Cheryl C. H. Yang, Sleep Research Center and Institute of Brain Science, National Yang-Ming University, No. 155, Sec. 2, Linong St, Taipei 11221, Taiwan. Tel.: +886-2-28267058; fax: +886-2-28273123; e-mail: firstname.lastname@example.org
We proposed that the higher incidence of sleep fragmentation, sympathovagal imbalance and baroreceptor reflex impairment during quiet sleep may play a critical role in late-sleep-related cardiovascular events. Polysomnographic recording was performed through wireless transmission using freely moving Wistar–Kyoto rats over 24 h. The low-frequency power of arterial pressure variability was quantified to provide an index of vascular sympathetic activity. Spontaneous baroreflex sensitivity was assessed by slope of arterial pressure–RR linear regression. As compared with early-light period (Zeitgeber time 0–6 h), rats during the late-light period (Zeitgeber time 6–12 h) showed lower accumulated quiet sleep time and higher paradoxical sleep time; furthermore, during quiet sleep, the rats showed a lower δ% of electroencephalogram, more incidents of interruptions, higher σ% and higher β% of electroencephalogram, raised low-frequency power of arterial pressure variability value and lower baroreflex sensitivity parameters. During the light period, low-frequency power of arterial pressure variability during quiet sleep had a negative correlation with accumulated quiet sleep time and δ% of electroencephalogram, while it also had a positive correlation with σ% and β% of electroencephalogram and interruption events. However, late-sleep-related raised sympathetic activity and sleep fragmentation diminished when an α1-adrenoceptor antagonist was given to the rats. Our results suggest that the higher incidence of sleep fragmentation and sympathovagal imbalance during quiet sleep may play a critical role in late-sleep-related cardiovascular events. Such sleep fragmentation is coincident with an impairment of baroreflex sensitivity, and is mediated via α1-adernoceptors.
The peak incidence of many cardiovascular events occurs at the end of the sleep period and before morning awaking rather than at any other period of the day (Argentino et al., 1990; Wolk et al., 2005). Frequent sleep/wake transitions, a higher number of arousals and longer rapid eye movement (REM) stage duration always occur during this period (Somers et al., 1993; Wolk et al., 2005). Neuronal control of the cardiovascular system, a possible etiology (Goff et al., 2010; Iellamo et al., 2004), also dramatically changes in association with these sleep structure changes (Yang et al., 2002; Zoccoli et al., 2001). Evidences related to the above have been obtained from a number of prospective studies (Li et al., 2010; Muller et al., 1985), including 30-min interval blood pressure recording during the whole day, and assessment of the biological index or evoked cardiovascular responses during a diurnal protocol (Hu et al., 2011; Scheer et al., 2009). It seems clear that sleep-related events are important (Legramante et al., 2003), and whether sleep-related autonomic nervous system (ANS) changes trigger cardiovascular events towards the end of sleep (as morning approaches in humans) needs to be explored.
Evidences suggest that sleep pattern changes, such as arousal from QS (Bangash et al., 2008; Catcheside et al., 2001, 2002; Kato et al., 2004; Somers et al., 1993) and transition from non-REM to REM (Guilleminault et al., 1984; Kuo and Yang, 2005; Somers et al., 1993), may result in significant pressor and sympathetic overactivity. Late-sleep seems to be correlated with longer waking, a higher PS time, a higher rate of arousal and a higher transition rate from non-REM to REM (Smolensky et al., 2007; Wolk et al., 2005); all of these events seem to occur in parallel. Evidences about the association of PS and higher incidence among the late-sleep have been particularly evaluated (Miki et al., 2003; Somers et al., 1993). However, the underlying mechanisms involved in such QS-related cardiovascular events during late-sleep are less well understood. We hypothesis that, during QS, the higher incidence of sleep fragmentation, the presence of sympathovagal imbalance and an impairment of the baroreceptor reflex may play a critical role in late-sleep-related cardiovascular events. The aim of this study was to explore the possible role of the end-of-sleep-related ANS and BRS changes in the sleep structure changes that occur during QS among unrestrained and freely moving rats. The intention was to further clarify if ANS malfunctioning (higher sympathetic activity) is correlated with sleep instability during this time. BRS impairment, higher sympathetic activity and sleep fragmentation during QS taken together may have a major impact on the high incidence of end-of-sleep-related cardiovascular disease.
Materials and Methods
Experiments were carried out on 10–13-week-old male Wistar–Kyoto (WKY) rats (n = 18). The rats were obtained from the Animal Center of National Yang-Ming University of Taiwan under the guidelines established by the Position of the American Heart Association on Research Animal Use. They were raised in a sound-attenuated room with a 12:12 hour light:dark cycle (08:30–20:30 lights on), and at an appropriate temperature (22 ± 2 °C) and humidity (40–70%) control. Zeitgeber time 0 (ZT 0) was designated as lights on, and ZT 12 was designated as lights off. The experimental procedures were approved by the Institutional Animal Care and Use Committee of National Yang-Ming University.
Electrodes implantation was performed when the rats were 8–10 weeks old. Under pentobarbital anesthesia (50 mg kg−1, i.p.), the dorsal surface of the skull was exposed and cleaned (Kuo and Yang, 2005; Kuo et al., 2008). Two stainless-steel screws were driven bilaterally into the skull overlaying the parietal cortex (2 mm posterior to and ±2 mm lateral to the bregma), and a reference electrode was implanted 2 mm caudal to the lambda. Care was taken to prevent the electrodes from penetrating the underlying dura. Two seven-strand stainless-steel microwires were inserted into the dorsal neck muscles to record EMG. Electrocardiogram (ECG) was recorded via a pair of microwires placed dorsally under the skin (one was between the cervical and thoracic levels, the other at the lumbar level). A telemetry transmitter (TA11PA-C40; Data Sciences, St Paul, MN, USA) was implanted in the abdomen to collect arterial pressure signals via the tip of an arterial catheter, which was inserted into the abdominal aorta. After surgery, the rats were given antibiotic (chlortetracycline) and housed individually for 1 week to allow recovery. To allow the rats to become habituated to the experimental apparatus, each animal was placed in the recording environment at least 7 days before testing.
The experiments started at least 7 days after surgery. The first part of the experiments (n = 9) consisted of the following. On Day 1, a wireless sensor was mounted on the head of the animal for habituation. On Day 2, 24-h recording was started at the beginning of lights on (08:30, ZT 0). In the second part of the experiment (n = 9; Fig. 7), the effects on sleep parameters and cardiovascular variability changes of treatment with vehicle or two distinct autonomic agents individually (the selective α1-blocker, prazosin, 10 mg kg−1, n = 9; and the selective β1-blocker, atenolol, 20 mg kg−1, n = 8; all from Sigma-Aldrich, St Louis, MO, USA) were evaluated. The time for autonomic agent application was 1 h after lights on. On Day 1, a wireless sensor was mounted on the head of the animal for habituation. On Day 2, 24-h recording started at the beginning of lights on (08:30, ZT 0), and then 1 h later (09:30, ZT 1) the animal was given an intraperitoneal injection of saline (saline, 1 mL kg−1). On Day 4, the animal was recorded using the same time window, but this time the animal was given an intraperitoneal injection of one of the two autonomic agents. On Day 6, the animal was recorded using the same time window and the animal was given an intraperitoneal injection of the other autonomic agent. All cardiovascular variables were analysed with respect to the corresponding sleep–wake stages. The data collected from drug injection to the end of the light phase (ZT 1–6, ZT 7–12) were analysed.
The electrophysiological signals were recorded by a wireless sensor (KY4C; K&Y Lab, Taipei, Taiwan; size: 25 × 2 × 13 mm; weight: 8.6 g), and performance of the telemetry system has been validated (Chen et al., 2011). The EEG, EMG and ECG signals were amplified 1000-, 1000- and 500-fold, respectively, and filtered at 0.16–48 Hz, 34–103 Hz and 0.72–103 Hz, respectively. The EEG, EMG and ECG signals were synchronously digitized by an analog–digital converter at different sampling rates (125, 250 and 500 Hz, respectively). The digitized signals were then wirelessly transmitted to a digital data recorder (KY3; K&Y Lab) at a radio frequency of 2.4 GHz. The arterial pressure signals were wirelessly transmitted to its specific receiver (CATALOG# 272-6001; Data Sciences), and were reconstructed into continuous waveforms by a pulse-interval to voltage converter (KY2; K&Y Lab). Next the arterial pressure waveforms were digitized (at 500 Hz) synchronously with the electrophysiological signals using the same data recorder (KY3). All digitized data were stored in a flash memory card for subsequent off-line analysis. We quantified the delta power (0.5–4 Hz), theta power (6–10 Hz), sigma power (11–16 Hz) and beta power (17–32 Hz; Bjorvatn et al., 1998) of EEG spectrogram based on fast Fourier transformation. Delta, sigma and beta power were used as a measure to evaluate depth of sleep. An augmentation of delta power and a suppression of beta/sigma power indicated increased sleep depth (Uchida et al., 1991).
Sleep–wake stages were classified based on the EEG and EMG records, and the details can be found in our previous reports (Kuo and Yang, 2005; Kuo et al., 2008). Continuous power spectral analysis was applied to the EEG and EMG signals using a Hamming window of 16 s (50% overlap), from which the mean power frequency of the EEG (MPF) and the power magnitude of the EMG were quantified. Thus, the time resolution of the sleep scoring was 8 s. For each time epoch, the conscious state was defined as AW if the corresponding MPF and EMG power were above a pre-defined MPF threshold (TMPF) and EMG power threshold (TEMG), respectively; as QS if the MPF and EMG power were below the TMPF and the TEMG, respectively; and as PS if the MPF was above the TMPF and the EMG power was below the TEMG. If the MPF was below the TMPF and the EMG power was above the TEMG, an erroneous epoch was identified and the corresponding cardiovascular signals were excluded from analysis. The TMPF and TEMG were adjusted using visual inspection by an experienced investigator for each animal and for every 2-h, 5-h and 6-h time window. The time series of MPF first underwent a histogram analysis, from which two separate populations, respectively, related to the AW/PS complex and QS could be identified. Thus, the TMPF could be set to discriminate these two populations. The histogram of the EMG time series also had two populations, but these were, respectively, related to AW and the QS/PS complex. Therefore, TEMG could be set to discriminate these two populations. In the present study, the range of the TMPF was 4.9–6.7 Hz for the early-light period and 5.2–6.9 Hz for the late-light period; the range of the TEMG was 3.6–5.1 ln(mV2) for the early-light period and 3.8–5.3 ln(mV2) for the late-light period. A sleep–wake stage was formed when there were at least six consecutive identical epochs, and an interruption was marked when consecutive epochs were <6. Because interruptions are intrinsic components of sleep, they were included within the sleep stage at which they occurred.
Cardiovascular variability analysis
The analytical procedures for cardiovascular variability have been detailed previously (Kuo and Yang, 2005; Yang et al., 1996). Briefly, the mean arterial pressure (MAP) was obtained by integration of the arterial pulse contour. The R–R interval (RR) was estimated continuously from the digitized ECG signals. The stationary MAP and RR were resampled and interpolated at 64 Hz to provide continuity in the time domain, and then were truncated into 16-s time segments with 50% overlap. These sequences were analysed by fast Fourier transform after application of a Hamming window. The low-frequency power (BLF; 0.06–0.6 Hz) of the MAP spectrogram, and the high-frequency power (HF; 0.6–2.4 Hz) and normalized low-frequency power (LF%; 0.06–0.6 Hz) of the RR spectrogram were quantified. BLF, LF% and HF provided estimates for sympathetic vasomotor activity, cardiac sympathetic modulation and cardiac vagal activity, respectively (Kuo et al., 2005; Task Force of the European Society of Cardiology and the North American Society of Pacing, Electrophysiology, 1996; Yang et al., 1996).
Spontaneous BRS was evaluated by MAP–RR linear regression, as described previously (Kuo and Yang, 2005). In brief, for each sequence analysis, the slope of the linear regression between the MAP and RR pairs that are ascending simultaneously was used to estimate the BrrA. The slope of the linear regression between the MAP and RR pairs that are descending simultaneously was used to estimate the BrrD. At least three beats were utilized to calculate the slope, and a slope was considered valid if the MAP was well correlated (r > 0.85) with the RR (Kuo and Yang, 2005). The data length for sequence analysis was 56 s, which was synchronous with the spectral analysis.
Notwithstanding the fact that each 24-h sleep experiment repeatedly produced a great deal of cardiovascular data, we analysed the original datasets over 24 h, and after vehicle and drug treatment over 10 h (5 h during early-light, ZT 2–6, and 5 h during late-light, ZT 7–12, periods); specifically, we then grouped these results into the three sleep–wake states and calculated their respective means. In this way we were able to focus on the various available cardiovascular parameters during QS. To quantify the QS interruption-induced cardiovascular and autonomic responses, the analysis was conducted on isolated interruptions. The increases of cardiovascular measures associated with an interruption event were determined by the differences between the peak values during the interruption period and the values of the epoch immediately preceding the interruption event.
HF and BLF were logarithmically transformed to correct for the skewness of their distribution (Kuo et al., 1999). The various different effects were assessed using non-parametric test by Wilcoxon two-sample test. Comparisons between two sets of data were carried out using Wilcoxon matched-pairs signed-rank test. Correlation between two parameters was assessed using linear regression analysis. The coefficient of the determinant (r-squared) was used to indicate the percentage of the variance in the dependent variable that can be explained by the regression equation. Differences between values and zero were assessed by 95% confidence interval analysis. Statistical significance was assumed for P < 0.05. The results are presented as means ± SEM.
To explore the possible roles of ANS and sleep fragmentation during late-sleep-related cardiovascular events, 24 h normal rhythm changes of cardiovascular neuronal parameters during QS together with sleep architecture changes were investigated without any other disturbance. In brief, we found that rats revealed significant diurnal changes in their sleep/wake rhythm (with more sleep time during the light period and more waking time during the dark period) as well as the presence of a diurnal rhythm of cardiovascular parameter changes (with lower MAP and higher RR in the light period, which was reversed in the dark period). Especially, the late stage of the light period (ZT 6–12) in rats revealed prominent sleep/wake pattern changes and sleep-related cardiovascular parameter changes compared with the early stage of the light period (ZT 0–6; Fig. 1).
Prominent sleep/wake distributions between late-light and early-light periods in rats
Quantitative analysis revealed that over a whole day (24 h) of recording, rats spend more time asleep (QS and PS) than AW during the 12-h light period. In contrast, rats spend more time AW than asleep during the 12-h dark period (Fig. 2). However, some differences between the early-light (ZT 0–6) and the late-light (ZT 6–12) periods could be identified. For rats, compared with the early stage of the light period, the accumulated AW and PS times were significantly higher, but the accumulated QS time was significantly lower during the late stage of the light period.
Prominent sleep architecture changes between late-light and early-light periods in rats
To demonstrate the differences in EEG activity and sleep architecture between the late-light and early-light periods, we investigated EEG activity and sleep architecture changes during QS using 2-h (Fig. 3, left panel) and 6-h windows (Fig. 3, right panel). As compared with the early-light period (the first point of light period, ZT 0–2 and the early-light period, ZT 0–6), the duration per QS stage and delta power percentage of QS during ZT 6–8, 8–10, 10–12 and ZT 6–12 were significantly decreased. However, the interruption events per minute during ZT 8–10, 10–12 and ZT 6–12, and MPF, sigma and beta power percentage during ZT 6–8, 8–10, 10–12 and ZT 6–12 were significantly increased during the late-light period compared with the early-light period. Our findings suggest that the peak of these changes in EEG activities and unstable sleep occur during the late-light period rather than at any other period of the day.
Prominent QS cardiovascular neural parameter changes between the late-light and early-light periods in rats
To demonstrate the possible role of sleep-related cardiovascular neuronal regulation during the late-light period, we investigated MAP, RR, vascular sympathetic activity index (BLF), cardiac sympathetic activity index (LF%) and cardiac vagal activity index (HF) changes during QS using 2-h (Fig. 4, left panel) and 6-h windows (Fig. 4, right panel). As compared with the early-light period during QS, the BLF and LF% during QS during the late-light period (ZT 6–8, 8–10, 10–12 and ZT 6–12) were significantly raised (Fig. 4). However, the changes in MAP, RR and HF were not significant for ZT 0–2 and ZT 0–6. Our findings suggest that the peak of changes in these cardiovascular neuronal parameters of the sympathetic limb occurs especially during the late-light period rather than at any other period of the day. When the event-related cardiovascular parameter changes were explored, it was found that there were significant phasic arterial pressure surges and tachycardia during the interruptions of QS during the late-light period (Fig. 5). The BLF and LF% also revealed significant phasic surges during the interruption of QS. When these events were compared with the early period, the rats showed more frequent arousals (brief awakening <56 s) in the late period (27.1 ± 6.8 versus 66.3 ± 7.4 events, n = 9, P < 0.05); however, no significant difference in arterial pressure surges (7.48 ± 2.30 versus 8.56 ± 1.19 mmHg, n = 9) and cardiac acceleration (−23.94 ± 4.18 versus −24.05 ± 6.20 ms, n = 9) between early and late periods was found.
Significant correlation between sleep architecture and sympathetic activity (BLF and LF%) during the light period of the QS stage in rats
In order to known whether there is any correlation between sleep indexes, EEG features and cardiovascular sympathetic activity changes, we applied linear regression analysis to these groups of parameters using 2-h windows over both the light and dark periods. Most of the sleep structures during QS in the light period were significantly correlated with the sympathetic indices (Fig. 6). During QS in the light period, accumulated QS time and delta power percentage were negatively correlated with BLF and LF%, while the MPF, sigma and beta power percentage and interruption events were positively correlated with BLF and LF%. Over the same period there was one additional negative correlation between duration of stage and BLF.
During QS in the dark period, the delta power percentage was negatively correlated with BLF and LF%, while the MPF was positively correlated with BLF and LF%. Our findings suggest that, especially in the light period during QS, EEG power and sleep architecture changes show a significant correlation with cardiovascular sympathetic activity. Higher sympathetic activity correlates with less sleep time, a lower sleep depth and higher fragmentation, especially during the light period.
The effects of α1- and β1-adrenoceptor antagonists on sleep structure changes of rats during the late-light period
To explore the role of adrenoceptors involved in sympathetic activity and sleep architecture during the late-light period, we applied α1- and β1-adrenoceptor antagonists to minimize the changes in vascular and cardiac sympathetic activity, respectively, during the light period. As compared with the early-light period during QS, the BLF and LF% during QS during the late-light period (ZT 7–12) of two vehicle groups were significantly raised (Table 1). As compared with vehicle group, prazosin (α1-adrenoceptor antagonist) and atenolol (β1-adrenoceptor antagonist) reduced the increase in BLF and LF% during the light period (Table 1). Interestingly, the increases in interruption events, MPF, sigma and beta power percentage as well as the decrease in delta power percentage were prevented by the α1-adrenoceptor antagonist, but not by the β1-adrenoceptor antagonist (Fig. 7). Our results suggested that the α1-adrenoceptor antagonist may simultaneously block the rise in sympathetic activity (BLF) and diminish late-sleep architecture changes. The higher incidences of sleep fragmentation and sympathetic hyperactivity during the late-light period thus seem to be mediated by α1-adrenoceptor neurotransmission.
Table 1. The average mean values for mean arterial pressure (MAP), R–R intervals (RR), low-frequency power (BLF) of APV, normalized low-frequency power (LF%) and high-frequency power (HF) of HRV during QS using a 5-h window after vehicle/prazosin treatment and after vehicle/atenolol treatment among WKY rats
Prazosin (10 mg kg−1, n = 9)
Atenolol (20 mg kg−1, n = 8)
The late-light period (ZT 7–12) is compared with the early-light period (ZT 1–6). Values are presented as mean ± SEM. ZT, Zeitgeber time.
*P < 0.05, **P < 0.001 versus vehicle control group for the same period.
†P < 0.05 versus the early stage (ZT 1–6) of the same period by Wilcoxon match-pairs signed-rank test.
104.80 ± 9.05
102.46 ± 7.24
107.76 ± 8.62
105.00 ± 25.30
94.38 ± 8.10*
98.11 ± 8.82†
96.44 ± 6.76
96.83 ± 8.42
167.90 ± 13.18
174.35 ± 16.34
174.17 ± 16.51
179.30 ± 18.44
158.04 ± 12.47*
159.74 ± 9.19*
192.88 ± 12.01*
185.42 ± 13.20
BLF, In (mmHg2)
0.18 ± 0.14
0.30 ± 0.20†
0.20 ± 0.20
0.33 ± 0.21†
−0.41 ± 0.33**
−0.45 ± 0.29**
0.16 ± 0.17
0.07 ± 0.27*
58.98 ± 4.04
63.84 ± 3.92†
58.95 ± 4.32
63.41 ± 5.59†
54.83 ± 9.36
59.46 ± 5.26
39.41 ± 5.52**
47.82 ± 5.05**
HF, In (ms2)
0.69 ± 0.54
0.87 ± 0.68
0.71 ± 0.59
0.83 ± 0.72
0.80 ± 0.46
0.76 ± 0.32
0.38 ± 0.46*
0.56 ± 0.39
Prominent BRS impairment during the late-light period of QS in rats
To demonstrate the possible role of central control of cardiovascular system during the late-light period, we investigated the BRS indices (BrrA and BrrD) during QS using 2-h (Fig. 8, left panel) and 6-h windows (Fig. 8, right panel). As compared with the early-light period (ZT 0–2 and ZT 0–6), BrrA and BrrD during ZT 6–8, 8–10, 10–12, and BrrA and BrrD during ZT 6–12 were significantly decreased (Fig. 8). Our findings suggest that the peak of changes in BRS parameters occurs particularly during the late-light period rather than at any other period of the day. This late-light period-related higher sympathetic activity may be partially explained by an impairment of central control of the baroreceptor reflex.
Based on simultaneous sleep/wake structure, HRV, APV and BRS analyses, the present study confirmed various previous findings that the late-light period of rats, as compared with other periods of the day, involves a lower depth of sleep (lower delta power percentage and higher sigma power percentage of EEG; Bjorvatn et al., 1998; Uchida et al., 1991), together with a lower accumulated QS time, a higher accumulated PS time, and a lower duration per QS stage (Alfoldi et al., 1991; Campbell and Feinberg, 1993; Rosenberg et al., 1976; Wolk et al., 2005). Focusing on the changes during the QS stage, we found a number of additional phenomena. These included the findings that the late-sleep period of rats showed a higher incidence of interruptions (event per min), showed more raised sleep-related sympathetic activity and showed lower BRS when compared with the early-sleep period. There are significant phasic arterial pressure surges and cardiac acceleration during the interruption of QS (Fig. 5). Additionally, linear regression analysis revealed that the changes in sleep pattern and changes in cardiovascular neuronal activity were significantly correlated with each other during the light period (Fig. 6). When we focused on temporal changes, our results revealed that, during QS, there was an increase in sympathetic activity that started 2 h earlier than the increase in sleep interruptions (Figs 3 and 4). On manipulating sympathetic activity of the rats by applying α1-adrenoreceptors antagonists, it was found that the rise in sympathetic activity and the sleep pattern changes were blocked during the late-light period (Fig. 7; Table 1). These findings suggest possible roles for ANS in QS during the late-light sleep of rats.
Although we found increases in QS sympathetic vasomotor activity and QS cardiac sympathetic activity during the late-light sleep period using a 2-h window length, there was no change in MAP and RR (Fig. 4). We speculate that this is because these findings are based on an average over the 2-h window, a large proportion of which was resting state, and thus the event-related changes may have been averaged out. To confirm at least some of the cardiovascular changes were still apparent during resting states, an additional analysis was performed on QS stages with interruptions excluded. We found that the increases in BLF and LF% during the late-light period remained significant, while the slopes of the increases were less steep than when interruption epochs were included (data not shown). Therefore, these cardiovascular responses are not exclusive to the interruption events. When the event-related cardiovascular changes were explored, it was found that there were significant phasic arterial pressure surges and cardiac acceleration during the interruptions of QS (Fig. 5). When these events are compared with the early period, the rats show more frequent interruption events in the late period; however, no significant differences in values of arterial pressure surges and cardiac acceleration events between early and late periods were found. We propose that there is a higher level of sympathetic modulation during the late-light period that produces unstable sleep and that this occurs simultaneously with cardiovascular surges during QS; together these are likely to be the mechanism that produces higher risk of cardiovascular accidents during the late-light period.
A number of human studies have revealed that blood pressure and sympathetic activity significantly increase in the morning (Kario et al., 2004; Kawano, 2011). Clinical data have revealed that the changes in blood pressure and sympathetic activity in the morning are greater in older hypertension patients than in younger subjects, which suggests that a surge in blood pressure and sympathetic hyperactivity may be the pathological mechanisms associated with morning-related cardiovascular events in older hypertension patients (Smolensky et al., 2007). However, there is no mention in these studies of simultaneous changes in sleep patterns, vagal activity and BRS control during the late-sleep period. On the other hand, it is well known that ANS and BRS indices are well correlated and dramatically change with an individual’s sleep/wake status (Kuo and Yang, 2005; Silvani et al., 2010; Yang et al., 2002; Zoccoli et al., 2001) As mentioned previously, evidence suggests that sleep pattern changes (Bangash et al., 2008; Catcheside et al., 2001, 2002; Kato et al., 2004) and sleep state transitions (Guilleminault et al., 1984; Kuo and Yang, 2005; Somers et al., 1993) may result in significant arterial pressure and ANS changes. Late-sleep seems to be correlated with longer waking, a higher PS time, a higher rate of arousal and a higher transition rate from non-REM to REM (Smolensky et al., 2007), and all these events seem to occur in parallel. Thus, without discriminating sleep/wake status, the phenomena of late-sleep-related pressor and sympathetic overactivity may occur either directly through excitation by central sympathetic control or indirectly through changes in the sleep pattern.
We have demonstrated in this study that EEG power and sleep architecture changes show a significant correlation with cardiovascular sympathetic activity. If the temporal changes in sympathetic indices (BLF and LF%; Fig. 4, left panel, ZT 6–12) and EEG power (delta%, sigma% and beta%; Fig. 3, left panel, ZT 6–12) are considered, we would suggest that raised sympathetic activity and a lower depth of sleep occur simultaneously. The raised sympathetic activity (BLF and LF%) during late-light period (Fig. 4, left panel, ZT 6–12) occurs first and may be a leading cause of the greater fragmentation (interruption, Fig. 3, left panel, ZT 8–12) during the late-light period. This hypothesis seems to be supported by the information presented in Figs 6 and 7, specifically the high correlation between sleep fragmentation and sympathetic indices. When the rise in sympathetic activity (BLF) during the late-light period is blocked by the presence of α1-adrenoceptor antagonist, this seems to diminish the late-light sleep architecture changes (Fig. 7; interruptions, MPF, delta%, sigma% and beta%). This finding is consistent with previous results, such as the fact the prazosin not only affects cardiovascular autonomic regulation, but also decreases AW and prolongs QS (Kleinlogel, 1989; Kuo et al., 2012). This adrenergic modulation of sympathetic activity and sleep may occur via alpha-receptors binding in the brain (Nelson et al., 2003; Pellejero et al., 1984; Saper et al., 2010).
When a 2-h window analysis is used, our results revealed that PS has higher sympathetic activity than QS (supplemental data Figs S1 and S2); however, there were no detectable cardiovascular or sympathetic changes during the late-light period PS compared with the early-light period PS. The phasic cardiovascular events of PS, such as the incidence and magnitude of arterial pressure and heart rate phasic surges during the late-light period, have been investigated in an earlier study as well (Miki et al., 2003). Our results support those findings, namely that a longer PS time (Fig. 2) is related to increases in blood pressure and to higher sympathetic activity (Fig. S2).
There are a number of limitations to this study. Firstly, morning-related cardiovascular events largely occur in old hypertensive patients, and age-related changes have not been explored in this study. Secondly, hypertension-related changes are also not investigated in this study. Thirdly, the sleep/wake rhythm of rats is the reverse of that of humans, and therefore possible differences in the effects of melatonin on rats make it difficult to directly answer some questions on morning-related cardiovascular events in humans. Fourthly, the role of PS is not explored in depth in the present study, and PS may be involved in other mechanisms that are associated with the higher incidence of cardiovascular events during late-sleep (Miki et al., 2003; Somers et al., 1993).
During the late-sleep period, rats show more sleep transitions and greater sleep-related sympathetic overactivity. This sympathetic overactivity during QS is distinct from the longer time spent in PS during late-sleep, where there is also higher sympathetic activity. In addition, impairment of central control of the ANS during QS together with QS-related α1-adrenoceptor sympathetic neurotransmitter hyperarousal, both of which give rise to sleep fragmentation, are quite possibly a cause of the high frequency of cardiovascular events that occur during late-sleep.
This work was supported by a grant (100AC-B3) from the Ministry of Education, Aim for the Top University Plan, a grant (NSC-98-2314-B-010-022) from the National Science Council (Taiwan) and a grant (99001-62-013) from Taipei City Hospital. The authors did not receive any other financial support from any manufacturer. We thank Ms Wen-Yi Wu and Ying-Hua Huang for their technical support, as well as their help with manuscript production.
This was not an industry supported study. The authors indicate no conflicts of interest.