Introduction

Despite increasing public awareness of the health risks of using tobacco products, approximately 1.2 billion people worldwide smoke tobacco daily. It is estimated that every year, 4.2 million people die from tobacco-related diseases: this number is predicted to approach 10 million by the year 2020 (WHO 2002). In the USA, approximately 70.3 million Americans aged 12 or older use tobacco products (SAMHSA 2001), and 20.8% of adults aged 18 and up smoke (CDC 2007). Thus, tobacco is one of the most widely abused substances, and it kills more than 438,000 US citizens each year (CDC 2005). Economically, smoking is responsible for about 7% of total US health care costs, an estimated $157.7 billion each year, of which $75.8 billion is direct medical costs.

Over the last decade, many large-sample twin studies in the US and other countries have demonstrated that genetic factors contribute to the risk of becoming a regular smoker (Carmelli et al. 1992; Heath et al. 1993; Heath and Martin 1993; Kendler et al. 1999; Madden et al. 1999; Maes et al. 1999; Swan et al. 1996, 1990a, b; True et al. 1999; Vink et al. 2005). Initial evidence for a genetic influence on nicotine dependence (ND) came from cross-sectional twin studies that showed a mean heritability of 0.53 (range 0.28–0.84) for cigarette smoking (Hughes 1986). Recently, we conducted a meta-analysis of genetic parameter estimates for ND based on 17 twin studies and determined that the weighted mean heritability for ND is 0.59 in male and 0.46 in female smokers, with an average of 0.56 for all smokers (Li et al. 2003a). Complex segregation analyses of smoking behavior in 493 three-generation families support a dominant major gene effect with residual familial correlation (Cheng et al. 2000). Together, these findings strongly suggest that ND, like many other physical and behavioral human disorders, is a complex trait (or disorder) that involves multiple genes and environmental risk factors, as well as interactions between genes or between genes and the environment. Further, the proportion of genetic versus environmental influences on different smoking stages differs by sex (Madden et al. 1999; Morley et al. 2006; Vink et al. 2006), with genetic factors appearing to have a larger role in initiation than in persistence in women, whereas the opposite is observed in men (Li et al. 2003a).

To identify susceptibility loci for ND, significant recent efforts have been made using an approach that tests for linkage of the disorder to polymorphic markers across the entire genome. We are aware of more than 20 published genome-wide linkage studies for smoking behavior (Bergen et al. 1999, 2003; Bierut et al. 2004; Duggirala et al. 1999; Ehlers and Wilhelmsen 2006, 2007; Gelernter et al. 2004, 2007; Goode et al. 2003; Li et al. 2003b, 2007a, 2006; Loukola et al. 2007; Morley et al. 2006; Pomerleau et al. 2007; Saccone et al. 2003, 2007; Swan et al. 2006; Vink et al. 2004, 2006; Wang et al. 2005). However, only a limited number of putative genomic linkages have been replicated in independent studies. A significant limiting factor in replicating these linkages is genetic heterogeneity, especially when the sample size is relatively small or participants of various ethnic origins are included. In addition, the size of the genetic effect, the density of markers, the definition and assessment of the phenotypes, and the statistical approaches might contribute to difficulty in replicating the findings of genome-wide linkage scans. Despite these concerns and limitations, significant progress has been made, particularly in the last few years. The primary purpose of this review is to provide an update on the progress made in identifying susceptibility loci for ND since our last review (Li et al. 2004).

Populations used in the reported linkage studies

Table 1 provides a detailed description of the populations employed in previous linkage studies for all ND-related measures. Collectively, these investigations have relied on 13 populations: the Collaborative Studies on the Genetics of Alcoholism (COGA), the Framingham Heart Study (FHS), the Mid-South Tobacco Family (MSTF) study, the Nicotine Addiction Genetics (NAG) project, the Finnish Twin Families (FTF), the Mission Indians in Southwest California, the Genetic Epidemiology Network of Arteriopathy (GENOA) study, the Smoking in Families Study (SMOFAM), the Netherlands Twin Register (NTR) study, the Genetics of Cocaine or Opioid Dependence (GCOD) study, the Christchurch sample of New Zealand, the Australian Twin Registry (ATR), and the Family Study of Panic Disorder (FSPD). Except for the MSTF (Li et al. 2006, 2007b), GCOD (Gelernter et al. 2007), and GEONA (Pomerleau et al. 2007) studies, which include substantial numbers of African-Americans (AA), and a Native American Mission Indian cohort (Ehlers and Wilhelmsen 2006, 2007), the participants of the other studies are of primarily European or European-American (EA) origin. The methods of assessing ND differ greatly from study to study, and include habitual smoking, regular and persistent tobacco use, smoking quantity (SQ), maximum number of cigarettes smoked in a 24-h period (MaxCigs24), the Heaviness of Smoking Index (HSI), the Fagerström Tolerance Questionnaire (FTQ), the Fagerström Test for ND (FTND), and DSM-IV or DSM-IV-like criteria.

Table 1 Sample characteristics used in genome-wide linkage scans for all ND-related measures
Full size table

Inclusion criteria for the current review

As described above, more than 20-linkage scans for various ND-related behaviors have been reported during the past several years, most since 2005. In reviewing the linkage regions nominated on the basis of these studies, one finds that almost all autosomal chromosomes (except for chromosome 15) have been implicated as harboring susceptibility loci for ND-related phenotypes. Considering that: (1) numerous genomic regions have been linked to various smoking phenotypes; (2) relatively detailed lists of these nominated regions (except for those loci reported during the past year or so) can be found in recent reviews (Ho and Tyndale 2007; Li et al. 2004; Vink et al. 2004); and (3) many of these results have not been replicated in independent studies, this review focuses primarily on those regions that show “suggestive linkage” in at least two independent samples or “significant” linkage in one study according to the rigorous criteria proposed by Lander and Kruglyak (1995), which define an LOD of >3.6 or a P value of <2.2 × 10−5 as a “significant linkage” and an LOD of >2.2 but <3.6 or a P value of 7.4 × 10−4 as a “suggestive linkage.” For those reports in which genome-wide empirical P values were provided using the permutation approach, “significant linkage” was declared if the genome-wide P value was ≤0.05 and “highly significant linkage” if the P value was ≤0.001. Under such criteria, many regions that show “suggestive” linkage to ND-related measures in a single study will not be discussed in the following sections. However, this is not to suggest that the linkage peaks reported in a single study are all false positives and should be ignored in future study. To be comprehensive and helpful for other researchers in the genetic study on tobacco and other substance abuses, Fig. 1 provides a graphical summary of most, if not all, of the regions that have been nominated for “suggestive” or “significant” linkage to ND in the literature. For detailed information on these nominated regions in a single study, please refer to Table 2. Also, to ensure the comparability of these loci across studies, the map position of each marker or marker pair that defines the linkage region in the original study was checked against the most recent version of the human linkage map through the website (http://www.ncbi.nlm.nih.gov/mapview/static/humansearch.html#marsh).

Fig. 1
figure 1

A graphical illustration of the chromosomal locations of peaks or intervals with “significant” or “suggestive linkage” to all ND-related measures in individual or repeated analyses of data from the Collaborative Studies on the Genetics of Alcoholism (COGA), the Framingham Heart Study (FHS), the Mid-South Tobacco Family (MSTF) study, the Nicotine Addiction Genetics (NAG) project, the Finnish Twin Families (FTF), the Mission Indians in Southwest California, the Genetic Epidemiology Network of Arteriopathy (GENOA) study, the Smoking in Families Study (SMOFAM), the Netherlands Twin Register (NTR) study, the Genetics of Cocaine or Opioid Dependence (GCOD) study, the Christchurch sample of New Zealand, the Australian Twin Registry (ATR), and the Family Study of Panic Disorder (FSPD)

Full size image
Table 2 Nominated chromosomal regions showing “significant” or “suggestive” linkage to smoking behavior in one sample
Full size table

“Significant” or “suggestive” susceptibility loci for ND found in at least two independent studies

According to the criteria described above, 13 linkage regions on 11 chromosomes have been identified. These linkage regions are summarized in Table 3 and Fig. 2. On inspection, several features become evident. First, except for chromosomes 5 and 9, for each of which two regions have been identified (defined as Regions 1 and 2), only one region was identified for other chromosomes. Second, the regions on chromosomes 9 (from 91.9 to 136.5 cM based on the Marshfield map), 10, 11, and 17 have received greater independent replication than the other regions. For example, the linkage region from 91.9 to 136.5 cM on chromosome 9 has been detected in four independent samples, namely, the FHS (Li et al. 2003b), COGA (Bergen et al. 1999), the EA sample of GCOD (Gelernter et al. 2007), and the AA sample of MSTF (Li et al. 2006). Within this linkage region, three genes, namely, γ-aminobutyric acid type B (GABAB) receptor subunit 2 (BABAB2), neurotrophic tyrosine kinase receptor 2 (NTRK2), and Src homology 2 domain-containing transforming protein C3 (SHC3), have been identified using family based association analysis and demonstrated to be significantly associated with ND in the MSTF sample (Beuten et al. 2005, 2007b; Li et al. 2007b). Also, the genomic region from 62 to 158 cM on chromosome 10 has been linked to ND in five independent populations: the Christchurch sample of New Zealand (Straub et al. 1999; Sullivan et al. 2004), FTF (Loukola et al. 2007), the EA sample of GCOD (Gelernter et al. 2007), the AA sample of MSTF (Li et al. 2006), and the EA sample of MSTF (Li et al. 2007a). Further, the region on chromosome 11 was detected by my research group in the FHS sample (Li et al. 2003b; Wang et al. 2005) and in both the AA (Li et al. 2006) and EA (Li et al. 2007a) samples of the MSTF cohort, as well as by Gelernter et al. (2004) in the FSPD sample, by Loukola et al. (2007) in the FTF sample, and by Morley et al. (2006) in the ATR sample. Because β-arrestin 1 is located in this region and is an important regulator of signal transduction mediated by opioid receptors through promotion of receptor desensitization and internalization (Bradaia et al. 2005; Cen et al. 2001; Gainetdinov et al. 2004), we were motivated to determine whether the β-arrestins 1 and 2 (located in a linked region to ND on chromosome 17; see below for details) are associated with ND. Our results indicated that these two genes are significantly associated with ND in European smokers (Sun et al. 2007). Furthermore, we found the strength of these associations to be higher after removal of the smoking quantity component from HSI and FTND scores in both the AA and EA samples, suggesting that these two genes play a critical role in biological processes involved in the regulation of smoking urgency (Sun et al. 2007).

Table 3 Nominated chromosomal regions with “significant” or “suggestive” linkage to smoking behavior in at least two independent samples
Full size table
Fig. 2
figure 2

Summary of chromosomal locations of nominated regions for all ND-related measures with “significant” or “suggestive” linkage score by at least two independent studies. Only chromosomes with positive linkages are shown. The linkage results were obtained from the following studies: AA/MSTF (Li et al. 2006); EA/MSTF (Li et al. 2007a); FHS (Li et al. 2003b; Wang et al. 2005); AA/GCOD and EA/GCOD (Gelernter et al. 2007), COGA (Bergen et al. 1999; Bierut et al. 2004; Duggirala et al. 1999), SMOFAM (Swan et al. 2006), FTF (Loukola et al. 2007), Mission Indians (Ehlers and Wilhelmsen 2007), FSPD (Gelernter et al. 2004), Christchurch (Straub et al. 1999; Sullivan et al. 2004), ATR (Morley et al. 2006), and Finnish/NAG and Australia/NAG (Saccone et al. 2007). Abbreviations: AA/MSTF African-American sample of the Mid-South Tobacco Family study, EA/MASTF European-American sample of the Mid-South Tobacco Family study, FHS Framingham Heart Study, GOCA Collaborative Studies on the Genetics of Alcoholism, Australia/NAG the Australia family sample of the Nicotine Addiction Genetics project, Finnish/NAG the Finnish family sample of the Nicotine Addiction Genetics project, FTF Finnish Twin Families, AA/GCOD African-American sample of Genetics of Cocaine or Opioid Dependence study, EA/GHCOD European-American sample of Genetics of Cocaine or Opioid Dependence study, Mission Indians Mission Indians in Southwest California, SMOFAM Smoking in Families Study, NTR Netherlands Twin Register study, Christchurch Christchurch sample of New Zealand, ATR Australian Twin Registry, and FSPD Family Study of Panic Disorder. Linkage peak marked with * on chromosomes 5, 10, 11, and 20 indicates a “significant linkage,” as reported in original study

Full size image

The region from 31.9 to 65 cM on chromosome 17 has been linked to ND in four studies based on three independent samples, FHS (Li et al. 2003b; Wang et al. 2005), COGA (Duggirala et al. 1999), and the EA sample of MSTF (Li et al. 2007a). Since the identification of linkage of the region on chromosome 17 to ND in our genome-wide linkage scan for SQ in the FHS sample, we have conducted candidate gene-based association analysis of this region as we did for the linked region on chromosomes 9 and 11. Our family-based association analysis revealed that GABA-A receptor-associated protein (GABARAP) (Lou et al. 2007), Discs, large homolog 4 (DLG4) or post-synaptic density protein-95 (Lou et al. 2007), protein phosphatase regulatory subunit B1 (PPP1R1B) or dopamine- and cAMP-regulated phosphoprotein, 32-KD; DARPP32 (Beuten et al. 2007a), and β-arrestin 2 (Sun et al. 2007) are significantly associated with ND in at least one of the two MSTF samples.

Third, of the 13 nominated loci listed in Table 3 and Fig. 2, four showed evidence of “significant” linkage to ND. They are located on chromosomes 5 with a genome-wide P value of 0.037 for FTND in the AA sample of GCOD (Gelernter et al. 2007), 10 with a maximum LOD score of 4.17 for SQ in the AA sample of MSTF (Li et al. 2006), 11 with a pointwise P value of 0.000001 for SQ in FHS (Li et al. 2003b), and 20 with a maximum LOD score of 4.22 for MaxCigs24 in the Finnish family sample of NAG (Saccone et al. 2007). Finally, although 13 susceptibility loci on 11 chromosomes are nominated here, we should not assume the regions identified in different populations are same set of genes or genetic variants. Rather, although these regions are more likely to harbor susceptibility loci for ND, the nature of the genetic variants may differ across samples.

Nominated “significant” susceptibility loci for ND in one study

As shown in Table 4, eight genomic regions, on chromosomes 1, 5, 10, 11, 12, 16, 20, and 22, have been nominated as “significant” loci for ND-related phenotypes. Of these loci, regions on chromosomes 1 and 5 were detected with the empirically genome-wide significance level determined by permutation analysis of at least 1,000 simulated genome-wide scans (Gelernter et al. 2007; Wang et al. 2005). The other six regions were detected with conventional one-round linkage analysis according to the theoretical threshold (Li et al. 2003b, 2007a, 2006; Saccone et al. 2007; Swan et al. 2006). Unlike the regions on chromosomes 1, 12, and 16, the regions on chromosomes 5, 10, 11, 20, and 22 have been replicated by other independent studies, although the LOD score or P value from other studies did not reach the threshold for “significant” linkage.

Table 4 Nominated chromosomal regions with “significant” linkage to smoking behavior
Full size table

Interestingly, although the significant region from 151.9 to 175.6 cM (based on the Marshfield map) on chromosome 1 (Wang et al. 2005) has received only limited support from two independent human studies (Bergen et al. 1999; Goode et al. 2003), it receives strong support from a linkage study for oral nicotine consumption in C57BL/6J × C3H/HeJ F2 intercross mice (Li et al. 2007c). Among the four detected significant quantitative trait loci (QTL), the locus with the largest LOD score, 15.7, was located at around 96 cM on chromosome 1 (Li et al. 2007c). This region of the mouse genome is syntenic with human chromosome 1 at around 169 cM. As for the “significant” linkage for ND on chromosomes 12 and 22, it has been detected only in the combined AA and EA samples of the MSTF cohort (Li et al. 2007a) and in the combined Australian and Finnish samples of NAG (Saccone et al. 2007). Given that plausible candidate genes with known biological functions in the etiology of dependence on nicotine and other substances of abuse are located within these regions, including ionotropic N-methyl d-aspartate glutamate receptor (NMDA) subunit 2B, and neurotrophin 3, GABA-A receptor-associated protein-like protein 1 on chromosome 12 and β-adrenergic receptor kinase 2 on chromosome 22, more linkage and position-based association studies are greatly needed to validate these linkage results.

Conclusions and perspectives

Despite inherent difficulties in conducting genetic studies on complex traits, significant progress has been made in recent years in the search for susceptibility loci for ND. By applying the same rigorous criteria for determination of “significant” or “suggestive” linkage to all reported linkage peaks for ND-related phenotypes, and requiring evidence from at least two independent studies, 13 regions on 11 chromosomes have been identified. Of these, the regions on chromosomes 9 (between 91.9 and 136.5 cM), 10, 11, and 17 have been detected by the greatest number of independent studies. In addition, a list of eight “significant” linkages on chromosomes 1, 5, 10, 11, 12, 16, 20, and 22 is provided. Considering that these regions have received the most support, it is suggested they be afforded the highest priority in searching for vulnerability genes for ND in future studies.

Several other issues deserve mention. Although different measures have been used to assess ND across studies, the linkage results appear similar, suggesting this phenomenon is robust. The genetic underpinnings of ND may not be critical to the ultimately observed phenotypic variation in ND characteristics. Although various linkage peaks were identified using different ND measures, smoking quantity (i.e., daily smoking rate or its variants) has yielded the most reproducible and strongest findings. Evidence that different genomic regions are definitively associated with specific ND qualities awaits further research. Second, sample size is likely to be a significant consideration in some of the failure to replicate across studies. Early investigations of genetic linkage or association studies of ND commonly included only a few to several hundred subjects, a sample size now considered unlikely to have the power required to yield conclusive results. Finally, although it has been common to conduct genetic studies in samples consisting of participants of various origins, recent findings suggest this may not represent an optimal approach, potentially producing results that are confusing or misleading. This point is well illustrated in a recent study in which several linkage peaks were detected in one ethnic sample, and pooling subjects across ethnic groups did not improve the statistical power to detect linkage (Li et al. 2007b). For example, by comparing the linkage results identified from the AA and EA samples and their combination, we found four overlapping regions on chromosomes 9 (two regions), 11, and 18 in the two samples. Furthermore, we identified five regions, on chromosomes 2, 4, 10, 12, and 17, that showed linkage to ND only in the EA sample and two regions, on chromosomes 10 and 13, that showed linkage only in the AA sample. This indicates that genetic differences underlying ND exist in these two ethnic populations. Given that populations of primarily European origin have been used in most reported studies on ND, more genetic studies with other ethnic groups such as AAs and Asians are greatly needed in the future.