Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Explore the Features of Brain-Derived Neurotrophic Factor in Mood Disorders

  • Fan-Chi Yeh,

    Affiliation Department of Public Health & Institute of Epidemiology and Preventive Medicine, College of Public Health, National Taiwan University, Taipei, Taiwan

  • Chung-Feng Kao,

    Affiliation Department of Public Health & Institute of Epidemiology and Preventive Medicine, College of Public Health, National Taiwan University, Taipei, Taiwan

  • Po-Hsiu Kuo

    phkuo@ntu.edu.tw

    Affiliations Department of Public Health & Institute of Epidemiology and Preventive Medicine, College of Public Health, National Taiwan University, Taipei, Taiwan, Research Center for Genes, Environment and Human Health, National Taiwan University, Taipei, Taiwan

Explore the Features of Brain-Derived Neurotrophic Factor in Mood Disorders

  • Fan-Chi Yeh, 
  • Chung-Feng Kao, 
  • Po-Hsiu Kuo
PLOS
x

Abstract

Objectives

Brain-derived neurotrophic factor (BDNF) plays important roles in neuronal survival and differentiation; however, the effects of BDNF on mood disorders remain unclear. We investigated BDNF from the perspective of various aspects of systems biology, including its molecular evolution, genomic studies, protein functions, and pathway analysis.

Methods

We conducted analyses examining sequences, multiple alignments, phylogenetic trees and positive selection across 12 species and several human populations. We summarized the results of previous genomic and functional studies of pro-BDNF and mature-BDNF (m-BDNF) found in a literature review. We identified proteins that interact with BDNF and performed pathway-based analysis using large genome-wide association (GWA) datasets obtained for mood disorders.

Results

BDNF is encoded by a highly conserved gene. The chordate BDNF genes exhibit an average of 75% identity with the human gene, while vertebrate orthologues are 85.9%-100% identical to human BDNF. No signs of recent positive selection were found. Associations between BDNF and mood disorders were not significant in most of the genomic studies (e.g., linkage, association, gene expression, GWA), while relationships between serum/plasma BDNF level and mood disorders were consistently reported. Pro-BDNF is important in the response to stress; the literature review suggests the necessity of studying both pro- and m-BDNF with regard to mood disorders. In addition to conventional pathway analysis, we further considered proteins that interact with BDNF (I-Genes) and identified several biological pathways involved with BDNF or I-Genes to be significantly associated with mood disorders.

Conclusions

Systematically examining the features and biological pathways of BDNF may provide opportunities to deepen our understanding of the mechanisms underlying mood disorders.

Introduction

Brain-derived neurotrophic factor (BDNF) is a member of the family of neurotrophic factors. It was first explored and purified from pig brain in 1982, and other members of the neurotrophin family were successively discovered [1]. In humans, the BDNF protein is first released to the Golgi apparatus in longer precursor forms (pro-BDNF), which are then cut by specific enzymes to obtain mature BDNF (m-BDNF) [2]. The m-BDNF and pro-BDNF forms have active but opposite functions in biological mechanisms. The function of m-BDNF is related to neuron survival, differentiation, and development through binding to tropomyosin-related kinase B (TrkB) receptor to induce a series of downstream pathways [3]. In contrast, Pro-BDNF binds to p75 neurotrophin receptor to induce cell apoptosis and long-term depression [1, 4, 5]. Because of their importance in maintaining the growth, survival, and function of neurons, exploring the potential roles of BDNF in brain disorders has been a long-standing focus, especially for mood disorders [6].

In past decades, various study approaches have been undertaken to explore the functions and associations of BDNF with mood disorders, such as assessing gene expression or the serum BDNF level, conducting linkage studies, or conducting individual and genome-wide association (GWA) studies. Inconsistent findings are often observed across or even within specific study designs. It is unlikely to achieve a comprehensive evaluation without combining information from different sources. Thus, exploration of the relationships between BDNF and mood disorders could benefit substantially from a systematic review of the currently available evidence from different study designs. Recently, more innovative approaches have been adopted to consider the complexity of molecular actions underlying complex traits, such as considering protein-protein interaction (PPI) or conducting pathway-based analyses. There have been some successful applications to psychiatric traits, such as schizophrenia and major depressive disorder [710]. For instance, candidate genes of schizophrenia are found to share common PPIs and act on the same biological processes [11]. Thus, it is essential to adopt strategies from systems biology and to incorporate PPI information and pathway analysis to enhance our understanding of the roles of BDNF in mood disorders.

Although a variety of genetic epidemiology designs have been used to study BDNF, other works focusing on its gene features, molecular evolution, and protein interactions with other molecules remain scarce and understudied. The adaptive selection and phylogenesis of the BDNF gene in 35 organisms have been previously examined in vertebrates [12]. However, the magnitude of natural selection in higher organisms and human beings requires further study. In this study, we adopted the systems biology perspective and applied several approaches to explore the features of BDNF in relation to mood disorders. We first evaluated its sequence conservation among species and the magnitude of positive selection in different human populations. Second, we reviewed curated evidence from a systematic review and meta-analysis of different study designs to evaluate the associations between BDNF and mood disorders, as well as evidence from basic experimental designs for the roles of pro-BDNF in emotion related traits. Finally, we investigated the protein interactome of BDNF and applied pathway-based analysis to examine the enrichment of biological pathways involved with BDNF in major depressive disorder (MDD) and bipolar disorder (BPD) separately, using two large GWA datasets.

Materials and Methods

1. Molecular evolution

Gene structure, sequence identity, and multiple alignment.

We compared 12 species in vertebrates, including 8 mammals (human, chimpanzee, macaque, dog, pig, cattle, mouse and rat) and 4 non-mammals (finch, turkey, chicken and zebrafish). We first searched the BDNF gene sequence, coding DNA sequence (CDS), and proteins in the NCBI Entrez Gene and Ensembl databases. For species with multiple transcripts in the database, the longest one was selected to include the most information. We used RepeatMasker (open version 3.3.0, http://www.repeatmasker.org/), a program to screen interspersed and low complexity DNA sequences, to identify different types of repeat sequence to analyze the underlying structure of the genes. The ABBlast searching engine was applied in this program for the advantages of time-saving and good sensitivity.

To compare the similarity between humans and the other species, we used the Basic Local Alignment Search Tool (BLAST) suites to calculate pairwise sequence identity rate for both the CDS nucleotides (blastn suite) and the amino acids of the protein sequence (blastp suite). BLAST is widely used in the comparison of query nucleotides or amino acid sequences and its sequence database, which calculates the similarity statistical significance by the Expected value (E value). The E value threshold was set to 10 by default, indicating a match number of 10 with a sequence of the same length in the database by chance. When the E value was smaller than 10, the pairwise identity and similarity significance were tested among species in BLAST. Multiple sequence alignment of the protein across species was performed by ClustalX (version 2.0) [13], illustrating sequence conservation. Finally, to evaluate the evolutionary relationships and distances between species, we used MEGA 5 to construct phylogenetic trees by neighbor-joining methods [14].

Estimation of positive selection.

The detection of adaptive selection across species was performed by PAL2NAL (version 14) [15]. The BDNF protein sequence and coding region DNA sequence were input into PAL2NAL. These sequences were converted into codons, and the ratio of the non-synonymous over the synonymous substitution rate (dN/dS) was pair-wise calculated among species using the codeml program in the Phylogenetic Analysis by Maximum Likelihood package. As most non-synonymous substitution is expected to occur under purifying selection, the gene is considered to experience positive selection when the dN/dS ratio is greater than 1. We also used Happlotter [16] to examine whether the BDNF gene has undergone recent positive selection during evolution in humans. This web application uses the HapMap phase II (release 21) allelic frequency data of three populations in its analysis, including Utah residents with northern and western European ancestry from the CEPH collection (CEU), Yoruban in Ibadan, Nigeria (YRI), and East Asians (ASN). The measurements of positive selection are calculated by the Integrated Haplotype Score (iHS), which is based on the difference in linkage disequilibrium between a selected and a background allele at the same position. Positive selection is considered to be significant when the iHS is higher than 2 [16].

2. Curated evidence from systematic literature review

Due to the great interest in studying BDNF and its protein in the context of mood disorders, a substantial amount of data has been accumulated. We performed a literature review by searching in the PubMed database for two topics: genomic studies of BDNF in humans and biological functions of pro-BDNF regardless of species. For the first topic, a number of meta-analysis and review articles have been published in the past decade. Articles were limited to English-language, full text, publication year between 2003 and 2013, human species, and review or meta-analysis as the article type. We also reviewed some of the references of these original articles when necessary. For the second topic, we focused on the biological functions of pro-BDNF in relation to emotion. Article types and species were not restricted. The search in the literature for mood disorder and emotion-related keywords used several MeSH terms, including mood disorder, major affective disorder, major depressive disorder, depression, bipolar disorder, mania, or stress. All the articles initially found in the systematic literature review process were manually checked for their relevance to our topics.

There are different designs in genomic studies of BDNF in human, and we divided these studies into categories of gene expression, linkage, genetic association and GWA studies, and serum/plasma BDNF level. A special note is for gene expression studies. Gene expression data and analysis results are affected by tissue types, experimental platforms, and characteristics of samples. Some gene expression results are deposited in the SMRI (Stanley Medical Research Institute), which collected brain tissues of patients with schizophrenia and bipolar disorder. Detailed experimental data of gene expression from12 studies were obtained from the SMRI database. In addition to literature review, we extracted gene expression data from the SMRI database to perform direct analysis, which provided the consensus fold change by weighted combination of individual probes mapping to genes across studies.

3. Protein interactome and pathway analysis

Protein-protein interaction (PPI).

To identify proteins that interact with BDNF, we used the STRING database (Search Tool for the Retrieval of Interacting Genes/Proteins, version 9.0). STRING includes more than 5 million known and predicted PPI from 1,133 organisms. It also provides an evidence score (ranging between 0 and 1) for the prediction of PPI from a variety of data sources, including experimental data on physical PPIs, annotated pathway sources, organism-specific or text-mining databases, and predicted associations via gene neighborhoods, fusion, co-occurrence, and co-expression. The higher the evidence score is, the greater is the confidence that the interaction exists. A molecule with an evidence score greater than 0.4 was considered to have medium confidence of interaction with BDNF, according to the suggestion of this web tool. Genes encoding proteins that interact with BDNF were named I-Genes in our study. Using the prediction results of STRING, we collected a total of 363 I-Genes based on predicted interactions with BDNF. We also plotted the PPI network of I-Genes with evidence scores greater than 0.9 in S1 Fig for easy visualization.

Pathway annotation and genome-wide association datasets.

We downloaded canonical pathway information from the Molecular Signatures Database (MSigDB, version 3.0) for pathway annotation. The MsigDB is a large integrated database from several available pathway sources, in which we focused on two main collections, the manually curated pathways based on expert knowledge and online pathway databases such as the Kyoto Encyclopedia of Genes and Genomes, BioCarta, Reactome, and Gene Ontology. In total, 4726 pathways were annotated. We further selected two sets of pathways to be included in our analysis for BPD and MDD: the BDNF gene-involving pathways (named BDNF-pw, see S2 Table.) and the BDNF PPI-related pathways that include I-Genes in the gene-set (named IGene-pw, see S2 Table.). The definition of IGene-pw is to have more than 10% I-Genes in the total gene number of a pathway. In total, there were 45 and 37 pathways for BDNF-pw and IGene-pw, respectively. Among them, two pathways overlapped. Pathways with extreme numbers of genes, i.e., more than 380 or less than 10 (i.e., the 10th percentile of pathway size distribution) were excluded from analysis to avoid stochastic bias or overly general biological processes. Finally, we included 29 pathways in BDNF-pw (consisting of 3,265 genes) and 34 in IGene-pw (consisting of 1,818 genes) in the following pathway analysis.

The GWA datasets of MDD and BPD for pathway analysis were accessed through the Genetic Association Information Network (GAIN). In brief, the MDD GWA data included 1,738 cases and 1,802 controls of Western European ancestry with 435,291 SNPs (single nucleotide polymorphisms); the BPD GWA data included 1,001 cases and 1,033 controls of European Americans with approximately 724,067 SNPs after data cleaning. More detailed information about the GAIN GWA datasets is available elsewhere [17, 18]. In total, there were 16,924 mapped genes in the GAIN-MDD dataset and 15,213 mapped genes in the GAIN-BPD dataset.

Pathway analysis.

We conducted pathway analyses using both competitive and self-contained methods. Competitive tests compare the statistics of certain genes in a given pathway with genes not in the pathway, while self-contained methods only compare specific genes in an interested pathway with non-associated background genome. Thus, we applied both the GSEA (Gene set enrichment analysis) and SumStat (sum statistic) methods for pathway analysis. Detailed information on these two statistics can be found elsewhere [19, 20]. In brief, the GSEA method calculates the enrichment score (ES) for the genes in pathways and compares the scores of other genes outside the pathway to evaluate the associations, while SumStat sums over all statistics in a gene set. The results were similar using the two methods, and thus, we report the results using the SumStat method in result Table.

To further consider the information gathered from PPI, we implemented a weighting scheme to give I-Genes twice the weight of the remaining genes in a given pathway. The significance of each examined pathway was calculated by 15,000 times permutation. The analysis flow chart for pathway analysis is shown in S2 Fig. To avoid the bias that larger genes may have smaller p-values by chance, we adopted the method of FOSCO (First Order Statistic Correction) to calculate gene-size corrected p-values for all genes [21, 22]. For multiple testing correction for the 61 examined pathways, we used the Benjamini-Hochberg method to adjust for p-values [23]. A padj-value less than 0.05 was considered significant. All the pathway analyses were performed using R (2.14.2).

Results

Identity rate and sequence alignments in BDNF gene

The DNA length of the BDNF gene varies widely across the examined vertebrates (Table 1), ranging from 770 base pairs in the finch to ~70,000 base pairs in primates. This variation may be partly due to DNA sequence repeats, as a higher repeat percentage was observed in longer DNA lengths. However, the DNA sequence of CDS and the amino acid sequence are directly related to protein coding, and non-coding sequences are removed during certain biological processes. Consequently, the variations in the DNA length of CDS and the amino acid length are relatively small. In molecular evolution, nucleotides are considered to be more informative, while amino acids are closer to the real biological function [24]. The higher proportion of repeats observed in nucleotide sequences was consistent with the complexity of the organisms we examined. On the other hand, non-coding region variations are suggested to play roles in advanced regulatory functions to affect the expression of coding genes [25]. Thus, variations in the non-coding region may also have impacts on BDNF availability.

Using the human sequence as the reference, the CDS sequence identity rate is high across species. For example, the identity rate is 85.7% between humans and chickens. The identity rate in the amino acid sequence between humans and other species is even higher, with an identity above 90% for chickens and a low in zebrafish of 70%. Except for zebrafish, the BDNF gene is highly conserved among the vertebrates examined. The high identity rate in amino acid sequences across species suggested that the functions of BDNF originated early in evolutionary history and are not specific to mammals with more complicated and sophisticated emotional expression. Fig 1b displays the evolutionary tree of the examined species based on BDNF protein sequences. Our results demonstrate a clear separation between mammals and non-mammals, as well as between primates and non-primates, which is consistent with the expectation for species relationships.

thumbnail
Fig 1. The multiple alignment (a) and Evolutionary tree of BDNF (b) in 12 species, and the positive selection in BDNF(c).

(a) The multiple alignment of BDNF sequence in 12 species by ClustalX. Arrowhead indicates the proteolytic site for specific enzymes to produce mature BDNF(*: Fully conserved) (b) Evolutionary tree of BDNF in 12 species produced by neighbor-joining methods in MEGA 5 (c) Plot showed the positive selection in BDNF by iHS score in Happlotter (Significant level: iHS ≧2; CEU: Utah residents with northern and western European ancestry from the CEPH collection; YRI: Yoruban in Ibadan, Nigeria; ASN: East Asians).

https://doi.org/10.1371/journal.pone.0128605.g001

The results of multiple alignments are shown in Fig 1a. The specific splicing site for producing mature BDNF was fully conserved across all examined species, and the sequence of mature BDNF was almost completely conserved in mammals. Similarly, several domains before the splicing site were also conserved in mammals, except the subtilisin site, which is consistent with findings in Tettamanti, Cattaneo (12]. The region after the specific splicing site is defined as a nerve growth factor domain, which has been found to be conserved in other neurotrophin family members [26]. In other words, multiple alignments revealed much variation in the pro-BDNF region, where the most studied polymorphism Val66Met (rs6265) is located [27]. This finding indicates the importance of studying pro-BDNF rather than m-BDNF alone. Considering the biological process from the gene level to different protein products might be necessary in studying BDNF and its phenotypes, including mood disorders.

Testing for recent positive selection

Human possess a higher level of emotional expressions and complex cognitional function; thus, we examined adaptive selection across species as well as several human populations. The dN/dS ratios in pairwise comparisons across species were all below one (S1 Table.), indicating that the BDNF gene did not experience strong natural selection in these species. Fig 1c showed the positive selection of BDNF variants in the CEU, YRI and ASN populations. In the BDNF gene region, the selection force was highest in the YRI population, with the highest differential linkage disequilibrium. However, the magnitude of positive selection in all three populations was non-significant (|iHS|<2). Moreover, we applied another program, dbPSHP [28], which utilized HapMap phase III data to examine selective signals in 11 populations. No significant positive selection was found for BDNF region in all examined populations.

The findings of non-significant natural selection in the evolutionary history across species and different human populations were consistent with results reported using the Selectome database [29]. The absence of significant selection for the BDNF gene across species and human populations suggested that the polymorphisms in this gene are unlikely to have dramatic deleterious impacts on species fitness and survival. Interestingly, a recent study reported high genetic diversity for the Val66Met polymorphism, of which the derived Met allele has frequency ranging from 0–72% in 58 global populations [30], suggesting the possibility for local selection. The same study then detected selection signal in a common 45% frequency haplotype carrying the Val66 allele in 12 European-derived CEPH pedigrees. Nevertheless, whether similar selection signals could be detected in other population data is unclear. The fact that we did not find positive selection among all examined populations in HapMap phase II and III data could be due to several reasons, such as marker density in the BDNF gene regions in the database and different analysis algorithms. However, from the evolutionary genetic point of view, mutations in the BDNF gene may still influence adaptions to environmental changes, especially for mental disorders. A polygenic mutation-selection balance theory (i.e., mutations in a large number of genes with accumulated influence on human behaviors) has been suggested to explain the development of mood disorders in human [31], further indicating the necessity of considering other genes and genes that interact with BDNF.

Systematic literature review of genomic studies on the relationship between BDNF and mood disorders

A decade ago, the neurotrophin hypothesis was proposed for the etiology of mood disorders, and BDNF is among the most studied neurotrophic factors [32]. A variety of studies have been conducted to investigate the relationships between BDNF and mood disorders. In this section, we included evidence from studies measuring BDNF serum levels, gene expression studies, and linkage and genetic association studies of BDNF. The results of this literature review of the genomic study designs of the BDNF gene and serum/plasma levels are listed in Table 2. For gene expression studies, no significantly different BDNF mRNA levels in peripheral blood were found between BPD patients and controls [33]. We also analyzed empirical gene expression data in postmortem brains of 12 studies in MDD and BPD from SMRI database. A consensus fold change could be obtained based on a weighted combination across different studies for each gene in the SMRI database. Results showed that the fold change of BDNF was small and p-values were non-significant for BPD (fold change = -1.03, p-value = 0.10) and MDD (fold change = 1.03, p value = 0.35). We found that BDNF mRNA level was neither associated with mood disorders nor exhibited substantial fold change between mood disorder patients and healthy controls.

thumbnail
Table 2. Systematic reviews of genetic studies of BDNF in review or meta- type articles (ordered by publication year).

https://doi.org/10.1371/journal.pone.0128605.t002

Linkage studies are often adopted to examine susceptible genetic regions and loci for the trait of interest. The BDNF gene is located on chromosome 11p. In seven review articles of linkage studies for MDD and BPD, only one study on BPD reported a suggestive linkage region at 11p (with LOD score of 2, near marker D11S1923) [34]. However, this region is distant from the BDNF gene (greater than 130 Mbp). Thus, the results from linkage studies provided negative evidence for mutations of the BDNF gene to have any direct and substantial effect on the familial inheritance of mood disorders, which may be due to the complex genetic etiology underlying MDD and BPD. The effect of the BDNF gene on mood disorder is too weak to be detected by family-based linkage study design.

For genetic association studies, two designs are commonly adopted: the candidate genes approach and the GWA approach. In the BDNF gene, the most tested genetic polymorphism is the Val66Met variant (rs6265). A few meta-analysis studies have been conducted for MDD and BPD, and the results are summarized in Table 2. Verhagen et al., (2010) selected 14 individual studies examining the Val66Met marker in 2,812 MDD cases and 10,843 healthy controls [35]. While the association between Val66Met and MDD in total samples was non-significant, gender stratification analysis showed an association in males only (OR = 1.67, P<0.01 in genotypic model). For BPD, an early meta-analysis with 11 studies did not find a significant effect of Val66Met [36]. Another study using 3,897 BPD cases and 6,807 controls found nominal significance for the marker Val66Met, but the association became non-significant after correcting for multiple testing [37].

Large scale GWA studies are expected to have a better chance of finding underlying susceptible genetic variants for mood disorders, especially meta- or mega-analysis of GWA studies that include even larger sample sizes with millions of markers distributed over the genome. Several GWA meta-analyses with thousands of individuals have been conducted for MDD and BPD, however, none of these studies reported suggestive signals for the association of the Val66Met marker or other genetic variants of BDNF with mood disorders [3842]. These somewhat inconsistent and majorly negative findings from genetic association studies are unlikely to result from the low power of individual studies, as a trend of no association was found in most of the meta-analysis studies. Petryshen et al. (2010) has observed considerable global diversity of BDNF SNPs. This genetic diversity somewhat decreases the sample size and reduces the analysis power in meta-analysis but is also unlikely to have caused the inconsistent results.

Some studies have examined the relationship between BDNF protein level and neuropsychiatric disorders in the hope of identifying useful peripheral biomarkers for predicting the risk of mood disorders. While most of the genomic studies (such as gene expression, linkage, or association studies) showed negative or inconsistent results, protein level studies seemed to demonstrate more consistent and positive findings. We systematically reviewed articles and meta-analysis studies for MDD and BPD considering both plasma and serum BDNF expression levels. For MDD, a significant association (P<6x10-8) between serum BDNF and depression status was also reported in Sen et al., (2008) [43]. In a meta-analysis with a larger sample size, findings in Bocchio-Chiavetto et al., (2010) supported a low serum BDNF level in MDD patients, and a similar pattern was found in the plasma BDNF level with 161 MDD patients and 211 controls [44]. For BPD, Lin et al., (2009) included 15 studies in a meta-analysis that showed significantly lower BDNF levels in blood in BPD patients than in controls (P = 1x10-4) [45].

In addition to the comparison between the blood level of BDNF in mood disorder patients and healthy controls, several studies further investigate the effect of different disorder statuses. After medication treatment, significantly increased BDNF levels were found in MDD patients [43, 46]. For BPD, two studies compared BDNF level in different mood episodic statuses with controls. Their results found that BDNF levels were significantly lower in patients in manic and depressed acute states but not in euthymic state [45, 47]. Moreover, the BDNF level increased significantly after treatment in patients with acute manic episodes [45], further supporting the direct association of BDNF protein level with mood episodic status. In contrast to genomic studies that investigate the relationship between gene and mood disorders, the protein level of BDNF seemed to be closer to behavioral expression and may have a more direct effect on mood disturbances. Based on reviewing the evidence in the literature, we considered that the more consistent findings for the involvement of BDNF in mood disorders come from protein level studies, which suggest roles for BDNF in the mechanisms of mood disorders.

Data integration

Although BDNF has been widely tested for mood disorders, the evidence was not consistent across different study designs and individual studies. A data integration strategy is necessary to evaluate the role of BDNF in mood disorders. We previously developed a data integration pipeline to combine multiple sources of genomic information (e.g., association, linkage, gene expression, regulatory pathway) to prioritize candidate loci for MDD [7]. With a modified algorithm, we found that BDNF was listed as the top depression candidate gene among the 151 identified DEPgenes (candidate gene list for MDD in Kao et al., (2011) [7]. Similarly, BDNF was identified in the top candidate gene list for BPD [22, 48]. Therefore, it is preferable to consider an integrative framework to accommodate a large amount of genomic data from different sources rather than focusing on one specific study design to weight the influences of BDNF in mood disorders.

Pathway analyses for BDNF and its interacting molecules

It is possible that BDNF interacts with other molecules in specific biological pathways to exert its effect on the trait of interest [49]. We further performed pathway analyses to examine BDNF-related pathways with MDD and BPD separately. Using GAIN GWA datasets for MDD and BPD with pathways collected in MSigDB, we reported significant pathways for mood disorders involved with BDNF (BDNF-pw) and its interacting molecules (I-Genes) in Table 3. In total, we identified 29 BDNF-pw and 34 IGene-pw. Two pathways overlapped between the IGene-pw and BDNF-pw sets: the MAPK signaling pathway and the neurotrophin signaling pathway (S2 Table.). The median gene number was 103 (ranging from 13 to 373) in the 29 BDNF-pw and 87.5 (ranging from 16 to 377) in the 34 IGene-pw.

thumbnail
Table 3. The significant pathways after gene size adjustment in 34 IGene-pw and 29 BDNF-pw of pathway analysis by using GAIN-MDD or GAIN-BPD GWAS data and SumStat methods.

https://doi.org/10.1371/journal.pone.0128605.t003

Table 3 shows that there were 19 nominally significant IGene-pw and 9 significant BDNF-pw that were associated with mood disorders. However, after correction of the multiple testing issue, none of the pathways were significant for MDD, while 11 IGene-pw and 2 BDNF-pw were still significant in BPD. Considering the high heritability in BPD, more loci might be involved in corresponding with much complex pathogenesis for BPD. This result implied the importance of considering interacting proteins to explore the roles of BDNF and the underlying mechanism of BPD. For both MDD and BPD, the results were very similar regardless of adding a weighting scheme to I-Genes or not. Adding extra weight to the genes that interact with BDNF in specified pathways seemed to be irrelevant and did not exert substantial influence on the pathway analysis results.

Comparing significant pathways of IGene-pw and BDNF-pw, we found that the results were quite different. The most significant IGene-pw were the calcium signaling pathway, neurological system process, and synaptic transmission for BPD (P<0.001), which were all related to nervous system functions. In contrast, the significant BDNF-pw were related to receptor binding, gene regulations of certain cells or organisms that were involved in cancer or physical health problems. Several previous studies have shown that cancer mortality rates were higher in mood disorder patients than in the general population [50]. The BDNF-pw results suggest a research avenue to study the link between mood disorders and cancers and to explore the common roles of BDNF in the two disease categories. Interestingly, the MAPK signaling pathway that included both the BDNF gene and more than 10 percent IGenes showed association with BPD under both the BDNF-pw and IGene-pw categories. The MAPK pathway is activated by BDNF through binding to the TrkB receptor, which further emphasizes the involvement of BDNF in mood disorders.

Some of the significant pathways that were identified in MDD and/or BPD in our studies were also reported by previous studies for mood disorders, such as calcium signaling pathway and glutamate receptor activity. Hasbi et al., (2009) reported that calcium mobilization to the plasma membrane was induced by the activation of the dopamine D1-D2 receptor heteromer, and the mobilization triggered the activation of CaMKIIα in the nucleus that enhanced BDNF expression [51]. BDNF also plays a role in the modification of glutamate synapses through regulating glutamate secretion presynaptically and changing the phosphorylation of glutamate receptor through postsynaptic actions [52]. A more detailed discussion of the roles of glutamate in emotion, cognition and depression can be found in a recent review [53]. To summarize, we identified a few significant biological pathways for mood disorders that involve BDNF and BDNF-interacting molecules. Studying these enriched pathways in future work might facilitate our understanding of the underlying etiology of mood disorders and how BDNF and its interacting molecules function to influence the risk of mood disorders.

The roles of pro-BDNF and m-BDNF in mood disorders

Neurotrophins are first produced and released in a precursor form and then proteolytically cleaved by several enzymes to create a mature form [27]. Like other neurotrophins, pro-BDNF is cleaved to m-BDNF within the endoplasmic reticulum by furin or in regulated secretory vesicles by proconvertase enzymes. When pro-BDNF is released to the extracellular environment, it is then cleaved by plasmin [54]. The activities of these cleavage enzymes critically decide the downstream actions, triggering cell survival by m-BDNF or apoptosis by pro-BDNF. Due to the opposite biological functions of pro-BDNF and m-BDNF in the neuronal system, it is important to discuss the roles of these two proteins in relation to mood disorders. In addition, the factors and contributors that lead to the cleavage of pro-BDNF to m-BDNF are also key points in understanding the mechanism of mood disorders. For example, it has been postulated that BDNF affects mood disorders through protein p11 by activating tissue plasminogen activator (tPA)/plasminogen activity, cutting pro-BDNF to increase the production of m-BDNF and leading to an antidepressant effect [55]. Therefore, enzymes that influence (tPA)/plasminogen activity are other potential targets to investigate for mood disorders.

On the other hand, the level of BDNF is found to be influenced by adverse environmental stimuli and stress, which is often linked with emotional, physiological and psychological reactions. Acute and chronic stresses are reported to increase the risk of developing mood disorders [32, 56], and several studies had showed the negative environmental stimuli effect on the relationship between BDNF and depression [57, 58]. A negative relationship between stress and BDNF blood level is found in depression patients [59]. Moreover, animal models are often applied to investigate the effects of stress on the level of BDNF. Using a zebrafish model (its brain organization is similar to higher vertebrates) to directly examine the effect of stress on neurotrophins, Tognoli et al., (2012) reported alterations of BDNF levels after acute stress, where the pro-BDNF level was increased and the m-BDNF level was decreased [60]. The authors further suggested that the pro-BDNF/total-BDNF ratio could accurately predict stress response. Several other studies have conducted experiments in rats. In one recent study, prenatal stress was employed in different rat strains, and changes in the protein levels of pro-BDNF and m-BDNF, as well as cleavage-related protease and BDNF downstream signaling proteins, were found after prenatal stress [61]. Notably, these changes depend on strains, revealing the critical role of genetic background. Another experiment used a chronic prenatal restraint stress model in rats to explore the effect of prenatal stress on hippocampal synaptic plasticity in the offspring [62]. Decreased protease tPA expression was found, which decreased the proteolytic cleavage from pro-BDNF to m-BDNF, leading to increased levels of pro-BDNF and decreased levels of m-BDNF in rats, thereby mediating the effect of prenatal stress on long-term potentiation and long-term depression in the hippocampus. These aforementioned study findings indicated the influence of stress on individuals by altering the levels of pro-BDNF and m-BDNF directly or indirectly through the cleavage enzymes. In addition, this evidence also supports the involvement of both pro-BDNF and m-BDNF in the etiological mechanisms of mood disorders. Thus, how individuals respond to acute or chronic stress may involve the regulation of BDNF-related pathways. Dysfunction of the BDNF-related pathways might be one of the key points leading to mood disorders in response to environmental stimuli.

Other observations regarding anti-depression treatment provide additional supporting evidence for the roles of BDNF in mood disorders. Segawa et al., (2013) examined the effects of electroconvulsive seizures (ECS) on the expression of BDNF and other protease enzymes in the hippocampus of rats [63]. The authors found that ECS increased the levels of pro-BDNF and extracellular protease (t-PA) simultaneously, as well as enhancing the pro-BDNF processing and thereby increasing the level of m-BDNF to exert its antidepressant effect. This process may partly explain the clinical effectiveness of ECS in treating depression compared to antidepressants. Despite the opposite functions of pro-BDNF and m-BDNF, they are rarely distinguished in previous studies examining the serum/plasma BDNF level. It was not until recently that one study tested pro-BDNF and m-BDNF levels in human saliva, and the authors also reported that the m-BDNF level was altered by the Val66Met polymorphism [64]. In 2012, Yoshida et al. used a newly available human pro-BDNF and m-BDNF ELISA kit to examine the BDNF serum level in MDD patients [65]. They found that m-BDNF, but not pro-BDNF, is lower in MDD patients than in healthy controls. However, the sensitivity of measuring pro-BDNF is much lower than the sensitivity of measuring m-BDNF in Yoshida et al.’s study. To overcome this measurement technique issue would assist in studying the mechanisms of BDNF-related molecules and pathways in mood disorders.

The current study had certain limitations. Because most of our analyses relied on data from publicly available and published literature reports, the accuracy of these results is influenced by the completeness of the databases and potential publication bias in the literature. This situation exists in most similar research settings, and we elected to use multiple databases to reduce the potential problems. We selected a so-called integrated database to reduce information bias and restricted our literature review to meta-analysis and review studies. In addition, we collected data from different data sources as completely as possible and developed a weighting system in the hope of obtaining more robust findings.

Conclusions and Future Directions

From a biological perspective, BDNF proteins participate in neuron development and stress responses, which are considered to play important roles in relation to neuropsychiatric disorders. Many previous studies examined the relationship between BDNF and mood disorders using a variety of study designs and analytical methods. We have systematically reviewed evidence from genomic studies and BDNF protein level studies, including pro-BDNF and m-BDNF. In addition, we performed sequence alignment among different species and tested positive selection to explore the evolutionary history of BDNF. Genes that interact with BDNF are evaluated at the pathway level and analyzed using two GWAS datasets of BPD and MDD.

To summarize, we observed a general pattern that most linkage and GWA association studies provided negative findings for BDNF with mood disorders, while studies examining serum/plasma levels of BDNF provided more consistent positive findings for its involvement in mood disorders. The lack of evidence from genetic studies may not negate the roles of BDNF in mood disorders. Most of the risk variants in GWA studies often explain very small proportion of heritability in complex disorders, and the negative findings may result from small main effect of BDNF [66]. In addition, multiple-gene interactions are not usually considered in most genetic studies. The significant findings in our pathway analysis (especially when we considered the I-genes) suggest that the association of BDNF with mood disorders may not be through the main effect. BDNF likely acts as a modifier gene or jointly with other genes in the same pathway to contribute to the risk of mood disorders. One example is a study that reported a significant interaction effect between BDNF Val66Met and DRD3 Ser9Gly genotypes, influencing bipolar disorder [67]. Because of the limitations in each type of study design, strategies that are designed to incorporate multiple lines of evidence are appreciated for the study of the underlying biological functions of targeted genes and molecules for the disorder of interest.

Stress is considered an important environmental stimulus that influences the risk of mood disorders by altering the expression of BDNF. Previous studies have suggested different roles for pro-BDNF and m-BDNF and how these proteins operate in response to stress. Examining the changes in pro-BDNF and m-BDNF blood levels and the pro-BDNF/m-BDNF ratio may be useful and important for the early detection of mood disorders. Moreover, the expression of the proteases that cut pro-BDNF to produce m-BDNF and of the proteins that interact with BDNF may also be important indicators for the risk of developing mood disorders, and further study is warranted to investigate their roles in the underlying mechanisms of mood disorders.

Supporting Information

S1 Fig. The network of proteins interact with BDNF which is based on the evidence score more than 0.9 produced by STRING.

https://doi.org/10.1371/journal.pone.0128605.s001

(DOCX)

S2 Fig. The flow chart of pathway analysis in 34 IGene-pw.

A. Genes interacts with BDNF by STRING. (I-Genes, #:363) B. Map I-Genes to the pathways in MsigDB, select the pathways which overlap rate is more than 10%, and exclude pathways with extreme gene numbers. (#:34) C. All the genes in 34 pathways were included for analysis. (#:1,818) D. Give weights for genes with evidence score provided by STRING. E. Use GSEA and SumStat as statistics.

https://doi.org/10.1371/journal.pone.0128605.s002

(DOCX)

S1 Table. Pairwise dN/dS ratio in the BDNF gene across species.

Significant level: dN/dS≧1.

https://doi.org/10.1371/journal.pone.0128605.s003

(DOCX)

S2 Table. The list of pathways that were recruited for pathway analysis in this study.

a The total number of genes in each pathway annotated by MsigDB. b Pathway was selected since it was related to BDNF. c Pathway was selected since it include more than 10 percent I-Genes. d Pathway was significant in at least one statistical way.—: Did not include in pathway analysis since the extreme pathway gene numbers.

https://doi.org/10.1371/journal.pone.0128605.s004

(DOCX)

Acknowledgments

The GWA data sets were accessed through the Genetic Association Information Network (GAIN). The datasets used for the analyses described in this manuscript were obtained from the database of Genotypes and Phenotypes (dbGaP) found at http://www.ncbi.nlm.nih.gov/gap through dbGaP accession number phs000020.v1.p1 for major depressive disorder(http://www.ncbi.nlm.nih.gov/projects/gap/cgi-bin/study.cgi?study_id=phs000020.v1.p1) and accession numbers phs000017.v1.p1 for bipolar disorder(http://www.ncbi.nlm.nih.gov/projects/gap/cgi-bin/study.cgi?study_id=phs000017.v1.p1).

Author Contributions

Conceived and designed the experiments: PHK. Performed the experiments: FCY. Analyzed the data: FCY CFK. Contributed reagents/materials/analysis tools: FCY CFK. Wrote the paper: FCY PHK.

References

  1. 1. Binder DK, Scharfman HE. Brain-derived neurotrophic factor. Growth Factors. 2004;22(3):123–31. Epub 2004/11/03. pmid:15518235; PubMed Central PMCID: PMC2504526.
  2. 2. Hashimoto K. Brain-derived neurotrophic factor as a biomarker for mood disorders: an historical overview and future directions. Psychiatry and clinical neurosciences. 2010;64(4):341–57. Epub 2010/07/27. pmid:20653908.
  3. 3. Numakawa T, Suzuki S, Kumamaru E, Adachi N, Richards M, Kunugi H. BDNF function and intracellular signaling in neurons. Histology and histopathology. 2010;25(2):237–58. Epub 2009/12/18. pmid:20017110.
  4. 4. Koshimizu H, Hazama S, Hara T, Ogura A, Kojima M. Distinct signaling pathways of precursor BDNF and mature BDNF in cultured cerebellar granule neurons. Neuroscience letters. 2010;473(3):229–32. Epub 2010/03/12. pmid:20219632.
  5. 5. Teng HK, Teng KK, Lee R, Wright S, Tevar S, Almeida RD, et al. ProBDNF induces neuronal apoptosis via activation of a receptor complex of p75NTR and sortilin. The Journal of neuroscience: the official journal of the Society for Neuroscience. 2005;25(22):5455–63. Epub 2005/06/03. pmid:15930396.
  6. 6. Martinowich K, Manji H, Lu B. New insights into BDNF function in depression and anxiety. Nature neuroscience. 2007;10(9):1089–93. Epub 2007/08/30. pmid:17726474.
  7. 7. Kao CF, Fang YS, Zhao Z, Kuo PH. Prioritization and evaluation of depression candidate genes by combining multidimensional data resources. PloS one. 2011;6(4):e18696. Epub 2011/04/16. pmid:21494644; PubMed Central PMCID: PMC3071871.
  8. 8. Menashe I, Maeder D, Garcia-Closas M, Figueroa JD, Bhattacharjee S, Rotunno M, et al. Pathway analysis of breast cancer genome-wide association study highlights three pathways and one canonical signaling cascade. Cancer research. 2010;70(11):4453–9. Epub 2010/05/13. pmid:20460509; PubMed Central PMCID: PMC2907250.
  9. 9. Savas S, Geraci J, Jurisica I, Liu G. A comprehensive catalogue of functional genetic variations in the EGFR pathway: protein-protein interaction analysis reveals novel genes and polymorphisms important for cancer research. International journal of cancer Journal international du cancer. 2009;125(6):1257–65. Epub 2009/06/06. pmid:19499547.
  10. 10. Sun J, Wan C, Jia P, Fanous AH, Kendler KS, Riley BP, et al. Application of systems biology approach identifies and validates GRB2 as a risk gene for schizophrenia in the Irish Case Control Study of Schizophrenia (ICCSS) sample. Schizophrenia research. 2011;125(2–3):201–8. Epub 2011/01/05. pmid:21195589; PubMed Central PMCID: PMC3031722.
  11. 11. Camargo LM, Collura V, Rain JC, Mizuguchi K, Hermjakob H, Kerrien S, et al. Disrupted in Schizophrenia 1 Interactome: evidence for the close connectivity of risk genes and a potential synaptic basis for schizophrenia. Molecular psychiatry. 2007;12(1):74–86. Epub 2006/10/18. pmid:17043677.
  12. 12. Tettamanti G, Cattaneo AG, Gornati R, de Eguileor M, Bernardini G, Binelli G. Phylogenesis of brain-derived neurotrophic factor (BDNF) in vertebrates. Gene. 2010;450(1–2):85–93. Epub 2009/11/03. pmid:19879341.
  13. 13. Larkin MA, Blackshields G, Brown NP, Chenna R, McGettigan PA, McWilliam H, et al. Clustal W and Clustal X version 2.0. Bioinformatics. 2007;23(21):2947–8. Epub 2007/09/12. pmid:17846036.
  14. 14. Tamura K, Peterson D, Peterson N, Stecher G, Nei M, Kumar S. MEGA5: molecular evolutionary genetics analysis using maximum likelihood, evolutionary distance, and maximum parsimony methods. Molecular biology and evolution. 2011;28(10):2731–9. Epub 2011/05/07. pmid:21546353; PubMed Central PMCID: PMC3203626.
  15. 15. Suyama M, Torrents D, Bork P. PAL2NAL: robust conversion of protein sequence alignments into the corresponding codon alignments. Nucleic acids research. 2006;34(Web Server issue):W609–12. Epub 2006/07/18. pmid:16845082; PubMed Central PMCID: PMC1538804.
  16. 16. Voight BF, Kudaravalli S, Wen X, Pritchard JK. A Map of Recent Positive Selection in the Human Genome. PLoS Biology. 2006;4(3):e72. pmid:16494531
  17. 17. Sullivan PF, de Geus EJ, Willemsen G, James MR, Smit JH, Zandbelt T, et al. Genome-wide association for major depressive disorder: a possible role for the presynaptic protein piccolo. Molecular psychiatry. 2009;14(4):359–75. Epub 2008/12/10. pmid:19065144; PubMed Central PMCID: PMC2717726.
  18. 18. Smith EN, Bloss CS, Badner JA, Barrett T, Belmonte PL, Berrettini W, et al. Genome-wide association study of bipolar disorder in European American and African American individuals. Molecular psychiatry. 2009;14(8):755–63. Epub 2009/06/03. pmid:19488044; PubMed Central PMCID: PMC3035981.
  19. 19. Efron B, Tibshirani R. On testing the significance of sets of genes. Annals of Applied Statistics. 2007;1(1):107–29.
  20. 20. Wang K, Li M, Bucan M. Pathway-based approaches for analysis of genomewide association studies. American journal of human genetics. 2007;81(6):1278–83. Epub 2007/10/30. pmid:17966091; PubMed Central PMCID: PMC2276352.
  21. 21. Mirina A, Atzmon G, Ye K, Bergman A. Gene Size Matters. PloS one. 2012;7(11):e49093. pmid:23152854
  22. 22. Kao C-F, Chuang L-C, Kuo P-H. Risk and information evaluation of prioritized genes for complex traits: Application to bipolar disorder. American Journal of Medical Genetics Part B: Neuropsychiatric Genetics. 2014;165(7):596–606. pmid:25123107
  23. 23. Benjamini Y, Hochberg Y. Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing. Journal of the Royal Statistical Society Series B (Methodological). 1995;57(1):289–300. citeulike-article-id:1042553.
  24. 24. Saccone C, Pesole G. Molecular Evolution. Handbook of Comparative Genomics: John Wiley & Sons, Inc.; 2005.
  25. 25. Shabalina SA, Spiridonov NA. The mammalian transcriptome and the function of non-coding DNA sequences. Genome biology. 2004;5(4):/2004/5/4/105-//5/4/105. pmid:15059262
  26. 26. Lanave C, Colangelo AM, Saccone C, Alberghina L. Molecular evolution of the neurotrophin family members and their Trk receptors. Gene. 2007;394(1–2):1–12. Epub 2007/03/24. pmid:17379456.
  27. 27. Lu B. Pro-Region of Neurotrophins: Role in Synaptic Modulation. Neuron. 2003;39(5):735–8. pmid:12948441
  28. 28. Li MJ, Wang LY, Xia Z, Wong MP, Sham PC, Wang J. dbPSHP: a database of recent positive selection across human populations. Nucleic acids research. 2014;42(Database issue):D910–6. Epub 2013/11/07. pmid:24194603; PubMed Central PMCID: PMC3965004.
  29. 29. Proux E, Studer RA, Moretti S, Robinson-Rechavi M. Selectome: a database of positive selection. Nucleic acids research. 2009;37(suppl_1):D404–7.
  30. 30. Petryshen TL, Sabeti PC, Aldinger KA, Fry B, Fan JB, Schaffner SF, et al. Population genetic study of the brain-derived neurotrophic factor (BDNF) gene. Molecular psychiatry. 2010;15(8):810–5. Epub 2009/03/04. pmid:19255578; PubMed Central PMCID: PMC2888876.
  31. 31. Keller MC, Miller G. Resolving the paradox of common, harmful, heritable mental disorders: which evolutionary genetic models work best? The Behavioral and brain sciences. 2006;29(4):385–404; discussion 5–52. Epub 2006/11/11. pmid:17094843.
  32. 32. Duman RS, Monteggia LM. A neurotrophic model for stress-related mood disorders. Biological psychiatry. 2006;59(12):1116–27. Epub 2006/04/25. pmid:16631126.
  33. 33. Munkholm K, Vinberg M, Berk M, Kessing LV. State-related alterations of gene expression in bipolar disorder: a systematic review. Bipolar disorders. 2012;14(7):684–96. Epub 2012/10/10. pmid:23043691.
  34. 34. Hayden EP, Nurnberger JI Jr. Molecular genetics of bipolar disorder. Genes, brain, and behavior. 2006;5(1):85–95. Epub 2006/01/27. pmid:16436192.
  35. 35. Verhagen M, van der Meij A, van Deurzen PA, Janzing JG, Arias-Vasquez A, Buitelaar JK, et al. Meta-analysis of the BDNF Val66Met polymorphism in major depressive disorder: effects of gender and ethnicity. Molecular psychiatry. 2010;15(3):260–71. Epub 2008/10/15. pmid:18852698.
  36. 36. Kanazawa T, Glatt SJ, Kia-Keating B, Yoneda H, Tsuang MT. Meta-analysis reveals no association of the Val66Met polymorphism of brain-derived neurotrophic factor with either schizophrenia or bipolar disorder. Psychiatric genetics. 2007;17(3):165–70. Epub 2007/04/10. pmid:17417060.
  37. 37. Seifuddin F, Mahon PB, Judy J, Pirooznia M, Jancic D, Taylor J, et al. Meta-analysis of genetic association studies on bipolar disorder. American journal of medical genetics Part B, Neuropsychiatric genetics: the official publication of the International Society of Psychiatric Genetics. 2012. Epub 2012/05/11. pmid:22573399.
  38. 38. Scott LJ, Muglia P, Kong XQ, Guan W, Flickinger M, Upmanyu R, et al. Genome-wide association and meta-analysis of bipolar disorder in individuals of European ancestry. Proceedings of the National Academy of Sciences of the United States of America. 2009;106(18):7501–6. Epub 2009/05/07. pmid:19416921; PubMed Central PMCID: PMC2678639.
  39. 39. McMahon FJ, Akula N, Schulze TG, Muglia P, Tozzi F, Detera-Wadleigh SD, et al. Meta-analysis of genome-wide association data identifies a risk locus for major mood disorders on 3p21.1. Nature genetics. 2010;42(2):128–31. Epub 2010/01/19. pmid:20081856; PubMed Central PMCID: PMC2854040.
  40. 40. Liu Y, Blackwood DH, Caesar S, de Geus EJ, Farmer A, Ferreira MA, et al. Meta-analysis of genome-wide association data of bipolar disorder and major depressive disorder. Molecular psychiatry. 2011;16(1):2–4. Epub 2010/03/31. pmid:20351715.
  41. 41. Wray NR, Pergadia ML, Blackwood DH, Penninx BW, Gordon SD, Nyholt DR, et al. Genome-wide association study of major depressive disorder: new results, meta-analysis, and lessons learned. Molecular psychiatry. 2012;17(1):36–48. Epub 2010/11/03. pmid:21042317; PubMed Central PMCID: PMC3252611.
  42. 42. Major Depressive Disorder Working Group of the Psychiatric GWAS Consortium. A mega-analysis of genome-wide association studies for major depressive disorder. Molecular psychiatry. 2012. Epub 2012/04/05. pmid:22472876.
  43. 43. Sen S, Duman R, Sanacora G. Serum brain-derived neurotrophic factor, depression, and antidepressant medications: meta-analyses and implications. Biological psychiatry. 2008;64(6):527–32. Epub 2008/06/24. pmid:18571629; PubMed Central PMCID: PMC2597158.
  44. 44. Bocchio-Chiavetto L, Bagnardi V, Zanardini R, Molteni R, Nielsen MG, Placentino A, et al. Serum and plasma BDNF levels in major depression: a replication study and meta-analyses. The world journal of biological psychiatry: the official journal of the World Federation of Societies of Biological Psychiatry. 2010;11(6):763–73. Epub 2010/03/26. pmid:20334574.
  45. 45. Lin PY. State-dependent decrease in levels of brain-derived neurotrophic factor in bipolar disorder: a meta-analytic study. Neuroscience letters. 2009;466(3):139–43. Epub 2009/09/30. pmid:19786073.
  46. 46. Brunoni AR, Lopes M, Fregni F. A systematic review and meta-analysis of clinical studies on major depression and BDNF levels: implications for the role of neuroplasticity in depression. The international journal of neuropsychopharmacology / official scientific journal of the Collegium Internationale Neuropsychopharmacologicum (CINP). 2008;11(8):1169–80. Epub 2008/08/30. pmid:18752720.
  47. 47. Fernandes BS, Gama CS, Cereser KM, Yatham LN, Fries GR, Colpo G, et al. Brain-derived neurotrophic factor as a state-marker of mood episodes in bipolar disorders: a systematic review and meta-regression analysis. Journal of psychiatric research. 2011;45(8):995–1004. Epub 2011/05/10. pmid:21550050.
  48. 48. Patel SD, Le-Niculescu H, Koller DL, Green SD, Lahiri DK, McMahon FJ, et al. Coming to grips with complex disorders: genetic risk prediction in bipolar disorder using panels of genes identified through convergent functional genomics. American journal of medical genetics Part B, Neuropsychiatric genetics: the official publication of the International Society of Psychiatric Genetics. 2010;153B(4):850–77. Epub 2010/05/15. pmid:20468069.
  49. 49. Jia P, Kao CF, Kuo PH, Zhao Z. A comprehensive network and pathway analysis of candidate genes in major depressive disorder. BMC Syst Biol. 2011;5 Suppl 3:S12. Epub 2012/07/20. pmid:22784618; PubMed Central PMCID: PMC3287567.
  50. 50. Roshanaei-Moghaddam B, Katon W. Premature mortality from general medical illnesses among persons with bipolar disorder: a review. Psychiatric services (Washington, DC). 2009;60(2):147–56. Epub 2009/01/30. pmid:19176408.
  51. 51. Hasbi A, Fan T, Alijaniaram M, Nguyen T, Perreault ML, O'Dowd BF, et al. Calcium signaling cascade links dopamine D1-D2 receptor heteromer to striatal BDNF production and neuronal growth. Proceedings of the National Academy of Sciences of the United States of America. 2009;106(50):21377–82. Epub 2009/12/02. pmid:19948956; PubMed Central PMCID: PMC2795506.
  52. 52. Carvalho AL, Caldeira MV, Santos SD, Duarte CB. Role of the brain-derived neurotrophic factor at glutamatergic synapses. British journal of pharmacology. 2008;153 Suppl 1:S310–24. Epub 2007/12/07. pmid:18059328; PubMed Central PMCID: PMC2268077.
  53. 53. Sanacora G, Treccani G, Popoli M. Towards a glutamate hypothesis of depression: an emerging frontier of neuropsychopharmacology for mood disorders. Neuropharmacology. 2012;62(1):63–77. Epub 2011/08/11. pmid:21827775; PubMed Central PMCID: PMC3205453.
  54. 54. Barker PA. Whither proBDNF? Nature neuroscience. 2009;12(2):105–6. Epub 2009/01/28. pmid:19172162.
  55. 55. Tsai S-J. The P11, tPA/plasminogen system and brain-derived neurotrophic factor: Implications for the pathogenesis of major depression and the therapeutic mechanism of antidepressants. Medical Hypotheses. 2007;68(1):180–3. pmid:16890384
  56. 56. Dunham JS, Deakin JF, Miyajima F, Payton A, Toro CT. Expression of hippocampal brain-derived neurotrophic factor and its receptors in Stanley consortium brains. Journal of psychiatric research. 2009;43(14):1175–84. Epub 2009/04/21. pmid:19376528.
  57. 57. Gatt JM, Nemeroff CB, Dobson-Stone C, Paul RH, Bryant RA, Schofield PR, et al. Interactions between BDNF Val66Met polymorphism and early life stress predict brain and arousal pathways to syndromal depression and anxiety. Molecular psychiatry. 2009;14(7):681–95. Epub 2009/01/21. pmid:19153574.
  58. 58. Juhasz G, Dunham JS, McKie S, Thomas E, Downey D, Chase D, et al. The CREB1-BDNF-NTRK2 pathway in depression: multiple gene-cognition-environment interactions. Biological psychiatry. 2011;69(8):762–71. Epub 2011/01/11. pmid:21215389.
  59. 59. Kunugi H, Hori H, Adachi N, Numakawa T. Interface between hypothalamic-pituitary-adrenal axis and brain-derived neurotrophic factor in depression. Psychiatry and clinical neurosciences. 2010;64(5):447–59. pmid:20923424
  60. 60. Tognoli C, Rossi F, Di Cola F, Baj G, Tongiorgi E, Terova G, et al. Acute stress alters transcript expression pattern and reduces processing of proBDNF to mature BDNF in Dicentrarchus labrax. BMC neuroscience. 2010;11:4. Epub 2010/01/16. pmid:20074340; PubMed Central PMCID: PMC2829032.
  61. 61. Neeley EW, Berger R, Koenig JI, Leonard S. Prenatal stress differentially alters brain-derived neurotrophic factor expression and signaling across rat strains. Neuroscience. 2011;187:24–35. Epub 2011/04/19. pmid:21497180; PubMed Central PMCID: PMC3114039.
  62. 62. Yeh CM, Huang CC, Hsu KS. Prenatal stress alters hippocampal synaptic plasticity in young rat offspring through preventing the proteolytic conversion of pro-brain-derived neurotrophic factor (BDNF) to mature BDNF. The Journal of physiology. 2012;590(Pt 4):991–1010. Epub 2011/12/14. pmid:22155932; PubMed Central PMCID: PMC3381323.
  63. 63. Segawa M, Morinobu S, Matsumoto T, Fuchikami M, Yamawaki S. Electroconvulsive seizure, but not imipramine, rapidly up-regulates pro-BDNF and t-PA, leading to mature BDNF production, in the rat hippocampus. The international journal of neuropsychopharmacology / official scientific journal of the Collegium Internationale Neuropsychopharmacologicum (CINP). 2013;16(2):339–50. Epub 2012/02/09. pmid:22310305.
  64. 64. Mandel AL, Ozdener H, Utermohlen V. Identification of pro- and mature brain-derived neurotrophic factor in human saliva. Archives of oral biology. 2009;54(7):689–95. Epub 2009/05/27. pmid:19467646; PubMed Central PMCID: PMC2716651.
  65. 65. Yoshida T, Ishikawa M, Iyo M, Hashimoto K. Serum levels of mature Brain-Derived Neurotrophic Factor (BDNF) and its precursor proBDNF in healthy subjects. The Open Clinical Chemistry Journal. 2012;5:7–12.
  66. 66. Manolio TA, Collins FS, Cox NJ, Goldstein DB, Hindorff LA, Hunter DJ, et al. Finding the missing heritability of complex diseases. Nature. 2009;461(7265):747–53. Epub 2009/10/09. pmid:19812666; PubMed Central PMCID: PMC2831613.
  67. 67. Lee SY, Chen SL, Chen SH, Chu CH, Chang YH, Lin SH, et al. Interaction of the DRD3 and BDNF gene variants in subtyped bipolar disorder. Progress in neuro-psychopharmacology & biological psychiatry. 2012;39(2):382–7. Epub 2012/08/11. pmid:22877924.
  68. 68. Craddock N, O'Donovan MC, Owen MJ. The genetics of schizophrenia and bipolar disorder: dissecting psychosis. Journal of medical genetics. 2005;42(3):193–204. Epub 2005/03/04. pmid:15744031; PubMed Central PMCID: PMCPMC1736023.
  69. 69. Serretti A, Mandelli L. The genetics of bipolar disorder: genome 'hot regions', genes, new potential candidates and future directions. Molecular psychiatry. 2008;13(8):742–71. Epub 2008/03/12. pmid:18332878.
  70. 70. Craddock N, Sklar P. Genetics of bipolar disorder: successful start to a long journey. Trends in genetics: TIG. 2009;25(2):99–105. Epub 2009/01/16. pmid:19144440.
  71. 71. Levinson DF. The genetics of depression: a review. Biological psychiatry. 2006;60(2):84–92. Epub 2005/11/23. pmid:16300747.
  72. 72. Lohoff FW. Overview of the genetics of major depressive disorder. Current psychiatry reports. 2010;12(6):539–46. Epub 2010/09/18. pmid:20848240; PubMed Central PMCID: PMCPMC3077049.