Voluntary locomotion linked with cerebral activation is mediated by vasopressin V1a receptors in free-moving mice

Authors


S. Masuki: Department of Sports Medical Sciences, Shinshu University Graduate School of Medicine, 3-1-1 Asahi Matsumoto 390-8621, Japan.  Email: masuki@shinshu-u.ac.jp

Key points

  • • Arterial blood pressure rises at the onset of voluntary locomotion, which is probably advantageous for increasing blood flow to contracting muscles without delay.
  • • We previously reported in free-moving mice that the feedback control of arterial blood pressure through peripheral baroreceptors, which was dominant at rest, was suppressed during activation of the cerebral cortex; however, no neurotransmitter for the mechanisms has been identified.
  • • Central suppression of the feedback control of arterial blood pressure at the onset of voluntary locomotion was abolished in vasopressin V1a receptor-deficient mice and by local infusion of a V1a receptor antagonist into the brainstem area, termed the nucleus tractus solitarii, of control mice.
  • • Thus, central vasopressin might play an important role as a neurotransmitter in the pressor response at the onset of voluntary locomotion.

Abstract  We previously reported that cerebral activation suppressed baroreflex control of heart rate (HR) at the onset of voluntary locomotion. In the present study, we examined whether vasopressin V1a receptors in the brain were involved in these responses by using free-moving V1a receptor knockout (KO, n= 8), wild-type mice locally infused with a V1a receptor antagonist into the nucleus tractus solitarii (BLK, n= 8) and control mice (CNT, n= 8). Baroreflex sensitivity (ΔHR/ΔMAP) was determined from HR response (ΔHR) to a spontaneous change in mean arterial pressure (ΔMAP) every 4 s during the total resting period, which was ∼8.7 h, of the 12 h measuring period in the three groups. ΔHR/ΔMAP was determined during the periods when the cross-correlation function (R(t)) between ΔHR and ΔMAP was significant (P < 0.05). Cerebral activity was determined from the power density ratio of θ to δ wave band (θ/δ) on the electroencephalogram every 4 s. Spontaneous changes in θ/δ were significantly correlated with R(t) during 62 ± 3% of the total resting period in CNT (P < 0.05), but only 38 ± 4% in KO and 47 ± 2% in BLK (vs. CNT, both P < 0.001). When R(t) and ΔHR/ΔMAP were divided into six bins according to the level of θ/δ, both were positively correlated with θ/δ in CNT (both P < 0.001), while neither was correlated in KO or BLK (all P > 0.05). Moreover, the probability that mice started to move after an increase in θ/δ was 24 ± 4% in KO and 24 ± 6% in BLK, markedly lower than 61 ± 5% in CNT (both P < 0.001), with no suppression of the baroreflex control of HR. Thus, central V1a receptors might play an important role in suppressing baroreflex control of HR during cerebral activation at the onset of voluntary locomotion.

Abbreviations 
AVP

arginine vasopressin

CBF

cerebral blood flow

CNT

control

EEG

electroencephalogram

HR

heart rate

MAP

mean arterial pressure

NTS

nucleus tractus solitarii

REM

rapid eye movement

R(t)

cross-correlation function between spontaneous changes in HR and MAP

V1a BLK

V1a receptor blockade

V1a KO

V1a receptor knockout

WT

wild-type

ZR(t)

transformed R(t)

ΔHR/ΔMAP

HR response to the spontaneous change in MAP

θ/δ

the power density ratio of θ to δ wave band in EEG

Introduction

Baroreflex control of heart rate (HR) is suppressed at the onset of voluntary exercise (Ebert, 1986; Fisher et al. 2007; Komine et al. 2003). Because the suppression has been reported to occur prior to taking hold of the handgrip dynamometer in humans (Ebert, 1986) and prior to pushing the lever of a feeder system in cats (Komine et al. 2003), this may be a feedforward mechanism to increase arterial pressure at the onset of exercise (Rowell et al. 1996). In addition to these observations under specific conditions focused on the response to a task, we have recently reported that baroreflex control of HR was markedly suppressed whenever spontaneous activation of the cerebral cortex occurred in mice during their daily life. Moreover, we found that after the suppression reached its peak, there was a 5-fold higher probability that the mice started to move in comparison to other time periods. These results suggest a tight link between cerebral activity, baroreflex control of HR and voluntary locomotion (Masuki & Nose, 2009); however, there have been no studies to assess possible mediators of these sequential and dynamic responses.

In the cardiovascular centre of the medulla, the feedback gain of baroreflexes is modulated by signals from the higher brain regions (Rowell et al. 1996). Since arginine vasopressin (AVP) V1a receptors have been reported to be richly expressed in the nucleus tractus solitarii (NTS) of the medulla (Koshimizu et al. 2006) and to regulate the activity of NTS neurons receiving baroreceptor input (Bailey et al. 2006), it is plausible that V1a receptors in the medulla might act as a mediator that receives signals from the higher brain regions and modulates baroreflex control of HR. If so, via this pathway, V1a receptors might significantly contribute to the start of voluntary locomotion, which we have reported to occur at high probability after voluntary cerebral activation (Masuki & Nose, 2009).

With these results as a background, in the present study, we hypothesized that the suppression of baroreflex control of HR after voluntary cerebral activation would be impaired in V1a receptor knockout (V1a KO) mice, and also accompanied by a marked reduction in probability of locomotion after cerebral activation. In addition, we hypothesized that if V1a receptors in the NTS mediated these responses, the findings in V1a KO mice would be confirmed after local infusion of a V1a receptor antagonist into the NTS of wild-type mice.

Methods

Animals

The generation of V1a receptor-deficient (i.e. V1a KO) mice has been described previously (Koshimizu et al. 2006). Mice littermates not deficient in V1a receptors (i.e. wild-type) were used as controls for V1a KO mice. The genetic background of V1a KO and wild-type mice was a mixture of 129sv and C57BL/6J mice. Adult males of these mice were used for the study at 13–22 weeks of age. Body weight was 29.9 ± 0.6 and 31.8 ± 1.2 g in V1a KO (n= 8) and in wild-type control mice (WT, n= 8), respectively, with no significant difference between groups in the 1st experiment (P= 0.17), and it was 33.3 ± 1.1 and 32.4 ± 1.4 g in wild-type mice treated with V1a receptor blockade (V1a BLK, n= 8) and vehicle control (CNT, n= 8), respectively, with no significant difference between groups in the 2nd experiment (P= 0.59) (see below for details of the protocol). They were housed at 25°C and 50% relative humidity with food and water available ad libitum under light conditions from 07:00 to 19:00 h. The procedures used were in accordance with the guiding principles for the care and use of animals in the field of physiological sciences published by the Physiological Society of Japan (2003) with prior approval of the Animal Ethics Committee of Shinshu University School of Medicine. All animals were killed with a pentobarbital overdose at the end of the study.

Drugs

Nonpeptide V1a receptor-selective antagonist (OPC-21268; Tocris Bioscience, Ellisville, MO, USA) (Oshikawa et al. 2004) was dissolved in dimethyl sulfoxide (DMSO). AVP (V9879; Sigma-Aldrich, St Louis, MO, USA) was dissolved in saline and used for the study.

Preparations

After anaesthesia with pentobarbital sodium (50 mg kg−1 body weight, i.p.), three stainless-steel screws (OD 1 mm) of electroencephalogram (EEG) electrodes were placed on the skull surface according to stereotaxic coordinates (Franklin & Paxinos, 2008); AP −1.0 and L +1.0, AP −3.0 and L −1.0 mm from bregma, and AP +1.0 and L +1.0 mm from lambda in mice for both experiments (Masuki & Nose, 2009). In addition, for the 1st experiment, a stainless-steel guide cannula (OD 0.80, ID 0.57, length 8.0 mm) was inserted through the skull so that the tip was positioned on the cortex surface (AP +1.5, L −1.0 and V +1.0 mm from bregma), which was used to hold a laser-Doppler flow probe for cerebral blood flow (CBF) measurement (Masuki & Nose, 2009).

For the 2nd experiment, a stainless-steel cannula (OD 0.36, ID 0.18 mm) was inserted through the skull so that the tip was positioned in the NTS (AP −3.2, L 0.0 and V +4.0 mm from lambda) (Franklin & Paxinos, 2008). The cannula was connected via 2 cm of silastic tubing to an Alzet osmotic pump (model 1002; Durect, Cupertino, CA, USA) that was placed in a subcutaneous cavity. The osmotic pump delivered either the V1a receptor antagonist (0.25 mm dissolved in 5% DMSO/95% artificial cerebrospinal fluid) or vehicle solution (5% DMSO/95% artificial cerebrospinal fluid) at a rate of 0.25 μl h−1 for 2 weeks into the NTS. The implantation was performed according to the method in previous studies (Blaha et al. 2000; Chakrabarty et al. 2009; Hiebert et al. 2002). For the 2nd experiment, the guide cannula for CBF measurement was not implanted due to limited space in the skull. The screws and the guide cannula or the infusion cannula in each mouse were fixed to the skull with dental cement.

In all mice for both experiments, a polyethylene catheter to measure mean arterial blood pressure (MAP) and HR was inserted into the left femoral artery so that the tip was positioned 5 mm below the left renal artery (Masuki et al. 2003a). The catheter was secured to the surrounding leg muscles. The arterial catheter and EEG electrodes were tunnelled subcutaneously and then exteriorized between the scapulae. The exteriorized catheter was connected to a cannula swivel (model TCS2–21; Tsumura, Tokyo, Japan), and the mouse was placed in a cage with a free-moving system (model TFM-170; Tsumura). The arterial catheter was flushed every day with 100 i.u. heparin in 0.2 ml saline. Surgery was performed at least 1 week before measurement (Masuki et al. 2005).

Protocols

The 1st experiment: effects of genetic deletion of V1a receptors To investigate the role of V1a receptors on the relationship between changes in cerebral activity and baroreflex control of HR, we continuously measured EEG, CBF, HR, MAP and activity in free-moving V1a KO and WT mice during the 12 h lights-on period, but not lights-off period when mice moved continuously without resting, which made it difficult to analyse transient changes from the resting to active state.

The 2nd experiment: effects of local infusion of a V1a receptor antagonist into the NTS To investigate whether V1a receptors in the NTS mediate the relationship between cerebral activity and baroreflex control of HR, the V1a receptor antagonist or vehicle was continuously infused into the NTS of wild-type mice using the osmotic pump. About 1 week after starting the infusion, EEG, HR, MAP and activity were continuously measured in free-moving V1a BLK and CNT mice during the 12 h lights-on period.

To confirm that the V1a receptor antagonist infused into the NTS did not leak into the peripheral circulation, we determined a change in MAP after an intra-arterial injection of AVP to evoke peripheral V1a receptor-mediated vasoconstriction. After the injection of 1 μg kg−1 AVP, MAP increased by 32 ± 2 and 29 ± 1 mmHg in V1a BLK and CNT mice, respectively, with no significant difference between groups (P= 0.21).

Measurements

EEG was measured through a band-pass filter of 0.5–30 Hz (Bioelectric Amplifier, model MEG-1200; Nihon Kohden, Tokyo, Japan). CBF was measured by laser-Doppler flowmetry (model FLO-C1 BV; Omegawave, Tokyo, Japan). The flow probe consisted of two glass fibres: one to insert the laser light and the other to detect the reflection. The tips of the fibres were glued together to be 0.5 mm in diameter and inserted through the guide cannula to a depth of 1.5 mm from the skull surface so that the tips were positioned close to the motor cortex (Jentink et al. 1991; Masuki & Nose, 2009).

MAP was measured through a catheter connected to a pressure transducer (model TP-400T; Nihon Kohden). HR was counted from the analog signal of the arterial pressure pulse with a tachometer (model AT-601G; Nihon Kohden) that calculated the inverse of the heart period taken from pulse wave maxima. Activity was monitored with locomotion sensors located on the rectangular frame of inner size 25.5 × 18.5 cm (model LCM-10M; Melquest, Toyama, Japan) in which a mouse plastic cage of outer size 20.8 × 15.5 cm was placed. The sensors were composed of three pairs of an infrared beam lamp and a confronting receiver on the longer frames and two more pairs on the shorter frames with ∼6.3 cm between each lamp or receiver. Mice were connected to the measuring instruments at least 12 h before the measurements.

Data acquisition

EEG, CBF, HR, MAP and activity were digitized and stored in a computer (Dimension 1100; Dell, Kawasaki, Japan) at 128 Hz with data acquisition software (Vital Recorder; Kissei Comtec, Matsumoto, Japan). HR and MAP were re-sampled at 10 Hz through a low-pass filter with an edge frequency of 1.5 Hz to remove pulsatile arterial pressure signals to determine baroreflex control of HR (see below).

Analyses

Data for analyses We analysed the data limited to the resting period in the present study because baroreflex control of HR during locomotion is affected by signals from exercising skeletal muscle and higher brain centres (McIlveen et al. 2001; Komine et al. 2003; Matsukawa et al. 2006; Sala-Mercado et al. 2007), which would make it difficult to assess the relationship between cerebral activity and baroreflex control of HR. Indeed, in the present study, although θ/δ of EEG, CBF and R(t) (see below for details of θ/δ and R(t)) in WT mice were closely linked during rest and just before locomotion, this linkage disappeared after the onset of locomotion (Fig. 1), suggesting that mechanisms other than cerebral activity also affect baroreflex control of HR during locomotion. The criterion to judge the resting period was zero counts of activity for 30 s. As a result, ∼520 min of 720 min was considered the resting period in each mouse and the data during this period were used for the following analyses.

Figure 1.

Typical examples of measurements of a wild-type control (WT) and a V1a receptor knockout (V1a KO) mouse in the free-moving state for 60 min 
Top to bottom: activity counts, ratio of θ to δ wave band in EEG (θ/δ), CBF, cross-correlation function (R(t)) between ΔMAP and ΔHR, ΔHR/ΔMAP, HR and MAP. *R(t) was transformed to ZR(t). θ/δ and ZR(t) determined every 4 s were averaged for a period from t– 40 to t+ 40 s (21 values) while moving t by an increment of 4 s. These values were used to determine the average correlation coefficient and the correlation period (Fig. 2A) and the probability of locomotion after voluntary cerebral activation during the resting period (Fig. 2C).

Baroreflex control of HR More details of the analyses were reported previously (Masuki et al. 2003a,b, 2005). Briefly, the slope of ΔHR/ΔMAP was determined from the HR response to the spontaneous change in MAP every 4 s using the cross-correlation function (R(t)). As shown in Fig. 1, R(t) above (red) and below (blue) the lines of P= 0.05 indicate significantly positive and negative correlations, respectively, which were used to determine positive (red) and negative (blue) ΔHR/ΔMAP. The formulae used for analyses are as follows:

display math

where R(t) is the cross-correlation coefficient between x (=MAP) and y (=HR) at the given time (t) after correction for the delay time (Δt= 0.6 s) of the HR response to MAP change. The inline image(t) and inline image(t) were averaged values of MAP and HR, respectively, from time inline image to inline image (inline image). The slope of inline image determined every 4 s was used as an index of baroreflex sensitivity of HR after R(t) was confirmed to be significant. Although we used an observational technique to assess HR responses to the change in MAP, in WT mice (n= 8), the occurrences of significantly positive and negative R(t) were 25 ± 1 and 58 ± 3% of the total resting period, respectively. By contrast, in a randomized data series (Laude et al. 2008), they were 5 ± 0 and 5 ± 0%, respectively (vs. original data series, both P < 0.001), indicating that correlations between MAP and HR represent physiological rather than random interactions.

Correlation period between θ/δ, CBF and R(t) EEG power density was calculated every 4 s in two frequency bands, δ (0.75–4.0 Hz) and θ (6.0–9.0 Hz), to determine the ratio of θ to δ wave band (θ/δ). CBF values were averaged every 4 s. To assess the relationship between R(t) and these variables quantitatively, we transformed R(t) to ZR(t) as follows (Fisher, 1915):

display math

Since an increase in CBF following an increase in θ/δ occurred at 4–8 min per cycle, their cross-correlation function was determined every 8 min after correction for the delay time by which the highest value was marked in each mouse as performed in the R(t) for the present and our previous studies (Masuki et al. 2003a,b, 2005). Similarly, a cross-correlation function between θ/δ and ZR(t) was determined. The results are shown in Figs 2A and 3A.

Figure 2.

θ/δ, CBF, ZR(t) and ΔHR/ΔMAP in free-moving wild-type control (WT) and V1a receptor knockout (V1a KO) mice 
Means and SE bars are presented for eight WT and eight V1a KO mice. A, the average correlation coefficient between θ/δ and ZR(t) determined every 8 min during the resting period (upper). The positive correlation period between θ/δ and ZR(t), presented as a percentage of the resting period (lower). Data during the resting period for ∼520 min in each mouse were used for the analyses. Total measuring period in each mouse was 720 min. *Values were averaged after z transformation. ***Significant difference from WT mice, P < 0.001. B, CBF, ZR(t) and ΔHR/ΔMAP in response to graded levels of θ/δ. Data used for the analyses were the same as in A (see Table 1 for further details). CBF was expressed as a percentage of the value at the lowest θ/δ. ΔHR/ΔMAP was determined when R(t) between ΔHR and ΔMAP was significant regardless of whether negative or positive. *Significant differences from values at the lowest θ/δ, P < 0.05. C, the probability of locomotion within 40 s after an increase in θ/δ. ***Significant difference from WT mice, P < 0.001. D, CBF, ZR(t), ΔHR/ΔMAP, HR, MAP and activity counts before and after an increase in θ/δ. Because locomotion started 12 s on average after an increase in θ/δ when it occurred, the time of 12 s after the increase was regarded as ‘0 s’, and variables were presented in the range of ±240 s from 0 s. Data were derived according to two criteria: (1) θ/δ increased to > threshold of 2 SD during the total resting period; (2) the increase was preceded by > a 240 s resting period. CBF was similarly expressed as in B. *Because some ΔHR/ΔMAP values were lacking when R(t) was not significant, they were interpolated from the next values and means and SE for eight mice in each group were calculated as in other variables. Red portions indicate significant differences from values at −240 to −200 s.

Figure 3.

θ/δ, ZR(t) and ΔHR/ΔMAP in a free-moving state after local infusion of vehicle control (CNT) or a V1a receptor antagonist (V1a BLK) into the nucleus tractus solitarii of wild-type mice 
Means and SE bars are presented for eight CNT and eight V1a BLK mice. A, the average correlation coefficient between θ/δ and ZR(t) determined every 8 min during the resting period (upper). The positive correlation period between θ/δ and ZR(t), presented as a percentage of the resting period (lower). Data during the resting period for ∼520 min in each mouse were used for the analyses. Total measuring period in each mouse was 720 min. *Values were averaged after z transformation. ***Significant difference from CNT mice, P < 0.001. B, ZR(t) and ΔHR/ΔMAP in response to graded levels of θ/δ. Data used for the analyses were the same as in A. *Significant differences from values at the lowest θ/δ, P < 0.05. C, the probability of locomotion within 40 s after an increase in θ/δ. ***Significant difference from CNT mice, P < 0.001. D, ZR(t), ΔHR/ΔMAP, HR, MAP and activity counts before and after an increase in θ/δ. Red portions indicate significant differences from values at −240 to −200 s. Details of analyses and other legends are provided in Fig. 2D.

Circulatory responses to graded levels of θ/δ To analyse CBF, ZR(t) and ΔHR/ΔMAP in response to graded levels of θ/δ, they were divided into subgroups belonging to four bins of θ/δ with 0.25 increment from 0.5 to 1.5 (see Table 1) with an additional two bins of ≤0.5 and >1.5 after correction for the delay time of their responses to an increase in θ/δ. Mean values of CBF, ZR(t) and ΔHR/ΔMAP in each bin were determined in each mouse and the results are presented as the means and SE for eight mice in each group in Figs 2B and 3B.

Table 1.  Data used to determine CBF, ZR(t) and ΔHR/ΔMAP in response to graded levels of θ/δ
  θ/δ level
≤0.500.51–0.750.76–1.001.01–1.251.26–1.50>1.50
 
  1. Values are the means ± SE. WT, wild-type control mice; V1a KO, V1a receptor knockout mice; CNT, wild-type mice treated with vehicle control in the nucleus tractus solitarii (NTS); V1a BLK, wild-type mice treated with V1a receptor blockade in the NTS; θ/δ, power density ratio of θ to δ wave band on electroencephalogram; CBF, cerebral blood flow; R(t), cross-correlation function between ΔHR and ΔMAP; ZR(t), transformed R(t). The number of data that were used to determine CBF, ZR(t) and ΔHR/ΔMAP in response to graded levels of θ/δ (Figs 2B and 3B) are shown. ΔHR/ΔMAP was determined when R(t) was significant. Significant differences from WT mice, *P < 0.05 and **P < 0.01.

WT (n= 8)
Number of data       
 CBF & ZR(t)2265 ± 3142791 ± 1461422 ± 147572 ± 94223 ± 51199 ± 81
 ΔHR/ΔMAP1137 ± 1461427 ± 93722 ± 110292 ± 67112 ± 2998 ± 43
V1a KO (n= 8)       
 CBF & ZR(t)1105 ± 192**2868 ± 1812103 ± 146**976 ± 122*449 ± 69*601 ± 113*
 ΔHR/ΔMAP476 ± 117**1229 ± 147902 ± 78411 ± 51185 ± 33243 ± 62
CNT (n= 8)       
Number of data       
 ZR(t)1965 ± 3202779 ± 1861510 ± 189637 ± 107285 ± 55301 ± 98
 ΔHR/ΔMAP1327 ± 2181822 ± 157967 ± 138392 ± 71161 ± 35164 ± 60
V1a BLK (n= 8)       
 ZR(t)1952 ± 3003230 ± 1471816 ± 106728 ± 75286 ± 43150 ± 33
 ΔHR/ΔMAP1194 ± 1971954 ± 1171100 ± 93432 ± 60173 ± 3098 ± 22

Probability of locomotion To estimate the probability that mice would start locomotion after an increase in θ/δ, we determined the time at which θ/δ decreased closest to but above the threshold of 2 SD determined during the total resting period after a transient increase above the threshold (Fig. 1). If locomotion, with more than zero counts for 30 s, occurred within 40 s after that time, we judged that it was associated with the increase in θ/δ. We used this time frame based on the results of Masuki & Nose (2009) and the delay time between an increase in θ/δ, a subsequent increase in ZR(t) and locomotion (Fig. 1). The probability was determined as (number of increases in θ/δ accompanied by locomotion/total number of increases in θ/δ) ×100%, during the total resting period. The results are shown in Figs 2C and 3C.

Circulatory responses before and after an increase in θ/δ When we analysed the transient changes in CBF and ZR(t) following an increase in θ/δ (Figs 2D and 3D), we extended the analyses of the data to after the start of locomotion, because it occurred after an increase in θ/δ at high probability. In the probability of locomotion analysis above, because locomotion started 12 s on average when θ/δ decreased closest to but above the threshold of 2 SD during the resting period after a transient increase in θ/δ, the time of 12 s was regarded as ‘0 s’. Variables are then presented in the range of ±240 s from 0 s. We derived results from the data used for calculating the probability of locomotion (Figs 2C and 3C) according to two criteria: (1) θ/δ increased more than the threshold of 2 SD during the total resting period; and (2) the increase was preceded by more than a 240 s resting period. The latter criterion was added to obtain a stable baseline before an increase in θ/δ. About nine episodes met the criteria for inclusion in each mouse, and these variables were averaged every 4 s in the range of ±240 s. They were adopted as representative values for each mouse, and then presented as the means and SE values for eight mice in each group. When some ΔHR/ΔMAP values were lacking immediately before and after an increase in θ/δ due to no significant R(t) during the period, they were interpolated from the next values, and the means and SE for eight mice in each group were similarly calculated for the other variables stated above.

Statistics

Values are expressed as the means ± SE. One-way ANOVA was used to examine any significant differences in the average correlation coefficient and the correlation period of θ/δ with CBF and ZR(t) (text, Figs 2A and 3A) and the probability of locomotion (Figs 2C and 3C) between groups. One-way ANOVA for repeated measures was used to examine any significant differences in the trend changes of variables from the baseline in each group (Figs 2B, D and 3B, D). Two-way ANOVA for repeated measures was used to examine any significant differences in circulatory responses to graded levels of θ/δ (Figs 2B and 3B), the number of data used for the analysis (Table 1), and circulatory responses before and after an increase in θ/δ (Figs 2D and 3D) between groups. Subsequent post hoc tests to determine significant differences in the various pairwise comparisons were performed using Fisher's least significant difference test. Standard regression analysis was used in Fig. 4. All P values < 0.05 were considered significant.

Figure 4.

Relationship between the average correlation coefficient of θ/δ with ZR(t)vs. the probability of locomotion 
The average correlation coefficient of θ/δ with ZR(t) every 8 min vs. the probability of locomotion after an increase in θ/δ in wild-type control (WT) and V1a receptor knockout mice (V1a KO), and after local infusion of vehicle control (CNT) or a V1a receptor antagonist (V1a BLK) into the nucleus tractus solitarii of wild-type mice on individual values of 32 mice (8 mice × 4 groups) for Figs 2A, C and 3A and C. *Values were averaged after z transformation. We found a significantly high correlation between them (R2= 0.66, P < 0.001), indicating that the probability of locomotion after cerebral activation is reduced when linkage between cerebral activity and baroreflex control of HR is impaired.

Results

The 1st experiment: genetic deletion of V1a receptors

Changes in θ/δ and CBF vs. baroreflex control of HR in V1a KO miceFigure 1 shows typical examples of activity counts, θ/δ, CBF, R(t), ΔHR/ΔMAP, HR and MAP every 4 s in the free-moving state for 60 min during the lights-on period. In a WT mouse, R(t) increased as θ/δ and CBF increased while it decreased as they decreased. On the other hand, in a V1a KO mouse, R(t) did not change in response to the changes in θ/δ and CBF. Accordingly, we performed cross-correlation analysis of θ/δ with CBF and ZR(t).

This analysis was performed for the resting period, which was 540 ± 23 and 498 ± 21 min of the total measuring period of 720 min in V1a KO and WT mice, respectively, with no significant difference between groups (P= 0.20). The analysis between θ/δ and CBF showed that the average correlation coefficient every 8 min was 0.36 ± 0.07 and 0.37 ± 0.03 in V1a KO and WT mice, respectively, with no significant difference between groups (P= 0.84). Also, the positive correlation period was 65 ± 6 and 68 ± 2% of the resting period in V1a KO and WT mice, respectively, with no significant difference between groups (P= 0.66). However, as shown in Fig. 2A, the analysis between θ/δ and ZR(t) showed that the average correlation coefficient and the positive correlation period in V1a KO mice were both about half of those in WT mice (P < 0.001).

Figure 2B shows CBF, ZR(t) and ΔHR/ΔMAP in response to graded levels of θ/δ. In WT mice, CBF, ZR(t) and ΔHR/ΔMAP increased as θ/δ increased (all P < 0.001); CBF, ZR(t) and ΔHR/ΔMAP at the highest θ/δ were 134 ± 7%, −0.10 ± 0.03 and −4.9 ± 1.5 beats min−1 mmHg−1, significantly higher than 100 ± 0%, −0.30 ± 0.03 and −12.4 ± 2.0 beats min−1 mmHg−1 at the lowest θ/δ, respectively (all P < 0.001). On the other hand, in V1a KO mice, although CBF increased as θ/δ increased (P < 0.001), neither ZR(t) nor ΔHR/ΔMAP increased (P= 0.07–0.43); ZR(t) and ΔHR/ΔMAP at the highest θ/δ were −0.14 ± 0.03 and −5.9 ± 1.6 beats min−1 mmHg−1, not different from −0.19 ± 0.04 and −4.9 ± 1.5 beats min−1 mmHg−1 at the lowest θ/δ, respectively. These results indicate that baroreflex control of HR in WT mice was suppressed as cerebral activity increased, whereas this response was abolished in V1a KO mice. Additionally, in the analyses, the number of data used to determine CBF and ZR(t) at the lowest θ/δ were significantly less in V1a KO than WT mice, whereas those at four higher bins of θ/δ were significantly more in V1a KO than WT mice (Table 1), indicating that the number of data distributed at a higher level of θ/δ in V1a KO than WT mice. Similarly, the number of data used to determine ΔHR/ΔMAP at the lowest θ/δ was significantly less in V1a KO than WT mice, whereas those at other bins were not different between groups (P= 0.07–0.27).

Increase in θ/δ and voluntary locomotion in V1a KO mice For WT mice, an increase in θ/δ was followed by an increase in CBF and R(t), and then the mouse started to move, whereas this sequential response was not observed in V1a KO mice (Fig. 1). Accordingly, we calculated the probability of locomotion after θ/δ increased beyond the threshold of 2 SD during the total resting period in each mouse. While the threshold was 1.35 ± 0.06 in V1a KO mice, significantly higher than 0.99 ± 0.07 in WT mice (P= 0.003), the frequency of θ/δ increasing beyond the threshold was 22 ± 3 and 21 ± 2 times during the total resting period in V1a KO and WT mice, respectively, with no significant difference between groups (P= 0.82). However, the probability that mice started to move after the increase in θ/δ within 40 s was 24 ± 4% in V1a KO, about one-third of 61 ± 5% in WT mice, as shown in Fig. 2C (P < 0.001).

Figure 2D shows CBF, ZR(t), ΔHR/ΔMAP, HR, MAP and activity counts before and after a transient increase in θ/δ. When values from −240 to −200 s were regarded as ‘baseline’, θ/δ and CBF in WT mice started to increase at about −160 s and peaked at −40 s and −32 s, respectively. Although θ/δ and CBF in V1a KO mice showed a similar temporal response to WT mice, the peak values of θ/δ and CBF were significantly higher in V1a KO than WT mice by 70% (P < 0.001) and 15% (P= 0.04), respectively. As θ/δ and CBF increased, ZR(t) and ΔHR/ΔMAP in WT mice increased, significantly higher than the baseline from −144 to 24 s for ZR(t) (P < 0.001) and from −96 to 28 s for ΔHR/ΔMAP (P < 0.001). In contrast, ZR(t) and ΔHR/ΔMAP in V1a KO mice did not increase significantly from the baseline at any time points (P= 0.06–0.17). In the data set adopted for the analyses, after these responses, mice started to move at the probability of 65 ± 5% in WT mice but only 24 ± 5% in V1a KO mice (vs. WT mice, P < 0.001), consistent with the results in Fig. 2C. HR and MAP in WT mice then started to increase from the baseline at 24 s (P= 0.02) and 64 s (P < 0.001), respectively, whereas HR and MAP in V1a KO mice did not increase at any time points (P= 0.39–0.95).

The 2nd experiment: local infusion of a V1a receptor antagonist into the NTS

Changes in θ/δ vs. baroreflex control of HR in V1a BLK mice The resting period for the analysis was 544 ± 22 and 498 ± 33 min of the total measuring period of 720 min in V1a BLK and CNT mice, respectively, with no significant difference between groups (P= 0.27). As shown in Fig. 3A, the analysis between θ/δ and ZR(t) showed that the average correlation coefficient and the positive correlation period were both significantly lower in V1a BLK than CNT mice (P < 0.001), similar to those in V1a KO mice in the 1st experiment.

Figure 3B shows ZR(t) and ΔHR/ΔMAP in response to graded levels of θ/δ. In V1a BLK mice, similar to the response in V1a KO mice, neither ZR(t) nor ΔHR/ΔMAP increased as θ/δ increased (P= 0.22–0.24); ZR(t) and ΔHR/ΔMAP at the highest θ/δ were −0.37 ± 0.03 and −10.0 ± 1.3 beats min−1 mmHg−1, not different from −0.38 ± 0.03 and −10.5 ± 1.3 beats min−1 mmHg−1 at the lowest θ/δ, respectively. Additionally, in the analyses, the number of data used to determine ZR(t) and ΔHR/ΔMAP was not different between V1a BLK and CNT mice at any bins of θ/δ (P= 0.27–0.82; Table 1).

Increase in θ/δ and voluntary locomotion in V1a BLK mice For the probability of locomotion analysis, the θ/δ threshold of 2 SD was 0.97 ± 0.04 and 1.10 ± 0.07 in V1a BLK and CNT mice, respectively, with no significant difference between groups (P= 0.14). Also, the frequency of θ/δ increasing beyond the threshold was 25 ± 3 and 24 ± 2 times during the total resting period in V1a BLK and CNT mice, respectively, with no significant difference between groups (P= 0.67). However, the probability that mice started to move after the increase in θ/δ within 40 s was 24 ± 6% in V1a BLK, about one-third of 72 ± 5% in CNT mice (P < 0.001) (Fig. 3C), similar to that in V1a KO mice.

Figure 3D shows ZR(t), ΔHR/ΔMAP, HR, MAP and activity counts before and after a transient increase in θ/δ. Although θ/δ in V1a BLK mice started to increase at −136 s and peaked at −60 s, ZR(t) and ΔHR/ΔMAP in V1a BLK mice did not increase from the baseline at any time points (P= 0.07–0.69). In the data set adopted for the analyses, after these responses, mice started to move at a probability of 28 ± 5% in V1a BLK mice, lower than the 74 ± 5% in CNT mice (P < 0.001), but consistent with the results in Fig. 3C. HR and MAP in CNT mice then started to increase from the baseline at 8 and 64 s (both P < 0.001), respectively, whereas HR and MAP in V1a BLK mice did not increase at any time points (P= 0.12–0.62), as observed in V1a KO mice.

Link between cerebral activity, baroreflex control of HR and probability of locomotion

Figure 4 shows the relationship between the average correlation coefficient of θ/δ with ZR(t) every 8 min vs. the probability of locomotion after an increase of θ/δ in individual mice. In total, 32 mice from the WT, V1a KO, CNT and V1a BLK groups (Figs 2A, C and 3A, C) were plotted (8 mice × 4 groups). The average correlation coefficient between θ/δ and ZR(t) varied from −0.02 to 0.51 among the mice, and it was highly correlated with the probability of locomotion after an increase in θ/δ (R2= 0.66, P < 0.001). As shown in the figure, mice with a tight linkage between θ/δ and baroreflex control had a high probability of locomotion whereas mice with limited linkage between θ/δ and baroreflex control had a low probability of locomotion.

Discussion

This study clearly shows a role for AVP V1a receptors in linking cerebral activity, baroreflex control of HR and voluntary locomotion in mice during their daily life. The major findings are: (1) ZR(t), an index of baroreflex control of HR, varied while synchronizing with cerebral activity in both V1a KO and WT mice – however, this occurred in V1a KO mice only half as frequently as in WT mice; (2) suppression of baroreflex control of HR in response to voluntary activation of the cerebral cortex was abolished in V1a KO mice; (3) the probability of locomotion after cerebral activation was markedly lower in V1a KO than WT mice with no suppression of baroreflex control of HR; (4) these findings in V1a KO mice were confirmed after local infusion of the V1a receptor antagonist into the NTS of wild-type mice; and (5) the probability of voluntary locomotion was highly correlated with the suppression of baroreflex control of HR after cerebral activation.

Cerebral activity

In the present study, we used the power density ratio of θ to δ wave band (θ/δ) in EEG as an index of cerebral activity. Lu et al. (2008) compared EEG waves between conscious and sedated rats and reported that θ wave activity increased during the conscious state while δ wave activity increased during sedation. Moreover, Tobler et al. (1997) compared EEG waves between awake and sleeping mice and reported that θ wave activity increased during the awake state while δ wave activity increased during non-rapid eye movement (non-REM) sleep. Additionally, cerebral activity has been suggested to be higher in the awake state than sedation or non-REM sleep (Nofzinger, 2006; Shulman et al. 2009). Therefore, in the present study, we used θ/δ as an index of cerebral activity.

We also used CBF as an index of cerebral activity, because it is well known that tissue blood flow in the brain is tightly coupled to regional neuronal activity (Raichle, 1987). Despite the two different techniques to measure cerebral activity, we found that θ/δ varied while synchronizing with CBF during ∼70% of the total resting period in both WT and V1a KO mice, which was identical to the results from freely moving mice in the previous study (Masuki & Nose, 2009). These results suggest that cerebral activity determined by these methods is reliable, and moreover, that the increase in metabolism close to the motor cortex during neuronal activation was as well preserved in V1a KO as in WT mice.

Cerebral activity and baroreflex control of HR

To assess how voluntary changes in cerebral activity for V1a KO mice influence baroreflex control of HR, baroreflex sensitivity (ΔHR/ΔMAP) was determined in the periods where the cross-correlation function (R(t)) between ΔHR and ΔMAP was significant. More negative R(t) and ΔHR/ΔMAP can be interpreted as the greater contribution of peripheral baroreflex control to arterial blood pressure regulation while less negative R(t) and ΔHR/ΔMAP can be interpreted as the greater contribution of other mechanisms (i.e. central pressor responses; Aslan et al. 2007; Silvani et al. 2011), which has been confirmed in freely moving mice before and after baroreceptor denervation (Masuki et al. 2003a). In the present study, ZR(t) and ΔHR/ΔMAP in WT mice varied while synchronizing with cerebral activity, which was consistent with the previous observation (Masuki & Nose, 2009), indicating a tight linkage between baroreflex control of HR and cerebral activity. However, we found in V1a KO mice that this linkage was impaired (Figs 1 and 2A) and neither ZR(t) nor ΔHR/ΔMAP increased as cerebral activity increased, indicating that the suppression of baroreflex control of HR in proportion to cerebral activation was abolished (Fig. 2B). Thus, the findings from the present study provide the first evidence that the normal linkage between baroreflex control of HR and dynamic changes in cerebral activity was absent when V1a receptors were absent.

Regarding the mechanisms, altered baroreflex sensitivity of HR has been reported in V1a KO mice (Koshimizu et al. 2006), AVP-deficient Brattleboro rats (Imai et al. 1983) and rabbits after AVP antagonist infusion into the brain (Hasser & Bishop, 1990). In addition, V1a receptors have been reported to be richly expressed in the NTS of the medulla (Koshimizu et al. 2006). These results suggest that the receptors were involved in controlling the feedback gain of baroreflex in the cardiovascular centre by signals from higher brain regions (Rowell et al. 1996). However, in the previous studies, because baroreflex sensitivity was determined only once under anaesthesia or after short recovery from surgery, it has remained unknown whether the receptors were involved in the suppression of baroreflex control in response to voluntary cerebral activation in a free-moving state. In the 1st experiment, we demonstrated this by continuously measuring cerebral activity and the baroreflex control of HR in free-moving V1a KO mice; however, it remained unknown which part of the brain expressing V1a receptors is involved in the suppression.

To examine this, the V1a receptor antagonist was locally infused into the NTS of wild-type mice in a free-moving state. The linkage between baroreflex control of HR and cerebral activity was impaired (Fig. 3A) and neither ZR(t) nor ΔHR/ΔMAP increased as cerebral activity increased (Fig. 3B), consistent with the results in V1a KO mice. In this protocol, 0.25 mm V1a receptor antagonist was infused into the NTS at 0.25 μl h−1. This dose was only 1/250th of previous studies in which the same antagonist, OPC-21268, was infused into the lateral ventricle (Chu et al. 2005; Kato et al. 2009), and the infusion volume per hour was 1/800th of cerebrospinal fluid volume (Simchon et al. 1999), suggesting that the effect of the antagonist was localized in the NTS and surrounding area. Therefore, our results from the V1a antagonist infusion suggest that V1a receptors in the NTS are involved in mediating baroreflex suppression in response to cerebral activation.

Alternatively, V1a receptors have been suggested to be expressed also in limbic regions such as in the amygdale, which reportedly enhances emotional reactions to external stress (Bielsky et al. 2004; Ferris et al. 2008; Bosch & Neumann, 2010). Since it is well known that emotional reactions suppress baroreflex sensitivity (Schlor et al. 1984; Knuepfer et al. 1991), the lack of central suppression of baroreflex control of HR observed in V1a KO mice might be caused by reduced emotional reactions to external stress due to the lack of V1a receptors in this region (Bielsky et al. 2004; Ferris et al. 2008; Bosch & Neumann, 2010); however, this is unlikely because we found that cerebral activation in V1a KO mice was not diminished but rather enhanced compared with WT mice (Table 1). These results suggest that the effects of external stress on cerebral activity and baroreflex control were minor in the present study. It might be more reasonable to think that the enhancement was a compensatory response to the impaired baroreflex response to cerebral activation, probably due to the lack of feedback signals from the cardiovascular centre to the higher brain regions. On the other hand, enhanced cerebral activation was not observed after infusion of the V1a antagonist (Table 1), suggesting that the enhancement might result from a long-term but not short-term adaptation to the impairment.

Also, V1a receptors expressed in the peripheral tissues (e.g. liver, kidney, blood vessels, muscle; Hirasawa et al. 1994; Hiroyama et al. 2007) might be involved in the linkage. However, we found that the change in baroreflex control of HR always occurred only several seconds after the change in cerebral activity, which is probably too short a time for circulating AVP to act on the peripheral tissues and link with the baroreflex. Moreover, in the present study, after the V1a antagonist infusion into the NTS of wild-type mice, we confirmed that the dose of the antagonist did not reduce peripheral V1a receptor-mediated vasoconstriction, indicating that impaired linkage (Fig. 3A, B) occurred in the face of preserved V1a receptor property in the peripheral tissues. Thus, because impaired linkage between baroreflex control of HR and cerebral activity was similarly observed after the V1a antagonist infusion into the NTS of wild-type mice and in V1a KO mice, the impairment was not caused by peripheral factors, but by lack of a central mediator, probably located in the NTS, that probably translates signals from the higher brain regions to the cardiovascular centre.

Cerebral activity and baroreflex control of HR at the onset of locomotion

Impaired linkage between baroreflex control of HR and cerebral activity might also disrupt voluntary locomotion. In this context we found that the probability that mice started moving after cerebral activation was markedly lower in V1a KO mice with no suppression of baroreflex control of HR in comparison to WT mice (Fig. 2C, D), and we confirmed this finding in V1a BLK mice (Fig. 3C, D). Moreover, we found that the probability of locomotion was highly correlated with the suppression of baroreflex control of HR after cerebral activation when individual values from all groups were pooled (Fig. 4). With respect to the preparation process for starting exercise, anticipating and/or imaging exercise has been suggested to increase regional CBF in several parts of the cerebral cortex (Thornton et al. 2001; Williamson et al. 2002; Michelon et al. 2006) and to suppress baroreflex control of HR (Ebert 1986; Komine et al. 2003; Masuki & Nose, 2009), consistent with the results from WT and CNT mice in the present study (Figs 2D and 3D). Together, these results suggest that the central effect of the V1a receptor significantly contributes to starting locomotion through suppression of baroreflex control of HR in response to cerebral activation when they intended to move.

However, cerebral activity increased 40 s before starting locomotion (Figs 2C and 3C), which might be too early for mice to have an intention. Denton et al. (1999) suggested in humans that activity in the cingulate, measured by positron emission tomography, was gradually enhanced during the development of thirst. Also, a similar temporal response is likely to occur during hunger (Tataranni et al. 1999; Hinton et al. 2004). Additionally, Vanderwolf (1969) reported in rats that θ wave frequency in EEG increased and peaked just before voluntary movement and suggested that this increase might organize or initiate sequential responses that produce motivated behaviours. Therefore, such a triggering signal for motivated locomotion might be responsible for the increased cerebral activity observed in this study.

Limitations

There are four limitations that deserve additional discussion. First, as mentioned above, the probability of locomotion after the increase in θ/δ was markedly lower in V1a KO than WT mice, whereas the total number of locomotions was 82 ± 9 in V1a KO mice, not different from 104 ± 10 in WT mice (P= 0.13), suggesting that locomotion initiated with motivation was especially impaired in V1a KO mice whereas other movement was relatively maintained. Therefore, if physiological desires had been greatly enhanced, such as by water or food deprivation, we would have seen a significantly reduced total number of locomotions in V1a KO compared with WT mice, which is also likely to occur in V1a BLK mice. Experimentally, Tsunematsu et al. (2008) reported that water deprivation increased locomotor activity in WT mice, but not in V1a KO mice.

Second, we determined the cardiac component of arterial baroreflex, but did not determine the vasomotor component. In anaesthetized rabbits, baroreflex sensitivity of renal sympathetic nerve activity as well as HR was reportedly altered after AVP antagonist infusion into the brain (Hasser & Bishop, 1990), but it is unknown whether V1a receptors are involved in any potential association between central suppression of the vasomotor component of baroreflex and voluntary locomotion. However, in the present study, we found that at least linkage between the cardiac component of baroreflex, cerebral activity and voluntary locomotion was impaired in conscious, freely moving V1a KO and BLK mice.

Third, we did not confirm whether the effect of the V1a receptor antagonist was specific to the NTS. For example, the NTS is located very close to the area postrema at which circulating AVP might act to modulate baroreflex control via the V1a receptors (Hasser & Bishop, 1990); therefore, the possibility that the infusion spread into the area postrema was not excluded in the present study.

Fourth, we did not confirm the sleep/waking state; therefore, an increase in θ/δ before locomotion (Figs 2D and 3D) might have overlapped with REM sleep, characterized by high θ wave activity in EEG (Tobler et al. 1997); however, Silvani et al. (2010) suggested that during REM sleep, ΔHR/ΔMAP was relatively maintained compared with other states in mice, whereas in the present study, ΔHR/ΔMAP before locomotion was markedly suppressed in WT mice, suggesting that the effect of REM sleep in the pre-locomotion period was minor.

In summary, the normal linkage between baroreflex control of HR and dynamic changes in cerebral activity was absent in V1a BLK as well as V1a KO mice with markedly lower probability of locomotion after cerebral activation. Thus, central V1a receptors might play an important role in facilitating voluntary locomotion after cerebral activation by suppressing the baroreflex control of HR.

Appendix

Additional information

Competing interest

No conflicts of interest, financial or otherwise, are declared by the authors.

Author contributions

S.M. and H.N. contributed to the study conception and design. S.M., E.S., T.K., J.Q., K.H., G.T. and H.N. contributed to data collection. S.M. analysed the data. S.M. and H.N. contributed to data interpretation and drafting the article. All authors revised the article for intellectual content and approved the final version.

Funding

This research was supported by grants from the Ministry of Education, Culture, Sports, Science, and Technology of Japan (Grant-in-Aid for Young Scientists to S.M., 21790224), from the Japan Society for the Promotion of Science (Grant-in-Aid for Challenging Exploratory Research to H.N., 21650168), and from the Uehara Memorial Foundation (to S.M.).

Acknowledgements

We thank Drs Yoshiaki Yamaguchi and Hitoshi Okamura (Kyoto University) for valuable suggestions to perform the pharmacological experiment from the viewpoint of neuroscientists. We also thank Dr Michael J. Joyner (Mayo Clinic) for English editing of the manuscript.

Ancillary