Sleep‐like behavior and 24‐h rhythm disruption in the Tc1 mouse model of Down syndrome

Using a conventional circadian wheel-running recording system, mice of both genotypes, Tc1 carrier mice and wild-type littermate controls, showed normal photoentrainment under a 12:12 LD cycle, with 90% of wheel-running activity occurring in the nocturnal phase and similar alpha (Table,anda). Comparison of wheel-running activity showed a trend towards a delay in the acrophase of wheel-running activity at the beginning of the dark phase for Tc1 animals, although this was not significant (a, wild-type acrophase at 22.00 ± 0.41 h and Tc1 at 20.96 ± 0.25 h,= 1.553,= 0.230). Furthermore, Tc1 animals ran significantly less during the dark phase of the LD cycle compared with their wild-type littermate controls resulting in lower average rotations per night (Table). Remarkably, however, video-tracking of Tc1 animals showed a 6-h period sustained wakefulness (0% immobility), with the onset of the dark phase. Inspection of videos during this time confirmed that mutant animals conduct all of the behaviours expected, including locomotor activity, vertical activity, climbing, feeding and grooming, more frequently than wild-type littermate controls. Controls were asleep for approximately 20% of the time with sleep increasing at later stages in the dark phase (b). Tc1 animals were not only more mobile than littermate controls but more active with greater distances travelled (= 100.91;< 0.001;c). Inspection of video recordings during this period of hyperactivity showed that the number of immobile episodes was significantly lower for Tc1 animals (= 45.88;< 0.001), which is consistent with a constant and uninterrupted wakefulness (e). Average estimated sleep bout length was also significantly shorter (g,= 0.842,= 0.0218). In the light phase of the LD cycle, no significant differences in distance travelled and percentage of immobility could be observed (). However, the number of immobile episodes during the light phase was significantly higher for Tc1 animals compared with littermate controls (= 5.72;= 0.018), while average estimated sleep bout length was not significantly different (= 7.367,= 0.38), which is indicative of greater sleep disruption/fragmentation in triploid animals (f). 1 Figs. 1 2 Fig. 2 1 Fig. 2 Fig. 2 Fig. 2 Fig. 2 Fig. 2 Fig. 2 F P F P F P F P F P F P 1,17 1,54 1,54 1,10 1,108 1,10

Under free-running conditions in DD, no significant differences in wheel-running amplitude or in average revolutions were identified. Neither was the circadian period significantly different between genotypes, but Tc1 animals displayed shorter alpha compared with wild-type littermates (Table). When investigated in DD, the acrophase of wheel-running activity again showed a non-significant advanced trend for Tc1 carriers with a more defined and narrow peak of activity in comparison to wild-type littermate controls (a,= 0.726,= 0.405). In LL conditions, the internal period lengthened for all animals as expected. Average, revolutions, and amplitude were not significantly different for wild-type and Tc1 animals, but alpha was shorter for Tc1 animals compared with that for littermate controls (Table). 1 Fig. 3 1 F P τ 1,18 LL

The results of the video-tracking in DD were similar to those described for LD conditions. At the beginning of the subjective night, Tc1 mice maintained their state of sustained wakefulness (displayed as 0% immobility) compared with littermate controls, although the length of this period was shortened to 3 h of intense activity. This increased wakefulness in Tc1 animals shows a non-significant advanced trend of 1–2 h relative to that of wild types (latency to first immobile episode in subjective night, wild-type 67.94 min, Tc1 185.40 min,= 1.338,= 0.277). Wild-type littermates exhibited a level of immobility of about 30% at this time, again decreasing towards the end of the subjective night (b). Distance travelled by Tc1 animals was significantly higher not only during subjective night (= 58.09;< 0.001,d) but also during the subjective day (= 10.02;= 0.002,c). The latter effect is likely related to the advanced phase of activity seen in Tc1 animals. Finally, by comparing the number of immobile episodes in DD conditions, no significant differences between genotypes were detected during the subjective day whereas during the subjective night mutant animals exhibited a significantly lower number of immobile episodes (= 44.68;< 0.001,f). F P F P F P F P 1,9 1,108 1,108 1,108 Fig. 3 Fig. 3 Fig. 3 Fig. 3

Introduction of an acute LP in the dark phase of the LD cycle should rapidly induce sleep in mice (Lupi; Muindi). Light-induced immobility-defined sleep was less pronounced for Tc1 animals during wheel-running (not significant, Fig. S1, Supporting information) and significantly delayed by around 20 min in Tc1 animals compared with that in their wild-type littermate controls in the video-tracking approach (a). This was confirmed as a significantly longer latency to the first occurrence of immobility-defined sleep (= 66.17;< 0.001,b). Analysis of the total amount of immobility-defined sleep and activity across the duration of the LP showed significant differences between genotypes with Tc1 animals travelling a greater distance (= 33.40;< 0.001,c) and spending less time asleep (= 7.92;= 0.020,d). Furthermore, comparing immobility between control littermates and Tc1 mutants in pre- and post-LP periods (darkness) showed much higher levels of activity for the Tc1 carrier group (a). This was consistent with the results of the 24-h LD and DD recordings (). et al. et al. F P F P F P ^{2008 2013} Fig. 4 Fig. 4 Fig. 4 Fig. 4 Fig. 4 Figs. 3 1,9 1,9 1,9

Using two diverse and complementary methodologies, conventional circadian wheel-running analysis and a novel validated video-tracking system that defines sleep based upon periods of immobility (Fisher), we have been able to gain new insights into the general activity, circadian function and sleep-related behaviours of the Tc1 mutant mouse line. Estimating periods of sleep based on immobility is a useful approach to assess sleep/wake behaviour in mice as it is a non-invasive technique that is both faster and more flexible for high-throughput analysis. It is also much more appropriate for mouse lines like Tc1, which show deficits in skilled motor function (Galante), a compromised health status and reduced survival rates after surgery. Moreover, as video-tracking results are so closely correlated with EEG data (Fisher), this approach has significant advantages over the use of more invasive tethered or telemetric EEG recording approaches as mutant animals may be unduly affected by anaesthesia and surgical procedures. In contrast, however, video-based assessment of sleep-like behaviour may be confounded by motor-skills differences in mutant mice and this possibility should be considered in studies such as this. et al. et al. et al. ^{2012 2009 2012}

The use of multiple tests in this study has allowed us to measure disturbances in general activity and motor function. Wheel-running is a non-invasive test that has been used traditionally to measure circadian activity as well as phasic responses to LD conditions. The majority of studies suggest no alterations in circadian locomotor activity for Ts65Dn animals during a 12:12-h LD cycle (Martinez-Cue; Reeves) or in constant lighting conditions (Ruby). Surprisingly, however, Ts65Dn mice exhibit a hyperactive phenotype visible predominantly in the early hours of the dark phase that is not evident during the light phase (Escorihuela; Martinez-Cue). Moreover, a single independent study found that Ts65Dn animals exhibit a significant advanced phase in activity onset of approximately 4 h compared with wild-type mice under a 12:12-h LD cycle (Stewart). Overall, we detected only very subtle circadian phase-associated disturbances in Tc1 animals using this behavioural assessment. However, we did detect a reduction in the average wheel-running activity of Tc1 mutant mice during the dark phase of the 12:12-h LD cycle. Motor disabilities are a common symptom in Down syndrome patients (Spano). In earlier studies, Tc1 mice were shown to have impairments in skilled motor functions although general movement, gait, grip strength and other simple motor functions were unaffected (Galante), and this is reflected in our study where mice are more active while engaging in less wheel-running activity. This distinction is not unprecedented as we have shown in an earlier study using principal component analysis that wheel-running performance is independent of locomotor activity (Mandillo). Of course, we cannot discount the fact that Tc1 animals were less motivated to run on wheels. Additional studies testing animal motivation would need to be carried out to comment on whether this might contribute to the low wheel-running we found. et al. et al. et al. et al. et al. et al. et al. et al. et al. ^{2002 1995 2010 1995 2013 2007 1999 2009 2014}

In contrast to the wheel-running study, Tc1 mice express a hyperlocomotor activity in the home cage as assessed using video-tracking. Hyperlocomotion had been previously recorded in Tc1 mice when they were assessed over short intervals in the open field test (Galante), but this is the first instance where consistent levels of hyperactivity have been recorded over long intervals in the home cage. This suggests that increased activity in Tc1 animals is not only precipitated by introducing mutant animals to a novel environment, as in the open field, but is expressed as an unprovoked behaviour in the familiar surroundings of the home cage. Galante() suggest that hyperactivity may be associated with a deficit in hippocampal function in Tc1 animals. Interestingly, impulsivity and hyperactivity are frequently reported for Down syndrome patients (Ekstein). et al. et al. et al. ^{2009 2009 2011}

Our data also highlight the influence of LD phases on Tc1 activity disturbances. Hyperactivity in Tc1 animals is most evident at night, while the duration of this hyperactive phase is shortened when they are maintained in DD. Similar observations have been noted for Ts65Dn animals. For example, activity levels have been significantly higher during the dark phase in LD compared with those in DD or LL for Ts65Dn animals when compared with wild types in an actimetry study (Ruby). Also, Ts65Dn mice tested under white and red light conditions in an open field arena only showed a significantly increased number of line crossings under white light conditions when compared with control animals (Escorihuela). Light at night-time also highlights significant differences between Tc1 animals and littermate controls. Analysis of the LP data emphasizes how changing environmental lighting conditions can modulate the behaviour of Tc1 animals as evidenced in the delayed onset and reduction of light-induced sleep for the duration of the LP. et al. et al. ^{2010 1995}

Down syndrome patients suffer from sleep disturbances such as increased daytime sleepiness, prolonged sleep latency at night, reduced amount and number of bouts of rapid eye movement (REM) sleep and sleep fragmentation (Carter; Diomedi; Grubar; Hamaguchi; Levanon). Sleep apnoea is thought to play a major role in causing sleep abnormalities in Down syndrome patients (Marcus) but is not the only cause (Levanon). In developing mouse models for Down syndrome, it is important that these sleep disturbances can be reflected accurately. Circadian and sleep/wake-related behaviour in the Ts65Dn mouse model shows similarities to Down syndrome patients with clear disturbances in EEG parameters showing decreased non-rapid eye movement (NREM) sleep and NREM bout durations associated with increased wakefulness during the light phase (Colas). Although not accompanied by EEG recordings, Tc1 animals exhibit a lower percentage of estimated sleep and delays in the onset of light-induced sleep in our study. Nevertheless, traits like sleep fragmentation and longer latencies for sleep onset in patients (Breslin; Carter) are reflected in delayed latency for light-induced sleep (LP) and a higher amount of immobile episodes in Tc1 animals. Conversely, major sleep disturbances are not found in Ts1Cje (Duchon), although they do show a delay in sleep rebound (Colas). These findings have prompted this group to make some assumptions on the contribution of loci to the sleep disturbance phenotype, suggesting that they should not be triplicated in Ts1Cje. In particular, they noted that APP transgenic mice show consistent sleep disturbances in multiple studies (Colas). However, recent findings in Tc1 mice that the final coding exon of APP is rearranged with no human APP protein detectable would argue that loci other than APP contribute to the sleep phenotypes in Down syndrome mutant models (Reinholdt). et al. et al. et al. et al. et al. et al. et al. et al. et al. et al. et al. et al. et al. et al. ^{2009 1999 1986 1989 1999 1991 1999 2008 2011 2009 2011 2008 2004 2011}

Although not fully investigated in Tc1 mice, data from human studies and from other mouse models would suggest that rest/activity and rhythm disturbances may arise as a consequence of either generalized synaptic deficits or disturbances in particular brain circuitries. MRI scans of individual Down syndrome patients have recorded numerous brain anomalies including changes in size of cerebellum, frontal, temporal and occipital cortical lobes and hippocampus (Roubertoux & Kerdelhue), while alterations in cortical lamination, dendritic branching and numbers of synapses have also been recorded (Roizen & Patterson). Aside from these general structural deficits, disturbances in cholinergic (Roizen & Patterson) and serotonergic function (Seidl) identified in Down syndrome patients would seem most likely to affect activity/sleep parameters. More specifically, deterioration in cholinergic basal forebrain neuronal function may be causative. Most data on neural correlates from mouse models have come from the Ts65Dn line. In line with the human MRI data, dendritic spine density is lower in hippocampus (Belichenko) and cortical pyramidal cells of environmentally enriched mutant animals (Dierssen). Disturbances in many neurochemical circuits are also evident in this model. Although there is no consensus as to when disturbances in particular circuits contribute to the numerous behavioural phenotypes (Granholm; Hunter; Seidl; Seo & Isacson), it is possible, for example, that subtle effects in cholinergic neurons from early adulthood may be affecting rest/activity patterns in mouse models. Finally, although neuroendocrine function is disturbed in Down syndrome (Roubertoux & Kerdelhue), there is no specific data from mouse work, suggesting that hypothalamic dysfunction contributes to the rest/activity or sleep fragmentation phenotypes seen in mouse models. Future studies into hypothalamic function in Tc1 mice may help in clarifying its contribution to rest/activity and sleep disturbances. ^{2006 2003 2003 1999 2004 2003 2000 2004 1999 2005 2006} et al. et al. et al. et al. et al. et al.

Sleep disturbances may contribute to cognitive dysfunctions in Down syndrome patients, while individuals with high ratings of sleep disruption have greater difficulties with executive functions (Chen). In a study using optogenetics tools to disrupt sleep in mice, sleep fragmentation in itself can impair mouse performance in an object recognition task without affecting the overall amount or intensity of sleep (Rolls). In general, poor sleep seems to impair memory consolidation (Brown; Stickgold) further exacerbating cognitive impairments in Down syndrome patients and warranting further investigation in mouse models. Tc1 mice display a number of these additional traits with deficits evident in a number of learning and memory paradigms (Morice; O'Doherty). et al. et al. et al. et al. et al. ^{2013 2011 2012 1998 2008 2005}

The investigation of sleep and rhythm-related disturbances in mouse models of Down syndrome shows consistently abnormal parameters, although the contribution of different loci on Hsa21 remains to be clarified. Nevertheless, the continued use of diverse phenotyping tools in different mouse models will be invaluable in furthering our understanding of sleep disturbances in Down syndrome patients.