Association and Linkage of Anxiety-Related Traits with a Functional Polymorphism of the Serotonin Transporter Gene Regulatory Re

Molecular Psychiatry (2000) 5, 216–219  2000 Macmillan Publishers Ltd All rights reserved 1359-4184/00 $15.00 www.nature.com/mp ORIGINAL RESEARCH ARTICLE Association and linkage of anxiety-related traits with a functional polymorphism of the serotonin transporter gene regulatory region in Israeli sibling pairs Y Osher1, D Hamer2 and J Benjamin3

1Beer Sheva Mental Health Center, Faculty of Health Sciences, Ben Gurion University of the Negev, POB 4600, Beer Sheva, Israel; 2Laboratory of Biochemistry, National Cancer Institute, National Institutes of Health, Bethesda, MD 20892, USA; 3Soroka Medical Center, Department of Psychiatry, Ben Gurion University of the Negev, POB 151, Beer Sheva, Israel

Keywords: neuroticism; harm avoidance; anxiety; genetics; to any TPQ traits, which led to the proposal that the serotonin transport effects of this gene might vary across ethnic groups or 11 A functional polymorphism in the regulatory region of cultures. In possible support of this idea, a United the serotonin transporter gene (5-HTTLPR) has been States study12 found a significant interaction between reported to be both associated and linked to anxiety- 5-HTTLPR genotype and race for NEO-PI-R neuroti- related personality measures, although other studies cism, which suggests that the relationship between the have not replicated these findings. The current study allele and anxiety-related traits may be obscured by the examines both association and linkage of the gene to effects of differing phenotypes and differing gene fre- two major anxiety-related personality measures, the quencies in different populations (population harm avoidance scale on the Tridimensional Personality stratification). Several other studies13–15 found no Questionnaire and the neuroticism scale of the NEO-PI- R, in a sample of 148 Israeli subjects comprising 74 association of the 5-HTTLPR with various personality same-sex sibling pairs. We replicated the reported measures. Unfortunately, none of the negative studies association between the short allele and higher scores included family material, which is essential to dis- on the TPQ harm avoidance scale (P = 0.03), including tinguish genuine genetic transmission, or its lack, from the subscale of shyness (P = 0.02), and also found artifacts of population stratification. We therefore association in the same direction between the short examined both association and linkage of the gene to allele and the NEO-PI-R neuroticism subscales of anxi- two major anxiety-related personality measures, the ety (P = 0.03) and depression (P = 0.04). Sib-pair linkage TPQ harm avoidance scale and the NEO neuroticism analysis, using the regression method, further sup- scale, in a sample of 148 healthy Israeli volunteers ported a role of the 5-HTTLPR in anxiety-related person- comprising 74 same-sex sibling pairs. ality traits. Molecular Psychiatry (2000) 5, 216–219. Allele frequencies were 48.9% for the long allele and 51.1% for the short allele. Genotype frequencies were Examination of the possible relationship between a in Hardy–Weinberg equilibrium (24.3% for long/long, functional polymorphism in the promoter region of the 49.3% for long/short, and 26.4% for short/short). serotonin transporter gene (5-HTTLPR) and clinical The association results are summarized in Table 1, anxiety disorders has yielded suggestive but inconclus- which shows the average TPQ and NEO-PI-R person- ive results.1–3 Studies of anxiety-related temperament ality test scores after stratifying the population by and personality dimensions, rather than overt psycho- genotype. Based on prior biochemical and behavioural pathology, have also yielded inconsistent findings. evidence,4,16 the statistical analysis focused on the Lesch et al4 originally reported an association in comparison between individuals with either one or United States subjects between the short form of the two short alleles (the ‘short’ group) as compared to polymorphism and higher scores on the Neuroticism individuals homozygous for the long allele (the ‘long’ scale of Costa and McRae’s NEO-PI-R5 and with higher group). The significance of association was determined estimated scores for Cloninger’s Tridimensional Per- by ANOVA with age, sex and ethnicity entered as sonality Questionnaire6 (TPQ) scale of harm avoidance. covariates. These findings were supported by a study7 of older The short genotype was significantly associated with United States subjects using the actual TPQ instrument higher overall harm avoidance scores (F = 4.61; and by two Japanese studies, one using the TPQ8 and P = 0.03) including its subscale of shyness (F = 5.93; the other using a self-report anxiety and depression P = 0.02). By contrast, between-group differences were questionnaire.9 Another study, in Finnish subjects, not significant for the other three TPQ scales of novelty found linkage but not association of the 5-HTTLPR seeking, reward dependence, and persistence. The with facets of harm avoidance.10 However, a study of short genotype also tended to be associated with higher Israeli subjects found no association of the 5-HTTLPR total neuroticism scores (F = 3.40; P = 0.06), and this 5-HTTLPR and anxiety traits Y Osher et al 217 Table 1 Personality scale scores (means ± standard deviation) by serotonin transporter regulatory region (5-HTTP) genotype

Genotype short/short short/long any short long/long

TPQ n = 39 n = 73 n = 112 n = 36 Novelty seeking 16.18 ± 4.53 14.88 ± 4.71 15.33 ± 4.67 16.44 ± 4.30 Harm avoidance 14.62 ± 6.93 14.37 ± 6.37 14.46 ± 6.54 11.81 ± 6.80** HA 1—Worry 4.00 ± 2.37 3.71 ± 2.33 3.81 ± 2.34 3.22 ± 2.72 HA 2—Fear of uncertainty 4.92 ± 2.16 4.62 ± 1.82 4.72 ± 1.94 4.08 ± 1.84 HA 3—Shyness 3.03 ± 1.97 2.92 ± 1.88 2.96 ± 1.90 2.11 ± 1.70** HA 4—Fatigability 2.67 ± 2.66 3.12 ± 2.33 2.96 ± 2.45 2.39 ± 2.42 Reward dependence 13.82 ± 3.88 13.52 ± 3.64 13.63 ± 3.71 14.36 ± 3.57 Persistence 4.92 ± 2.25 4.75 ± 2.01 4.81 ± 2.09 5.36 ± 2.10

NEO-PI-R n = 39 n = 70 n = 109 n = 35 Neuroticism 91.5 ± 25.6 92.8 ± 23.8 92.3 ± 24.3 85.1 ± 24.6* N 1—Anxiety 17.3 ± 6.1 17.6 ± 6.1 17.5 ± 6.1 15.1 ± 6.6** N 2—Angry hostility 14.5 ± 5.4 14.4 ± 5.1 14.4 ± 5.2 13.6 ± 5.4 N 3—Depression 14.2 ± 5.9 16.0 ± 5.5 15.4 ± 5.7 13.3 ± 6.2** N 4—Self-consciousness 16.8 ± 5.6 16.0 ± 4.5 16.2 ± 5.0 15.0 ± 4.9 N 5—Impulsiveness 17.7 ± 4.1 17.0 ± 4.4 17.3 ± 4.3 16.8 ± 5.3 N 6—Vulnerability 11.2 ± 5.6 11.6 ± 4.9 11.5 ± 5.1 10.9 ± 5.2 Extroversion 114.7 ± 24.6 111.0 ± 22.9 112.3 ± 23.5 120.3 ± 24.2 Openness 115.8 ± 18.1 114.2 ± 17.5 114.7 ± 17.6 111.7 ± 19.1 Agreeableness 114.9 ± 16.1 112.8 ± 17.2 113.5 ± 16.8 111.6 ± 19.8 Conscientiousness 125.8 ± 21.7 122.9 ± 20.8 123.9 ± 21.1 128.6 ± 17.1

Any short vs long/long (df = 1): **P Ͻ 0.05 after correcting for sex, age, and ethnic group as covariates. *P Ͻ 0.07 after correcting for sex, age, and ethnic group as covariates.

was significant for the two neuroticism subscales of Table 2 Linkage results anxiety (F = 4.86; P = 0.03) and depression (F = 4.15; = P 0.04). There was no significant association with any B3 P other NEO-PI-R scale. In agreement with previous results, the short allele had more of a dominant than TPQ recessive effect (based on analyses of s/s vs s/l vs l/l, Novelty seeking −0.041 ns not shown; see Table 1). Harm avoidance 0.289 0.07 Since population-based association studies are sub- HA1 0.004 ns ject to false positive findings due to population strati- HA2 0.281 0.06 fication, within-family analyses are also important. HA3 0.327 0.04 HA4 0.246 0.09 Analysis of discordant sib-pairs revealed that sibs with Reward dependence −0.010 ns the short allele tended to have slightly higher harm Persistence −0.285 ns avoidance scores than did sibs homozygous for the long allele, but the difference was not significant NEO (HA = 14.55 ± 4.67 for short vs HA = 13.27 ± 5.95 for Neuroticism 0.292 0.0652 long, n = 22 pairs; paired t = 0.838, df = 21, P = 0.41). N1 0.114 ns Neuroticism scores in the discordant sib-pairs also did N2 0.122 ns not differ significantly (N = 93.10 ± 18.67 for short vs N3 0.127 ns − N = 92.86 ± 22.35 for long, n = 21 pairs; paired t = 0.41, N4 0.294 ns df = 20, P = 0.97). N5 0.531 0.0007 N6 0.211 ns Linkage is an alternative method to assess the Extroversion 0.020 ns relationship between a gene and a trait within families. Openness −0.221 ns Sib-pair linkage analysis by the regression method of Agreeableness 0.028 ns 17 Cardon et al gave the results shown in Table 2. There Conscientiousness −0.041 ns was suggestive evidence for linkage with total HA = = scores (P 0.07), with the HA shyness subscale B3 regression coefficient for linkage. = = (P 0.04), and with two other HA subscales (fear: P one-sided significance level of B3. P = 0.06 and fatigue: P = 0.09). This finding is some- ns = P Ͼ 0.1.

Molecular Psychiatry 5-HTTLPR and anxiety traits Y Osher et al 218 what consistent with the report of Mazzanti et al,10 Methods who found linkage to two of the Harm Avoidance sub- Sample scales (fear and anticipatory worry) but not to the HA Participants were volunteers recruited by poster and scale as a whole. Similarly, overall N scores show a media advertisements. Preliminary screening excluded trend towards linkage (P Ͻ 0.07) which was significant Ͻ major mental illness or psychiatric disability, and for the N subscale of impulsiveness (P 0.0007). ascertained a literacy level adequate to allow compre- This study replicates the reported relationship hension of questionnaire items. Subjects were 148 Isra- between the short allele of 5-HTTLPR and several anxi- elis, mean age 30.7 (SD 11.6 years), range 17–78 years ety-related personality measures, both by association old. The majority of subjects were from in and around analysis and, at the trend level, by linkage analysis. the Beer Sheva area, and both university and nursing The association data gave estimated 5-HTTLPR-specific school students, along with members of the hospital 2 Х heritabilities in the range of h 0.02 (for N) to 0.04 staff, were somewhat over-represented. Subjects were (for HA3), consistent with the limited effect of a single almost evenly split between Ashkenazi and non-Ash- gene on complex personality traits. The estimated heri- kenazi ethnic groups (49.4% vs 50.6%), and women tabilities from the linkage data were similar or greater outnumbered men by approximately two to one (65.7% (h2 Х 0.14 for N and HA), suggesting that the observed vs 34.3%). association is due to genuine transmission rather than population stratification. However, because of the Measures much lower power of linkage compared to association, Participants completed two questionnaires: a Hebrew far larger sib samples would be required to accurately version19 of Cloninger’s Tridimensional Personality partition the within-family and across-family vari- Questionnaire, which measures four temperament ances. traits with hypothesized mediation by the following No association or linkage of the 5-HTTLPR was neurotransmitters: Novelty Seeking—dopamine, Harm found to any of the other TPQ or NEO-PI-R scales in Avoidance—serotonin, Reward Dependence and Per- these two instruments, which argues for the specificity sistence—noradrenaline;6 and a validated Hebrew ver- of the relationship between this gene and anxiety- sion20,21 of Costa and McCrae’s NEO-PI-R, which meas- related measures. While some studies have failed to ures the so-called ‘big five’ personality factors of find association between the 5-HTTLPR and measures Neuroticism, Extroversion, Openness to new experi- of anxiety, it should be noted that none of these have ence, Agreeableness, and Conscientiousness.5 reported significant overall associations in the ‘wrong’ direction, and other evidence suggests that the effects Procedure may be seen only through interactions with other poly- All subjects gave informed consent, and the protocol morphisms.18 It is difficult to assess the probability of was approved by the Ben Gurion University Helsinki false negatives in such studies since experimental and Committee. Volunteers filled out the questionnaires measurement errors in genotyping or phenotyping, and donated 20 cc of blood by venipuncture. The which will increase type II error, are not readily esti- majority of subjects were interviewed at an office in the mated. We believe that, overall, the accumulated evi- medical center, although in some cases the interviewer dence supports the hypothesis of a real relationship went to the subjects’ homes. Test materials and bloods between the short allele of the 5-HTTLPR gene and were coded by number to ensure confidentiality. anxiety-related personality traits. Cross-cultural studies are an important tool in disen- Genotyping tangling genetic and environmental influences on per- DNA was extracted and genotypes determined as pre- 11 sonality. The present finding of an association of the viously described. 5-HTTLPR with anxiety-related traits in Israelis sug- gests that the effects of the serotonin transporter gene Statistical analyses are not limited to one particular ethnic group or cul- Individuals with two short alleles were grouped ture. However, a potentially interesting difference together with individuals with one short and one long between this study and Lesch et al’s analysis of United allele, and this group (short) compared to individuals States subjects is that the Israelis did not display having two long alleles (long) (as suggested in Lesch et 4 association of the 5-HTTLPR with the NEO-PI-R trait of al ). Between-group differences for mean scale scores agreeableness, which is a measure of social affiliation, were tested with ANOVA (SPSS), with age, sex, and ethnic group as covariates and using a significance cooperativeness and affability. The Israeli population Ͻ also failed to show the usual negative phenotypic cor- level of P 0.05. 5-HTTLPR-specific heritabilities were estimated from association data by h2 = r 2, where r relation between neuroticism and agreeableness (data GP GP is the Pearson correlation between dichotomized geno- not shown). These observations suggest that the type and phenotype. Genetically discordant sib-pairs sources of variance in socially agreeable personality were examined using a paired t-test.4 Linkage analysis traits may be different in Israelis and Americans. It was conducted by the regression method of Cardon et would be interesting to test this hypothesis in a al17 for unselected samples according to the equation larger sample. = + + ␲ + ␲ C B0 B1P B2 B3P , where C is the phenotype of

Molecular Psychiatry 5-HTTLPR and anxiety traits Y Osher et al 219 the first sib, P is the phenotype of the second sib, and K. Anxiety traits associated with a polymorphism in the serotonin ␲ is genotype sharing at the 5-HTTLPR (␲ = 0 for transporter gene regulatory region in the Japanese. J Hum Gen 1999; ␲ = 44: 15–17. short/long and long/short pairs, 1 for short/short 10 Mazzanti CM, Lappalainen J, Long JC, Bengel D, Naukkarinen H, and long/long pairs). The test for linkage is provided Eggert M et al. Role of the serotonin transporter promoter polymor- by B3. Simulation experiments show that when the fre- phism in anxiety-related traits. Arch Gen Psychiatry 1998; 55: quency of long and short alleles is approximately 936–940. 11 Ebstein RP, Gritsenko I, Nemanov L, Frisch A, Osher Y, Belmaker equal, as is the case in this study, the specific herita- RH. No association between the serotonin transporter gene regulat- bility due to the 5-HTTLPR can be estimated from link- ory region polymorphism and the tridimensional personality ques- 2 Х age data by h B3/2. tionnaire (TPQ) temperament of harm avoidance. Mol Psychiatry 1997; 2: 224–226. 12 Gelernter J, Kranzler H, Coccaro EF, Siever LJ, New AS. Serotonin Acknowledgements transporter protein gene polymorphism and personality measures in African American and European American subjects. Am J Psy- This research was supported by grant No. 95–98 of the chiatry 1998; 155: 1332–1338. 13 Jorm AF, Henderson AS, Jacomb PA, Christensen H, Korten AE, United States-Israel Binational Science Foundation Rodgers B et al. An association study of a functional polymorphism (BSF), Jerusalem, Israel (JB and DH). of the serotonin transporter gene with personality and psychiatric symptoms. Mol Psychiatry 1998; 3: 449–451. 14 Flory JD, Manuck SB, Ferrell RE, Dent KM, Peters DG, Muldoom References MF. Neuroticism is not associated with the serotonin transporter (5-HTTLPR) polymorphism. Mol Psychiatry 1999; 4: 93–96. 1 Ohara K, Nagai M, Suzuki Y, Ochiai M, Ohara K. Association 15 Ball D, Hill L, Freeman B, Eley TC, Strelau J, Riemann R et al. The between anxiety disorders and a functional polymorphism in the serotonin transporter gene and peer-rated neuroticism. Neuroreport serotonin transporter gene. Psychiatry Res 1998; 81: 277–279. 1997; 8: 1301–1304. 2 Ishiguro H, Arinami T, Yamada K, Otsuka Y, Toru M, Shibuya H. 16 Greenberg BD, Tolliver TJ, Huang SJ, Li Q, Bengel D, Murphy DL. An association study between a transcriptional polymorphism in Genetic variation in the serotonin transporter promotion region the serotonin transporter gene and panic disorder in a Japanese affects serotonin uptake in human blood platelets. Am J Med Genet population. Psychiatry Clin Neurosci 1997; 51: 333–335. 1999; 88: 83–87. 3 Deckert J, Catalano M, Heils A, Di Bella D, Friess F, Politi E et al. 17 Cardon LR, Smith SD, Fulker DW, Kimberling WJ, Pennington BF, Functional promoter polymorphism of the human serotonin trans- DeFries JC. Quantitative trait locus for reading-disability on chro- porter: lack of association with panic disorder. Psychiatr Genet mosome-6. Science 1994; 266: 276–279. 1997; 7: 45–47. 18 Benjamin J, Osher Y, Kotler M, Gritsenko I, Nemanov L, Belmaker 4 Lesch KP, Bengel D, Heils A, Sabol SZ, Greenberg BD, Petri S et RH et al. Association between tridimensional personality question- al. Association of anxiety-related traits with a polymorphism in naire (TPQ) traits and three functional polymorphisms: dopamine the serotonin transporter gene regulatory region. Science 1996; 274: receptor D4 (DRD4), serotonin transporter promoter region (5- 1527–1531. HTTLPR) and catechol O-methyltransferase (COMT). Mol Psy- 5 Costa PTJ, McCrae RR. Revised NEO Personality Inventory (NEO- chiatry 2000; 5: 96–100. PI-R) and NEO Five Inventory (NEO-FFI) Professional Manual. 19 Ebstein RP, Novick O, Umansky R, Priel B, Osher Y, Blaine D et Psychological Assessment Resources: Florida, 1992. al. Dopamine D4 receptor (D4DR) exon III polymorphism associa- 6 Cloninger CR, Przybeck T, Svrakic D. The Temperament and ted with the human personality trait of Novelty Seeking. Nat Genet Character Inventory: A Guide to its Development and Use. Center 1996; 12: 78–80. for Psychobiology of Personality, Washington University: St Louis 20 Schwimmer M, Montag I. Sensation seeking and the five-factor 1994, pp 1–184. model. Man and Work 1996; 7: 22–45. In Hebrew; English abstract 7 Ricketts MH, Hamer RM, Sage JI, Manowitz P, Feng F, Menza MA. 70–71. Association of a serotonin transporter gene promoter polymor- 21 McCrae R, Costa P. Personality trait structure as a human universal. phism with harm avoidance behaviour in an elderly population. American Psychologist 1997; 52: 509–516. Psychiatr Genet 1998; 8: 41–44. 8 Katsuragi S, Kunugi H, Sano A, Tsutsumi T, Isogawa K, Nanko S Correspondence: J Benjamin, Soroka Medical Center, Dept of Psy- et al. Association between the serotonin transporter gene polymor- chiatry, POB 151, Beer Sheva 84101, Israel. E-mail: yonatanȰbgu- phism and anxiety-related traits. Biol Psychiatry 1999; 45: 368–370. mail.bgu.ac.il 9 Murakami F, Shimomura T, Kotani K, Ikawa S, Nanba E, Adachi Received 1 June 1999; revised and accepted 6 August 1999

Molecular Psychiatry