Executive and social behaviors under nicotinic receptor regulation

Abstract

Nicotine enhances several cognitive and psychomotor behaviors, and nicotinic antagonists cause impairments in tasks requiring cognitive effort. To explore the contribution of nicotinic receptors to complex cognitive functions, we developed an automated method to investigate sequential locomotor behavior in the mouse and an analysis of social behavior. We show that, in the β2^-/- mutant, the high-order spatiotemporal organization of locomotor behavior, together with conflict resolution and social interaction, is selectively dissociated from low-level, more automatic motor behaviors. Such deficits in executive functions resemble the rigid and asocial behavior found in some psychopathological disorders such as autism and attention deficit hyperactivity disorder.

Nicotine and nicotinic agonists enhance cognitive and psychomotor behaviors in various species (1–5). Conversely, nicotinic antagonists (6, 7) and loss of nicotinic acetylcholine receptors (nAChRs) under various pathological conditions impair cognitive performance (8–10). These pharmacological actions of nicotine are mediated by a variety of nAChR subtypes with different distribution patterns in the brain (11, 12). To unravel their respective contribution to brain functions, mice lacking defined nAChR subunits have been generated (for a review see ref. 13).

Brain nAChRs are pentameric oligomers composed of protein subunits arranged in various combinations of α (2–9) and β (2–4). The principal combinations that predominate in the brain are α4β2 and α7 subunits (13, 14). Mice lacking the β2 subunits show abnormal passive avoidance (15) and impaired nicotine self-administration (16) and drug discrimination (17) and exhibit a reduced nociceptive response to nicotine (18) and decreased visual acuity (19). On the other hand, general spatial memory tested in the water-maze task is not affected (9).

The aim of this article is to explore the contribution of nAChRs to complex cognitive functions referred to as executive processes. The management of these processes provides the maintenance of goal representation, the appropriate adaptation of behavior in a changing environment, the organization of sequences of actions over time, and the inhibition of prepotent or previous responses (20). Former work in humans and animals has shown that these processes require prefrontal and/or cingulate activation (21–24). Behavioral protocols known to rely on the integrity of these structures were adapted to mice.

We developed an automated analytical procedure for locomotor behavior (25) in the mouse. This method makes possible the distinction and quantitative evaluation of high-level executive components from low-level motor behavior. Furthermore, to study an index of adapted responses to a context that potentially leads to conflict resolution, we designed a procedure aimed at the distinction between several types of sequential behaviors in a social context. Interestingly, the “supervisory planning” organization of mouse locomotor behavior and conflict resolution were found selectively impaired in the β2^-/- mice challenged in different spatial learning and social paradigms. The implications of these results for human pathology and drug screening are discussed.

Methods

Animals. Thirty-six C57BL/6 and β2^-/- mutant mice were used. The animals were housed individually when food-deprived and before the social experiment and were handled during ≈10–15 days for 10 min a day before the beginning of the experiments. When food-deprived, their weights were adjusted at 90% of normal body weight. All animals had access to water ad libitum.

Behavioral Tasks. Control measures. We measured (i) the weight of WT and β2^-/- mice for several days either when animals were maintained on free food or when they were food-deprived (2 grams given daily), (ii) food intake over three 2-min blocks in a new environment, and (iii) the levels of anxiety in WT and β2^-/- mice by using the light/dark device in four groups of mice, two of which were food-deprived 1 week before the experiment. Spatial learning. During the learning period of the food location, the apparatus was equipped with four dissociable arms (north, south, east, and west) arranged in a cross, with two arms 45 cm long facing each other and two other arms 34 cm long (Fig. 1a). A cup containing the animals' usual food was presented at the end of the north arm of the maze, whereas a similar cup with unreachable food was placed at the end of the south arm. This procedure was designed to prevent the animals to be guided by the odor of the food, because it has been shown that both rats and mice primarily use olfaction to locate food. The experimental room contained several extramaze cues, and the maze was cleaned daily and between each trial to prevent olfactory intramaze cues. After habituation to the maze environment and the consumption of food inside the maze, the learning was initiated.

Fig. 1.

Spatial learning measured in a modified cross maze in which a place strategy was reinforced. (a) The location of the food reward (goal) was constant, and the position of the animal to reach the goal (west or east) was variable. (b) After completion of learning, a test phase evaluated whether mice used intramaze and/or extramaze cues to reach the goal. (c) β2^-/- mice reached the goal more quickly than WT for days 1–4 [group effect: F(1,14) = 14.9 and P = 0.0017; interaction group × session effect: F(6,84) = 4.76 and P = 0.0003; post hoc analyses (t test, df = 14): day 1 (t = 3.8 and P = 0.018), day 2 (t = 4.21 and P = 0.009), and day 3 (t = 2.46 and P = 0.028)] but showed no significant difference for subsequent days including the test phase (all P values > 0.06).

During that phase, each animal received three trials a day. The starting arm was either the west or east arm and randomly chosen such that on 2 consecutive days, each animal started by the east arm half of the time and by the west arm the other half. The south arm was never used as a starting point. Entries into the unbaited arms were scored as incorrect responses, and entries into the baited arm (north) were scored as correct responses if the mice consumed the food. The time to collect the food was also scored. The animal was allowed to eat for 10 sec. If an animal failed to reach the goal after 2 min, it was stopped and put back in the waiting cage and the maximal time (120 sec) was scored for the current trial. Learning continued until a stable time within two consecutive experimental sessions (six trials) was needed to reach the goal. Animals then were submitted to a procedure that challenged and questioned the way they were managing the task, i.e., test phase. The test consisted of reaching the goal from usual arms displaced in modified starting points. The west arm was rotated 45°, making a 160° angle with the goal arm. The east arm, which was used for the test start, was at 21° of the goal arm. The south arm was rotated 25° such that it made a 205° angle with the goal arm. The position of the goal arm in the absolute space remained unchanged (Fig. 1b).

The time to reach the goal was measured per day and statistically compared between groups with repeated-measures ANOVA (STATVIEW software) and Student's t tests. Statistical analyses emphasized group effects with two levels, WT versus knockout (KO) animals, session effect for learning phase, and the effect of the spatial test for the two groups.

Exploratory behavior and navigation. A camera, fixed to the ceiling above the circular open field, was connected to a videotrack system (View-point, Lyon, France), allowing the experimenter to record and observe the behavior out of the sight of the animals. The apparatus was cleaned between each animal. Mice were placed in the center of the empty arena, and their trajectories were recorded for 6 min. The videotrack system was set up to break down each trajectory into navigation (i.e., large movements at a speed ≥14.4 cm/sec), fast exploration (i.e., small movements at a speed between 6.8 and 14.4 cm/sec), and slow exploration (i.e., very small movements at a speed ≤6.8 cm/sec).

The behavioral measures consisted of the time spent for different types of displacements (navigation and fast and slow exploration) and the distance covered during navigation (i.e., at a speed ≥14.4 cm/sec) and exploration (i.e., at a speed ≤14.4 cm/sec). Statistical analyses were performed with repeated-measures ANOVA (STATVIEW) to evaluate the group effect (WT versus KO), the effect of the type of displacements (navigation versus fast exploration versus slow exploration) and the effect of distance covered (navigation versus exploration). Post hoc analyses were carried out with Student's t tests.

Symbolic quantification of trajectories. The two-dimensional paths in the arena were normalized and transformed such that position of the animal was defined by radial position, R(t) = [1 - (x² + y²)]^1/2. R varied from 0 (periphery of the arena) to 1 (center). A second parameter was the instantaneous velocity, i.e., the first derivative of position v(t) = [(dx/dt)² + (dy/dt)²]^1/2. These two continuous time series were converted into sequences of symbols by using thresholds. Instantaneous velocities were partitioned into two categories: those above [fast (F)] and below [slow (S)] threshold, respectively. The arena was divided into two regions: a central zone (B) and an annulus (A). The symbols were combined to obtain sequences representing four states, namely AF (periphery-fast), BF (center-fast), AS (periphery-slow), or BS (center-slow). Two-dimensional paths then were transformed into ordered sequences of transitions between these four states and quantified by using the entropy (H) of conditional probability transitions [p(.|.)], an index of the distribution of the probability of transitions from a given state to another [H(.|AS) = -Σ_j = AF,BS,BS p(j|AS) × ln p(j|AS)].

Social behavior. We conducted a social-interaction test between a test resident mouse (either WT or KO) and a social WT intruder of the same sex. Test resident mice were isolated previously for at least 4 weeks, whereas social intruders were reared in a social cage. Thirty minutes before the introduction of the intruder, the resident mouse was placed in a new, large, transparent experimental box containing clean sawdust. The social-interaction test lasted for 4 min. We divided the interactions into four major types depending on which animal initiated the social contact and the escape behaviors: (i) the social intruder animal initiates approach and escape behaviors while the resident mouse shows a neutral reaction; (ii) the resident animal initiates approach and escape behaviors while the social intruder shows a neutral reaction; (iii) the social mouse initiates the approach while the resident shows escape behavior; and (iv) the resident initiates the approach while the social mouse shows escape behavior. We measured the frequency of each interaction type within the 4-min test.

Results

nAChR β2^-/- Mice Show Improved Spatial Learning. We statistically compared the time to reach the goal in β2^-/- and WT mice over days of learning and showed (Fig. 1c) that, unexpectedly, β2^-/- mice learned the task more rapidly than WT mice. They reached the goal more quickly than WT for days 1–4. Furthermore, for β2^-/- mice, time to reach the goal decreased after the first session but not for subsequent days, including the test phase. These results suggest that although the two groups learned the task differently, the processes relying on this spatial learning (intramaze cues versus extramaze cues) are likely to be identical in both groups. It is, at a glance, quite surprising that β2^-/- mice exhibit a more efficient spatial learning (or foraging behavior) than the WT, even though these results are compatible with the apparent “improved” learning reported with the avoidance learning test (15), which becomes amplified with age (9).

To eliminate the possibility of altered reward mechanisms or a decreased anxiety, which would enhance eating behavior in a novel environment, we checked that the change in the weights of animals from both groups did not show any difference in either the free-feeding state or food-deprived state [group effect F < 1, not significant; interaction group × time: F(4, 72) = 1.96 and P = 0.11]. The measures of food consumption showed no group effect (F < 1, not significant), indicating that β2^-/- mice did not have weight- or appetite-regulation troubles that could explain a modified learning curve. In addition, WT and KO animals did not differ significantly for the time spent in the anxiogenic area of the light/dark device [no significant group effect, F < 1, not significant; no-interaction group × motivational state, F < 1, not significant] or for the number of transitions between both compartments [F(1,14) = 1.99 and P = 0.18; no-interaction group × motivational state, F < 1, not significant], indicating that the level of anxiety in KO mice is not different from that of WT mice.

nAChR β2^-/- Mice Exhibit Altered Organization of Behavior. Exploratory activity is a spontaneous behavior that does not involve any explicit reinforcement (26); it serves to gather and store spatial information, which allows allocentric coding of space, itself necessary for flexible navigational processes. A large, circular, empty, open field (see Methods) was used to compare WT and β2^-/- mice for both fast navigatory behavior and slower exploratory behavior in a food-deprived condition and under a free-food condition referred to as the “motivational state.” Statistical analyses revealed that WT and β2^-/- mice display similar locomotor reaction to food deprivation [significant group effect: F(1,18) = 6.46 and P = 0.02 for the free-food condition and F(1,18) = 10.98 and P = 0.004 and for food-deprived condition] and no effect of motivational state and no-significant-interaction group × motivational state (all P values > 0.26), but a modification in the organization of displacements was noted in β2^-/- mice. It reveals a modification of the balance between navigatory behavior (i.e., fast, large movements) devoted to acquire general information about the environment and exploratory behavior (i.e., slow, local movements) devoted to more precise investigation of the environment (Fig. 2a).

Fig. 2.

Exploratory behavior in a circular open field. (a) Balance between navigatory (Left) and exploratory (Right) behaviors is statistically different between WT and KO mice [t = 2.63 and 3.78, df = 38; P = 0.012 and 0.0005, respectively]. (b Left) Density of trajectories in the arena for WT (Upper) and β2^-/- (Lower) mice. (Right) The density of segments of slow trajectories shows accumulation in the center for WT (Upper) but not β2^-/- (Lower) mice. (c) Significant difference for entropy of conditional probability transition from the BF state between WT and KO (t test, df = 18, t = 4.05, and P = 0.0008). (d) Modification of entropy from BF indicated that β2^-/- mice favor transition from BF to AF instead of BS (indicated by dashed line). This explained the reduction of density of slow trajectories in the center of the arena observed in β2^-/- mice.

To refine this analysis, an automatic method was used to decompose mice trajectories into successions of so-called symbols representing both the level of velocity and the position in the arena of the free-behaving animal (ref. 25; see Methods). The advantage of this procedure is that the probabilities, or rates, become independent of the units used to quantify activity; thus it provides a scale-independent measure of the behavior. Four states labeled AS, AF, BS, and BF were distinguished, and the probability of transition between them was estimated (see Methods). β2^-/- mice showed a reduction of density of trajectories in the center of the arena (Fig. 2b Left) corresponding to a diminution of BS states (Fig. 2b Right). Quantification of transitions from BF indicated a displacement of equilibrium from BF to AF (Fig. 2c), meaning that β2^-/- mice tended to cross over the arena at high speed without slowing down in the center (Fig. 2d). Furthermore, when a single object was presented in the arena, the time spent in the object area and in another empty area of the same surface did not differ (group effect: F < 1, not significant), indicating that β2^-/- mice normally explored an object placed in the environment. In other words, they did not show gross sensory or memory impairment that could alter recognition processes.

To evaluate more sophisticated features of exploratory behaviors (27–29), the same apparatus was used, but two objects were placed in the environment. The results (Fig. 3a Upper) indicate that β2^-/- mice have exploratory behavior different than WT mice, but their reaction to spatial change is normal. Symbolic analysis revealed a striking reorganization of the transitions between sequences of actions in WT mice, as shown by modification of entropy from BF states, that is absent in β2^-/-. WT mice increased the number of trajectories between the two objects, a path that passes through the center of the open field. The number of BS states increased in session 3 for WT but not for KO (Fig. 3b). The reorganization observed in WT reflects the normal adaptive processes of becoming familiar with the environment, which are not exhibited by β2^-/- mice.

Fig. 3.

Exploration of two different objects in the open field. Objects remained at the same place for sessions 1–3, and their places were exchanged for the test. (a Upper) The duration of exploration of the objects' areas was not different for WT and KO animals for sessions 1 and 2 and test but differed for session 3 {repeated measures of variance for the four sessions revealed a significant group effect [F(1,10) = 6.28 and P = 0.03) and a significant session effect [F(3,30) = 3.77 and P = 0.02]}. Post hoc analyses demonstrated that the two groups differed only for session 3, i.e., before spatial change (t = 3.6, df = 10, and P = 0.005; all other P values > 0.13). (a Lower) A similar result was obtained with entropy {repeated measures of variance for the four sessions revealed a significant group effect [F(1,10) = 7.09 and P = 0.024] and a marginally significant session effect [F(3,30) = 2.8 and P = 0.057]}. Post hoc analyses demonstrated that the two groups differed only for session 3, i.e., before the spatial change (t = 3.53, df = 10, and P = 0.0054; all other P values > 0.08). WT mice exhibited habituation, which is absent in β2^-/- mice. (b) Illustration of trajectories at slow speed at the center of the open field (red points) for WT (Upper) and KO (Lower) mice for session 3.

nAChR β2^-/- Mice Display Altered Conflict Resolution. The deficit in basic exploratory behavior and the lack of flexibility of the β2^-/- mice was analyzed further in a conflict-behavior paradigm that specifically requires strategic choices. Animals were submitted to a “mild” conflict-resolution problem that consisted of increasing the normal rate of exploratory behavior by including a new and unexpected object in a maze devoted to finding a rewarding goal. The same protocol and apparatus as in the learning experiment (Fig. 1 a and b) was used except that objects were inserted inside the maze to serve as attentional distractors during learning. We measured the time to reach the food and determined that the two groups did not react in the same way to the successive changes of the maze constitution (Fig. 4a). WT mice reacted to novelty by increasing exploratory activity, and therefore, increasing the time to reach the food, whereas β2^-/- mice did not adapt their behavior to a change in the environment.

Fig. 4.

Mice were subjected to conflict-resolution situations. (a) During spatial learning, objects were inserted in the maze (see Methods for details). The two groups differed significantly during learning {group effect: F(1,12) = 13.92 and P = 0.003; session effect: [F(10,120) = 9.1 and P < 0.0001]; and interaction group × session: [F(10,120) = 2.83 and P = 0.0034]}. The post hoc Student's t test indicated that the two groups were different for the first session (t = 3.0, df = 12, and P = 0.01) and the two first sessions with objects (t = 2.57 and 3.1, df = 12, and P = 0.024 and 0.0097, respectively) but not for the subsequent sessions with the same-sized or smaller object. The time to reach the goal was different for the two groups when a bigger object was inserted within the maze (t = 2.54, df = 12, and P = 0.026) or for a challenge (Fig. 1b)(t = 2.25, df = 12, and P = 0.029). (b) Sequences of social interactions between a test resident mouse and a social intruder. The number of behavioral sequences differed significantly between KO and WT mice for the first and fourth types of sequence (t =,df = 18, and P = 0.046 and 0.0009, respectively).

We then analyzed the sequences of interactions between a resident test mouse and a social intruder of the same sex (Fig. 4b). We divided the interactions into four types, depending on which animal initiated the approach and the escape behaviors, and measured the frequency of each interaction type in 4 min (see Methods). Fig. 4b illustrates a significantly higher rate of approach behaviors and lower rates of escape behaviors in β2^-/- mice as compared with WT mice. These results indicate that social-interaction sequences are disturbed in β2^-/- mice with enhanced social interactions and an impaired capacity for interrupting ongoing behaviors.

Discussion

Previous studies with mice deleted for β2-containing nAChRs did not reveal significant impairments in spatial memory except in aged animals (9, 17, 30). The present article confirms that β2-containing nAChRs are not necessary for memory formation, food motivation, or anxiety control. To investigate a possible contribution of this subunit to more complex behaviors, we first developed an analytical model describing sequences of locomotor activities by using a symbolic decomposition of unconstrained trajectories (25) with the underlying assumption that a change in the temporal organization of the symbolic description reflects a change in the underlying sequences of behaviors. Second, we developed a conflict-situation paradigm to address the adapted behavioral response to changing external stimuli. We decomposed social interactions into different possible sequences of actions, pointing to the succession of initiation and termination of social contacts as an index of adapted behaviors.

Quantification of movement transitions indicated that WT mice gradually and spontaneously adapted their displacements from fast navigation at the periphery of an open field to slow displacements toward the center of the arena. This behavior, reflecting normal and spontaneous flexibility in WT animals, may also contribute to other behavioral changes. It is likely that, for example, the higher level of exploratory activity exhibited by WT mice contributes to the time spent exploring the maze at the beginning of learning, because it constitutes a new environment. In contrast, KO mice did not adapt their displacements over time in the open field as the environment became more familiar. This lack of exploratory interest therefore may contribute to the apparent faster spatial learning exhibited by these animals. Similarly, when confronted by objects, KO mice did not spontaneously adapt their displacements by progressively visiting the center but maintained rigid runs along the wall of the arena. Note, however, that KO mice did not show any memory or recognition impairments in this paradigm, therefore reinforcing previous results obtained with these mice (9). In a different paradigm, when confronted by mild conflict situations, KO mice did not modify their routine behavior, whereas the presence of new objects or reconfiguration of the maze elicited longer exploration periods for WT animals. KO mice did not show any sensitization to novelty and, again, did not react to drastic changes of maze configuration. In contrast, the social-interaction paradigm requires organized sequences of actions that create a high load on conflict resolution, mainly because the animal has to adapt rapidly to the unpredicted behavior of another animal. Our results show a clear distinction between WT and KO animals. Indeed, although WT mice exhibited an alternation between approach and escape behaviors, KO mice exaggeratedly initiated approach but rarely affected escape behaviors. KO mice, again, behaved in what might be called a “rigid” manner, although they were fully able to memorize the environmental characteristics and the objects therein. Altogether, these results reveal that elimination of β2-containing nAChRs causes a clear-cut dissociation between the executive organization of locomotor behavior and lower level elementary behaviors themselves, e.g., recognition, memory, or anxiety.

The temporal dynamics of exploratory behaviors, the adaptive balance between different components of displacements, and the succession and duration of contacts between conspecifics are processes commonly referred to as “executive functions,” within which planning plays a crucial role (31). They mobilize, in the primate and human brain, areas such as the dorsolateral prefrontal cortex, the anterior and posterior cingulate, and the posterior parietal cortex (32, 33). In rats, planning and flexible behaviors also rely on prefrontal cortex integrity (21, 22, 34–36). Such “supervisory” executive processes (37) may plausibly mobilize sets of neurons with long-range axons belonging to a global workspace. These “global-workspace” neurons differentially modulate specialized perceptual, motor, long-term memory, evaluative, and attentional processors in a top-down manner (38, 39). Neurons of this type are abundant in prefrontal and cingulate cortices. Action monitoring, behavioral shifting, and temporal organization of action sequences are anticipated to enhance the activation of these territories (23, 40). Indeed, human brain-imaging studies show activation of prefrontal and cingulate cortices in cognitive tasks involving conflict (e.g., ref. 41), suggesting that the prefrontal and anterior cingulate cortices are both engaged during effortful conflict resolution and action monitoring during effortful task resolution (23, 42, 43). Such situations are still difficult to investigate in mice by brain-imaging techniques. Yet, the present analysis reveals that mice do far more than simply react to sensory information. They may engage in complex and extended behaviors geared toward far-removed goals: They use processes that override, or differentially select, routines and orchestrate locomotor behaviors according to defined intentions or plans.

Our results further demonstrate that acetylcholine binding to high-affinity, widespread β2-containing nAChRs gates this integrative process and, in addition, reconciles puzzling results concerning the global role of nicotine in cognitive functions. This pharmacological action of nicotine as a psychostimulant is an important component of the process of tobacco addiction (44). Conversely, it is of potential importance for the design of nicotinic agents acting on psychiatric or neurologic diseases that hit nAChRs. Various cognitive dysfunctions such as those associated with Alzheimer's disease (45), schizophrenia (46), and autism (10) have been reported to differentially affect diverse subtypes of nicotinic receptors. The phenotypes observed with the β2^-/- mice are characterized by alterations of behavioral flexibility and adaptive behaviors coupled with unimpaired memory and anxiety. We thus may suggest that they reproduce some of the cognitive impairments observed in autism and attention deficit hyperactivity disorder (ADHD) (47–49). This interpretation is supported by the following lines of evidence: (i) the behavior of β2^-/- mice resembles that exhibited by rats with lesions of the prefrontal and cingulate cortices and of the amygdala; (ii) autistic and ADHD patients show a reduced activation of these brain structures (47, 50); and (iii) the relative proportion of α4β2-containing nicotinic receptors is substantially decreased in the frontal cortex of autistic patients (10), whereas nicotine treatment improves ADHD symptoms (51). The β2^-/- mice thus might represent a useful animal model for the study of autism and ADHD.

The automatic procedure to analyze locomotor behavior, together with the detailed analyses of the sequences of social behavior, proved useful to pin down cognitive deficits of β2^-/- mice. These behavioral procedures might be extended to other nAChR mutant mice (13) and help define their respective contribution to cognitive function, and dysfunction, a critical knowledge required for the development of relevant nicotinic medications.