|Home | About | Journals | Submit | Contact Us | Français|
To develop a rodent model of the attentional dysfunction caused by sleep loss.
The attentional performance of rats was assessed after 4, 7, and 10 hours of total sleep deprivation on a 5-choice serial reaction time task, in which rats detect and respond to brief visual stimuli.
The rats were housed, sleep deprived, and behaviorally tested in a controlled laboratory setting.
Ten male Long-Evans rats were used in the study.
Rats were trained to criteria and subsequently tested in daily sessions of 100 trials at approximately 4:00 pm (lights on 8:00 am-8:00 pm). Attentional performance was measured after 4, 7, 10 hours of total sleep deprivation induced by gentle handling.
Sleep deprivation produced a monotonic increase in response latencies across the 4-hour, 7-hour, and 10-hour deprivations. Sleep deprivation also led to increased omission errors, but the overall number of perseverative and premature responses was unchanged. Subgroups of rats were differentially affected in the number of omission errors and perseverative responses.
The effects of sleep deprivation on rats are compatible with a range of human findings on the effects of sleepiness on selective attention, psychomotor vigilance, and behavioral control. Rats also exhibited differential susceptibility to the effects of sleep deprivation, consistent with ‘trait-like’ susceptibility that has been found in humans. These findings indicate the feasibility of using the 5-choice serial reaction time task as an animal model for investigating the direct links between homeostatic sleep mechanisms and resulting attentional impairments within a single animal subject.
THE EFFECTS OF SLEEP DEPRIVATION ON BEHAVIOR AND COGNITION HAVE GENERALLY BEEN UNDERSTOOD IN TERMS OF A HOMEOSTATIC PROCESS that increases the drive for sleep with extended wakefulness, a process whose neurochemical origins have recently begun to be elucidated.1-3 As decrements in vigilance and attention are among the first signs of sleep loss and sleepiness, it has been hypothesized that common neural systems are involved in the processes of homeostatic sleep drive and sustained attention.4 For example, the activity of the basal forebrain cholinergic system has been proposed to promote cortical arousal and wakefulness5-6 as well as attention.7-10 Furthermore, we and others have hypothesized that an inhibition of this cortically projecting cholinergic system mediates the sleepiness associated with prolonged wakefulness. Although the majority of biologic studies investigating sleep processes have used animal models, we only know of 1 study that has used an attentional measure (false alarms) in a behavioral assessment of sleep-deprived animals.11 Hence, the goal of the present study was to develop a rodent model of the attentional dysfunction caused by sleep loss that could be used to elucidate the effects of sleep drive on neurobiologic processes of attention.
The effect of sleep deprivation on attention has been well documented in humans. Total sleep loss in a single night leads to impairments in sustained attention tasks that require selective attention to an auditory channel12 and psychomotor vigilance to unpredictable visual stimuli.13-15 Sleep deprivation also reduces performance when stimulus demands are higher, such as tasks that divide attention between arithmetic and verbal information.16 More recently, it has been suggested that the pattern of deficits resulting from sleep loss reflects a dysregulation of behavioral control processes that rely on the prefrontal cortex,15,17,18 such as the attention to relevant cues,19 flexible thinking,20 and cognitive perseveration.21,22 Total sleep deprivation, however, spares some aspects of executive attention, such as task shifting, response inhibition,23 and tests of cognitive flexibility.24 The heterogeneous effects of sleep deprivation on attentional performance indicate that sleep deprivation has different effects on underlying mechanisms of attention. Thus, an important feature of an animal model would entail the ability to measure the separate effects of sleep deprivation on behavioral measures that assess vigilance, selective attention, and behavioral control.
The current study assessed the consequences of 4, 7, or 10 hours of total sleep loss in a well-developed rodent model of selective attention, the 5-choice serial reaction time task (5CSRTT). The task was designed to assess major aspects of attentional control by measuring the ability of rats to continually monitor the location of a brief visual stimulus presented in one of five spatial locations.25,26 The task is based on a 5CSRTT for humans that has been used assess the effects of stressors on attention27,28 and it bears important similarities to several human sustained-attention tasks, including continuous performance test and the Psychomotor Vigilance Test (PVT).29 The 5CSRTT uses operational measures to assess the breakdown of performance according to different demands of controlled attention, including response latencies, omission errors, commission errors, impulsive responding, and perseverative responding.
A considerable benefit of the 5CSRTT is that it has been used extensively in studies that have investigated the neurobiologic mechanisms of attention in rats (reviewed by Robbins, 200226). Thus, there is an established literature that has systematically outlined the effects of neurochemical and anatomic alterations on the behavioral measures of the task. These investigations have included studies of monoamine systems that contribute to homeostatic sleep onset and cortical structures that have been associated with impaired attentional processing in sleep-deprived humans, such the parietal and frontal cortex.30
Importantly, the rodent 5CSRTT uses measures that are analogous to measures in vigilance tasks that are affected by even a few hours of total sleep loss in humans, such as slowed reaction times and omission errors,14 which are called “lapses” in the PVT literature.15 The separate measures of inhibitory control could also allow a nuanced analysis of executive dysfunction that has been hypothesized to result from sleep loss. Whereas premature responses have been attributed to an impulsive disturbance in preparatory mechanisms, perseveration has been attributed to a compulsive inability to disengage from responding once it has been initiated. Much like humans, these distinct impairments of executive control have been shown to result from the disruption of separate anatomic regions of the rodent prefrontal cortex in the 5CSRTT and other behavioral tests.31,32
All procedures and animal care adhered strictly to Association for Assessment and Accreditation of Laboratory Animal Care International, Society for Neuroscience, and institutional guidelines for experimental animal health, safety, and comfort. Ten male Long-Evans hooded rats (Charles River, Raleigh NC) began behavioral testing. They were maintained in a temperature-controlled room with a 12-hour light-dark cycle (lights on 8:00 am - 8:00 pm). Water was available ad libitum, but rats were food restricted to maintain body weight at 85% of their free-feeding weight for motivation by a food reward.
Rats were tested using two 5-hole operant chambers (Cambridge Cognition, Cambridge, UK). Each aluminum chamber was 25 × 25 × 25 cm and was equipped with a Plexiglas flap door that allowed access to a magazine where 45-mg Noyes sucrose pellets were mechanically dispensed (Research Diets Inc., New Brunswick, NJ). Five evenly spaced apertures were located on the opposite wall of the chamber that registered nose pokes via the interruption of infrared photocell beams. At the rear of each aperture was a 3-watt bulb that served as a stimulus. A 3-watt bulb at the top of the chamber provided illumination (the house light). Each behavioral chamber was contained in a larger sound-attenuating chamber with a small fan that provided ventilation and background white noise. All stimulus presentations and behavioral measures were made with Cambridge Cognition Control 1.17 software on a PC Pentium IV class computer.
The 5CSRTT has been described previously.25,26 Rats were trained to nose poke in an aperture following the brief presentation of a visual stimulus within that aperture (Figure 1). Each stimulus was presented pseudorandomly in 1 of the 5 apertures and was thus spatially unpredictable. The rats were initially allowed 60 seconds to respond to each stimulus and were rewarded with a sucrose pellet immediately following a response into the illuminated port. Upon attaining 70% correct responses on 2 consecutive days, the stimulus duration was shortened until the rats reattained 70% accuracy. The stimulus was shortened repeatedly in this fashion until the rats were able to respond correctly within 3 seconds to a 500-millisecond stimulus for 70 of 100 trials (8 rats reached this criterion).
The measures obtained in the 5CSRTT include the following 5 measures (Figure 2): (1) correct response—defined as a nose poke response into the aperture in which the stimulus was just presented; (2) incorrect response—a response into an aperture in which the stimulus was not presented; (3) omission error—when no response was made in the 3-second period after the presentation of the stimulus; (4) premature response—when a response was made during the 5-second intertrial interval prior to the presentation of a stimulus; (5) perseverative response—when 2 or more responses were made into an aperture after a single stimulus.
To facilitate learning, the house light was turned off for 5 seconds immediately following an incorrect response, an omission, or a premature response (a time out, Figure 2). After a correct response or a time out, the rat initiated the next trial by opening and releasing the reward-tray door. Each stimulus was presented for 5 seconds after egress from the reward magazine for both the training and testing periods to prevent the accumulation of omission errors during testing from possible lapses of performance, a variation from other 5CSRTT protocols.10,25
Following training, rats were divided in 2 groups and tested on the 5CSRTT at approximately 4:00 pm on 6 successive days of the week (Sunday-Friday) for 4 weeks (n=8). On alternating weeks, 1 group of rats underwent sleep deprivation prior to the behavioral test, while the other group was left undisturbed in the colony room before testing. On the weeks of deprivation, the rats were kept awake before testing on alternating days (Monday, Wednesday, and Friday) for 4, 7, and 10 hours, respectively, allowing a day for recovery between deprivations (Table). The rats were kept awake by the gentle handling method,33 which entailed continuous visual observation and gentle handling, using sensory, auditory, and tactile stimulation whenever the rats became prone or immobile. Electroencephalographic confirmation has indicated that the technique keeps rats awake 98% of the time over a 6-hour deprivation (~2% of the time, they transition to sleep) while undisturbed controls are awake 25% of the time.34
As the anticipation of daily feeding in food-restricted rats can reduce behavioral motivation in instrumental tasks and shift circadian rhythms, the rats were fed no less than 2 hours after the last rats finished behavioral testing (~8 pm). It has been shown that the body temperature and behavioral activity of food-restricted rats increases approximately 1 hour before the daily consumption of rat chow35,36 and before regular feedings of midday sucrose solution (~20 kcal) presented 8 hours before daily feeding.37 Although the current behavioral task provided smaller quantities of sucrose (~11 kcal) at less regular intervals (5 days weekly), it is possible that the rats’ waking schedule was similarly shifted before testing and that they were sleep deprived for as little as 3, 6, and 9 hours from their actual sleeping times. Novel stimuli and daytime physical activity produce small and transitory changes on sleeping schedules in rodents,38,39 so the periodic deprivations were not expected to significantly affect the sleeping schedule of the rats on the days of testing.
A within-subject comparison of performance was made between the 3 days of incremental sleep deprivation and the corresponding 3 days on the weeks that the rats were not deprived. Data from recovery days after sleep deprivation were not analyzed. Results were analyzed using JMP 4.0.2 (SAS Institute, Cary, NC). Following satisfaction of the Mauchly sphericity criterion that assesses a possible confound of within-subject correlations,40 a repeated measures F test was used to assess the effects of deprivation, hours, and subject for each of the 5 behavioral measures (correct responses, incorrect responses, omission errors, premature response, and perseverative responses) across 4 weeks of behavioral data (2 weeks deprivation, 2 weeks control). A mixed-model design was used to account for the within- and between-subject variance that might arise from the interindividual differences between subjects.41 Posthoc comparisons were made using Fisher HSD. To address PVT data on the consistency of omission errors performed by individuals on repeated testing following sleep deprivation,43 an interclass correlation coefficient was also calculated between omission errors made during the two 10-hour deprivations. The interclass correlation coefficient computes the ratio of between-subject variance to the sum of between-subject and within-subject variance and thus directly assesses the proportion of interindividual variability in the repeated measures of the two 10-hour deprivation conditions.43 One subject was not tested on the first 7-hour control day following a finger injury.
Sleep deprivation caused significant behavioral impairments across a number of performance measures of the 5CSRTT. Sleep deprivation produced a significant 33.14-millisecond increase in the mean latency of correct responses (F1,62 = 6.89, P = .0109). There was also a significant interaction between the duration of sleep deprivation and the latency of correct responses (F2,62 = 3.32, P = .0426) (Figure 3). Posthoc comparisons showed that there were no significant differences between the three 0-hour conditions over both control weeks (P > .05), but that, within deprivation conditions, the longest deprivation produced significantly slower responses than the shortest deprivation (10 hours vs 4 hours, P < .05). These results indicate a dose-dependent effect between the increasing length of sleep deprivation and the increasing speed of accurate responding. There was also a significant reduction in the number of correct responses following sleep deprivation (F1,62 = 5.20 P = .026), indicating the absence of the potential confound of a speed-accuracy trade-off.
Sleep deprivation produced a significant increase of omission errors (F1,62 = 14.47, P = .0003) (Figure 4), during which rats did not respond to a stimulus within 3 seconds. In addition to the group effects on omission errors, there was also a significant interaction between the duration of sleep deprivation and individual rats for omission errors (F7,62 = 2.24, P = .0422) (Figure 5), indicating an inconsistency in the effect of deprivation on the performance of individual rats. The interclass correlation coefficient of omission errors performed by the rats in the two 10-hour deprivation conditions was almost significant (r = 0.703; P = .0518) (Figure 6). These data suggest a differential effect of sleep deprivation on individual rats on their ability to reliably execute a response within a short window of time. The variance of individual data was somewhat larger than has been reported for human data42 and may reflect differences in motor preparedness between the PVT and the 5CSRTT, during which rats move freely within the behavioral chamber.
Sleep deprivation did not affect the overall measures of behavioral inhibitory control. The number of premature responses was not significantly changed by sleep deprivation (F1,62, P > .1). Although the majority of rats made more perseverative errors following deprivation, the average number of perseverative responses into the same port was not significantly different (F1,62, P > .1). However, there was a significant interaction of rats and deprivation (F7,62 = 2.94, P = .01), as sleep deprivation produced mixed effects in the different rats and dramatic increases of perseverative responding in 1 rat (that also made increased omission errors).
Ten hours of total sleep deprivation in rats produced a pattern of behavioral impairments in the 5CSRTT that is broadly consistent with the effects of sleep deprivation on vigilant attention performance in humans. Sleep deprivation produced a significant increase in the latency of correct responses in a dose-dependent manner, consistent with a monotonic effect of sleep debt on attention. Sleep deprivation also led to an overall increase in the number of omission errors, during which a rat did not respond to the stimulus within a brief period. The same measures are comparably affected in the PVT following similar deprivation lengths,4,15 thus the behavioral effects of sleep deprivation closely resemble the findings in human studies using the PVT to assess vigilance and attention deficits after sleep deprivation. In the current task, care was taken to limit possible lapses of performance from sleeping by requiring the rats to behaviorally initiate each trial. While the increases of omissions could be due to failures in the detection of brief stimuli,26 it is nonetheless possible that the rats had “microsleeps” in the brief period before the presentation of the stimulus that prevented both detection and responding, an effect of sleep deprivation that has been proposed to contribute to lapses of psychomotor vigilance.15
Sleep deprivation, however, had a minimal effect on aspects of the task that require inhibitory control. Consistent with human findings in a visuospatial reaction time task, there was little difference in the number of premature responses between deprivation and control conditions.23 In addition, sleep deprivation did not produce a significant group effect in the number of perseverative responses. This finding is consistent with human performance on cognitive tasks such as the Wisconsin Card Sorting Task following short-term total sleep deprivation,24 although some studies have reported a nonsignificant trend toward more perseverative errors on the task following chronic partial sleep deprivation.44,45 Perseverative responding has been shown to arise directly from lesions of medial prefrontal cortex both in the 5CSRTT and in a rodent task that requires attentional set-shifts equivalent to those of the Wisconsin Card Sorting Task in humans.46 The region also shares functional and anatomic features with the lateral prefrontal cortex in primates, the damage to which has been identified as a focal source of perseverative responding in the Wisconsin Card Sorting Task.31 The sparing of perseverative behavior in the current task may thus reflect a comparable limited effect of sleep deprivation on regional processing in the prefrontal cortex
Interestingly, individual rats appeared to be differentially susceptible to the effects of sleep deprivation. Although there was an overall group effect of sleep on omission errors, there was also a reliable difference between individual rats in the effects of deprivation on number of omission errors. Some rats exhibited almost no change in the number of omission errors in response to sleep deprivation, suggesting different compensatory mechanisms across individual rats that allowed them to maintain consistent performance on the task. Moreover, the degree of impairment for each rat was consistent across the two 10-hour deprivations (Figure 6), and there was a strong trend toward a significant interindividual performance difference between the rats. These effects are consistent with “trait-like” differences between individuals in their vulnerability to the effects of sleep deprivation on omission errors in the PVT.42
A similar individual pattern arose for the performance of perseverative responses. Following sleep deprivation, most rats made a greater number of perseverative responses into the same port, indicating a compulsive impairment of inhibitory control. Some rats, however, showed a decrease in the number of perseverative responses. Although there was not a significant main effect of perseverative responses, the significant interaction of sleep deprivation and individual performance could also be indicative of a selective susceptibility in subpopulations to the effects of sleep deprivation.
These data provide a systematic demonstration of attentional impairments in animals resulting from sleep loss. The pattern of effects on selective attention, behavioral control, and individual susceptibility is consistent with findings from human studies in which similar behavioral demands are made following sleep deprivation. These findings suggest that there is a broad preservation of sleep effects on attention and behavioral control across species.
The correspondence of human and animal impairments is significant in light of the paucity of data regarding the ways that sleep deprivation affects attentional processes in the brain. The extensive use of the 5CSRTT for neurobiologic investigations of attention in rats, however, provides a framework for understanding specific behavioral alterations in terms of their possible neurochemical and structural sources.
Much like sleep processes, some mechanisms of attention have been understood in terms of large-scale processes of the cortex that are regulated by ascending neuromodulatory systems. Attentional studies based on the 5CSRTT include studies of cortical structures that exhibit activity changes in sleep deprived humans30,47 and studies of neurochemical systems that help establish the sleeping and waking states of the brain, such as norepinephrine, dopamine, serotonin, and acetylcholine. It is noteworthy that each of the monoamine systems is also the site of orexin modulation that contribute to the stable maintenance of wakefulness.48 The basal forebrain cholinergic system receives inputs from the monoamines and orexin and is the major site of action of adenosine, a major homeostatic sleep factor.1,2,49-51 The considerable overlap of attention and sleep processes across the same neural systems highlights the potential value of the 5CSRTT as a model for relating the mechanisms of homeostatic sleep drive with processing changes in known systems subserving attention in the mammalian brain
Of the neurochemical manipulations that have been tested with 5CSRTT—including systemic noradrenergic manipulations and dorsal noradrenergic bundle lesions,25,52-55 cholinergic basal forebrain antagonists7 and lesions,10 striatal dopamine lesions,56,57 and global serotonin depletion in the forebrain58—the effects of sleep deprivation are most consistent with effects that arise from disruptions of basal forebrain cholinergic function and, to some degree, from drug manipulations of adrenergic systems. α2-Receptor agonists reduce the number of premature responses while also increasing omission errors on the 5CSRTT.52,54 Both the local blocks of acetylcholine and selective lesions of basal forebrain cholinergic neurons lead to a reduction in choice accuracy and lengthening of the latency of correct responses. Cholinergic disruption also leads to an increased number of premature and perseverative responses, an effect that corresponds directly to the reduced cholinergic modulation of the medial frontal cortex.10 Direct lesions of the medial prefrontal cortex produce a similar pattern of effects on latency, accuracy, and perseverative responding.32 Interestingly, 2 major targets of the cholinergic basal forebrain, the frontal and parietal cortex, show increased activity in a divided-attention task in humans following 35 hours of total sleep deprivation, possibly indicating compensatory processing as a result of altered subcortical modulation.16,59
The multiple effects of total sleep loss on the 5CSRTT are thus consistent with known effects of acetylcholine and norepinephrine on attentional performance and could partially be due to the combined suppression of these systems during sleep onset.60 The parallels of cholinergic function are particularly interesting in light of known processes in the basal forebrain that underlie the promotion of sleep. Adenosine directly inhibits cholinergic neurons in the basal forebrain of cats and rats, and its concentration increases in the extracellular space of the basal forebrain in a manner that increases linearly with the duration of total sleep loss.51,61 It could thus play an underlying role in the decreases of attentional function that accompany sleep loss. Increasing sedation is also likely to result from noradrenergic suppression,62-66 altered thalamocortical activity, hormones, and metabolic changes. Each of these possibilities, however, remains to be tested.
While the current findings suggest a broad preservation of sleep effects across species, they only reveal behavioral parallels, and many more studies will be needed to establish the biologic nature of attentional alterations that result from sleep loss. In light of considerable, but remarkably separate, scientific knowledge of sleep and attention processes in the brain, the successful use of the 5CSRTT for assessing attentional dysfunction in sleep-deprived animals provides new opportunities for directly investigating the links between neurochemical sleep processes and our cognitive and behavioral impairments that result from sleep loss.
We thank Kara Mulkern for technical assistance and Doug Nitz for comments on an earlier version of the paper. This research was supported by an NSF 0094377 to A.A.C., The Department of Veterans Affairs Medical Research Service Awards to R.E.S. and R.W.M., USPH grants HL60292, MH39683, and MH070959 to C.C.
Disclosure Statement This was not an industry supported study. Dr. Strecker has received research support from Cephalon. Drs. Cordova, Said, McCarley, Baxter, and Chiba have indicated no financial conflicts of interest.