CHRNA5 Variants Moderate the Effect of Nicotine Deprivation on a Neural Index of Cognitive Control

Abstract

Individuals with reduced attention and memory cognitive control-related processes may be motivated to smoke as a result of the cognitive enhancing effects of nicotine. Further, nicotine deprivation-induced reductions in cognitive control may negatively reinforce smoking. Minor allele carriers at rs16969968 in the nicotinic acetylcholine receptor α5 subunit gene (CHRNA5) have been shown to exhibit both reduced cognitive control and greater nicotine dependence. It is therefore of interest to see if variants in this gene moderate the influence of nicotine deprivation on cognitive control. P3b and P3a components of the event-related brain potential waveform evoked by a 3-stimulus visual oddball task are widely viewed as positive indices of cognitive control-related processes. We tested the hypothesis that individuals possessing at least one minor allele at rs16969968 in CHRNA5 would show greater nicotine deprivation induced reductions in P3b and P3a amplitude. The sample included 72 Non-Hispanic, Caucasian heavy smokers (54 men and 18 women) with a mean age of 36.11 years (SD = 11.57). Participants completed the visual oddball task during counterbalanced nicotine and placebo smoking sessions. Findings indicated that rs16969968 status did not moderate nicotine effects on P3b or P3a, whereas variation in other CHRNA5 SNPs, which are not as well characterized and are not in linkage disequilibrium with rs16969968, predicted nicotine deprivation induced reduction of P3a amplitude: rs588765 (F[1,68] = 7.74, p = 0.007) and rs17408276 (F[1,67] = 7.34, p = 0.009). Findings are interpreted in the context of vulnerability alleles which may predict nicotine effects on cognitive control.

Corresponding Author: David E. Evans, Ph.D. Department of Health Outcomes and Behavior H. Lee Moffitt Cancer Center & Research Institute Departments of Oncologic Sciences & Psychology University of South Florida david.evans@moffitt.org tel: 813-745-4373; fax: 813-449-8226

Introduction

Cognitive control refers to a wide range of attention and memory-related processes relevant to the performance of daily activities which require effort (Lavie, 2010; Pontifex et al., 2011). The effects of nicotine on cognitive control may contribute to the reinforcing effects of tobacco use (Heishman et al., 2010.). Individuals with impaired cognitive control may be especially susceptible to the cognitive enhancing effects of nicotine (Newhouse et al., 2004), and/or experience greater nicotine deprivation-induced reductions in cognition (Evans & Drobes, 2013). Identifying individuals whose smoking is motivated by the cognitive effects of nicotine could allow for the development of behavioral and/or pharmaceutical treatment programs tailored to the needs of this subgroup of smokers (Evans & Drobes, 2009).

Genome-wide association studies (GWAS) have identified nicotinic receptor polymorphisms which predict smoking behavior. Presence of a minor allele at rs16969968, a non-synonymous coding polymorphism of the nicotinic acetylcholine receptor α5 subunit gene (CHRNA5), positively predicts smoking status, cigarettes per day, nicotine dependence, and respiratory disease (Saccone et al., 2010). Carriers of a minor allele at rs169669968 also exhibit poorer performance on a variety of cognitive control-related tasks, which include subtests of the Wechsler Adult Intelligence Test (WAIS-IV), the N-back task, and a continuous performance task (Winterer et al., 2010). As such, it is possible that the rs169669968 variant impacts the reinforcement of smoking by moderating the cognitive effects of nicotine intake and/or deprivation.

A challenge to pharmacogenomic research involves the development of intermediate phenotypes with well-defined biological underpinnings (i.e., endophenotypes) which can help explain the effects of candidate genes. The traditional P300 (P3b) event-related brain potential (ERP) component is a robust neural marker of cognitive control-related processes (Polich, 2007). Identification of infrequently presented target stimuli evokes the P3b, a positive voltage deflection peaking around 400-500 ms post-stimulus onset at parietal/central scalp sites, which has been associated with the allocation of attentional resources and the updating of working memory (Polich, 2007). An additional P3 component (i.e., the P3a) can be elicited via inclusion of vivid distracting stimuli presented at the same frequency as the rare targets that evoke P3b (Polich, 2007). The P3a is a positive deflection peaking at about 300-400 ms post-stimulus onset at central/frontal scalp sites which indexes orienting of attention to stimuli not relevant to current task goals. Nicotine deprivation has been shown to reduce P3b amplitude among smokers, and also P3a amplitude among individuals lower in trait cognitive control (Evans et al., 2013). As indices of cognitive control (Evans et al., 2013), P3a/P3b measures may tap into aspects of cognition that are influenced by variation in CHRNA5, and serve as endophenotypes for cognitive processes relevant to smoking.

The present study follows up on data reported by Evans et al. (2013) by examining the association of CHRNA5 variants with nicotine deprivation-induced changes in P3b and P3a amplitudes. It was hypothesized that minor allele carriers in rs16969968 would show greater nicotine deprivation-induced reductions in P3b and P3a amplitudes. Non-coding CHRNA5 SNPs were also explored as potential moderators of nicotine deprivation effects on P3b/P3a.

Material and Methods

Participants

Buccal cells were collected as a genomic DNA source from 72 Caucasian Non-Hispanic participants from the parent P3 nicotine deprivation study (see Evans et al., 2013). The sample consisted of 54 men (mean age = 36.07, SD = 10.94, range 19-56) and 18 women (mean age = 36.22, SD = 13.64, range 21-62). Education level ranged from 6 to 18 years (mean = 12.82, SD =1.90). Mean score on the FTND was 5.63 (SD = 1.88), which reflects a moderate to high level of nicotine dependence. Mean daily number of cigarettes smoked per day was 22.36 (SD = 5.97). Analyses were limited to Caucasian Non-Hispanic participants to avoid confounds produced by the differential frequency of rare CHRNA5 variants across races. Eligible participants were between the ages of 18 and 70 years, and smoked 15 cigarettes or more per day for the past two years, with smoking status verified biochemically. Participants were excluded for recent use of nicotine products other than smoking, neurological conditions, medication use that could affect physiological responding, significant head injury/concussion, other serious medical conditions (cardiopulmonary problems) or respiratory-related illness exacerbated by smoking (e.g. bronchitis, emphysema, asthma), currently using psychoactive substances (as assessed by a urine drug test), vision problems, pregnancy or breast feeding, current psychosis, mood disorders, or non-nicotine substance dependence disorders . A more detailed description of the inclusion/exclusion criteria are reported in Evans et al. (2013).

Procedure

This study was approved by the internal review board at the University of South Florida. Written informed consent was obtained at the beginning of an initial screening session, followed by confirmation of eligibility according to the above criteria. Eligible participants next completed a questionnaire assessing nicotine dependence (Fagerström Test for Nicotine Dependence/FTND; Heatherton et al., 1991), and buccal cells were collected for DNA extraction and subsequent genotyping. Participants were then scheduled to attend two 2.5-hour experimental sessions (3 to 14 days apart) following overnight (12-hour) nicotine deprivation, which was confirmed biochemically (see Evans et al., 2013). Four cigarettes were smoked (Quest, Vector Tobacco, Inc.) during each experimental session, with two of these cigarettes smoked prior to performance of the visual oddball task, with initiation of smoking separated by 40 minutes. These cigarettes contained nicotine (.60 mg nicotine yield) at one session and placebo (< .05 mg nicotine yield) at the other session, with nicotine content administered in a double-blind fashion and in counterbalanced order across the sessions. Participants completed the Wisconsin Smoking Withdrawal Scale (WSWS; Welsch et al., 1999) near the beginning (i.e., pre-smoking) and end (i.e., post-smoking) of both experimental sessions. Two of the WSWS subscales (sleep and diet) were omitted from the scale because they are not relevant to overnight deprivation.

Cognitive tasks

An established version of the three-stimulus visual oddball task (see Hagen et al., 2006) was presented using E-Prime software (Psychology Software Tools, Inc.). Stimuli were presented for 100 ms duration every 1000 ms for a total of 400 trials split into 2 blocks (separated by a 30 second break). Participants were presented with a standard stimulus (3.0 cm diameter blue circle; 70% of trials), a target stimulus (3.5 cm diameter blue circle; 15% of trials), and a distracting stimulus (18 cm² black and white checkerboard; 15% of trials). Instructions were given to respond by pressing a button as quickly as possible to target stimuli.

Electroencephalogram (EEG) recording and data processing

The 64-channel electrode array included the 10-20 montage. Neuroscan Synamps 2 system and its accompanying SCAN 4.3 software system were used to record EEG data.

Genotyping

Buccal cell samples were used to extract Genomic DNA (Gentra Puregene tissue kit; Valencia, CA) according to the manufacturer’s instructions. The Illumina GoldenGate™assay was used to genotype the samples which were called using the BeadStudio algorithm through the Molecular Genomic Core at the Moffitt Cancer Center. Ten CHRNA5 SNPs were originally genotyped (rs3829787, rs588765, rs637137, rs17408276, rs11637635, rs17486278, rs569207, rs684513, rs4275821, and rs16969968), but the number of SNPs was reduced to five (rs3829787, rs588765, rs637137, rs17408276 and rs16969968) as SNPs with a bivariate R² of .80 or greater with other variants were removed until the remaining SNPs had R²s below this level. Valid genotyping data ranged from between 68 and 72 participants across the five SNPs. Table 1 reports SNP characteristics, including minor allele frequencies and number of participants per genotype. A linkage disequilibrium (LD) plot for these SNPs is reported in Figure 1. Genotype distributions of the five SNPs followed Hardy Weinberg equilibrium.

Table 1.

SNP Characteristics and Frequencies

SNP (N)	Allele (minor/major)	Location on gene	MAF	Genotype* MajH/Het/MinH	Genotype* MajH/MinC
rs3829787 (68)	G/A	5′ near gene	.41	20/40/8	20/48
rs588765 (72)	T/C	Intron 1	.45	19/41/12	19/53
rs637137 (68)	A/T	Intron 2	.21	43/21/4	43/25
rs17408276 (71)	C/T	Intron 4	.39	24/38/9	24/47
rs16969968 (72)	A/G: Asn/Asp	Exon 5	.31	33/33/6	33/39

^*Frequency of individuals with a genotype. Abbreviations: MAF = minor allele frequency, MajH = major homozygote, Het = heterozygote, MinH = minor homozygote, MinC = minor allele carrier.

Figure 1.

LD table of CHRNA5 SNPs. Darker shades indicative of greater association between SNPs.

Data processing

Incorrect responses on the oddball task were omitted prior to EEG data processing. ERP epochs for each trial were from 0 to 1000 milliseconds. Epoching, baseline correction, filtering, eye-blink and spherical spline artifact correction of EEG data are reported in detail in Evans et al. (2013). Midline raw waveforms at electrodes Pz (parietal), Cz (central), and Fz (frontal) for all stimulus conditions across both nicotine satiation and deprivation sessions for all participants is presented in Figure 2. A temporal-spatial principal component analysis (PCA) was used to derive factor scores representing P3b and P3a ERP component amplitude (see Evans et al., 2013), using data from the larger parent sample of 121 participants to enhance scoring reliability. Figures 3 and 4 are the same as were presented in Evans et al. (2013). Work by Dien (2010) supports that temporal-spatial PCA is optimal for identifying and parsing the variance associated with specific ERP components.

Figure 2.

ERP waveforms of averaged trials by trial type at midline electrode sites. Black = nicotine condition, gray = placebo condition. Dotted, solid, and dashed lines correspond to standard, target, and distracter trial types, respectively.

Figure 3.

Temporal factor loadings. Time is from 0 to 1000 ms.

Black line is temporal factor 2 loadings indicative of P3b amplitude, and gray line is temporal factor 3 loadings indicative of P3a amplitude. Reprinted from Evans et al. (2013).

Figure 4.

Spatial factor loadings. Reprinted from Evans et al. (2013).

Statistical analyses

Repeated measures analysis of variance (ANOVA) was first employed to demonstrate that the P3b and P3a effects reported by Evans et al. (2013) were similar among this smaller subsample. Trial type included standards versus targets for P3b, and standards versus distracters for P3a. Repeated measures ANOVA was also used to test the current hypotheses, which focused on genotype x nicotine deprivation (i.e., cigarette type) x trial type interactions, and also genotype x nicotine deprivation x trial type x P3-type (P3b vs. P3a) interactions, with separate tests for each SNP. The inclusion of P3-type in the second interaction allows for distinguishing the unique influence of genetic moderation on P3b and P3a. That is, full support for nicotine deprivation-induced reductions in P3b and P3a may or may not show unique effects for the P3 components. Genotypes were dichotomized into minor allele carriers versus noncarriers (major homozygote), thereby maximizing the number of participants per genotype group and strengthening the validity of tests. Age and nicotine dependence (i.e., FTND) scores were included as covariates for all between subject genetic analyses, as these variables may spuriously be associated with genotype differences in ERP components. Simpler effects analyses were employed to explore significant associations between genetic variants and P3b/P3a amplitudes. To reduce the probability of type 1 errors, Bonferonni correction for multiple tests were applied to the omnibus tests for each SNP.

Results

Smoking/nicotine deprivation and withdrawal

Mean breath carbon monoxide (CO) was 30.67 (13.70) parts per million (ppm) at the baseline session. Relative to baseline, there were highly robust and significant reductions in carbon monoxide (CO) following overnight deprivation at the start of both the nicotine deprivation and nicotine satiation sessions: t’s (71) = 15.13 and 14.96, p’s < .0001, respectively. There were no CO differences at the start of nicotine deprivation (mean CO = 10.42 ppm, SD = 4.56) and nicotine satiation (mean CO = 10.24, SD = 4.71) sessions.

Across both sessions, there was a main effect for time on self-reported smoking withdrawal (WSWS), with pre-smoking (start of session) ratings (deprivation: mean = 1.84 [SD = .73], satiation: mean = 1.82 [SD = .74]) significantly higher than post-smoking (end of session) ratings (deprivation: mean = 1.32 [SD = .64], satiation: mean = 1.21 [SD = .57]), F(1,71) = 54.86, p < .001. However, consistent with the parent study, there was no significant nicotine deprivation (nicotine vs. placebo) x time effect on this measure (F[1,71] = 1.08, p = 0.30), suggesting that smoking behavior regardless of nicotine content was sufficient to reduce self-reported withdrawal (see Perkins et al., 2010). Similarly, there were no significant nicotine effects observed for any of the WSWS subscales administered (i.e., craving, concentration, and distress). Further, none of the genotypes associated with the five SNPs moderated the effect of nicotine deprivation on the overall WSWS or subscale scores.

Oddball Task Behavioral Data

Overall task accuracy was close to 100%, with slightly greater accuracy during the nicotine satiation (99.25%) versus deprivation (98.73%) session, t(71) = 3.58, p < .001. Reaction time to accurate trials was significantly reduced in the satiated condition (436 ms satiated vs. 448 ms deprived), t(71) = 3.51, p < .001. None of the five SNPs moderated the effects of nicotine deprivation on reaction time or accuracy.

P3b/P3a

All analyses of P3 were tested in the metric of temporal-spatial factor scores that were noted above as indicative of P3b and P3a amplitudes. As expected, target trials evoked greater P3b amplitude than standard trials, F(1,71) = 118.56, p < 0.001. The nicotine deprivation x trial type interaction was significant ([F1,71] = 11.94, p < 0.001, partial eta² = .144), with the deprived condition eliciting reduced P3b amplitude. As expected, distracter trials evoked greater P3a amplitude than standard trials, F(1,71) = 591.18, p < 0.001. The nicotine deprivation x trial type interaction approached significance (F[1,71] = 2.86, p = 0.095), with the deprived condition eliciting reduced P3a amplitude. Including age as a covariate allowed for examining the moderating influence of age on P3s. The age x trial type interactions were significant for both P3b (F[1,70] = 17.15, p < 0.001) and P3a (F[1,70] = 15.29, p < 0.001), with increased age resulting in reduced P3b and P3a amplitudes, thereby further supporting the inclusion of age as a covariate.

The hypothesized genotype x nicotine deprivation x trial type x P3-type interaction was not significant at rs16969968 (F[1,68] = 0.08, p = 0.78), nor was there a significant effect for the non-coding SNP rs637137 (F[1,64] = 0.08, p = 0.78). However, three of the non-coding SNPs showed significant genotype x nicotine deprivation x trial type x P3-type interactions: rs588765 (F[1,68] = 9.30, p = 0.003), rs17408276 (F[1,67] = 11.57, p = 0.001), and rs3829787 (F[1,64] = 4.88, p = 0.03), with the effects for rs588765 and rs17408276 remaining significant after Bonferonni correction.

The first level of follow up analyses on significant effects involved breaking down the analyses into separate predictions of P3b and P3a. None of the three non-coding SNPs (i.e., including follow-up analysis on the SNP that did not pass Bonferonni correction) showed significant genotype x nicotine deprivation x trial type interactions in predicting P3b amplitude: rs3829787 F[1,64] = 0.40, p = 0.53), rs588765 (F[1,68] = 2.08, p = 0.15), and rs17408276 (F[1,67] = 3.36, p = 0.07). However, effects were observed for these SNPs with respect to predicting P3a amplitude: rs588765 (F[1,68] = 7.74, p = 0.007), rs17408276 (F[1,67] = 7.34, p = 0.009), and rs3829787 (F[1,64] = 7.22, p = 0.009), albeit the latter SNP did not pass Bonferonni correction. The next level of follow-up testing at each of the SNPs involved examining nicotine x trial type interactions separately by minor allele carriers versus noncarriers. Figure 5 shows the raw waveform for responses to the distracter stimuli across nicotine and placebo sessions separately for each of the genotypes. Note that for each SNP, the nicotine deprivation condition shows reduced P3a compared to nicotine satiation in the raw waveform for the minor allele carriers, whereas this is not the case with non-carriers. Consistent with the data observed in the raw waveforms, Figure 6 shows mean factor scores indicative of P3a amplitude (based on distracters minus standards) for minor allele carriers and non-carriers for each of these SNPs. There was no effect of nicotine deprivation among homozygotes of the major allele, but significant nicotine deprivation effects for minor allele carriers at each SNP: rs588765 F(1,50) = 7.04, p = 0.011, rs17408276 F(1,44) = 5.96, p = 0.019, and rs3829787 F(1,45) = 6.92, p = 0.012. Associations between these three SNPs were positive, with an R² among bivariate comparisons ranging from .69 to .79, indicating substantive LD, whereas associations between rs16969968 and these three SNPs were negative, with R² ranging from 0.13 to 0.37 (minimal LD).

Figure 5.

ERP waveforms of averaged trials in response to P3a-evoking checkerboard stimulus at site Cz (vertex) for each of the SNPs that significantly predicted P3a amplitude. Black = nicotine condition, gray = placebo condition. Solid line corresponds to minor allele carrier and dotted line to major homozygote.

Figure 6.

Temporal-spatial factor scores indicative of P3a amplitude for minor allele carriers versus major homozygotes across nicotine satiation and deprivation conditions at SNPs rs58876, rs17408276, and rs3829787. * = significant difference (p < .05) between paired bars.

Discussion

This study examined CHRNA5 variants as moderators of nicotine deprivation effects on ERP indices of cognitive control. Contrary to our hypothesis, genotype at rs16969968 did not moderate the effects of nicotine deprivation on either P3b or P3a amplitude. However, substantially correlated non-coding SNPs rs588765 and rs17408276 moderated the effect of nicotine deprivation on P3a. rs3829787 also moderated this effect, but this latter effect was not significant after Bonferonni correction. Specifically, for each of these SNPs, carriers of a minor allele produced reductions in P3a amplitude when deprived of nicotine, whereas this effect was absent among smokers homozygous for the major allele.

In contrast with the rs16969968 variation, the functional effects of these non-coding polymorphisms are not as well characterized. However, recent evidence suggests that rs588765 is predictive of CHRNA5 expression. In a study of 43 elderly European Americans, individuals homozygous for the minor allele expressed more than twice the level of α5 subunit mRNA in frontal cortex as those homozygous for the major allele (Wang et al., 2009).

A discussion of a joint analyses of rs588765 and rs16969968 in predicting light (10 or fewer cigarettes per day) versus heavy smoking (greater than 20 cigarettes per day) may shed some light on interpreting the direction of the effect regarding the minor allele at rs588765 (and/or the correlated SNP effects) predicting greater nicotine deprivation-induced reduction of P3a amplitude. Saccone et al., (2010) conducted a meta-analysis of over 24,000 smokers showing that the minor allele at rs588765 predicts light smoking and could therefore be regarded as protective against smoking quantity (OR = .93). However, after controlling for rs16969968 in a joint analysis, the minor allele at rs588765 robustly predicted heavy smoking (OR = 1.15). This reversal in the direction of predicting heavy smoking even reached genome wide significance. Analysis of rs588765 in isolation leads to the vulnerability effect of the minor allele being masked by the negative association (r = −.64) between rs588765 and rs16969968 risk alleles. Ultimately, this masking effect is driven by rs16969968 having a substantially larger effect on smoking (OR = 1.33; Saccone et al., 2010) than rs588765. In the absence of controlling for the larger effect from rs16969968, possessing the risk minor allele at rs588765 is also predictive of having the protective allele at rs16969968, thus masking (and even reversing) risk allele status at rs588765.

The suppressant effect of rs16969968 on the minor allele at rs588765 in predicting smoking intensity may be translatable to nicotine deprivation effects on cognitive control as well. That is, rs16969968 might be expected to suppress the minor allele effect at rs588765 as a predictor of P3a. However, rs16969968 had no effect on P3a amplitude in the current data. Thus, an rs16969968 suppressant effect on rs588765 can be ruled out in this data set. In the absence of an rs16969968 suppressant effect, the minor allele at rs588765 would be expected to be a vulnerability allele, and would thus also be expected to predict greater nicotine deprivation-induced reduction in P3a amplitude, as was the case. Nevertheless, as this is a small sample and a post hoc finding that was not originally considered, we duly note that our interpretation is speculative and in need of replication.

In addition to rs588765, SNP rs17408276 also moderated nicotine deprivation-induced reductions in P3a amplitude, as was also the case for rs3829787 before Bonferonni correction. These three SNPs were highly correlated, and the effect may therefore be driven by any single, or any combination, of these or other related SNPs. rs588765 was chosen to represent locus 3 in CHRNA5 in the Saccone et al. (2010) meta-analysis. This SNP has also been associated with increased mRNA expression in cortex and lungs, thus leading us to focus on this SNP in the current study.

Previous research in our lab has shown that individuals lower in self-reported trait cognitive control exhibited greater nicotine deprivation induced reduction of P3a amplitude, suggesting individual variability in nicotine deprivation-induced reductions in cognitive control that is consistent with nicotine self-medication of cognitive control (Evans et al., 2013; see Evans & Drobes, 2009). In the absence of an rs16969968 suppressant effect, the current findings show rs588765 to predict greater nicotine deprivation-induced reductions in P3a amplitude which may also be interpreted as somewhat consistent with a nicotine self-medication of cognitive control effect. In contrast, these SNPs did not moderate nicotine deprivation influences on P3b amplitude. This pattern of findings may be preliminarily interpreted as suggesting that the P3a response evoked by vivid distracting stimulus might be more greatly influenced by individual differences than the P3b response associated with the identification of a priori targeted stimuli. Specifically, the P3a is relevant to the flexible orienting of attention (Polich & Criado, 2006). For example, one may be occupied by the task of reading a book, but the smell of smoke irrelevant to the current task may be adaptive and alert the individual to potential danger. At the more mundane level of performing tasks throughout the day, the capacity to flexibly shift attention to unanticipated stimuli is important to the ongoing updating of task priorities amid a changing environment. Capturing specific processes influenced by nicotine that are also moderated by individual differences (e.g., processes underlying generation of the P3a) may assist in identifying subgroups of individuals who find smoking more reinforcing for cognitive reasons. Tailored treatment programs may then seek to match the needs of individuals who are more reinforced by smoking for cognitive reasons (e.g., cognitive training and/or pharmacotherapy of cognitive control).

We did not find evidence that any of the SNPs predicted smoking/nicotine dependence or self-reported withdrawal. This is in contrast with a large scale meta-analysis of GWAS studies (e.g., Saccone et al., 2010), that demonstrated that the minor alleles in both rs16969968 and rs588765 predicted greater nicotine dependence and cigarettes smoked per day. It should be noted that the goal of the present study was to examine neurocognitive substrates of nicotine deprivation as potential intermediate phenotypes (or endophenotypes) for smoking. The potential benefit of this approach is that the link between variants on a specific gene may have more proximal associations with a specific biologically plausible endophenotype than a very general phenotype such as cigarettes per day (see Lerman, Perkins, & Gould, 2009.). Interestingly, as reported in the parent study (Evans et al., 2013), smoking, but not nicotine content, predicted self-reported withdrawal, whereas nicotine deprivation was highly predictive of reduced P3b among all participants and P3a among individuals lower in trait cognitive control. These findings suggest that P3 amplitudes may provide more sensitive indices of nicotine deprivation, relative to self-report. In terms of the clinical relevance of P3 among smokers, larger scale clinical outcome studies are needed to determine if P3 indices of nicotine deprivation predict smoking cessation.

Limitations and Future Directions

In this paper, we have shown that presence of a minor allele at rs588765, and two other highly correlated SNPs, predict nicotine deprivation-induced reductions of P3a amplitude. However, a number of limitations deserve attention. First, we did not make a priori predictions for the non-coding SNPs that showed significant effects. Rather, our post hoc interpretation of the observed interaction between CHRNA5 variants and nicotine deprivation was guided by Saccone et al.’s (2010) prior findings regarding rs588765. In addition, the well characterized non-synonymous coding SNP rs16969968, which was hypothesized to moderate the influence of nicotine deprivation on both P3b and P3a, did not show any significant effects. The above limitations could be addressed by conducting a larger study that includes a priori predictions for all CHRNA5 SNPS.

Another limitation of the current design was that we divided the sample into minor allele carriers versus major allele homozygotes, a genotyping approach that implies a dominant model. The motivation for this strategy was to maximize the number of participants per genotype group, thereby strengthening the power of statistical tests given the limited sample size. Future studies should be sufficiently powered to examine all three genotypes (major homozygote, heterozygote, and minor homozygote) for each SNP. Larger scale studies should also facilitate examination of CHRNA5 haplotypes as predictors of nicotine deprivation-induced reductions in cognitive control. These CHRNA5 haplotypes may then be used to predict treatment effects on nicotine deprivation-induced reductions in cognitive control. For example, it may be that individuals who possess CHRNA5 haplotypes associated with cognitive vulnerability in the context of nicotine withdrawal may benefit from agonist medications that are specific to nicotinic receptor subtypes.