Nicotine Modulation of Fear Memories and Anxiety: Implications for Learning and Anxiety Disorders

Abstract

Anxiety disorders are a group of crippling mental diseases affecting millions of Americans with a 30% lifetime prevalence and costs associated with healthcare of $42.3 billion. While anxiety disorders show high levels of co-morbidity with smoking (45.3% vs. 22.5% in healthy individuals), anxiety disorders are also more common among the smoking population (22% vs. 11.1% in the non-smoking population). Moreover, there is clear evidence that smoking modulates symptom severity in patients with anxiety disorders. In order to better understand this relationship, several animal paradigms are used to model several key symptoms of anxiety disorders; these include fear conditioning and measures of anxiety. Studies clearly demonstrate that nicotine mediates acquisition and extinction of fear as well as anxiety through the modulation of specific subtypes of nicotinic acetylcholine receptors (nAChRs) in brain regions involved in emotion processing such as the hippocampus. However, the direction of nicotine’s effects on these behaviors is determined by several factors that include the length of administration, hippocampus-dependency of the fear learning task, and source of anxiety (novelty-driven vs. social anxiety). Overall, the studies reviewed here suggest that nicotine alters behaviors related to fear and anxiety and that nicotine contributes to the development, maintenance, and reoccurrence of anxiety disorders.

1) Introduction

Anxiety disorders, such as panic disorders, phobias, generalized anxiety disorder, and post-traumatic stress disorder (PTSD), are a group of crippling mental disorders affecting 40 million Americans with an 18% 12-month prevalence and a 30% lifetime prevalence [1, 2]. These statistics suggest that anxiety disorders are one of the most common mental disorders [3]. This group of disorders is associated with an estimated $42.3 billion in psychiatric and non-psychiatric treatment costs, indirect workplace costs, mortality-related costs, and prescription costs; this comprises approximately one third of the total mental health budget of the U.S. [4]. Moreover, anxiety disorders are associated with impaired workplace performance [4] and increased risk of morbidity and mortality related to cardiovascular problems [5–7]. Therefore, in addition to the devastating personal and social costs, anxiety disorders are responsible for a significant burden to the U.S. economy and taxpayers.

Anxiety disorders can develop following a single traumatic event or a highly stressful period [8]. Although different disorders under the anxiety disorder umbrella are defined by different symptoms, they are often characterized by an exaggerated fear response to cues and contexts that are not dangerous [9]. Common symptoms that most anxiety disorders share include re-experiencing (intrusive memories and nightmares), avoidance (avoidance of trauma-associated stimuli and memories), emotional numbing (loss of interest and interpersonal detachment), and hyperarousal (irritability, difficulty concentrating, hypervigilance, insomnia; [10]). For example, patients with PTSD usually re-experience the traumatic event through intrusive fearful memories, nightmares, and flashbacks, which evoke physiological distress, arousal and fear responses. Similarly, individuals with panic disorder usually experience episodes of intense fear even in situations where real danger is absent. Finally, patients with specific phobias experience similar exaggerated fear-related physiological responses to specific objects or situations [11].

Given the central role of fear response in anxiety disorders, many etiological accounts of anxiety disorders proposed that malfunctioning Pavlovian conditioning mechanisms are responsible for the development of anxiety disorders. One of the first accounts was offered by Watson and Rayner in 1920 as an explanation for phobias [12]. In their seminal “Little Albert” study, a baby was presented a white rat, which initially did not elicit a fear response, together with a fear-eliciting loud noise. These white rat-loud noise pairings resulted in a conditioned fear response to the white rat that eventually generalized to other furry objects. This early attempt to explain phobias suggested that the fear elicited by a loud noise was associated with another neutral stimulus present at the time (e.g., a white rat) and therefore, the fear of the rat developed based on this Pavlovian association.

Although anxiety is a different behavioral output than fear, based on the same principles, Pavlovian associations may result in “learned anxiety” to a conditioned cue or context. This “learned anxiety” may be generalized to other similar objects or settings causing anxiety [13]. The Pavlovian conditioning explanation has also been adopted for other anxiety disorders (see [14] for a review). For instance, later theories of PTSD argue that uncontrollable stress during severe trauma leads to re-experiencing symptoms triggered by the reminders of the traumatic event (e.g., [8, 15, 16]). According to these accounts, the elicited unconditioned fear response becomes associated with a variety of neutral cues and contexts during trauma and after the traumatic event these previously neutral stimuli trigger negative emotional responses, such as fear and anxiety, by reminding the person of the trauma. In support of these theories, evidence from animal research suggests that animals exposed to uncontrollable and unpredictable electric shock or a socially superior conspecific develop PTSD-like symptoms such as heightened fear and anxiety responses to stimuli associated with these experiences (e.g., [17–19]).

One of the most commonly used animal models of learned fear is fear conditioning, a Pavlovian conditioning paradigm involving an aversive unconditioned stimulus (US, e.g., mild foot shock) and a neutral cue or context akin to the design used by Watson and Rayner [12]. In fear conditioning, the neutral stimulus is paired with an aversive US that elicits a fear response. After the pairings, the previously neutral stimulus will elicit a fear response. Therefore, fear conditioning models the encoding and stimulus-based retrieval of the traumatic memories [14]. Similarly, there are several animal models of innate anxiety-related symptoms. The elevated plus maze (EPM), developed by Handley and Mithani [20], is the most utilized animal model of anxiety-related symptoms associated with anxiety disorders. The EPM, and variations such as the elevated zero maze [21] and elevated T-maze [22], take advantage of the innate exploratory behavior of rodents and their tendency to avoid bright open areas. Animals are placed in an apparatus that has open and closed arms and are allowed to explore the apparatus freely. The amount of time spent in the closed arms is an indicator of anxiety-like behavior and the time spent exploring the open arms indicates reduced anxiety. Evidence showing that classical anxiolytic drugs, such as benzodiazepines, reduce the time spent in the closed arms in rodents [23] suggest that these tasks are assessing anxiety. Other rodent models of anxiety also use innate behaviors indicative of anxiety such as marble burying (increased duration of burying an object; e.g., glass marbles) and social anxiety (social interaction test; decreased time spent by pairs of male rats in social interaction; [24]). However, it is important to note that the anxiety paradigms described above measure the innate anxiety response to a novel context or a conspecific as opposed to “learned anxiety” [13]. Nevertheless, animal models of anxiety are commonly utilized to understand the mechanisms responsible for the anxiety response independent from the learning component and therefore, provide invaluable information about anxiety-related symptoms in psychological disorders.

While the animal paradigms described above model the development and maintenance of anxiety disorder symptoms, fear extinction, a safety learning paradigm, is a model for exposure-based therapies of anxiety disorders. Mechanisms of extinction of conditioned responses were first proposed by Ivan Pavlov in his seminal work in 1920’s [25]. Following Pavlov, both animal and human studies have shown that repeated presentations of fear eliciting cues and contexts without the fear eliciting stimuli lead to extinction of negative emotional response in fear conditioning paradigms [26, 27]. Importantly, there are several studies suggesting that fear extinction is impaired in individuals with anxiety disorders [28–31]. Therefore, to promote extinction learning, the patient in exposure therapy is repeatedly given actual (in vivo exposure) or imaginary exposure to the trauma-associated cues or contexts followed by relaxation in order to reduce the negative emotional response elicited by these stimuli [32–34]. Exposure therapies have been successfully used to treat anxiety disorders since the work of Wolpe and his colleagues in 1960’s ([35, 36]; see [37] for a historical perspective) and they continue to be the most effective treatment method for a variety of anxiety disorders. The critical premise of the exposure therapies is that trauma-related memories should be reactivated during the process for successful extinction. In line with the importance of memory reactivation in exposure therapies, evidence from animal studies suggests that fear memories become labile and can be modified during memory retrieval, a process known as reconsolidation [38], which leads to a reduction of fear response in the absence of aversive stimulus. While extinction is different from the reconsolidation process [39], modification of memories during reconsolidation can also contribute to the success of exposure therapy. Although exposure therapies are relatively successful for treatment of anxiety disorde, relapse of the symptoms after a successful exposure therapy is common (19%–62%; [40, 41]). The release of symptoms may occur because extinction is an active inhibitory learning process that does not erase original fear memories and therefore, fear response resurfaces in the contexts other than the extinction context (renewal) or simply after a period of time (spontaneous recovery; see [42]for a review).

Overall, there are mulitple parallels between anxiety disorder symptoms and fear learning, suggesting that some symptoms of anxiety disorders arise from malfunctioning learning and memory processes. The mechanisms responsible for formation, consolidation, and retrieval of fear memories as well as anxiety-like behavior have been shown to be affected by many factors ranging from neurobiological mechanisms, such as protein synthesis (e.g., [38]), to outside factors such as stress (e.g., [17]). Numerous studies suggest that fear and anxiety-related learning and consequently, anxiety disorders, are especially vulnerable to the effects of drugs of abuse such as nicotine (e.g., [43]). In line with this idea, nicotine dependence has been shown to be critically involved in the symptom severity of anxiety disorders [44]. Therefore, this review focuses on the relationship between smoking and anxiety disorders based on the evidence from human studies as well as the studies reporting the effects of nicotine exposure on fear and anxiety-related behavior in animal models of anxiety disorders.

2) Co-morbidity of Nicotine and Anxiety Disorders

Patients with anxiety disorders are susceptible to unhealthy behaviors such as drug abuse [45, 46], which results in increased health costs [4], decreased quality of life [47], and poor health behaviors [48]. Patients with anxiety disorder show a particularly high rate of comorbidity with nicotine dependence [3, 49–52]. Specifically, the rates of nicotine dependence have been shown to be significantly higher in the population with anxiety disorders (45.3%) in comparison to healthy individuals (22.5%; [53, 54]), in addition anxiety disorders are more prevalent among the individuals who smoke (22%) than in the non-smoking population (11.1%, [55, 56]). In line with the high co-morbidity observed between smoking and anxiety disorders, prior smoking history has been found to be a predictor of a increased vulnerability to developing PTSD [52] and panic disorder [57] following a traumatic experience. Moreover, both smoking initiation and daily smoking rates have been shown to increase after a trauma [3, 49] and in patients with social phobia [58]. Patients with PTSD show lower rates of successful quitting [53, 59] and higher rates of relapse to smoking behavior [60] than the non-clinical population, which might be related to the more severe withdrawal symptoms they experience [44]. These results suggest a strong bi-directional relationship between anxiety disorders and nicotine dependence.

In line with the link between anxiety disorders and smoking, numerous studies have identified a significant association between smoking and anxiety disorder symptoms such as re-experiencing, avoidance, emotional numbing, and hyperarousal [48, 50, 61–63]. These results suggest that smoking behavior may be a modulator of the anxiety disorder symptoms. For example, several studies showed that heavy smoking behavior in PTSD patients was positively correlated with PTSD symptom severity [4, 48, 63, 64]. Importantly, while smoking is positively correlated with PTSD symptoms, there is also evidence showing that PTSD patients who smoke experience more severe nicotine withdrawal symptoms when they encounter trauma-related stimuli than non-PTSD smokers [65]. In addition, PTSD patients who smoke experience craving symptoms when presented with anxiety-triggering stimuli [66]. These studies suggest that although smoking increases the chances of developing PTSD and PTSD patients that smoke experience increased intensity of PTSD symptoms compared to non-smoking PTSD patients, smoking may serve as a means of alleviating some PTSD symptoms. The “self-medication” hypothesis of drug abuse is a well-supported theory that can explain high levels of smoking-anxiety disorder co-morbidity [67–72]. According to the self-medication hypothesis, a patient with an anxiety disorder acquires a substance abuse disorder to alleviate the anxiety symptoms. In support, specific PTSD symptoms are also linked to self-medication via smoking. For example, emotional numbing has been shown to be significantly associated with smoking, which is in line with the data showing that positive mood induction is more successful following nicotine administration in smokers who experience anhedonia, lack of pleasure towards natural reinforcers [73]. Similarly, hyperarousal symptom also strongly correlated with nicotine dependence, which may be because nicotine can reduce anxiety-like symptoms in humans [74–78] and nicotine withdrawal can aggregate PTSD symptoms [44]. These results suggest that nicotine mediates specific symptoms associated with anxiety disorders as nicotine provides short-term reduction of these symptoms but leads to increased baseline severity of anxiety disorder symptoms.

Nevertheless, the extent of nicotine’s effects on anxiety disorder symptoms varies between specific anxiety disorder subtypes. For example, while panic disorder has the highest prevalence of smoking among anxiety disorders and smoking is shown to increase the risk of initiation and maintenance of panic attacks [49, 79], OCD patients show the lowest rates of smoking and prospective studies could not find any predictive power of smoking on the development of OCD (e.g., [80]). Thus, different neural pathways may be involved in the different subtypes of anxiety disorders and nicotine’s effect on these pathways may vary. The next section reviews neural underpinnings of fear, anxiety, and other symptoms specific to anxiety disorder subtypes.

3) Neurobiology of anxiety disorders

The neurobiological mechanisms underlying the symptomatology of anxiety disorders are complex as they involve several distinct systems including pathways responsible for emotional processing as well as different neuroendocrine and neurotransmitter signaling pathways (see [81] for an extended review). A central hypothesis explaining the general problems with emotional processing observed in patients with anxiety disorders is that following a traumatic event or a highly stressful period, the balance between the cognitive and emotional control over behavior shifts towards emotional control [81]. The limbic cortex, which is the primary pathway responsible for emotional processing, and the prefrontal cortex (PFC), which controls higher level cognition (e.g. decision making, executive control), are interconnected and compete for functional control of behavioral ([82–84]; see [85] for a review). On the one hand, the limbic cortical regions, including the amygdala and hippocampus, process and integrate sensory and emotional information to control internal bodily responses through the modulation of the hypothalamic–pituitary–adrenal (HPA) axis, a neuroendocrine system responsible for stress response ([86]; Figure 1). Specifically, the hippocampus assumes control over the HPA axis by forming a negative feedback loop and regulating the sensitivity of the HPA system to corticosteroids [87]. On the other hand, the ventromedial PFC is responsible for modulating visceral emotional responses [88]. For example, the medial PFC has been shown to reduce fear response to previously fear-eliciting stimulus during extinction by inhibiting activity in the amygdala [82, 84]. Similarly, while the prelimbic subregion of the medial PFC reduces its spontaneous firing rate in the presence of a fear eliciting stimulus in parallel to the increased neuronal activity in the amygdala [89], lesions of the infralimbic subregion of the medial PFC have been shown to prevent recall of fear extinction [84]. In addition, lesions to the anterior cingulate cortex (ACC) have been shown to increase fear response to a conditioned stimulus [82]. Together, these results indicate that the interplay between prefrontal regions and the limbic cortex determines the behavioral outcome of the previously learned emotional associations.

Figure 1. Brain regions involved in the anxiety disorders.

The hypothalamic–pituitary–adrenal (HPA) axis is directly influenced by the amygdala and hippocampus whereas the prefrontal cortex indirectly modulates HPA axis activity through its inhibitory control over the amygdala.

In line with the hypothesis suggesting that the balance between the limbic and prefrontal regions is disturbed in patients with anxiety disorders, there is evidence suggesting that activation patterns of several prefrontal and limbic structures are altered in patients with anxiety disorders. For example, the amygdala, a limbic region central for fear learning and regulation, is hyperactive in patients with PTSD, which results in further difficulties in successful treatment [90]. Also, amygdala activation has been shown to correlate with the severity of PTSD symptoms [91]. While there is evidence for increased activity in fear learning-related structures, Bryant et al. [90] showed that rostral ACC volume in PTSD patients predict the success of the exposure therapies. Therefore, decreased prefrontal cortical function may be associated with impaired extinction learning observed in PTSD patients [28–31, 81]. Furthermore, Morey et al. [92] showed that ventromedial PFC activation was positively correlated with symptom severity in Iraq war veterans with PTSD and that PTSD patients exhibited exaggerated ventromedial PFC activity to combat-related stimuli. The increased PFC activity may be a direct response to the hyperactive limbic cortex activity triggered by trauma-related stimuli. While imaging studies of PTSD patients usually utilize a fear-related task, anxiety-related tasks are used with generalized anxiety disorder (GAD) patients. As in PTSD, GAD patients showed increased activity in the limbic circuitry, especially in the amygdala and insula [93, 94]. Also similar to PTSD patients, patients with GAD showed increased right amygdala activity when viewing angry faces; this activity was negatively correlated with ventrolateral PFC activity [95]. Finally, changes in prefrontal-limbic activity may modulate symptoms of panic disorder. Compared to healthy individuals, patients with panic disorder showed decreased PFC and increased cingulate gyrus and amygdala activity after administration of the respiratory stimulant doxapram, which induces panic attacks [96]. These results clearly suggest that the PFC-amygdala relationship determines anxiety disorder symptoms.

As a modulator of the HPA axis and with its strong connections with both the PFC and the amygdala, the hippocampus appears to have an especially important role in modulating anxiety disorder symptoms (Figure 1). Studies suggest that in patients with PTSD, hippocampal activity is increased [97–102] and this activity is positively correlated with PTSD symptoms [103, 104]. Moreover, hippocampal volume has been found to be decreased in PTSD patients [105–117] and the size of the hippocampus was negatively correlated with symptom severity [107, 108, 114]. These results suggest that hippocampal dysfunction contributes to PTSD symptomatology. Hippocampal abnormalities have also been linked to panic disorder [118, 119]. Thus, the hippocampus may modulate prefrontal cortical-amygdala interactions and changes in the connectivity between and the function of these areas may determine the symptomatology of anxiety disorders. Therefore, it is possible that the action of nicotine on these brain regions significantly contributes to smoking-related changes in anxiety disorders.

4) Neurobiology of nicotine and nicotinic receptors

Nicotinic acetylcholine receptors (nAChRs) are pentameric ligand-gated ion channels with seventeen known subunit combinations expressed in central and peripheral nervous systems [120–123]. Nicotinic receptors either contain only α subunits (α7–α10) or a combination of α (α2–α6) and β (β2–β4) subunits ([124–126]; see [127] for a review). In the central nervous system, α7 and α4β2* nAChRs (* designates potential additional subunit) are the most commonly expressed nAChRs [128–133]. Importantly, in terms of their affinity, functional changes, and localization, α7 and α4β2* nAChRs manifest different characteristics. For example, α4β2* nAChRs show higher affinity for nicotine, desensitize relatively slowly but upregulation of these receptors is long lasting, whereas α7 nAChRs show lower affinity to nicotine but desensitize rapidly [134, 135]. In addition, α7 and α4β2* nAChRs show differential cellular localization. While α7 nAChRs are mainly expressed in the granule and pyramidal cells of the hippocampus as well as other limbic system regions such as the amygdala and hypothalamus [136–138], α4β2* nAChRs are located in the dentate gyrus and CA1 sub-regions of the hippocampus [132]. Therefore, because of their different characteristics, α7 and α4β2* nAChRs may play different roles in the modulation of behavior.

While nAChRs gate sodium and/or calcium (Ca²⁺, [132, 139, 140]), they can also stimulate the release of neurotransmitters including acetylcholine, serotonin, GABA, dopamine, norepinephrine, and glutamate [141–148]. The pre- and post-synaptic locations on neurons increase the functional diversity of nAChR effects. For example, while nAChRs located on glutamatergic neurons modulate excitatory circuitries, those located on GABAergic neurons mediate inhibitory processes [137, 143]. Similarly, nAChRs expressed in pre-synaptic neurons trigger neurotransmitter release [137, 143, 149–153] whereas nAChRs in post-synaptic neurons contribute to depolarization and activation of secondary messenger systems [137, 153, 154]. By activating secondary messenger systems and cell-signaling cascades, nAChRs can be critically involved in the processes underlying synaptic plasticity and formation of long-term memory ([132, 139, 140, 155–159]; see [160] for a review). For example, nAChR activation can directly induce long-term potentiation (LTP; [161–163]), a form of synaptic plasticity thought to underlie formation of long-term memories [164]. LTP occurs through simultaneous activation of pre- and post-synaptic N-methyl-D-aspartic acid (NMDA) receptors. In support of the role of NMDA-dependent synaptic plasticity in fear learning, there are numerous studies showing that inhibition of NMDA receptors prevents formation of long-term fear memories (e.g. [165–169]; see [170] for a review). As a result of NMDA receptor activation, Ca²⁺ flows into the cell [171, 172], which triggers depolarization as well as changes in protein activation, mRNA synthesis, and protein translation [173–179]. Specifically, Ca²⁺ activates a variety of different cell-signaling cascades. One of the primary cascades activated during the LTP process is the CREB-dependent gene transcription pathway, which is initiated by the conversion of intracellular adenosine diphosphate (ADP) to cyclic adenosine monophosphate (cAMP, [180]). In turn, cAMP phosphorylates protein kinase A (PKA) can activate cAMP-response element binding protein (CREB) either directly or indirectly through activation of the mitogen activated protein kinases (MAPKs; e.g., extracellular-regulated kinase 1/2; ERK1/2). CREB activation leads to protein synthesis, which plays a crucial role in the induction of LTP and long-term memory formation [181–183].

As described above, nAChR activation can also lead to Ca²⁺ influx into the cell [184], which can modulate PKA and ERK1/2 activation [185]. Thus, nAChRs have the ability to trigger CREB-dependent gene transcription pathway and modulate synaptic plasticity (Figure 2). Several studies reported that the activation of nAChRs via nicotine and other agonists can enhance [186–188] or directly induce hippocampal LTP [161–163]. Specifically, Matsuyama and Matsumoto [162] showed that both nicotine and the α4β2* nAChR agonist epibatidine induced LTP in the mouse dentate gyrus whereas α7 nAChR activation was required for the epibatidine-induced LTP to reach the size of the nicotine-induced LTP. Also, both acute and chronic nicotine have been shown to enhance LTP induction while acute nicotine administration in hippocampal slices chronically treated with nicotine resulted in further enhancement of LTP [189]. Withdrawal from chronic nicotine abolished the nicotine-induced reduction of LTP threshold [190]. Also, Mann & Greenfield [186] showed that nicotine application in hippocampal slices decreased the threshold for LTP induction and this effect was reversed by the α7 nAChR-specific antagonist α-bungarotoxin, but not by the α4β2* nAChR antagonist dihydro-β-erythroidine (DHβE) or the non-specific nicotinic antagonist mecamylamine. In contrast, Fujii et al. [191] showed that inhibition of α7 nAChRs facilitated LTP-induction in the hippocampal slices not treated with nicotine, which suggests that α7 and non-α7 nAChRs may have opposing roles in the nicotine-induced enhancement of LTP. Specifically, while α4β2* nAChR activation enhanced LTP, activation of α7 nAChRs downregulated LTP induction. In addition, nAChRs seem to modulate LTP process through NMDA receptors (NMDARs), as there is evidence showing that cholinergic lesions impaired LTP and nicotine rescued LTP impairment by facilitating NMDAR activity [192]. There is also evidence showing that nAChR agonists nicotine and epibatidine can reverse the GABAergic inhibition of LTP in the hippocampal CA1 sub-region [191]. Similarly, activation of α4β2* nAChRs through epibatidine can also reverse amyloid β-protein induced suppression of hippocampal LTP [193]. Interestingly, nicotine may destabilize already consolidated LTP [194], which indicates differential effects of nicotine when administered before or after memory consolidation. These results clearly show that nicotine has a modulatory effect on LTP and several nAChR subtypes may be involved in the process.

Figure 2. Nicotinic modulation of the CREB-dependent gene transcription pathway.

Glutamate binds to NMDARs, leading to Ca²⁺ influx and PKA phosphorylation, which in turn, directly and indirectly through ERK1/2, activates CREB. Acetylcholine also triggers a Ca²⁺ influx by binding to the nAChRs, which can modulate this cell signaling cascade.

5) Involvement of nicotine and nAChRs in modulation of fear and safety learning: Evidence from animal studies

By modulating neurotransmitter release and cell signaling cascade activation, nAChRs may have direct effects on behavior such as learning and memory (see [195] for a review). There are several animal models used to investigate the effects of nicotine and other nAChR agonists/antagonists on fear-related symptoms observed in anxiety disorders. Specifically, fear conditioning has been widely utilized as a translational model to study traumatic experience, and safety learning paradigms such as fear extinction have been used to model exposure therapies [196]. Research on the neurobiology of fear conditioning has identified different types of fear memories mediated by different brain regions; i.e., hippocampus-dependent contextual and trace fear conditioning and hippocampus-independent cued fear conditioning [197–201]. Utilizing these animal paradigms, the effects of nicotine on fear conditioning have been extensively studied (see [127] for a review). These studies repeatedly showed that in mice, acute nicotine administration enhanced hippocampus-dependent contextual [43, 202–206] and trace [204, 207] fear conditioning but had no effect on hippocampus-independent fear conditioning or on general freezing behavior (e.g., [43, 205]). While these results show that nAChR activation is required for enhancement of hippocampus-dependent fear conditioning, Gould and Higgins [43] also showed that the non-specific nAChR antagonist mecamylamine had no effect on contextual fear conditioning. This indicates that nAChR activation is not critical for normal contextual fear learning. These studies suggest that acute nicotine administration, which models the initial effects of smoking, may enhance fear memories and potentially aggravate fear-related anxiety disorder symptoms such as intrusive memories. There is also evidence from multiple studies showing that the enhancing effects of acute nicotine on hippocampus-dependent memory are mediated by β2-containing nAChRs. Specifically, studies show that systemic administration of an α4β2* nAChR antagonist, DhβE, reversed the acute nicotine-induced enhancement of contextual fear conditioning, while a low-affinity α7 nAChR antagonist, methyllycaconitine (MLA), had no effect [208]. In line with this study, studies also showed that acute nicotine had no effect on contextual and trace [207] fear conditioning in knock-out (KO) mice that lack β2-containing nAChRs. These results strongly suggest that β2-containing nAChRs are necessary for the enhancing effects of acute nicotine on hippocampus-dependent learning.

Although acute nicotine seems to have profound effects on fear memory, many of the anxiety disorder patients that smoke are heavy smokers with established nicotine dependence [209]. The chronic effects of nicotine can be modeled in mice with continuous subcutaneous (s.c.) administration of nicotine via osmotic mini-pumps (e.g. [208, 210]). Using this model, several studies have investigated the effects of both chronic nicotine and withdrawal from chronic nicotine on fear conditioning in mice [210–214]. These studies showed that while chronic nicotine had no effect on fear conditioning, withdrawal from chronic nicotine resulted in deficits in contextual and trace fear conditioning. There are several changes occurring in the brain during chronic nicotine administration and withdrawal that may be responsible for these effects. First, during chronic nicotine administration hippocampal nAChRs desensitize and upregulate [215–218]. Therefore, it is suggested that in the absence of nicotine during withdrawal, upregulated nAChRs re-sensitize, which results in a hypersensitive cholinergic system and deficits in hippocampus-dependent learning [215]. This hypothesis was supported by Gould et al.’s [216] findings that the duration of withdrawal deficits in fear conditioning paralleled the duration of nAChR upregulation in the hippocampus. Moreover, in support of a hypersensitive cholinergic system during withdrawal, Wilkinson and Gould [218] found that acute nicotine administration during nicotine withdrawal resulted in an even greater enhancement of contextual fear conditioning compared to acute nicotine administration in nicotine naïve mice. Finally, Gould et al. [219] showed that tolerance to the effect of nicotine on contextual fear conditioning takes a shorter duration of chronic nicotine exposure to develop than nicotine withdrawal effects and that during tolerance nAChRs are not significantly upregulated. Thus, it is possible that nAChR desensitization, which occurs faster than receptor upregulation, underlies tolerance whereas slowly developing nAChR upregulation is necessary for withdrawal.

In sum, studies investigating the effects of acute, chronic, and withdrawal from chronic nicotine demonstrate that while acute nicotine enhances hippocampus-dependent fear learning, withdrawal results in impaired fear learning and chronic nicotine has no effect. Therefore, although nicotine withdrawal seemingly results in reduced fear learning it does not mean it is beneficial in terms of anxiety disorder symptoms. This is because nicotine withdrawal is also associated with many other psychological symptoms such as irritability, depression, restlessness, insomnia, anxiety, hunger, and poor concentration [220] and anxiety disorder patients that smoke experience more severe nicotine withdrawal symptoms when they encounter trauma-related stimuli than smokers without anxiety disorders [65]. Consequently, as discussed previously, patients with anxiety disorders show lower rates of successful quitting [53, 59] and an increased level of relapse to smoking [60] than non-anxiety disorder smokers. Although re-initiation of smoking may help patients with anxiety disorders reduce the withdrawal symptoms, Wilkinson and Gould’s [218] results suggest that an increased sensitivity of the cholinergic system during withdrawal may enhance fear learning during relapse of nicotine use, which could worsen the condition of the patient.

Similar to their role in acute nicotine’s effects on hippocampus-dependent learning, β2-containing nAChRs have also been shown to be necessary for withdrawal effects. For example, Portugal et al. [221] demonstrated that KO animals that lack the β2 subtype of nAChRs did not show withdrawal deficits in contextual fear conditioning. Furthermore, there is evidence showing that infusions of the α4β2 nAChR antagonist DhβE into the dorsal hippocampus precipitated deficits in contextual [222] and trace [214] fear conditioning in mice treated chronically with nicotine. Importantly, the involvement of the β2-containing nAChRs may also add translational value to the animal studies of fear conditioning. This is because the β2-containing nAChRs have been implicated to play a critical role in anxiety disorder symptomatology in humans. Specifically, using the radiotracer [¹²³I]5-IA-85380 and single-photon emission computed tomography (SPECT), Czermak et al. [223] investigated the relationship between nAChRs and PTSD and found that compared to non-smoker healthy individuals, non-smoker PTSD patients showed a significantly higher density of β2 nAChRs in the mesiotemporal cortex including the amygdala and hippocampus, regions previously linked with PTSD pathogenesis [224, 225]. In addition, the same study also found that the density of the β2 nAChRs in the thalamus and mesiotemporal cortex was significantly correlated with PTSD symptoms such as re-experiencing. This suggests that β2-containing nAChRs, which are largely responsible for the effects of nicotine on hippocampus-dependent fear learning and memory, are already upregulated in PTSD patients and they may be responsible for some symptoms of PTSD. Therefore, further upregulation of these receptors through smoking may worsen fear-related symptoms. If true, this offers a neurobiological mechanism linking animal studies showing that nicotine enhances fear-related responses and human studies suggesting that smoking worsens the symptoms of anxiety disorders. Together with studies showing that β2 nAChRs are upregulated as a response to chronic nicotine, Czermak et al.’s [223] results suggest that upregulation of β2-containing nAChRs may be responsible for the smoking-related exacerbation of PTSD symptoms.

In addition to the acquisition, consolidation, and retrieval of fear memories, several studies have investigated the effects of nicotine on safety learning, which may be informative on exposure therapy processes in humans [226–229]. For example, Elias et al. [226] investigated the effects of acute nicotine on cued fear extinction and contextual modulation of the extinction process. The results of this study showed that acute nicotine administered during extinction enhanced cued extinction and reduced renewal of cued fear in a novel context, an indicator of weakened contextual control over extinction memories. Furthermore, the same study found that when nicotine was administered during both training and extinction it impaired extinction and enhanced renewal. Although Elias et al. [226] studied acute nicotine’s involvement in contextual control over cued extinction, Kutlu and Gould [227] directly tested the effects of acute nicotine on contextual fear extinction, which is also hippocampus-dependent [230]. Our study demonstrated that acute nicotine administered during extinction impaired contextual fear extinction while not affecting general freezing behavior. In addition, Tian et al. [229] found that prior chronic nicotine administration impaired subsequent cued but not contextual fear extinction. However, nicotine administration used by Tian et al. [229] was given 2 weeks prior to the training and extinction phases and therefore, the results of this study are unlikely to be due to the direct effects of nicotine.

Using another safety learning paradigm, we also investigated the effects of acute nicotine on contextual safety discrimination [227]. The result of this study showed that acute nicotine also impaired this form of safety learning. Overall, these results suggest that nicotine differentially modulates extinction and safety learning based on the timing of the nicotine administration. Critically, several epidemiological accounts of anxiety disorders attributed the development of these disorders to impaired safety learning [231, 232]. In support, multiple studies have shown that encoding and retrieval of fear extinction memories are impaired in PTSD patients [28–30, 233]. For example, Milad et al. [30] found that although PTSD patients did not show altered extinction performance, retrieval of extinction memories was impaired in individuals with PTSD. Similarly, Garfinkel et al. [233] showed that PTSD patients showed increased fear response and enhanced amygdala activity during extinction retrieval in a “safe” context. In contrast to the results suggesting that PTSD patients show normal extinction learning performance, there are studies showing impaired cued and contextual fear extinction learning in PTSD patients [28, 234]. Also, Michael et al. [29] demonstrated that individuals with panic disorder showed resistance to fear extinction. Furthermore, in a series of experiments, Jovanovic and colleagues demonstrated that PTSD patients also exhibited diminished safety discrimination learning, where fear-potentiated startle response in individuals with PTSD was enhanced during a safety signal ([235–237]; for a review, see [238]). Therefore, our results showing impaired contextual fear extinction and safety discrimination as a result of acute nicotine administration [227, 228] suggest that smoking may further disrupt already impaired safety learning in patients with anxiety disorders and potentially prolong the course of the disorder. In support, there is evidence from human studies showing that PTSD patients showed altered contextual information processing during extinction [239]. Furthermore, Calhoun et al. [240] showed that the exaggerated fear response to the trauma context observed in PTSD patients was aggravated with smoking. Although the underlying neurobiological mechanisms of nicotine’s effects on safety learning are unknown, further investigation of these mechanisms will aid in understanding the effects of smoking on the exposure therapy processes used in anxiety disorder treatment.

Overall, in line with the human studies suggesting that smoking and withdrawal have substantial effects on anxiety disorder symptomatology, the evidence from animal studies also suggest that acute, chronic, and withdrawal from chronic nicotine have differential effects on hippocampus-dependent and hippocampus–independent fear learning as well as on extinction and safety discrimination. In addition, human and animal studies converge on the potential involvement of hippocampal β2-containing nAChRs in the effects of nicotine on PTSD symptoms. Furthermore, there is translational evidence suggesting that nicotine may impair anxiety disorder treatment processes by blocking safety learning. Therefore, translational research investigating the neurobiological mechanisms involved in the nicotinic modulation of fear learning will inform on potential treatment mechanisms to alleviate nicotine’s negative effects on anxiety disorders.

6) Involvement of nicotine and nAChRs in modulation of anxiety: Evidence from animal studies

While animal fear conditioning paradigms inform on learned fear symptoms of anxiety disorders, rodent models of anxiety-like behaviors have been widely utilized for the neurobiological and pharmacological investigations of innate anxiety symptoms [241]. As mentioned, these behavioral models usually take advantage of the innate anxiety-like behaviors of rodents such as avoidance of open spaces and model anxiety disorder symptoms such as hyperarousal. Therefore, the rodent models of anxiety have long been utilized to understand the effects of nicotine on anxiety as well as involvement of specific nAChRs in these behaviors. However, the effects of nicotine and other nicotinic agonists/antagonists on anxiety-like behavior in rodents are less clear than the effects of nicotine on fear conditioning (Table 1). For instance, there is evidence showing that subcutaneous injections of acute nicotine (0.1 mg/kg) 30 minutes prior to EPM increased anxiety-like behavior [242]. Also, there is evidence for acute nicotine’s anxiogenic effects with higher doses (0.5 and 1 mg/kg) when injected intraperitoneally (i.p.; [243]). Interestingly, Irvine et al. [242] also showed that repeated s.c. injections of nicotine, similar to chronic administration, 5 mins prior to EPM testing had anxiolytic effects after 7 days of administration, whereas the same repeated administration regimen resulted in increased anxiety when testing occurred during withdrawal 24 hrs after the last injection. This anxiogenic effect of nicotine withdrawal has been shown to be reversed by nicotine challenge [244]. In addition, Irvine et al. [245] previously showed that acute nicotine injections 5 mins before the behavioral test decreased social interaction, an anxiogenic effect. However, the same study also showed that nicotine injections 30 min before the task increased social interaction, an anxiolytic effect. Moreover, employing a social interaction task, File et al. [246] found that while the lower doses of acute nicotine (0.01, and 0.1 mg/kg) decreased anxiety, higher doses (0.5 and 1.0 mg/kg) increased anxiety. Thus, acute nicotine has differential effects on anxiety-like behavior, which may be due to the involvement of different neurobiological substrates depending on the task and the nicotine delivery method. In support, File et al. [247] found that while local nicotine infusion into both the dorsal hippocampus and lateral septum resulted in anxiogenic effects in the social interaction test, in EPM nicotine had anxiogenic effects only when infused in the lateral septum but not when injected into the dorsal hippocampus.

Table 1.

Effects of acute nicotine on anxiety

Irvine et al. (2001)	Hooded Lister rats	EPM	0.1 mg/kg s.c.	30 mins	Anxiogenic
Ouagazzal et al. (1999)	Hooded Lister rats	EPM	0.5 and 1 mg/kg i.p.	5 mins	Anxiogenic
Elliott et al., (2004)	Sprague – Dawley rats	EPM	0.1, 0.5 and 1 mg/kg s.c.	10 mins	Anxiogenic
O’Neill and Brioni (1994)	Wistar rats	EPM	0.3 mg/kg i.p.	30 mins	Anxiolytic
Ericson et al. (2000)	CD1 mice	EPM	0.35 mg/kg s.c.	Immediate	Anxiolytic
Irvine et al. (1999)	Hooded Lister rats	Social Interaction	0.1 mg/kg s.c.	30 mins	Anxiolytic
				5 mins	Anxiogenic
File et al. (1998)	Hooded Lister rats	Social Interaction	0.01 and 0.1 mg/kg i.p.	3 mins	Anxiolytic
			0.5 and 1 mg/kg i.p.		Anxiogenic

Because lower doses decrease anxiety-like behavior in the social interaction task but not in EPM, it is possible that these paradigms have different sensitivities to nicotine’s effects and/or measure different aspects of anxiety. In line with increased sensitivity of EPM to the anxiogenic effects of acute nicotine, Elliot et al. [248] found that s.c. injections of both low and high doses of nicotine (0.1, 0.5, and 1.0 mg/kg) resulted in reduced open arm duration, an anxiogenic effect, in EPM. Interestingly, there is also evidence from studies using EPM that both i.p. and s.c. acute nicotine injections decreased anxiety [244, 249]. Nevertheless, the contradictory results showing different effects of acute nicotine on EPM may be explained by different species and rat strains used in the studies. Specifically, Irvine et al. [242] and Ouagazzal et al. [243] showed the anxiogenic effects of acute nicotine using Hooded Lister rats, whereas studies showing anxiolytic effects nicotine used Wistar rats and CD1 mice [244, 249]. This suggests strain/genetic differences contribute to the effects of nicotine on anxiety. In support, differences in anxiety-like behavior between Wistar and Hooded Lister rats have been documented [250]. These studies suggest that although nicotine may be used as a mean to self-medicate anxiety-related symptoms in anxiety disorders, its effects on anxiety may be dependent on a variety of factors including the source or type of anxiety the patient experiences and genetic background. Therefore, it is possible that nicotine-based self-medication may be effective only for some individuals with specific types of anxiety and/or specific genotypes. In support there is evidence showing that the prevalence of self-medication with alcohol and other drugs among different subtypes of anxiety disorders vary from 3.3% in specific phobias and panic disorder to 18.3% in generalized anxiety disorder [72].

Specific nAChRs seem to have different roles in the effects of nicotine on anxiety (Table 2). Similar to the role of the high-affinity α4β2 nAChRs in nicotine’s effects on hippocampus-dependent fear conditioning, these nAChR subtypes also play a modulatory role in anxiety-like behavior. For example, ABT-418, an α4β2 nAChR agonist, has been shown to have anxiolytic effects in EPM while both ABT-418 and an α4β2 nAChR partial agonist, ABT-089, have been shown to reverse the anxiogenic effects observed during nicotine withdrawal [124, 251, 252]. Also, Brioni et al. [251] showed that the anxiolytic effects of ABT-418 can be reversed by the injections of non-specific nAChR antagonist mecamylamine. In addition to studies using α4β2 nAChR agonists, McGranahan and colleagues [253] showed that selective deletion of α4β2 nAChRs located on dopaminergic neurons also reversed nicotine’s anxiolytic effects. However, studies showed that KO mice lacking the β2 subtype of nAChRs showed normal levels of anxiety in EPM as well as the light/dark box and mirrored chamber tasks [254], which suggests that β2-containing nAChRs may be required for the effects of nicotine on anxiety but may not be directly involved in anxiety. In contrast, there is evidence showing that α4 KO and heterozygous mice showed increased anxiety in the EPM paradigm [255, 256]. This shows that α4-containing nAChRs may modulate anxiety.

Table 2.

Involvement of specific nAChRs in anxiety-like behavior

Brioni et al., (1994)	α4β2	α4β2 nAChR agonist, ABT-418	EPM	Anxiolytic
McGranahan et al. (2011)	α4β2	Selective α4β2 nAChR lesions	EPM	Anxiogenic
Picciotto et al. (1997)	β2	β2 KO mice	EPM / Light/Dark Box / Mirrored Chamber	No Effect
Labarca et al. (2001)	α4	α4 heterozygous mice	EPM	Anxiogenic
Ross et al. (2000)	α4	α4 KO mice	EPM	Anxiogenic
Paylor et al. (1998)	α7	α7 KO mice	Open Field	Anxiolytic
Pandya and Yakel (2013)	α7	α7-selective agonist, PNU-282987	Open Field	Anxiogenic
Yohn et al. (2014)	α7	Partial α7 nAChR agonist, ABT-107	Novelty-Induced Hypophagia	Anxiolytic
	α4β2	Partial α4β2 nAChR agonist, ABT-089		Anxiogenic

Paylor et al. found that in contrast to α4 KO mice, α7 KO mice showed decreased anxiety in EPM [257]. In line with the Paylor et al. [257] results, Pandya and Yakel [258] demonstrated that an α7-selective agonist, PNU-282987 increased anxiety-like behavior in the open field paradigm. Finally, Yohn et al. [252] found that a partial α7 nAChR agonist, ABT-107, which desensitizes and effectively inhibits further activation of α7 nAChRs, reversed the anxiogenic effects of nicotine withdrawal in a novelty-induced hypophagia paradigm, an anxiety paradigm measuring the novelty-induced reduction of feeding behavior. Also in this task, Yohn et al. [252] found that a partial agonist of α4β2 nAChR, ABT-089, resulted in the reversal of withdrawal-related anxiety but produced anxiogenic effects in nicotine naïve animals. Therefore, these results show that α4β2 and α7 nAChRs differentially mediate both anxiety and nicotine’s effects on anxiety. In sum, while elimination or inhibition of α4β2 nAChRs reverses the anxiolytic effects of nicotine, agonists at these receptors results in a decrease in anxiety-like behavior. In contrast, activation of α7 nAChRs results in an anxiogenic effect while inhibiting α7 activation reverses anxiogenic effects of nicotine withdrawal. Given that nicotine differentially binds to these two classes of nAChRs, pharmacological differences between α7 and α4β2 nAChRs may determine the outcome of nicotine administration in terms of its anxiolytic and anxiogenic effects.

7) Conclusion: Towards understanding the relationship between nicotine and anxiety disorders

In summary, the studies reviewed here suggest a strong relationship between nicotine and anxiety and fear learning, which underlie some of the most common symptoms of anxiety disorders. Human studies suggest that although smoking may be an attempt to self-medicate, it is also correlated with symptom severity. In line with these results, animal studies suggest that acute nicotine enhances hippocampus-dependent fear-learning in animals that are nicotine-naïve and this effect is greater in animals withdrawn from chronic nicotine. There is evidence from multiple studies that the enhancing effects of nicotine on hippocampus-dependent fear learning are modulated by the β2-containing nAChRs in the hippocampus. Moreover, these studies also suggest that nicotine impairs contextual fear extinction and contextual safety learning, which suggests that nicotine may reduce the effectiveness of exposure therapy and prolong the course of the disorder. Together with the human studies suggesting a positive correlation between smoking and anxiety disorder symptom severity and data showing that contextual fear is enhanced with smoking, these results clearly indicate that nicotine may worsen fear-related symptoms of anxiety disorders. This effect may be mainly mediated by the high-affinity nAChRs in the hippocampus. Although the effects of nicotine on learned fear are relatively clear, nicotine’s effects on other anxiety symptoms are variable. These effects are dependent on both α7 and α4 subtypes of nAChRs. The animal studies employing models of anxiety suggest that acute nicotine may have both anxiolytic and anxiogenic effects at different doses based on the type of anxiety (novelty-driven or social anxiety) and potentially genetic background. This may explain the different rates of self-medication prevalence observed in the subtypes of anxiety disorders. Further work in this area may aid in the development of behavioral and pharmacological interventions for anxiety disorders and nicotine addiction.