The prefrontal cortex, a cortical area essential for working memory and higher cognitive functions, is modulated by a number of neurotransmitter systems, including acetylcholine; however, the impact of cholinergic transmission on prefrontal activity is not well understood. We relied on systemic administration of a muscarinic receptor antagonist, scopolamine, to investigate the role of acetylcholine on primate prefrontal neuronal activity during execution of working memory tasks and recorded neuronal activity with chronic electrode arrays and single electrodes. Our results indicated a dose-dependent decrease in behavioral performance after scopolamine administration in all the working memory tasks we tested. The effect could not be accounted for by deficits in visual processing, eye movement responses, or attention, because the animals performed a visually guided saccade task virtually error free, and errors to distracting stimuli were not increased. Performance degradation under scopolamine was accompanied by decreased firing rate of the same cortical sites during the delay period of the task and decreased selectivity for the spatial location of the stimuli. These results demonstrate that muscarinic blockade impairs performance in working memory tasks and prefrontal activity mediating working memory.
- persistent activity
- principal sulcus
the prefrontal cortex plays an important role in executive function, allowing the control of behavior in accordance to prior information and future goals (Miller and Cohen 2001). Accordingly, neural correlates of processes such as working memory, decision-making, and planning have been observed in neurophysiological studies of primates trained to perform behavioral tasks (Barraclough et al. 2004; Funahashi et al. 1989; Fuster and Alexander 1971; Kim and Shadlen 1999; Mansouri et al. 2007; Wallis et al. 2001). Importantly, the precise level of firing rate of prefrontal neurons has been shown to correlate with psychophysical measures of these functions (Constantinidis et al. 2001; Kim and Shadlen 1999). A number of neuromodulator systems are known to regulate both executive functioning and the activity of prefrontal neurons, including dopamine, norepinephrine, and acetylcholine (Arnsten and Li 2005; Robbins and Arnsten 2009). The acetylcholine system in particular has been implicated in the profound deficits in cognitive function associated with aging, dementia, and Alzheimer's disease (Bartus et al. 1982; Ebert and Kirch 1998; Heise 1984; Mesulam 2004; Sarter and Parikh 2005). Decrease in the action of acetylcholine, for example, by administration of muscarinic receptor antagonists, impairs cognitive abilities in humans and animals (Bartus and Johnson 1976; Buccafusco et al. 2007; Deutsch 1971; Drachman 1977; Everitt and Robbins 1997). Conversely, enhancement of acetylcholine levels through systemic administration of acetylcholinesterase inhibitors is used widely for the treatment of decline in executive function in human patients (Birks and Harvey 2003; Courtney et al. 2004; Doody et al. 2001; Rogers et al. 1998; Winblad et al. 2001). Such drugs also have been shown to improve performance in delayed matching tasks in young adult and older monkeys (Buccafusco and Terry 2004; Terry and Buccafusco 2003). Enhancement of acetylcholine levels by stimulation of the nucleus basalis also improves capacity for executive function in humans (Freund et al. 2009).
Few studies have tested effects of acetylcholine on activity of cortical neurons during execution of cognitive tasks in primates. The effects of acetylcholine levels on neuronal activity have been studied more extensively in the context of plasticity: cholinergic depletion blocks cortical plasticity (Juliano et al. 1991; Webster et al. 1991), whereas an increase in acetylcholine levels or direct iontophoretic application can induce plastic changes (Bakin and Weinberger 1996; Froemke et al. 2007; Kilgard and Merzenich 1998; Metherate et al. 1988). One of the few studies that has addressed the role of acetylcholine on executive function in monkeys reported that scopolamine degrades behavioral performance in a working memory task but, paradoxically, causes an increase of neuronal discharges during the presentation of stimuli in the monkey inferior parietal cortex (Miller and Desimone 1993). Evidence from human functional imaging studies suggests that the prefrontal cortex may be a critical site of action of acetylcholine, since acetylcholinesterase inhibitors can affect the patterns of prefrontal cortical activation during working memory (Furey et al. 2000). However, the exact effects of acetylcholine application on prefrontal neuronal discharges are unknown.
We relied on scopolamine to block acetylcholine actions in the central nervous system (CNS). Scopolamine is an alkaloid muscarinic receptor antagonist, nonspecific for the M1 and M2 families of muscarinic receptors that comprise M1, M3, M5 and M2, M4 receptors, respectively (Bolden et al. 1992; Klinkenberg and Blokland 2010). The primary CNS effects attributable to the muscarinic receptors are changes in EEG activation patterns and cortical activity that endogenously accompany changes in alertness or arousal (Ebert and Kirch 1998). The stimulation of the ascending reticular pathways or the nucleus basalis alters EEG activation patterns to suppress low-frequency activity and enhance higher frequencies (Metherate et al. 1992; Steriade and Amzica 1996), and these effects are blocked by muscarinic antagonists. Our experiments tested how systemic blockade of acetylcholine action affects neuronal activation in the monkey lateral prefrontal cortex as it relates to performance of visual working memory tasks.
Three male rhesus monkeys (Macaca mulatta) weighing 5–12 kg were used in this study. All animal experiments were performed in compliance with the guidelines set forth by the National Institutes of Health, as reviewed and approved by the Wake Forest University Institutional Animal Care and Use Committee. Data analysis was performed using the MATLAB computational environment (The MathWorks, Natick, MA).
Details of the experimental setup have been published previously (Qi et al. 2010). Briefly, the monkeys sat in a primate chair with their head restrained and viewed a computer screen positioned 60 or 68 cm away. They were required to maintain fixation while visual stimuli were presented on the screen. An infrared tracking system (ISCAN, Burlington, MA) sampled eye position at 240 Hz. Correct completion of a trial resulted in delivery of a liquid reward (fruit juice). In-house software (Meyer and Constantinidis 2005) was used to monitor behavior online.
We trained two monkeys to perform a variant of a delayed response task (Fig. 1, A–C). This was a spatial working memory task, which required them to remember the spatial location of a cue stimulus flashed on a screen for 0.5 s. The cue was a 3° white square stimulus that could appear at one of eight locations arranged on a circle of 12° eccentricity. After a 3.5-s delay period, the fixation point was extinguished and the monkey was trained to make an eye movement to the remembered location of the cue within 0.6 s. The saccade needed to terminate on a 6° radius window centered on the stimulus (within 2–3° from the edge of the stimulus), and the monkey was required to hold fixation within this window for 0.2 s. Zero, one, or two distractors could appear between the cue and the saccade, with equal probability, although the total duration of the trial was the same for all trial types. Distractors were incorporated in the delayed response task so that we could test the effects of scopolamine at versions of the task of varying difficulty in terms of attentional demands. Breaking fixation at any point before the offset of the fixation point aborted the trial and resulted in no reward. These errors were also excluded from analysis.
As a control, the same animals were trained to perform a visually guided saccade task. This used the same size stimulus, presented at the same eccentricity as in the delayed response task. However, in this task the fixation point disappeared simultaneously with the appearance of the cue, and the monkey was rewarded for making a saccade to the cue and for maintaining fixation for 0.2 s. The duration of trials in this task was much shorter, and we were able to collect more than one set of trials in each daily session (typically 2) to approximate the same total length of the data collection session in the two tasks.
One monkey was trained in a match/nonmatch task (Fig. 1D). This was also a spatial working memory task, which required animals to remember the spatial location of one stimulus flashed on the screen, to observe a second stimulus presented after a delay period, and to indicate whether the two stimuli appeared at the same or different locations by making a saccade toward a green or blue target, respectively, which appeared after a second delay period (Qi et al. 2011). Stimuli for this experiment appeared on a 3 × 3 grid of 10° distance between each other.
A saline solution of scopolamine was administered intramuscularly 30–60 min before the beginning of a recording session. Scopolamine crosses the blood-brain barrier with peak CNS action between 30 min and 3 h after administration and with peak peripheral action in the first 30 min (Renner et al. 2005). Effective dosage was determined separately for each monkey during an initial period before the collection of neurophysiological data. The doses used were 5 and 10 μg/kg, 2.5 and 5 μg/kg, and 10 μg/kg for the three animals, respectively (see Fig. 2). Scopolamine was administered 1 or 2 days a week, separated by at least a 48-h period. Control data were collected on the days between scopolamine administrations.
Surgery and neurophysiology.
A chronic microelectrode array (Spingath et al. 2011) was implanted on the lateral prefrontal cortex of two monkeys (see Fig. 4). The implant comprised 64 parylene-c insulated, Pt/Ir, or pure Iridium microelectrodes arranged on a grid, spaced 0.75 mm apart from each other, thus covering an ∼5 × 5-mm surface area. During implantation surgery, a <20-mm-diameter craniotomy was performed and a titanium holder was anchored onto the skull with bone screws. Subsequently, the dura under the craniotomy was removed to expose the underlying cortex. The implant core was lowered into the titanium holder, and the electrodes were advanced into the cortex. The implant's design allowed repositioning of the electrodes after implantation. Neuronal recordings were obtained through a set of connectors attached on a circuit board, from up to 16 channels at a time. The electrical signal from each electrode was amplified, band-pass filtered between 500 and 7 kHz, and recorded with a data acquisition system (APM system; FHC, Bowdoin, ME). Waveforms were sampled at 25-μs resolution, digitized, and stored.
Neurophysiological recordings with single electrodes were performed in one monkey as described previously (Qi et al. 2010). Briefly, a 20-mm-diameter craniotomy was performed over the prefrontal cortex, and a recording cylinder was implanted. We used glass-coated tungsten electrodes of 250 μm in diameter with an impedance of 1 MΩ at 1 kHz (Alpha-Omega Engineering, Nazareth, Israel). Arrays of up to 8 microelectrodes spaced 0.2–1.5 mm apart were advanced into the cortex through the dura with a microdrive system (EPS drive; Alpha-Omega Engineering).
Behavioral trial analysis.
To analyze the patterns of eye movements in the delayed response task, we first determined the end point of saccades performed after the fixation point was turned off. Trials resulting in fixation errors were omitted from this and any subsequent analysis. During the 0.6-s interval following the offset of the fixation point, we determined the eye position corresponding to the time point of maximum eye velocity. The end point of the saccade was determined as the eye position at the last point of monotonic deviation away from the fixation point, following the point of maximum velocity. We compared a number of eye movement variables in control and scopolamine sessions: Response latency was defined as the interval between the offset of the fixation point and the time that eye position entered the target window. Saccadic error was defined as the standard deviation of the saccadic end point around its mean position for each target. Peak saccadic velocity was determined from the maximum eye position distance between successive samples during the saccade period, smoothed with an 8-ms triangular filter, and resampled at 4 ms.
We classified error trials in four types: 1) those with eye position end points falling within 4° from the fixation point, indicating that the monkey did not perform a saccade at all; 2) those falling within 6° of the edge of the cue, suggesting a saccade in the correct direction, or a correct saccade that was not followed by a 0.2 fixation of the target; 3) those falling within 6° of a distractor; and 4) a saccade in any other direction. Error trials in the match/nonmatch task were of only a single type, indicating that the monkey saccaded toward the blue target after the appearance of a match stimulus or the green target after the appearance of a nonmatch. Errors due to breaks in fixation were also omitted from analysis.
Neuron data analysis.
Multiunit records were extracted from each electrode of the chronic implant, using a fixed threshold of 3 standard deviations from the signal baseline. Firing rate of multiunits was then determined for each of the task epochs in each trial. Analysis of single-electrode recordings was performed as described previously (Qi et al. 2010). Briefly, action potential waveforms were sorted into separate units using an automated cluster analysis method based on the KlustaKwik algorithm (Harris et al. 2000).
We identified multiunit records selective to the task, evidenced by significant increase in firing rate during a stimulus presentation (paired t-test, P < 0.05). We further selected units with robust delay period activity by averaging responses from all multiunit records with visual responses recorded from each electrode (with no respect to whether they were collected during a control or scopolamine sessions), rank ordering sites with respect to mean firing rate in the delay period following the best cue location, and selecting the top quartile of sites. Analysis was only performed for electrode sites where task-selective recordings were obtained from both the control and scopolamine sessions. Since unequal numbers of recordings were obtained from the scopolamine (n = 26) and control conditions (n = 46), we used all possible pairs, always matching recordings from the same electrode (n = 114). All analyses and comparisons of responses were performed for spike counts from multiunit records paired in this fashion. Peristimulus time histograms (PSTHs) were obtained by averaging responses, after each spike train was converted to a spike-density function by being convolved with a Gaussian kernel of 80-ms standard deviation. A cumulative spike count measure was calculated by integrating PSTH rates across the time course of the trial, after the average response rate computed in the fixation period was subtracted.
Firing rate of single units obtained with movable electrodes was computed in each task epoch and compared with the baseline firing rate recording during the fixation interval. Neurons responding to the stimuli were selected for further analysis, evidenced by a significant increase in firing rate above the fixation period (paired t-test, P < 0.05). Population PSTHs and cumulative spike counts were constructed as described above. Neural data from trials resulting in correct behavioral responses are only presented in the article.
We administered scopolamine systemically to two monkeys trained to perform a spatial working memory task (Fig. 1, A–C) and collected neural data with a chronic implant. Each monkey was tested with two doses, determined in a series of experiments before implantation. The task was a variant of the delayed response task (Funahashi et al. 1989), requiring the animals to remember the spatial location of a stimulus presented on the screen and, after a delay period, to perform a saccade toward the remembered location. Zero, one, or two distracting stimuli appeared after the cue in trials that were randomly interleaved. The total duration was the same for all types of trials.
Scopolamine degraded behavioral performance in the delayed response task. We collected data from 56 sessions of scopolamine administration while recording neural data: 12 and 19 sessions were obtained under a low dose and 13 and 12 sessions under a high dose for the two monkeys, respectively. These were compared with 69 and 56 control sessions. We evaluated performance in the control and scopolamine sessions by determining the percentage of correct trials, eliminating trials in which the animal broke fixation before the end of the last delay period. Scopolamine decreased performance in a dose-dependent manner (Fig. 2). The effect was highly significant in each monkey (2-way ANOVA, main effect of dose, P < 10−5). There was no significant interaction between number of distractors and dose (2-way ANOVA, P > 0.4 for each of the monkeys), suggesting that scopolamine caused an equal decrement of performance for each task type.
Scopolamine can cause a number of peripheral side effects, most importantly mydriasis (pupil dilation) caused by the blockade of muscarinic receptors in the peripheral nervous system. To minimize effects on visual stimulation, we conducted all experiments in darkness with only the monitor providing ambient illumination. Furthermore, to ensure that errors in the task under scopolamine administration were not caused by inaccuracies in visual perception, motor response, or other nonspecific effects, we trained the same animals to perform a visually guided saccade task involving identical stimulus presentations. We obtained 7 and 12 sets of trials at a low dose and 8 and 9 sets of trials at a high dose of scopolamine for the two monkeys, respectively. These were compared with 22 and 31 control sets. Performance under scopolamine for any monkey and dose was within 3% of control performance (Fig. 2). A two-way ANOVA comparing performance in the one-stimulus delayed response task and visually guided saccade task (which involved identical stimulus presentation) showed a significant interaction between scopolamine dose and task (P < 0.001 for both animals), indicating that scopolamine disproportionally impacted the working memory task.
To gain insights on the nature of effects caused by scopolamine administration, we analyzed a number of eye movement metrics in correct trials of the delayed response task and compared data from control and scopolamine sessions. We observed subtle but statistically significant changes in mean response latency, which increased from 272 to 277 ms during scopolamine administration and saccadic error (standard deviation of saccade end point), which also increased from 1.66° to 1.82° (t-test, P < 0.005 for both measures). On the other hand, peak saccadic velocity was not significantly different between control and scopolamine sessions (t-test, P > 0.1).
We also distinguished between four error types in the delayed response task, depending on the end point of eye movements following the offset of the fixation point (see methods). These were 1) small eye deviations from the fixation point, suggesting no behavioral choice, 2) eye movements in the direction of the cue, suggesting correct recall of the cue location but inaccurate saccade, 3) eye movements in the direction of a distractor, suggesting susceptibility to the interfering effect of the distractor, and 4) random eye movements. During control trials, the monkeys made approximately equal numbers of the four types. Unexpectedly, during scopolamine administration, error rates increased for all the categories except for type 3, suggesting that the monkeys were no more distracted by the appearance of additional stimuli after the cue (Fig. 3). The number of errors expected in each type, assuming that control and scopolamine conditions were drawn from the same underlying distribution, was significantly different from the number of errors we observed (χ2 test, P < 10−5).
Chronic recordings in delayed response task.
While performing the task, neuronal recordings were collected from the two monkeys with a chronic electrode implant (Fig. 4). Multiunit activity was collected from each electrode. We identified 200 recordings that exhibited task-related effects during the control condition and 106 during scopolamine administration from 47 cortical sites (black circles in Fig. 4). Of those, we identified 12 sites with robust delay period activity (filled circles in Fig. 4). We then compared activity in the same sites during the control experiments, when task-related responses were also recorded. A total of 46 recordings in the control condition were identified in this way and were compared with 26 recordings under scopolamine. Of the scopolamine recordings, 17 were recorded with the low dose and 9 with the high dose, 15 and 11 from the two monkeys, respectively. Differences in firing rate were not evident between recordings with low and high doses of scopolamine or between the two monkeys, and data were pooled together to increase the power of statistical comparisons with the control condition.
Scopolamine administration affected neuronal responses, mainly by decreasing delay period activity, an effect that was evident when we compared the averaged firing rate from the control and scopolamine sessions of the same cortical sites (Fig. 5, A and B). Since baseline firing rate differed slightly between conditions, we compared average firing rates after subtracting baseline responses. Delay period activity following the best stimulus location was significantly lower in the scopolamine than the control condition (t-test, P < 0.001). This resulted in lower cumulative discharge under scopolamine (Fig. 5C).
Similar effects of scopolamine were observed in the delay period of single-unit responses tested with the match/nonmatch task. To ensure that these effects of scopolamine were not specific to the behavioral task used and to neuron selection with the chronically implanted electrodes, we collected data from a third monkey trained on a spatial match/nonmatch task (Fig. 1D). Neuronal recordings were obtained using movable microelectrodes that were advanced in the cortex and removed at the end of each recording session. Single units were identified in these recordings. We compared responses of dorsolateral prefrontal neurons responding to the visual stimuli during scopolamine administration (n = 51) and during control sessions (n = 236), obtained at different cortical sites, which were sampled randomly. Averaged responses are shown in Fig. 5, D–F. As was the case in the chronic experiment, we saw a significant decrease in firing rate during the delay period following the best cue location, after subtracting the baseline firing rate (t-test, P < 0.01).
In this experiment we also observed a second type of delay period activity: anticipatory activity, rising before appearance of the second stimulus (Fig. 6). This activity was typically highest after the cue had appeared out of the receptive field, before appearance of a match or nonmatch stimulus, whose location could not be predicted from trial to trial (Qi et al. 2010). Scopolamine administration produced a significant decrease of anticipatory activity, as well. The mean firing rate observed before appearance of the second stimulus in the receptive field (Fig. 6, C and D) or out of the receptive field (not shown), after the baseline firing rate was subtracted, was significantly lower in the scopolamine condition (t-test, P < 0.05).
Analysis of single neuron recordings also allowed us to determine the effects of scopolamine on directional selectivity of prefrontal responses. We computed a selectivity index defined as (Max − Min)/(Max + Min), comparing average firing rates with the best (Max) and worst (Min) location (Meyer et al. 2011). This index was computed separately in the cue period and delay period for each neuron and compared across conditions. We found that scopolamine was associated with a significant decrease of the average selectivity index in the delay period (from 0.36 in the control to 0.29 in the delay period; t-test, P < 0.05). In contrast, mean selectivity indexes computed in the cue period were not significantly different (0.50 vs. 0.45; t-test, P > 0.2). The results indicate that scopolamine reduced stimulus selectivity specifically in the delay period.
Our study demonstrates that systemic application of scopolamine, a muscarinic acetylcholine antagonist, decreases delay period firing rate in the dorsolateral prefrontal cortex in concert with degrading behavioral performance in working memory tasks in primates. Neural correlates of a series of cognitive functions have been observed in the activity of prefrontal neurons (Constantinidis and Procyk 2004; Goldman-Rakic 1995), and diminished firing rate in lateral prefrontal cortex has been associated with lower levels of performance in working memory and decision making tasks (Constantinidis et al. 2001; Kim and Shadlen 1999). Changes in the patterns of lateral prefrontal activity also have been suggested for schizophrenia and other mental illnesses (Rolls et al. 2008). Our finding demonstrates that effects on working memory and executive function by cholinergic modulation also have a neural correlate in the lateral prefrontal cortex.
Behavioral effects of scopolamine.
Scopolamine administration increased the percentage of errors in the working memory tasks in a dose-dependent manner, as previously reported (Buccafusco et al. 2007; Everitt and Robbins 1997). Previous studies have revealed a significant effect of muscarinic antagonists on working memory accuracy for both auditory and visual tasks, and lack of an interaction with delay period length (Chudasama et al. 2004; Penetar and McDonough 1983; Plakke et al. 2008; Taffe et al. 1999; Yamamoto et al. 2011). If, however, accuracy at zero-delay length trials is also compared with variable longer delays, an effect of delay length is found (Buccafusco et al. 2007). These findings suggest that the actions of scopolamine are specific to a working memory task and that the deficit caused by scopolamine relevant to task performance is most likely in the conversion of the sensory information into a working memory trace. The effect we observed was indeed evident in all working memory tasks used, including delayed response trials with no distractors, one distractor, or two distractors, as well as in the match/nonmatch task. The effect of scopolamine on error rate was uniform regardless of the presence of distractors in the trials, resulting in a downward shift of performance curves with no significant interaction (Fig. 2). The pattern of errors could not be explained by a mere decline in visual acuity, e.g., due to possible mydriasis caused by the blockade of muscarinic receptors in the peripheral nervous system, since inaccurate eye movements directed in the general direction of the cue were not impacted disproportionately (Fig. 3) and the monkey performed a visually guided saccade essentially error free (Fig. 2, A and B). The increased errors we observed in the delayed response task were not specifically directed to the distractor, either.
Our results are consistent with the neural effects of acetylcholine described in other cortical areas. The main source of acetylcholine projecting to the cerebral cortex and thalamus is the nucleus basalis of Meynert (Everitt and Robbins 1997). Increased activity of individual neurons in this region is associated with reinforcement and arousal (Richardson and DeLong 1986, 1990, 1991). Cholinergic innervation is diffuse across the cerebral cortex, allowing the system to control the excitability of most cortical areas; however, the modulation may be specific for particular areas and stimuli (Sarter and Parikh 2005). The stimulation of nucleus basalis leads to cortical depolarization and a shift in membrane potential oscillations reflective of the EEG changes, shifting to a state with more low-frequency energy and less neuronal depolarization; these changes are also blocked by muscarinic antagonists (Metherate et al. 1992).
Neural effects of scopolamine have been described recently in primate area V1, in the context of an attention paradigm. Firing rate of V1 neurons increases when attention is directed at the receptive field, an effect that is blocked by scopolamine (Herrero et al. 2008). The effect suggests a top-down influence, possibly from the prefrontal cortex, and our results indeed suggest that prefrontal activity is diminished after scopolamine administration. On the other hand, the change in prefrontal activity that we observed cannot be explained as an effect of impaired visual processing on prefrontal activity itself. Monkeys were able to perform a visually guided saccade task at high levels of performance, and error rates in the delay response task under scopolamine administration did not increase disproportionately for trials that involved one or two distractors.
Beyond the primary visual cortex, there has been little evidence of cholinergic modulation of association areas in primates. A previous study in the inferior temporal cortex reported a decrease in performance in a working memory task but a seemingly paradoxical increase in firing rate during the stimulus presentation under scopolamine administration (Miller and Desimone 1993). Our present results revealed an effect that can account for the behavioral effects observed, namely, a decrease in firing rate specifically during the delay period of the working memory task, while responses to the stimuli were relatively unaffected and neuronal selectivity for stimulus location remained unchanged. Our results illustrate the neural basis of cholinergic modulation of executive function.
This work was supported by National Institute of Mental Health Grant R21 MH083209.
No conflicts of interest, financial or otherwise, are declared by the author(s).
We thank Fumi Katsuki for help with experiments and Alvin Terry and Albert Compte for helpful comments.
- Copyright © 2011 the American Physiological Society