Role of genetic variants in the CHRNA5–CHRNA3–CHRNB4 cluster in nicotine dependence risk: importance of gene–environment interplay

The CHRNA5CHRNA3CHRNB4 locus on chromosome 15q25.1 contains three genes encoding nicotinic acetylcholine receptor (nAChR) subunits, which lie very close to, and in strong linkage disequilibrium with, each other. A plethora of recent evidence suggests an association of this gene cluster with several nicotine dependence (ND) phenotypes. Saccone et al.1 tested 348 candidate genes in a mixed gender sample of 1050 cases of European ancestry with scores 4 on the Fagerstrom test for nicotine dependence (FTND) and 879 controls who had smoked at least 100 cigarettes in their lifetimes but had an FTND of 0 during the heaviest period of smoking. Seven single-nucleotide polymorphisms (SNPs) (see Supplementary Table 1) located within the CHRNA5CHRNA3CHRNB4 cluster were associated with ND (P-values 0.01–0.0003). Biologically, the most plausible evidence for a risk variant was for the non-synonymous SNP, rs16969968, in the CHRNA5 gene (P=0.00064), and for rs578776 in the 3′-UTR of CHRNA3 (P=0.0003). A recent paper from the same group2 validated and refined the finding in a large European ancestry sample. Associations with habitual smoking (defined as 20 cigarettes per day for 6 months or more) vs <10 cigarettes per day were found, for rs16969968 (P=0.0007) and rs578776 (P=0.0009). Although the value of D′ between the two SNPs was high (D′=1), the correlation was low (r2<0.15), rendering them statistically independent. As expected, other SNPs that were highly correlated (r2>0.79) with rs16969968 were found to be associated with habitual smoking.2 Functional studies showed that the rs16969968 polymorphism, which causes a change from aspartic acid to asparagine, affects maximal receptor response to nicotine agonist.2

Association of rs1051730 in the CHRNA3 gene and other highly correlated SNPs with smoking quantity was shown in a genome-wide association study in 10 995 Icelanders (P=5 × 10−16).3 When rs1051730 was tested in a subsample of ND individuals (defined as FTND4, or fulfillment of at least three out of seven Diagnostic and Statistical Manual for Mental Disorders-IV criteria) vs low-quantity smokers, a measure of ND resembling that reported by Saccone et al.,1 the association was significant (P=7 × 10−15). Association of smoking quantity with other SNPs in the region was reported by Berretini et al.4 in a genome-wide association study of three European samples containing 15 000 individuals (Supplementary Table 1). Furthermore, recent reports show an association of the CHRNA5CHRNA3CHRNB4 cluster with other ND-related conditions, such as alcohol dependence,5, 6 lung cancer and peripheral artery disease.3, 7, 8

Although genetic factors play a significant etiological role in ND-related behavior, other factors, such as family background, life experience, personality traits and cognitive factors, are likely to be implicated directly and/or interactively.9, 10 None of the papers that the reported association of CHRNA5CHRNA3CHRNB4 SNPs with ND took such factors into account. For several years, our group has been studying the interaction between genetic factors contributing to ND phenotypes, and psychological, environmental and cognitive factors. Our sample includes 501 Jewish, Israeli female undergraduate students (aged 20–30 years at the time of recruitment). Coding smokers with a score 6 on the Fagerstrom Tolerance Questionnaire as high ND and smokers with a score 4 as low ND, we performed a candidate gene study of 11 nAChR genes. On univariate analysis, we found a trend for the association of SNP rs1051730 in CHRNA5 with the severity of ND.10 Our best logistic regression model (P=2.24 × 10−7; Nagelkerke pseudo R2=0.40) predicting high ND included the effect of parental smoking, lifetime duration of smoking, age at first cigarette and neuroticism, as well as four nAChR SNPs, including rs680244, in CHRNA5.10

In view of recent findings regarding the association of SNPs in the CHRNA5CHRNA3CHRNB4 cluster with ND and related phenotypes, we extended our earlier study of SNPs in this region. We genotyped 19 SNPs in 90 cases (FTND 4) and 108 controls (smoked at least 100 cigarettes in their lifetimes, but FTND 0 during the heaviest period of smoking) defined according to the phenotype of Saccone et al.1 Genotyping was performed at the Harvard Medical School Partners Healthcare Center for Genetics and Genomics with the Sequenom MassARRAY system using iPLEX chemistry. For quality control, 20% of samples were genotyped twice on all assays; the match rate was 99%. Phenotypic data, including background, life experience, personality and cognitive variables, were available on these subjects, as described in our earlier reports.10, 11 Data were analyzed using Haploview (version 4.0), PLINK (version 0.991) and SPSS (version 15). No significant deviation from Hardy–Weinberg equilibrium was detected. An initial logistic regression model with SNP genotype as the independent measure and controlling for age and ethnic origin (Ashkenazi/non-Ashkenazi) showed a statistically significant association between ND and the CHRNA3 3′-UTR SNP, rs578776 (P=0.0006), which survived Bonferroni correction for multiple testing. Two SNPs highly correlated with rs578776 (rs3743078 and rs6495308; r2>0.78) and one moderately correlated SNP (rs637137; r2=0.69) were nominally associated with ND (Supplementary Table 1). Haploview was used to perform haplotype association analysis, but significance levels were not stronger than those of individual SNPs. On the other hand, the regression model showed only a suggestive trend for association with ND of the second highly intercorrelated (r2>0.97) three-SNP group: rs16969968, rs1051730 and rs951266 (P=0.05–0.063).12 As the minor allele frequencies of these SNPs in our sample are similar to the minor allele frequencies reported by Saccone et al.1 and Bierut et al.2 in their European ancestry population, it is plausible that small sample size limited our power to detect the association of this second SNP group with ND. Similar to the findings of Bierut et al.,2 the absolute value of D′ between all the above-mentioned SNPs is high (D′>0.95), but the correlation between the two distinct SNP groups (marked by rs578776 and by rs16969968/rs1051730) is low (r2=0.2). We did not detect any significant association with ND of other SNPs, reported earlier to be associated with ND-related phenotypes. These include rs8034191 (located within gene, upstream of CHRNA5), which was earlier associated with both ND3 and lung cancer.7, 8 As we studied females only (as in our original study design10), our replication is limited to that gender. However, no gender effect was found by Saccone et al.1 in their mixed sample with regard to any of the SNPs in this gene cluster.

We extended the logistic regression model to include additional variables associated with severity of ND in our univariate analyses (Supplementary Table 2), the following variables serving as predictors: (1) background variables: religious observance, parental smoking and life-time experience of trauma; and (2) cognitive variables: performance speed and impulsivity factors, which are composite variables produced by confirmatory factor analysis. The analysis indicated that all the single-test reaction time measures are loaded on one factor (performance speed) and most of the single-test false-alarm measures loaded on a second factor (impulsivity). The score for each factor was calculated by regression method score: (3) personality variables: TPQ-novelty-seeking and reward dependence scales; and (4) genetic variables: SNP rs578776 (the two SNPs that were nominally significant on univariate analysis were omitted due to multicolinearity and lower association with the phenotype); percentage of lifetime smoking was forced into the model as a controlling variable. Three variables remained in the model (Table 1): parental smoking (P=0.01; odds ratio=3.21), lifetime experience of trauma (P=0.006; odds ratio=2.77) and SNP rs578776. Carrying one minor allele of rs578776 lowered the probability of being in the ND group by 3.44 (P=0.002) and carrying two minor alleles lowered the probability by 8.33 (P=0.02). No other direct or interactive effects remained in the model. Nagelkerke pseudo R2 of the model was 0.25 (P<0.0001) and it correctly classifies 70% of the cases.

Table 1 Logistic regression model for predicting nicotine dependence

Our findings replicate the association of the CHRNA5CHRNA3CHRNB4 cluster with ND using the phenotype of Saccone et al.,1 support the notion that at least two independent genetic risk factors for ND are found in this gene cluster and show the importance of an integrated approach that takes into account background, personality, life experience and cognitive factors in addition to genetic variables.


  1. 1

    Saccone SF, Hinrichs AL, Saccone NL, Chase GA, Konvicka K, Madden PA et al. Hum Mol Genet 2007; 16: 36–49.

  2. 2

    Bierut LJ, Stitzel JA, Wang JC, Hinrichs AL, Grucza RA, Xuei X et al. Am J Psychiatry 2008; 165: 1163–1171.

  3. 3

    Thorgeirsson TE, Geller F, Sulem P, Rafnar T, Wiste A, Magnusson KP et al. Nature 2008; 452: 638–642.

  4. 4

    Berrettini W, Yuan X, Tozzi F, Song K, Francks C, Chilcoat H et al. Mol Psychiatry 2008; 13: 368–373.

  5. 5

    Schlaepfer IR, Hoft NR, Collins AC, Corley RP, Hewitt JK, Hopfer CJ et al. Biol Psychiatry 2008; 63: 1039–1046.

  6. 6

    Wang JC, Grucza R, Cruchaga C, Hinrichs AL, Bertelsen S, Budde JP et al. Mol Psychiatry 2008 (e-pub ahead of print, doi: 10.1038/mp.2008.42).

  7. 7

    Hung RJ, McKay JD, Gaborieau V, Boffetta P, Hashibe M, Zaridze D et al. Nature 2008; 452: 633–637.

  8. 8

    Amos CI, Wu X, Broderick P, Gorlov IP, Gu J, Eisen T et al. Psychol Med 1999; 29: 299–308.

  9. 9

    Kendler KS, Neale MC, Sullivan P, Corey LA, Gardner CO, Prescott CA . Psychol Med 1999; 29: 299–308.

  10. 10

    Greenbaum L, Kanyas K, Karni O, Merbl Y, Olender T, Horowitz A et al. Mol Psychiatry 2006; 11: 312–322.

  11. 11

    Yakir A, Rigbi A, Kanyas K, Pollak Y, Kahana G, Karni O et al. Eur Neuropsychopharmacol 2007; 17: 339–351.

  12. 12

    Weiss RB, Baker TB, Cannon DS, von Niederhausern A, Dunn DM, Matsunami N et al. PLoS Genet 2008; 4: e1000125.

Download references

Author information



Corresponding author

Correspondence to B Lerer.

Additional information

Supplementary Information accompanies the paper on the Molecular Psychiatry website (

Supplementary information

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Greenbaum, L., Rigbi, A., Teltsh, O. et al. Role of genetic variants in the CHRNA5–CHRNA3–CHRNB4 cluster in nicotine dependence risk: importance of gene–environment interplay. Mol Psychiatry 14, 828–830 (2009).

Download citation

Further reading