INTRODUCTION

Tobacco use is the leading cause of preventable death in developed countries. According to the 2005 Centers for Disease Control and Prevention estimates, smoking annually causes 18% of total deaths, or approximately 440 000 people, in the United States alone, with more than US$168 billion in direct and indirect medical costs. The addictive potential of tobacco is exemplified by the difficulty in quitting. Nearly 35 million smokers make a serious attempt to quit each year, but less than 7% who try to quit on their own remain tobacco free more than a year. Of the roughly 4000 ingredients in cigarette smoke, nicotine is the primary addictive component. Studies have demonstrated that individual vulnerability to develop nicotine dependence (ND) is influenced by a combination of genetic and environmental factors, with an estimated heritability of at least 50% (Li et al, 2003; Sullivan and Kendler, 1999).

The addictive effects of nicotine operate through the dopaminergic system in the brain. Exposure to nicotine increases neurotransmitter dopamine release (Nisell et al, 1994; Pontieri et al, 1996). In particular, the dopaminergic mesocorticolimbic reward pathways have been frequently implicated in the etiology of nicotine and other drug addictions, as well as non-drug addictive behaviors. Consequently, a great deal of attention has been devoted to determining whether variations in genes within a dopaminergic system could account for the heritable factors in susceptibility to ND (Ho and Tyndale, 2007; Li, 2006; Li et al, 2004). Since dopamine receptors mediate effects of dopamine, the five G-protein-coupled receptors, classified into D1-like (D1 and D5) and D2-like (D2, D3, and D4) groups, are considered viable candidates for the study of the genetic bases of ND. Of the five dopamine receptors, D1 and D2 emerge as two key components of the dopaminergic system. They are largely expressed in the nucleus accumbens and ventral caudate-putamen, two reward-related brain regions, and their concurrent absence in the early postnatal period is lethal (Kobayashi et al, 2004). Previously, we have reported the significant association of DRD1 with ND in an independent study (Huang et al, 2008). In this study, we will focus on determination of a possible association of dopamine D2 receptor encoding gene DRD2 with ND.

DRD2 has been extensively investigated regarding its association with various psychiatric disorders, including nicotine addiction, with the majority focusing on a polymorphism known as Taq IA (rs1800497 in this study and in the NCBI dbSNP database). On the basis of a large number of studies, conflicting results have been documented with respect to smoking behavior. Some reported positive findings (Anokhin et al, 1999; Cinciripini et al, 2004; Comings et al, 1996; Erblich et al, 2004), but not all (Bierut et al, 2000; Johnstone et al, 2004; Singleton et al, 1998). A similar pattern of conflicting results exists regarding the association of DRD2 with alcoholism (Munafo et al, 2007). Although our previous meta-analysis revealed a significant association of Taq IA polymorphism with smoking behavior (Li et al, 2004), which was verified by another meta-analysis (Munafo et al, 2004), significant heterogeneity was noted across studies. When a random-effects model was applied to account for between-study differences, no significant association remained (Munafo et al, 2004).

The Taq IA polymorphism is located 9.5 kb downstream from DRD2. Although the DRD2 Taq IA polymorphism notation has been well established in the literature, a recent report indicated that it actually resides in the neighboring gene ANKK1 (ankyrin repeats and kinase domain containing 1 gene) (Neville et al, 2004). The encoded protein ANKK1, previously named as SgK288, belongs to a family of serine/threonine kinases in a branch of human kinome, which share a highly homologous amino- (N-) terminal kinase domain (see Supplementary Figure 1 for details) (Manning et al, 2002). Although their carboxyl (C) termini differ, this family of proteins including receptor-interacting protein kinases (RIPKs) 1–4 have been shown to be involved primarily in activation of transcription factor NF-κB (Meylan and Tschopp, 2005). ANKK1 is highly similar to RIPK4, sharing not only the N-terminal kinase domain, but also the C-terminal ankyrin repeat domain. The ‘C’ to ‘T’ transition of Taq IA polymorphism causes an amino-acid change (Glu713Lys) in the C-terminal ankyrin repeat domain, which has been suggested to mediate protein–protein interaction (Michaely et al, 2002; Neville et al, 2004).

Interestingly, a recent family-based association study revealed that ANKK1 is significantly associated with ND in both African-American (AA) and European-American (EA) populations, while DRD2 showed a relative weaker association (Gelernter et al, 2006). Little evidence was found for association of the Taq IA polymorphism with ND. Although the Taq IA polymorphism is a non-synonymous base substitution, it was suggested to be potentially attributable to linkage disequilibrium (LD) with a more centromeric genetic variant within DRD2 or ANKK1 (Gelernter et al, 2006).

Since a region near DRD2 on human chromosome 11q23 has demonstrated significant linkage with cigarette consumption and ND in genome-wide scans (Gelernter et al, 2007; Morley et al, 2006) and DRD2 has an established close relationship with its neighboring ANKK1 in genetic studies, we evaluated possible associations of both DRD2 and ANKK1 with ND in our Mid-South Tobacco Family (MSTF) cohort. Our results not only reveal a stronger association of ANKK1 than that of DRD2 with ND in both AA and pooled samples, but also identify a possible causative single nucleotide polymorphism (SNP) rs2734849. Moreover, we demonstrate that the rs2734849 polymorphism is functional in the ANKK1 negative regulation of transcription factor NF-κB.

MATERIALS AND METHODS

Study Participants

The subjects in this study were of either AA or EA origin and were recruited from the mid-south states in the United States during 1999–2004, designated as the ‘MSTF’ cohort (Li, 2006; Li et al, 2005). Proband cigarette smokers were required to be at least 21 years old, to have smoked for at least the last 5 years, and to have consumed an average of 20 cigarettes per day for the last 12 months. Siblings and parents of a smoking proband were recruited whenever possible, regardless of their smoking status. The cohort included 2037 subjects in 602 nuclear families, with 671 subjects in 200 EA families and 1366 subjects in 402 AA families. Extensive clinical data were collected from each participant, including demographics (eg, sex, age, race, biological relationships, weight, height, years of education, and marital status), medical history, smoking history, current smoking behavior, ND, and personality traits, assessed by various questionnaires, available at National Institute on Drug Abuse Genetics Consortium website (http://zork.wustl.edu/nida). Individuals exhibiting substance abuse other than for alcohol were excluded. All participants provided informed consent. The study protocol and forms/procedures were approved by all participating Institutional Review Boards.

For each smoker, the degree of ND was ascertained by the three most commonly used measures: smoking quantity (SQ, defined as the number of cigarettes smoked per day); the Heaviness of Smoking Index (HSI, 0–6 point scale), which includes SQ and smoking urgency (ie, how soon after waking up does the participant smoke the first cigarette); and the Fagerström test for ND score (FTND, 0–10 point scale) (Heatherton et al, 1991). All three measures have been used consistently in our previous genetic studies on ND; please refer to these papers for detailed demographic and clinical characteristics of the MSTF cohort (eg, Beuten et al, 2006, 2007; Li et al, 2005, 2007).

DNA Extraction, SNP Selection, and Genotyping

Genomic DNA samples were prepared from peripheral blood tissue provided by each participant using the Maxi blood DNA extraction kit from Qiagen (Valencia, CA). In total, 7 SNPs for ANKK1 and 16 SNPs for DRD2 were selected from the NCBI dbSNP database to cover the entire genomic region of the ANKK1 and DRD2 genes. Prior published reports addressing SNPs within DRD2 and ANKK1, minor allele frequency, functional potential, and validation evidence were considered in determining final SNP selection. Information regarding these SNPs, including their IDs, allelic variants, contig positions, heterozygosities, and site functions is shown in Table 1. The TaqMan primer–probe set for each SNP was purchased from Applied Biosystems (Foster City, CA) and their sequences are provided in Supplementary Table 1. PCR reactions for genotyping were performed in a 384-well microplate format, using TaqMan universal PCR master mix. A standard PCR amplification protocol with 2 min at 50°C and 10 min at 95°C followed by 40 cycles of 25 s at 95°C and 1 min at 60°C was applied. The following allelic discrimination analysis was carried out using the ABI Prism 7900HT Sequence Detection System (Applied Biosystems).

Table 1 Information Regarding SNPs Selected from NCBI dbSNP Database and their Allele Frequencies in AA, EA, and Pooled Samples
Full size table

SNP- and Haplotype-Based Association Analysis

The PedCheck program (O’Connell and Weeks, 1999) was used to identify any inconsistent Mendelian inheritance, non-paternity, and/or genotyping errors. The detected inconsistencies were excluded from subsequent association analyses. To verify data quality, we also checked the genotyping results for any significant departure from Hardy–Weinberg equilibrium (HWE). Pairwise LD between all SNPs was evaluated using the Haploview program (Barrett et al, 2005) with the option of determining haplotype blocks according to the criteria defined by Gabriel et al (2002).

Individual SNP association was determined using the Pedigree-Based Association Test (PBAT) program employing generalized estimating equations (Lange et al, 2004). Haplotype-based association was determined using the Family-Based Association Test (FBAT) program with the option of computing P-values of the Z statistic based on Monte Carlo sampling (Horvath et al, 2004). The AA and EA samples were analyzed separately, with gender and age included as covariates in both PBAT and FBAT analyses. Ethnicity was employed as a covariate for analyses of the pooled sample. All three ND measures (SQ, HSI, and FTND) were analyzed under the additive and dominant models. All associations found to be significant were corrected for multiple testing according to the SNP spectral decomposition (SNPSpD) approach (Nyholt, 2004) for individual SNP analysis, and using Bonferroni correction by dividing the significance level by the number of major haplotypes (frequency >5.0%) for haplotype-based association analysis.

Vector Construct

The full-length cDNA of ANKK1 was fused with a FLAG tag at N terminus, cloned into pcDNA 3.1 expression vector (Invitrogen, Carlsbad, CA), and sequenced to ensure it contained a G-allele for SNP rs2734849 (encoding arginine, R490). The A-allele of rs2734849 (for arginine to histidine mutation, R490H) was obtained by a site-directed mutation using QuikChange II XL mutagenesis kit (Stratagene, La Jolla, CA) and the following designed primers: 5′-gaccccaacctgcatgaggctgagggc-3′ (forward) and 5′-gccctcagcctcatgcaggttggggtc-3′ (reverse). The lysine to arginine mutation (K51R) at ANKK1 kinase domain was also realized with a site-directed mutation using the following designed primers: 5′-gagtacgccatcaggtgcgccccctgc-3′ (forward) and 5′-gcagggggcgcacctgatggcgtactc-3′ (reverse). All constructs were confirmed by DNA sequencing.

Cell Transfection and Luciferase Assay

Human neuroblastoma SH-SY5Y cells were purchased from American Type Culture Collection (Manassas, VA) and cultured in a 1 : 1 mixture of minimum essential medium and F12 medium with fetal bovine serum added to a final concentration of 10%. The transfections were performed with Lipofectamine 2000 reagent (Invitrogen), in accordance with the manufacturer's protocol. Each ANKK1 expression vector (R490, R490H, or K51R) was co-transfected with the PathDetect pNF-κB-Luc cis-reporter plasmid (Stratagene) with a 3 : 1 ratio, using pcDNA 3.1 vector as a mock control. The luciferase activities in SH-SY5Y cells after 48 h cell transfection were analyzed using the Luciferase Assay System with 20/20n luminometer method (Promega, Madison, WI). In each experiment, quadruplicate transfections were made for each plasmid and their luciferase activities were averaged. Three independent experiments were conducted to assure findings were replicable.

RESULTS

Analysis of SNPs with HWE and Pairwise LD Tests

We genotyped a total of 23 SNPs, 7 within ANKK1 and 16 within DRD2. Tests for HWE indicated that no SNP deviated significantly from HWE in either the AA or EA sample (minimum P=0.164 and 0.475 for the AA and EA samples, respectively), confirming the high quality of the genotyping data. The allele frequencies for each SNP in the AA, EA, and pooled samples are shown in Table 1. In light of ethnic-specific characteristics of SNPs (Gabriel et al, 2002; Wall and Pritchard, 2003) and known ethnic differences in ND and nicotine metabolism (Benowitz et al, 1999; Perez-Stable et al, 1998), we performed a separate association analysis on each ethnic sample. Figure 1 demonstrates the pairwise LD structures among the 23 SNPs in the AA and EA samples. The r2 values were calculated using the Haploview program (Barrett et al, 2005) and haplotype blocks were determined according to the criteria specified by Gabriel et al (2002). We observed two LD blocks of ANKK1 in AAs, which were different from those in EAs. In addition, three LD blocks of DRD2 observed in the EA sample were not found in the AA sample. Notably, the rs1800497 polymorphism of ANKK1 was not in a LD block of ANKK1 in AAs, but was in a LD block crossing ANKK1 and DRD2 in EAs (Figure 1).

Figure 1
figure 1

Haploview-generated LD patterns for 23 SNPs within ANKK1 and DRD2 in the AA and EA samples. Pairwise LD between all SNPs was evaluated using the Haploview program (Barrett et al, 2005) with the option of determining haplotype blocks according to the criteria defined by Gabriel et al (2002). The number in each box represents the r2 value for each SNP pair.

Full size image

Association Analysis of Individual SNPs

Program PBAT-GEE (Lange et al, 2004) was employed for individual SNP analysis. For DRD2, we found that rs7131056 and rs4274224 were significantly associated with all three ND measures (SQ, HSI, and FTND) (P=∼0.036–0.048), rs4648318 with FTND (P=0.041), and rs6278 with both HSI (P=0.040) and FTND (P=0.039) in the EA sample. In AAs, only SNP rs6589377 yielded a significant association with FTND (P=0.049). However, all these associations were nonsignificant after correction for multiple testing on the basis of the SNPSpD approach (Nyholt, 2004). No SNP was found significantly associated with any ND measure in the pooled sample (Table 2).

Table 2 Association P-Values of Each SNP with Three ND Measures under the Additive (Top Line) and Dominant (Bottom Line) Models in AA, EA, and Pooled Samples
Full size table

For ANKK1, three SNPs (rs1800497, rs2734849, and rs11604671) were significantly associated with ND. Of these, rs1800497 showed significant associations with HSI under the dominant model in both EA and pooled samples (P=0.038 and 0.042, respectively) and with FTND in EAs (P=0.043). Also, SNP rs11604671 showed significant associations under the dominant model with SQ (P=0.028) and HSI (P=0.047) in AAs and under the additive model with SQ (P=0.023), HSI (P=0.0091), and FTND (P=0.050) in the pooled sample. All these associations, except rs11604671 with HSI, were rendered nonsignificant after correction for multiple testing. In contrast, rs2734849 had significant associations with all three ND measures in both AA and pooled samples under the dominant model even after correction for multiple testing (P<0.010). Under the additive model, rs2734849 yielded significant associations with HSI in AAs (P=0.023); it was also associated with SQ (P=0.023), HSI (P=0.0064), and FTND (P=0.027) in the pooled sample (Table 2).

Haplotype-Based Association Analysis

We employed the FBAT program (Horvath et al, 2004) to assess haplotype-based associations for all five contiguous SNPs using a sliding window approach (Lin et al, 2004). For haplotypes formed by rs2075654–rs2587548–rs2075652–rs1079596–rs4586205 in DRD2, haplotype G-G-C-G-T had significant associations under the dominant model with FTND in EAs (Z=−2.13, P=0.033, at a frequency of 56%) and with all three ND measures in AAs (frequency=11%). However, only the association with HSI in AAs remained significant after Bonferroni correction (Z=2.67, P=0.0075) (Table 3).

Table 3 Z-Scores and Permutation P-Values for Two Major Haplotypes of rs2075654–rs2587548–rs2075652–rs1079596–rs4586205 in DRD2 with Three ND Measures under the Additive (Top Line) and Dominant (Bottom Line) Models in AA, EA, and Pooled Samples
Full size table

In ANKK1, haplotype T-T-C-G-A, formed by rs10891545–rs7945132–rs4938013–rs7118900–rs11604671, yielded significant associations under the additive model with all three ND measures in EAs (frequency=47%) and in the pooled sample (frequency=20%). Significant associations were identified under the dominant model for SQ in AAs (frequency=6%) and with all three ND measures in the pooled sample as well. After Bonferroni correction, the significant associations remained for SQ (Z=2.70, P=0.0071) and HSI (Z=2.72, P=0.0066) in the pooled sample under the additive model (Table 4). Another haplotype C-G-A-G-G, formed by rs4938013–rs7118900–rs11604671–rs2734849–rs1800497, was significantly associated with all three ND measures under both additive and dominant models in both AA (frequency=5%) and pooled (frequency=19%) samples. Most of these associations remained significant after Bonferroni correction (Z>2.64, P⩽0.0083) (Table 5).

Table 4 Z-Scores and Permutation P-Values for Some Major Haplotypes of rs10891545–rs7945132–rs4938013–rs7118900–rs11604671 in ANKK1 with Three ND Measures under the Additive (Top Line) and Dominant (Bottom Line) Models in AA, EA, and Pooled Samples
Full size table
Table 5 Z-Scores and Permutation P-Values for Some Major Haplotypes of rs4938013–rs7118900–rs11604671–rs2734849–rs1800497 in ANKK1 with Three ND Measures under the Additive (Top Line) and Dominant (Bottom Line) Models in AA, EA, and Pooled Samples
Full size table

Function of ANKK1 in NF-κB-Regulated Gene Expression

Sequence analysis revealed that ANKK1 has apparent similarity to the NF-κB-activating RIPKs with a highly homologous N-terminal kinase domain (see Supplementary Figure 1 for details). In sharing the C-terminal ankyrin repeat domain (Figure 2a), ANKK1 has overall identity of 37% and similarity of 52% with RIPK4. In view of RIPK's function in regulating transcription activity of NF-κB, we investigated the capacity of ANKK1 to activate an NF-κB-luciferase reporter. Surprisingly, we found that an overexpression of ANKK1 (R490) in human neuroblastoma SH-SY5Y cells resulted in a weak suppression (∼5%) of NF-κB-reporter gene expression (Figure 2b), which differed from RIPK4 activation of NF-κB (Meylan et al, 2002). Compared with the critical role of RIPK4 N-terminal kinase domain in activation of NF-κB (Meylan et al, 2002), we found that the kinase domain of ANKK1 was not involved in suppression of NF-κB activity. We demonstrated that a lysine-to-arginine mutation of ANKK1 (K51R), predicated to reduce ATP binding and catalytic activity of the kinase domain (Hanks and Hunter, 1995), yielded no significant change in the reporter gene suppression (Figure 2b). This suggests that ANKK1 has a biological function different from RIPK4, despite sharing similar protein domains.

Figure 2
figure 2

rs2734849 affects suppression of ANKK1 on NF-κB-regulated gene expression. (a) A subfamily of serine/threonine kinases, which share high similarity in their N-terminal kinase domains, but possess distinct C termini. RIPK1 has a so-called death domain (DD or DEATH); RIPK2 has a related caspase recruitment domain (CARD); RIPK4 has 10 ankyrin repeats; and ANKK1 has 12 ankyrin repeats. (b, c) NF-κB-reporter assay. Vector pcDNA3.1 was used as a mock control, and mutant K51R was used as a mutation control (b). ANKK1/R490 is encoded by ANKK1 with rs2734849/G allele (b, c). Mutant R490H is encoded by ANKK1 with rs2734849/A allele (c). Each plasmid construct was co-transfected with pNF-κB-Luc reporter vector, and their luciferase activities in human neuroblastoma SH-SY5Y cells were measured after 48 h of cell transfection. The result is representative of three independent experiments. Data are shown as mean±SD (N=4). (d) R490H mutation site resides on protein surface in the three-dimensional (3D) structure of ankyrin repeat domain. The 3D structure of 12 ankyrin repeats is modified from 1N11 in Protein Data Bank (PDB), using Cn3D 4.1 software (NCBI). R490H mutation site is determined by a protein sequence alignment.

Full size image

Since our PBAT and FBAT analysis indicated that rs2734849 is significantly associated with ND, and rs2734849 is a non-synonymous polymorphism with G to A transition causing an alteration from arginine to histidine at amino-acid residue 490 (R490H) in ANKK1, we further investigated whether rs2734849 represents a functional polymorphism to change ANKK1 suppression on NF-κB activity. Using the NF-κB-luciferase reporter assay, we found that the ‘A’ allele of rs2734849 in ANKK1 had greater suppression (∼30%) on NF-κB-regulated luciferase activity than the ‘G’ allele of rs2734849 (Figure 2c). This indicates that rs2734849 is a functional polymorphism that may be responsible, at least partly, for the observed association of ANKK1 with ND. In a protein sequence alignment and crystal structure comparison with a 12-ankyrin repeat domain (Michaely et al, 2002), we found that the residue at 490 position resides on the surface of protein (Figure 2d), a prerequisite for its involvement in mediating protein–protein interaction in the signal-transduction processes leading to inhibition of NF-κB activity.

DISCUSSION

In this study, we show that (1) ANKK1 is significantly associated with ND in our family-based PBAT and FBAT analysis; (2) the strength of this association is greater than for DRD2; (3) SNP rs2734849 in ANKK1 is significantly associated with ND in both AA and pooled samples, and a likely causative polymorphism for ND; and (4) SNP rs2734849 in ANKK1 is functional in yielding differential suppression of NF-κB-regulated gene expression. Since transcription factor NF-κB is a necessary and sufficient signal to induce DRD2 expression (Bontempi et al, 2007; Fiorentini et al, 2002), this suggests that variants of ANKK1, specifically rs2734849, may function to affect DRD2 expression.

As a G-protein-coupled receptor in dopaminergic neurons, the dopamine D2 receptor plays a prominent role in reward-mediating mesocorticolimbic pathways. As such, DRD2 variants have been the focus of many genetic association studies of addictive behaviors. Of the loci at DRD2 studied, the Taq IA polymorphism (rs1800497) has received the most attention, and has been implicated in smoking behavior and alcoholism (Li et al, 2004; Munafo et al, 2004, 2007), as well as in smoking cessation in pharmacogenetic studies (Berlin et al, 2005; David et al, 2007). However, its importance is controversial given inconsistent results from association studies. Recently, significant efforts have been expended to investigate additional genes in the DRD2 region, including TTC12 (tetratricopeptide repeat domain 12), ANKK1, and NCAM1 (neural cell adhesion molecular 1) (Gelernter et al, 2006; Yang et al, 2007). In a recent family-based association analysis using two distinct American populations, relatively weak evidence emerged for association of ND with markers at DRD2 and NCAM1, whereas strong evidence for multiple SNPs at TTC12 and ANKK1 was noted (Gelernter et al, 2006). In conjunction with our results, the Taq IA polymorphism likely contributes to LD with ANKK1, rather than adjacent DRD2. This may explain previous inconsistent findings on ND and other psychiatric disorders regarding this polymorphism.

Our current family-based association study provides an independent replication of Gelernter et al (2006) in that ANKK1 has a stronger association with ND than DRD2. However, some discrepancies exist at the individual SNP level. Potential reasons for these discrepancies include differences in sample characteristics across studies, as well as the assessment of ND. Gelernter et al's sample was originally recruited for opioid and/or cocaine dependence, whereas the MSTF participants were recruited based on ND, specifically excluding other substance dependence except for alcohol. Another factor may be the statistical power. Our AA sample (N=1366 in 402 families) was larger than that of Gelernter et al (N=854 subjects in 319 families), whereas our EA sample was slightly smaller (N=671 in 200 families in this study; N=761 subjects in 313 families in Gelernter et al's study). In addition, we employed SQ, HSI, and the FTND to assess ND, measures emphasizing the amount, frequency and pattern of tobacco consumption, while Gelernter et al used FTND and DSM-IV criteria, the latter of which addresses withdrawal, impairment, difficulty quitting, and other factors in a diagnostic framework.

In contrast with the weak association of DRD2 with ND in the EA sample, based on both SNP and haplotype association analyses (Tables 2 and 3), we detected strong associations of ANKK1 with ND in the AA and pooled samples, specifically for rs2734849. Evidence suggests polymorphism rs11604671 may also contribute to this association (Tables 2, 4 and 5). In addition, the haplotypes with either ‘A’ allele of rs11604671 or ‘A-G’ by rs11604671 and rs2734849 yielded positive Z-scores in the haplotype analyses (Tables 4 and 5), implicating a protective function in the development of ND. Given both rs11604671 and rs2734849 are non-synonymous polymorphisms, possessing ‘G’ to ‘A’ transitions that cause amino-acid changes from arginine to either glycine or histidine in the C-terminal ankyrin repeat domain of ANKK1, we suspect these SNPs represent two causative polymorphisms in the observed association of ANKK1 with ND in the AA and pooled samples.

ANKK1 is classified as one of the RIPKs, which have emerged as essential sensors of cellular stress, integrating both extracellular stress signals transmitted by various cell surface receptors, and signals emanating from intracellular stress (Meylan and Tschopp, 2005). Although RIPKs share high homolog kinase domains (Supplementary Figure 1) and have been shown to affect the NF-κB pathway, the kinase domain of RIPK4 is the only one required to activate NF-κB (Meylan et al, 2002). Despite the similarity between ANKK1 and RIPK4, ANKK1 appears to be a negative regulator of NF-κB (Figure 2b and c), suggesting that its biological function differs from that of RIPK4. Given the K51R mutant at the kinase domain showed no effect on suppression (Figure 2b), we concluded that the C-terminal ankyrin repeats acts as an inhibitory domain, just like its counterpart in RIPK4 (Meylan et al, 2002). As activation of transcription factor NF-κB is dependent on the formation of a multi-protein complex, the C-terminal ankyrin repeat domain in ANKK1 (or RIPK4) may function to block some protein interactions, and thus trigger a partial inhibitory effect. Variant R490H, caused by the rs2734849 polymorphism, may induce differential protein-binding affinity of ANKK1 in this NF-κB-inhibition process, leading to enhanced inhibition (Figure 2c). Since the promoter region of DRD2 contains NF-κB-binding sites (Bontempi et al, 2007; Fiorentini et al, 2002), our findings provide a putative connection between ANKK1 variants and DRD2 expression. In some in vitro tests, nicotine has been shown to inhibit transcriptional activity of NF-κB (Yoshikawa et al, 2006; Zhang et al, 2001). However, it is not clear whether nicotine inhibits NF-κB signaling via ANKK1.

Expression of DRD2 determines density and availability of the dopamine D2 receptor in dopaminergic neurons. Recent positron emission tomography (PET) studies have indicated that decreased dopamine D2 receptor availability in the striatum of brain may be a predisposing neurobiological trait for substance dependence (Dalley et al, 2007; Morgan et al, 2002; Nader et al, 2006), as opposed to solely a consequence of chronic exposure to abused drugs (Heinz et al, 2004; Martinez et al, 2004; Volkow et al, 1997, 2004). The finding suggests that individual differences in DRD2 expression (or dopamine D2 receptor availability) relate to a specific behavioral process that confers addiction vulnerability to abused drugs. Specifically, low DRD2 expression in the striatum predicts increased consumption of abused drugs (Dalley et al, 2007; Nader et al, 2006).

Although functional PET imaging studies have indicated that the Taq IA polymorphism is associated with reduced dopamine D2 receptor density in the striatum as well (Jonsson et al, 1999; Pohjalainen et al, 1998), this finding is not universally accepted (Laruelle et al, 1998), similar to its genetic association with smoking behavior or alcohol dependence (Li et al, 2004; Munafo et al, 2004, 2007). In this report, we report a significant association of ANKK1 with ND, but also identify a possible causative genetic variant, rs2734849, which functions to regulate NF-κB signaling, including DRD2 expression. We suspect that polymorphism rs2734849 represents a more centromeric variant in ANKK1 than the Taq IA polymorphism, which may account for conflicting reports on the association of the Taq IA polymorphism with ND. We expect future functional PET scans will confirm the association of rs2734849 in ANKK1 with dopamine D2 receptor availability, as well as the association of ANKK1 with ND.