Constitutional mechanisms of vulnerability and resilience to nicotine dependence

Abstract

The core nature of nicotine dependence is evident in wide variations in how individuals become and remain smokers. Individuals with pre-existing behavioral traits are more likely to develop nicotine dependence and experience difficulty when attempting to quit. Many molecular factors likely contribute to individual variations in the development of nicotine dependence and behavioral traits in complex manners. However, the identification of such molecules has been hampered by the phenotypic complexity of nicotine dependence and the complex ways molecules affect elements of nicotine dependence. We hypothesize that nicotine dependence is, in part, a result of interactions between nicotine and pre-existing behavioral traits. This perspective suggests that the identification of the molecular bases of such pre-existing behavioral traits will contribute to the development of effective methods for reducing smoking dependence and for helping smokers to quit.

Introduction

According to a recent estimate, 1.3 billion individuals or one-fifth of the global population smokes. Tobacco smoking poses a grave health hazard, with half of the continuing smokers being expected to die prematurely due to tobacco-related diseases.¹ Cigarette smoking is one of the most persistent types of substance dependence, comparable to cocaine dependence. A survey of 10 343 daily smokers and 107 daily cocaine users aged 12 years or older revealed that % of smokers and 72% of daily cocaine users try to cut down on their use. However, 80% of smokers and 66% of daily cocaine users feel unable to cut down on their use, despite their efforts. The majority of daily smokers (85%) and daily cocaine users (63%) feel they are dependent, and 37% of daily smokers and 49% of daily cocaine users feel sick when they stop or cut down on the use of the drug.²

Antidepressants and some forms of nicotine replacement therapy increase tobacco cessation rates.^3,4 However, the majority of quitters relapse, despite the use of these aids or even a combination of these aids. More effective tobacco cessation treatments depend upon a more complete understanding of the mechanisms underlying dependence. Although chemicals other than nicotine might contribute to the development of continued smoking,⁵ evidence suggests that nicotine is a major determinant of dependence. When switched to denicotinized cigarettes, smokers consume fewer denicotinized cigarettes within 1 week.⁶ Smoking denicotinized cigarettes causes smokers to experience withdrawal symptoms.⁷ Conversely, nicotine replacement therapy reduces withdrawal symptoms and the urge to smoke in smokers during abstinence.⁸ Nicotine replacement also increases the success of cessation efforts.⁴ Varenicline, a partial nicotinic receptor agonist,⁹ increases the success rates of long-term smoking cessation efforts.¹⁰ Nicotine alone is sufficient to sustain self-administration behavior in humans.

In this review, we discuss the difficulty in defining nicotine dependence, individual variation in smoking and pre-existing traits for smoking. We also discuss potential molecular correlates of individual susceptibility to nicotine dependence in humans and experimental animals. We propose a hypothetical framework to better understand nicotine dependence and discuss its implications for future research directions.

Many definitions of nicotine dependence

The clinical definition of nicotine dependence depends on the measure that is used. The Fagerström Test for Nicotine Dependence (FTND) and the Diagnostic and Statistical Manual (DSM)-III or DSM-IV-Revised are commonly used for studies of nicotine dependence. Both measures define nicotine dependence in terms of behavioral and clinical manifestations. The FTND is a revised version of the earlier Fagerström Tolerance Questionnaire. As the name indicates, this scale assumes that physical dependence, including withdrawal and tolerance, motivates compulsive smoking. It assesses the degree of dependence on a continuous scale.¹²

The DSM focuses on salient behavioral and physiological features of smoking. These features include (1) impaired control over use, (2) use greater than intended, (3) withdrawal, (4) use despite harm and problems and (5) tolerance. DSM criteria not readily applicable to nicotine dependence are (6) large amounts of time spent obtaining, using and recovering from the substance and (7) forgoing other activities to use the substance.¹³ Each item of the DSM is judged to be present or absent, and overall dependence is categorically judged to be present or absent if a smoker meets three or more of these criteria at any time during a 12-month period. This diagnosis is useful for reliably determining the prevalence of smoking. However, a shortcoming of the DSM-based diagnosis is that it does not take into account the gradual development of smoking. Even when a smoker is nearing the threshold of a diagnostic cutoff, he would be judged as non-dependent. Moreover, a single diagnosis does not reveal the combinations of the multidimensional symptoms that are present. At different stages of smoking, smokers might exhibit distinct sets of characteristics or more of one particular symptom. Grouping smokers into a single ‘dependent’ group ignores the potential heterogeneity among these individuals.

Not surprisingly, the FTND and DSM do not correlate well. The FTND defines smoking on a continuous dimension and has no clear-cut point that divides dependent and non-dependent individuals. Cutoffs can be arbitrarily set to divide non-dependent and dependent smokers or to define low, medium or high levels of dependence. However, no matter how a cutoff is set, the FTND-defined nicotine dependence does not agree well with nicotine dependence defined by the DSM-III-R-based scale.¹⁴ The FTND is a better predictor of smoking cessation than the DSM-III-R.^15,16

Whether nicotine dependence should be categorically defined as a state or a continuous process and whether nicotine dependence involves a single process or multiple processes are matters of debate.¹⁷ The Wisconsin Inventory of Smoking Dependence Motives (WISDM-68) was developed to comprehensively capture the multidimensional reasons for smoking on a continuous scale.¹⁸ This inventory includes the following potential motives for smoking: a strong emotional attachment to smoking and cigarettes (that is, affiliative attachment); smoking without awareness or intention (that is, automaticity); smoking among all options despite constraints or negative consequences (that is, behavioral choice-melioration); smoking to enhance cognitive functioning (that is, cognitive enhancement); frequent urge, inability to ignore urge to smoke and intensified and intolerable nature of urge to smoke during abstinence (that is, craving); strong impact of smoking-associated cues on urge to smoke and smoking (that is, cue exposure-associative processes); a sense of loss of volitional control over smoking (that is, loss of control); the ability of smoking to improve mood, lift a ‘down’ mood, relieve irritability and increase the ability to cope with stress (that is, negative reinforcement); the desire to smoke to experience pleasure or to enhance an already positive feeling or experience (that is, positive reinforcement); social stimuli or contexts that either model or invite smoking (that is, social-environmental goads); the need to experience the sensory and gustatory effects of smoking (that is, taste and sensory properties); the need to smoke increasing amounts over time or the ability to smoke large amounts without acute toxicity (that is, tolerance); and use of cigarettes to control body weight or appetite (that is, weight control).

This test is likely to uncover distinct underlying motives for various aspects of smoking. Variation in how much one smokes is correlated with tolerance, loss of control, craving, automaticity, behavioral choice-melioration, affiliative attachment, cognitive enhancement, cue exposure-associative processes and negative reinforcement; the number of cigarettes smoked is less correlated with positive reinforcement, weight control, taste and sensory properties and social-environmental goads.¹⁸ Relapse also can be predicted by some of these motives. Smokers subjectively feel that they relapse because they automatically smoke without thinking about it (that is, automaticity), they smoke to enhance cognitive function (that is, cognitive enhancement), they smoke to alleviate stress or withdrawal (that is, negative reinforcement) and their social environments are conducive to smoking (that is, social-environmental goads).¹⁸ In a follow-up study, tolerance, automaticity, social-environmental goads and craving were found to be predictive of abstinence.¹⁶

Many behavioral and physiological features are potentially related to the mechanisms underlying nicotine dependence. The symptomatic complexity challenges the assumption that a single mechanism governs the many aspects of nicotine dependence. This raises the question as to how to model ‘dependence’ as a whole and whether such an attempt is a reasonable endeavor. In addition, imposing an artificial division between dependent and non-dependent individuals is conceptually limiting. Smoking is more appropriately characterized by the degree of dependence on a continuous, multidimensional scale.

Myth of the average smoker

The interindividual variation in nicotine intake is large. Among smokers aged 18 or older in the United States, the majority (64%) consume between 5 and 24 cigarettes per day. A sizable population of smokers (23%) smoke only on some days or smoke 1–4 cigarettes per day. Heavy smokers who smoke at least 25 cigarettes per day make up 12% of the total population of smokers.¹⁹

These different levels of smoking represent a cross-section of how individuals’ smoking habits change over a lifetime. Most smokers initiate smoking during adolescence, and their smoking habits change through distinct trajectories into adulthood. Although there is inconsistency among studies regarding how many trajectories exist, prospective studies have identified four primary trajectories: (1) persistent low-level smoking, (2) early initiation of smoking with rapid progression to heavy, persistent smoking, (3) early initiation of smoking with slow progression to moderate smoking and eventual cessation, (4) later initiation of smoking with progression to moderate to heavy smoking.^20–36 Those who become persistent, heavy smokers comprise approximately one-third of those who have ever tried cigarettes.

Like adolescent smokers, adult smokers show various trajectories. Among college students examined over a course of 4 years, light smokers quit (45%), became occasional (35%) or daily smokers (20%). The majority of those who were daily smokers at baseline quit or reduced smoking (54%). The remaining individuals became low-level smokers (6%) or increased their daily smoking (27%).³⁷ Among adults over 24 years old who have smoked for at least 5 years, 37% of daily low-level smokers (⩽five cigarettes per day) quit, 36% retain the same level of smoking, 21% increase their smoking to more than five cigarettes per day and 6% become occasional smokers over a period of 2 years.³⁸

Light or intermittent smokers are often in transition to either quitting or becoming regular smokers. However, a subpopulation of intermittent and light smokers remains at this level for more than 20 years without changing their smoking status.³⁹ Even after smoking tens of thousands of cigarettes over 20 years, these smokers, dubbed ‘chippers’, do not experience craving, mood disturbance, reduced arousal, or sleep disturbance.⁴⁰ Chippers feel the urge to smoke only in settings where they usually smoke, such as in settings related to relaxation, socializing, eating and drinking alcohol; they do not feel the urge to smoke or experience positive feelings upon smoking outside of their usual smoking settings.^41–43

The intensity and time course of withdrawal symptoms following quitting differs from individual to individual. Some individuals experience intense withdrawal immediately following the onset of abstinence, with the symptoms gradually or quickly waning. Others experience a rebound in withdrawal symptoms later.^44,45 Lapses are highly correlated with the individual time course of withdrawal intensity,⁴⁶ and lapsers have higher levels of withdrawal symptoms (for example, negative affect) than abstainers.⁴⁴

There is also heterogeneity in nicotine dependence among smokers. During adolescence, different smoking trajectories are correlated with different levels of nicotine dependence. Adolescents whose smoking rates rapidly escalate tend to develop stronger nicotine dependence, with 95% becoming dependent (as defined by the International Statistical Classification of Diseases and Related Health Problems or ICD). In contrast, only 12% of non-progressing, low-intensity adolescent smokers become nicotine dependent, whereas 79% of both slow and moderate smoking escalators become dependent.³⁰ Similarly, by early adulthood, 51% of early stable smokers meet the FTND-defined criteria for nicotine dependence, compared to 26% of late stable smokers and 5% of low-level smokers.⁴⁷ However, individual smokers with different trajectories during adolescence might eventually reach similar levels of nicotine dependence by the time they reach adulthood.³⁵

Individual smokers markedly differ in the trajectory of smoking, rate of smoking, relapse timing, withdrawal and nicotine dependence. The mechanisms of nicotine dependence need to explain the reasons for this variation.

Pre-existing conditions for smoking progression

The individual variation in nicotine dependence raises the question as to what may be unique about those who are more vulnerable to nicotine dependence. One way of viewing this variation is that it simply reflects the amount of nicotine exposure. That is, the more one is exposed to nicotine, the more likely they will develop dependence. According to this explanation, any person who is exposed to a sufficient amount of nicotine will develop dependence. However, this view cannot adequately explain why chippers maintain smoking without developing dependence and why only one-third of novice light smokers increase their level of smoking.

Evidence suggests that those susceptible to heavy smoking are not simply individuals who happened to be exposed to large amounts of nicotine. Longitudinal studies have shown that individuals with higher levels of novelty/sensation seeking tend to have a greater susceptibility to smoking and nicotine dependence.^33,48–55 Novelty seeking is ‘a heritable tendency toward intense exhilaration or excitement in response to novel stimuli or cues for potential rewards or potential relief of punishment, leading to frequent exploratory activity in pursuit of potential rewards’.⁵⁶ Sensation seeking is ‘a trait by which an individual seeks varied, novel, complex and intense sensations and experiences and is willing to take physical, social, legal and financial risks for the sake of such experiences’.⁵⁷ These two scales are highly correlated.⁵⁸ If a cigarette is perceived as a novel object during adolescence, then novelty/sensation seeking could easily manifest itself as initiation of smoking. However, because novelty seeking might not necessarily be a predictor for sustained smoking,⁵⁵ this behavioral trait may not influence all aspects of nicotine dependence.

Novelty seeking predicts the onset of disruptive behaviors, such as oppositional defiant disorder and conduct disorder. These behavioral traits also are associated with susceptibility to nicotine dependence.⁵² The presence of conduct disorder in childhood and adolescence is a strong predictor of daily smoking and DSM-defined nicotine dependence.^55,59–61 Such traits are likely to contribute to poor choices despite negative future consequences (that is, behavioral choice-melioration).¹⁸

Novelty seeking also predicts the future incidence of attention-deficit hyperactivity disorder (ADHD).⁵² Longitudinal studies have shown that ADHD predicts progression to regular smoking and nicotine dependence.^59,62–67 ADHD symptoms of inattention^68,69 and hyperactivity/impulsivity⁶⁰ are associated with smoking and nicotine dependence, although these two classes of symptoms might contribute to nicotine dependence during different developmental periods.⁷⁰ Because nicotine improves attention in smokers with ADHD⁷¹ and cognitive/behavioral inhibition in non-smoking adolescents with ADHD,^72–76 negative reinforcement might underlie nicotine dependence in smokers with ADHD.⁶⁸

Pre-existing conditions that may lead a person to smoke not only include motivational and behavioral traits, but also include more pathological conditions. Although smoking withdrawal might cause anxiety and depression-like symptoms, the opposite can also be true. The presence of depressive disorders and some forms of anxiety disorders may predict the likelihood that a person will initiate smoking and develop dependence.^77,78 However, it is unclear if there is a causal relationship between depressed mood or anxiety and various aspects of nicotine dependence.⁷⁹

Genetic origin of variation in nicotine dependence and pre-existing traits

Multiple factors likely contribute to individual variation in pre-existing behavioral traits and susceptibility and resilience to nicotine dependence. Environmental factors are thought to contribute to variation in nicotine dependence. However, the precise identification of the environmental factors is difficult, and such factors might not exert as robust an influence as genetic factors.⁸⁰ Twin studies have consistently shown that heritable factors account for a substantial amount of the variance in susceptibility to smoking initiation, quantity, persistence, regular use, dependence and cessation.^81–87 However, heritability estimates vary depending upon the particular aspect of smoking and the definition of dependence. Different genetic factors appear to affect smoking initiation, regular use and nicotine dependence (as defined by the FTND) as well as DSM-IV-defined nicotine withdrawal.^80,88

Although it is clear that there is a genetic influence on nicotine dependence, the exact genes that contribute to dependence have not been clearly delineated. Genome-wide linkage analyses have implicated almost all chromosomes in various aspects of smoking. However, the involvement of each locus has not been consistently demonstrated. Several factors are thought to contribute to this problem. First, studies have used different criteria, including the Fagerström Tolerance Questionnaire, number of cigarettes smoked per day, maximum number of cigarettes smoked in any 24-h period, age at which the first cigarette was smoked, FTND, cessation, frequency, withdrawal severity and DSM. These different measures have led to the identification of linkages to different chromosomes within the same population sample. For example, linkage was found on chromosome 5 when smoking quantity was used as a criterion, chromosome 6 with FTND, chromosome 7 with a DSM-based continuous scale, chromosome 8 with a dichotomous DSM classification, chromosome 16 with cessation and chromosome 19 with smoking frequency.⁸⁹ Second, different population bases show different linkage loci even when the same measures are used. Linkage with FTND scales are located on chromosomes 9, 10, 11 and 13 in an African-American sample but on chromosomes 4 and 9 in a sample from Americans of European origin.^90,91

A genome-wide association study found that alleles in the neurexin 1 and the β3 nicotinic receptor genes may be associated with FTND-defined nicotine dependence.⁹² However, another genome-wide association study, examining far more SNPs, identified a different set of genes that were associated with both nicotine and other substance dependence.⁹³ By contrast, a number of genes have been found to be consistently associated with smoking cessation in three different sample populations.⁹⁴ Some of these genes encode proteins that are involved in cell adhesion and cell signaling.

Association studies examining individual candidate genes have uncovered variants of many monoamine-related genes that might contribute to smoking initiation, persistence, cessation, consumption and therapeutic response (see refs. 95–99 for reviews).

Associations between specific genes and variation in behavioral traits have also been examined. Polymorphisms in the dopamine D4 receptor have been implicated in novelty/sensation seeking, and polymorphisms in the serotonin transporter have been linked to anxiety-related traits.¹⁰⁰ Although some studies do not support the association between D4 receptor alleles and novelty/sensation seeking, the association is more consistent in subjects under 35 years old.¹⁰¹ Like genes associated with nicotine dependence, variants in monoamine-related genes have been reported to be associated with susceptibility to ADHD. These genes include the dopamine transporter, dopamine D4 receptor, tryptophan hydroxylase 2, phenylethanolamine N-methyl-transferase, adrenergic β2 receptor and monoamine oxidase A (MAOA).¹⁰² The serotonin receptor 2A, adrenergic α2A receptor, adrenergic α2C receptor, phenylethanolamine N-methyltransferase, catechol-O-methyltransferase, serotonin transporter and MAOA have been reported to be associated with conduct disorder, oppositional defiant disorder or antisocial behavior.^103–105

In some cases, human association studies are difficult to replicate. Many reasons are likely to contribute to this problem, but the following are some of the potential reasons relevant to the gene side of association. First, the impact of each gene on nicotine dependence or novelty/sensation seeking is thought to be small. A meta-analysis could potentially reveal an overall trend in a large number of studies but might overestimate the impact of a gene allele on smoking and associated behavioral traits. This is because meta-analyses are affected by publication bias; more studies with positive associations are published than those with no association and more significant measures are reported than those that are less or not significant even within individual studies.¹⁰⁶ Second, reported gene alleles might not be the genuine source of phenotypic variation. Rather, allelic association with a phenotype could reflect the impact of other gene alleles that are linked to the reported gene alleles. Third, the phenotypic expression of a single gene allele may be influenced by alleles in other genes that differ in different ethnic groups.

A problem at the phenotypic side of correlation is that the scales used to categorize phenotypes might not match genetic influences perfectly. For example, categorizing nicotine dependence and pre-existing behavioral traits using self-reported questionnaires may not fit well with the way single genes influence behavior. If a specific gene affects only a specific element/aspect of nicotine dependence or pre-existing behavioral traits, the overall association between gene polymorphisms and traits could fail to reach significance. Another complicating factor is that variance of each motive for smoking is large and thus each smoker might have a unique set of motives for smoking.¹⁸ Moreover, various subscales of novelty/sensation seeking do not necessarily correlate with each other, and thus what is globally defined as novelty/sensation seeking might contain more than one psychological process.¹⁰⁷ In mice, behaviors in various behavioral tasks that involve novel elements do not necessarily co-vary with genetic background, suggesting a complex influence of genetic factors on behaviors in response to novel stimuli.^108,109

Different elements of nicotine dependence and behavioral traits might have non-identical, albeit partially overlapping, genetic bases. The reproducibility of human association studies might be improved if they focused on associations between single genes and distinct elements of nicotine dependence and behavioral traits.

Phenotypic refinement: translational models of nicotine dependence

Experimental animals provide a complementary approach to precisely characterize genetic mechanisms of nicotine dependence independently of the subjective assessment of smokers. The ultimate value of rodent models lies in their ability to predict nicotine dependence in humans, but the translation of rodent data into human studies and vice versa must be done cautiously. Caution must be used when translating animal data to humans, particularly in terms of highly subjective processes such as craving, urge and pleasure, as there are no unequivocal, objective ways to measure these psychological processes in rodents.

Another limitation of rodent models of nicotine dependence is that they are not designed to mimic the entire process of nicotine dependence. Ironically, this limitation may be advantageous in identifying behavioral elements that might be affected by molecules. Models that focus on specific, distinct aspects of nicotine dependence are advantageous when testing underlying mechanisms. A single molecule may not contribute equally to all elements of nicotine dependence, and distinct molecules may affect different elements of nicotine dependence. Feature-specific models are less likely to be contaminated by mechanistically heterogeneous processes and thus will allow us to tease apart distinct mechanisms.

Three paradigms are widely used to determine the role of molecules in nicotine dependence using knockout (KO) mice: self-administration, conditioned place preference and withdrawal. Because mice cannot reliably or robustly establish intravenous self-administration, the oral route has been used to evaluate how much nicotine a mouse consumes. Nicotine consumption has a clear counterpart in smokers. Each smoker has a unique optimal blood nicotine concentration.^110–112 The amount of nicotine taken in during smoking is the best predictor of relapse in humans.¹⁵ ‘Tolerance’, as defined in the WISDM-68 inventory, essentially characterizes how heavily one consumes nicotine.¹⁸

Cue reactivity or cue exposure-associative processes have been examined using the place-conditioning paradigm in mice. This Pavlovian conditioning procedure assesses an animal’s approach, termed conditioned place preference (CPP), toward sensory cues that are associated with a substance that has dependence potential.^113–115 Cue reactivity is a prominent element of nicotine dependence in humans. Non-nicotine cues that are associated with smoking (for example, smoking paraphernalia and smoking-association sensory cues and context) contribute to relapse in ex-smokers and to persistent smoking in smokers.^116,117 Cue reactivity correlates with the amount of smoking during early, developing stages of nicotine dependence, as well as in late stages of dependence.¹⁸ Cue reactivity is established through Pavlovian conditioning.¹¹⁸ When neutral cues are paired with smoking under Pavlovian experimental conditions, new sensory cues acquire the ability to induce urge within only a few trials.¹¹⁹ Individuals differ in the way they react to these cues during relapses.¹²⁰

In humans, withdrawal symptoms of nicotine dependence are divided into two major classes: somatic and affective withdrawal. During abstinence, smokers experience somatic symptoms, including insomnia, difficulty concentrating, restlessness, decreased heart rate, increased appetite and weight gain.¹²¹ Affective withdrawal includes feelings of depression, heightened anxiety, increased irritability or anger and frustration.^122,123 Smokers report that avoidance of withdrawal is one of the subjective motives for heavy smoking and relapse (that is, negative reinforcement).¹⁸ Because a reliable method to demonstrate affective withdrawal signs has not been developed in mice until recently,^124,125 somatic signs have been used widely to evaluate withdrawal in mice. Somatic withdrawal symptoms in mice include paw tremor, head shaking, scratching, body tremor, teeth chatter, ptosis, backing and jumping.

Other possible contributors to smoking are difficult to model in experimental animals. For example, it is not feasible to establish that a mouse considers nicotine to be a ‘friend’ (affiliative attachment) or makes decisions in light of future negative consequences (behavioral choice-melioration). The latter is not a simple choice between a drug and other reinforcers but reflects a complex cognitive process that compares the immediate value of smoking with future reinforcers in the presence of other reinforcers and constraints on cigarettes. Determining whether animals feel that they have lost control over their behavior (loss of control) or whether they act without being aware (automaticity) is not feasible. Similarly, although craving may be an underlying motive, it is not possible to unequivocally attribute the cause of behavioral alterations to this subjective feeling in experimental animals.

A systematic comparison between preclinical and clinical studies is difficult for several reasons. Rodent models focus on specific elements of nicotine dependence, whereas human association studies use more global definitions of nicotine dependence. A future challenge is to increase the translationability of animal and human studies by focusing on specific elements of human nicotine dependence, as exemplified in the WISDM-68. For example, examining the association between subjective feelings and activities of specific brain regions during exposure to smoking-related cues in humans with specific genetic alleles would provide a human equivalent of rodent cue reactivity, as measured in CPP and withdrawal-associated conditioned place aversion (CPA). Moreover, a better understanding of the functional consequences of gene alleles (for example, expression and activity levels of proteins) in humans would be essential for comparing mouse KO studies to human studies.

Multiple molecules for distinct aspects of nicotine dependence

Complex behaviors, including dependence-related behaviors, are thought to be influenced by multiple genes. Rodent quantitative trait loci for complex behavioral traits generally suggest that the average percentage of total phenotypic variance that is explained by each sequence variant is 5% or less.¹²⁶ Such a small effect is difficult—if not impossible—to detect.

The constitutive, complete deletion of genes using KO mice provides a unique opportunity to assess the maximum potential impact of a single gene on behavior. Another advantage of constitutive genetic manipulation is that it is expected to induce developmental and compensatory alterations. From the susceptibility perspective, secondary alterations originating from a single gene deletion do not limit interpretation and should not be considered a contaminant, because any molecular alteration would be expected to have some effects on related molecules and developmental processes in the human brains, as well as in the mouse brains.^127–130 Developmental alterations cannot be recapitulated by pharmacological means that are applied to adult or even adolescent rodents.

Evidence suggests that both shared and distinct molecular bases exist for elements of nicotine dependence. Both nicotine CPP and somatic withdrawal are reduced in mice that are defective for preproenkephalin¹³¹ or the μ-opioid receptor.¹³² Other molecules influence some but not all models of nicotine dependence. Genetic deletion of the cannabinoid CB1 receptor or adenosine A2A receptor attenuates nicotine CPP but not somatic withdrawal signs.^133,134 Mice deficient for the nicotinic acetylcholine receptor subtype β2 do not show withdrawal-induced CPA, but are normal in somatic withdrawal signs; α5 subunit KO mice show the opposite pattern of phenotypes.¹²⁴ As the impact of single genes on elements of nicotine dependence would be difficult to detect in a model that incorporates many aspects of dependence symptoms, rodent models that focus on specific elements of nicotine dependence likely are the most effective models for understanding the genetic determinants of nicotine dependence.

Distinct molecular mechanisms might also exist for different phases of exposure to nicotine. Mice deficient for the transcription factor FosB are impaired in behavioral alterations that are induced by repeated or prolonged exposure to nicotine in voluntary nicotine intake, nicotine CPP and nicotine-induced motor suppression, but these mice respond normally to acute single nicotine exposure.¹⁰⁹ Concomitant with the behavioral effects of FosB deletion, FosB proteins increase in the brain of wild-type mice following repeated injections of nicotine, but not after a single, acute nicotine exposure.¹⁰⁹

Gene variation is likely to influence nicotine dependence through many molecular pathways. First, molecules may influence nicotine dependence through direct involvement in signaling cascades activated by nicotine (Figure 1, circles). The nicotinic acetylcholine receptor is one such example (Figure 1, green circle).^{124,135–137} Alternatively, even if they are not part of a cascade that is directly activated by nicotine, other molecules still could influence nicotine dependence either by influencing the molecular cascades activated by nicotine (Figure 1, triangle) or independently of the nicotine-activated cascades (Figure 1, star). Specific examples of these factors are discussed below.

Figure 1.

Possible effects of molecular variation on the development of nicotine dependence. The degree of nicotine dependence may be determined by molecular variation in the cascades that are activated directly (green circle) or indirectly (gray and red circles) by nicotine, by molecular variation that indirectly affects nicotine-regulated molecular cascades (triangle), or by variation in molecules that influence the degree of dependence independently of these cascades (star).

Pleiotropy of nicotine dependence and pre-existing behavioral traits

Why is there an association between nicotine dependence and pre-existing traits? One plausible biological explanation is that some molecules commonly affect both nicotine dependence and pre-existing behavioral traits.¹³⁸ This phenomenon, called pleiotropy, could be one of the underlying mechanisms of nicotine dependence. Although human association studies have not provided genes that clearly exert pleiotropic effects on nicotine dependence and pre-existing traits, animal studies have provided a few examples.

FosB is induced along the mesencephalic dopaminergic systems and other limbic brain regions by nicotine and other addictive substances in rodents (see Figure 1, red circle).^109,139 FosB is also induced in limbic and other brain regions by external stimuli that evoke stress and alter affect in rodents.^140,141 FosB KO mice are impaired in nicotine intake and CPP, and also are abnormal in behavioral traits related to tasks in which stress levels are considered high.¹⁰⁹ Another transcription factor, cAMP-response element binding protein (CREB), has been implicated in behaviors relevant to learning, depression and anxiety, as well as nicotine dependence.^142–144 These transcription factors could have a common influence on elements of nicotine dependence and pre-existing behavioral traits.

The pleiotropic action of these genes raises the question of whether molecules exist that are selectively involved in the development of nicotine dependence. Lack of behavioral selectivity does not necessarily indicate that the dependence phenotype in a KO mouse is a contaminant of some general behavioral abnormality. Instead, such lack of behavioral selectivity could provide a means to explore a causal relationship between pre-existing behavioral traits and substance dependence. In more general terms, it is difficult to assume that the organisms have evolved, during their evolutionary history, to express molecules that are selectively designed to respond to addictive substances.

Indirect influence of molecular variation on nicotine dependence

Constitutive levels of molecules and their activities that are not regulated by nicotine nonetheless could influence smoking and relapses (Figure 1, triangle or star). Nicotine at physiological concentrations does not reliably alter the activity of MAOA, but the constitutive level of MAOA nevertheless might affect smoking in humans and behavioral responses to nicotine in mice.

The activity levels of MAOA vary widely among individuals. In cultured skin fibroblasts, individual responses range from extremely low, almost undetectable, levels to levels 30-fold higher.^145–149 Differences in MAOA activity of up to seven-fold have been noted in postmortem tissues of the human frontal cortex.¹⁴⁹ These individual differences in MAOA activity exist at ages before smoking usually starts.^145,149,150

The genetic origin for variance in MAOA activities is poorly understood. High levels of MAOA activities are associated with either a T allele, compared to a C allele, at position 1460 in exon 14 or with either 3.5 and 4 repeats, compared to other repeat numbers, at the variable number tandem repeat in the promoter region of the human MAOA gene.^{146,151–153} However, the impact of these polymorphisms on MAOA activity is weak and no known single alleles or their haplotypes adequately account for the large inter-individual differences in basal MAOA activity.^{146,149,151–153} A PET study also demonstrated that MAOA alleles are not strongly correlated with MAOA activity in vivo.¹⁵⁴

The weak impact of genetic alleles on MAOA activities might be one of the reasons why association studies of human MAOA and nicotine dependence have not been consistent. Some studies have shown a positive association between the high-activity alleles and an increased risk of nicotine dependence,^155–157 but other studies have failed to confirm this association.^158,159 These studies have many procedural differences, including the gender, age and other characteristics of the sample population and the definition of dependence, which make direct comparisons difficult.

Constitutive Maoa-deficient mice are impaired in nicotine intake and nicotine CPP.¹⁶⁰ This effect cannot be generalized to the isoenzyme MAOB in mice, as constitutive Maob-deficient mice exhibit normal nicotine intake.¹⁶¹ Although species differences should be taken into account in generalizing this finding to humans,¹⁶² constitutively high and low MAOA activities might confer susceptibility and resilience, respectively, to levels of nicotine intake and cue reactivity in smokers.

Although it still is possible that some unidentified alleles within MAOA or in other genes influence MAOA activities, non-genetic, environmental factors also might alter MAOA activities. Tobacco smoke contains chemicals that inhibit MAOA,^163–166 and brain MAOA levels are reduced in smokers.¹⁶⁷ Simultaneous, irreversible inhibition of both MAOA and MAOB by non-selective MAOA/B inhibitors (for example, tranylcypromine or phenelzine) increases nicotine self-administration in rats.^168,169 Clorgyline, an irreversible MAOA inhibitor, also promotes nicotine self-administration in rats.^169,170 These studies in rats suggest that MAO activities have an inhibitory effect on nicotine intake. However, this possibility is inconsistent with the report that the MAOA inhibitor moclobemide does not increase smoking and rather has some subjective beneficial effects on reducing smoking.¹⁷¹ Similarly, none of the MAOB inhibitors tested so far lead to increased smoking; rather, they result in reduced smoking and increased cessation rates.^172–174

These studies with Maoa KO mice and pharmacological inhibition of MAOA in rats should not be compared directly. First, the developmental impact of constitutive deletion of MAOA expression on the behavioral effects of nicotine likely occurs through processes different from acute or semi-chronic pharmacological inhibition of MAOA during adulthood. In fact, the constitutive inactivation of MAOA and the pharmacological inhibition of MAOA by clorgyline induce many different and even opposite effects on various behaviors in mice. For example, acoustic startle response is reduced in Maoa KO mice, compared to wild-type mice, but is either unaffected or slightly potentiated by clorgyline in a dose-dependent manner in mice.¹⁷⁵ Second, many MAO inhibitors lack selectivity and exert a myriad of effects. It is unclear what actions of these drugs influence nicotine-regulated behaviors in rats. The non-selective MAOA/B inhibitor tranylcypromine inhibits CYP2A6, the principle enzyme responsible for metabolizing nicotine into cotinine.¹⁷⁶ Tranylcy-promine and clorgyline inhibit monoamine uptake in various brain regions.^177–180 This action of these inhibitors is likely to contribute to the potentiation of nicotine self-administration, as a dopamine transporter inhibitor has been shown to enhance nicotine-induced behavior.¹⁸¹ Clorgyline also binds to the σ-opioid receptor.^182–184 This is problematic because the σ-opioid receptor is involved in some behavioral actions of nicotine.¹⁸⁵

Constitutional basis for nicotine dependence

Exposure to nicotine does not necessarily result in smoking. Even after smoking has become established, individuals exhibit varying degrees of nicotine dependence. Individuals who develop dependence are not necessarily those who happened to be exposed to a large amount of nicotine for a long period of time; those who develop nicotine dependence often exhibit pre-existing traits. Human and mouse studies suggest that molecular variation provides a basis for both pre-existing behavioral traits and susceptibility and resilience to elements of nicotine dependence.

One way to understand the complex picture of nicotine dependence is to hypothesize that nicotine acts to direct pre-existing behavioral traits toward nicotine or to provide some beneficial effects for the host (Figure 2). According to this constitutional view, nicotine dependence is determined in part by pre-existing molecular variation that provides a basis for quantitative variance in behavioral traits. However, pre-existing traits are unlikely to be all-or-none determinants for nicotine dependence; it is more probable that they probabilistically increase the chances of developing nicotine dependence.

Figure 2.

Three possible modes of the influence of molecular variation on nicotine dependence. Molecular variation, which involves altered levels or function of a molecule, is caused by genetic variants or non-genetic factors such as developmental anomalies or environmental factors. Molecular variation could increase tendencies toward novelty seeking affiliative attachment, loss of control, cue reactivity, decision-making deficits and automaticity. After exposure, nicotine could serve as a new target for these traits (re-directed target mode). Alternatively, molecular variation could set a basal tone of affect (for example, negative affect, depression, anxiety and stress vulnerability). Nicotine then could provide a ‘therapeutic’ or ‘repair’ function and may potentiate the intensity of negative reinforcement (defect repair mode). Molecular variation also could independently influence behavioral traits and susceptibility or resilience to nicotine dependence (shared, independent influence).

Pre-existing traits may confer susceptibility through various modalities. Smoking may become a target of some pre-existing behavioral traits (see Figure 2, redirected target mode). Compared to other age groups, adolescents tend to seek novel sensations and cigarettes and smoking might be perceived as novel. A pre-existing automatic and stereotypical behavioral pattern also might be directed toward smoking. A possible gene that contributes to behavioral choice-melioration, a condition in which smokers are unwilling to give up cigarettes even when faced with negative consequences (for example, illness), is an allele linked to the dopamine D2 receptor. Healthy volunteers with the A1 allele are impaired in their general ability to learn from negative consequences.¹⁸⁶ The A1 allele linked to the dopamine D2 receptor gene is associated with current smoking, as well.⁹⁶

In other instances, nicotine may be perceived as therapeutic for an individual’s pre-existing motivational or affective defects (see Figure 2, defect repair mode). Nicotine may induce this effect by normalizing negative affective states. Alternatively, nicotine may induce some subjective effects that counteract negative affective states. Affective abnormalities such as depressive symptoms, high levels of anxiety and altered stress reactivity exist before smoking starts in a subpopulation of smokers.⁷⁷ The ability of anti-depressants to reduce smoking and aid smoking cessation¹⁸⁷ is consistent with this mode. Alternatively, molecular variation may independently influence pre-existing behavioral traits and nicotine dependence (see Figure 2, independent influence).⁷⁹

The plasticity-based dependence model posits that addictive substances induce behavioral dependence (that is, addiction) because they cause pathologically persistent molecular and cellular alterations that, in a normal form, serve to establish long-term associative memories.¹⁸⁸ Rodent studies have provided an ample body of evidence that addictive substances, including nicotine, modulate many molecules that normally subserve learning-related events in the rodent brain.^188–190 The plasticity-based hypothesis of addiction/dependence provides a plausible explanation as to why nicotine consolidates cue reactivity or promotes self-administration, but it does not explain why such processes widely differ among individuals.

The ultimate evidence would be to demonstrate that such plastic alterations actually occur in the brains of heavy smokers, but not in chippers or non-smokers, as a causal event for the development of nicotine dependence. Needless to say, it is technically difficult to demonstrate drug-induced molecular alterations as a causal event in humans. Demonstration of altered levels or activities of molecules in the postmortem brains of current and former smokers, compared to non-smokers, could suggest that such alterations are induced by smoking and are long-lasting. However, equally plausible is an alternative interpretation that pre-existing molecular levels or activities increase the probability of initiating smoking. Levels or activities of molecules that are different in smokers than in former smokers and non-smokers could be equally interpreted as a suggestion that former smokers were able to quit because they lacked the pre-existing molecular substrates for addiction or that nicotine only transiently induces these molecular alterations in former smokers. Functional imaging might provide an alternative, reliable way to obtain correlative evidence that altered responsiveness of regions in the brain predates or follows the initiation of drug exposure in both humans and experimental animals.¹⁹¹

Setting the dividing line between smokers and non-smokers is another difficult issue in determining the molecular correlates of nicotine dependence in humans. Artificially drawing a line between smokers and non-smokers would create a barrier to revealing the underlying mechanisms. ‘Smokers’ may include heterogeneous samples, if distinct molecular substrates exist for heavy smokers, moderate smokers and chippers. Similarly, it is important to consider that non-smokers might include a combination of individuals with and without susceptibility factors. Nicotine dependence develops only if nicotine is available and one initiates smoking. Even if one is burdened with many susceptibility factors, nicotine dependence will not develop unless smoking is initiated. ‘Non-smokers’ might be heterogeneous and may not serve as a reliable control. Unless initial exposure to nicotine is equal among smokers and non-smokers, the molecular basis for susceptibility to nicotine dependence will be difficult to identify. This point is well illustrated by the seminal study that demonstrated that the genetic contribution to tobacco consumption was more prominently revealed when historical social restrictions in Sweden changed to allow more women to smoke.¹⁹²

One possible difficulty associated with the constitutional hypothesis is that cue reactivity or cue exposure-associative processes occur only after exposure to nicotine and cues, and therefore, these processes cannot be a pre-existing trait. However, pre-existing molecular variation could potentially influence the response to nicotine, the general capability to form cue association or both. Consistent with this possibility, smokers with the A1 allele of the dopamine D2 receptor exhibit stronger cue-induced cravings than those without the allele, even though these two genotype groups are indistinguishable in the number of cigarettes smoked per day, years of smoking and levels of nicotine dependence.¹⁹³ This study suggests that pre-existing molecular variation could selectively affect the capability of cue reactivity without altering the direct effects of nicotine. Moreover, inbred rat strains that differ in susceptibility to behavioral alterations induced by nicotine exhibit differences in basal levels of many molecules in the brain, suggesting that pre-existing molecular variation could alter the behavioral effects of nicotine.^194–196 Alternatively, both pre-existing molecular variation and induced alterations might determine the ultimate degree of dependence.

Individual variations in the susceptibility to dependence and pre-existing traits are not unique to nicotine dependence and are also seen with stimulants, opiates and alcohol dependence.^{138,197–200} However, any given molecule could exert different effects in response to diverse addictive substances. For example, the constitutive absence of FosB or CREB renders mice more sensitive to cocaine, but less sensitive to nicotine in the place-conditioning paradigm.^{109,143,201,202}

A future challenge is to evaluate how each molecule contributes to pre-existing traits and the susceptibility and resilience to substance dependence. Such an endeavor will hopefully result in the identification of appropriate molecular targets for medication to more effectively prevent substance dependence disorders.