Experimental sleep fragmentation impairs spatial reference but not working memory in Fischer/Brown Norway rats


Christopher P. Ward, Department of Psychology, University of Houston-Clear Lake, 2700 Bay Area Blvd., Box 22, Houston, TX 77058, USA. Tel.: 281 283 3303; fax: 281 283 3406; e-mail: wardchris@uhcl.edu


Sleep fragmentation is a common symptom in sleep disorders and other medical complaints resulting in excessive daytime sleepiness. The present study seeks to explore the effects of sleep fragmentation on learning and memory in a spatial reference memory task and a spatial working memory (WM) task. Fischer/Brown Norway rats lived in custom treadmills designed to induce locomotor activity every 2 min throughout a 24-h period. Separate rats were either on a treadmill schedule that allowed for consolidated sleep or experienced no locomotor activation. Rats were tested in one of two water maze-based tests of learning and memory immediately following 24 h of sleep interruption. Rats tested in a spatial reference memory task (eight massed acquisition trials) with a 24-h follow-up probe trial to assess memory retention showed no differences in acquisition performance but were impaired on the 24 h retention of the platform location. In contrast, the performance of rats tested in a spatial WM task (delayed matching to position task) was not impaired. Therefore, sleep fragmentation prior to testing impairs the ability to retain spatial reference memories but does not impair spatial reference memory acquisition or spatial WM in Fischer-Norway rats.


There is growing evidence that sleep plays an important role in the consolidation of memories (Stickgold and Walker, 2007). Most studies that have explored this question have focused on the role of sleep after new information has been learned. On the other hand, relatively little research has looked at the role of sleep prior to learning. The disruption of sleep is associated with decreased learning capacity in students as well as impairments in declarative and procedural learning (Curcio et al., 2006). Additionally, patients suffering from primary insomnia have increased difficulty in acquiring procedural memories (Nissen et al., 2006).

Previous research in rodents has shown that hippocampal-dependent memory processes are especially sensitive to disruptions in sleep. Selective rapid eye movement (REM) sleep deprivation impairs place acquisition, but not cued learning of the water maze task (Youngblood et al., 1997) nor the working memory (WM) component of a radial arm task (Smith and Rose, 1996). Sleep deprivation also impaired performance on hippocampal-dependent contextual fear task following the first 5 h after acquisition (Graves et al., 2003). Total sleep deprivation prior to learning can also impair retention of a spatial water maze task (Guan et al., 2004). Other disruptions of sleep such as 24 h of sleep fragmentation prior to learning also impair water maze performance (Tartar et al., 2006). Deficits in hippocampal long-term potentiation (LTP) have also been found following selective REM sleep deprivation (Davis et al., 2003; McDermott et al., 2003) and sleep fragmentation (Tartar et al., 2006). Sleep deprived human participants show decreased hippocampal activity during learning (Yoo et al., 2007).

In many sleep disorders such as obstructive sleep apnea, as well as other diseases and disorders, sleep is fragmented throughout the night leading to daytime symptoms resembling those of total sleep loss (Bonnet and Arand, 2003). In experiments with normal humans, sleep that is interrupted every minute produces deficits in thought and vigilance tasks (Bonnet, 1986). Given the relevance of sleep fragmentation in human lives, rodent models of sleep interruption (SI) have not been sufficiently studied. Prior research in our laboratory has indicated that sleep fragmentation can impair spatial memory and hippocampal synaptic plasticity (Tartar et al., 2006). Additionally, we found that 24 h of sleep fragmentation can impair cognitive flexibility in an attentional set shifting task (McCoy et al., 2007a). In an attempt to more closely model human experience, the present experiment explores the effects of sleepiness on two types of spatial learning tasks. The sleep of rats was interrupted every 2 min for a 24-h period prior to the acquisition of a spatial reference learning task or prior to performing a spatial WM task.



Adult male Fischer/Brown Norway (FBN) F1 (275–325 g; Harlan, Indianapolis, IN, USA) were used in behavioral testing (n = 63). Rats were housed in groups of three in standard cages (unless otherwise noted) under constant temperature (23 °C) and a 12 : 12 light dark cycle (lights on at 07:00 hours) with food and water available ad libitum. All animals were treated in accordance with the National Institute of Health Guide for the Care and Use of Laboratory Animals (NIH Publications No. 80–23) revised 1996. All procedures were approved by the institutional animal care and use committee of the VA Boston Healthcare System.

Sleep interruption procedure

The experimental sleep fragmentation procedures have been described and characterized in detail previously (McCoy et al., 2007a; McKenna et al., 2007; Tartar et al., 2006). This procedure has provided consistent and reliable disruptions of sleep in rats. Polysomnography was not recorded to avoid the invasive surgery required for the unnecessary replication of data. Briefly, during the SI protocol, rats lived in a treadmill cage (L × W × H; 50.8 × 16.51 × 30.48 cm) with free access to food and water. The floor is a horizontal rubber belt automatically programmed to move slowly at a rate of 0.02 m s−1. The treadmill ran at this slow speed for 30 s, followed by no treadmill movement for 90 s. This 30 s on/90 s off schedule produced 30 interruptions of sleep per hour continuously for 24 h. In order to habituate the rats to the treadmill movement, rats were placed in the treadmills 2 days prior to the experiment. Treadmills were turned on (5 min on followed by 5 min off) for 1 h on each of the 2 days. As a control for the non-specific effects of locomotor activity, an exercise control (EC) group was included in this study. In this group, rats obtained an equivalent amount of treadmill movement/exercise, but with a treadmill on/off schedule of 10 min on/30 min off, allowing for longer periods of undisturbed sleep. Cage control (CC) rats lived in the same cage without any treadmill movement. Fig. 1a graphically represents the timeline of the treadmill procedure along with the subsequent memory testing.

Figure 1.

 (a) Schematic of experimental time line. Rats were habituated to the treadmill on days 1 and 2 for 1 h with the treadmill moving for 5 min and off for 5 min. On day 3, sleep interruption (SI) was produced through treadmill movements of 30 s on and 90 s off. Exercise control rats experienced 10 min on and 30 min off. Immediately following time spent on the treadmill, rats were tested in either a spatial reference memory (RM) or working memory (WM) task. The RM task consisted of acquisition (Acq.) trials followed by a 24 h recall (Rec.) trial. Open bars indicate lights on. (b) The diagram represents the nine novel platform locations used in the delayed matching to position task to assess spatial WM (adapted from Steele and Morris, 1999).


All animals were trained in a pool 2.0 m in diameter and 0.4 m in depth, containing water made opaque by non-toxic, water-soluble paint at room temperature (23 °C). The pool was in a 3 × 5 m room with several distinctive spatial cues (e.g. signs, laboratory furniture). A 10 cm diameter platform submerged approximately 1 cm below the surface of the water. To cue the platform location during the WM trials, a flag was attached that extended approximately 10 cm above the water surface. Rodent performance was tracked with a video tracking system (EzVideo Multi Track System, AccuScan, Columbus, OH, USA).

Spatial reference memory

In the reference memory water maze task, rats (n = 12 SI, n = 12 EC, n = 11 CC) were tested in one training session consisting of eight consecutive trials with a 60-s inter-trial interval (Packard and Teather, 1997). This version of the water maze task allows rats to be fully trained rapidly, which is necessary for manipulations that cannot be given on multiple days, such as 24-h sleep fragmentation. This protocol also denied rodents time to sleep between testing trials. Rats were tested in the water maze during the last 2 h of the 12-h lights-on period. On each trial, rats were placed in the WM facing the wall in one of three quadrants that did not contain the hidden platform. The starting position was in a semi-random order so that no start point was repeated and no point was used more than three times. The location of the hidden platform remained constant. If the animal did not find the hidden platform within 60 s, the rat was guided to the platform by the experimenter and allowed to remain on the platform for approximately 15 s before being placed in a dry holding cage for an additional 60 s. Following the last training trial, rats were returned to their home cages and were left undisturbed until the probe trial. A probe trial was given 24 h after the last learning trail. During the probe trial, the platform was removed and each rat had a 30 s free swim in the pool. All rats were started in the quadrant across from the target quadrant.

Spatial working memory

A separate cohort of rats (n = 17 SI, n = 11 CC) were tested in the spatial WM water maze task. The EC group was not utilized because no significant differences were eventually found between CC and SI groups. Initially, rats were trained in a standard water maze protocol as described above. During this training, the location of the hidden platform remained constant across all trials. This training occurred approximately 1 week before spatial WM testing and allowed rats to learn the basic form of the water maze task.

Following the water maze training, rodents were tested in the spatial WM protocol. The spatial WM task consisted of pairs of trials. In the first trial, a cued platform marked with a flag was placed in a novel location in the tank. The nine novel positions used in the task were modified from Steele and Morris (1999) (see Fig. 1b). Rats were released facing the wall from one of the three quadrants that did not contain the platform. Rats were given 120 s to locate the cued platform before the experimenter would guide the rat to the platform. After 15 s on the platform, rats were removed to an individual holding cage. During this time, the flag was removed from the platform so that it was hidden but in the same location as in the first trial. After the rodent spent 1, 5 or 10 min in the holding cage, the rat was placed in the tank in one of the two quadrants that did not contain the platform and was not the starting quadrant from the first trial. Rats were once again given 120 s to find the hidden platform before the experimenter would guide the rat. After approximately 5 s on the platform, the rat was placed back in the holding cage for 10 min before a new pair of trials (cued followed by hidden platform) with a novel platform location was given. Three rats were tested at a time. The order of delays was counterbalanced so that each rat was tested three times at 1, 5 or 10 min delays between the cued and hidden platforms. Therefore, each rat was tested with nine pairs of trials with the platform in a novel location for each pair. The total testing time for three rats was <2.5 h.

Data analysis

The main dependent variables collected in both tests were latency and path distance rats took to find the platform in the water maze. In probe trial data, the main dependent variables were percent time and swim distance spent in the quadrant of the pool that formerly contained the hidden platform. Water maze testing was analysed by factorial repeated measures analysis of variance (anova). Trend analysis of data was performed using Tests of Within-Subjects Contrasts and sphericity was verified using Mauchly’s Test. Probe trial data were analysed by anova followed by Dunnett’s post hoc analysis. All data analysis was conducted utilizing spss (version 13.0, SPSS Inc., Chicago, IL, USA) with an alpha level of 0.05.


Reference memory water maze task

Twenty-four hours of SI prior to testing did not alter acquisition of platform location in massed trial learning, but did impair the memory of platform location 24 h later (see Fig. 2). In acquisition trials (Fig. 2a), no significant differences in learning were observed among the three different groups of rats as indicated by either latency [F(2,32) = 2.139, P > 0.05] or swim distance to reach the platform [F(2,33) = 2.148, P > 0.05], nor was there a significant trial × group interaction for latency or distance [F(14,224) = 1.287, P > 0.05; F(14,224) = 1.086, P > 0.05 respectively]. There was a significant effect across trials [latency: F(7,224) = 24.683, P < 0.05; distance: F(7,224) = 15.308, P < 0.05] indicating the animals learned the location of the platform. Twenty-four hours after the last trial, rats were tested in a single probe trial to test memory of the platform location. The percentage of total time and distance the animals spent searching in the quadrant that formerly contained the hidden platform was calculated (Fig. 2b). There was a significant difference among the groups in percent time [F(2,32) = 4.405, P < 0.05] and percent distance [F(2,32) = 10.750, P < 0.05] indicating rats with SI prior to initial learning overall spent significantly less time in the target quadrant than did CC (P < 0.05), but EC was not significantly different from CC. Differences in probe performance cannot be attributed to motor impairments as swim velocity was not significantly altered [F(2,32) = 0.404, P > 0.05] (data not shown).

Figure 2.

 The recall of the platform location was impaired by 24 h of sleep interruption (SI) prior to acquisition. (a) Mean (±SEM) latency (left) and path length (right) to reach the hidden platform over eight consecutive acquisition trials. No significant differences in learning were observed among the three different groups of rats. (b) Mean (+SEM) percent time (left) and distance (right) during probe trial that rats spent searching in the target quadrant that formerly contained the hidden platform. SI rats spent significantly less time in the target quadrant than did cage control (CC), but exercise control (EC) was not significantly different from CC. *P < 0.05.

Spatial working memory water maze task

In the spatial WM task, control rats took increasingly longer time periods and distances to reach the hidden platform as delay periods lengthened from 1 to 10 min (see Fig. 3). In control rats, there were no significant differences in the latency [F(2,44) = 0.084, P < 0.05] or swim distance [F(2,44) = 0.015, P > 0.05] to reach the cued platform among the different delay intervals. However, there was a significant effect of delay time on rodents latency [F(2,44) = 7.814, P < 0.05] and path length [F(2,44) = 6.829, P = 0.005] to find the hidden platform. The pattern of responses showed a significant linear trend [F(1,11) = 11.594, P < 0.05] with the shortest mean latencies following the 1-min delay and the longest average latency following the 10-min delay. The same linear pattern of responses was observed for swim distance [F(1,11) = 9.945, P = 0.009] where once again, the shortest average distances were observed after the 1-min delay and the longest average distances followed the 10-min delay. There was not a significant main effect for the trial number in latency [F(1.318,44.0) = 3.083, P > 0.05] or swim distance [F(1.263,44.0) = 3.762, P > 0.05] indicating that rats performed similarly across the three trials of the same delay. The delay × trial interaction was also not significant for latency [F(4,44) = 1.220, P > 0.05] or distance [F(4,44) = 1.204, P > 0.05].

Figure 3.

 The delayed matching to position task tested spatial working memory in control rats. Mean (±SEM) latency (top) and distance (bottom) for cage control (CC) rats to find the cued platform in the first trial compared with finding the hidden platform in the second trial. Rats spent significantly longer periods of time finding the previously indicated platform location as delays increased from 1 to 5 to 10 min. *, comparison between cued and hidden platform P < 0.05; +, comparison between delays P < 0.05.

Twenty-four hours of SI prior to testing did not impair performance on the spatial WM task (see Fig. 4). There were no significant differences between SI and CC groups in either latency [F(1,26) = 0.216, P > 0.05] or in the path length to find the hidden platform [F(1,26) = 0.236, P > 0.05]. Motivation to escape the water maze was not affected by the prior 24 h of SI. There were no significant differences in latency [F(2,33) = 0.977, P > 0.05] or swim distance [F(2,33) = 1.513, P > 0.05] to find the cued platform (data not shown). Additionally, as seen in the previous reference memory experiment, swim speed velocity was not affected by prior SI [F(2,33) = 0.046, P > 0.05] (data not shown).

Figure 4.

 Twenty-four hours of sleep interruption (SI) prior to testing did not impact performance on the spatial working memory task. Mean (±SEM) latency (top) and distance (bottom) for rats to find the hidden platform for cage control (CC) versus SI rats. No significant differences were observed between groups.


The present set of experiments demonstrate that 24 h of sleep fragmentation prior to acquisition causes a deficit in the 24 h retention of a spatial reference memory task, but no impairments in a spatial WM task. There were also no impairments observed in the learning of the location of the hidden platform in the spatial reference memory task. This suggests that prior sleep disturbances selectively interfere with the consolidation of spatial memories.

When compared with previous findings, the present data indicate that subtle differences in water maze testing protocols and/or the use of different rat strains can influence the outcome of studies investigating the effect of sleep disruption on learning and memory. Previous research from our laboratory demonstrated that water maze acquisition was impaired when training followed 24 h of SI (Tartar et al., 2006). The differences in results are likely because of the choice of rat strains. Prior research in our laboratory utilized Sprague-Dawley (SD) rats while the present study utilized FBN rats, which are superior performers in spatial learning tasks when compared with SD rats (Harker and Whishaw, 2002). The idea that different strains could be impacted by sleep loss differently could be as a result of genetic differences such as suspected to underlie individual variations in response to sleep loss in humans (Van Dongen et al., 2005).

Another major difference between the present study and previous findings (Tartar et al., 2006) is changes in testing protocol. Prior research utilized a protocol in which rats were allowed 1 h breaks between three sessions of four trials/session. The present protocol does not utilize any breaks between the eight consecutive acquisition trials as described by Packard and Teather (1997). This rapid protocol allows testing to be completed in a minimal amount of time after being removed from SI. The pattern of results observed in the present experiment are consistent with a finding in another laboratory where rats were not given a break between trials and only showed deficits in the recall of the platform location and not acquisition following total sleep deprivation (Guan et al., 2004).

These results propose an interesting hypothesis that the choice of water maze protocol is important in that the two protocols used in our sleep fragmentation studies gave different results. If sleepiness produced by sleep fragmentation selectively impairs the consolidation of spatial memories, it is possible that utilizing a protocol that allows for 1 h breaks between blocks of sessions relies more on the consolidation of memories in prior blocks as opposed to keeping the memories ‘fresh’ during continuous training. The results of the spatial WM task could support this argument.

It is likely that there is an interaction effect of rat strain and task difficulty. Additional research in our laboratory (McCoy et al., 2007b) in which FBN rats were tested utilizing the same protocol as previously reported with SD rats (Tartar et al., 2006) demonstrated that FBN rats show no learning or recall deficits following 24 h of sleep fragmentation (McCoy et al., 2007b). The SD rats showed sleep loss-related deficits in the easier water maze task (i.e. three sessions of four trials/session) and, in pilot studies, performed very poorly on the harder water maze task (eight consecutive trials and 24-h recall) without any manipulations. On the other hand, FBN rats do not show performance deficits following sleep disruption in the easy water maze task (McCoy et al., 2007b) but do show deficits in the 24 h retention of spatial memory. This could suggest that FBN rats have a greater cognitive reserve so that deficits as a result of prior sleep disruption are not noticed unless task difficulty is increased.

The impact of the SI procedure utilized in the present experiment on sleep architecture has been characterized in prior research from our laboratory. This procedure has been used with both SD (McKenna et al., 2007; Tartar et al., 2006) and FBN (McCoy et al., 2007a) rats yielding similar results in both strains, especially during the final 12 h of SI and during recovery sleep. During the 24 h of SI, there is a reduction in sleep bout length, and a reduction in total amount of sleep, mainly at the expense of REM sleep. Total amount of non-REM sleep is not greatly affected (McCoy et al., 2007a; McKenna et al., 2007; Tartar et al., 2006). Rats show an increase in sleepiness following 24-h SI as defined by decreased time to fall asleep in a multiple sleep latency test and increased extracellular levels of the inhibitory neuromodulator, adenosine, in the basal forebrain (McKenna et al., 2007). Basal forebrain levels of adenosine have previously been shown to increase during prolonged wakefulness and dissipate during recovery sleep (Porkka-Heiskanen et al., 1997, 2000). Additionally, electroencephalogram measures indicate an increase in non-REM delta power in recovery sleep following 24 h of SI (McCoy et al., 2007a; McKenna et al., 2007; Tartar et al., 2006). Non-REM delta power has been proposed as a homeostatic marker of sleep drive (Franken et al., 1991). Future experiments should explore the role of delta power following sleep disruption on spatial learning tasks.

The evidence for sleepiness persists for approximately 3 h following the termination of the SI protocol (McCoy et al., 2007a; McKenna et al., 2007; Tartar et al., 2006), therefore, the spatial memory tests utilized in the present study were specifically designed to be completed within this time frame. Standard protocols in the reference memory version of the water maze task typically divide training over several days (Morris et al., 1982; Sutherland et al., 1982). Rapid training protocols for this task have been previously developed (Frick et al., 2000; Packard and Teather, 1997). On the other hand, to the author’s knowledge, no spatial WM task that can be completed within 3 h has been developed. Much like the standard water maze task, typical WM protocols take many days to train (Steele and Morris, 1999; Whishaw, 1985). The delayed matching to sample task used to assess spatial WM in the present study is a novel task that allows for completion in under 3 h while rodents are still sleepy (McKenna et al., 2007). This task shows increased memory demands on rodents as delay periods increase (see Fig. 3). Additionally, as a spatial WM task defines (Olton, 1977; Steele and Morris, 1999; Whishaw, 1985), each cued/non-cued trial pairing is in a unique spatial location, rodents must disregard previous platform locations and recall only the most recent cued location. The task also includes a built-in test for motivation, sensory and motor ability as the initial trial is cued by a visible target. While this task is promising as a rapid measure of spatial WM in rodents, additional test validation is needed to demonstrate that the test is sensitive to known manipulations that disrupt spatial WM.

While there is still some debate about the role of sleep in memory consolidation (Vertes, 2004), there is growing evidence that sleep does play an important role in the consolidation of memories (Stickgold and Walker, 2007) even though the mechanisms are not fully understood (Frank and Benington, 2006). While the role of sleep after learning a new task has generated a great deal of research, experiments exploring learning a new task while sleepy have received less attention (Guan et al., 2004; Stern, 1971; Yoo et al., 2007). In humans, a recent functional imaging study showed that participants with a single night of sleep deprivation had impaired hippocampal activity during episodic memory encoding (Yoo et al., 2007). Prior research from our laboratory has suggested that sleepiness resulting from 24 h of SI causes a deficit in the acquisition of a spatial learning task that correlated with an absence of hippocampal LTP (Tartar et al., 2006). Other laboratories have also noted impaired hippocampal LTP following selective REM sleep deprivation (Davis et al., 2003; McDermott et al., 2003). Prior sleep deprivation has also been shown to reduce phosphorylated extracellular signal-regulated kinase in the hippocampus (Guan et al., 2004). Sleepiness in rodents also caused a deficit in cognitive flexibility in a test of executive functioning (McCoy et al., 2007b). It is a possibility that increases in basal forebrain adenosine because of sleep fragmentation (McKenna et al., 2007) inhibit excitatory projections from the basal forebrain to the frontal cortex and hippocampus.

Another possible hypothesis is that memory impairments following sleep fragmentation are a secondary effect, and that deficits in attention are the primary effect. Sleep loss impacts measures of attention in humans (Dinges et al., 1997) and rats (Cordova et al., 2006; Godoi et al., 2005). If attentional functions are compromised, this will affect later encoding of memories. While the impact of sleepiness on attention cannot be overlooked, evidence from impaired hippocampal function following sleep loss suggests that there are also deficits in the memory circuit. Additionally, stress produced by the treadmill-induced SI procedure is an unlikely explanation for the memory deficit because prior research has shown that EC and SI both have elevated levels of circulating corticosterone, though only SI rats indicate difficulty in the recall of the platform location (Tartar et al., 2006).

In conclusion, 24 h of sleep fragmentation prior to learning effectively impairs rodents’ ability to retain a reference memory. This finding is especially relevant to patients with sleep disorders that fragment sleep such as sleep apnea. Combined with our previous work the findings indicate that sleep disruption more easily impairs spatial reference memory in SD rats compared with FBN rats; and, the effect of sleep disruption on spatial reference memory is influenced by the trial frequency of the water maze protocol used (massed versus spaced acquisition trials). This work adds to the growing literature indicating that sleep disruption can severely impact the consolidation of hippocampal-dependent memories.


The authors wish to thank Matthew Heard and Amy Blanchette of Stonehill College for their excellent technical assistance. This research was supported by NIH HL060292 and the Department of Veterans Affairs.