Molecular Psychiatry (2006) 11, 312–322 & 2006 Nature Publishing Group All rights reserved 1359-4184/06 $30.00 www.nature.com/mp ORIGINAL ARTICLE Why do young women smoke? I. Direct and interactive effects of environment, psychological characteristics and nicotinic cholinergic receptor L Greenbaum1,3, K Kanyas1,3, O Karni1, Y Merbl2, T Olender2, A Horowitz2, A Yakir2, D Lancet2, E Ben-Asher2 and B Lerer1 1Biological Psychiatry Laboratory, Department of Psychiatry, Hadassah-Hebrew University Medical Center, Jerusalem, Israel and 2Department of Molecular Genetics, Weizmann Institute of Science, Rehovot, Israel

Despite the health hazards, cigarette smoking is disproportionately frequent among young women. A significant contribution of genetic factors to smoking phenotypes is well established. Efforts to identify susceptibility genes do not generally take into account possible interaction with environment, life experience and psychological characteristics. We recruited 501 female Israeli students aged 20–30 years, obtained comprehensive background data and details of cigarette smoking and administered a battery of psychological instruments. Smoking initiators (n = 242) were divided into subgroups with high (n = 127) and low (n = 115) levels of dependence based on their scores on the Fagerstrom Tolerance Questionnaire and genotyped with noninitiators (n = 142) for single nucleotide polymorphisms (SNPs) in 11 nicotinic cholinergic receptor genes. We found nominally significant (P < 0.05) allelic and genotypic association with smoking initiation of SNP rs2072660 and multilocus haplotypes (P < 0.007–0.05) in CHRNB2 and nominal (P < 0.05) allelic or genotypic association of SNPs in CHRNA7 (rs1909884), CHRNA9 (rs4861065) and CHRNB3 (rs9298629) with nicotine dependence. Employing logistic regression and controlling for known risk factors, the best- fitting model for smoking initiation encompassed a 5 SNP haplotype in CHRNB2, neuroticism and novelty seeking (P = 5.9 Â 10À14, Nagelkerke r2 = 0.30). For severity of nicotine dependence, two SNPs in CHRNA7 (rs1909884 and rs883473), one SNP in CHRNA5 (rs680244) and the interaction of a SNP in CHRNA7 (rs2337980) with neuroticism, were included in the model (P = 2.24 Â 10À7, Nagelkerke r2 = 0.40). These findings indicate that background factors, psychological characteristics and genetic variation in nicotinic cholinergic receptors contribute independently or interactively to smoking initiation and to severity of nicotine dependence in young women. Molecular Psychiatry (2006) 11, 312–322. doi:10.1038/sj.mp.4001774; published online 13 December 2005 Keywords: smoking initiation; nicotine dependence; nicotinic cholinergic receptor genes; environment; life experience; personality

Introduction women typically serve in the army for 18–24 months, representing the age group of 18–21 years. Since Despite global trends indicating an overall decline, smoking behavior is considered to be fixed by 25 cigarette smoking is increasing among women in years,4 these figures are alarming. high-income countries and is highest among women Research on smoking behavior has focused on risk of reproductive age.1,2 In the US, nearly 30% of female factors for both smoking initiation (SI) and nicotine high school seniors were smoking in 2000.2 In Israel, dependence (ND). Especially among adolescents, SI 21.9% of Jewish women aged 21–34 years smoked in has been associated with family risk factors, such as 2003.3 The Israel Defense Force (IDF) reports an parental smoking5,6 and family protective factors, increase of 35% in smoking over the last decade such as a strong parent–child relationship.7,8 Psycho- among women at induction and demobilization with logical factors, such as extraversion,9 novelty seek- 39.1% of female soldiers smoking regularly.3 In Israel, ing,10,11 neuroticism12 and external locus of control13 have also been associated with SI. Having two parents Correspondence: Professor B Lerer, Biological Psychiatry Labora- who smoked has also been related to adult ND7,14 tory, Department of Psychiatry, Hadassah-Hebrew University as has harm avoidance.10 High harm avoidance and Medical Center, Ein Karem, Jerusalem 91120, Israel. neuroticism were related to severity of withdrawal E-mail: [email protected] 15 3These authors contributed equally to this work. symptoms from nicotine. Received 5 August 2005; revised 6 October 2005; accepted 11 A substantial body of literature indicates that October 2005; published online 13 December 2005 nicotine is the major component in cigarette smoke Environment, personality, genes and smoking L Greenbaum et al 313 that leads to addiction.16 The development of ND in for the beta4 nicotinic receptor subunit were demon- vulnerable individuals requires exposure to nicotine, strated, but not in mice null for the beta2 .32 A which occurs by voluntary initiation of cigarette role of CHRNB2 in nicotine reinforcement was smoking. Thus, consideration of genetic factors that recently demonstrated by specifically re-expressing influence cigarette smoking should take into account this gene in the ventral-tegmental area of mutated factors that influence the likelihood of smoking the mice.33 A few studies have shown association of first cigarette, the development of a regular smoking nAChR genes with smoking in schizophrenic patients habit and, once smoking behavior is established, the and have raised the possibility of common neuro- degree to which the smoker is dependent on nicotine. biological mechanisms. CHRNA7 was found to be Extensive evidence from twin and adoption studies associated with smoking34 and linkage of CHRNA2 supports a significant contribution of genetic factors and CHRNB2 was reported,35 based on the results of a to SI as well as progression to ND, with heritability of genome scan of schizophrenia patients. CHRNA7 was the order of 50–60%.17–19 shown to be involved in auditory sensory processing The physiological and behavioral effects of nicotine deficits in schizophrenia patients.36,37 Nicotine may are primarily mediated by neuronal nicotinic choli- ameliorate some of these deficits.38 nergic receptors (nAChRs), which modulate the In this study, we examined the association of 11 release of dopamine in the mesolimbic system. Other nAChR genes, including CHRNA4 and CHRNAB2, addictive drugs such as cocaine and amphetamine are with SI and ND in a case–control sample of young thought to act through this central reward path- female students, taking into account the contribution way.20,21 Neuronal nAChRs are ligand gated, penta- of background, life experience and psychological meric ion channels. A number of different subtypes of characteristics of the subjects and the possible nAChR exist, each with individual pharmacological interaction of these variables with nAChR gene and physiological profiles and distinct anatomical variants. distribution in the brain.22 Each nAChR is composed of five subunits arranged in either homomeric or Materials and methods heteromeric complexes of alpha or beta subunits. The different combination of subunits is responsible for Subjects the unique properties of the receptor.23 In all, 12 genes A total of 501 female subjects were recruited from the coding for neuronal nAChR subunits have been Jerusalem area between September 2002 and May cloned: nine neuronal alpha-subunits (alpha2– 2004. Subjects were recruited through advertisements alpha10) and three neuronal beta-subunits (beta2– at institutions of higher education reflecting both beta4). They encode peptides that have a relatively secular and religious environments and various fields hydrophilic extracellular amino terminal portion, of study. Inclusion criteria were age (20–30 years), followed by three hydrophobic trans-membrane doma- enrollment in an institute of higher learning, Jewish ins (M–M3), a large intracellular loop, and then a origin with both parents Ashkenazi or non-Ashkenazi fourth hydrophobic trans-membrane domain (M4).22 and no history of psychiatric treatment of any kind. The two binding sites for acetylcholine (both of which Subjects who were not born and raised in Israel were need to be occupied to cause the channel to open) excluded in order to prevent cultural bias. reside at the interface between the extracellular After screening by telephone, each subject was domain of each of the alpha-subunits and its invited to visit the laboratory where she received an neighbor. Acetylcholine acts on the postsynaptic explanation of the project and was asked to provide receptor to cause a large increase in its permeability signed informed consent as approved by the Helsinki to cations and depolarization of the membrane.24 The Committee (Internal Review Board) of the Hadassah primary brain nAChR subtype with high affinity for Medical Organisation. All consenting subjects com- nicotine contains alpha4 and beta2 subunits.25 pleted a booklet containing psychological and back- Studies to date on the association of nAChRs with ground measures and provided 30–50 ml of fresh cigarette smoking and ND have focused on the genes blood for DNA extraction and lymphoblast transfor- encoding the alpha4 and beta2 subunits. A recent mation. All subjects received monetary compensation study26 reported significant protective effects against for their time and travel expenses. ND in Chinese men of two single nucleotide poly- Subjects were coded as smoking initiators (SI) if morphisms (SNPs) and a multilocus haplotype in the they had smoked at some time during their lifetime CHRNA4 gene. Ethnic- and gender-specificity in the and as noninitiators (NI) if they reported never having association of CHRNA4 allelic variants and haplo- smoked a single cigarette. The current analysis types with ND was also reported recently.27 Mice included only SI who had smoked daily for at least genetically modified to express a mutation in the 1 year (Table 1). Among SI, those receiving a score of 6 alpha4 subunit, which confers higher affinity for or above on the Fagerstrom Tolerance Questionnaire nicotine, were found to have lowered thresholds for (FTQ)39 were coded as high nicotine dependent the induction of ND.28,29 No associations with ND (HND), those receiving scores of 4 or below were were found in four independent studies of CHRNB2 coded as low nicotine dependent (LND). Subjects gene polymorphisms and haplotypes.26,27,30,31 De- scoring 5 were excluded from the current analysis in creased signs of nicotine withdrawal in mice null order to maximize differentiation between the groups.

Molecular Psychiatry Environment, personality, genes and smoking L Greenbaum et al 314 Table 1 Background and life experience variables in SI v. NI subjects and LND vs HND subjects

Variable SI (n = 242) NI (n = 142) Sig. LND (n = 115) HND (n = 127) Sig. Mean/Freq (s.d.)Mean/Freq (s.d.) (P) Mean/Freq (s.d.)Mean/Freq (s.d.) (P)

Agea 24.3 (2.3) 23.2 (2.1) 1.70 Â 10À6 24.2 (2.3) 24.3 (2.3) > 0.05 Education (years)a 13.6 (1.7) 13.7 (1.7) > 0.05 13.7 (1.9) 13.5 (1.6) > 0.05 Ethnicity (Ashkenazi)b 50.4% 48.0% > 0.05 51.3% 49.6% > 0.05 Religious observanceb 7.4% 32.4% 1.67 Â 10À10 7.8% 7.1% > 0.05 Parental smokingb 80.3% 57.6% 1.97 Â 10À6 73.9% 85.8% 0.02 Lifetime experience of traumaa 4.2 (2.1) 3.5 (1.9) 0.004 4.1 (2.1) 4.3 (2.2) > 0.05 Highest lifetime BMIa 24.0 (4.2) 22.9 (4.2) 0.02 23.5 (3.9) 24.4 (4.4) > 0.05 Age at first cigarettea 16.0 (2.4) — — 16.5 (2.5) 15.6 (2.2) 0.006 Percentage of lifetime smokinga,c 0.24(0.10) — — 0.21 (0.10) 0.27(0.09) 1.28 Â 10À6

Significant (P < 0.05) results are shown in bold. SI, smoking initiators; NI, noninitiators; HND, smoking initiators who scored 6 or above on the FTQ; LND, smoking initiators who scored 1-4 on the FTQ. aContinuous variable, compared by t-test. bCategorical variable compared by w2 test. cYears of smoking/age.

Evaluation instruments 105) and three in the non-coding region of exon 7. Out Subjects completed questionnaires covering extensive of the newly idenified SNPs, two were genotyped in background information, smoking behavior including the full study sample: Exon 4, position 5338 and exon the FTQ39 and alcohol and drug consumption. A 7, position 15418. The second exon 7 SNP did not Lifetime Experiences of Trauma questionnaire was yield any call in the genotyping assay and the exon 4 compiled by the research team and covered various coding SNP had too low a frequency (0.5%). types of experiences that may be considered trau- For large-scale genotyping of our sample we matic, before and after the age of 17 years. Seven selected SNPs that fulfilled the following criteria: psychological measures were included: Brief Symp- (1) located within the gene of interest or no more then tom Index (BSI),40 a general measure of psychiatric 20 000 bases upstream or downstream; (2) reported complaints; Eating Attitudes Test (EAT-26),41 a mea- heterozygosity > 0.10. The heterozygosity of the sure of attitudes toward eating and weight; Relation- selected SNPs was checked by genotyping 24 Jewish, ship Questionnaire (RQ),42 a measure of attachment Israelis, unrelated to our subjects. Altogether 53 SNPs security; Perceived Social Support – Friends and fulfilled these conditions. Genotypes were tested for Family (PSS-Fr, PSS-Fa),43 a measure of perceived Hardy–Weinberg equilibrium (HWE) and for differ- social support; Parental Bonding Index (PBI),44 a ences in allele frequency between subjects of Ashke- measure of parent–child relationship, Family Adapt- nazi and non-Ashkenazi origin. SNPs that showed ability and Cohesiveness Evaluation (FACES-II),45 a significant deviation from HWE (n = 8) and/or signi- family perception measure and the Tridimensional ficant differences between Ashkenazi and non-Ash- Personality Questionnaire (TPQ),46 a well-known kenazi subjects (n = 5) or minor allele frequency < 0.05 measure of personality. (n = 1) were excluded from further analysis. A list of the 39 SNPs that were included in the analysis, with SNP selection details of their location and minor allele frequency in The SNPs used in this study were selected from four our sample, is provided in Supplementary Table A. It different databases: dbSNP, Ensembl Genome Brow- should be noted that the loci for CHRNA3, CHRNA5 ser, Celera Genomics and Sequenom RealSNP. We and CHRNB4 overlap, so they were considered as a also resequenced the coding regions of CHRNA4 in single locus in this study. The same was carried out four smokers with the highest and lowest scores on for CHRNA6 and CHRNB3. the FTQ,39 respectively. The CHRNA4 gene consists of seven exons, which were designed for sequencing Genotyping including B50 bases flanking sequences. When SNP genotyping was performed with a high-through- necessary, as for exon 7, three overlapping segments put system of chip-based mass spectrometry (matrix- were designed for PCR of about 500 base pairs each. assisted laser desorption/ionization time-of-flight; Primer design was performed with the program MALDI-TOF) (Sequenom, San Diego, CA, USA). The PRIMER 3; sequences were assembled and compared allele determination in the sampled DNA was based using both the SEQUENCHER and STADEN Pack- on MALDI-TOF mass spectrometry of allele-specific age.47 Four novel SNPs were identified, one in exon 4, primer products.48,49 Genotyping assays were de- causing an amino-acid change (Arg105His at position signed as multiplex reactions using SpectroDE-

Molecular Psychiatry Environment, personality, genes and smoking L Greenbaum et al 315 SIGNER software version 2.0.7 (Sequenom). Primers were chosen as candidates for the logistic regres- were synthesized by Integrated DNA Technologies sion.53 A series of logistic regressions were performed (Coralville, IA, USA). The detailed PCR and primer to identify possible interactions between the psycho- extension reactions were according to the protocol for logical factors and these SNPs for both SI and ND. The high multiplex homogeneous MassEXTEND (hME) 5 SNP haplotype in CHRNB2 which showed a procedure (Sequenom application notes, and de- nominally significant protective effect on SI on scribed in McCullough50). The high-throughput liquid univariate analysis, was examined according to the handling was performed with the aid of a MULTIMEK same procedure outlined for the SNPs. The first step 96 automated 96-channel robot (Beckman Coulter, of the logistic regression for each of the dependent Fullerton, CA, USA). Primer extension products were variables involved entering all the background vari- loaded onto a 384-element chip (SpectroCHIP; Seque- ables, which had been determined as clinically nom) by nanoliter pipetting robot (SpectroPOINT, significant, in the first block. These variables re- Sequenom) and analyzed with a MassARRAY mass mained fixed. A second block of psychological factors spectrometer (Bruker Daltonik, Bremen, Germany). was entered using a forward stepwise procedure The resulting mass spectra were processed and based on likelihood ratio. A third block included analyzed for peak identification and allele determina- those SNPs identified by the univariate analysis and tion with the MassARRAY TYPER version 3.1.4.0 the 5 SNP CHRNB2 haplotype. Once the main effects software (Sequenom). About 10% of the total calls were determined, the possible interactions were were given a low score by the Sequenom caller entered; the main effects of variables that comprised software, and were inspected manually for the correct interactions were retained in the model even if they call. were not significant in themselves. The result was specific models for both SI and ND, which took into Statistical analysis account relevant background variables and delineated Data were analyzed using SPSS 12.0.1. Direct rela- the contributions of psychological factors, SNPs and tionship between SI and ND and the background, life haplotypes and relevant interactions. experience and psychological variables was examined using Student’s t-tests for continuous measures and Pearson’s w2 for categorical data. For the purpose of Results data reduction, a principal components analysis Demographic and life experience variables (PCA) was conducted on six of the seven psycholo- gical measures using the entire sample (n = 501). The In all, 10 background variables were examined in TPQ was not included in the PCA as its components relation to SI and ND (Table 1). SI were slightly but significantly older than NI (P = 1.70 Â 10À6), were less were originally derived from a similar procedure. À10 Promax rotation with Kaiser normalization provided likely to be religiously observant (P = 1.67 Â 10 ), the best solution. Three factors with an eigenvalue > 1 reported more lifetime traumatic events (P = 0.004) were selected. All items loaded for > 0.59 on one and had higher lifetime body mass index (BMI, not factor. Adjusted factor scores were calculated for each including pregnancy, P = 0.02) than NI. Additionally, SI were more likely to have had a parent who smoked subject for each factor. The factors were then inter- À6 preted by assessing the attributes reflected by the (P = 1.97 Â 10 ). No differences were observed bet- variables for which the factor loadings were high. ween SI and NI subjects on education, ethnicity or Linkage disequilibrium among the different mar- military service. SI who exhibited higher levels of ND kers and haplotype block structure were analyzed (HND) were more likely than LND smokers to have with Haploview V.3.12. Haploview was also used to had a parent who had smoked (P = 0.02), to have had detect significant departure from HWE, calculate their first cigarette at an earlier age (P = 0.006) and to have a longer duration of smoking, controlling for age minor allele frequency, determine allele frequency À6 differences between Ashkenazi and non-Ashkenazi (P = 1.28 Â 10 ). subjects and to perform single SNP association tests (SI vs NI, HND vs LND) and haplotype association Single SNPs and haplotypes in nAChR genes tests. Individual haplotype estimations from the Details of the 39 SNPs that we analyzed in 11 population genotype data were assessed using the nicotinic receptor genes are given in Supplementary program, PHASE V.2.51,52 Genotypic association be- Table A. Table 2 shows P-values for comparison of tween individual SNPs in the 11 nAChR genes and SI allele and genotype frequencies in SI compared to and ND was examined by Pearson’s w2 or Fisher’s NI and HND compared to LND smokers. There was exact tests. significant allelic (P = 0.02) and genotypic (P = 0.015) For logistic regression, each SNP was coded for association of SNP rs2072660 in CHRNB2 with genotype according to three categories: homozygosity SI. There was also evidence (P < 0.05) for allelic for the major allele, heterozygosity or homozygosity or genotypic association of SNPs in CHRNA7 for the minor allele. The third category was elimi- (rs1909884), CHRNA9 (rs4861065) and CHRNB3 nated for SNPs where the frequency was < 5%. (rs9298629) with ND. None of these significant values Individual SNPS with a P-value of < 0.25 for their would withstand correction for the 39 SNPs that we univariate relationship with the dependent variables analyzed.

Molecular Psychiatry Environment, personality, genes and smoking L Greenbaum et al 316 Table 2 Association of SNPs in nicotinic, cholinergic receptor genes with SI and ND

Gene Location/Cluster SNP-ID p-SI/NI Allele p-SI/NI-Gen p-HND/LND Allele p-HND/LND-Gen

CHRNA2 8p21.1 rs891398 0.34 0.52 0.96 0.65 27,340,175-27,358,668 bp rs2565065 0.87 0.83 0.27 0.32

CHRNA4 20q13.33 rs2236196 0.15 0.26 0.96 0.87 61,446,465-61,463,192 bp rs6062899 0.36 0.19 0.59 0.47 rs3827020 0.71 0.45 0.43 0.76 rs1044397 0.44 0.20 0.68 0.69 rs1044396 0.54 0.55 0.98 0.73 rs2273502 0.89 0.39 0.65 0.47 rs2273504 0.94 0.53 0.84 0.88 rs6090387 0.67 0.62 0.78 0.76

CHRNA7 15q13.3 rs883473 0.82 0.77 0.42 0.23 30,038,782-30,177,289 bp rs904951 0.29 0.35 0.88 0.24 rs904952 0.20 0.16 0.82 0.73 rs1909884 0.44 0.12 0.20 0.016 rs2337980 0.54 0.47 0.64 0.89

CHRNA9 4p14 rs4861065 0.45 0.85 0.038 0.10 40,253,011-40,272,515 bp rs7669882 0.31 0.54 0.07 0.09 rs4241706 0.80 0.33 0.22 0.14

CHRNA10 11p15.4 rs2741862 0.29 0.16 0.29 0.31 3,651,127-3,656,923 bp rs2741868 0.28 0.35 0.29 0.24 rs4575303 0.26 0.37 0.28 0.20

CHRNB2 1q21.3 rs4845652 0.36 0.22 0.36 0.75 151,353,330-151,362,156 bp rs2280781 0.28 0.21 0.86 0.87 rs4845378 0.16 0.16 0.63 0.63 rs12072348 0.07 0.053 0.93 0.36 rs2072659 0.90 10.00 0.89 0.82 rs2072660 0.020 0.015 0.77 1 rs1127314 0.36 0.52 0.15 0.41 rs3766927 0.22 0.38 0.16 0.40

CHRNA5 15q25.1 rs684513 0.16 0.31 0.55 0.54 76,573,725-76,602,277 bp rs601079 0.31 0.80 0.32 0.15 A5-A3-B4 rs680244 0.68 0.82 0.22 0.22 CHRNA3 rs621849 0.34 0.60 0.48 0.43 CHRNB4 rs1051730 0.68 0.54 0.08 0.17

CHRNB3 8p11.21 rs6474414 0.41 0.14 0.94 0.99 42,569,930-42,609,219 bp rs9298629 0.17 0.34 0.18 0.035 B3-A6 rs2304297 0.88 0.13 0.58 0.88 CHRNA6 rs2217732 0.09 0.20 0.44 0.11 rs1072003 0.37 0.34 0.26 0.053

Nominally significant (P < 0.05) results are shown in bold. P-values were derived from Pearson’s x2 or Fisher’s exact test, as appropriate.

Haplotype analysis was performed for genes where 2) and the two SNPs between the blocks (rs2072659 nominally significant allelic and/or genotypic asso- and rs2072660). Haplotype analysis was performed ciation of SNPs was observed. Analysis of CHRNB2 on the tagging SNPs. This showed a 5 SNP haplotype revealed the most consistent findings. As shown in (CACTA) that exerted a significant protective effect Supplementary Figure A, two haplotype blocks were against SI, showing excess representation in NI identified in CHRNB2. Block 1 encompassed SNPs subjects (frequency SI = 0.13 vs NI = 0.19; P = 0.036). rs4845652, rs2280781, rs4845378 and rs12072348 and Altogether there were 5 haplotype combinations in Block 2, SNPs rs1127314 and rs3766927. The haplo- CHRNB2 that showed nominally significant, protec- type tagging SNPs (htSNPs) identified were tive association with SI (P < 0.007 to 0.05). rs2280781 and rs12072348 (block 1), rs3766927 (block All included the A allele of SNP rs12072348 and the

Molecular Psychiatry Environment, personality, genes and smoking L Greenbaum et al 317 T allele of SNP rs2072660 that individually showed (P = 1.20 Â 10À8) and lower on persistence (P = 0.03) allelic and/or genotypic association with SI. These on the TPQ.46 The only difference on these measures associations do not withstand correction for multiple between HND and LND smokers was on novelty testing. seeking (TPQ)46 where HND smokers tended to score higher (P = 0.05). Psychological measures PCA was performed on six of the seven measures Scores on the seven psychological instruments were (the TPQ was not included, as its factors were derived compared between SI and NI, and between HND and from a similar procedure). A promax rotation pro- LND subjects (Table 3). Compared to NI, subjects who duced three factors, which accounted for 56.4% of the initiated smoking perceived their fathers as less variance. The factor loadings for individual items are caring (PBI,44 paternal warmth; P = 0.0004), reported shown in Table 4. The three factors were interpreted more psychological symptoms (BSI;40 P = 0.01), were as representing three psychological dimensions: more concerned with their eating and weight (EAT- Factor 1 accounted for 33.4% of the variance and 26;41 P = 0.02) and scored higher on novelty seeking was termed ‘family supportiveness.’ Factor 2 ac-

Table 3 Comparison on psychological measures between smoking initiators vs non-initiators, and high nicotine dependent vs low nicotine dependent smokers

Variable SI (n = 242) NI (n = 142) Sig. (P) HND (n = 115) LND (n = 127) Sig. (P) Mean (s.d.) Mean (s.d.) Mean (s.d.) Mean (s.d.)

PSS-Fr 17.1 (3.1) 16.5 (3.6) > 0.05 16.8 (3.4) 17.4 (3.2) > 0.05 PSS-Fam 14.0 (5.9) 14.7 (5.4) > 0.05 14.2 (5.9) 13.7 (5.8) > 0.05 RQ-Secure Attachment 3.1 (1.1) 3.1 (1.2) > 0.05 3.1 (1.1) 3.2 (1.2) > 0.05 PBI – maternal warmth 27.1 (8.1) 27.4 (8.5) > 0.05 26.9 (8.3) 27.4 (7.9) > 0.05 PBI – paternal warmth 24.2 (9.1) 27.2 (7.1) 0.0004 24.2 (9.0) 24.2 (9.2) > 0.05 PBI – maternal protection 11.1 (7.3) 11.0 (7.7) > 0.05 11.4 (7.1) 10.8 (7.6) > 0.05 PBI-paternal protection 9.8 (7.3) 9.45 (7.1) > 0.05 9.7 (7.5) 9.8 (7.2) > 0.05 FACES-II Cohesiveness 35.5 (7.6) 36.5 (7.4) > 0.05 35.2 (7.5) 35.7 (7.7) > 0.05 FACES-II Adaptability 27.1 (6.3) 26.5 (5.8) > 0.05 27.5 (5.5) 26.6 (6.9) > 0.05 TPQ – novelty seeking 18.1 (4.9) 15.1 (5.2) 1.20 Â 10À8 18.7 (5.2) 17.5 (4.4) 0.05 TPQ – harm avoidance 12.8 (6.3) 13.1 (6.1) > 0.05 13.2 (6.7) 12.3 (5.9) > 0.05 TPQ – reward dependence 14.5 (3.2) 14.4 (3.9) > 0.05 14.2 (3.2) 14.8 (3.4) > 0.05 TPQ – persistence 4.9 (2.1) 5.47 (2.9) 0.03 4.9 (2.0) 5.0 (2.2) > 0.05 BSI11 44.2 (26.7) 37.4 (22.9) 0.01 46.0 (24.4) 42.2 (28.9) > 0.05 EAT-2612 7.3 (8.0) 5.6 (5.1) 0.02 7.4 (8.1) 7.2 (7.8) > 0.05

Significant (P < 0.05) results are shown in bold. PSS-Fr, Perceived Social Support – Friends; PSS-Fam, Perceived Social Support – Family; RQ, Relationship Questionnaire; PBI, Parental Bonding Instrument; FACES-I I, Family Adaptability and Cohesiveness Evaluation Scales-II; TPQ, Tridimensional Personality Questionnaire; BSI, Brief Symptom Index; EAT-26, Eating Attitudes Test-26.

Table 4 Factor loadings of psychological dimensions

Variable 1 2 3 Family supportiveness Sociabilityl Neuroticism (33.4%) (12.9%) (10%)

Eating Attitudes Test (EAT-26) 0.09 À0.23 0.81 Brief Symptom Inventory (BSI) À0.10 À0.29 0.59 Perceived Social Support – Friends (PSS-Fr) 0.05 0.85 À0.03 Perceived Social Support – Family (PSS-Fam) 0.70 0.09 À0.20 Relationship Questionnaire (RQ) – Secure Attachment À0.24 0.71 À0.18 Parental Bonding Instrument (PBI) – Maternal Care 0.70 À0.16 À0.26 Parental Bonding Instrument (PBI) – Paternal Care 0.61 0.22 0.27 Parental Bonding Instrument (PBI) – Maternal Protection À0.71 0.25 0.09 Parental Bonding Instrument (PBI) – Paternal Protection À0.66 À0.12 À0.32 Family Adaptation and Cohesiveness Evaluation (Faces-II) 0.69 0.01 À0.17 – Cohesiveness Family Adaptation and Cohesiveness Evaluation (Faces-II) 0.71 À0.00 0.15 – Adaptation

Molecular Psychiatry Environment, personality, genes and smoking L Greenbaum et al 318 counted for 12.9% of the variance and was termed TPQ dimension (novelty seeking) were found to be ‘sociability.’ Factor 3 accounted for 10% of the directly associated with SI. Supplementary Table B1 variance and was termed ‘neuroticism.’ presents the model with background variables only (P = 1.5 Â 10À7, Nagelkerke r2 = 0.20). Supplementary Logistic regression models for SI and ND Table B2 presents the model with both background Table 5 presents the best model predicting SI among and psychological variables (P = 4.7 Â 10À11, Nagelk- the subjects in our study. This was significant erke r2 = 0.29). (P = 5.9 Â 10À14) and explained 30% of the variance The best model predicting HND as compared to (Nagelkerke r2). While controlling for the background LND appears in Table 6. Taking into account the effect variables, a 5 SNP haplotype (CACTA) in CHRNB2, of the background variables, three SNPs (rs1909884 one psychological dimension (neuroticism) and one and rs883473 in CHRNA7, and rs680244 in CHRNA5)

Table 5 Logistic regression predicting smoking initiation

Variable B s.e. P (df) Odds ratio Wald statistic 95% C.I. for Exp(B)

Lower Upper

Religious orientation (nonobservance) 1.30 0.38 0.001 (1) 3.68 11.69 1.74 7.77 Age 0.10 0.06 0.123 (1) 1.10 2.38 0.97 1.24 HBMI 0.02 0.03 0.516(1) 1.02 0.42 0.96 1.09 Parental nonsmoking À0.90 0.29 0.002 (1) 0.41 9.38 0.23 0.72 Lifetime experience of trauma 0.04 0.08 0.565 (1) 1.04 0.33 0.90 1.21 Noncarrier of CHRNB2 CACTA haplotype 0.70 0.31 0.023 (1) 2.02 5.16 1.10 3.70 TPQ novelty seeking 0.12 0.03 0.00002 (1) 1.13 17.83 1.07 1.20 Neuroticism 0.42 0.18 0.022 (1) 1.52 5.25 1.06 2.17 Constant À5.71 1.67 0.001 (1) 11.71 11.71

Significant (P < 0.05) results are shown in bold. Model is significant (P = 5.9 Â 10À14), À2 log likelihood = 337.72, Nagelkerke r2 = 0.30.

Table 6 Logistic regression predicting high nicotine dependence

Variable B s.e. P (df) Odds ratio Wald statistic 95% C.I. for Exp(B)

Lower Upper

Parental smoking 1.52 0.54 0.005 (1) 4.57 8.03 1.60 13.05 Lifetime smoking duration (years of smoking/age) 0.67 0.24 0.004 (1) 1.96 8.18 1.24 3.11 Age at first cigarette -0.19 0.09 0.055 (1) 0.83 3.68 0.68 1.00 CHRNA5 rs680244 0.047 (2) 6.20 CHRNA5 rs680244 (1) 1.74 0.70 0.014 (1) 5.69 6.11 1.43 22.67 CHRNA5 rs680244 (2) 1.26 0.67 0.059 (1) 3.52 3.56 0.95 12.98 CHRNA7 rs883473 0.026 (2) 7.30 CHRNA7 rs883473 (1) 0.32 0.69 0.644 (1) 1.37 0.21 0.36 5.28 CHRNA7 rs883473 (2) 1.39 0.71 0.052 (1) 4.01 3.78 0.99 16.27 CHRNA7 rs1909884 0.016 (2) 8.31 CHRNA7 rs1909884 (1) -2.29 1.01 0.023 (1) 0.10 5.16 0.14 0.73 CHRNA7 rs1909884 (2) -2.85 1.03 0.005 (1) 0.06 7.71 0.01 0.43 Neuroticism -0.31 0.50 0.528 (1) 0.73 0.40 0.28 1.94 CHRNA7 rs2337980 0.250 (2) 2.77 CHRNA7 rs2337980 (1) 0.12 0.54 0.825 (1) 1.13 0.05 0.39 3.22 CHRNA7 rs2337980 (2) 0.72 0.47 0.130 (1) 2.05 2.29 0.81 5.18 Neuroticism  CHRNA7 rs2337980 0.223 (2) 3.00 Neuroticism  rs2337980 (1) 1.10 0.65 0.092 (1) 2.97 2.85 0.84 10.54 Neuroticism  rs 2337980 (2) 0.83 0.59 0.160 (1) 2.29 1.97 0.72 7.32 Constant 1.88 1.97 0.338 (1) 6.60 0.92

Significant (P < 0.05) results are shown in bold. Model is significant (P = 2.24 Â 10À7, À2 log likelihood = 171.52, Nagelkerke r2 = 0.40. (1) Refers to homozygotes for major allele, as compared to homozygotes for minor allele. (2) Refers to heterozygotes for major allele as compared to homozygotes for minor allele.

Molecular Psychiatry Environment, personality, genes and smoking L Greenbaum et al 319 were found to be significantly associated. Although rs1044396 and rs2236196). The CHRNB2 SNP not significant in itself, the interaction of SNP (rs2072660), which was associated with SI in our rs2337980 in CHRNA7 with the psychological dimen- study, was examined previously26,27 but no associa- sion, neuroticism, was included in the model. The tion was observed. The association reported by Feng model was significant (P = 2.24 Â 10À7) and predicted et al.26 of CHRNA4 with ND was in Chinese males; 40% (Nagelkerke r2) of the variance. Supplementary that observed by Li et al.27 showed both ethnic and Table C1 presents the model with background vari- gender specificity and was strongest in African- ables only (P = 2.1 Â 10À5, Nagelkerke r2 = 0.18). Sup- American females. Our sample was made up of plementary Table C2 presents the model with Jewish, Israeli females. We observed nominally sig- background and psychological variables only nificant association of SNPs and haplotypes in (P = 1.3 Â 10À6, Nagelkerke r2 = 0.18). CHRNB2 with SI. Only one30 of the four previous studies that reported negative results for CHRNB2, Discussion examined SI as specifically defined by us. One previous study found evidence for linkage of CHRNB2 Although previous studies have indicated the impor- to smoking in schizophrenia patients.35 The CHRNB2 tance of background, life experience and psycho- SNPs making up the haplotype that we found to be logical factors in SI and ND, little is known about the nominally associated with SI are nonfunctional. relationship between these factors and genes that may rs2072659 and rs2072660 are located in the 30 UTR; confer susceptibility to these two complex pheno- rs2280781 is in the 50 UTR, rs12072348 is intronic, types. Most studies of the potential role of genetic and rs3766927 is in the downstream flanking region polymorphisms have focused on the direct effects of (Supplementary Figure A). The other SNPs for which candidate genes. Noteworthy exceptions in this we observed nominally significant allelic and/or regard are several studies that examined the possible genotypic association are also intronic (rs1909884- interactive relationship of personality features11,54 CHRNA7, rs4861065-CHRNA9) or intergenic or neuroticism,55,56 and polymorphisms in the sero- (rs9298629-CHRNB3). Thus, the most likely explana- tonin transporter and tryptophan hydroxylase1 tion of any association is linkage disequilibrium with genes57 to smoking behavior. Interactive effects of yet to be discovered functional variants. It should be the dopamine D2 receptor gene and depression on emphasized that none of the associations we observed smoking among adolescents have also been repor- would withstand correction for multiple testing and ted.58 It has been suggested59 that personality corre- require independent replication. Furthermore, our lates of smoking ‘mediate the association between sample size, particularly for the HND vs LND genetics and SI.’ comparison limits the power to detect associations The present study applied an integrated approach with a small effect size. that took into account the independent contribution Overall, the finding that background variables alone of background variables and psychological character- predict SI and, to a lesser extent, ND is not surprising. istics, polymorphisms in a series of nicotinic choli- Both parental smoking5,6 and religious orientation60 nergic receptor genes and their interactive effects have previously been cited as related to SI. Among with psychological variables. The results obtained Jewish women, being religious was recently found to should be considered with appropriate reservations be negatively associated with smoking.61 However, considering the sample size. For this reason and given two lesser-known risk factors for SI were identified: the application of multivariate statistical methods, higher lifetime BMI and experience of traumatic replication is important. The most striking observa- events. In a recent study in Israel, smoking was found tion that emerges from the analysis is a strong to be related to obesity in a representative sample of contribution of background and psychological factors IDF personnel aged 20–21 years.62 Increased smoking to SI, accounting for the bulk of the variance among young women may stem from the widely held explained, with a relatively small contribution of notion that smoking controls weight.63 Evidence variation in nicotinic cholinergic receptor genes. The linking traumatic experience and smoking has been contribution of background and psychological char- previously reported. Women with a lifetime history of acteristics to severity of ND is smaller but the assault were found to be 1.8 times more likely to be contribution of genetic factors is larger as is the active smokers.64 Furthermore, studies of two major overall proportion of the variance explained by the traumatic events found increased smoking among best model encompassing background and psycholo- individuals who sought treatment – after the Okla- gical characteristics and genetic variants in nicotinic, homa bombings65 and among residents of Manhattan cholinergic receptor genes. after the 9/11 attacks.66 This link, between smoking In contrast to two previous studies,26,27 we did not and traumatic experience, warrants further explora- find direct association of CHRNA4 with either tion. Interestingly, we have previously reported67 smoking phenotype that we examined (SI or ND). that normal subjects with a history of parental loss All six CHRNA4 SNPs studied by Feng et al.26 were before the age of 16 years had a substantially stronger included in the present study. Three of the six lifetime history of cigarette smoking. In the current CHRNA4 SNPs that yielded significant results in the study, despite the overall relationship between trau- study of Li et al.27 were included (rs2273504, ma and SI, no specific relationship was found

Molecular Psychiatry Environment, personality, genes and smoking L Greenbaum et al 320 between early parental loss and smoking behavior. Acknowledgments Finally, we were unable to explore the relationship of smoking behavior to military service as only a small Research described in this article was supported in proportion (3.8%) of the subjects in our sample did part by Philip Morris USA Inc. and Philip Morris not serve in either military or civilian national International (investigator designed, independently service, and any difference observed between military reviewed grant), the Genome Infrastructure Program service and national service could be accounted for by of the Israeli Ministry of Science and Technology and religious observance. the Crown Center at the Weizmann In addition to the TPQ46 variables (novelty seeking, Institute of Science. The authors thank Naomi harm avoidance, persistence and reward depen- Boumard and Professor Benny Yakir for their helpful dence), three psychological dimensions (family sup- advice. portiveness, sociability and neuroticism) were identified in this study. Family supportiveness was not found to be related to either SI or ND although References previous research has demonstrated a protective 1 World Health Organization, Regional Office for Europe. WHO effect of family support among adolescents against European Country Profiles on Tobacco Control, WHO, Geneva, SI.7,8 Sociability was also not found to be related to 2005. either smoking phenotype although extraversion, a 2 US Department of Health and Human Services. Women and smoking: a report of the Surgeon General. Centers for Disease Control related psychological factor has been previously and Prevention, National Center for Chronic Disease Prevention and 9 found to be associated with SI. A third psychological Health Promotion, Office on Smoking and Health. 2001. dimension, neuroticism, was found to be related to SI 3 Israel Ministry of Health. Report on Smoking in Israel 2003–2004. (directly) and by interaction with genetic variants to Israel Ministry of Health, May 2004. 4 US Department of Health and Human Services. Preventing tobacco ND. This finding is consistent with previous observa- use among young people. US Department of Health and Human tions regarding the relationship between neuroticism Services, Public Health Service, Centers for Disease Control and and smoking, including interaction with genetic Prevention, National Center for Chronic Disease Prevention and variants.12,55–57 Of the TPQ46 variables, novelty seek- Health Promotion, Office on Smoking and Health. 1994. ing was found to be related to SI in our sample, when 5 Barman SK, Pulkkinen L, Kaprio J, Rose RJ. Inattentiveness, parental smoking and adolescent smoking inititiation. Addiction background variables were accounted for. This is 2004; 99: 1049–1061. 10,11 consistent with previous reports. No contribution 6 Tyc VL, Hadley W, Allen D, Ey S, Rai SN, Lensing S. Predictors of of any of the TPQ variables was found in prediction of smoking intentions and smoking status among nonsmoking and HND vs LND. smoking adolescents. Addict Behav 2004; 29: 1143–1147. 7 Harakeh Z, Scholte RH, Vermulst AA, de Vries H, Engels RC. In conclusion, the results of this study support a Parental factors and adolescents’ smoking behavior: an extension significant contribution of background, life experi- of the theory of planned behavior. Prev Med 2004; 39: 951–961. ence and psychological characteristics to SI and to a 8 Tilson EC, McBride CM, Lipkus IM, Catalono RF. Testing the lesser extent to ND. Interaction of psychological interaction between parent-child relationship factor and parent variables with genetic variants in nicotinic choliner- smoking to predict youth smoking. J of Adol Health 2004; 35: 182–189. gic receptor genes was demonstrated. Nominally 9 Wilkinson D, Abraham C. Constructing an integrated model of the significant direct association of SI with CHRNB2 antecedents of adolescent smoking. Br J Health Psychol 2004; and of ND with CHRNA7 and CHRNA5 was observed. 9(Part 3): 315–333. Overall, the contribution of background and environ- 10 Pomerleau CS, Pomerleau OF, Flessland KA, Basson SM. Relation- ship of Tridimensional Personality Questionnaire scores and mental factors and of psychological characteristics to smoking variables in female and male smokers. J Subst Abuse SI emerges as greater than that of genetic variation in 1992; 4: 143–154. nicotinic cholinergic receptor genes. The contribution 11 Kremer I, Bachner-Melman R, Reshef A, Broude L, Nemanov L, of genes to these two complex phenotypes should be Gritsenko I et al. Association of the serotonin transporter gene considered in the context of background, environ- with smoking behavior. Am J Psychiatr 2005; 162: 924–930. 12 Kendler KS, Neale MC, Sullivan P, Corey LA, Gardner CO et al. mental and psychological characteristics and not A population-based twin study in women of smoking initiation alone as is the case in many studies. and nicotine dependence. Psychol Med 1999; 29: 299–308. 13 Kim YH. Korean adolescents’ smoking behavior and its correlation with psychological variables. Addict Behav 2005; 30: 343–350. Electronic database information 14 Kardia SL, Pomerleau CS, Rozek LS, Marks JL. Association of parental smoking history with nicotine dependence, smoking rate, DbSNP: www.ncbi.nlm.nih.gov; Ensemble Genome and psychological cofactors in adult smokers. Addict Behav 2003; Browser: www.ensembl.org; Celera Genomics: 28: 1447–1452. www.celera.com; Sequenom RealSNP: www.realsnp. 15 Madden PA, Madden F, Bucholz KK, Dinwiddie SH, Slutske WS, com; Primer 3 (Rozen S, Skaletsky HJ, 1996, Bierut LJ et al. Nicotine withdrawal in women. Addiction 1997; 92: 889–902. 1997,1998) http://www genome.wi.mit.edu/genome_ 16 Stolerman IP, Jarvis MJ. The scientific case that nicotine is software/other/primer3.html; GeneCodes addictive. Psychopharmacology (Berlin) 1995; 117: 2–10. Corporation: http://www.genecodes.com/; Seque- 17 Sullivan PF, Kendler KS. The genetic epidemiology of smoking. nom, San Diego, California: http://www.sequenom. Nicotine Tobacco Res 1999; 1(Suppl 2): S51–S57; discussion S69–S70. com; Haploview, version 3.12: www.broad.mit.edu/ 18 Li, MD, Cheng R, Ma JZ, Swan GE. A meta-analysis of estimated mpg/haploview; PHASE V2.0: http://www.stat. genetic and environmental effects on smoking behavior in male washington.edu/stephens/ and female adult twins. Addiction 2003; 98: 23–31.

Molecular Psychiatry Environment, personality, genes and smoking L Greenbaum et al 321 19 Niu T, Chen C, Ni J, Wang B, Fang Z, Shao H et al. Nicotine 41 Garner DM, Garfinkel PE. The Eating Attitudes Test: an index for dependence and its familial aggregation in Chinese. Int J the symptoms of anorexia nervosa. Psychol Med 1979; 9: 273–279. Epidemiol 2000; 29: 248–252. 42 Bartholomew K, Horowitz LM. Attachment styles among young 20 Koob GF. Neural mechanisms of drug reinforcement. Ann N Y adults: a test of a four-category model. J Person Social Psychol Acad Sci 1992; 654: 171–191. 1991; 61: 226–244. 21 Rossing MA. Genetic influences on smoking: candidate genes. 43 Procidano ME, Heller K. Measures of perceived social support Environ Health Perspect 1998; 106: 231–238. from friends and from family: three validation studies. Am J Comm 22 Gotti C, Clementi F. Neuronal nicotinic receptors: from structure to Psych 1983; 11: 1–21. pathology. Prog Neurobiol 2004; 74: 363–396. 44 Parker G, Tubling H, Brown LB. A parental bonding instrument. 23 Laviolette SR, van der Kooy D. The neurobiology of nicotine British J Med Psych 1979; 52: 1–10. addiction: bridging the gap from molecules to behaviour. Nat Rev 45 Olson DH, Bell R, Portner J. Family Adaptability and Cohesion Neurosci 2004; 5: 55–65. Evaluation Scales, FACES-II, Family Inventories Manual: Life 24 Rang HP, Dale MM, Ritter JM, Moore PK. Pharmacology, 5th edn Innovations, 1992. Livingstone: Churchill, 2003. 46 Cloninger CR, Przybeck TR, Svrakic DM. Tridimensional Person- 25 Hogg RC, Raggenbass M, Bertrand D. Nicotinic acetylcholine ality Questionnaire: US Normative Data. Psychol Rep 1991; 69: receptors: from structure to brain function. Rev Physiol Biochem 1047–1057. Pharmacol 2003; 147: 1–46. 47 Bonfield JK, Smith KF, Staden R. A new DNA sequence assembly 26 Feng Y, Niu T, Xing H, Xu X, Chen C, Peng S et al. A common program. Nucl Acids Res 1995; 23: 4992–4999. haplotype of the nicotine alpha 4 subunit 48 Little DP, Braun A, Darnhofer-Demar B, Frilling A, Li Y, McIver Jr gene is associated with vulnerability to nicotine addiction in men. RT et al. Detection of RET proto-oncogene codon 634 mutations Am J Hum Genet 2004; 75: 112–121. using mass spectrometry. J Mol Med 1997; 75: 745–750. 27 Li MD, Beuten J, Ma JZ, Payne TJ, Lou XY, Garcia V et al. Ethnic- 49 Little DP, Braun A, Darnhofer-Demar B, Koster H. Identification of and gender-specific association of the nicotinic acetylcholine apolipoprotein E polymorphisms using temperature cycled primer receptor {alpha}4 subunit gene (CHRNA4) with nicotine depen- oligo base extension and mass spectrometry. Eur J Clin Chem Clin dence. Hum Mol Genet 2005; 14: 1211–1219. Biochem 1997; 35: 545–548. 28 Tapper AR, McKinney SL, Nashmi R, Schwarz J, Deshpande P, 50 McCullough RM, Cantor CR, Ding C. High-throughput alternative Labarca C et al. Nicotine activation of alpha4* receptors: sufficient splicing quantification by primer extension and matrix-assisted for reward, tolerance, and sensitization. Science 2004; 306: laser desorption/ionization time-of-flight mass spectrometry. Nucl 1029–1032. Acids Res 2005; 33:99. 29 Hogg RC, Bertrand D. Neuroscience. What genes tell us about 51 Stephens M, Smith NJ, Donnelly P. A new statistical method for nicotine addiction. Science 2004; 306: 983–985. haplotype reconstruction from population data. Am J Hum Genet 30 Silverman MA, Neale MC, Sullivan PF, Harris-Kerr C, Wormley B, 2001; 68: 978–989. Sadek H et al. Haplotypes of four novel single nucleotide 52 Stephens M, Donnelly P. A comparison of Bayesian methods for polymorphisms in the nicotinic acetylcholine receptor beta2- haplotype recontstruction from population genotype data. Am J subunit (CHRNB2) gene show no association with smoking Hum Genet 2003; 73: 1162–1169. initiation or nicotine dependence. Am J Med Genet 2000; 96: 53 Hosmer D, Lemeshow S. Applied Logistic Regression, 2nd edn 646–653. Wiley Interscience: New York, 2000. 31 Lueders KK, Hu S, McHugh L, Myakishev MV, Sirota LA, Hamer 54 Gerra G, Garofano L, Zaimovic A, Moi G, Branchi B, Bussandri M DH. Genetic and functional analysis of single nucleotide poly- et al. Association of the serotonin transporter promoter poly- morphisms in the beta2-neuronal nicotinic acetylcholine receptor morphism with smoking behavior among adolescents. Am J Med gene (CHRNB2). Nicotine Tob Res 2002; 4: 115–125. Genet B Neuropsychiatr Genet 2005; 135: 73–78. 32 Salas R, Pieri F, De Biasi M. Decreased signs of nicotine 55 Hu S, Brody CL, Fisher C, Gunzerath L, Nelson ML, Sabol SZ et al. withdrawal in mice null for the beta4 nicotinic acetylcholine Interaction between the serotonin transporter gene and neuroti- receptor subunit. J Neurosci 2004; 10: 10035–10039. cism in cigarette smoking behavior. Mol Psychiatr 2000; 5: 33 Maskos UBE, Molles S, Pons M, Besson BP, Guiard JP, Guilloux A 181–188. et al. Nicotine receptors reinforcement and cognition restored by 56 Lerman C, Caporaso NE, Audrain J, Main D, Boyd NR, Shields PG. targeted expression of nicotinic. Nature 2005; 436: 103–107. Interacting effects of the serotonin transporter gene and neuroti- 34 De Luca V, Wong AH, Muller DJ, Wong GW, Tyndale RF, Kennedy cism in smoking practices and nicotine dependence. Mol JL. Evidence of association between smoking and alpha7 nicotinic Psychiatr 2000; 5: 189–192. receptor subunit gene in schizophrenia patients. Neuropsycho- 57 Reuter M, Hennig J. Pleiotropic effect of the TPH A779C pharmacology 2004; 29: 1522–1526. polymorphism on nicotine dependence and personality. Am J 35 Faraone SV, Su J, Taylor L, Wilcox M, Van Eerdewegh P, Med Genet B Neuropsychiatr Genet 2005; 134: 20–24. Tsuang MT. A novel permutation testing method implicates 58 Audrain-McGovern J, Lerman C, Wileyto EP, Rodriguez D, Shields sixteen nicotinic acetylcholine receptor genes as risk factors PG. Interacting effects of genetic predisposition and depression for smoking in schizophrenia families. Hum Hered 2004; 57: on adolescent smoking progression. Am J Psychiatr 2004; 161: 59–68. 1224–1230. 36 Freedman R, Leonard S, Gault JM, Hopkins J, Cloninger CR, 59 Munafo M, Johnstone E, Murphy M, Walton R. New directions in Kaufmann CA et al. Linkage disequilibrium for schizophrenia at the genetic mechanisms underlying nicotine addiction. Addict the 15q13–14 locus of the alpha7-nicotinic acetyl- Biol 2001; 6: 109–117. choline receptor subunit gene (CHRNA7). Am J Med Genet 2001; 8: 60 Van den Bree MB, Whitmer MD, Pickworth WB. Predictors of 20–22. smoking development in a population-based sample of adoles- 37 Leonard S, Gault J, Hopkins J, Logel J, Vianzon R, Short M et al. cents: a prospective study. J Adolesc Health 2004; 35: 172–181. Association of promoter variants in the alpha7 nicotinic acet- 61 Baron-Epel O, Haviv-Messika A, Tamir D, Nitzan-Kaluski D, ylcholine receptor subunit gene with an inhibitory deficit found in Green M. Mutiethnic differences in smoking in Israel: pooled schizophrenia. Arch Gen Psychiatr 2002; 59: 1085–1096. analysis from three national surveys. Eur J Public Health 2004; 14: 38 Martin LF, Kem WR, Freedman R. Alpha-7 nicotinic receptor 384–389. agonists: potential new candidates for the treatment of schizo- 62 Zimlichman E, Kochba I, Mimouni FB, Schochat T, Grotto I, Kreiss phrenia. Psychopharmacology 2004; 174: 54–64. Y et al. Smoking habits and obesity in young adults. Addiction 39 Fagerstrom KO. Measuring degree of physical dependence to 2005; 100: 1021–1025. tobacco smoking with reference to individuation of treatment. 63 Delnevo CD, Hrywna M, Abatemarco DJ, Lewis MJ. Relationships Addict Behav 1978; 3: 235–241. between cigarette smoking and weight control in young women. 40 Derogatis LR 1993)). Brief Symptom Inventory (BSI): Administra- Family Community Health 2003; 26: 140–146. tion, Scoring and Procedures Manual, 3rd edn, National Computer 64 Acierno RA, Kilpatrick DG, Resnick HS, Saunders BE, Best CL. Systems: Minneapolis, MN. Violent assault, posttraumatic stress disorder, and depression.

Molecular Psychiatry Environment, personality, genes and smoking L Greenbaum et al 322 Risk factors for cigarette use among adult women. Behav Modif Manhattan, New York, residents after the September 11 terrorists 1996; 20: 363–384. attacks. Am J Epidemiol 2002; 155: 988–996. 65 Pfefferbaum B, Vinekar SS, Trautman RP, Lensgraf SJ, Reddy C, 67 Agid O, Shapira B, Zislin J, Ritsner M, Hanin B, Murad H et al. Patel N et al. The effect of loss and trauma on substance Environment and vulnerability to major psychiatric illness: a case use behavior in individuals seeking support services after control study of early parental loss in major depression, bipolar the 1995 Oklahoma City bombing. Ann Clin Psychiatr 2002; 14: disorder and schizophrenia. Mol Psychiatr 1999; 4: 163–172. 89–95. 68 Gabriel SB, Schaffner SF, Nguyen H, Moore JM, Roy J, Blumenstiel 66 Vlahov D, Galea S, Resnick H, Ahern J, Boscarino JA, Bucuvalas M B et al. The structure of haplotype blocks in the human genome. et al. Increased use of cigarettes, alcohol and marijuana among Science 2002; 296: 2225–2229.

Supplementary Information accompanies the paper on the Molecular Psychiatry website (http://www.nature.com/mp).

Molecular Psychiatry