1. ¹Department of Psychiatry, Neurogenetics Section, Centre for Addiction and Mental Health, University of Toronto, 250 College street R-30, Toronto, Ontaria Canada M5T 1R8
2. ²Instituto Gulbenkian de Ciência, Oeiras, Portugal

Correspondence: JL Kennedy, E-mail: James_Kennedy@camh.net

Converging lines of evidence suggest that the brain-derived neurotrophic factor (BDNF) could be implicated in the neurodevelopmental abnormalities found in schizophrenia brain.¹ BDNF is involved in a variety of trophic and neuromodulatory effects that include an important role in the development and survival of dopaminergic and serotonergic neurons. BDNF levels are increased in hippocampus and anterior cingulate cortex of schizophrenia brains,² a rat model of schizophrenia shows altered expression of the BDNF gene (BDNF), BDNF knock-out mice exhibit dysfunction of dopaminergic and serotonergic transmission, and neuroleptics appear to influence BDNF expression.³ BDNF is expressed throughout life, and thus altered BDNF expression or modified protein structure determined by variations in the gene could have a cumulative effect and be at least partly responsible for the subtle anomalies described in schizophrenia brains.¹

BDNF has been mapped to chromosome 11p13 and a (GT)_n dinucleotide repeat is present 1.04 kb upstream from the transcription initiation site.⁴ This polymorphism has been tested for association with schizophrenia in five independent studies.^5,6,7,8,9 Three of these studies^6,7,8 used a case–control association strategy, one⁹ used a family-based approach, while the other study used both the family and the case–control approach.⁵ None of these data sets showed association between the BDNF dinucleotide repeat and schizophrenia. However, Krebs et al⁶ reported the presence of an association between a group of BDNF dinucleotide long alleles (172–176 bp) and late-onset schizophrenia patients who responded to neuroleptics. In the present study, we tested for association between the BDNF dinucleotide polymorphism and schizophrenia in a sample of nuclear families. The DNAs from 89 schizophrenia patients and both respective parents were available. The DNAs were extracted from lymphocytes of patients assessed at the Department of Neuropsychiatric Science of the San Raffaele Hospital in Milan, the Department of Clinical Psychiatry at the University of Rome, Italy, and at the Centre for Addiction and Mental Health, Clarke Site, an affiliated hospital of the University of Toronto, Canada. All patients were diagnosed according to DSM-III-R criteria, by at least two psychiatrists, following administration of the Diagnostic Interview Schedule for the Italian sample and of the Structured Clinical Interview for DSM (SCID) in the Canadian sample. The dinucleotide polymorphism was genotyped using the method originally described by Proschel et al.⁴ We analyzed our families for the presence of association using the transmission disequilibrium test for multiallelic polymorphisms as in ETDT (v1.8). The ETDT allele-wise and genotype-wise analyses showed biased transmissions of the BDNF (GT)_n alleles from the parents to the schizophrenia probands (allele-wise ²=7.9, 3 df, P=0.04; genotype-wise ²=12.2, 6 df, P=0.05, and empirical P values via Monte Carlo ETDT: allele-wise P = 0.04; genotype-wise P=0.08, single allele P=0.02). The analysis of transmissions of the individual alleles (see Table 1) showed that allele 3 (170 bp) was more often transmitted (42 times) to the probands than nontransmitted (21 times). The same pattern of biased transmissions was observed when the data set was subdivided according to the two ethno-cultural groups present in our sample. Allele 3 showed 28 transmissions vs 15 nontransmissions in the Canadian families who were primarily of northern European ancestry, while the Italians showed 14 transmissions vs six nontransmissions. The analysis of the allele transmissions stratified for the parents' sex showed a parent of origin effect (POE) since preferential transmissions of allele 3 derived from the maternal meiosis were observed (maternal 21 transmissions vs six nontransmissions, P=0.003; paternal 16 transmissions vs 10 nontransmissions, P=0.2).

Table 1 - ETDT analysis for individual alleles of the BDNF (GT)_n polymorphism in 89 triad families of schizophrenia probands.

Full table

Our sample showed modest evidence for an association between the 170 bp dinucleotide repeats BDNF allele and schizophrenia. The increase in the number of transmissions for the allele 3 continues to be marginally significant (P=0.03) after Bonferroni correction for four multiple tests performed to analyze the transmissions of each allele independently. However, most previous studies did not show the presence of association. Several nonmutually exclusive factors can produce conflicting results in association studies of common complex diseases including the presence of genetic, clinical and population heterogeneity. One of the previous studies was conducted in Japanese patients,⁷ one in Roscommon County in Ireland,⁵ one in 48 triads from Iowa,⁹ and one in Spanish Caucasians.⁸ The study by Krebs et al⁶ showed evidence for association between a group of BDNF dinucleotide alleles (ranging from 172 to 176 bp) and a small sample of French Caucasian patients with late onset of the disease and good response to neuroleptics.⁶ The association reported in this study could explain the apparently contrasting results between the present report and previous studies.

The POE we observed in the transmission of the BDNF allele could be due to a parent-specific expression of BDNF, named genomic imprinting. BDNF is located between two imprinted loci: 24 Mb centromeric from the oxysterol binding protein homologue 1 gene (OBPH1) on 11p15.5 and 4 Mb telomeric from Wilson tumor 1 gene (WT1) on 11p13, a brain expressed zinc finger protein maternally expressed in the brain.¹⁰ However, the presence of imprinting has never been demonstrated for BDNF thus far.

The functions of BDNF described are strongly supportive of a role of this neurotrophin in processes that may be disrupted in schizophrenia.¹ A sustained qualitative or quantitative abnormality in the synthesis of BDNF determined by a defective gene could be involved in the anomalies described in post-mortem studies of schizophrenic brains. Furthermore, a defective BDNF could determine an altered BDNF synthesis in response to numerous environmental factors that have a demonstrated ability to modify BDNF expression.

In conclusion, a role for BDNF as a predisposing gene for schizophrenia is suggested by our result; however, its meaning in the context of several other negative studies requires further evaluation. The GT polymorphism is in proximity of the BDNF promoter region. A direct functional role of the repeat is needed considering the numerous lines of evidence showing the function of VNTRs in non-coding regions as transcriptional activating elements.¹¹ Furthermore, several regulatory elements within BDNF have been described¹² and the presence of linkage disequilibrium between the GT repeat and polymorphisms in these regions need also to be addressed.