Skin sympathetic nerve activity component synchronizing with cardiac cycle is involved in hypovolaemic suppression of cutaneous vasodilatation in hyperthermia

Abstract

Non-technical summary

Thermoregulatory responses during exercise are reduced following thermal dehydration. If individuals do not rehydrate adequately, it could lead to heat exhaustion or stroke with the worst case scenario being death. Plasma volume loss during dehydration has been suggested to suppress cutaneous vasodilatation in response to hyperthermia via a baroreflex-mediated reduction in active vasodilator activity rather than enhanced active vasoconstrictor activity. However, no changes in the electrical signals of the efferent neural pathway have ever been identified. In the present study, we found a component of efferent skin sympathetic nerve activity that was synchronized with the cardiac cycle in thermally stressed individuals. This nerve activity increased with an increase in oesophageal temperature and the increase was significantly suppressed by hypovolaemia. Thus, this component of skin sympathetic nerve activity might represent the active vasodilator signals that regulate skin blood flow during hyperthermia in humans.

Abstract

Although cutaneous vasodilatation in hyperthermia was suppressed during hypovolaemia, the efferent neural pathway mediating this suppression has not been identified. To determine the electrical nerve signals which account for the suppression of cutaneous vasodilatation during hypovolaemia, skin sympathetic nerve activity (SSNA; microneurography) from the peroneal nerve, laser-Doppler blood flow (LDF) on the ipsilateral dorsal foot, mean arterial pressure (MAP; sonometry) and oesophageal temperature (T_oes) were measured before and during 45 min of passive warming in 20 healthy subjects during normovolaemia (n = 10) or hypovolaemia (n = 10) conditions. Hypovolaemia was achieved by diuretic administration. Cutaneous vascular conductance (CVC = LDF/MAP), SSNA burst frequency and total SSNA obtained from rectified and filtered SSNA signal increased as T_oes increased by ∼0.5°C by the end of warming in both groups. The increase in CVC was significantly lower in hypovolaemia than normovolaemia (P < 0.0001), but with no significant difference in the increase in burst frequency and total SSNA between groups (P > 0.32). However, using an alternative analysis that constructed spike incidence histograms from the original signal using 0.05 s bins during the 5 s following a given R-wave, we found a SSNA component synchronized with the cardiac cycle with a 1.1–1.3 s latency. This component increased with an increase in T_oes and the increase was significantly suppressed by hypovolaemia (P < 0.0001). In conclusion, hypovolaemic suppression of cutaneous vasodilatation during hyperthermia might be caused by a reduction in the SSNA component synchronized with cardiac cycle.

Introduction

Thermoregulatory responses during exercise are reduced following thermal dehydration. If individuals do not rehydrate adequately, it could lead to heat exhaustion or heat stroke with the worst case scenario being death (Rothstein & Towbin, 1947). A reduction in plasma volume (PV) loss has been suggested to suppress cutaneous vasodilatation in response to hyperthermia via baroreflex (Nadel et al. 1980; Ikegawa et al. 2011) -mediated changes in an efferent pathway that resulted in a reduction in active vasodilator drive rather than an increase in active vasoconstrictor activity (Edholm et al. 1957; Kellogg et al. 1990). However, neurophysiological data supporting this hypothesis are lacking.

Since both skin blood flow (SkBF) and sweat rate (SR) increase simultaneously with a rise in body temperature, it has been difficult to distinguish between the neural signals driving active vasodilatation and that of sweating.

We have recently reported that in exercising humans under warm environmental conditions, cutaneous vasodilatation was reduced in hypovolaemia whereas the sweating response was sustained (Kamijo et al. 2005b). Therefore, we surmised that the response of the neural pathway for active vasodilatation would differ from that for sweating under conditions of combined hypovolaemia and hyperthermia. Furthermore, the neural efferent signal riding in the skin sympathetic nerve activity (SSNA) is probably controlled by an afferent input to the cardiovascular centre via baroreceptors. As such, it would be enhanced/suppressed by loading/unloading of receptors in a beat-by-beat fashion with cardiac cycles as is known to occur in muscle sympathetic nerve activity (MSNA) (Vallbo et al. 1979).

Therefore, in the present study, we hypothesized that there is a component of the SSNA that is synchronized with the cardiac cycle that increases with an increase in thermal stress and cutaneous vasodilatation. In addition, the increase in this component of SSNA during thermal stress would be suppressed in hypovolaemia with reduced cutaneous vasodilatation. To examine these hypotheses, we analysed raw SSNA signals in relation to cardiac cycles continuously during 45 min of passive warming in normovolaemia and hypovolaemia conditions.

Methods

Subjects

The procedures in this study conformed to the guidelines of the Declaration of Helsinki and were approved by the Review Board on Human Experiments, Shinshu University School of Medicine. After the experimental protocols had been fully explained, 20 healthy and non-smoking young men gave their written informed consent before participating in this study. Subjects were divided into two groups, normovolaemia (n = 10) and hypovolaemia (n = 10), with no significant differences in any physical characteristics between the two groups: 23 ± 2 and 24 ± 5 years of age (mean ± SD), 171 ± 3 and 173 ± 7 cm tall, and 62.9 ± 5.9 and 64.9 ± 7.6 kg body weight, respectively, with no significant differences between groups (P > 0.304).

Experimental protocol

In the hypovolaemia trial, subjects reported to the laboratory at 06.30 and sat in a test room, at an ambient temperature (T_a) of ∼28°C and ∼50% relative humidity (RH). Subjects ate a light meal with 500 ml tap water 30 min before arrival. After the first blood sample was taken at 07.00 through an 18-G Teflon catheter placed in the left antecubital vein, subjects took 20 mg furosemide, and rested in the room for an additional 3.5 h voiding the bladder when necessary. During the waiting period, they were allowed to drink water ad libitum, but without any food, for the first 2.5 h to prevent plasma hyperosmolality but were restricted from drinking from this point until the end of the experiment. The fluid balance ( = total water intake − total urine volume) during this period was −1433 ± 92 ml in hypovolaemia, significantly higher than −217 ± 148 ml in normovolaemia (P < 0.0001). The protocol in normovolaemia was the same as that in hypovolaemia except for taking furosemide. After the resting period, they wore a tube-lined perfusion suit to cover the entire skin surface except for the face, hands, feet and a leg where SSNA was measured. Subjects in both groups then entered an experimental chamber at ∼28°C T_a and ∼20% RH and sat in a semi-recumbent position while all measurement devices were applied. All measurements were made continuously during the experimental trial. During the 10 min thermoneutral condition, 34°C water was perfused through the suit. During the 45 min passive warming period, 47°C water was perfused through the suit. We took the 2nd blood sample in the 5th minute of the thermoneutral condition and the 3rd in the last minute of warming. After each trial, we determined maximal SkBF by heating the skin site to ∼44°C for 30 min using an incandescent lamp.

Measurements

Microneurography

A tungsten microelectrode with an impedance of 4 MΩ at 1 kHz, <5 μm tip and 200 μm shaft diameter, was inserted percutaneously into cutaneous nerve fascicles in the peroneal nerve at the posterior aspect of the head of the fibula to record multiple-unit postganglionic SSNA. An Ag–AgCl electrode (Vitrode Bs; Nihon Kohden Corporation, Tokyo, Japan) was placed on the skin surface ∼5 cm from the active electrode as a reference. The nerve signal was pre-amplified 10,000-fold (DAM80; WPI Inc., Sarasota, FL, USA), transferred to a digital tape recorder, and sent in parallel to a loud speaker, passed through a band-pass filter of 700–2000 Hz as in previous studies, then rectified and filtered (τ = 0.1 s). We confirmed that the signals completely met the following criteria for SSNA: (1) subjects reported that they felt paraesthesia in the dorsal foot without any numbness, (2) a loud sound was evoked by a deep breath, sudden arousal or gentle touch within the innervated area, but (3) was not evoked by a Valsalva's manoeuver (Bini et al. 1980; Kamijo et al. 2005a).

SkBF and SR

SkBF was measured by laser-Doppler velocimetry (τ = 0.1 s; ALF21, ADVANCE, Tokyo) at the site of the ipsilateral dorsal foot, innervated by the peroneal nerve, but devoid of any superficial cutaneous veins. SR was measured by a capacitance hygrometer (Hygro Flex1, Rotronic Inst., Huntington, NY, USA). A small capsule was placed on the dorsal foot (0.79 cm² area) beside the laser-Doppler probe to detect sweating. Dry air was ventilated through a small capsule placed on the foot at 60 ml min⁻¹. Relative humidity and temperature in the sweat capsule were monitored and SR was calculated from them. SkBF and SR were shown as averaged values every minute. Cutaneous vascular conductance (CVC) was calculated as SkBF/mean arterial pressure (MAP) and is presented as a percentage of the maximal SkBF (% max).

Body temperatures and cardiovascular and respiratory variables

Oesophageal temperature (T_oes) and skin temperature were monitored with thermocouples. Mean skin temperature (T_sk) was determined by Robert's equation (Roberts et al. 1977). Heart rate (HR) was calculated by averaged R-R intervals of the ECG every minute. Systolic (SBP) and diastolic (DBP) blood pressures were measured by automated brachial auscultation. Pulse pressure (PP = SBP − DBP) and MAP ( = PP/3 + DBP) were calculated. Respiratory rate (RR) was calculated from voluntary respiratory movement (see Supplemental material for more information).

Blood composition

A 1 ml aliquot of each 11 ml blood sample was used to determine the haematocrit (in %) and haemoglobin concentration (in g dl⁻¹), which were used to calculate the per cent change (Δ) in PV (Greenleaf et al. 1979). Three milliliters of the aliquot were immediately centrifuged at room temperature and the aliquots of plasma were used to determine plasma osmolality (P_osmol, in mosomol (kg H₂O)⁻¹). The remaining 7 ml of aliquots were placed in a chilled tube (EDTA2Na, 1.5 mg ml⁻¹), which was centrifuged at 4°C to determine plasma adrenaline, noradrenaline, arginine vasopressin and angiotensin II concentrations ([Ad]_p, [NA]_p, [AVP]_p and [AGII]_p, respectively). These samples were stored at −85°C until hormone assays were performed (see Supplemental material for more information).

Data analyses

Bursts frequency (BF) and total SSNA (tSSNA) were determined by methods described previously and as shown in the Supplemental material (Kamijo et al. 2005a).

We also determined a component of the SSNA signal that was synchronized with the cardiac cycles. Spike incidence from the original signal was counted in 0.05-s bins from a given R-wave over the following 5 s (Fig. 1A and B) and was presented as a histogram as reported by Macefield & Wallin (1996). The same number of the 5 s incidence histograms as HR were averaged every minute from the 5th minute of the thermoneutral condition to the 45th minute of warming (Fig. 1C). Since we found more than three peaks in the averaged histogram, we determined peak-to-peak intervals for SSNA spike incidence (T_S) and that for R-wave incidence (T_R) where the maximal value of auto-correlation function was observed. Similarly, we determined the latency of SSNA firing after a given R-wave (T_RS) by cross-correlation analysis. We confirmed that the maximal values of autocorrelation and cross-correlation functions were significant (P < 0.05). If maximal correlation coefficients were not significant in some periods, we interpolated the values from the values at the closest times; however, only ∼1% of the data fell into this category. In addition, we calculated the area above the valley of the periodic cycle of the averaged histogram for 5 s: upper area (UA), as a component synchronized with the cardiac cycle and that below the valley, lower area (LA), as a component non-synchronized with the cardiac cycle (Fig. 1D). We calculated UA and LA per minute (UA_min and LA_min, respectively) by multiplying UA and LA for 5 s by 60 s/5 s and also calculated UA and LA per one heart beat (UA_beat and LA_beat, respectively) by dividing the UA and LA for 5 s by 5 s/T_S.

Figure 1. The procedure of SSNA analyses.

A, ECG (upper) and original recording of skin sympathetic nerve activity (SSNA; bottom). B, R-wave and spike incidence histograms from the time of a given R-wave of ECG in the 5-s window with 0.05 s bin. The number of 1, 2 and 3 …. on the histograms corresponds to the numbers of 5-s windows in A. C, the averaged histograms for every minute (left) and auto-correlation functions (right) of R-wave incidence (upper) and SSNA spike incidence (bottom), respectively. T_R and T_S, the peak-to-peak intervals of the R-wave and the SSNA spike incidences, respectively, determined from the averaged histograms of each minute in individual subjects by auto-correlation analysis; T_RS, the time between a given peak of R-wave incidence and the following peak of SSNA spike incidence by cross-correlation analysis. D, the upper (UA) and the lower (LA) areas, which are the components synchronized and non-synchronized with the cardiac cycle, respectively, are shown in one of each subject in normovolaemia (Nor, left) and hypovolaemia (Hypo, right) at a minute of thermoneutral condition (upper) and 45th minute warming (bottom).

Statistics

Two-way ANOVA for repeated measures was used. The standard least-square method was used to determine regression equations between changes in UA_min (ΔUA_min) and CVC (ΔCVC) and between changes in LA_min (ΔLA_min) and SR (ΔSR) from their thermoneutral values during warming (Fig. 6). All values are means ± SEM. The null hypothesis was rejected at P < 0.05.

Figure 6. Relationships between changes in cutaneous vascular conductance (ΔCVC) from thermoneutral values and those in upper area per minute (ΔUA_min; A) and between changes in sweat rate (ΔSR) and those in lower area per minute (ΔLA_min; B) from thermoneutral values during warming in normovolaemia (open circles) and hypovolaemia (filled circles).

Values are means for 10 subjects determined every minute. SEM bars are shown at 0, 15th, 30th and 40th minute of warming.

Results

As shown in Table 1, PV was significantly lower in hypovolaemia than normovolaemia in the thermoneutral condition (P < 0.0001), but there was no significant differences in P_osmol between groups (P > 0.62). The reduction in PV persisted throughout the passive warming portion of the experiment.

Table 1.

Changes in plasma volume and osmolaity in the two groups

					P values
	Groups	BL	TN	W45	Group	Time	G × T
ΔPV (% baseline)	Nor	0.0 ± 0.0	−0.3 ± 1.0	−5.9 ± 1.5
	Hypo	0.0 ± 0.0	−9.0 ± 0.9	−14.5 ± 0.9	< 0.0001	< 0.0001	< 0.0001
Posmol (mosmol (kgH2O)−1)	Nor	292 ± 2	290 ± 1	293 ± 1
	Hypo	291 ± 1	289 ± 1	292 ± 1	0.618	0.007	0.892

Values are means ± SEM for 10 subjects in normovolaemia (Nor) and in hypovolaemia (Hypo). BL, baseline values before the start of measurement in Nor and before administration of a diuretic in Hypo; TN, thermoneutral condition; W45, at the 45th minute of passive warming; ΔPV, changes in plasma volume from BL; P_osmol, plasma osmolality. P values from two-way [1 between (Group) × 1 within (Time)] ANOVA for repeated measures; Group (Nor vs. Hypo), Time (BL, TN and W45), G × T (an interactive effect of [Group × Time]).

Figure 2 shows thermoregulatory responses during the thermoneutral condition and passive warming. T_sk and T_oes increased during warming in both groups but with no significant differences between groups (both P > 0.5) and also with no [Group × Time] interaction (both P > 0.3). Similarly, BF, tSSNA, CVC and SR increased with an increase in T_oes with no significant differences between groups (both P > 0.15). There were no significant [Group × Time] interaction for BF, tSSNA and SR (all P > 0.6). However, we found the significant [Group × Time] interaction for CVC (P < 0.0001).

Figure 2. Thermoregulatory responses during thermoneutral condition and passive warming in both groups.

Mean skin (T_sk; A) and oesophageal temperatures (T_oes; B), burst frequency of skin sympathetic nerve activity (BF; C), total skin sympathetic nerve activity (tSSNA; D), cutaneous vascular conductance CVC; E) and sweat rate (SR; F) are shown as means ± SEM for 10 subjects in normovolaemia (Nor) and hypovolaemia (Hypo). Although statistical analyses were performed on the values every minute, only values in the –5th minute (i.e. 5 min before warming began) under thermoneutral conditions (TN), and the 15th (W15), 30th (W30) and 45th minute of warming (W45) are presented to avoid confusion. P values by two-way [1 between (Group) × 1 within (Time)] ANOVA for repeated measures are described in the figures; G, Groups (Nor vs. Hypo); T, time (50 min); G×T, an interactive effect of [Group × Time].

There were no significant main effects or [Group × Time] interactions for cardiovascular and respiratory variables shown in Fig. 3.

Figure 3. Cardiovascular and respiratory responses.

Heart rate (HR; A), systolic (SBP; B) and diastolic blood pressures (DBP; C), pulse pressure (PP; D), mean arterial pressure (MAP; E) and respiratory rate (RR; F) are shown as the means ± SEM for 10 subjects in normovolaemia (Nor) and hypovolaemia (Hypo). Although statistical analyses were performed on the values every minute, only values in the −5th minute under thermoneutral conditions (TN), and the 15th (W15), 30th (W30) and 45th minute of warming (W45) are presented to avoid confusion. P values by two-way ANOVA for repeated measures are described in the figures; G, groups (Nor vs. Hypo); T, time (50 min); G×T, an interactive effect of [Group × Time].

The maximal auto-correlation coefficient for T_R during the thermoneutral and warming phases in each subject ranged from 0.292 to 0.982 and 0.398 to 0.970 in normovolaemia and hypovolaemia, respectively, with significances (all P < 0.008). Also, T_S ranged from 0.233 to 0.960 in normovolaemia and 0.234 to 0.947 in hypovolaemia with significances (all P < 0.046). Further, the maximal cross-correlation coefficient for T_RS every minute in each subject ranged from 0.219 to 0.705 in normovolaemia and 0.211 to 0.696 in hypovolaemia with significances (all P < 0.05). T_R, T_S and T_RS in both groups significantly decreased with a rise in T_oes (all P < 0.0001) but with no significant differences between groups (all P > 0.22) and also no interactive effect of [Group × Time] (all P > 0.27) (Fig. 4). T_S (x) was highly correlated with T_R (y) in normovolaemia (r = 0.966, P < 0.0001, y = 0.940x+ 0.055) and hypovolaemia (r = 0.958, P < 0.0001, y = 0.948x+ 0.031).

Figure 4. Number of spike, T_R, T_S and T_RS during thermoneutral condition and passive warming.

Number of spikes (A), an R-R interval of ECG (T_R; B), a spike incidence interval of skin sympathetic nerve activity (SSNA) synchronized with the cardiac cycle (T_S; C) and a latency of SSNA spike incidence synchronized with cardiac cycle after a given R-wave of ECG (T_RS; D) are shown as means ± SEM for 10 subjects in normovolaemia (Nor) and hypovolaemia (Hypo). Although statistical analyses were performed on the values every minute, only values in the −5th minute under thermoneutral conditions (TN), and the 15th (W15), 30th (W30) and 45th minute of warming (W45) are presented to avoid confusion. P values by two-way ANOVA for repeated measures are described in the figures; G, groups (Nor vs. Hypo); T, time (50 min); G×T, an interactive effect of [Group × Time].

As shown in Fig. 5, although LA_min and UA_min increased with an increase in T_oes in both groups (P < 0.0001), we found a significant interactive effect of [Group × Time] on UA_min (P < 0.0001), but no significant interactive effect on LA_min (P = 0.999). Similarly, although UA_beat increased during warming in both groups (P = 0.0006), we found a significantly interactive effect of [Group × Time] on UA_beat (P = 0.026) but with no significant difference in LA_beat (not shown) between groups (P = 0.327), and also no significant interactive effect on LA_beat (P = 0.993).

Figure 5.

Lower (LA_min; A), upper (UA_min; B) areas per minute, and upper area per one heart beat (UA_beat; C) during thermoneutral (−5–0 min) and warming (0–45 min) periods in normovolaemia (open circles) and hypovolaemia (filled circles). UA_min and LA_min indicate the components of SSNA spike synchronized and not synchronized with the cardiac cycle, respectively. Values are means and SEM bars for 10 subjects. P values by two-way ANOVA for repeated measures are described in the figures; G, groups (Nor vs. Hypo); T, time (50 min); G×T, an interactive effect of [Group × Time].

As shown in Fig. 6, we found that ΔUA_min (x) was significantly correlated with ΔCVC (y) in normovolaemia (r = 0.961, P < 0.0001, y = 2.200x+ 4.048) and in hypovolaemia (r = 0.795, P < 0.0001, y = 1.519x+ 1.377). The correlation coefficient (P < 0.0001), slope (P = 0.00175) and intercept (P < 0.0001) was smaller in hypovolaemia than normovolaemia. Similarly, we found that ΔLA_min was significantly correlated with ΔSR in normovolaemia (r = 0.981, P < 0.0001, y = 0.006x+ 0.038) and in hypovolaemia (r = 0.993, P < 0.0001, y = 0.005x+ 0.021). The correlation coefficient (P < 0.0001), slope (P = 0.0036) and intercept (P < 0.0001) was again smaller in hypovolaemia than normovolaemia.

As shown in Fig. 7, [Ad]_p, [NA]_p, [AVP]_p and [AGII]_p in both groups increased at the end of warming (all P < 0.033), but with no significant differences between groups (P > 0.1). Also, there were no significant interactive effects of [Group × Time] (P > 0.5) except for [AVP]_p (P = 0.007).

Figure 7. Circulatory vasoactive substances throughout experiment in the both groups.

Plasma adrenaline ([Ad]_p; A), noradrenaline ([NA]_p; B), arginine vasopressin ([AVP]_p; C) and angiotensin II concentrations ([AGII]_p; D) are shown as means ± SEM for 6 subjects in normovolaemia (Nor) and 8 subjects hypovolaemia (Hypo). BL, baseline values before the start of measurement in Nor and before administration of a diuretic in Hypo; TN, thermoneutral condition; W45, at the 45th minute of passive warming. P values by two-way ANOVA for repeated measures are described in the figures; G, groups (Nor vs. Hypo); T, time (50 min); G×T, an interactive effect of [Group × Time].

Discussion

In the present study, we identified a component of SSNA that was synchronized with the cardiac cycle that increased with cutaneous vasodilatation during hyperthermia. More importantly, the increase in this component of SSNA during hyperthermia was suppressed in hypovolaemia and was associated with a reduction in cutaneous vasodilatation. We found no differences in circulating hormones that could potentially suppress cutaneous vasodilatation during hypovolaemia.

We found that the response of T_S before and during warming was almost identical to T_R in normovolaemia and hypovolaemia (Fig. 4). These data indicate that SSNA signals were synchronized with the cardiac cycle whereas earlier reports using rectified and filtered SSNA signals noted that SSNA bursts occurred randomly without synchronizing with the cardiac cycle (Hagbarth et al. 1972; Vallbo et al. 1979). However, Bini et al. (1981) suggested that SSNA recorded from the posterior cutaneous antebrachial nerve included a component synchronized with the cardiac cycle in passively warmed subjects although the burst incidence and the latency after R-wave time was not stable compared with those of MSNA. Macefield & Wallin (1996) suggested that the signals recorded from 3 of 8 single sudomotor fibres had some latency after a given R-wave time although the variance of latency was 2–fold greater than that in MSNA. However, they did not indicate if the synchronizing component was significantly associated with the thermoregulatory responses during warming. In the present study, we found that ∼20% of SSNA spikes synchronized with the cardiac cycle and that they were closely associated with cutaneous vasodilatation in hyperthermia.

T_RS in the present analysis was 1.3 s under thermoneutral conditions, similar to that in MSNA (Sundlof & Wallin, 1978; Fagius & Wallin, 1980), and the suggested time required for electrical signal conduction through the baroreflex loop in normothermia. However, T_RS decreased to ∼1.1 s by the end of warming. The decrease in T_RS might be caused by the direct effect of increased body temperature on the signal conductance speed (Halar et al. 1981). The latency after a given R-wave and the effect of increased body temperature are unlikely to be different for SSNA and MSNA. However, the electrical signal conductance of SSNA is known to be ∼40% faster than that of MSNA (Fagius & Wallin, 1980). Therefore, considering that MSNA signals were triggered by the diastolic pressure wave immediately after the second R-wave of ECG before the burst (Sundlof & Wallin, 1978; Fagius & Wallin, 1980), we speculate that the synchronizing component for SSNA should be triggered by another pressure wave after the diastolic pressure wave.

One of the candidate pressure waves would be the systolic pressure wave ∼0.2 s after the diastolic pressure wave. However, Wilson et al. (2001) examined the responses of rectified and filtered SSNA to pharmacologically induced alteration of arterial pressure in passively heated subjects and suggested that the SSNA was not affected by altered arterial pressure. Crandall et al. (1996) examined the effects of unloading the carotid baroreceptor on CVC by applying 45 mmHg pulsatile neck pressure over the carotid sinus area 0.05 s after a given R-wave for a duration of 0.5 s for 3 min and suggested that CVC was not reduced despite elevated MAP and HR. Thus, since neither pharmacological nor pulsatile stimulation of arterial baroreceptors altered SSNA and CVC, the synchronizing component identified in the present study may not be triggered by a systolic pressure wave.

Another candidate would be the atrial pressure wave caused by atrial filling. There are two types of baroreceptors in the atria: one bursting at the timing of atrial constriction and the other bursting at the timing of atrial filling (Honig, 1981). As shown in Fig. 5, we found that UA_min, a product of HR and UA_beat, increased with a rise in T_oes for both groups; however, in hypovolaemia, the increases in UA_min and UA_beat were significantly attenuated compared to normovolaemia. Since the decrease in T_R, equal to 60 s/HR, during warming was not significantly different between the two groups (Fig. 4), the smaller increase in UA_min in hypovolaemia than normovolaemia was mainly caused by the smaller increase in UA_beat. Since atrial filling occurs at the time of a T-wave of ECG, ∼0.3 s after an R-wave, and since the electrical signal conductance of SSNA was ∼40% faster than that of MSNA, we speculate that the synchronizing component would be evoked by the stretching of atrial baroreceptors, which would be reduced by lowered atrial filling pressure in hypovolaemia.

There have been several studies supporting the idea that cardiopulmonary baroreflex contributes to the control of cutaneous vasodilatation more than arterial baroreflexes (Mack et al. 1988; Crandall et al. 1996; Nagashima et al. 1998). Ahmad et al. (1977) examined the effects of reduced right atrial pressure by applying graded lower body negative pressure to resting and hyperthermic subjects in the supine position and suggested that above −20 mmHg, forearm SkBF was reduced with reduced right atrial pressure while MAP and PP remained unchanged, and below this pressure, the decrease in forearm SkBF was blunted while PP started to decrease with a gradual increase in HR. Nagashima et al. (1998) showed in exercising subjects in a warm environment that cutaneous vasodilatation was enhanced by continuous negative-pressure breathing despite an ∼10 mmHg reduction of MAP, suggesting that the loading of cardiopulmonary baroreceptors increased SkBF more than the unloading of arterial baroreceptors reduced SkBF. These results support our idea that the synchronizing component was evoked by the stretching of atrial baroreceptors, rather than that of arterial baroreceptors. Furthermore, these data support the role of atrial baroreceptors in hypovolaemia-induced suppression of SkBF during hyperthermia.

The increase in UA_beat during warming in normovolaemia might be caused by an increase in the signals generated by the thermoregulatory centre in the hypothalamus as body temperature increased. Regarding the mechanisms for the synchronization of UA_beat with the cardiac cycle, the afferent signals from atrial baroreceptors may control a UA_beat outflow in the cardiovascular centre in the medulla like a gate: passing the efferent signals when baroreceptors are loaded while blocking them when they are unloaded, as suggested in the control of MSNA outflow so that it synchronizes with arterial pressure waves (Kienbaum et al. 2001). In hypovolaemia, since the stretching of atrial baroreceptors decreases due to a reduction in venous return to the heart, a UA_beat outflow may be blocked more than in normovolaemia, resulting in a smaller increase in UA_beat during warming even though the original UA_beat signals generated by the thermoregulatory centre increase as body temperature increases.

Although ΔUA_min was significantly correlated with ΔCVC, the slope and the intercept of regression equations were significantly smaller in hypovolaemia than normovolaemia (Fig. 6). The reasons for this remain unknown; however, ΔUA_min might include SSNA components other than just active vasodilator and active vasoconstrictor signals (Vissing, 1997; Kamijo et al. 2005a). However, the variations of ΔUA_min and ΔCVC in hypovolaemia were only ∼50% of those in normovolaemia, and may have affected the regression analyses. On the other hand, the regression equation between ΔLA_min and ΔSR in normovolaemia was almost identical to that in hypovolaemia; even the slope and the intercept of regression equations were both significantly smaller in hypovolaemia than normovolaemia, suggesting that the sweating response was largely explained by LA_min in both groups.

If the microneurography recording included MSNA spikes then UA_min and UA_beat could demonstrate synchronization with the cardiac cycle; however, this is not likely since the SSNA signal was confirmed by the three criteria stated in the Methods. In addition, we found that UA_min and UA_beat increased with a rise in T_oes; moreover, the increases were attenuated in hypovolaemia compared to those in normovolaemia. Thus, the responses of UA_min and UA_beat to hyperthermia in hypovolaemia occurred in the opposite direction of the predicted responses by MSNA to hypovolaemia (Rowell & Seals, 1990; Kamiya et al. 2003; Kimmerly & Shoemaker, 2003; Convertino et al. 2004). Therefore, we are certain that UA_min and UA_beat were not contaminated with MSNA signals.

We failed to find any significant differences in BF and tSSNA calculated from rectified and filtered SSNA measurements as shown in the Supplemental material, which has been used by other previous studies (Cui et al. 2004; Kamijo et al. 2005a). However, BF in MSNA is reported to be highly reproducible from day to day (Kimmerly et al. 2004). The inter-individual differences in SSNA burst amplitude might be caused by the population of activated sympathetic fibres as well as by the location of the tip of the active electrode at the sympathetic nervous fascicles (Macefield et al. 2002; Young et al. 2009). Therefore, we calculated tSSNA from the BF and burst area, which was expressed as per cent of the maximal value during thermoneutral conditions in each subject. Furthermore, we focused our attention on changes in BF and/or tSSNA in response to an increase in T_oes, but not on the SSNA level between the groups (Young et al. 2009). One possible reason for failing to find any significant differences in the variables between groups despite such careful SSNA measurements was that the signals of active vasodilator fibres were too sparse and small to be detected compared with those of sudomotor fibres by using the previous method. Indeed, in the present study, we successfully identified the signal and found that it was only ∼20% of total SSNA signal.

Sugenoya et al. (1995) measured SkBF and SR continuously on the dorsal foot by laser-Doppler flowmetery and hygrometry, respectively, in passively heated humans and found that a transient increase of SkBF was accompanied by a high probability of sweat expulsion, suggesting that active vasodilator activity was closely related to sudomotor activity. However, they measured these two parameters during a small increase (∼0.1°C) of tympanic temperature and tracked only small transient increases in SkBF and a sweat pulsation while controlling SkBF and SR very close to the pre-warming baselines. In the present study, SkBF increased by 3–fold and SR by 7–fold from the pre-warning levels with an increase of T_oes of ∼0.5°C. These results suggest that a suppression of SkBF response to hyperthermia occurs in the situation where arterial pressure regulation would be threatened, such as when a large amount of blood flow is demanded by the skin vasculature due to hyperthermia, and we thought that the suppression would be enhanced by hypovolaemia. In the present study, we confirmed that a reduction of SkBF in hypovolaemia was not accompanied by a reduction in SR and have successfully identified the possible efferent signal of SSNA for this regulation.

In conclusion, we found a component of the SSNA spike synchronized with the cardiac cycle that increased with a rise in T_oes during passive warming, and was suppressed in hypovolaemia compared with normovolaemia. This component might be an active vasodilator signal for skin vessels and the efferent path for active cutaneous vasodilatation modulated by baroreflex.

Glossary

Abbreviations

[Ad]_p

plasma adrenaline concentration

[AGII]_p

plasma angiotensin II concentration

[AVP]_p

plasma arginine vasopressin concentration

BF

bursts frequency

CVC

cutaneous vascular conductance

DBP

diastolic blood pressure

HR

heart rate

LA

lower area below the valley of the averaged spike incidence histogram for 5 s as a component non-synchronized with the cardiac cycle

LA_beat

lower area per a heartbeat

LA_min

lower area per minute

MAP

mean arterial pressure

MSNA

muscle sympathetic nerve activity

[NA]_p

plasma noradrenaline concentration

P_osmol

plasma osmolality

PP

pulse pressure

PV

plasma volume

RH

relative humidity

RR

respiratory rate

SBP

systolic blood pressure

SkBF

skin blood flow

SR

sweat rate

SSNA

skin sympathetic nerve activity

T_a

ambient temperature

T_oes

oesophageal temperature

tSSNA

total skin sympathetic nerve activity

T_R

an interval for R-wave incidence

T_RS

a firing timing of SSNA spike after R-wave

T_S

an interval for SSNA spike incidence

T_sk

mean skin temperature

UA

upper area above the valley of the averaged spike incidence histogram for 5 s as a component synchronized with the cardiac cycle

UA_beat

upper area per a heartbeat

UA_min

upper area per minute

Author contributions

Y.-i.K. and H.N. designed the study; Y.-i.K., Y.O., S.I., K.O. and M.G. performed the experiments; Y.-.K., H.N., Y.O. and S.I. analysed the data; Y.-i.K. and H.N. wrote the manuscript. All authors approved the final version.

Supplementary material

Supplemental Material

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer-reviewed and may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors