal formation), MD – mediodorsal nucleus of thalamus, STR – striatum, CDS – coding sequence, CPM – counts per million.

Our analysis further shows that BDNF exon IV-containing transcripts are the most ubiquitously expressed in different brain regions, while the proportions of BDNF exon VI-containing transcripts are low (Figure 3A). Notably, in the hindbrain and cerebellum BDNF exon IIc transcripts comprise a remarkable proportion of total BDNF mRNA, and this proportion increases during development, whereas the proportion of BDNF exon I transcripts is very low in these brain regions (Figure 3A). In contrast, in the mediodorsal nucleus of thalamus, BDNF exon I-containing transcripts notably contribute to the total BDNF mRNA pool throughout development and a high proportion of exon I transcripts is also seen in the amygdala and adult hippocampus (Figure 3A). Overall, it appears that BDNF exon IV-containing transcripts are the main transcripts in most human brain regions (except the cerebellum), whereas other transcripts show brain region-specific expression patterns.

Next, we analyzed the mRNA levels of TRKB and P75NTR. Our analysis shows that TRKB-T1 is the main isoform expressed in most brain regions (except the cerebellum) across all developmental stages (Figure 3B). Furthermore, TRKB-T1 mRNA levels remarkably decrease in the adult hippocampus, thalamus, and striatum compared to levels in young adults (Figure 3B). Interestingly, the mRNA levels of TRKB-FL isoform increase in the forebrain and dorsolateral prefrontal cortex during development (Figure 3B). In other studied brain regions, the mRNA levels of TRKB-FL are low and TRKB-SHC expression is almost undetectable (Figure 3B). Finally, we focused on P75NTR mRNA levels in the human brain and determined that it is expressed at low levels, except in the embryonic hindbrain and cerebellum, which showed high expression levels (Figure 3C). Our results suggest that in adult humans the main BDNF receptor in the brain is TRKB.

BDNF, TRKB, and P75NTR mRNA levels in human non-neural tissues during development

Next, we investigated BDNF expression in human non-neural tissues. Our analysis shows that BDNF is expressed at low levels and its expression decreases in the heart, kidney, and ovary, and increases in the testis during development (Figure 4A; Supplementary Table 11). In adult human, BDNF mRNA expression is highest in the nervous system, but also in the arteries and heart, prostate, and lung (Supplementary Figure S6A; Supplementary Table 12). BDNF exon IV and VI-containing transcripts are the main BDNF transcripts expressed in human non-neural tissues (Figure 4A; Supplementary Figure S6A).

Figure 4. The expression levels of mRNAs encoding BDNF and its receptors TRKB (NTRK2 mRNAs) and p75NTR (NGFR mRNAs) in human non-neural tissues. The expression levels of BDNF, NTRK2 (encoding TRKB), and NGFR (encoding P75NTR) mRNAs in early embryo [0–9 postcoital week (PCW), embryo (10–19 PCW), late embryo (20–39 PCW), infant (younger than 12 months), toddler (1–4 years old), school age (7–8 years old), adolescent (10–17 years old), young adult (18–29.99 years old), adult (30–39.99 years old), and middle-aged (40–58 years old) humans are shown (data from Cardoso-Moreira et al. (2019)]. (A) Total BDNF mRNA levels measured by levels of BDNF coding sequence (CDS, upper panel) and distribution of levels of BDNF 5′ exons FIGURE 4 (Continued)(lower panel) are shown as depicted on the schematics of human BDNF gene structure. The exons indicated with gray color were not included in the analysis of 5′ exons. (B) The mRNA levels of different TRKB isoforms based on the levels of unique 3′ exons of NTRK2 gene (exon 16, 19, 24). The 3′ exons specific for TRKB isoforms TRKB-T1, TRKB-SHC, and TRKB-FL are shown on the scheme (adapted from Luberg et al. (2010)). Asterisks on the scheme mark stop-codons. (C) Total NGFR mRNA expression levels. All used datasets and underlying data are shown in Supplementary Table 11. Data from individual animals are shown as small dots, circles indicate mean values and error bars represent standard error of the mean (SEM). CDS – coding sequence, CPM – counts per million.

The BDNF receptor TRKB-T1 is the only TRKB isoform that is expressed in human non-neural tissues, with the highest levels in embryonic kidney, but found at lower levels also in the heart, testis, and ovary (Figure 4B). During development, the levels of TRKB-T1 decrease in the kidney, testis, and ovary (Figure 4B). Interestingly, in adult human TRKB-T1 is also expressed in non-neural tissues where BDNF is expressed, e.g., in arteries, mammary tissue, adipose tissue, and thyroid (Supplementary Figure S6B). Finally, P75NTR is expressed in the heart, liver, and ovary at similar and stable levels throughout development, while the expression of P75NTR decreases during development in the kidney and testis (Figure 4C). In adult human, the mRNA expression of P75NTR is notably higher in non-neural tissues than in different brain regions, and surprisingly shows the highest expression in the tibial nerve (Supplementary Figure S6C).

Evolutionary analysis of BDNF gene expression in mammalian development

Next, to study whether the expression of BDNF is evolutionarily conserved among mammals, we focused on BDNF mRNA levels and proportions of its transcripts throughout mammalian development. Our results show that during development total BDNF mRNA levels generally increase and reach similar levels in human, rhesus macaque, mouse, rat, and rabbit forebrain and hindbrain, but reach remarkably higher levels in the adult opossum forebrain and hindbrain (note different scale for opossum in the figure) (Figure 5; Supplementary Table 13). Although total BDNF mRNA levels are quite similar in the forebrain of different mammals, we noted significant differences in the expression of different BDNF transcripts (Figure 5A). In human and rhesus macaque forebrain, vast majority (> 75%) of BDNF transcripts are exon IV mRNAs, followed by lower levels of exon I and IIc mRNAs and almost no contribution by exon VI mRNAs (Figure 5A). Surprisingly, in mouse and rat forebrain, all Bdnf 5′ exons contribute to Bdnf mRNA levels (Figure 5A). Although exon VI mRNAs are also prevalent in the early development of rabbit and opossum forebrain, the proportion of exon VI-containing transcripts drops quickly during development to almost undetectable levels in the adult organism. It appears that the proportion of transcripts arising from the first cluster of BDNF exons (exon I, IIc, and III) has decreased during mammalian evolution, with the highest levels seen in opossum and lowest in primates (Figure 5A). In human and rhesus macaque hindbrain, BDNF transcripts containing exon IIc and IV are the main transcripts, while in mouse, rat, rabbit, and opossum Bdnf exon VI also contributes notably to the total pool of BDNF (Figure 5B). Collectively, our results show evolutionary differences in the expression of different BDNF transcripts in mammalian brain.

Figure 5. BDNF mRNA expression levels in the brain of different mammals. Total BDNF mRNA levels in forebrain (A) and hindbrain (B) measured by levels of BDNF coding sequence (CDS, upper panels) and distribution of levels of BDNF 5′ exon-specific mRNAs (lower panels) are shown with the exons depicted below in the schematics of murine Bdnf gene structure. The exons indicated with gray color were not included in the analysis of expressed 5′ exons. BDNF mRNA levels in humans are shown in early embryo (0–9 postcoital week (PCW)), embryo (10–19 PCW), late embryo (20–39 PCW), infant (younger than 12 months), toddler (1–4 years old), school age (7–8 years old), adolescent (10–17 years old), young adult (18–29.99 years old), adult (30–39.99 years old), and middle-aged (40–58 years old); in rhesus macaques in embryo (E93-E109), late embryo (PE112-E130), postnatal (P0-P24), pre-adolescent (0.5–1 years old), adolescent (2–3 years old), young adult (8–11 years old), adult (14–15 years old), and middle-aged (20–26 years old) animals; in mouse and rats in embryo (E10.5-E14.5/E11-E15 mouse/rat), late embryo (E15.5-E18.5/E17-E20 mouse/rat), postnatal (P0-P14), adolescent (P22-56), and adult (P62+) animals; in rabbits in embryo (E12-E19.5), late embryo (E21-E27), postnatal (P10-P14), adolescent (P84), and adult (P186-P548) animals; in opossums in late embryo (E13.5), young postnatal (P0-P6), late postnatal (P10-P21), young adolescent (P28-P60), adolescent (P90-P120), and adult (P150-P180) animals [data from (2019)]. All used datasets and underlying data are shown in Supplementary Table 13. Data from individual animals are shown as small dots, circles indicate mean values and error bars represent standard error of the mean (SEM). CDS – coding sequence, CPM – counts per million.

Next, we studied BDNF mRNA levels in non-neural tissues during mammalian development. Total BDNF mRNA levels are mostly similar within one tissue among the different studied mammals (Supplementary Figure S7; Supplementary Table 14). Some exceptions are rhesus macaque heart, mouse and rabbit kidney, and opossum and rhesus macaque testis, where BDNF mRNA levels are higher than in the respective tissue in other studied mammals (Supplementary Figures S7A,C,D, respectively). Overall, the majority of BDNF transcripts contain exon VI, however, BDNF exon IV also contributes to BDNF mRNA levels in these mammals, although at a much lower level in murines when compared to the other studied mammals (Supplementary Figure S7).

Expression of BDNF transcripts with alternative 3’ UTRs

In addition to different promoter regions and the resulting 5′ exons in BDNF gene, BDNF exon IX contains alternative polyadenylation sites that can result in transcripts with either a short or long 3′ untranslated region (UTR) (Timmusk et al., 1993b; Fukuchi and Tsuda, 2010). BDNF transcripts with short 3’ UTR are more stable (Castrén et al., 1998) and are more associated with polysomes than transcripts with long 3’ UTR, suggesting better translatability (Timmusk et al., 1994b). As there are functional differences between the transcripts with different 3’ UTR lengths (Lekk et al., 2023), we set out to analyze the expression levels of BDNF transcripts with different 3’ UTRs in different tissues and to see whether the choice of the polyadenylation signal is evolutionarily conserved.

Our analysis shows that in the forebrain of human, rhesus macaque, mouse, rat, rabbit, and opossum the relative levels of BDNF transcripts with long 3’ UTR slightly decrease during development, dropping down to 30% of total BDNF transcripts in adulthood (Figure 6A; Supplementary Table 15). In contrast, in hindbrain the relative levels of transcripts with long 3’ UTR slightly increase during development, reaching up to 50% of all Bdnf transcripts in adult animals, and the highest increase is observed in middle aged rhesus macaques (up to 80%) (Figure 6B). In non-neural tissues, the relative levels of BDNF transcripts with long 3’ UTR is either stable with around ~50% of all transcripts or slightly decrease during development in the heart, liver, kidney, and ovary, and substantially decrease in the testis of the studied animals (except in rabbit) (Supplementary Figure S8; Supplementary Table 16). Collectively, our analysis indicates that the usage of BDNF alternative polyadenylation signals is relatively conserved in mammalian evolution.

Figure 6. The proportion of BDNF transcripts with long 3′ untranslated region in the brain of different mammals. The ratio of BDNF transcripts with long 3′ untranslated region (UTR) is shown in forebrain (A) and hindbrain (B) relative to total BDNF mRNA levels measured by levels of BDNF coding sequence (CDS). All used datasets and underlying data are shown in Supplementary Table 15. For more information on age classifications, see legend of Figure 5 (data from Cardoso-Moreira et al., 2019). Data from individual animals are shown as small dots, circles indicate mean values and error bars represent standard error of the mean (SEM). CDS – coding sequence.

Next, we conducted a comprehensive adult meta-analysis of different datasets to further describe BDNF transcripts with alternative 3’ UTRs in several different tissues in adult human, mouse, and rat. In agreement with our previous results, the highest proportion of long 3’ UTR transcripts is observed in human hindbrain (Supplementary Figure S9A; Supplementary Table 17), mouse cerebellum, bone and bone marrow, spleen, small intestine, and ovary (Supplementary Figure S9B), and rat cerebellum, uterus, adrenal gland, and lung (Supplementary Figure S9C). Altogether, our data suggests differential regulation of BDNF alternative polyadenylation in different tissues.

BDNF expression in different cell types based on single cell RNA sequencing

We next used publicly available single cell RNA-sequencing data accessible through the CellxGene database (Chan Zuckerberg Initiative, n.d.) to describe the cell types where BDNF is expressed. As expected, in the brain the most prevalent expression of BDNF is in glutamatergic neurons in both mouse (Table 1) and human (Table 2). Unfortunately, the studied CellxGene single cell datasets did not contain any data on sensory neurons, where the function of BDNF protein was discovered (Barde et al., 1978, 1980, 1982) and which are substantially lost in Bdnf knock-out animals (Ernfors et al., 1994; Jones et al., 1994). To further elucidate the expression of Bdnf, TrkB and p75NTR expression in the nervous system, we used single cell RNA sequencing data of the adolescent mouse nervous system by Zeisel et al. (2018) (Web tool accessible from http://mousebrain.org/adolescent/). According to this data, the Bdnf gene is mainly expressed in excitatory neurons in the central nervous system, and in sensory neurons in the peripheral nervous system (Figure 7). In contrast, TrkB shows widespread expression in different neurons throughout the nervous system, as well as in astrocytes (Figure 7). The expression of p75NTR is mainly constrained to certain subtypes of cholinergic neurons, and sympathetic and sensory neurons (Figure 7).

Table 1. Bdnf mRNA expression in mouse tissues according to single cell RNA sequencing.

Table 2. BDNF mRNA expression in human tissues according to single cell RNA sequencing.

Figure 7. The expression of Bdnf, TrkB (Ntrk2 gene), and p75NTR (Ngfr gene) mRNA in adolescent mouse nervous systems according to single cell RNA sequencing. Single cell RNA sequencing data of postnatal day 10–30 mice from Zeisel et al. (2018) was visualized using Mouse Brain Atlas from the Linnarson lab (http://mousebrain.org/). A deeper blue color indicates stronger expression of the indicated gene. The different cell types and their origins are shown above.

In mouse non-neural tissues, Bdnf is expressed in luminal epithelial cells in the mammary gland, in cardiac fibroblasts, in various cell types in the lung, in smooth muscle cells and in bladder cells (Table 1). In human non-neural tissues, BDNF is mainly expressed in epithelial and mesothelial cells, e.g., in the colon and lung (Table 2). In addition, we determined widespread BDNF gene expression in megakaryocytes, pericytes and smooth muscle cells of the vasculature (Table 2). In the human heart, BDNF is expressed in both myocytes and myoblasts (Table 2). In summary, our data show that only certain types of specialized cells produce BDNF.

BDNF protein levels in mouse and rat brain during development

BDNF mRNA and protein