More than Smoke and Patches: The Quest for Pharmacotherapies to Treat Tobacco Use Disorder

A. Self-Administration

Self-administration is a behavioral assay that has traditionally served as the cornerstone for examining the abuse liability of drugs and can be used to evaluate potential pharmacotherapies for substance use disorders. In this assay, animals are trained to make an operant response, typically either a lever press or a nose poke, to receive a drug infusion. Typically, drugs in self-administration procedures are delivered via an intravenous injection, and the nicotine literature is no exception (Goodwin et al., 2015); however, vapor chambers have recently been used to deliver nicotine in animal studies (George et al., 2010), although no published studies have used this for nicotine self-administration to date. Nonetheless, nicotine can serve as a positive reinforcer and is self-administered by rodents (Collins et al., 1984; Corrigall and Coen, 1989; Donny et al., 1995; Valentine et al., 1997; Picciotto et al., 1998), nonhuman primates (Goldberg et al., 1981; Sannerud et al., 1994), and humans (Henningfield and Goldberg, 1983; Henningfield et al., 1983).

Early nicotine self-administration studies were successful in demonstrating that nicotine could serve as a reinforcer, but they were insufficient to characterize nicotine as a drug of abuse because nicotine did not maintain rates of behavior comparable to drugs such as cocaine (Deneau and Inoki, 1967). Although it has now been shown repeatedly and definitively that under certain conditions nicotine maintains rates of behavior equivalent to cocaine, the experimental variables that impact this behavior have been a consistent topic of study. For example, pretraining with food or a drug like cocaine is often used to facilitate acquisition of nicotine self-administration (Griffiths et al., 1979; Goldberg et al., 1981; Yanagita et al., 1983; Slifer and Balster, 1985). Also, because nicotine self-administration in animals is typically done intravenously, it has been argued that nicotine infusion rate is an important factor in nicotine self-administration, with slower rates supporting more robust self-administration in rats (Sorge and Clarke, 2009). However, it has also been shown that slower infusion rates of nicotine decrease self-administration (Wakasa et al., 1995; Wing and Shoaib, 2013). Furthermore, length of self-administration session has also been manipulated, and rats show signs of withdrawal when nicotine is removed from extended access sessions of self-administration (O’Dell et al., 2007a); however, limited access sessions have been shown to produce similar levels of dependence when sessions are conducted 7 days a week (Paterson and Markou, 2004). Thus, the relative importance of some variables in nicotine self-administration is open to debate.

One experimental variable that clearly influences the reinforcing effectiveness of nicotine is the schedule of reinforcement. A second order schedule of reinforcement was the first to demonstrate intravenous self-administration of nicotine at high rates (Goldberg et al., 1981). Another commonly used schedule of reinforcement in self-administration is the progressive ratio schedule of reinforcement (Donny et al., 1999; Brunzell et al., 2010; Cohen et al., 2012; Le Foll et al., 2012; Weaver et al., 2012; Gamaleddin et al., 2013; Garcia et al., 2014); this is typically used to estimate the maximum reinforcing effectiveness of a drug. In progressive ratio experiments, the number of responses required for a single drug infusion increases by some amount with each successive infusion instead of remaining constant. This allows for determination of a “breakpoint,” the point at which the response demand is high enough that the animal will no longer work to receive drug infusions. Thus, a favorable outcome for a potential pharmacotherapy for smoking cessation in a progressive ratio experiment would be to decrease the breakpoint or, in other words, make the animal less willing to work to receive an infusion of nicotine.

Another variable that is uniquely important to the study of nicotine self-administration is pairing of the nicotine infusion with some other unconditioned stimuli. In fact, some groups have specifically sought to better understand this relationahip and developed other variations on self-administration of nicotine that integrate both Pavlovian and operant conditioning components to study its effectiveness as a reinforcer. For example, it has been shown in rats that when a nonrewarding conditioned stimulus is paired with nicotine (the unconditioned stimulus), the nonrewarding conditioned stimulus is reinforced (Bevins and Palmatier, 2004). Furthermore, pairing this conditioned stimulus with access to nicotine increases the amount of nicotine that is self-administered. Among other benefits, studies like these can serve to help in our understanding of how other effects of cigarettes, such as the subjective feeling of holding a cigarette or drawing cigarette smoke into the lungs, might be related to smoking cessation and relapse.

Importantly, in the abovementioned self-administration procedures, the primary dependent variable is typically a measure of the rate of responding. One limitation of many traditional preclinical self-administration procedures when evaluating potential pharmacotherapies for tobacco use disorder has been the integration of control experiments that allow for a distinction to be made between drug effects that selectively decrease the rate of responding for a self-administered drug as opposed to effects that produce generalized suppression of behavior. However, self-administration studies in humans have a long history of using choice experiments to examine a variety of abused drugs, including nicotine (Johnson and Bickel, 2003; Bisaga et al., 2007; Odum and Baumann, 2007; Stoops et al., 2011; Green and Lawyer, 2014; Cassidy et al., 2015). In preclinical choice procedures, in addition to a rate-dependent measure of responding, rate-independent data about the allocation of responses for the drug as opposed to the nondrug reinforcer can also be collected (see Banks and Negus, 2017, for review). For this reason, drug versus nondrug choice experiments are becoming a more frequently used method for studying changes in preclinical self-administration behavior, although relatively few studies have used nicotine in choice paradigms to date (Lesage, 2009; Panlilio et al., 2015; Huynh et al., 2017; Bagdas et al., 2019). This may reflect a difference between nicotine and other drugs that has been noted in the human literature; although an alternative reinforcer (e.g., money) is effective for reducing cigarette smoking in humans (Bisaga et al., 2007), a meta-analysis revealed that the effectiveness of alternative options has a relatively weaker effect in studies with nicotine as compared with heroin or cocaine (Prendergast et al., 2006). However, it may also simply result from the nature of nicotine as a reinforcer that is typically not self-administered as robustly as other drugs of abuse in preclinical studies.

B. Intracranial Self-Stimulation

Intracranial self-stimulation (ICSS) is another operant procedure in which behavior is maintained by pulses of electrical brain stimulation (for review see Carlezon and Chartoff, 2007; Negus and Miller, 2014). When this procedure is used for evaluating the abuse potential of drugs, an electrode is most commonly implanted, targeting the medial forebrain bundle at the level of the hypothalamus. Following electrode implantation, the animal is trained to complete an operant response to produce an electrical stimulation that can be modified in terms of both amplitude and frequency. ICSS procedures have been performed in mice (Johnson et al., 2008; Fowler et al., 2013), rats (Schaefer and Michael, 1992; Panagis et al., 2000; Kenny et al., 2009), and nonhuman primates (Routtenberg et al., 1971) to study the ability of drugs (Negus and Miller, 2014; Freitas et al., 2016) and physiologic conditions (Freitas et al., 2015) to produce increases or decreases in baseline ICSS responding. Many drugs of abuse produce increases in measures of baseline ICSS responding; this is typically interpreted as an abuse-related effect (Bonano et al., 2014) and is correlated with alterations in dopamine signaling (Bauer et al., 2013). Furthermore, both drugs of abuse as well as drugs that do not produce abuse-related effects in animals are able to produce decreases in measures of baseline ICSS responding given sufficiently large doses; this is typically interpreted as an abuse-limiting effect (Bauer et al., 2013). Nicotine produces dose-dependent biphasic effects in ICSS, increasing responding at lower doses of nicotine and decreasing responding at higher doses (Schaefer and Michael, 1986; Huston-Lyons and Kornetsky, 1992; Bauco and Wise, 1994; Spiller et al., 2009; Freitas et al., 2016), similar to effects seen in the self-administration assay (Lau et al., 1994; Valentine et al., 1997; Le Foll et al., 2007). Thus, a favorable outcome for a potential pharmacotherapy for smoking cessation might be to attenuate nicotine-induced increases in ICSS, as seen with the N-methyl-D-aspartate receptor antagonist LY235959 (Kenny et al., 2009); however, this type of ICSS procedure is not the most commonly used for evaluating pharmacotherapies for tobacco use disorder.

An alternative ICSS procedure uses discrete trials that vary the current intensity to determine a threshold amplitude that will maintain operant responding. In these types of procedures, the reward-enhancing effects of acute nicotine are observed in the form of decreases in brain reward threshold (Bespalov et al., 1999; Nakahara, 2004; Paterson, 2009). Furthermore, following a regimen of chronic nicotine administration, both spontaneous and precipitated withdrawal produce increases in brain reward threshold, an anhedonia-like effect (Epping-Jordan et al., 1998; Bruijnzeel et al., 2007; Johnson et al., 2008). This increase in brain reward threshold is typically interpreted as diminished sensitivity to reward and decreased motivation for previously rewarding stimuli under conditions of nicotine withdrawal, and it is considered to be relevant insofar as preventing withdrawal plays an important role in successfully maintaining abstinence from smoking (Bruijnzeel and Gold, 2005; Hughes, 2006; Koob, 2008).

C. Drug Discrimination

Drug discrimination is another behavioral assay that is often used to examine compounds for abuse potential and to evaluate potential pharmacotherapies. Commonly, subjects are trained to make some response (e.g., pressing a lever) when they receive vehicle and some other response (e.g., pressing a different lever) when they receive the training dose of a drug. The training dose of the training drug then sets the occasion for responding on the drug-paired lever, and, with training, animals accurately choose the appropriate response lever even though there may be no other observable measures to indicate that they have received the training drug. In humans, a drug can be trained as a discriminative stimulus; simultaneously, subjects can be asked to respond on a variety of standardized questionnaires and rating scales (e.g., measures of “good” or “bad” drug effect) to collect subjective effects and discriminative stimulus effects simultaneously, which can be dissociable (Lamb and Henningfield, 1989). However, in humans that have been trained to discriminate nicotine from saline, the discriminative stimulus effects of nicotine are directly correlated with its subjective effects (Perkins et al., 1999).

The nicotine discriminative stimulus was one of the first studied in the operant discrimination procedure that is most commonly used today (Morrison and Stephenson, 1969). Thus, it should be no surprise that nicotine has been trained as a discriminative stimulus in a variety of species, including mouse (Gommans et al., 2000), rat (Zaniewska et al., 2006), monkey (Takada et al., 1988), and human (Perkins et al., 1996). If a test compound shares discriminative stimulus effects with nicotine, then it might serve as an effective substitution pharmacotherapy; however, there is also the potential for the test compound to have abuse liability itself.

Drug discrimination is a pharmacologically selective bioassay that was used in the past for elucidating the receptor pharmacology of nicotine in vivo (Pratt et al., 1983; Stolerman et al., 1999; Rollema et al., 2007). The nicotine cue is thought to be mediated centrally, and this is supported by the fact that a peripherally restricted nicotinic agonist, methylcarbamylcholine, does not substitute for nicotine (Desai et al., 1999). Specific brain regions can also be implicated by targeted injections of nicotine into the brain; in rats, nicotine injected into the dorsal hippocampus, but not the nucleus accumbens, produces nicotine-like discriminative stimulus effects (Shoaib and Stolerman, 1996).

One feature of drug discrimination is that the dose of the drug that is selected for training as a discriminative stimulus is known to impact the pharmacological selectivity of the resulting discrimination. For example, the discrimination of a relatively small training dose can lack pharmacological selectivity because the magnitude of the difference between the presence of a “drug effect” versus its absence is relatively small and difficult to detect. Lack of pharmacological selectivity is evidenced by substitution of test drugs with mechanisms of action distinct from the training drug. In contrast, sufficiently large training doses can result in discriminations that are relatively selective for test drugs that share a mechanism of action with the training drug [for examples with nicotine as a training drug, see Smith and Stolerman (2009) and Cunningham and McMahon (2013)].

D. Place Conditioning

Place conditioning is different from the operant assays discussed previously because it uses classic conditioning to measure preference for or avoidance of a location that has been paired with a drug stimulus. Both two- and three-chamber variations are common, in which the third chamber is a neutral, unpaired chamber that connects the first and second chambers. One of the chambers is typically paired with a dose of a drug, whereas a separate, distinct chamber is paired with the administration of the drug vehicle alone. After some number of pairings of the drug in one compartment and the absence of drug in the other, the animal is placed in the apparatus without an injection of drug or vehicle, and the amount of time spent in the two chambers previously paired with either drug or vehicle is measured. Most drugs of abuse produce a conditioned place preference. That is, animals will spend more time in the chamber previously paired with an injection of drug compared with the time they spend in the chamber previously paired with drug vehicle. Nicotine produces a place preference in both rats and mice at smaller doses (Fudala et al., 1985; Vastola et al., 2002; Walters et al., 2006) but an aversion to the place paired with larger doses of nicotine in mice, resulting in an inverted U-shaped dose-response curve (Risinger and Oakes, 1995).

One variation of this procedure is conditioned place aversion, in which instead of pairing one chamber with a drug, one chamber is paired with antagonist-precipitated withdrawal. Under these conditions, animals typically spend less time in the withdrawal-paired chamber (i.e., it is “avoided”). Conditioned place aversion studies of both mice and rats have found that adolescents, as compared with adults, have a smaller response in terms of avoidance of a chamber previously paired with nicotine withdrawal (O’Dell et al., 2007b; Jackson et al., 2009). For further review on conditioned place assays, please see Prus et al. (2009).

A. Absorption and Distribution

Once inhaled from a cigarette, nicotine reaches the brain within 10–20 seconds (Benowitz, 1990, 1996). This rapid rise in nicotine concentration, which allows for the titration of nicotine dose on a puff by puff basis, contributes to the high abuse liability inherent in this form of nicotine administration (Benowitz, 1990; Henningfield and Keenan, 1993).

Although nicotine is most commonly inhaled through cigarette smoke, translating this to preclinical studies has inherent difficulties. Monkeys can be taught to smoke cigarettes (Ando and Yanagita, 1981), but the variables described above limit the ability to deliver a specific, predetermined dose of nicotine via inhalation of tobacco smoke. Recent advances in technology have yielded vapor chambers for the reliable delivery of inhaled nicotine in preclinical studies. However, intravenous administration with chronic indwelling catheters remains the most common route for nicotine delivery in preclinical monkey, rat, and mouse administration procedures, in addition to extensive utilization in human studies (Goldberg et al., 1981; Spealman and Goldberg, 1982; Henningfield et al., 2016).

Nicotine delivered intravenously has 100% bioavailability compared with inhaled nicotine, 80%–90% of which is absorbed during smoking (Armitage et al., 1975). However, nicotine delivered by the intravenous route does not reach the brain as quickly as inhaled nicotine (Benowitz, 1990, 1996). Nevertheless, intravenous nicotine takes less than 60 seconds to reach the brain and provides the closest approximation of inhalation that allows for precise delivery of a specific dose. Humans report differences in the subjective effects of nicotine based on the route of administration (Henningfield and Keenan, 1993). However, of the routes of administration typically used in animal studies, only intravenous nicotine has been shown to share subjective effects with cigarette smoking in humans (Henningfield and Keenan, 1993); the subjective effects of subcutaneous nicotine in humans are modest at best (Le Houezec et al., 1993). Additionally, inhaled nicotine and intravenous nicotine follow a comparable time course in regard to onset and duration of action (Henningfield et al., 1985; Mello et al., 2013). This is opposed to subcutaneous administration of nicotine, which reaches a peak blood concentration between 20 and 25 minutes after injection in humans, although this route of administration also appears to offer 100% bioavailability (Le Houezec et al., 1993).

Solutions intended for use in ENDS that are currently available for consumers typically label nicotine content as a concentration of nicotine per total volume of liquid, and these concentrations range from 0 to 30 mg/ml. However, individual differences in inhalation variables, such as puff duration and velocity, that impact nicotine delivery from cigarettes also apply to nicotine delivered from ENDS, meaning that nicotine yield from ENDS can vary by more than 50-fold (Talih et al., 2015). Additional factors such as output voltage, other components of the nicotine solution (e.g., propylene glycol, vegetable glycerin, flavors), and the pH of the solution also impact nicotine exposure with ENDS, so it is not surprising that studies using different procedures often report different results. For example, some studies report that ENDS deliver less nicotine than a cigarette (Trehy et al., 2011; Farsalinos et al., 2013; Yingst et al., 2019) and increased latencies to reach peak nicotine concentration in blood (Farsalinos et al., 2014). Several important limitations of these studies are worth noting. First, many early studies of ENDS used experienced smokers that were relatively naïve to vaping (Schroeder and Hoffman, 2014), and it has been shown that different inhalation strategies used by naïve compared with experienced ENDS users may be responsible for lower nicotine delivery from ENDS, and once sufficiently experienced in the use of ENDS products, users may achieve higher concentrations of nicotine in blood (Farsalinos et al., 2014; Schroeder and Hoffman, 2014). Furthermore, the increased latency to peak nicotine concentration in blood may be a result of significant buccal absorption in vaping-naïve ENDS users as opposed to primarily pulmonary absorption in experienced cigarette users (Schroeder and Hoffman, 2014). More recent studies suggest that experienced ENDS users alter their inhalation strategy to achieve similar peak levels of nicotine with ENDS use as they achieve with cigarette use, independent of the concentration of nicotine solution used (St Helen et al., 2016b), and that the time course of nicotine in blood is very similar to cigarette smoking, with peak nicotine concentrations within 2–5 minutes of vaping (St Helen et al., 2016a). However, even if ENDS users receive similar amounts of nicotine, the absence of toxins present in combusted smoke have led to the generally accepted conclusion that ENDS are less harmful than cigarettes (https://www.cdc.gov/tobacco/basic_information/e-cigarettes/about-e-cigarettes.html), although they are not approved pharmacotherapies for smoking cessation. Studies in rats have found that experimental vapor chambers reliably produce air-nicotine concentrations of 4–12 mg/m³ and that within 60 minutes of exposure, animals have levels of nicotine in blood equivalent to the average concentration observed in human smokers (Gilpin et al., 2014).

All formulations of FDA-approved nicotine replacement therapy are absorbed more gradually than either inhaled or intravenous nicotine, resulting in slower increases in nicotine blood levels (Henningfield et al., 1985; West et al., 2000). This more gradual increase in nicotine concentration results in lower relative abuse liability, as slower absorption produces modest increases of dopamine over time in key areas of the brain related to substance use disorders in contrast to the corresponding quick spike of dopamine release and subsequent downstream signaling events produced by cigarette smoking (Dani and De Biasi, 2001; Nestler, 2005). Evidence suggests that simultaneous smoking may slow transdermal absorption, as was found to be the case when nicotine was administered intravenously to nicotine patch wearers (Benowitz et al., 1992). Thus, absorption kinetics, as opposed to simply absorption route, is a critical factor in determining the therapeutic potential of a nicotine replacement strategy.

Nicotine absorption is dependent on the pH of the vehicle used for administration as well as the environment it is administered into (e.g., liquid of the oral cavity for buccal absorption) (Le Houezec, 2003; Hukkanen et al., 2005; U.S. Department of Health and Human Services, 2010; Pickworth et al., 2014). However, once nicotine is in the bloodstream at a physiologic pH, it is distributed extensively to body tissues. In autopsies of smokers, the highest affinity for nicotine was found in the liver, kidney, spleen, and lung, and the lowest was found in adipose tissue (Hukkanen et al., 2005).

The time course of nicotine, its accumulation in various organs of the body, and its pharmacologic effects are highly dependent on the route of administration and rate of dosing. The concentration of nicotine in blood after smoking a cigarette can reach 100 ng/ml but is generally in the range of 20–60 ng/ml (Armitage et al., 1975; Henningfield and Keenan, 1993; Gourlay and Benowitz, 1997; Rose et al., 1999; Lunell et al., 2000). Blood levels of nicotine peak after smoking a cigarette and fall rapidly over the subsequent 20 minutes; the average distribution half-life of nicotine is about 8 minutes (Hukkanen et al., 2005). Over the course of a day, smokers typically demonstrate trough concentrations of nicotine in blood from 10 to 35 ng/ml and peak concentrations between 20 and 50 ng/ml (Schneider et al., 2001). The average elimination half-life of nicotine in plasma is the same for both inhaled and intravenous nicotine, approximately 100–150 minutes (Benowitz and Jacob, 1993, 1994). Thus, typical patterns of cigarette smoking result in considerable accumulation of nicotine over the course of a day, which then diminishes overnight, resulting in very low nicotine levels upon waking in the morning.

Nicotine present in saliva is often used as a convenient proxy for the amount of nicotine present in blood. However, in nicotine skin patch users, nicotine in saliva was a factor of 8.13-times greater than nicotine in plasma (Rose et al., 1993). This accumulation is likely due to ion trapping of nicotine in saliva when in ionized form (Hukkanen et al., 2005).

B. Metabolism and Elimination

Metabolism of nicotine takes place primarily in the liver, and nicotine has six primary metabolites, although numerous others have also been identified, including cotinine, trans-3-hydroxycotinine, nicotine N-oxide, nornicotine, norcotinine, and cotinine N-oxide (Hukkanen et al., 2005). Cotinine is the primary metabolite in both humans and nonhuman primates; 70%–80% of nicotine is metabolized to cotinine in the liver in humans (Benowitz and Jacob, 1994), whereas rhesus macaques metabolize 80% of nicotine to cotinine (Poole and Urwin, 1976). Mice, rabbits, and dogs also metabolize nicotine into cotinine at a rate similar to humans and nonhuman primates. However, rats and guinea pigs metabolize nicotine equally into nicotine-N-oxide, cotinine, and trans-3-hydroxycotinine (Matta et al., 2007). Cotinine and trans-3-hydroxycotinine are the primary metabolites identified in urine for all mammalian species studied to date (Jenner et al., 1973; Nwosu and Crooks, 1988; Kyerematen et al., 1990). Half-lives of nicotine and cotinine appear to be similar in humans and macaques (Seaton et al., 1991). The half-life of nicotine is generally 45 minutes in rats and between 6 and 7 minutes in mice. This is considerably shorter than the 2-hour half-life of nicotine observed in humans and nonhuman primates (Matta et al., 2007). Thus, an important consideration for nicotine studies in rodents is that a higher dose of nicotine is needed to achieve equivalent human physiologic levels.

The enzyme responsible for both metabolism of nicotine to cotinine and cotinine to trans-3-hydroxycotinine in both humans and rhesus monkeys is CYP2A6 (Murphy et al., 1999; Hecht et al., 2000; Hukkanen et al., 2005). In mice, CYP2A5 is the functional homolog of human CYP2A6. In rats, CYP2A6 is inactive. Instead, CYP1B1/2 is the enzyme responsible for nicotine metabolism (Hammond et al., 1991; Nakayama et al., 1993). Additionally, cigarette smoking is known to accelerate the metabolism of some drugs (Zevin and Benowitz, 1999), although it appears to slow the metabolism of nicotine itself (Benowitz and Jacob, 1993). In humans, differences in metabolism based on both ethnicity and sex have been reported, including faster nicotine and cotinine clearance in women than in men (Pérez-Stable et al., 1998; Hukkanen et al., 2005; Benowitz et al., 2006, 2009; Tanner et al., 2015).

Cotinine has a longer elimination half-life than nicotine, averaging about 770–1130 minutes (Benowitz and Jacob, 1994), but the elimination half-life of trans-3-hydroxycotinine falls between nicotine and cotinine at about 400 minutes (Benowitz and Jacob, 2001). The longer elimination half-life of cotinine relative to nicotine results in less variability in cotinine concentrations measured over the course of the day. This has resulted in the wide-spread use of cotinine concentration as a biomarker for daily tobacco consumption (Benowitz et al., 1996), and several studies have demonstrated that experienced ENDS users achieve levels of cotinine similar to cigarette smokers (Etter and Bullen, 2011; Caponnetto et al., 2013). Furthermore, the ratio of trans-3-hydroxycotinine to cotinine present in plasma or saliva can be used as a marker of CYP2A6 activity (Dempsey et al., 2004). A genetic polymorphism in the CYP2A6 gene results in individuals who may be broadly categorized as fast or slow metabolizers, and this ratio is a predictor of cigarette consumption (Benowitz et al., 2003).

Nonrenal clearance accounts for the majority of nicotine elimination. Renal clearance is, on average, about 35–90 ml/min, which accounts for about 5% of total nicotine clearance (Hukkanen et al., 2005).

A. Receptor Pharmacology

Acetylcholine is the endogenous neurotransmitter for acetylcholine receptors, which fall into two major groups: nAChRs and muscarinic acetylcholine receptors (mAChRs) (Albuquerque et al., 1995; Gotti and Clementi, 2004; Eglen, 2005; Dani and Bertrand, 2007). Muscarinic receptors are metabotropic, G protein-coupled seven transmembrane receptors that were originally defined with activation by muscarine, a product of the Amanita muscaria mushroom (Eugster et al., 1965). There are five subtypes, labeled M1 through M5 (Hulme et al., 1990; Fredriksson et al., 2003). Like nicotinic receptors, mAChRs are located both centrally and in the periphery on neuronal and nonneuronal cells. However, in comparison with nicotinic receptors, which are rapidly activated (i.e., microseconds), activation of mAChRs is generally slower (i.e., milliseconds). For a review of mAChR pharmacology, see Kruse et al. (2014). Although there is no evidence that nicotine binds to muscarinic receptors, effects mediated by muscarinic receptors may be an important consideration in the development of potential pharmacotherapies that target endogenous acetylcholine.

Nicotinic receptors are ionotropic, ligand-gated ion channels that were originally defined by activation with nicotine, an alkaloid produced by plants in the nightshade family, but traditionally associated with plants of the genus Nicotiana, otherwise known as tobacco plants. Nicotinic receptors are composed of five subunits (Cooper et al., 1991), which, together, form a pore in the cell membrane that allows for the passage of ions in and out of the cell.

Nicotinic receptors can be generally divided into two populations: muscle-type and neuronal. Muscle-type nAChRs were identified first and are found at the neuromuscular junction, where ion conductance through the channel produces excitatory postsynaptic potentials that are characteristic of muscle contraction. Neuronal nAChRs can be further subdivided into those that serve the autonomic nervous system (i.e., ganglionic) and those that are present in the brain (i.e., central). Like muscle-type nAChRs, ganglionic nAChRs are generally located postsynaptically and transmit fast excitatory postsynaptic potentials that are often the first signal in a serial circuit followed by slow excitatory postsynaptic potentials mediated by mAChRs. Compounds restricted to the periphery by poor penetration of the blood-brain barrier act selectively at these receptors, and ganglionic receptors may be responsible for some side effects of cholinergic drugs.

Nicotinic receptors in the brain are of primary interest for the study of tobacco use disorder. Like muscle-type and ganglionic receptors, nicotinic receptors in the brain are ligand-gated ion channels composed of five subunits (Fig. 1A). The subunits that have been identified in mammalian brain are notated as α2 through α7 and β2 through β4. Although in theory, many different possible combinations of subunits could come together to form an ion channel, there are apparent limitations. One of these limitations is that certain α subunits are required for a functional binding site; thus, the β subunits are sometimes referred to as accessory subunits. Both homomeric subtypes, which include five of the same α subunit and heteromeric subtypes, containing both α and β subunits, have been identified. In mammalian brain, homomeric receptors are thought to be limited to those containing five α7 subunits, whereas several distinct heteromeric subtypes have been identified, each with different pharmacological characteristics (Fig. 1B). Although they serve distinct functions, nicotine binds to all subtypes of nAChRs in the brain; however, the affinity of nicotine for the nAChR varies by subtype.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Basic structure of neuronal nAChRs. (A) A nAChR is composed of five subunits. Each subunit contains four transmembrane domains (M1–M4). An amine functional group is located at the end of the M1 transmembrane domain, whereas a carboxyl group is located at the end of the M4 transmembrane domain (insert). When an agonist (e.g., acetylcholine, nicotine) binds an orthosteric site, extracellular sodium and calcium enter the cell. (B) Diagram of the pentameric structure of neuronal nAChRs, which can be heteromeric (e.g., α4β2, α4α5β2) or homomeric (e.g., α7) in composition.

B. Tolerance

There are three different types of drug tolerance: pharmacokinetic, pharmacodynamic, and behavioral. Tolerance is said to occur when a larger dose of drug is required to achieve the same level of effect previously achieved by a smaller dose of the drug or when the same dose of drug produces a smaller magnitude of effect with subsequent administration. Tolerance to the effects of abused drugs often occurs in individuals with substance use disorders, but tolerance itself is not indicative of substance use disorder. In the context of the current review, the word “tolerance” will be used exclusively to describe pharmacodynamic tolerance to effects of nicotine. Importantly, nicotine produces at least two distinct types of tolerance: chronic tolerance, which develops over a period of days and can be observed in experienced smokers even following a period of abstinence, and acute tolerance, which develops over a period of minutes to hours. In clinical studies, it is generally presumed that chronic tolerance has developed in habitual smokers, allowing for comparisons to be made between groups both with (i.e., smokers) and without (i.e., nonsmokers) chronic tolerance to nicotine (Perkins et al., 1993). Acute tolerance, on the other hand, occurs as rapidly as after one dose of nicotine (Stolerman et al., 1973) and clinically can be observed in both experienced smokers as well as nicotine-naïve individuals (Perkins et al., 1993). Evidence suggests that both acute and chronic tolerance to nicotine are important considerations in the development of pharmacotherapies for tobacco use disorder (Balfour, 1994; Benowitz, 2008).

Nicotine in the context of cigarette smoking is typically dosed repeatedly at semiregular intervals over the course of the day. This leads to cycles of receptor activation and desensitization, and accumulating evidence suggests that both states contribute to the reinforcing effects of cigarette smoking [see Picciotto et al. (2008) for review]. This pattern of behavior is generally disrupted overnight, when sleeping precludes continued nicotine dosing and leads to a period of nighttime abstinence, with subsequent withdrawal symptoms occurring when waking in the morning. Thus, the first cigarette is smoked on a baseline that differs from every other cigarette during the day.

Acute tolerance to the subjective effects of nicotine is thought to be responsible for the finding that smokers report the first cigarette of the day as the most pleasurable (Fant et al., 1995). Experimentally, acute tolerance to effects of nicotine has been examined in both monkeys and humans, and there is evidence for tolerance to both physiologic (e.g., cardiovascular) effects of nicotine (Perkins et al., 1991) as well as the discriminative stimulus effects of nicotine (Perkins et al., 1996; Moerke and McMahon, 2018) and the subjective effects of nicotine (Perkins et al., 1993). Furthermore, acute tolerance to the subjective effects of nicotine is reported both in smokers and subjects who have never smoked (Perkins et al., 1993), although these two groups rate the subjective effects of nicotine differently.

However, tolerance to effects of nicotine that develops over days and years has historically been of more interest for the purpose of developing pharmacotherapies for the treatment of tobacco use disorder (Kauer and Malenka, 2007; Kalivas et al., 2009). Thus, chronic tolerance to the effects of nicotine on locomotor activity has been studied extensively in rodents (Keenan and Johnson, 1972; Stolerman et al., 1973, 1974; Hubbard and Gohd, 1975; Hatchell and Collins, 1977; Clarke and Kumar, 1983a,b; Marks et al., 1983, 1985), and chronic tolerance to the subjective effects of nicotine has been studied in humans (Perkins et al., 1993, 1994, 2002). Although the precise mechanisms underlying tolerance to locomotor activity and subjective effects of nicotine remain unclear, it has long been known that nicotine receptor binding is increased in the brains of animals after chronic exposure to nicotine (Marks et al., 1983; Schwartz and Kellar, 1983; Sanderson et al., 1993) as well as in human smokers compared with nonsmokers (Benwell et al., 1988; Breese et al., 1997; Court et al., 1998; Perry et al., 1999). Indeed, receptor upregulation is considered one hallmark of chronic tolerance to nicotine (Benwell et al., 1988; Cairns and Wonnacott, 1988). This is in opposition to what is typically observed after chronic drug treatment, which is downregulation of receptors (Overstreet and Yamamura, 1979; Creese and Sibley, 1981). Currently, this exception to the general rule is thought to be related to nicotine’s ability to inactivate nAChRs (Marks et al., 1983; Schwartz and Kellar, 1985).

In vitro studies suggest that upregulation of nicotinic receptors varies based on the subtype of receptor. Whereas the α4β2* subtype is readily upregulated after nicotine exposure, the α3β4 subtype is upregulated to a lesser degree. Furthermore, the α4β2* subtype appears to recover more slowly from upregulation than does the α3β4 subtype (Peng et al., 1994; Wang et al., 1998; Fenster et al., 1999; Meyer et al., 2001; Harkness and Millar, 2002). Evidence for this difference has been demonstrated in vivo as well. In rats treated with nicotine for 14 days, upregulation of α4β2* binding sites was shown to be between 20% and 100%, depending on brain region; however, similar upregulation of α3β4 receptors was not apparent (Nguyen et al., 2003).

Interestingly, altered transcription does not seem to play a role in this receptor upregulation, as mRNA levels remain constant over time with nicotine exposure (Marks et al., 1992; Peng et al., 1994; Ke et al., 1998). Furthermore, chronic treatment with nicotine does not appear to modify metabolism (Hatchell and Collins, 1977; Marks et al., 1983). In preclinical rodent studies of chronic tolerance, 1 mg/kg nicotine administered twice daily for 5 days is sufficient for receptor upregulation to reach its half-maximal state (Marks et al., 1985; Schwartz and Kellar, 1985).

The subjective effects of nicotine in cigarette smokers are influenced by both chronic and acute tolerance to nicotine. By extension, the subjective effects of nicotine differ as a function of duration of abstinence and thus are likely to change over the course of smoking-cessation treatments. For example, both chronic and acute nicotine tolerance can lead to decreases in some subjective measures of smoking. As mentioned above, nicotine acts as an array of nicotinic receptors and has a wide range of physiologic effects. Also, as mentioned above, selective targeting of specific nAChR subtypes is common in the development of pharmacotherapies for smoking cessation. Thus, an agonist at α4β2* used as a smoking-cessation pharmacotherapy may produce tolerance to the effects of nicotine that are mediated by α4β2* and may or may not alter nicotine’s actions at other nAChR subtypes (de Moura and McMahon, 2017). Furthermore, both acute and chronic tolerance will alter the effectiveness of treatments for smoking cessation, insomuch as these treatments are or are not affected by cross-tolerance. Thus, these factors must be considered when evaluating potential pharmacotherapies for tobacco use disorder.

C. Physiologic Dependence and Withdrawal

Although substance use disorders are often associated with the positive subjective effects of the abused drug, continued drug use can also be motivated by the desire to avoid negative effects associated with drug withdrawal. Discontinuation of nicotine use in dependent individuals leads to increases in stress, appetite, insomnia, anxiety, and irritability as well as disruptions in cognition (Hughes, 2007; American Psychiatric Association, 2013; Wesnes et al., 2013). Many pharmacotherapies for smoking cessation attenuate the withdrawal effects of nicotine discontinuation, and this is thought to be one important aspect of their effectiveness. Experimental paradigms in rodents and nonhuman primates can recapitulate many of the symptoms associated with nicotine withdrawal in humans, such as increases in stress-like, anhedonia-like, and anxiety-like behaviors and disruptions in learned memory tasks (Malin et al., 1994; De Biasi and Salas, 2008). For example, nicotine withdrawal in rats was found to increase corticotropin-releasing factor levels in the central nucleus of the amygdala, an area of the brain known to modulate stress response (George et al., 2007). Furthermore, nicotine withdrawal increased anxiety-like behavior through activation of a subset of corticotrophin-releasing factor receptors. Of note, repeated exposure to inhaled nicotine vapor (as with ENDS), upon abrupt discontinuation, produces signs of tobacco use disorder and withdrawal in both humans (Morean et al., 2018) and rodents (George et al., 2010, 2011). Clinically, the reported severity of nicotine withdrawal effects by smokers is a potential predictor of relapse (West et al., 1989). Thus, an additional consideration in developing potential pharmacotherapies for smoking cessation is the treatment of nicotine withdrawal–related effects.

A. Nicotine Replacement Therapies

In the United States, nicotine replacement is the most widely available and easily accessible pharmacotherapy for smoking cessation, as most formulations can be purchased over the counter and without a prescription. Nicotine replacement was added to the WHO List of Essential Medicines in 2009, and a wide variety of nicotine replacement products are marketed around the world. There are a number of legal and regulatory differences between countries as well as differences in attitude about nicotine replacement as a harm-reduction measure; thus, the availability (i.e., by prescription only or from a pharmacy) of different formulations varies from country to country. For example, nicotine gum (Nicorette), the first nicotine replacement therapy approved by the FDA for use in the United States in 1984, was originally available by prescription only. Similarly, the transdermal nicotine patch (NicoDerm CQ), approved by the FDA in 1991, originally required a prescription. These formulations only became available “over the counter” in the U.S. in 1996, when nicotine nasal spray (Nicotrol NS), followed by the nicotine inhaler (Nicotrol), became available by prescription. The nicotine lozenge (Commit), approved in 2002, was the only nicotine replacement therapy that never required a prescription in the United States. ENDS started appearing on the market around 2003, and use in the United States and elsewhere has grown dramatically in recent years; however, there is no current consensus on the safety or efficacy of these products, so regulations vary widely. In the United States, they are regulated by the FDA and are legal to buy for anyone 18 years of age or older; however, they are not considered a pharmacotherapy for smoking cessation.

One nicotine reduction strategy was part of an effort by tobacco companies to lower perceived harm of their product by marketing so-called “light” cigarettes. Although low-yield nicotine cigarettes are considered attractive measures for decreasing nicotine consumption (Benowitz and Henningfield, 1994), these products were primarily intended to reduce nicotine content with filter ventilation, not by changing the nicotine content of the tobacco used in cigarettes. Furthermore, for these types of cigarettes, evidence suggests that smokers will alter variables of nicotine intake, including puff volume, depth of inhalation, extent of dilution with room air, rate of puffing, and intensity of puffing, to compensate (U.S. Department of Health and Human Services, 1981). More recently, this approach has been revived by researchers studying low-yield nicotine cigarettes in clinical trials, indicating that cigarettes with reduced nicotine content do have potential for use in smoking-cessation treatments (Hatsukami et al., 2010; Donny et al., 2015; Pacek et al., 2016; Tidey et al., 2017), but this strategy is not currently approved by the FDA.

Nicotine replacement therapies, regardless of formulation, work on the same principle executed slightly differently; they promote smoking cessation because they are substitution therapies, continuing to provide nicotine by another route of administration and without other tobacco constituents after an individual has quit using tobacco. These therapies are aimed at reducing tobacco cravings and withdrawal symptoms by delivering therapeutic doses of nicotine that generally have a slower onset and are thus less likely to be abused than inhaled nicotine. Furthermore, nicotine is delivered without the additional toxins that are inhaled along with tobacco smoke. With the exception of the nicotine patch, which delivers nicotine at a constant low rate, these formulations can be taken on an as-needed basis, although they do have recommended dosing regimens. Ideally, an individual will taper the magnitude of the dose of nicotine they are self-administering, as nicotine replacement is only intended to be used for 2–3 months. This is one potential drawback to nicotine replacement therapy, as many people find themselves unable or unwilling to gradually decrease their dose and continue long-term use (Hajek et al., 1988; Hughes et al., 1991; Johnson et al., 1991; Hurt et al., 1995). Despite this pattern of behavior, studies to date have not examined the effects of these pharmacotherapies over extended periods of time, so the relative safety of long-term use of nicotine replacement therapies is unknown. Additionally, even among patients receiving a prescription for a product available over the counter (i.e., nicotine patch, nicotine gum), only 67% are given any instructions for usage by their physician, and 77% receive no follow-up (Shiffman et al., 2007). Nicotine replacement therapies may also produce side effects, such as insomnia, dizziness, and headaches, that compromise compliance (Jorenby et al., 1995; Hurt et al., 1998; Hajek et al., 1999) as well as have the potential for nicotine toxicity at large doses (Dale et al., 1995).

Despite these potential disadvantages, nicotine replacement therapies generally appear to approximately double a smoker’s chance of remaining abstinent 6 months after quitting over placebo, and there do not appear to be differences among the different formulations (Cahill et al., 2013). However, a conflicting report suggests that after properly adjusting for bias, there is no indication that use of any nicotine replacement formulation improves outcomes for smoking cessation over placebo (Stanley and Massey, 2016). This claim does not extend to findings that suggest that using two nicotine replacement therapies in combination appears to be about twice as effective as using any one of the formulations alone (Cahill et al., 2013, 2014). The benefit of combining two different forms of nicotine replacement therapy is similar to what has been reported for varenicline alone.

B. Varenicline (Chantix)

Varenicline is a novel compound developed specifically for use in smoking cessation. Approved by the FDA in 2006, varenicline is only available by prescription. The recommended course of treatment is to begin taking varenicline a week before planning to quit smoking. For the first 3 days, 0.5 mg is taken as a tablet once per day and then twice per day for the rest of that week. At this point, when no longer smoking, the dose is increased to 1 mg twice daily for an additional 12 weeks. This course of treatment can be repeated or extended for those who relapse. Varenicline was identified among a series of compounds synthesized based on the structure of cytisine, a natural compound and partial nAChR agonist marketed for smoking cessation in Europe as Tabex (Coe et al., 2005). Like cytisine, it is designated as a partial, or low-efficacy, agonist. Theoretically, as a partial agonist, varenicline is effective as a pharmacotherapy for smoking cessation in two complimentary ways: 1) it functions as a substitution therapy like nicotine replacement, reducing cravings and withdrawal effects; 2) it also functions as an antagonist therapy, reducing or preventing the reinforcing effects of tobacco if an individual continues to smoke while taking it (Coe et al., 2005).

Electrophysiological studies in transfected cells demonstrate that varenicline does not produce the same maximum effect as nicotine at α4β2* nAChRs (Coe et al., 2005; Rollema et al., 2007). Furthermore, varenicline antagonizes nicotine’s effects at this receptor subtype (Coe et al., 2005). Varenicline also demonstrates lower efficacy than nicotine to stimulate dopamine release from rat brain slices (Rollema et al., 2007). In vivo varenicline does not alter ICSS thresholds; however, varenicline does block nicotine-induced facilitation of ICSS in rats (Vann et al., 2011). In many drug discrimination studies, including those in mice (Cunningham and McMahon, 2013), rats (Rollema et al., 2007; Jutkiewicz et al., 2011), and monkeys (Cunningham et al., 2012; Moerke et al., 2017) discriminating nicotine, varenicline fully substitutes for the nicotine discriminative stimulus, although additional studies in rats found this was only the case at relatively short pretreatment times (i.e., 5–40 minutes), whereas longer pretreatment times (i.e., 2–4 hours) resulted in very low levels of generalization with the nicotine discriminative stimulus (Le Foll et al., 2012). Under conditions in which varenicline only partially substitutes for the nicotine discriminative stimulus, varenicline sometimes, but not always, antagonizes the discriminative-stimulus effects of nicotine (LeSage et al., 2009; Cunningham and McMahon, 2013). When varenicline fully substituted for the nicotine discriminative stimulus in monkeys, combinations of nicotine and varenicline were synergistic (Cunningham et al., 2012). These observations, including full substitution of varenicline for nicotine, do not exclude the possibility that varenicline has lower efficacy than nicotine. Instead, it may simply be that for the training doses of nicotine studied in these discrimination assays, the efficacy demand might be sufficiently low, such that even a low-efficacy agonist can mimic the effects of a higher-efficacy agonist. Another possibility is that varenicline pretreatment results in acute tolerance through desensitization of the receptors where nicotine is acting to produce discriminative-stimulus effects; experimentally, acute cross-tolerance from varenicline to nicotine would not readily be distinguishable from antagonism of nicotine. These results are summarized in Table 1.

Although classified as a partial agonist at α4β2* nAChRs, as described above, varenicline has also been characterized as a full agonist at homomeric α7 nAChRs (Mihalak et al., 2006). In mice, varenicline dose-dependently blocks a conditioned place preference for nicotine (Bagdas et al., 2018a). Interestingly, α5 but not α7 nAChR knockout mice display an attenuation of varenicline’s effects on nicotine conditioned place preference (Bagdas et al., 2018a). That is, although varenicline is classified as an α7 nAChR full agonist, varenicline’s actions at this receptor in vivo seem somewhat dispensable. Meanwhile, it appears that nAChRs containing the α5 subunit may mediate at least some of varenicline’s effects.

Early reports of the side effects of varenicline prompted the FDA to require a black box warning for depression, suicidal thoughts, and suicidal actions on varenicline in 2009. However, several meta-analyses have found no evidence for an increase in any adverse neuropsychiatric events beyond sleep disturbances, which have been well-documented, and the warning has since been removed (Harrison-Woolrych and Ashton, 2011; Thomas et al., 2015). Other side effects, predominantly nausea, are reported more frequently than similar side effects with nicotine replacement therapies (Cahill et al., 2013, 2016). In addition to being generally unpleasant, side effects can also compromise compliance as well as promote relapse (Williams et al., 2007; Faessel et al., 2009; Kasliwal et al., 2009; Harrison-Woolrych and Ashton, 2011; Jimenez-Ruiz et al., 2013). Varenicline appears to approximately double one’s chances of remaining abstinent for a year over placebo alone (Gonzales et al., 2006), but it does not appear to be any more or less effective than multiple nicotine replacement therapies used in combination (Cahill et al., 2013, 2016; Baker et al., 2016).

C. Bupropion (Zyban)

Bupropion was approved for use in the United States as an antidepressant under the trade name Wellbutrin as early as 1985 but not as a pharmacotherapy for smoking cessation until 1997 (Zyban). In some countries, such as the United Kingdom, its only approved indication is as a smoking-cessation aid. It is considered an atypical antidepressant in the sense that it does not appear to function as a selective serotonin reuptake inhibitor, as is the case for many commonly prescribed antidepressants. Chemically, it is a substituted cathinone with a complex mechanism of action that is not completely understood. Thus, the exact mechanism(s) by which bupropion functions as a smoking-cessation aid is unknown, as it has effects at many targets in the central nervous system.

Bupropion shares many of its effects with other psychostimulants, and it is typically characterized as a dual norepinephrine and dopamine reuptake inhibitor (Stahl et al., 2004). In drug discrimination studies, both amphetamine, a structurally related compound, and cocaine, a structurally distinct compound, fully crossgeneralize with bupropion in both rats and monkeys (Jones et al., 1980; Blitzer and Becker, 1985; Kamien and Woolverton, 1989; Kleven et al., 1990; Lamb and Griffiths, 1990; Terry and Katz, 1997; Bondarev et al., 2003). Furthermore, bupropion supports self-administration behavior in rats as well as monkeys (Bergman et al., 1989; Lamb and Griffiths, 1990; Tella et al., 1997). This does not appear to completely translate to humans, however, as human studies have shown that although bupropion produces some subjective effects similar to abused drugs like amphetamine, it does not have abuse liability different from placebo (Griffith et al., 1983; Miller and Griffith, 1983).

Other data from studies comparing bupropion and nicotine have also produced mixed results. For example, in animals trained to discriminate nicotine, the ability of bupropion to substitute for nicotine has varied widely between studies. In rhesus monkeys, bupropion did not substitute for nicotine (Cunningham et al., 2012), and in mice, it only partially substituted for nicotine (Damaj et al., 2010); however, in rats, bupropion fully substituted (Wiley et al., 2002; Young and Glennon, 2002), partially substituted (Desai et al., 2003), or did not substitute (Shoaib et al., 2003), depending on the study. These results are summarized in Table 1.

Understanding the mechanism that underlies the therapeutic effects of bupropion for smoking cessation is further complicated by other actions of bupropion itself in addition to its active metabolites at a variety of central nervous system targets. One possibility is that bupropion functions as a smoking-cessation aid through antagonism of nAChRs (Slemmer et al., 2000). However, as bupropion is predominately metabolized by the CYP2B6 enzyme into its active metabolites R,R-hydroxybupropion, S,S-hydroxybupropion, threo-hydrobupropion, and erythro-hydrobupropion, these metabolites may also play an important role in its therapeutic effects (Cooper et al., 1984). Preclinical studies in mice found that both R,R-hydroxybupropion and S,S-hydroxybupropion were similar to bupropion in reversal of antagonist-precipitated nicotine withdrawal symptoms (Damaj et al., 2010), and the clinical effectiveness of bupropion treatment has also been linked to its active metabolites (Zhu et al., 2012). Clinical trials suggest that although bupropion is not as effective for smoking cessation as varenicline, it does work better than placebo (Hughes et al., 2014; Anthenelli et al., 2016; Stead et al., 2016; Windle et al., 2016). It is, however, a potent inhibitor of the CYP2D6 enzyme (Güzey et al., 2002; Jefferson et al., 2005), which is necessary for the metabolism of a variety compounds; thus, it may have adverse effects in combination with other drugs that rely on this mechanism for their clearance.

D. Off-Label Pharmacotherapies for Smoking Cessation

Clonidine and nortriptyline are considered “second line” medications for smoking cessation. Although these drugs appear to be more effective than placebo at maintaining abstinence at 6 months, they are associated with an increased risk of adverse side effects as compared with “first line” pharmacotherapies (Cahill et al., 2013). Furthermore, relatively little preclinical research has addressed the potential mechanism of action that these pharmacotherapies use to produce therapeutic effects in tobacco use disorder. Clonidine attenuates footshock-induced reinstatement of nicotine-seeking behavior in rats self-administering nicotine (Zislis et al., 2007; Yamada and Bruijnzeel, 2011) but, to our knowledge, has not been tested with nicotine in drug discrimination. Nortriptyline has no effect on nicotine discrimination and is only effective in decreasing nicotine self-administration in rats at doses sufficiently high to also depress responding for food (Wing and Shoaib, 2012). Mecamylamine, the nAChR antagonist, was originally approved for use in hypertension but has also been used both alone and in combination with nicotine as a potential therapy for tobacco use disorder. Alone, mecamylamine is ineffective for smoking cessation (Lancaster and Stead, 2000), but there is very limited evidence that it might be slightly more effective in combination with nicotine replacement therapy than nicotine replacement therapy alone (Rose et al., 1998). More recently, ENDS have been proposed for use in smoking cessation, as previously discussed. There is still a lack of evidence, but ENDS technologies might also help improve outcomes for smoking cessation, although the extent to which ENDS might be more effective than currently approved pharmacotherapies is unknown (Malas et al., 2016).

A. Allosteric Modulation of Nicotinic Acetylcholine Receptors

Allosteric modulation has long been recognized as the primary mechanism of action for FDA-approved drugs (e.g., benzodiazepines, barbiturates), but only more recently has it inspired widespread interest as an alternative strategy for targeting nAChRs, not just for smoking cessation but also for a variety of disorders characterized by dysfunction of the central nervous system. Allosteric interactions were originally proposed in the context of enzymatic reactions (Monod et al., 1965), but it was not long before this model was extended to include ligand binding at biologic membranes (e.g., nAChRs) (Changeux et al., 1967). Importantly, the word “allosteric” has been used somewhat ambiguously as a descriptor of ligands in the literature. For the sake of simplicity in the current review, we consider “orthosteric ligands” of nAChRs to include both acetylcholine and nicotine as well as other ligands that share the same canonical binding site as acetylcholine and nicotine; “allosteric ligands” will be used to refer to compounds that bind to nAChRs at sites that are distinct from the canonical “orthosteric” site. It has been theorized that one important feature of an allosteric modulator for use in smoking cessation might be that it does not produce effects alone in the absence of an orthosteric ligand. Instead, only when the orthosteric ligand is present would the allosteric modulator serve to produce a change in either the affinity and/or the efficacy of the orthosteric ligand (Uteshev, 2014) (Fig. 2). This is particularly promising in terms of a therapeutic strategy, as the effects of an allosteric modulator alone should be minimal, reducing the potential for unwanted side effects like those that often occur with approved substitution-type pharmacotherapies. For example, a negative allosteric modulator (NAM) might serve to reduce the reinforcing effects of nicotine, whereas a positive allosteric modulator (PAM) in combination with a more traditional orthosteric agonist therapy could reduce the dose of agonist required to produce a therapeutic effect. Decreasing the minimal effective therapeutic dose of an orthosteric ligand through combination with a positive allosteric ligand could both retain therapeutic effectiveness and decrease the potential for toxicity and other side effects, particularly when substitution pharmacotherapies are taken in larger quantities or more often than the recommended clinical indication. Furthermore, agonist activation of nAChRs also perpetuates cycles of activation and desensitization, which are also of importance in tobacco use disorder [see Picciotto et al. (2008) for review]. In this case, smaller doses of an orthosteric agonist in combination with a PAM should result in less receptor desensitization produced by the orthosteric ligand (Williams et al., 2011).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Neuronal nAChR positive allosteric modulation. (Left) A neuronal nAChR with agonist bound at the orthosteric binding site allows for the influx of sodium and calcium into the cell. (Center) A positive allosteric modulator (PAM) bound to an allosteric site of a neuronal nAChR. By itself, it does not trigger the influx of ions into the cell. (Right) A neuronal nAChR PAM bound to an allosteric site, with an agonist at the orthosteric binding site, allows for an increased influx of sodium and calcium into the cell.

PAMs of nAChRs are proposed to increase the binding affinity and/or efficacy of an orthosteric agonist (Pandya and Yakel, 2013), whereas NAMs theoretically do the opposite by decreasing the binding affinity and/or efficacy of an orthosteric agonist. However, there are several different mechanisms by which allosteric modulators may accomplish these effects, which we will discuss briefly with a focus on heteromeric nAChRs. The prevailing consensus is that nAChRs are composed of dynamic proteins that are capable of multiple different states. However, for simplification, here we limit discussion to three possible states: closed, open, and desensitized. Furthermore, heteromeric nAChRs can potentially have 0, 1, or 2 agonist molecules bound. The likelihood of the receptor remaining stable in any one of the three possible states and the rates at which one state shifts to another state depend on the level of agonist occupancy (Changeux and Edelstein, 1998; Auerbach, 2010). At baseline equilibrium, when exogenous and endogenous signaling does not occur and no agonist is bound, nAChRs remain preferentially in the closed state. However, with an extremely low probability of shifting to the open state, the receptor may exist with some equilibrium between the closed state and the desensitized state (Williams et al., 2011). If a high concentration of agonist sufficient to saturate all the possible binding sites is rapidly applied, α4β2 nAChRs have an 80% probability of simultaneously shifting transiently into the open state before reaching a new equilibrium in the desensitized state (Li and Steinbach, 2010). Furthermore, it is known that once receptors are in the desensitized state, agonists bind with higher apparent affinity.

One way that a PAM of nAChRs may exert its effects would be to increase the agonist binding to the closed state of the receptor, which is experimentally represented by an increase in the potency of the agonist. This type of modulation might be most advantageous under a condition in which there is a low concentration of agonist, which would not otherwise produce a maximal response. By increasing the potency of the agonist, lower concentrations would be able to produce the maximum response. However, it would remain impossible to exceed the maximum response if the modulator is only changing the potency of the orthosteric ligand.

An alternative effect observed with PAMs is that they are often observed to increase the efficacy of an agonist. One way a PAM might accomplish this is via shifting the equilibrium between the open and closed states, making it easier to move from closed to open state. This would result in not only more receptors moving to the open state but also a greater likelihood that they might move from closed to open more than once before shifting to the desensitized state. Thus, this would yield a concurrent decrease in the rate of desensitization. Functionally, this could be observed as an overall increase in the time spent in the open state. Having this effect, a PAM can produce a transient increase in efficacy. The reverse of these conditions is hypothesized with the use of a NAM, leading to a transient decrease in agonist efficacy. However, neither a NAM nor a PAM in this scenario would alter how favored one state is over another, only the rate of change between states. Thus, receptors would eventually end up preferentially in the desensitized state, as under normal conditions.

Finally, if instead of altering the rate of change between states, as described above, a modulator functioned to shift equilibrium away from the desensitized state, this would result in yet another unique profile of observable effects. The result of this type of modulator would likely not be manifest as an increase in efficacy or the maximum effect, nor would it be observable under conditions in which the agonist interaction with the receptor pool is brief (e.g., acetylcholine broken down by acetylcholinesterase, rapid agonist application). Instead, it would be most apparent under equilibrium conditions, producing a significant amount of steady-state current (Williams et al., 2011).

The above is a simplified explanation of a theoretical model. Although there is experimental evidence that is consistent with these scenarios, it is not possible to prove or disprove them; they simply are not violated by what is currently known. Furthermore, it has been reported that, at least under some conditions, only a small percentage of available nAChRs are capable of being activated at once (McNerney et al., 2000; Li and Steinbach, 2010). Thus, the ability of a modulator to change receptors from the inactive to the active state is yet another possible mechanism by which it might enhance agonist activity. Finally, and perhaps most importantly, a similar but not identical model can be applied to homomeric nAChRs. Homomeric nAChRs are different from heteromeric nAChRs because they contain only one type of subunit; thus, they can potentially have 0, 1, 2, 3, 4, or 5 orthosteric agonist molecules bound to a single receptor (i.e., assemblies of five of the same subunit yield five essentially equal interfaces where ligand binding is possible). Heteromeric nAChRs, as described above, will have at the very most two similar interfaces for ligand binding, and it is likely that binding to one of these interfaces is not exactly interchangeable with binding to the other. Furthermore, an additional “fast” desensitization state has been described for homomeric α7 nAChRs, which is concentration-dependent and thought to be possible under conditions in which more agonist molecules are bound to a single receptor than is possible for heteromeric receptors (Papke et al., 2000). However, regardless of the exact mechanism, both negative and positive allosteric modulation of nAChRs represents an attractive therapeutic strategy that may circumvent the limitations inherent in targeting the canonical orthosteric site. A list of current nAChR NAMs and PAMs are shown in Tables 2 and 3, respectively, and are discussed below with regard to potential application for the treatment of tobacco use disorder.

B. Acetylcholinesterase Inhibitors

Identifying novel targets for smoking cessation and developing drugs that produce a desirable profile of action at these targets is a process that can take more than 10 years and an investment of several billion dollars before a new drug ever reaches the market. One way to circumvent this process is to consider repurposing drugs that have already been approved by the FDA and examining their potential to produce therapeutic effects outside of their current indications.

Acetylcholinesterase (AChE) inhibitors represent one such class of drugs, as three are currently approved by the FDA for the treatment of cognitive deficits associated with Alzheimer disease: donepezil, rivastigmine, and galantamine. AChE inhibitors produce their effects by preventing AChE, an endogenous enzyme that helps regulate cholinergic neurotransmission, from breaking down acetylcholine. In theory, this would lead to increases in receptor activation by endogenous acetylcholine and might produce nicotine-like effects. Early experiments that attempted to use this strategy in rats found that the nicotine discriminative stimulus could not be mimicked or potentiated by the AChE inhibitor physostigmine (Rosecrans and Meltzer, 1981; Pratt et al., 1983). However, when physostigmine was trained as a discriminative stimulus, although nicotine did not substitute, oxotremorine and arecoline, both mAChR agonists, did (Jung et al., 1988). Thus, AChE inhibitors were not studied further as potential treatments for smoking cessation.

Following FDA approval for Alzheimer disease, however, a small study of patients undergoing treatment of alcohol dependence indicated that galantamine was effective in reducing smoking in this population (Diehl et al., 2006), renewing interest in AChE inhibitors as pharmacotherapies for smoking cessation. Discrimination studies in rats (Giarola et al., 2011) and monkeys (Moerke and McMahon, 2019) provided evidence that donepezil and galantamine partially substituted for the nicotine discriminative stimulus, and several groups found that both donepezil and galantamine were effective at reducing nicotine self-administration in rats (Hopkins et al., 2012; Liu, 2013; Kimmey et al., 2014; Ashare et al., 2016). Results from preclinical studies using self-administration and drug discrimination assays are summarized in Table 1.

All three FDA-approved AChE inhibitors have been examined in smokers, and thus far galantamine has shown the most potential for smoking cessation. Rivastigmine, like galantamine, reduced tobacco consumption in alcohol-dependent individuals (Diehl et al., 2009) but had no effect on cigarette smoking in methamphetamine-dependent individuals (De la Garza and Yoon, 2011). Furthermore, donepezil did not effectively decrease cigarette smoking in a pilot study (Ashare et al., 2012). In comparison, only one clinical study using galantamine found no decrease in smoking among individuals with schizophrenia (Kelly et al., 2008). Several other clinical studies have shown that galantamine can attenuate cigarette craving and reduce smoking satisfaction in addition to reducing overall tobacco consumption (Sofuoglu et al., 2012; Ashare et al., 2016; MacLean et al., 2018).

One point of interest is that of the three AChE inhibitors approved for Alzheimer disease and discussed here, only galantamine has been shown to also have activity as a positive allosteric modulator of nAChRs (Maelicke et al., 2001; Farlow, 2003). However, recent studies suggest that galantamine does not functionally act at human α4β2* or α7 nAChRs as a PAM (Kowal et al., 2018). Thus, although it seems unlikely that action as a PAM is responsible for the relative success of galantamine to decrease smoking in some populations, it cannot be ruled out entirely. Clearly, much more work needs to be done both to evaluate AChE inhibitors in general as potential smoking-cessation pharmacotherapies as well as to determine if galantamine offers any benefits in comparison with existing pharmacotherapies for smoking cessation.

C. Psilocybin

Psychedelics, including psilocybin, were actively studied for their potential in treating substance use disorders from the 1950s into the early 1970s. Passage of the Controlled Substances Act in 1970, however, which classified psilocybin as a Schedule I compound, effectively halted any further development as a pharmacotherapy for decades.

Psilocybin is a prodrug that in vivo is rapidly metabolized to psilocin, an agonist at serotonin (5HT) receptors, and there is overwhelming evidence to support that 5HT2A is the most important 5HT receptor subtype for mediating the effects of classic psychedelics [see Nichols (2016) for review]. Recent study of psilocybin in the preclinical literature, however, suggests that 5HT2A receptors are only partially responsible for mediating the discriminative stimulus effects of psilocybin; in rats trained to discriminate 0.5 mg/kg psilocybin, M100907, a 5HT2A receptor antagonist, did not fully antagonize the discriminative stimulus effects of psilocybin up to doses that suppressed responding (Winter et al., 2007). In humans, however, there is abundant evidence that 5HT2A receptors mediate the subjective effects of psilocybin (Vollenweider et al., 1998; Kometer et al., 2012, 2013; Quednow et al., 2012).

A recent pilot study administered psilocybin over the course of a 15-week program of cognitive behavior therapy for current smokers. Participants received either two or three administrations of psilocybin under guided supervision in the treatment setting, and 12 of the 15 (80%) participants were confirmed abstinent by measure of urine cotinine levels at a 6-month follow-up; this is a notably higher percentage than is typically seen for smoking interventions, which is generally less than 35% (Johnson et al., 2014). Furthermore, follow-up at 2.5 years revealed that 60% of the participants had remained abstinent (Johnson et al., 2017). Such positive results should be interpreted with caution given the small sample size of the study and the fact that there was no comparison group; nonetheless, further trials are underway with psilocybin for not only tobacco use disorder but also alcohol use disorder, cocaine use disorder, depression, anorexia, and obsessive-compulsive disorder.

D. Hypocretin Receptor Antagonists

Hypocretins (Hcrt), also referred to as orexins, are neuropeptides produced within a small population of neurons located in the hypothalamus with projections to many other regions of the brain, including the limbic system (Peyron et al., 1998). Two G-protein coupled receptors, hcrtR1 and hcrtR2, have been identified, and the hypocretin system plays an important role in the homeostatic regulation of adaptive behaviors associated with arousal. Hypocretin signaling is necessary for the regulation of sleep cycles (Nishino et al., 2000; Thannickal et al., 2000) in addition to playing an important role in feeding (Haynes et al., 2000; Inutsuka et al., 2014) and mating (Muschamp et al., 2007) behaviors. Furthermore, hypocretin has also been implicated in behaviors of overconsumption, including substance use disorders (Barson and Leibowitz, 2017). For a current review on hypocretin receptors, please see Wang et al. (2018).

Accumulating preclinical evidence focuses on hypocretin receptor antagonists in reducing the reinforcing effects of nicotine. For example, both the hcrtR1 antagonist SB-334867 as well as a dual hcrtR1/hcrtR2 antagonist, almorexant, dose-dependently reduced nicotine self-administration in rats (Hollander et al., 2008; LeSage et al., 2010). However, 2-SORA, a hcrtR2 selective antagonist, had no effect on nicotine self-administration in rats, suggesting hcrtR1-mediated activity was responsible for decreases in self-administration (Uslaner et al., 2014). In tests of nicotine reinstatement, on the other hand, both hcrtR1 and hcrtR2 have been implicated but different groups have reported results that make interpretation difficult. Specifically, one group has reported attenuation of cue-induced reinstatement of nicotine-seeking behavior with hcrtR1 antagonist SB-334867 but not hcrtR2 antagonist TCSOX229 (Plaza-Zabala et al., 2013), whereas a second group has reported that cue-induced reinstatement of nicotine-seeking behavior is blocked by hcrtR2 antagonist 2-SORA (Uslaner et al., 2014), and a third group has reported no effect on cue-induced reinstatement of nicotine-seeking behavior with dual hcrtR1/hcrtR2 antagonist TCS1102 (Khoo et al., 2017). Furthermore, SB-334867 did not block footshock-induced reinstatement of nicotine seeking (Plaza-Zabala et al., 2010), and 2-SORA did not block nicotine-induced reinstatement of nicotine-seeking behavior (Uslaner et al., 2014). Results from preclinical studies of hcrtR compounds in nicotine self-administration and nicotine discrimination assays are summarized in Table 1. Nonetheless, a clinical trial is currently planned to evaluate the effects of suvorexant, a dual hcrtR1/hcrtR2 antagonist currently approved for the treatment of insomnia, in treatment of tobacco use disorder.