Modulation of sympathetic vasoconstriction is critical for the effects of sleep on arterial pressure in mice

Abstract

Key points

• While values of arterial pressure during sleep are predictive of cardiovascular risk, the autonomic mechanisms underlying the cardiovascular effects of sleep remain poorly understood.

• Here, we assess the autonomic mechanisms of the cardiovascular effects of sleep in C57Bl/6J mice, taking advantage of a novel technique for continuous intraperitoneal infusion of autonomic blockers.

• Our results indicate that non‐REM sleep decreases arterial pressure by decreasing sympathetic vasoconstriction, decreases heart rate by balancing parasympathetic activation and sympathetic withdrawal, and increases cardiac baroreflex sensitivity mainly by increasing fluctuations in parasympathetic activity.

• Our results also indicate that REM sleep increases arterial pressure by increasing sympathetic activity to the heart and blood vessels, and increases heart rate, at least in part, by increasing cardiac sympathetic activity.

• These results provide a framework for generating and testing hypotheses on cardiovascular derangements during sleep in mouse models and human patients.

Abstract

The values of arterial pressure (AP) during sleep predict cardiovascular risk. Sleep exerts similar effects on cardiovascular control in human subjects and mice. We aimed to determine the underlying autonomic mechanisms in 12 C57Bl/6J mice with a novel technique of intraperitoneal infusion of autonomic blockers, while monitoring the electroencephalogram, electromyogram, AP and heart period (HP, i.e. 1/heart rate). In different sessions, we administered atropine methyl nitrate, atenolol and prazosin to block muscarinic cholinergic, β₁‐adrenergic and α₁‐adrenergic receptors, respectively, and compared each drug infusion with a matched vehicle infusion. The decrease in AP from wakefulness to non‐rapid‐eye‐movement sleep (N) was abolished by prazosin but was not significantly affected by atropine and atenolol, which, however, blunted the accompanying increase in HP to a similar extent. On passing from N to rapid‐eye‐movement sleep (R), the increase in AP was significantly blunted by prazosin and atenolol, whereas the accompanying decrease in HP was blunted by atropine and abolished by atenolol. Cardiac baroreflex sensitivity (cBRS, sequence technique) was dramatically decreased by atropine and slightly increased by prazosin. These data indicate that in C57Bl/6J mice, N decreases mean AP by decreasing sympathetic vasoconstriction, increases HP by balancing parasympathetic activation and sympathetic withdrawal, and increases cBRS mainly by increasing fluctuations in parasympathetic activity. R increases mean AP by increasing sympathetic vasoconstriction and cardiac sympathetic activity, which also explains, at least in part, the concomitant decrease in HP. These data represent the first comprehensive assessment of the autonomic mechanisms of cardiovascular control during sleep in mice.

Key points

• While values of arterial pressure during sleep are predictive of cardiovascular risk, the autonomic mechanisms underlying the cardiovascular effects of sleep remain poorly understood.

• Here, we assess the autonomic mechanisms of the cardiovascular effects of sleep in C57Bl/6J mice, taking advantage of a novel technique for continuous intraperitoneal infusion of autonomic blockers.

• Our results indicate that non‐REM sleep decreases arterial pressure by decreasing sympathetic vasoconstriction, decreases heart rate by balancing parasympathetic activation and sympathetic withdrawal, and increases cardiac baroreflex sensitivity mainly by increasing fluctuations in parasympathetic activity.

• Our results also indicate that REM sleep increases arterial pressure by increasing sympathetic activity to the heart and blood vessels, and increases heart rate, at least in part, by increasing cardiac sympathetic activity.

• These results provide a framework for generating and testing hypotheses on cardiovascular derangements during sleep in mouse models and human patients.

Introduction

Sleep exerts dramatic effects on cardiovascular control. On passing from wakefulness (W) to non‐rapid‐eye‐movement sleep (N), which is typically the entrance into sleep and constitutes most of sleep time, arterial pressure (AP) decreases by approximately 10% while heart period (HP, i.e. the reciprocal of heart rate) increases. On passing from N to rapid‐eye‐movement sleep (R), AP and HP return towards the values they have during W (Silvani, 2008). Higher values of AP during nocturnal sleep predict (Roush et al. 2014) and may even cause (Hermida et al. 2011) higher cardiovascular risk. In spite of the scientific and clinical relevance of these cardiovascular effects of sleep, their autonomic mechanisms are still unclear (Silvani & Dampney, 2013).

AP may decrease during N because of a decrease in cardiac output (Khatri & Freis, 1967; Schneider et al. 1997), due to an increase in HP driven by an increase in cardiac parasympathetic activity (Baust & Bohnert, 1969). However, sympathetic activity to skeletal muscles (Somers et al. 1993), kidneys (Miki et al. 2003) and skin (Takeuchi et al. 1994) may also decrease during N compared to W, raising the hypothesis that the decrease in AP during N is due to a decrease in peripheral vascular resistance. Both hypotheses rest on limited pieces of evidence obtained on different species, and are not mutually exclusive (Silvani, 2008). The uncertainty concerning these hypotheses also concerns the autonomic mechanisms that increase AP during R, which might be parsimoniously explained by the simple reversal of the changes in autonomic activity that decrease AP during N.

The changes in HP associated with either sleep state are the opposite of those in AP. This is inconsistent with the operating logic of the arterial baroreflex (Somers et al. 1993; Silvani, 2008). The sleep‐related cardiovascular changes may thus result primarily from central autonomic commands (Silvani, 2008), and be permitted by resetting of the arterial baroreflex (Bristow et al. 1969; Nagura et al. 2004; Silvani, 2008). Increases in the gain of the baroreflex control of HP (cardiac baroreflex sensitivity, cBRS) have been reported during N (Smyth et al. 1969), although there is contrasting evidence (Silvani, 2008), and the underlying mechanisms remain unclear. In human subjects, changes in cBRS among conditions including N and wakefulness in the lying, sitting and standing positions closely mirror in magnitude those of fluctuations in cardiac parasympathetic activity (Silvani et al. 2017). However, this conclusion was reached based on the analysis of heart rate variability, and more direct evidence on the autonomic mechanisms of cBRS during N and R is lacking.

Mice are arguably the mammalian species of choice to study the functional genomics of sleep and autonomic cardiovascular control. However, information on the autonomic mechanisms of sleep‐related cardiovascular control in mice is presently lacking, and its availability is hindered by substantial technical constraints. The study of mice with mutations that impact on autonomic control may be exacting in terms of time and resources for mouse breeding. Administration of autonomic receptor blockers by intraperitoneal injection (Silvani et al. 2014) entails restraint and injection pain, which may cause long‐lasting sleep disruption. Surgical implantation of miniature pumps (Tan et al. 2011) may solve this problem. However, non‐refillable osmotic pumps are hardly applicable in within‐subject experimental designs, while the greater weight of refillable electronic pumps is a significant limitation to the study of freely behaving mice already loaded with AP transducer and electrodes.

The overarching aim of our study was to determine the physiological autonomic mechanisms of cardiovascular control during sleep in mice. In particular, our study tested the hypotheses that (a) the decrease in AP during N results from an increase in parasympathetic activity on the heart; (b) the decrease in AP during N results from a decrease in sympathetic vasoconstriction; (c) the increase in AP during R is explained by reversal of the changes in autonomic activity that decrease AP during N; and (d) the values of cBRS during sleep critically depend on the parasympathetic activity on the heart.

To these aims, we developed and applied an original technical approach, which consisted of long‐term (8 h) continuous infusions of autonomic receptor blockers through a lightweight, flexible catheter implanted in the peritoneal cavity and connected to a remote precision pump. We performed these infusions in freely behaving mice instrumented for continuous sleep and cardiovascular recordings (Silvani et al. 2009). In different recording sessions on each mouse, we infused atropine methyl nitrate, atenolol and prazosin, which block acetylcholine muscarinic, β₁‐adrenergic and α₁‐adrenergic receptors, respectively. These receptors mediate parasympathetic control of the heart, sympathetic control of the heart and sympathetic vasoconstriction, respectively. For each mouse, we compared the results obtained during infusion of each of these drugs with results obtained during matched infusions of saline (drug vehicle), which closely preceded (2 days before) each drug infusion.

Methods

Ethical approval

The study protocol complied with the EU Directive 2010/63/EU for animal experiments. The experiments were carried out according to the guidelines of the animal welfare committee at the University of Bologna, Italy, and conformed to the principles and regulations for animal experiments reported in The Journal of Physiology (Grundy, 2015). All efforts were made to minimize animal suffering.

Mice

Experiments were performed on 12 male wild‐type mice of the C57Bl/6J strain, which is reportedly the most widely used inbred strain and the first to have its genome sequenced (cf. www.jax.org).

Mice were bred from a colony expanded at the Department of Biomedical and Neuromotor Sciences of the University of Bologna, Italy, from founder mice bought from Charles River Italy (Calco, Italy). Mice were housed under a 12:12‐h light–dark cycle with ambient temperature set at 21–23°C and free access to water and food (4RF21diet; Mucedola, Settimo Milanese, Italy).

Overview of experimental protocol

Mice underwent surgery for implantation of electroencephalographic (EEG) and electromyographic (EMG) electrodes, a telemetric AP transducer, and a catheter in the peritoneal cavity at the age of 32.3 ± 1.5 weeks. The body weight at surgery was 31.7 ± 0.7 g.

After 3 weeks’ recovery from surgery, including 12 days’ habituation to the recording environment and apparatus, each mouse was scheduled to undergo six sessions of simultaneous recordings of sleep and cardiovascular variables, each one taking place during the first 8 h of the light cycle. Three test sessions were performed with intraperitoneal infusion of atropine methyl nitrate, atenolol or prazosin, in random order at 1‐week intervals. The other three control sessions were performed with intraperitoneal infusion of saline, which was the vehicle of each drug, 2 days before each test session. After the sixth recording session, mice were euthanized by anaesthetic overdose (isoflurane 4% in O₂) and autopsied.

Surgery

Surgery was performed under isoflurane anaesthesia (1.8–2.4% in O₂, inhalation route) with intra‐operative analgesia (Carprofen 0.1 mg subcutaneously, Pfizer Italy, Latina). Mice were instrumented with two screw electrodes for EEG recordings (frontal‐parietal differential lead) and two wire electrodes in the neck muscles for EMG recordings. A calibrated telemetric AP transducer (TA11‐PAC10, DSI, New Brighton, MN, USA) was implanted subcutaneously, with the catheter tip advanced via the femoral artery until below the renal arteries (Silvani et al. 2009). In addition, a catheter was inserted with one end in the peritoneal cavity through a small incision along the left subcostal margin. The other end of the catheter was tunnelled subcutaneously to the mouse head and included in an acrylic cap (Respal NF, SPD, Mulazzano, Italy) together with the electrodes. This catheter was crafted in the laboratory from medical‐grade silicone (Silclear Degania, Defries Industries, Australia) with length of 82 mm and volume of approx. 20 μL.

Recordings

The AP signal was recorded via telemetry by an antenna located below the mouse cage. The EEG/EMG signals were recorded via a lightweight electrical cable connected to a rotating electrical swivel (Plastics One, Roanoke, VA, USA) mounted on a balanced suspensor arm. This montage allowed unhindered movement to the mice, which were left undisturbed and unrestrained except for the electrode tether. The AP and EEG/EMG signals were recorded simultaneously inside the mouse home cages with ambient temperature set at 25°C as previously described (Silvani et al. 2009). Data acquisition was performed with LabVIEW software (National Instruments, Austin, TX, USA).

Intraperitoneal infusion

The set‐up for intraperitoneal infusion is illustrated in Fig. 1. A remote infusion pump (model 22 multiple syringe pump, Harvard Apparatus, Holliston, MA, USA) was connected to the intraperitoneal catheter by means of a long (188 cm) extracorporeal tube made of two segments of polyethylene of unequal length (Smiths Medical, Ashford, UK), one closer to the mouse (segment M) and the other closer to the pump (segment P).

Figure 1. Schematic diagram of the experimental set‐up for continuous intraperitoneal infusion in freely behaving mice.

The diagram shows the experimental set‐up for recording two mice at a time with simultaneous sleep and cardiovascular monitoring. See ‘Methods’ for details. EC, extracorporeal.

Segment M had a length of 28 cm, internal diameter of 0.58 mm and volume of 75 μL. Segment M was first connected to the intraperitoneal catheter at the beginning of the habituation to the recording apparatus. At the same time, the electrical cable was connected to the electrodes on the mouse head. Segment M and the electrical cable were left connected to the mouse until the completion of the whole experimental protocol to avoid dishabituation to the recording apparatus and manipulation stress. Segment M was filled with saline at the time of its first connection to the intraperitoneal catheter, and again the week before the first recordings session and after each recording session (see below).

Segment P had a length of 160 cm, internal diameter of 0.86 mm and volume of approximately 1 mL, and was pre‐filled with either sterile saline or drug solutions before each recording session.

Segment M was connected to segment P by means of a short silicone junction, and to the intraperitoneal catheter by means of a 22G blunt needle. Twisting of segment M was prevented by stitching it to the electrical cable that connected the mouse electrodes to the electrical swivel. Segment P could easily accommodate twisting because of its greater length, making a fluid swivel unnecessary.

The free end of segment P was passed through a hole in the electrical swivel before connecting it to a syringe, which was filled with distilled water and mounted on the pump. This set‐up allowed the overall weight of both segments of extracorporeal tubing to be supported by the same counterbalanced suspensor arm that supported the weight of the electrical cable and swivel. A small air bubble was introduced at the pump end of segment P to avoid mixing of the distilled water in the syringe with the saline/drug solutions in segment P. The progress of this bubble along segment P, whose internal diameter was known, was periodically measured with a caliper to check that the actual and nominal infusion rates coincided.

Before the start of each recording session, a rapid infusion of saline or drug solution was performed for 5 min at a rate of 30 μL min⁻¹, so as to fill segment M and the intraperitoneal catheter with the saline or drug solution that was originally contained in segment P. This infusion also resulted in the transfer of approx. 50 μL of this saline or drug solution to the peritoneal cavity in approx. 1.7 min. After this initial bolus loading, recordings were started and infusion of saline or drug solution was performed at 100 μL h⁻¹ for 7 h. Recordings were then stopped. Segment P was replaced with another one filled with sterile saline, taking care to insert a small air bubble at the pump end, as previously described. A fast infusion of 30 μL min⁻¹ was performed for 5 min, so as to fill segment M and the intraperitoneal catheter with saline. Segment P was eventually disconnected, and segment M was plugged. The mouse was then left undisturbed in its recording cage, which was its home cage, with the cable connected to the electrical swivel and segment M stitched to the cable, until the next recording session.

Drugs

Atropine methyl nitrate (SML0732, Sigma‐Aldrich, St Louis, MO, USA), a muscarinic acetylcholine receptor antagonist, was infused at 0.5 mg mL⁻¹ to block parasympathetic nervous activity on the heart. Atenolol (A7655, Sigma‐Aldrich), a hydrophilic selective β₁‐adrenergic receptor antagonist with limited blood–brain barrier transport, was infused at 0.25 mg mL⁻¹ to block sympathetic nervous activity on the heart. Prazosin hydrochloride (P7791, Sigma‐Aldrich), an α₁‐adrenoceptor antagonist, was infused at 0.25 mg mL⁻¹ to block sympathetic vasoconstrictor activity. Each drug was dissolved in sterile saline, and drug solutions were prepared fresh before each recording session.

The drugs and their concentrations were selected arbitrarily based on previous published evidence on cardiovascular control in mice. We did not measure the mouse weight before each recording session because this would have involved stressing the mouse due to handling and tubing/cable disconnection and connection, with consequences for sleep and cardiovascular control during the subsequent recordings. As a consequence, we did not scale drug concentrations or infusion rate to the mouse weight. Nonetheless, based on the average body weight measured at surgery (31.7 g, see above) and on our infusion rate of 100 μL h⁻¹, the atropine methyl nitrate concentration of 0.5 mg mL⁻¹ corresponded to a dose of 1.6 mg atropine per hour and per kg body weight. This implied administering every hour a dose of atropine 20% lower than that (2 mg kg⁻¹) which was shown to be safe and effective as a single bolus injection on parasympathetic cardiac control in mice (Baudrie et al. 2007; Laude et al. 2008). Similarly, our atenolol and prazosin concentrations of 0.25 mg mL⁻¹ corresponded to doses of 0.8 mg atenolol or prazosin per hour and per kg body weight, which were 20% lower than those (1 mg kg⁻¹) which were shown to be safe and effective on sympathetic control as single bolus injections in mice (Gross et al. 2005; Baudrie et al. 2007; Laude et al. 2008).

Sleep scoring and beat‐to‐beat cardiovascular data

Data analysis was performed with software written in Matlab (Mathworks, Natick, MA, USA). The states of W, N and R were discriminated with 4‐s time resolution based on inspection of raw EEG and EMG signals by trained investigators following published criteria (Silvani et al. 2009). Following these criteria, we did not attempt to discriminate between active wakefulness and quiet wakefulness based on neck EMG recordings. Scoring R based only on EEG and EMG recordings is a standard practice in mice, and has been validated with concomitant electrooculographic recordings (Fulda et al. 2011). Sleep structure was assessed by computing the percentage of recording time spent in each wake–sleep state. The average duration and number of spontaneous wake–sleep episodes were also computed taking into account only episodes with duration ≥12 s as previously described (Bastianini et al. 2011), in order to limit the confounding effect of the very short wake–sleep bouts, which are common in mice (McShane et al. 2010).

The values of mean AP (MAP, the average AP in each cardiac cycle), systolic AP (the peak AP in each cardiac cycle) and HP (the time between the onset of successive systolic upstrokes, akin to the electrocardiogram‐derived R–R interval) were computed from the raw AP signal for each heartbeat (Silvani et al. 2009).

Analysis of cardiovascular changes across state transitions

The effect of each drug on cardiovascular control during wakefulness and sleep was assessed by computing the changes in MAP and HP across spontaneous transitions between wake–sleep episodes as previously described (Lo Martire et al. 2012). The analysis involved transitions from N to W (i.e. awakening from N), from W to N (i.e. falling asleep), from N to R (i.e. sleep state transitions) and from R to W (i.e. awakenings from R). Direct transitions from W to R and from R to N were not contemplated because they are absent or exceptional in wild‐type mice. In this study, the minimum duration of the wake–sleep episodes before and after transitions was set to 20 and 48 s, respectively. These time constraints were selected as a compromise between the duration and the number of the episodes analysed. State transitions occurring in the first hour after infusion were also excluded from the analysis as a precaution to avoid the effects of the initial transient in the circulating drug concentration.

Quantitative estimation of the parasympathetic and sympathetic contributions to the control of HP across state transitions

The contributions (effect sizes) of cardiac parasympathetic and sympathetic activity to the sleep‐related changes in HP (∆HP_P and ∆HP_S, respectively) were estimated by analogy to a published procedure developed for phasic autonomic nervous system responses (Berntson et al. 1994). The procedure considers ∆HP_P and ∆HP_S as additive because HP is linearly related to either parasympathetic or sympathetic activity (Berntson et al. 1995; Ursino, 1998). This would not hold for heart rate, which is related non‐linearly to HP and, therefore, to parasympathetic and sympathetic activities.

The analysis was based on grand means (averaged within mouse, then over mice) of the differences (∆) in HP across wake–sleep state transitions, focusing on their last 16 s period (Fig. 2).

Figure 2. Definition of residual and subtractive estimates of the autonomic contributions to sleep‐related changes in heart period.

The panels illustrate the definitions of the residual (∆HP_Sres) and subtractive (∆HP_Ssub) estimates of the contribution of cardiac sympathetic activity to the difference in heart period (∆HP) across state transitions, and of the corresponding estimates (∆HP_Pres and ∆HP_Psub, respectively) concerning cardiac parasympathetic activity. These definitions are illustrated only for transitions from non‐rapid‐eye‐movement sleep (N) to wakefulness (W) for the sake of brevity. A, grand means of ∆HP during atropine methyl nitrate (atropine) infusions and their preceding saline (vehicle) infusions. B, corresponding recordings during atenolol infusions and their preceding vehicle infusions. The dotted vertical lines indicate the time at state transition. The dotted horizontal lines correspond to the average ∆HP during the last 16 s of the transition, indicated by the horizontal black segment.

The ∆HP observed during atropine infusion, which tended to abolish parasympathetic effects on ∆HP, estimated the ‘residual’ contribution of sympathetic activity to ∆HP (∆HP_Sres). The parasympathetic contribution to ∆HP, which tended to be lost during atropine infusion, was estimated by ‘subtraction’ (∆HP_Psub) as the difference between the ∆HP during control saline infusion and ∆HP_Sres. Similarly, the analysis of ∆HP during atenolol infusion, which tended to abolish the sympathetic effects on HP, provided a ‘residual’ estimate of the parasympathetic contribution to ∆HP (∆HP_Pres) and a ‘subtractive’ estimate of the sympathetic contribution to ∆HP (∆HP_Ssub).

Ideally, ∆HP_Pres and ∆HP_Sres should equal ∆HP_Psub and ∆HP_Ssub, respectively. In practice, residual estimates differ from subtractive estimates because of random and systematic (bias) errors. Grand‐averaging of ∆HP limits the impact of random errors, but does not protect from error biases. Sources of error biases include insufficient drug doses causing incomplete receptor blockade, non‐selective drug actions on the central nervous system, physiological interactions between parasympathetic and sympathetic activities, and drug effects exerted on the unblocked autonomic branch through the arterial baroreflex, e.g. in response to changes in AP or because of drug modulation of the baroreflex itself. To limit the impact of error bias, ∆HP_P may be estimated by averaging ∆HP_Pres and ∆HP_Psub, and ∆HP_S by averaging ∆HP_Sres and ∆HP_Ssub.

The error bias may be indexed by half its total range, which is the absolute value of the difference between subtractive and residual estimates:

$$\mathrm{\Delta }\mathrm{H}{\mathrm{P}}_{\mathrm{Pbias}}=\left|\mathrm{\Delta }\mathrm{H}{\mathrm{P}}_{\mathrm{Psub}}-\mathrm{\Delta }\mathrm{H}{\mathrm{P}}_{\mathrm{Pres}}\right|/2$$

$$\mathrm{\Delta }\mathrm{H}{\mathrm{P}}_{\mathrm{Sbias}}=\left|\mathrm{\Delta }\mathrm{H}{\mathrm{P}}_{\mathrm{Ssub}}-\mathrm{\Delta }\mathrm{H}{\mathrm{P}}_{\mathrm{Sres}}\right|/2$$

The validity of ∆HP_P and ∆HP_S may then be evaluated with the coefficients:

$$\mathrm{\Delta }\mathrm{H}{\mathrm{P}}_{\mathrm{Pval}}=\left|\mathrm{\Delta }\mathrm{H}{\mathrm{P}}_{\mathrm{P}}\right|/\left(\left|\mathrm{\Delta }\mathrm{H}{\mathrm{P}}_{\mathrm{P}}\right|+|\mathrm{\Delta }\mathrm{H}{\mathrm{P}}_{\mathrm{Pbias}}|\right)$$

$$\mathrm{\Delta }\mathrm{H}{\mathrm{P}}_{\mathrm{Sval}}=\left|\mathrm{\Delta }\mathrm{H}{\mathrm{P}}_{\mathrm{S}}\right|/\left(\left|\mathrm{\Delta }\mathrm{H}{\mathrm{P}}_{\mathrm{S}}\right|+|\mathrm{\Delta }\mathrm{H}{\mathrm{P}}_{\mathrm{Sbias}}|\right)$$

which are bounded by 0 and 1, and indicate valid estimates if >0.5 (Berntson et al. 1994).

Analysis of cBRS in each wake–sleep state

The cBRS was computed with the sequence technique, employing an algorithm adapted for mice (Laude et al. 2008) and based on beat‐to‐beat values of systolic AP and HP during full‐blown wake–sleep episodes ≥60 s, as previously described (Silvani et al. 2012).

Statistical analysis

The statistical analysis was performed with SPSS Statistics (IBM Corp., Armonk, NY, USA) and the level of significance (alpha level) taken as P < 0.05. The results of each drug infusion session were compared with those of the vehicle infusion session that took place 2 days before. Data were analysed with repeated‐measures ANOVA with Huyn–Feldt correction and two within‐subject factors: the wake–sleep state (3 levels: W, N and R, or 2 levels: pre‐ vs. post‐state transition) and drug/vehicle (2 levels). The statistical analysis of state transitions focused on the average absolute values of MAP and HP during two 16 s periods at the beginning and at the end of transitions, and on the differences (∆) in MAP and HP between these two 16 s periods. In case of significant interaction effects, simple effects of the wake–sleep state and drug vs. vehicle were tested with Student's paired‐sample t test. Data are reported as means ± SEM.

Results

Missing data

Because of battery failure of AP telemetric transducers, we obtained data with each drug infusion from 10 mice out of 12. The analysis of cardiovascular changes across state transitions yielded <3 transitions in the following instances: from W to N during prazosin infusion in one mouse, from N to W during prazosin infusion in two mice, from N to R during atropine methyl nitrate infusion in one mouse, and from R to W during each drug infusion in every mouse but one. For greater robustness, the analysis thus focused on mice with ≥3 transitions from W to N, from N to W and from N to R, and excluded transitions from R to W. The analysis of cBRS could not be performed in one mouse during infusion of atropine methyl nitrate, in which movement artefacts prevented the analysis of episodes of W ≥ 60 s.

Analysis of wake–sleep structure

The results of the analysis of wake–sleep structure are reported in Table 1. Atropine methyl nitrate, atenolol and prazosin did not significantly alter the percentage of recording time spent in each wake–sleep state compared with the respective time‐matched vehicle infusions. However, the average duration of W and N episodes was significantly shortened by each drug, and that of R episodes was significantly shortened by prazosin. Each drug also increased significantly the number of episodes of W and N per hour of recording time (data not shown).

Table 1.

Wake–sleep structure during infusion of each drug and its vehicle

		W	N	R
Atropine	%	31 ± 3 (31 ± 2)	60 ± 3 (58 ± 1)	7 ± 1 (9 ± 1)
	D (s)	71 ± 8 (114 ± 11)*	68 ± 4 (80 ± 6)*	55 ± 5 (56 ± 5)
Atenolol	%	33 ± 1 (32 ± 2)	56 ± 1 (56 ± 2)	8 ± 0 (9 ± 1)
	D (s)	83 ± 5 (120 ± 13)*	59 ± 4 (76 ± 4)*	51 ± 4 (52 ± 3)
Prazosin	%	33 ± 2 (34 ± 3)	56 ± 2 (55 ± 3)	7 ± 1 (9 ± 1)
	D (s)	79 ± 7 (127 ± 20)*	56 ± 5 (74 ± 6)*	44 ± 5 (56 ± 3)*

N, non‐rapid‐eye‐movement sleep; R, rapid‐eye‐movement sleep; W, wakefulness. Data are means ± SEM of the percentage of recording time spent in each state and of the average duration (D) of each wake–sleep state, with n = 10 during infusion of atropine methyl nitrate (atropine), atenolol, prazosin, or saline. Values in parentheses refer to data during the saline (vehicle) infusion session that occurred 2 days before each drug infusion session. ^* P < 0.05, vehicle vs. drug (t test).

Cardiovascular changes across state transitions during infusions of saline

The spontaneous state transitions during infusion of saline (vehicle) 2 days before infusion of atropine methyl nitrate entailed the expected physiological pattern of cardiovascular changes, which was readily evident at the inspection of raw recordings (Fig. 3). This pattern consisted of an increase in AP and a decrease in HP on passing from N either to W (Fig. 3 A and B) or to R (Fig. 3 E and F), and of the opposing changes (i.e. a decrease in AP and an increase in HP) on passing from W to N (Fig. 3 C and D). The raw tracings during infusions of saline 2 days before atenolol or prazosin showed the same pattern, and are not displayed for brevity.

Figure 3. Representative raw recordings across state transitions during vehicle infusion.

Representative raw recordings of electroencephalogram (EEG), electromyogram (EMG) and arterial pressure (AP) during spontaneous transitions from non‐rapid‐eye‐movement sleep (N) to wakefulness (W) (A and B), from W to N (C and D), and from N to rapid‐eye‐movement sleep (R) (E and F). B, D and F show the first and last 4 s epochs of the transitions in A, C and E, respectively, at higher temporal resolution to highlight signal details and the duration of the pulse waves, which corresponded to heart period. All recordings were obtained in sessions with infusion of saline (vehicle) 2 days before infusion of atropine methyl nitrate, but were representative also of recordings during infusions of saline before atenolol and prazosin.

Cardiovascular changes across state transitions during infusions of atropine methyl nitrate

The increase in AP on passing from N either to W (Fig. 4 A and B) or to R (Fig. 4 E and F) and the decrease in AP on passing from W to N (Fig. 4 C and D) were still evident in raw recordings during infusion of atropine methyl nitrate. However, the ∆HP across state transitions was blunted during infusion of atropine methyl nitrate, and could hardly be appreciated at the inspection of raw recordings (Fig. 4 B, D and F).

Figure 4. Representative raw recordings across state transitions during infusion of atropine methyl nitrate.

Panels and abbreviations have the same meaning as in Fig. 3.

These qualitative results were supported by the quantitative analysis of MAP and HP (Fig. 5). In particular, atropine methyl nitrate significantly increased MAP (Fig. 5 A–C) and decreased HP (Fig. 5 G–I) in each wake–sleep state, and significantly reduced ∆HP (Fig. 5 J–L), but not ∆MAP (Fig. 5 D–F), across each state transition.

Figure 5. Analysis of the effects of atropine methyl nitrate on the cardiovascular changes across state transitions.

A–C, absolute values of mean arterial pressure (MAP) during spontaneous transitions from non‐rapid‐eye‐movement sleep (N) to wakefulness (W) (A), from W to N (B), and from N to rapid‐eye‐movement sleep (R) (C) in the course of infusions of atropine methyl nitrate (atropine, black lines) and of the preceding infusions of saline (vehicle, grey lines). The statistical analysis was performed on averages over the time periods indicated by the horizontal bars. D–F, corresponding differences (∆) in MAP across state transitions, computed considering the averages over the horizontal bars to the left of A–C as the respective baseline values. The statistical analysis for D–F was performed on averages over the time periods indicated by the horizontal bars to the right. G–I and J–L show corresponding values of heart period (HP) and ∆HP, respectively. Data are means ± SEM with n = 10 for transitions from W to N and from N to W, and with n = 9 for transitions from N to R. ^* and †, P < 0.05 drug vs. vehicle or pre‐ vs. post‐transition, respectively (t test), in presence of a significant drug/vehicle × pre/post transition interaction effect at ANOVA. [^*] and [^†], P < 0.05, main effects of drug/vehicle and of pre/post transition, respectively (ANOVA), in the absence of a significant drug/vehicle × pre/post transition interaction effect.

Cardiovascular changes across state transitions during infusions of atenolol

As was the case during atropine methyl nitrate infusion, inspection of raw recordings during infusion of atenolol revealed that state transitions still showed the increase in AP on passing from N either to W (Fig. 6 A and B) or to R (Fig. 6 E and F) and the decrease in AP on passing from W to N (Fig. 6 C and D). The ∆HP across state transitions was also blunted during atenolol infusion and not readily evident at visual inspection (Fig. 6 B, D and F).

Figure 6. Representative raw recordings across state transitions during infusion of atenolol.

Panels and abbreviations have the same meaning as in Fig. 3.

The quantitative analysis of MAP and HP indicated that atenolol significantly decreased MAP across transitions from W to N (Fig. 7 B) and from N to R (Fig. 7 C), and significantly reduced the ∆MAP from N to R (Fig. 7 F). In addition, atenolol significantly increased HP in each wake–sleep state (Fig. 7 G–I), decreased the ∆HP from W to N (Fig. 7 K), and virtually abolished the ∆HP from N to R (Fig. 7 L). The effect of atenolol on ∆HP from N to W was not statistically significant (P = 0.053, Fig. 7 J).

Figure 7. Analysis of the effects of atenolol on the cardiovascular changes across state transitions.

Abbreviations and symbols have the same meaning as in Fig. 5 and refer to infusions of atenolol (black lines) and to the preceding infusions of saline (vehicle, grey lines). Data are shown as means ± SEM with n = 10.

Cardiovascular changes across state transitions during infusions of prazosin

Raw recordings during prazosin infusion still showed the increase in AP on passing from N to R (Fig. 8 E–F). However, increases in AP were not readily evident in transitions from N to W (Fig. 8 A–B). On the other hand, the decrease in HP on passing from N either to W (Fig. 8 B) or to R (Fig. 8 F) and the increase in HP from W to N (Fig. 8 D) appeared preserved.

Figure 8. Representative raw recordings across state transitions during infusion of prazosin.

Panels and abbreviations have the same meaning as in Fig. 3.

These qualitative results were supported by quantitative analyses (Fig. 9). In particular, prazosin significantly decreased MAP during W (Fig. 9 A and B) and R (Fig. 9 C), and significantly decreased MAP during N after transitions from W (Fig. 9 B). Prazosin also abolished the increase in MAP from N to W (Fig. 9 D) and the decrease in MAP from W to N (Fig. 9 E), and significantly blunted the increase in MAP from N to R (Fig. 9 F). On the other hand, prazosin did not exert any significant effect on HP across state transitions (Fig. 9 G–L).

Figure 9. Analysis of the effects of prazosin on the cardiovascular changes across state transitions.

Abbreviations and symbols have the same meaning as in Fig. 5 and refer to infusions of prazosin (black lines) and to the preceding infusions of saline (vehicle, grey lines). Data are shown as means ± SEM with n = 8, n = 9 and n = 10 for transitions from N to W, from W to N and from N to R, respectively.

Quantitative estimation of the parasympathetic and sympathetic contributions to the control of HP across state transitions

The estimates of the sympathetic and parasympathetic contributions to ∆HP across transitions between W and N were valid (validity coefficient >0.5) and virtually the same for the two autonomic branches (Table 2). On the other hand, across the transition from N to R, only the estimate of the sympathetic contribution to ∆HP was valid, with an effect size marginally lower than that across transitions between W and N (Table 2).

Table 2.

Quantitative estimation of parasympathetic and sympathetic contributions to the control of HP across state transitions

	N → W		W → N		N → R
Estimates	∆HPP	∆HPS	∆HPP	∆HPS	∆HPP	∆HPS
Residual (ms)	−6	−6	5	3	1	−3
Subtractive (ms)	−6	−7	7	8	−7	−8
Effect size (ms)	−6	−6	6	6	−3	−5
Error bias (ms)	0	1	1	2	4	2
Validity coefficient	0.98	0.93	0.85	0.70	0.45	0.69

∆HP_P and ∆HP_S, contributions of the parasympathetic and sympathetic activity on the heart, respectively, to the change (∆) in heart period (HP) across transitions from non‐rapid‐eye‐movement sleep (N) to wakefulness (W), from W to N and from N to rapid‐eye‐movement sleep (R). Data are means over mice, with n = 9.

Analysis of cBRS in each wake–sleep state

The values of cBRS were significantly and dramatically reduced in each wake–sleep state during infusion of atropine methyl nitrate (Fig. 10 A), whereas they were not significantly altered during atenolol infusion (Fig. 10 B), and they increased slightly but significantly during prazosin infusion (Fig. 10 C).

Figure 10. Analysis of cardiac baroreflex sensitivity in each wake–sleep state.

A, values of cardiac baroreflex sensitivity (cBRS) during wakefulness (W), non‐rapid‐eye‐movement sleep (N), and rapid‐eye‐movement sleep (R) in the course of infusions of atropine methyl nitrate (atropine, black) and of the preceding infusions of saline (vehicle, grey). B and C, corresponding values in the course of infusions of atenolol (B) and prazosin (C) and of the respective vehicle control infusions. Data are means ± SEM with n = 10 for B and C, and with n = 9 for A. ^* and †, P < 0.05 drug vs. vehicle or W/R vs. N, respectively (t test), in presence of a significant drug/vehicle × wake–sleep state interaction effect at ANOVA. [^*] and [^†], P < 0.05, main effects of drug/vehicle and of wake–sleep state, respectively (ANOVA), in the absence of a significant drug/vehicle × wake–sleep state interaction effect.

Discussion

Overview of the experimental findings

The main findings of our study may be summarized as follows: (a) the increase in MAP from N to W, the decrease in MAP from W to N, and the increase in MAP from N to R were significantly blunted by the α₁‐adrenergic receptor blocker prazosin; (b) the increase in MAP from N to R was also significantly blunted by the β₁‐adrenergic receptor blocker atenolol; (c) these changes in MAP across wake–sleep state transitions were not significantly affected by the muscarinic receptor blocker atropine methyl nitrate; (d) atropine methyl nitrate and atenolol exerted strikingly balanced and significant effects on the increase in HP from W to N; (e) atropine methyl nitrate significantly blunted and atenolol virtually abolished the decrease in HP from N to R; and (f) the values of cBRS were significantly decreased in each wake–sleep state by atropine methyl nitrate, whereas they were not significantly affected by atenolol and were slightly increased by prazosin.

The autonomic mechanisms of the effects of N on the mean values of AP and HP

We tested whether the decrease in AP during N results from an increase in parasympathetic activity on the heart or from a decrease in sympathetic vasoconstriction. Our results (Table 3) supported the second hypothesis but not the first one, challenging previously held views. An early report on cats attributed the increase in HP during N to an increase in parasympathetic activity to the heart (Baust & Bohnert, 1969). Conversely, our data demonstrated a striking balance of parasympathetic activation and sympathetic withdrawal, each making approximately the same contribution to the increase in HP during N (Table 2). This discrepancy suggests that sympathetic withdrawal also contributes to the increase in HP from W to N in other species including human subjects. On passing from N to W, the effect of atenolol on ∆HP was not statistically significant (P = 0.053), likely because of enhanced ∆HP variability (Fig. 7 J). Nonetheless, estimates of the parasympathetic and sympathetic contributions to ∆HP on passing from N to W were valid and almost identical to those on passing from W to N (Table 2). Thus, the effect of atenolol on ∆HP on passing from N to W is worth re‐evaluating in experiments with greater sample size to increase statistical power.

Table 3.

Summary of the effects of autonomic blockers on cardiovascular changes across wake–sleep transitions

		PH	SH	SV
∆MAP	↑ (N → W)			×
	↓ (W → N)			×
	↑ (N → R)		×	×
∆HP	↓ (N → W)	×	(×)
	↑ (W → N)	×	×
	↓ (N → R)	?	×

∆MAP and ∆HP, changes in mean arterial pressure and heart period, respectively, attributable to the effects of the parasympathetic activity on the heart (PH), sympathetic activity on the heart (SH) and sympathetic activity on blood vessels (SV) across transitions from non‐rapid‐eye‐movement sleep (N) to wakefulness (W), W to N and from N to rapid‐eye‐movement sleep (R). ‘×’ indicates a significant contribution, and ‘(×)’ indicates that a valid estimate of the effect of SH on HP could be obtained (Table 2) even though the effect of atenolol on ∆HP was not statistically significant (P = 0.053). The symbol ‘?’ indicates an open question, given that parasympathetic withdrawal during R was apparently abolished by atenolol infusion (see text for details).

The occurrence of a decrease in cardiac output during N was investigated with conflicting results by two early studies on human subjects (Khatri & Freis, 1967; Bristow et al. 1969). A later study on dogs indicated that the decrease in MAP during N was paralleled by a decrease in cardiac output and by an increase in HP (Schneider et al. 1997). A recent report on mice without sleep monitoring also indicated that the decrease in AP during the light (rest) period was caused by a decrease in cardiac output. In turn, this decrease in cardiac output was determined primarily by an increase in HP and, marginally, also by a decrease in stroke volume, in the face of a significant increase in peripheral vascular resistance (Kurtz et al. 2014). These results closely mirrored those of previous studies on rats (Smith et al. 1987), monkeys (Engel et al. 1993) and human subjects (Veerman et al. 1995) without sleep monitoring. In monkeys, the lack of an increase in stroke volume during the night‐time rest period in the face of an increase in HP was interpreted as an indication of relative hypovolaemia, caused by the suppression of drinking during night‐time sleep. The corresponding increase in peripheral vascular resistance was unaffected by prazosin and was, accordingly, interpreted as a result of blood flow autoregulation triggered by hypovolaemia (Talan & Engel, 1989). At variance with these results, we found that prazosin was the only drug that dramatically blunted ∆MAP across transitions between W and N in mice. Our results thus indicated that the decrease in MAP during N in mice resulted from a decrease in sympathetic vasoconstrictor activity, and was unaffected by blunting ∆HP with atropine methyl nitrate and atenolol infusions. One possibility to reconcile this discrepancy is that different autonomic and haemodynamic mechanisms contribute to control AP during the rest period at different time scales. Short‐term (seconds to a few minutes) changes in AP between W and N may be driven by changes in sympathetic vasoconstrictor activity (our study), and be superimposed on long‐term (hours) haemodynamic changes associated with hypovolaemia and flow autoregulation (Smith et al. 1987; Talan & Engel, 1989; Engel et al. 1993; Veerman et al. 1995; Kurtz et al. 2014). Accordingly, the difference in AP between W and N is somewhat stable throughout the light–dark cycle in mice, but the values of AP during both W and N are lower during the rest (light) period than during the dark (active) period (Bastianini et al. 2012).

The autonomic mechanisms of the effects of R on the mean values of AP and HP

We tested the hypothesis that the increase in AP during R is explained by the reversal of the changes in autonomic activity that decrease AP during N. Our data did not support this hypothesis, indicating that the mechanisms of cardiovascular control during R are more complex than those during N. In particular, as discussed in the previous paragraph, our results indicated that the decrease in AP during N results from a decrease in sympathetic vasoconstrictor activity. Conversely, we found that the increase in MAP from N to R was significantly blunted both by prazosin (Fig. 9 F) and by atenolol (Fig. 7 F), indicating its dependence not only on an increase in sympathetic vasoconstriction via α₁‐adrenergic receptors, but also by an increase in the sympathetic activity on the heart via β₁‐adrenergic receptors. Accordingly, N decreases sympathetic activity to the skeletal muscles (Somers et al. 1993), kidneys (Miki et al. 2003) and skin (Takeuchi et al. 1994), whereas R increases sympathetic activity to the skeletal muscle (Somers et al. 1993) and at the same time decreases sympathetic activity to the kidneys (Nagura et al. 2004) and, possibly, also to the intestines and skin (Futuro‐Neto & Coote, 1982).

We found that the decrease in HP on passing from N to R was significantly blunted but not abolished by atropine methyl nitrate (Fig. 5 L), which would indicate a role of both parasympathetic withdrawal and sympathetic activation. However, we also found that the decrease in HP on passing from N to R was virtually abolished by atenolol infusion (Fig. 7 L), which would indicate an exclusive role of sympathetic activation. We have no conclusive explanation for this mismatch (see critique of the methods, below), which led to invalid estimates of the parasympathetic contribution to ∆HP during R (Table 2). Nonetheless, our data support a contribution of sympathetic activation to the ∆HP during R, whereas a role of parasympathetic withdrawal remains an open question (Table 3). Our data, therefore, challenge early findings on cats that cardiac sympathetic withdrawal occurs on passing from N to R (Baust & Bohnert, 1969; Futuro‐Neto & Coote, 1982).

The autonomic mechanisms of cBRS in each wake–sleep state

We tested the hypothesis that the values of cBRS during sleep critically depend on the parasympathetic activity on the heart. This hypothesis was supported by our data (Fig. 10). In particular, we found that cBRS during saline infusion was the highest during N, in agreement with previous work on mice (Silvani et al. 2012) and with early work on human subjects (Smyth et al. 1969). During atropine infusion, not only did cBRS decrease markedly and significantly in each wake–sleep state, but the differences in cBRS between N and either W or R were also abolished or substantially blunted (Fig. 10 A).

We found that atenolol did not significantly alter cBRS (Fig. 10 B). Conversely, prazosin increased cBRS (ANOVA main effect, Fig. 10 C), possibly because of central enhancement of parasympathetic activity (Boychuk et al. 2011). These results are in broad agreement with findings on mice without wake–sleep monitoring, in which atenolol was also ineffective on cBRS, and an increase in cBRS after prazosin was noted, but was not statistically significant (Laude et al. 2008). In our study, however, sleep‐related differences in cBRS lost significance during atenolol infusion (Fig. 10 B), and a conspicuous variability of cBRS was evident during prazosin infusion particularly during N and R (Fig. 10 C). Further studies with increased sample size and statistical power may thus be required to exclude subtle effects of atenolol and prazosin on cBRS during sleep.

Critique of the method

Atropine methyl nitrate, atenolol and prazosin (Gross et al. 2005; Baudrie et al. 2007; Laude et al. 2008) are standard drugs employed in mice to discriminate the autonomic mechanisms of cardiovascular control. The low lipophilicity of atropine methyl nitrate and atenolol minimizes drug transport across the blood–brain barrier, and, thus, the likelihood of central nervous system effects. In our study, neither of these drugs had any significant effects on the total amount of time spent in each wake–sleep state, but each significantly decreased the average duration of W and N episodes (Table 1) and correspondingly increased episode number. Fragmentation of N by brief bouts of W may have been elicited at the level of the central nervous system as a result of minimal blood–brain transport of atropine methyl nitrate and atenolol, although these drugs target very different (cholinergic muscarinic vs. β₁‐adrenergic) receptors. Alternatively, alterations in sleep structure may have arisen as a result of changes in baroreceptor afferent activity resulting from the peripheral effects of atropine methyl nitrate and atenolol on MAP (Silvani et al. 2015). Nonetheless, our analysis of state transitions with pre‐set time constraints on episode duration, and our analysis of cBRS on wake–sleep episodes ≥60 s were protected from the confounding effects of altered wake–sleep structure. Moreover, neither atropine methyl nitrate nor atenolol altered significantly ∆MAP across transitions between W and N (Fig. 5 D and E and Fig. 7 D and E). These drugs also yielded valid estimates of the parasympathetic and sympathetic contributions to ∆HP across transitions between W and N (Table 2), which would have been unlikely in case of significant confounding effects of either drug on the central nervous system (Berntson et al. 1994). However, we were unable to obtain valid estimates of the parasympathetic contribution to ∆HP across the transition from N to R (Table 2) because atenolol infusion blunted ∆HP (Fig. 7 L) in excess of what was expected based on atropine methyl nitrate infusion (Fig. 5 L). Since a minimal blood–brain barrier transport of atenolol does occur (Neil‐Dwyer et al. 1981), one possibility is that cardiovascular control during R is particularly sensitive to central β₁‐adrenergic receptor blockade. Excitatory and inhibitory synapses on cardiac vagal motor neurons of the nucleus ambiguous are, indeed, subjected to presynaptic modulation by β₁‐adrenergic receptors (Bateman et al. 2012).

Prazosin administered peripherally is well known to exert central nervous system effects (Menkes et al. 1981). These effects may explain why prazosin is effective in the treatment of post‐traumatic stress disorder (Taylor et al. 2008) and causes much less tachycardia compared to the expected baroreflex response to the fall in AP that it produces (Hardey & Lokhandwala, 1979). This lack of tachycardia may result from presynaptic disinhibition of cardiac vagal motor neurons in the nucleus ambiguous caused by α₁‐adrenergic receptor blockade (Boychuk et al. 2011), and may explain why we did not find any significant effect of prazosin on HP (Fig. 9 G–I) even in the face of substantial decreases in MAP (Fig. 9 A–C). The lack of tachycardia in the face of the decreases in MAP during prazosin infusion cannot be explained based on the slight increase in cBRS (Fig. 10 C), and points to a resetting of the arterial baroreflex control of HP. On the other hand, prazosin did not exert significant effects on the total amount of time spent in each wake–sleep state, but decreased the average duration of wake–sleep episodes (Table 1) and increased significantly the number of W and N episodes, suggesting fragmentation at least of N by brief bouts of W. Nonetheless, prazosin did not significantly alter ∆HP across state transitions (Fig. 9 J–L). This suggests that the central nervous system effects of prazosin did not alter the central autonomic commands responsible for cardiac control during sleep.

As detailed in ‘Methods’, we administered every hour by continuous intraperitoneal infusion drug doses approximately 20% lower than those previously shown as safe and effective on mice by intraperitoneal bolus injection (Gross et al. 2005; Baudrie et al. 2007; Laude et al. 2008). A limitation of our study is that we did not directly demonstrate the adequacy of our drug doses based on the lack of responses to co‐administered receptor agonists. Nonetheless, the validity of the estimates of sympathetic and parasympathetic contributions to ∆HP during N and W (Table 2) provides indirect evidence that our doses of atropine methyl nitrate and atenolol were adequate at least to approximate complete receptor blockade (Berntson et al. 1994). Similarly, adequacy of our prazosin dose is suggested by the occurrence of dramatic decreases in MAP during W and R (Fig. 9 A–C).

We did not compare the effects of the different drugs, which would have required a dose–response curve for each one. Nonetheless, our approach allowed us to compare the effects of each drug between different wake–sleep states. We randomized the order of administration of the different drugs and matched as precisely as possible (2‐day intervals) drug and vehicle infusions. However, we invariably performed vehicle infusions before drug infusions. This limitation was because of lack of information on the pharmacokinetics of atropine methyl nitrate, atenolol and prazosin in mice, which did not allow us to exclude short‐term (2 days) carryover effects of the drugs. These effects may have been more prominent for prazosin, which behaves as a slowly reversible antagonist due to endothelial binding and/or accumulation (Doggrell, 1992).

The lack of available information on the variability of cardiovascular measurements in each wake–sleep state during the infusion of autonomic receptor blockers in mice precluded an a priori statistical power analysis. As discussed above, this limitation may have impacted on our ability to assess the effect of atenolol on ∆HP across transitions from N to W and the effects of atenolol and prazosin on sleep‐related changes in cBRS.

Our finding that atropine methyl nitrate and prazosin altered cBRS (Fig. 10 A and C) begs the question of the potential role of drug‐induced baroreflex alterations in determining our findings on AP and HP across state transitions. A conclusive answer to this question would require a complete characterization of cardiac and sympathetic baroreflex sigmoid functions during each wake–sleep state. Nonetheless, as previously discussed, the sleep‐related changes in AP and HP cannot be primarily explained in terms of baroreflex control (Silvani, 2008), and the validity of the estimates reported in Table 2 for W and N suggests that the baroreflex modulation by atropine methyl nitrate and atenolol was limited, if any.

Sleep transitions may differ in many respects between mice and human subjects. As mentioned in ‘Methods’, mice commonly show transitions between wake–sleep bouts that are much shorter than those of humans, with duration in the order of few seconds to minutes (McShane et al. 2010). On the other hand, the changes in cardiovascular function during the sleep onset period in humans follow a complex time course over several minutes, which is significantly affected by the occurrence of brief arousals from sleep (Carrington et al. 2005). The changes in AP and HP that we analysed in mice may be a better model of the short‐term changes than of the steady‐state changes in these variables that occur during the different wake–sleep states in humans.

The main limitation of our study was that we indirectly inferred changes in autonomic activity from the results of autonomic receptor blockade, instead of measuring autonomic nerve activity directly. During sleep, however, such direct measurements would be extremely challenging and have never been performed on small model organisms such as mice.

Conclusions

Our results on C57Bl/6J mice indicated that N elicited a decrease in MAP essentially by decreasing sympathetic vasoconstriction compared to W. The concomitant increase in HP during N did not contribute significantly to the decrease in MAP, and resulted from a balance of parasympathetic activation and sympathetic withdrawal. On the other hand, R elicited an increase in MAP because of increased sympathetic vasoconstriction and because of increased sympathetic activity to the heart. The concomitant decrease in HP during R resulted, at least in part, from this increase in sympathetic activity to the heart. Finally, the values of cBRS during each wake–sleep state and their increase during N critically depended on parasympathetic activity.

Perspectives

This study provided the first comprehensive assessment of the autonomic mechanisms whereby sleep impacts on MAP and HP in mice, taking advantage of a newly developed technique for continuous intraperitoneal infusion of autonomic blockers. These mechanisms have the potential for translation to human subjects. On the other hand, mice are arguably the species of choice to study the functional genomics of sleep and the autonomic nervous system. Our results and novel infusion technique provide a framework for generating and testing hypotheses on the autonomic mechanisms of cardiovascular alterations during sleep in mouse models of human diseases.

Additional information

Competing interests

The authors declare the absence of competing interests.

Author contributions

Conception or design of the work: A.S. and G.Z. Acquisition of data for the work: V.L.M., S.A., S.B., C.B. and A.V. Analysis and interpretation of data for the work: AS. Drafting the work: A.S. Revising the work critically for important intellectual content: G.Z., V.L.M., S.A., S.B., C.B. and A.V. All authors approved the final version of the manuscript and agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. All persons designated as authors qualify for authorship, and all those who qualify for authorship are listed.

Funding

The work was supported by the University of Bologna (RFO funds to A.S., C.B. and G.Z.; post‐doc salary co‐funding to V.L.M.).

Biography

Viviana Lo Martire, a senior post‐doctoral researcher, and Alessandro Silvani, an associate professor of physiology, enjoy a long‐standing collaboration at the University of Bologna, Italy, where both earned their degrees. Their joint research activity has mainly focused on cardiovascular and respiratory control during sleep in mouse models of obesity and of narcolepsy type 1 (NT1). They contributed to the first demonstrations that NT1 may entail high‐amplitude bursts of electroencephalographic theta waves and blunted sleep‐related reductions in blood pressure and heart rate during sleep, which were subsequently reported for patients. Currently, they are collaborating to understand the mechanisms of these cardiovascular anomalies and to expand their research to models of rapid‐eye‐movement sleep behaviour disorder and the restless legs syndrome.