Met66 in the brain-derived neurotrophic factor (BDNF) - Nature

0 downloads 0 Views 159KB Size Report
precursor is associated with anorexia nervosa restrictive type ... were found in two patients with anorexia nervosa (AN) and consisted of single-nucleotide.
Molecular Psychiatry (2003) 8, 745–751 & 2003 Nature Publishing Group All rights reserved 1359-4184/03 $25.00 www.nature.com/mp

ORIGINAL RESEARCH ARTICLE

Met66 in the brain-derived neurotrophic factor (BDNF) precursor is associated with anorexia nervosa restrictive type M Ribase´s1,2, M Grataco`s1,3, L Armengol1,2, R de Cid1,2, A Badı´a4, L Jime´nez4, R Solano4, J Vallejo4, F Ferna´ndez4 and X Estivill1,2,5 1

Genes ad Disease Program, Center for Genomic Regulation, Barcelona, Spain; 2Medical and Molecular Genetics CenterFIRO, Barcelona, Spain; 3Fundacio´ August Pi i Sunyer, Ciutat Sanita`ria Bellvitge, L’Hospitalet de Llobregat, Barcelona, Spain; 4Servei de Psiquiatria, Hospital de Bellvitge, Ciutat Sanita`ria Bellvitge, L’Hospitalet de Llobregat, Barcelona, Spain; 5 Departament de Cie`ncies de la Salut i de la Vida, Universitat Pompeu Fabra, Barcelona, Spain Several lines of evidence support a role for brain-derived neurotrophic factor (BDNF) alterations in the etiology of eating disorders (EDs). BDNF heterozygous knockout mice show alterations in eating behavior, increased body weight and adipocyte hypertrophy. BDNF also regulates the synaptic efficiency through the modulation of key neurotransmitter systems previously known to be involved in ED. These findings, together with the fact that this neurotrophin is expressed in the hypothalamus nuclei associated with weight regulation and feeding control, led us to propose BDNF as a candidate gene for ED. To investigate the possible involvement of this neurotrophin in eating behavior, we screened the BDNF gene in 95 ED patients and identified four sequence variants. Two of them, 374A/T and 256G/A, were found in two patients with anorexia nervosa (AN) and consisted of single-nucleotide mutations within the 50 untranslated region (50 UTR). The other two polymorphisms resulted in a C to T transition located at the 50 UTR of the BDNF gene and an amino-acid substitution within the BDNF precursor protein (Val66Met). We performed a case–control study for these two Single-nucleotide polymorphisms in a sample of 143 ED patients and 112 unrelated controls and found a strong association of restricting AN (ANR) with the Met allele of the Val66Met BDNF polymorphism (2p ¼ 0.002). There was also evidence for a significant effect of this sequence variant on the minimum body mass index (MBMI) (2p ¼ 0.006). These results suggest that the BDNF Met66 variant may be a susceptibility factor to ED, mainly to ANR and low MBMI. Molecular Psychiatry (2003) 8, 745–751. doi:10.1038/sj.mp.4001281 Keywords: anorexia; bulimia; brain-derived neurotrophic factor; BDNF; single-nucleotide polymorphism

Eating disorders (EDs) are complex and multifactorial, with the involvement of both psychosocial and biological factors.1–6 A genetic contribution in ED is widely accepted6–9 and genes involved in eating behavior, weight regulation, satiety and metabolism, as well as those under hormonal control are considered candidate genes for their role in anorexia nervosa (AN) and bulimia nervosa (BN). Among these candidates, we propose brain-derived neurotrophic factor (BDNF), which encodes for a neurotrophic factor with an essential role in neuronal survival and differentiation, and is also involved in synaptic efficiency and neuronal plasticity.10–13 BDNF maps

Correspondence: Dr X Estivill, Genes and Disease Program, Center for Genomic Regulation, Passeig Marı´tim 37-49, E08003-Barcelona, Catalonia, Spain. E-mail: [email protected] Received 9 May 2002; revised 26 July 2002, 27 August 2002 and 19 September 2002; accepted 29 September 2002

to 11p13–p1414 and, although the structure of the coding exon is well known,15 the genomic organization of the 50 noncoding exons has not been described. Several lines of evidence indicate a role of this neurotrophin in eating behavior and suggest that alterations in its function or expression pattern could be considered susceptibility factors to ED. To test this hypothesis we have characterized the genomic structure of the human BDNF gene. BDNF is composed of, at least, four 50 nontranslated exons and a single 30 coding exon (exon 5; Figure 1b). Since mRNA transcripts were different at their 50 ends, we determined the existence of four 50 untranslated exons that give rise to four distinct mRNA isoforms (named A, B, C and D), as previously described in the rat BDNF gene.16,17 Exon 1 generates the mRNA isoform A and is 85% identical to the rat exon 3. Exon 2 gives rise to the mRNA isoform B and corresponds to a previously described promoter region of the human BDNF gene.18 Moreover, mRNA

BDNF variant in eating disorders M Ribase´s et al

746

Figure 1 Variants and gene structure of the BDNF gene. (a) Changes in the SSCP patterns corresponding to the 270C/T, 196G/A (Val66Met), 256G/A and 374T/A BDNF variants. Genotyping of three of the identified sequences variants by PCRRFLP. Comparison of DNA sequences of both normal and abnormal SSCP fragments for both of the novel SNPs (256G/A and 374T/A). (b) BDNF gene structure based on alignments of mRNA, EST and genomic sequences of the GenBank database. The BDNF gene is composed of, at least, four 50 nontranslated exons and a single 30 coding exon that give rise to four alternative transcripts different at their 50 ends (Isoforms A to D). CL: cleavage site.

and EST sequences gi: 1336884 and gi: 179404 revealed the presence of two other exons (exons 3 and 4); while isoform C results from alternative splicing of exon 4, the 50 region of isoform D corresponds to both exons 3 and 4. In order to identify sequence variants, we screened the BDNF gene by PCR-SSCP in 95 patients with ED. By sequencing the PCR products showing aberrant SSCP patterns, four nucleotide variants were identified (Figure 1a). We first detected a T to A transversion in exon 3 located 374 bp upstream from the BDNF coding region. This sequence variant does Molecular Psychiatry

not affect any consensus sequence and was found in heterozygosity in a patient with ANR (22 years, minimum body mass index (MBMI) 17.2 kg/m2, age of onset of weight loss (AAOC) 17 years and age of onset of amenorrhea (AAOA) 18 years). The second change identified within exon 3 consisted of a G to A transition located 256 bp upstream from the BDNF coding region. This variant affects the splicing donor consensus site and was found in heterozygosity in a patient with binge-eating/purging anorexia (ANBP; 37 years, MBMI 11.5 kg/m2, and AAOC 18 years). We genotyped 65 unrelated controls for these two muta-

BDNF variant in eating disorders M Ribase´s et al

tions but none of them was detected. The third sequence variant was a previously described singlenucleotide polymorphism (SNP) consisting of a C to T transition within exon 2, located 270 bp upstream from the BDNF coding region.19 This SNP does not affect any consensus site and was found in 8% of the ED patients. We also detected a previously described SNP consisting of a missense change (196G/A, Val66Met)15 in the coding exon of the BDNF gene, which was present in 44% of patients with ED. A case–control study was carried out to determine the potential role of these two SNPs in ED. The initially screened group was increased to a total sample of 143 patients. This group and 112 unrelated controls matched for ethnicity and sex were genotyped for the 270C/T and Val66Met SNPs. A power analysis was performed and revealed a statistical power of 99 and 100% for the 270C/T and Val66Met SNPs, respectively. Under the hypothesis that BDNF may confer susceptibility to the ED subtypes in different ways and to reduce heterogeneity, our patients were subdivided into four main diagnostic groups that have been shown to differ in various biological and psychopathological features.20–22 Thus, AN, ANR, ANBP and BN diagnoses were separately considered in the statistical analyses. To examine the association of both BDNF SNPs with ED, we compared allele and genotype frequencies between the patient and control groups. Both polymorphisms followed a distribution in Hardy– Weinberg equilibrium. No significant associations were found in any of the ED groups when both allele and genotype frequencies of the 270C/T SNP were considered. Nevertheless, genotype frequencies of the Val66Met polymorphism did significantly differ between the ED (w2 ¼ 8.55; d.f. ¼ 2; P ¼ 0.012), the AN (w2 ¼ 9.43; d.f. ¼ 2; P ¼ 0.008) and the ANR (w2 ¼ 11.64;

d.f. ¼ 2; P ¼ 0.002) groups and controls (Table 1). Moreover, once allele frequencies were compared, there was an increased number of ED patients carrying the Met66 allele with respect to controls, reaching statistical significance in the ED group (w2 ¼ 4.97; d.f. ¼ 1; P ¼ 0.032; OR ¼ 1.67; 95% CI ¼ 1.06–2.64), the AN group (w2 ¼ 4.42; d.f. ¼ 1; P ¼ 0.045; OR ¼ 1.78; 95% CI ¼ 1.03–3.07) and the ANR group (w2 ¼ 7.7; d.f. ¼ 1; P ¼ 0.005; OR ¼ 2.71; 95% CI ¼ 1.36–5.38) (Table 2). After the most conservative multiple comparison correction, the Met66 variant was still positively associated with ANR, but no associations were found in any of the other groups. Haplotype analysis gave no significant P-values (P ¼ 0.13 for the model-free analysis and P ¼ 0.11 for the heterogeneity model). No linkage disequilibrium between both BDNF SNPs was observed when ED patients (w2 ¼ 2.22; P ¼ 0.69) or controls were considered (w2 ¼ 0.63; P ¼ 0.97). This lack of linkage disequilibrium may explain why we observed association of ANR with the Val66Met but not with the 270C/T SNP. We also examined the relation between the Val66Met variant and ED-related phenotypes, such as MBMI and AAOC (Table 3). Thus, mean scores were compared between ED subjects carrying the Met66 allele and those homozygous for the wild-type allele of the Val66Met variant. While no significant effect on AAOC was observed, the mean MBMI was significantly lower in the Met66 carriers than in noncarriers (P ¼ 0.006). The major finding of this study is a strong association of ANR and low MBMI with the Met66 allele of the Val66Met BDNF polymorphism. These results support our initial hypothesis about a contribution of BDNF to alterations in eating behavior and body weight regulation. Two different mechan-

747

Table 1 Distribution of genotypes for the BDNF 270C/T and 196G/A (Val66Met) SNPs in 143 patients with EDs and 112 control subjects BDNF 270C/T n Genotypes (%) Phenotype Eating disorders (n=143) Anorexia nervosa (n=64) Anorexia restrictive (n=26) Anorexia binge-eating purging (n=36) Bulimia nervosa (n=70) Controls (n=112)

C/C 131 (91.6) 58 (90.6) 25 (96.1) 31 (86.1) 65 (92.8) 102 (91)

C/T 11 (7.7) 5 (7.8) 1 (3.9) 4 (11.1) 5 (7.2) 10 (9)

T/T 1 (0.07) 1 (1.6) 0 1 (2.8) 0 0

BDNF Val66Met w2 P (2 d.f.) w2=0.87 2p=0.89 w2=1.7 2p=0.53 w2=0.74 2p=0.69 w2=3.32 2p=0.26 w2=0.18 2p=0.78

n Genotypes (%) Val/Val 81 (56.6) 34 (53.1) 10 (38.4) 23 (63.9) 40 (57.1) 82 (73.2)

Val/Met 58 (40.6) 29 (45.3) 15 (57.8) 13 (36.1) 27 (38.6) 26 (23.2)

Met/Met 4 (2.8) 1 (1.6) 1 (3.84) 0 3 (4.3) 4 (3.6)

w2 P (2 d.f.) w2=8.55 2p=0.012* w2=9.43 2p=0.008* w2=11.6 2p=0.002** w2=2.77 2p=0.19 w2=5.24 2p=0.07

d.f.: degrees of freedom; OR: odds ratio; CI: confidence interval; *P-values statistically significant (Po0.05); **P-values statistically significant after Bonferroni correction (Po0.006). Molecular Psychiatry

BDNF variant in eating disorders M Ribase´s et al

748

Table 2 Distribution of alleles for the 270C/T and the 196G/A (Val66Met) SNPs in 143 patients with EDs and 112 control subjects 270C/T BDNF n Alleles (%) Phenotype Eating disorders Anorexia nervosa Anorexia restrictive Anorexia binge- eating/purging Bulimia nervosa Controls

C 275 (95.5) 121 (94.5) 51 (98.1) 66 (91.6) 135 (96.4) 214 (95.5)

T 13 (4.5) 7 (5.5) 1 (1.9) 6 (8.4) 5 (3.6) 10 (4.5)

Val66Met BDNF

w2; P (1d.f.) OR (95% CI)

w2; P (1d.f.) OR (95% CI)

n Alleles (%) Val 220 (77.0) 97 (75.8) 35 (67.3) 59 (82.0) 107 (76.4) 190 (84.2)

w2=0.00; 2p=1.00 0.98 (0.42–2.29) w2=0.03; 2p=0.79 0.8 (0.30–2.17) w2=0.21; 2p=0.69 2.3 (0.29–19.0) w2=0.92; 2p=0.23 0.51 (0.18–1.46) w2=0.02; 2p=0.79 1.26 (0.42–3.77)

Met 66 (23.0) 31 (24.2) 17 (32.7) 13 (18.0) 33 (23.6) 34 (15.2)

w2=4.97; 2p=0.032* 1.67 (1.06–2.64) w2=4.42; 2p=0.045* 1.78 (1.03–3.07) w2=7.7; 2p=0.005** 2.71 (1.36–5.38) w2=0.15; 2p=0.58 1.23 (0.61–2.48) w2=3.5; 2p=0.05 1.72 (1.01–2.94)

d.f.: degrees of freedom; OR: odds ratio; CI: confidence interval; *P-values statistically significant (Po0.05); **P-values statistically significant after Bonferroni correction (Po0.006).

Table 3 Means and standard deviations of age at onset of weight loss (AAOC) and minimum body mass index (MBMI) according to the Val66Met variant in patients with EDs Val66Met BDNF genotypes

Phenotypes AAOC (years) MBMI (kg/m2)

Val/Val (n=72)

Val/Met or Met/Met (n=53)

18.27 (4.98) 17.95 (3.25)

17.9 (3.49) 16.51 (2.53)

P-value*

Mean difference

Standard error difference

95% CI

0.628 0.006**

0.36 1.4

0.74 0.51

1.11 to 1.84 0.41 to 2.46

*T-test for two independent samples (Val66 allele carriers versus noncarriers).

**P-values considered statistically significant (Po0.05).

isms could explain this association. First, the Met66 variant could be in linkage disequilibrium with a yet unknown nearby susceptibility gene directly involved in ANR or low MBMI. Alternatively, as it is located in the prodomain of the BDNF precursor this variant itself could exert a direct biological action and be a susceptibility factor to ED. Thus, the Met66 allele could genetically determine alterations in its function or stability, in the folding of the mature protein,23,24 in its sorting to either constitutive or regulated secretor pathways or in the neurotrophin spatio-temporal expression pattern. Moreover, the BDNF Met66 variant could also determine variations in the biosynthesis and the post-translational processing of the BDNF precursor.25 The role of BDNF in ED is unknown, but several lines of investigation suggest that increases in the levels of this neurotrophin could be involved in some ANR characteristics, such as dietary restriction behaviors and body weight reduction. Thus, BDNF heterozygous ‘knockout’ mice develop obesity, Molecular Psychiatry

showing eating behavior alterations, increased body weight and adipocyte cellular hypertrophy.1 These results, together with the fact that BDNF increases extracellular serotonin levels3,4,26 and that its expression is under estrogen control,27,28 suggest that BDNF variants associated with ANR may lead to a gain-of-function related to increases in the neurotrophin levels, function or stability. BDNF could influence the susceptibility to ANR and low MBMI through direct effect on food consumption and control of body weight, as it is expressed in the hypothalamic centers involved in these activities. Alternatively, since BDNF has also a key role in the modulation of different neurotransmitters, it could indirectly participate in ANR and low MBMI through changes in these systems previously involved in ED.2–5 The contribution of distinct susceptibility factors to the different ED subtypes, which is supported by other studies where differential personality and psychopathological traits were found between purging (ANBP or BN) and nonpurging ED (ANR),20–22

BDNF variant in eating disorders M Ribase´s et al

could explain the lack of association of the Val66Met polymorphism with ANBP and BN. Moreover, because of its clinical homogeneity, the ANR sample may represent a distinctive subgroup showing a higher genetic homogeneity, therefore making it easier to identify those genetic factors that contribute to this clinical subtype. Alternatively, our negative results in the ANBP or BN subgroups could also be explained by the relative small sample size and the reduced statistical power of the test obtained once the ED group was subdivided according to diagnoses. Neither association was found in any ED group when the 270C/T SNP was considered, suggesting that it may not be involved in eating behavior alterations. Nevertheless, it would be necessary to test it in a larger sample because of its low frequency in the studied populations. The other two BDNF variants identified in two AN patients, 374A/T and 256G/ A, were located at the 50 noncoding region and could also be involved in the modulation of stability, transcription levels or expression pattern of the BDNF isoforms. Moreover, 256G/A may be implicated in the splicing function as it affects the splicing donor site of exon 3. In conclusion, the results of our study suggest that the BDNF Met66 variant may be involved in the susceptibility to ED and that its participation is highly significant in ANR and low MBMI. However, they should be considered as preliminary until replicated in larger samples, other populations and family trios to avoid population stratifications that lead to false-positive associations. Once we confirm this association, functional studies will be necessary to determine mechanisms by which alterations in the BDNF system could contribute to the development of the ED.

Materials and methods Study subjects The clinical sample consisted of 143 Spanish Caucasian patients with ED that were consecutively admitted in the Psychiatric Unit of the Hospital de Bellvitge between April 1999 and December 2001. The group consisted of 70 BN cases (49%; 63 bingeeating/purging vs seven nonpurging subtype), 64 AN cases (45%; 36 bingeing/purging, 26 restricting subtype and two anorexia cases with insufficient time elapsed (o3 years restricting illness) to classify them as restricting or binge-eating/purging subtype) and nine patients with ED not otherwise specified (EDNOS; 6%). All subjects fulfilled DSM-IV criteria for these disorders (APA, 1994) and were diagnosed using the Structured Clinical Interview for Mental Disorders, research version 2.0, (SCID-I). Diagnosis was blind to genotype. Most of the patients were female (N ¼ 130; 91%), and 20% of the studied AN sample has been described and assessed in a previous report.29 The lifetime MBMI was 18.9 kg/m2 (SD ¼ 2.78) for BN patients, 15.14 kg/m2 (SD ¼ 1.51) for AN patients and 19.18 kg/m2 (SD ¼ 3.63) for

EDNOS patients. The average age at assessment was 25.4 years old (SD ¼ 4.89) for BN patients, 24.8 years old (SD ¼ 5.8) for AN patients and 25.1 years old (SD ¼ 4.34) for EDNOS patients. The average duration of the disorder was 7.15 years (SD ¼ 4.64) for BN patients, 6.15 years (SD ¼ 3.81) for AN patients and 8.3 years (SD ¼ 4.84) for EDNOS patients. The control sample consisted of 112 white Spanish Caucasian unrelated subjects matched for ethnicity and sex. The study was approved by the ethics committee of our Institution (Ciutat Sanitaria i Universitaria Bellvitge) and written informed consent was obtained from all subjects.

749

BDNF genomic organization BDNF genomic organization was determined in silico using the sequence similarity search program PSIBLAST 2.2. Exon/intron boundaries were predicted from the public database sequences by alignment of the RP11-651M4 clone sequence of human chromosome 11 (GenBank accession nos. 9838303 and 13518154) and the genomic sequence (GenBank accession no. 2174122), cDNA sequences (GenBank accession nos. 3828269, 987871, 1336884 and 179404) and EST sequences (GenBank accession no. 1367883) of the BDNF gene. The results obtained were also compared with the exon sequences described for the rat BDNF gene (GenBank accession nos. 311226 and 557909). PCR-SSCP and sequencing Eleven primer pairs were designed covering the entire coding region (exon 5), the exon/intron boundaries, including the splicing sites and the 50 nontranslated regions of the BDNF gene. Primer pair sequences consisted of pair 1: 50 -GATCAAATGGAGCTTCTCGC30 and 50 -AGTAACCATCAAGGCAGCTG-30 ; pair 2: 50 -CTCCCCCCTTTAATTAAGCG-30 and 50 -AAGCCACACGCTTTCTAGCC-30 ; pair 3: 50 -CCCGCGCGAGGAAAAG-30 and 50 -TAAAGCCCCCCGCGCAGG-30 ; pair 4: 50 -ATCGAAGCTCAACCGAAGAGC-30 and 50 GTCCACACAAACCTCACGGG-30 ; pair 5: 50 -GAGCCAGAATCGGAACCACG-30 and 50 -TCTACCGGAGGGGAGGAAAG-30 ; pair 6: 50 -GCTGTTCCTGTCACACATTC-30 and 50 -AATTCCAGGCCATTCTGCAG-30 ; pair and 50 7: 50 -TGTGTCTGGTGCAGCTGGAG-30 CCTTGTCCTCGGATGTTTGC-30 ; pair 8: 50 -AGGTGAGAAGAGTGATGACC-30 and 50 -CTGGACGTGTACAAGTCTGC-30 ; pair 9: 50 -CTGACACTTTCGAACACGTG-30 and 50 -CCTTTTCAAGGACTGTGACC-30 ; pair 10: 50 -GAGCTGAGCGTGTGTGACAG-30 and 50 -CTATCCATGGTAAGGGCCCG-30 ; and pair 11: 50 -TGGAACTCCCAGTGCCGAAC-30 and 50 -CAGTTCTTGGCAACGGCAAC-30 . Fragments were amplified from DNA samples of 95 patients with ED by PCR reaction in a total volume of 10 ml containing 50 ng of template DNA, 50 mM KCl, 10 mM Tris-HCl pH 8.3, 1.5 mM MgCl2, 200 mM of dNTP (Pharmacia Biotech), 10 pmol of each oligonucleotide (Life Technologies) and 0.25 U of Taq DNA polymerase (Boehringer Mannheim). Amplification conditions consisted of an Molecular Psychiatry

BDNF variant in eating disorders M Ribase´s et al

750

initial 4 min denaturation step at 941C, 32 cycles of 30 s at 941C, 30 s at the annealing temperature according to the melting temperature of each primer and 30 s at 721C, followed by a final extension of 10 min at 741C. For the SSCP analysis, 2 ml of each PCR reaction was diluted in formamide containing buffer (v/v), denatured at 941C during 4 min and electrophoresed on polyacrylamide gels in 1  TBE buffer (41C, 300 V). Subsequently, gels were silver stained and those samples showing an aberrant SSCP banding pattern were re-amplified, sequenced with a commercial kit (Big Dye Terminator Cycle Sequencing Ready Reaction Kit, Applied Biosystems) and analyzed on an automatic sequencer (ABI PRISM 337XL). Once sequence variants were identified, their effect on the restriction map was determined by the DNA Strider 1.0.1 software. Then, 3 ml of the corresponding PCR products were digested overnight at 371C in a total volume of 10 ml with 5 U of HinfI (Biolabs) for 270C/T, DdeI (Roche) for 256G/A and NlaIII (Biolabs) for Val66Met and resolved on 12% polyacrylamide gels (19 : 1). Statistical analysis The power analysis was performed post hoc on the ED group as a whole with the Power Calculator software (Department of Statistics of the University of Los Angeles; http://ebook.stat.ucla.edu/calculators/powercalc) assuming a lifetime risk of 6% for first relatives of a patient with ED and a significance level of 0.05. Average minimum BMI, age at assessment and duration of the disorder in each studied group were measured by the statistical package SPSS 10.0. The distribution of genotypes for the different populations was tested for the Hardy–Weinberg equilibrium by a w2 analysis using the INSTAT Graphpad software. As the sample size was too small to use a standard w2 test, the comparison of both allelic and genotypic frequencies between the control and the ED groups was assessed by the two-tailed Fisher’s exact test using the statistical package SPSS 10.0. Patients with ED were analyzed as a unique group and also subgrouped according to the clinical subtype (AN, ANR, ANBP and BN) in all statistical tests. As we subdivided our ED patients into four diagnostic subgroups and two genetic markers were analyzed, the Bonferroni correction was used for w2 test and corrected P-values were considered significant when below 0.006. Haplotype association analysis and linkage disequilibrium tests were performed using the pmplus program.30 Differences in mean scores of EDs-related phenotypes, such as MBMI and AAOC, were assessed by t-tests using the statistical package SPSS 10.0. Significance levels were set at Po0.05.

Acknowledgements We thank the patients for their participation in the study, and M Dierssen and H Kruyer for their helpful comments on the manuscript. MR is supported by the Molecular Psychiatry

Fondo de Investigaciones Sanitarias de la Seguridad Social (FISS), Instituto de Salud Carlos III. MG is supported by the EU (ECF5000916). XE is a Senior Scientist of the Centre de Regulacio´ Geno`mica (CRG). Financial support was received from the European Union (Framework-V Multicentre Research Grant, QLK1-1999-916) for FF and XE, from the Red FIS G03/184 and from the Departament d’Universitats Recerca i Societat de la Informacio´ (DURSI) and the Departament de Sanitat for XE.

References 1 Kernie SG, Liebl DJ, Parada LF. BDNF regulates eating behavior and locomotor activity in mice. EMBO J 2000; 19: 1290–1300. 2 Knusel B, Winslow JW, Rosenthal A, Burton LE, Seid DP, Nikolics K et al. Promotion of central cholinergic and dopaminergic neuron differentiation by brain-derived neurotrophic factor but not neurotrophin 3. Proc Natl Acad Sci USA 1991; 88: 961–965. 3 Mamounas LA, Blue ME, Siuciak JA, Altar CA. Brain-derived neurotrophic factor promotes the survival and sprouting of serotonergic axons in rat brain. J Neurosci 1995; 15: 7929–7239. 4 Mossner R, Daniel S, Albert D, Heils A, Okladnova O, Schmitt A et al. Serotonin transporter function is modulated by brain-derived neurotrophic factor (BDNF) but not nerve growth factor (NGF). Neurochem Int 2000; 36: 197–202. 5 Guillin O, Diaz J, Carroll P, Griffon N, Schwartz JC, Sokoloff P. BDNF controls dopamine D3 receptor expression and triggers behavioural sensitization. Nature 2001; 411: 86–89. 6 Garner DM. Pathogenesis of anorexia nervosa. Lancet 1993; 341: 1631–1635. 7 Strober M. Family-genetic studies of eating disorders. J Clin Psychiatry 1991; 52 (Suppl): 9–12. 8 Kendler KS, MacLean C, Neale M, Kessler R, Heath A, Eaves L. The genetic epidemiology of bulimia nervosa. Am J Psychiatry 1991; 148: 1627–1637. 9 Grice DE, Halmi KA, Fichter MM, Strober M, Woodside DB, Treasure JT et al. Evidence for a susceptibility gene for anorexia nervosa on chromosome 1. Am J Hum Genet 2002; 70: 787–792. 10 Snider WD, Johnson Jr EM. Neurotrophic molecules. Ann Neurol 1989; 26: 489–506. 11 Barde YA. Neurotrophins: a family of proteins supporting the survival of neurons. Prog Clin Biol Res 1994; 390: 45–56. 12 Lo DC. Neurotrophic factors and synaptic plasticity. Neuron 1995; 15: 979–981. 13 Thoenen H. Neurotrophins and neuronal plasticity. Science 1995; 270: 593–598. 14 Hanson IM, Seawright A, van Heyningen V. The human BDNF gene maps between FSHB and HVBS1 at the boundary of 11, p13– p14. Genomics 1992; 13: 1331–1333. 15 Maisonpierre PC, Le Beau MM, Espinosa III R, Ip NY, Belluscio L, de la Monte SM et al. Human and rat brain-derived neurotrophic factor and neurotrophin-3: gene structures, distributions, and chromosomal localizations. Genomics 1991; 10: 558–568. 16 Ohara O, Gahara Y, Teraoka H, Kitamura T. A rat brain-derived neurotrophic factor-encoding gene generates multiple transcripts through alternative use of 50 exons and polyadenylation sites. Gene 1992; 121: 383–386. 17 Nakayama M, Gahara Y, Kitamura T, Ohara O. Distinctive four promoters collectively direct expression of brain-derived neurotrophic factor gene. Brain Res Mol Brain Res 1994; 21: 206–218. 18 Shintani A, Ono Y, Kaisho Y, Igarashi K. Characterization of the 50 -flanking region of the human brain-derived neurotrophic factor gene. Biochem Biophys Res Commun 1992; 182: 325–332. 19 Kunugi UA, Otsuka M, Isse K, Hirasawa H, Kato N, Nabika T et al. A novel polymorphism of the brain-derived neurotrophic factor (BDNF) gene associated with late-onset Alzheimer’s disease. Molec Psychiatry 2001; 6: 83–86. 20 Bulik CM, Sullivan PF, Carter FA, Joyce PR. Initial manifestations of disordered eating behavior: dieting versus binging. Int J Eat Disord 1997; 22: 195–201.

BDNF variant in eating disorders M Ribase´s et al

751 21 Bulik CM, Sullivan PF, Kendler KS. An empirical study of the classification of eating disorders. Am J Psychiatry 2000; 157: 886–895. 22 Gendall KA, Joyce PR, Sullivan PF, Bulik CM. Personality and dimensions of dietary restraint. Int J Eat Disord 1998; 24: 371–379. 23 Suter U, Heymach Jr JV, Shooter EM. Two conserved domains in the NGF propeptide are necessary and sufficient for the biosynthesis of correctly processed and biologically active NGF. EMBO J 1991; 10: 2395–2400. 24 Rattenholl A, Ruoppolo M, Flagiello A, Monti M, Vinci F, Marino G et al. Pro-sequence assisted folding and disulfide bond formation of human nerve growth factor. J Mol Biol 2001; 305: 523–533. 25 Mowla SJ, Farhadi HF, Pareek S, Atwal JK, Morris SJ, Seidah NG et al. Biosynthesis and post-translational processing of the precursor to brain-derived neurotrophic factor. J Biol Chem 2001; 276: 12660–12666.

26 Siuciak JA, Clark MS, Rind HB, Whittemore SR, Russo AF. BDNF induction of tryptophan hydroxylase mRNA levels in the rat brain. J Neurosci Res 1998; 52: 149–158. 27 Jezierski MK, Sohrabji F. Region- and peptide-specific regulation of the neurotrophins by estrogen. Brain Res Mol Brain Res 2000; 85: 77–84. 28 Sohrabji F, Miranda RC, Toran-Allerand CD. Identification of a putative estrogen response element in the gene encoding brainderived neurotrophic factor. Proc Natl Acad Sci USA 1995; 92: 11110–11114. 29 Gorwood P, Ades J, Bellodi L, Cellini E, Collier DA, Di Bella D et al. The 5-HT(2A) 1438G/A polymorphism in anorexia nervosa: a combined analysis of 316 trios from six European centres. Mol Psychiatry 2002; 7: 90–94. 30 Zhao JH, Curtis D, Sham PC. Model free analysis and permutation tests for allelic associations. Hum Hered 2000; 50: 133–139.

Molecular Psychiatry