Molecular Psychiatry (2010) 15, 831–843 & 2010 Macmillan Publishers Limited All rights reserved 1359-4184/10 www.nature.com/mp ORIGINAL ARTICLE Differences and similarities in the serotonergic diathesis for suicide attempts and mood disorders: a 22-year longitudinal gene–environment study J Brezo1, A Bureau2,3,CMe´rette2,4, V Jomphe2, ED Barker5,6,7, F Vitaro5,MHe´bert8, R Carbonneau5, RE Tremblay5,9,10 and G Turecki1,10 1The McGill Group for Suicide Studies, Douglas Hospital Research Centre, McGill University, Montreal, QC, Canada; 2Centre de recherche Universite´ Laval Robert-Giffard, Quebec City, QC, Canada; 3Department of Social and Preventive Medicine, Laval University, Quebec City, QC, Canada; 4Department of Psychiatry, Laval University, Quebec City, QC, Canada; 5Research Unit on Children’s Psychosocial Maladjustment, University de Montre´al, Montreal, QC, Canada; 6Social, Genetic and Developmental Psychiatry Centre (MRC), King’s College London, London, UK; 7Department of Psychology, Center for the Prevention of Youth Behavior Problems, University of Alabama, Tuscaloosa, AL, USA; 8Department of Sexology, Universite´ du Que´bec, Montreal, QC, Canada and 9International Laboratory for Child and Adolescent Mental Health Development (INSERM U669, Paris, France; University College Dublin, Ireland; University of Montreal, Canada) To investigate similarities and differences in the serotonergic diathesis for mood disorders and suicide attempts, we conducted a study in a cohort followed longitudinally for 22 years. A total of 1255 members of this cohort, which is representative of the French-speaking population of Quebec, were investigated. Main outcome measures included (1) mood disorders (bipolar disorder and major depression) and suicide attempts by early adulthood; (2) odds ratios and probabilities associated with 143 single nucleotide polymorphisms in 11 serotonergic genes, acting directly or as moderators in gene–environment interactions with childhood sexual or childhood physical abuse (CPA), and in gene–gene interactions; (3) regression coefficients for putative endophenotypes for mood disorders (childhood anxiousness) and suicide attempts (childhood disruptiveness). Five genes showed significant adjusted effects (HTR2A, TPH1, HTR5A, SLC6A4 and HTR1A). Of these, HTR2A variation influenced both suicide attempts and mood disorders, although through different mechanisms. In suicide attempts, HTR2A variants (rs6561333, rs7997012 and rs1885884) were involved through interactions with histories of sexual and physical abuse whereas in mood disorders through one main effect (rs9316235). In terms of phenotype-specific contributions, TPH1 variation (rs10488683) was relevant only in the diathesis for suicide attempts. Three genes contributed exclusively to mood disorders, one through a main effect (HTR5A (rs1657268)) and two through gene–environment interactions with CPA (HTR1A (rs878567) and SLC6A4 (rs3794808)). Childhood anxiousness did not mediate the effects of HTR2A and HTR5A on mood disorders, nor did childhood disruptiveness mediate the effects of TPH1 on suicide attempts. Of the serotonergic genes implicated in mood disorders and suicidal behaviors, four exhibited phenotype-specific effects, suggesting that despite their high concor- dance and common genetic determinants, suicide attempts and mood disorders may also have partially independent etiological pathways. To identify where these pathways diverge, we need to understand the differential, phenotype-specific gene–environment interactions such as the ones observed in the present study, using suitably powered samples. Molecular Psychiatry (2010) 15, 831–843; doi:10.1038/mp.2009.19; published online 21 April 2009 Keywords: suicide attempts; mood disorders; gene–environment interactions; gene-gene interactions; major depression; childhood sexual abuse Introduction difficulties for individuals, families and public health systems.1–3 SBs are closely, but not exclusively, Suicidal behaviors (SBs) and mood disorders (MDs) associated with MDs and represent their most serious cause considerable psychosocial and economic complications.4 Personal and family histories of MDs predict both attempted and completed suicide.5–7 Correspondence: Dr G Turecki, The McGill Group for Suicide Heritability figures indicate a multifactorial mode of Studies, Douglas Hospital Research Centre, McGill University, inheritance, with genes explaining 17–55%5 of the 6875, Boul La Salle, Montreal, QC H4H 1R3, Canada. variance in suicide attempts and 30–80% in major E-mail: [email protected] 8–10 10These authors share senior authorship of this work. depression and bipolar disorder. Several lines of Received 2 September 2008; revised 20 November 2008; accepted evidence point toward emotional, cognitive and beha- 14 January 2009; published online 21 April 2009 vioral deficits in both SBs and MDs.11,12 These deficits Serotonergic diathesis for MDs and suicide attempts J Brezo et al 832 have, in turn, been linked to alterations in the activity of important early environmental risk factor for SBs and serotonin—a master regulator of other neurotransmitters MDs.38–41 For example, suicide attempts and bipolar and a contributor to early brain development.13,14 Lower disorder severity have been linked to childhood concentration, binding, neurotransmission, and re- sexual (CSA) and physical abuse (CPA).42–44 Primate uptake of serotonin and its metabolites are markers of studies suggest that early life adversity alters stress the risk for suicidality and MDs.15,16 sensitivity—a feature of both SBs and MDs—compro- Several serotonergic genes (5HTT—also known as mising serotonergic neurotransmission.45,46 Stress is, SLC6A4, TPH1, TPH2, HTR1A, HTR2A, HTR2C and for example, associated with SLC6A4 mRNA level47 HTR7) have been examined as candidates for bipolar and harsh parenting to HTR2A receptor density.48 disorder, major depression and suicide attempts, most Recognizing the interconnectedness of genes and often with inconsistent findings.17–19 High concor- environments in susceptibility to MDs and SBs, we dance of MDs and SBs has made it difficult to investigated single nucleotide polymorphisms (SNPs) disentangle shared and unique genetic contributions in 11 serotonergic genes in a cohort of young French to these phenotypes.5 For example, it is unclear if Canadians followed since kindergarten. Our main variation in the gene coding for the serotonin objective was to identify variants associated with transporter—an extensively investigated candidate MDs and attempted suicide—a strong predictor of gene whose protein product controls the magnitude completed suicide49,50—that act either directly, as and duration of the serotonergic neurotransmis- moderators, or through mediators, that is, endophe- sion20—shows specificity for the diathesis for major notypes. Our main theoretical premise was that SNPs depression, bipolar disorder or suicide attempts, or involved in susceptibility to SBs and MDs would, whether it is relevant for all three.21,22 Similar directly or through linkage disequilibrium (LD), lead questions could be raised regarding all other seroto- to a reduction in serotonergic activity in a way that is nergic genes. conditional on the function of the particular gene Poor replicability of significant findings, methodo- (through dysregulation in uptake, binding or the logical and population heterogeneity, low statistical removal of serotonin from the synaptic cleft) and its power and limited knowledge about the precise roles interactions with two salient environmental factors of candidate genes23–25 have been invoked as reasons and with other examined serotonergic genes. behind the relative lack of success and poor under- We also set out to resolve differences in the standing of the diathesis of SBs and MDs despite two diatheses for MDs and suicide attempts, which, decades of active molecular research. In addition, despite their relatedness, may exist at the level of most candidate genes are rarely investigated within genes, environmental factors and endophenotypes. the context of other genes (for example, epistasis or We hypothesized that childhood disruptiveness, a gene–gene interactions), environments (gene–envir- construct related to impulsive aggression, and an- onment interactions and correlations) or possible xiousness, respectively, mediate genetic diatheses for endophenotypes. suicide attempts and MDs. Several lines of evidence We26–28 and others24,29–33 have shown that behavioral point toward emotional, cognitive and behavioral dysregulation reflected in impulsive aggression and deficits in both SBs and MDs.11,12 These deficits have, related traits may be an endophenotype because it in turn, been linked to alterations in the activity of has been linked to increased risk for SBs and to serotonin—a master regulator of other neurotransmit- the activity and variation of the serotonergic system. ters and a contributor to early brain development.13,14 Anxiousness, neuroticism and internalizing behav- Lower concentration, binding, neurotransmission, iors have also been linked to serotonergic variants as and reuptake of serotonin and its metabolites are well as to MDs.34–36 Personality traits such as these markers of the risk for suicidality and MDs.15,16 have been suggested as candidate endophenotypes33 In terms of the specific moderating effects, guided because of their quantitative nature, moderate-to-high by literature and theoretical evidence, we tested heritability and a closer temporal relationship