Differential effects of metaboreceptor and chemoreceptor activation on sympathetic and cardiac baroreflex control following exercise in hypoxia in human

Authors


Corresponding author M. Gujic, Department of Cardiology, Erasme University Hospital, 808 Lennik road, B-1070 Brussels, Belgium. Email: marko.gujic@ulb.ac.be

Abstract

Muscle metaboreceptors and peripheral chemoreceptors exert differential effects on the cardiorespiratory and autonomic responses following hypoxic exercise. Whether these effects are accompanied by specific changes in sympathetic and cardiac baroreflex control is not known. Sympathetic and cardiac baroreflex functions were assessed by intravenous nitroprusside and phenylephrine boluses in 15 young male subjects. Recordings were performed in random order, under locally circulatory arrested conditions, during: (1) rest and normoxia (no metaboreflex and no chemoreflex activation); (2) normoxic post-handgrip exercise at 30% of maximum voluntary contraction (metaboreflex activation without chemoreflex activation); (3) hypoxia without handgrip (10% O2 in N2, chemoreflex activation without metaboreflex activation); and (4) post-handgrip exercise in hypoxia (chemoreflex and metaboreflex activation). When compared with normoxic rest (−42 ± 7% muscle sympathetic nerve activity (MSNA) mmHg−1), sympathetic baroreflex sensitivity did not change during normoxic post-exercise ischaemia (PEI; −53 ± 9% MSNA mmHg−1, P= 0.5) and increased during resting hypoxia (−68 ± 5% MSNA mmHg−1, P < 0.01). Sympathetic baroreflex sensitivity decreased during PEI in hypoxia (−35 ± 6% MSNA mmHg−1, P < 0.001 versus hypoxia without exercise; P= 0.16 versus normoxic PEI). Conversely, when compared with normoxic rest (11.1 ± 1.7 ms mmHg−1), cardiac baroreflex sensitivity did not change during normoxic PEI (8.3 ± 1.3 ms mmHg−1, P= 0.09), but decreased during resting hypoxia (7.3 ± 0.8 ms mmHg−1, P < 0.05). Cardiac baroreflex sensitivity was lowest during PEI in hypoxia (4.3 ± 1 ms mmHg−1, P < 0.01 versus hypoxia without exercise; P < 0.001 versus normoxic exercise). The metaboreceptors and chemoreceptors exert differential effects on sympathetic and cardiac baroreflex function. Metaboreceptor activation is the major determinant of sympathetic baroreflex sensitivity, when these receptors are stimulated in the presence of hypoxia.

The peripheral chemoreceptors are located in the carotid bodies and aortic arch (Gonzalez et al. 1994) and respond primarily to hypoxia. Their activation leads to an increase in ventilation, muscle sympathetic nerve activity (MSNA) and blood pressure (Kara et al. 2003).

During exercise, accumulation of metabolites within the active skeletal muscle, stimulates group III and IV afferent neurons which evoke a reflex increase in MSNA, known as the muscle metaboreflex (Houssiere et al. 2005; Smith et al. 2006).

Both chemoreceptors and metaboreceptors are activated when subjects perform a handgrip exercise during hypoxia (Houssiere et al. 2005, 2006a,b). It was previously reported that chemoreceptors and metaboreceptors exert differential effects on the cardiorespiratory and autonomic responses during this hypoxic exercise (Houssiere et al. 2005). The muscle metaboreflex is the major determinant of MSNA and blood pressure increases during exercise in hypoxia (Houssiere et al. 2005). However, the muscle metaboreflex does not play an important role in the heart rate increase during handgrip in hypoxia. This is shown by the sudden reduction in heart rate when peripheral chemoreflex activation is discontinued, while metaboreflex activation is maintained during a local circulatory arrest after isometric exercise (Houssiere et al. 2005). This manoeuver allows persisting metaboreflex activation after exercise cessation. Whether this differential effect of exercise in hypoxia is accompanied by specific changes in sympathetic and cardiac baroreflex control is not known.

Sympathetic and cardiac baroreflex function have been assessed during hypoxia in human (Bristow et al. 1971; Eckberg et al. 1982; Sagawa et al. 1997; Halliwill & Minson, 2002; Halliwill et al. 2003) and animal (Pelletier & Shepherd, 1975; Iriki et al. 1977; Pisarri & Kendrick, 1984; Malpas et al. 1996) studies. The human studies reported no change in baroreflex sensitivity of MSNA, with a reset of the operating point to higher pressure, during acute exposure to mild hypoxia (Halliwill & Minson, 2002; Halliwill et al. 2003), while animal studies have shown an increase in baroreflex sensitivity of sympathetic outflow to the kidney (Iriki et al. 1977; Malpas et al. 1996) and skeletal muscle vascular beds during more severe hypoxia (Pelletier & Shepherd, 1975). Moreover, a decrease in cardiac baroreflex sensitivity was reported in some human (Sagawa et al. 1997) and animal (Pisarri & Kendrick, 1984) studies, but not in all (Bristow et al. 1971; Korner et al. 1973; Eckberg et al. 1982). Differences in species and different levels of hypoxia (Bristow et al. 1971; Korner et al. 1973; Pelletier & Shepherd, 1975; Iriki et al. 1977; Eckberg et al. 1982; Pisarri & Kendrick, 1984; Malpas et al. 1996; Sagawa et al. 1997; Halliwill & Minson, 2002; Halliwill et al. 2003) could explain these diverging results.

Several studies have also assessed baroreflex regulation during post-isometric exercise ischaemia (Iellamo et al. 1999; Kamiya et al. 2001; Cui et al. 2001; Ichinose et al. 2002, 2004). These studies (Cui et al. 2001; Kamiya et al. 2001; Ichinose et al. 2002, 2004) have reported an increase in baroreflex sensitivity of MSNA. Most reported, however, no changes in cardiac baroreflex sensitivity (Cui et al. 2001; Ichinose et al. 2002; Iellamo et al. 1999), with a reset of the operating point to higher pressure (Iellamo et al. 1999; Cui et al. 2001; Ichinose et al. 2004).

Effects of combined metaboreflex and chemoreflex activation on sympathetic and cardiac baroreflex function are unknown. We did not maintain isocapnia in our study because we wanted to investigate baroreflex regulation in experimental conditions similar to those that occur during exercise in altitude or in hypoxaemic patients. In addition, two previous studies, on the effects of hypoxia on baroreflex control, performed either during isocapnic hypoxia or hypocapnic hypoxia, reported the same changes in baroreflex function (Halliwill & Minson, 2002; Halliwill et al. 2003). Because of previous evidence that muscle metaboreceptors are the main determinants of changes in MSNA during exercise in hypoxia, but play little influence on R–R interval, we decided to test the hypothesis that metaboreceptors are the main determinants of sympathetic baroreflex control, but do not contribute importantly to cardiac baroreflex control during exercise in hypoxia.

Methods

Subjects

We recruited 15 young healthy male subjects taking no drugs and aged between 20 and 36 years (mean 25 years). The subjects were instructed not to drink coffee or tea 48 h before the study. They were asked to empty their bladder before the beginning of the recordings. Written informed consent was received from all our subjects before their participation in the study. The study protocol conformed to the standards set by the Declaration of Helsinki and was approved by the Ethical committee of our institution.

Measurements

The subjects were studied in the supine position. A peripheral venous access was placed in the non-dominant arm to permit drug administration. We obtained continuous recordings of electrocardiogram (Siemens), systolic (SBP) and diastolic (DBP) blood pressure (Finometer), ventilation (Pneumotachometer), end-tidal CO2 (inline image; Capnometer, Datex Normocap) and oxygen saturation (inline image; Nellcor N-100 C Pulse Oxymeter).

Breathing was performed through a mouth piece with the use of a nose clip to ensure exclusive mouth breathing (Velez-Roa et al. 2003; Houssiere et al. 2005, 2006a,b).

MSNA was recorded continuously by obtaining multiunit recordings of postganglionic sympathetic activity, measured from a nerve fascicle in the peroneal nerve posterior to the fibular head in all of our subjects (Velez-Roa et al. 2003; Houssiere et al. 2005, 2006a,b). Electrical activity in the nerve fascicle was measured using tungsten microelectrodes (shaft diameter 200 μm, tapering to a non-insulated tip of 1–5 μm). A subcutaneous reference electrode was inserted 2–3 cm away from the recording electrode, which was inserted into the nerve fascicle. The neural signals were amplified, filtered, rectified and integrated to obtain a mean voltage display of sympathetic nerve activity.

All variables were recorded on a Fujitsu-Siemens computer with Chart 5 data acquisition software (AD Instruments).

Interventions

We determined the maximal voluntary contraction of the dominant arm for every subject, in triplicate, with a handgrip dynamometer, at the beginning of the recording session.

Interventions: Normoxic rest, normoxic exercise, resting hypoxia and hypoxic exercise We recorded the above-mentioned variables in four different conditions, for 3 min, in a randomized order: (1) normoxia without exercise (normoxic rest); (2) isometric handgrip exercise of the dominant arm at 30% of maximum voluntary contraction in normoxia (normoxic exercise); (3) hypocapnic hypoxia (10% O2 in N2) in the absence of exercise (resting hypoxia); and (4) isometric handgrip exercise at 30% of maximum voluntary contraction during hypocapnic hypoxia (hypoxic exercise). Each intervention was preceded by a 3 min baseline period with stable ventilation, while subjects were breathing room air. Subjects were requested to minimize any muscle contraction in the resting muscle during handgrip. All interventions were followed by a forearm local circulatory arrest for 3.5 min (Fig. 1).

Figure 1.

Study design
After 3 min of baseline measurement, the subjects performed one of the four randomized interventions (normoxic exercise, resting hypoxia and hypoxic exercise) for 3 min. Each intervention was followed by 3.5 min of local circulatory arrest of the arm performing the handgrip. After 1 min of local circulatory arrest, we administrated a bolus injection of 150 μg sodium nitroprusside (nitro) followed 1 min later by a bolus injection of 150 μg phenylephrin HCl (phen). A resting period of 15 min was observed between each intervention.

Local circulatory arrest Local circulatory arrest was performed by inflating a standard blood pressure cuff at 240 mmHg on the exercising arm, 5 s before the end of each intervention (Fig. 1). The subjects were instructed to relax their grip after the cuff was inflated. This procedure traps metabolites released by the muscle contraction and maintains metabolically sensitive muscle afferents activation independently of muscle contraction (mechanoreceptor reflex) and volitional effects (central command) (Gandevia & Hobbs, 1990). Subjects continued to breathe the hypoxic gas after the local circulatory arrests during resting hypoxia and hypoxic exercise, but breathed room air during normoxic rest and normoxic exercise. Subjects were not allowed to hold their breath during the four interventions. A resting period of 15 min, with subjects breathing room air, was observed between each intervention.

Assessment of baroreflex control of sympathetic outflow and R–R intervals A bolus of 150 μg sodium nitroprusside was given intravenously after the first minute of local circulatory arrest which followed each condition (i.e. normoxic rest, normoxic exercise, resting hypoxia, hypoxic exercise, Fig. 1). This bolus was followed 1 min later by a bolus of 150 μg phenylephrine HCl. The range of pressures produced by the sequential injections of nitroprusside and phenylephrine was similar during the post-handgrip normoxic rest (26 ± 3 mmHg), normoxic exercise (26 ± 3 mmHg), resting hypoxia (30 ± 2 mmHg) and hypoxic exercise (30 ± 3 mmHg). We determined baroreflex control by measuring MSNA and R–R intervals during the arterial pressure changes induced by these vasoactives drugs. Briefly, each MSNA recording was normalized by assigning the largest sympathetic burst under resting condition at an amplitude of 1. The MSNA recorded during the interventions and the local circulatory arrest was then expressed as a percentage of the baseline value. For each heart beat, MSNA tracings were isolated during a 2 s window, centred 1.4 s after each R wave. The corresponding DBP was recorded simultaneously. MSNA was then averaged within intervals of a 3 mmHg range. This technique has been validated in previous studies (Rudas et al. 1999; Halliwill & Minson, 2002; Halliwill et al. 2003). Baroreflex control of MSNA was estimated from the linear relationship between MSNA and diastolic pressure (Rudas et al. 1999).

Baroreflex control of R–R interval was determined from the relationship between the R–R interval and systolic pressure (Rudas et al. 1999). Changes in R–R interval were associated with changes in SBP induced by nitroprussiate and phenylephrine. Similarly to MSNA determination, intervals of 3 mmHg pressure were used to associate SBP and R–R interval. The slope of the linear regression was used to determine cardiac baroreflex sensitivity. Averaged arterial pressure, R–R interval and sympathetic nerve activity during the first minute following the local circulatory arrest provided the operating point of the baroreceptors.

Statistical analysis

Comparisons between levels at baseline, 3rd min of intervention and 1st min after local circulatory arrest during each intervention were tested using a one-way ANOVA for repeated measures, followed by a pair-wise multiple comparison procedure (Tukey's test).

The effects of combined exercise and hypoxia on blood pressure, heart rate and baroreflex indexes were tested using a two-way ANOVA for repeated measures with exercise and hypoxia as factors, followed by a pair-wise multiple comparison procedure (Tukey's test). As the distribution of the variables did not match a normal distribution, ANOVA on ranks were performed. All values are reported as means ±s.e.m.P values < 0.05 were considered statistically significant.

Results

Baseline values

Comparison of baseline parameters before each intervention revealed that there was no difference in terms of ventilation, oxygen saturation, end-tidal CO2, heart rate, systolic blood pressure, diastolic blood pressure and MSNA between the baselines of each interventions (0.3 < P < 0.9).

The average baseline values were: ventilation, 7.3 ± 0.3 l min−1; oxygen saturation, 97 ± 1%; end tidal CO2, 40 ± 1 mmHg; SBP, 120 ± 3 mmHg; DBP, 72 ± 3 mmHg; heart rate, 64 ± 3 beats min−1.

Effects of normoxic exercise, resting hypoxia and hypoxic exercise on ventilatory, cardiovascular and sympathetic parameters (Fig. 2)

Figure 2.

Changes in ventilatory and haemodynamic parameters between baseline recording, the third minute of intervention and the first minute of local circulatory arrest (LCA), during the four interventions
Values are shown as means +s.e.m.+ Difference from baseline; * difference between intervention and local circulatory arrest; +P < 0.05, ++P < 0.01, +++P < 0.001; *P < 0.05, **P < 0.01, ***P < 0.001).

When compared with baseline, ventilation increased during normoxic exercise, resting hypoxia and hypoxic exercise (P < 0.001 for all). As expected, a decrease in inline image occurred during resting hypoxia (from 97 ± 1 to 80 ± 1%; P < 0.001), when compared with baseline. This occurred also during hypoxic exercise (from 97 ± 1 to 82 ± 1%, P < 0.001). The hyperventilation induced by normoxic exercise, resting hypoxia and hypoxic exercise decreased end-tidal CO2 (all P < 0.01). Exercise induced an increase in systolic blood pressure during normoxic exercise and hypoxic exercise (both P < 0.001). This was also observed for diastolic blood pressure, which increased from 72 ± 3 to 83 ± 4 mmHg during normoxic exercise, and from 74 ± 3 to 79 ± 5 mmHg during hypoxic exercise (both P < 0.05). MSNA increased during normoxic exercise and during hypoxic exercise (both P < 0.001). Normoxic exercise, resting hypoxia and hypoxic exercise increased heart rate (all P < 0.001). No other significant changes were observed.

Effect of circulatory arrest on ventilatory, cardiovascular and sympathetic parameters (Fig. 2)

When compared with the levels during intervention, ventilation decreased after circulatory arrest during hypoxic exercise (P < 0.05). There was no change in oxygen saturation or end-tidal CO2 when comparing each intervention to the subsequent local circulatory arrest. SBP and DBP increased slightly, and MSNA decreased slightly, during the local circulatory arrest after normoxic rest (all P < 0.05). Heart rate decreased during the local circulatory arrests which followed normoxic exercise and hypoxic exercise (both P < 0.001).

When compared with the baseline periods, parameters which had increased during normoxic exercise, resting hypoxia and hypoxic exercise, remained elevated during the subsequent local circulatory arrests, except for heart rate, which returned to baseline values after normoxic exercise (P= 0.66).

Effect of combined hypoxia and exercise on sympathetic and cardiac baroreflex estimates

Baroreflex control of MSNA ( Table 1, Fig. 3) When compared with normoxic rest, baroreflex sensitivity of MSNA did not change during normoxic PEI (P= 0.5) but increased during resting hypoxia (P < 0.01). Metaboreflex activation suppressed this increase when associated with hypoxia, and sympathetic baroreflex sensitivity returned to levels recorded during normoxic PEI (P= 0.16). However, there was a reset of the operating point to higher diastolic pressure and MSNA during normoxic PEI and hypoxic PEI (both P < 0.001). End-tidal CO2 was absolutely not related to sympathetic baroreflex sensitivity (Pearson correlation test, P= 0.5) and metaboreflex activation remained an independent determinant of the resetting of the operating points for diastolic blood pressure (standardized beta coefficient = 0.262; P= 0.048) and MSNA (standardized beta coefficient = 0.434; P < 0.001) after controlling for changes in end-tidal CO2 levels, in a multivariate regression model.

Table 1.  Baroreflex sensitivity and operating points of baroreflex control of MSNA, during local circulatory arrest following the four interventions: normoxic rest, normoxic exercise, resting hypoxia, hypoxic exercise
 Normoxic RestNormoxic PEIResting HypoxiaHypoxic PEI
  1. * Comparison with normoxic rest; **P < 0.01, ***P < 0.001; comparison with resting hypoxia; †††P < 0.001.

Baroreflex sensitivity (% MSNA mmHg−1)−42 ± 7−53 ± 9     −68 ± 5**−35 ± 6†††  
Operating point (DBP, mmHg) 70 ± 280 ± 3***69 ± 379 ± 4†††
Operating point (MSNA,% increase from baseline) 93 ± 2168 ± 16***105 ± 5 200 ± 26†††
Figure 3.

Changes in operating points of sympathetic baroreflex during the four conditions: normoxic rest, normoxic PEI, resting hypoxia and hypoxic PEI
Operating points are represented as mean ±s.e.m.

Baroreflex control of R–R interval ( Table 2, Fig. 4) Cardiac baroreflex sensitivity did not change during normoxic PEI (P= 0.09), but decreased during resting hypoxia (P < 0.05). Cardiac baroreflex sensitivity was lowest during PEI in hypoxia (4.3 ± 1 ms mmHg−1, P < 0.01 versus hypoxia without exercise; P < 0.001 versus normoxic PEI). There was a reset of the operating point to a shorter R–R interval during resting hypoxia and hypoxic PEI, as well as to higher systolic blood pressure during normoxic PEI and hypoxic PEI (P < 0.001). Changes in cardiac baroreflex sensitivity in our study became non-significant after controlling for changes in end tidal CO2 in a multivariate analysis (P= 0.08), but the same multivariate analysis revealed that hypoxia and metaboreflex activation were, respectively, responsible, for the resetting of operating points of R–R interval (standardized beta coefficient =−0.550; P < 0.001) and for the resetting of operating points for systolic blood pressure (standardized beta coefficient = 0.395; P= 0.002), independently of changes in end-tidal CO2.

Table 2.  Baroreflex sensitivity and operating points of baroreflex control of R–R interval, during local circulatory arrest following the four interventions: normoxic rest, normoxic exercise, resting hypoxia, hypoxic exercise
 Normoxic RestNormoxic PEIResting HypoxiaHypoxic PEI
  1. * Comparison with normoxic rest; *P < 0.05, **P < 0.01, ***P < 0.001; comparison with normoxic PEI; ††P < 0.001, †††P < 0.001.

Baroreflex sensitivity (ms/mmHg)11.1 ± 1.78.3 ± 1.3 7.3 ± 0.8*   4.3 ± 1††  
Operating point (SBP, mmHg)121 ± 3 141 ± 4***124 ± 4   145 ± 7   
Operating point (R–R interval, ms)909 ± 39 909 ± 52  714 ± 16***714 ± 27†††
Figure 4.

Changes in operating points of cardiac baroreflex during the four conditions: normoxic rest, normoxic PEI, resting hypoxia and hypoxic PEI
Operating points are represented as mean ±s.e.m.

Discussion

The main new finding of our study is that metaboreceptors and chemoreceptors exert differential effects on sympathetic and cardiac baroreflex regulation.

Metaboreceptors are the main determinants of sympathetic baroreflex sensitivity when this muscle reflex is activated in the presence of hypoxia. Moreover, the metaboreceptors are predominantly responsible for a reset of the operating point of the baroreceptor to a higher pressure, while the chemoreceptors reset this operating point to a shorter R–R interval. These results are in accordance with our previous studies (Houssiere et al. 2005, 2006a), in which we observed that the muscle metaboreflex is the major determinant of MSNA and blood pressure increases, while the peripheral chemoreflex is responsible for the rise of heart rate, during handgrip in hypoxia.

Studies on metaboreflex or chemoreflex activation on sympathetic and cardiac baroreflex regulation

We did not observe an increase in sympathetic baroreflex sensitivity during PEI in contrast to previous studies (Cui et al. 2001; Kamiya et al. 2001; Ichinose et al. 2002, 2004). Variations in the experimental settings, as well as in the methods used to assess baroreflex sensitivity, probably explain these differences. In contrast to previous studies (Cui et al. 2001; Kamiya et al. 2001; Ichinose et al. 2002, 2004), we assessed baroreflex control of MSNA and cardiac interval after local circulatory arrest during all the interventions, to prevent any effect of local ischaemia per se on our measures. Moreover, while some have used vasoactive drug infusion techniques to assess baroreflex control (Cui et al. 2001), others have assessed spontaneous fluctuations in blood pressure and MSNA (Kamiya et al. 2001; Ichinose et al. 2004) and others have also used neck chamber techniques (Ichinose et al. 2002). These techniques determine baroreflex control over different ranges in blood pressure. Some investigate sensitivity mainly around the operating point (Kamiya et al. 2001; Ichinose et al. 2004), or over wider ranges in blood pressure (Cui et al. 2001), or in response to either increases or decreases in blood pressure (Ichinose et al. 2002). Moreover, some of these techniques assess only carotid baroreflex control (Ichinose et al. 2002), and all these techniques have their specific limitations (Parati et al. 2000), which may render comparisons between studies difficult.

More important is also the fact that the protocols during which baroreflex control was assessed are also highly variable. In some studies respiration was controlled (Pelletier & Shepherd, 1975; Iriki et al. 1977; Ichinose et al. 2002, 2004), while in other studies respiration was uncontrolled (Malpas et al. 1996; Kamiya et al. 2001; Halliwill & Minson, 2002; Halliwill et al. 2003; Cooper et al. 2005). Moreover, even within a given experimental protocol, expressing baroreflex control of MSNA during handgrip in terms of changes in burst frequency, burst amplitude or total activity, yielded different results (Malpas et al. 1996; Ichinose et al. 2004). This is also evident when effects of hypoxia on baroreflex control are assessed on MSNA (Halliwill & Minson, 2002; Halliwill et al. 2003) or on vascular resistance (Cooper et al. 2005). These studies reported, respectively, that arterial baroreflex sensitivity was unchanged (Halliwill & Minson, 2002; Halliwill et al. 2003) or decreased (Cooper et al. 2005). Further differences between studies concern the levels of oxygen desaturation achieved, as some have used a inline image (fraction of inspired oxygen) of ±12% (Bristow et al. 1971; Eckberg et al. 1982; Halliwill & Minson, 2002; Halliwill et al. 2003; Cooper et al. 2005), and others 10% or less (Pelletier & Shepherd, 1975; Pisarri & Kendrick, 1984; Malpas et al. 1996; Sagawa et al. 1997). Moreover in some studies hypoxia is accompanied by hypocapnia (Iriki et al. 1977; Sagawa et al. 1997; Halliwill & Minson, 2002), but not in others (Pelletier & Shepherd, 1975; Halliwill et al. 2003; Cooper et al. 2005). Also the duration of hypoxia is variable, ranging between 2 (Pelletier & Shepherd, 1975) to 60 min (Sagawa et al. 1997). Furthermore, the range of exercise intensities vary from 30 (Kamiya et al. 2001) to 50% (Ichinose et al. 2002, 2004) of maximal voluntary contraction, even within the studies who assessed only the effect of sustained handgrip exercise. Last, the durations of these exercises, as well as of the post-handgrip periods, vary between 1 (Ichinose et al. 2002, 2004) and 2 min (Cui et al. 2001). A higher percentage of maximal contraction during handgrip (Cui et al. 2001; Ichinose et al. 2002, 2004) increases the level of metaboreflex activation proportionally to metabolite accumulation and pH fall within the muscle. This may enhance the MSNA response and may exert an effect on the assessment of sympathetic baroreflex sensitivity during handgrip exercise. Indeed, higher levels of DBP and MSNA can displace the operating points to a steeper part of the full baroreflex curve, resulting in larger baroreflex sensitivity.

Comparison of the effects of exercise or hypoxia on baroreflex control between these previous studies and our study is therefore difficult. Even when considering only studies assessing baroreflex sensitivity by the modified Oxford technique during hypoxia and during post-exercise ischaemia (Cui et al. 2001; Halliwill & Minson, 2002; Halliwill et al. 2003), differences in the intensity of chemoreflex and metaboreflex activation are due both to the severity of hypoxia used and the percentage of maximal contraction performed during exercise persist, and could probably explain the diverging results in baroreflex sensitivity. The changes in baroreflex control we observed tend mostly to reflect those that were observed in animal studies, where hypoxia decreased baroreflex sensitivity of heart rate (Pisarri & Kendrick, 1984; Sagawa et al. 1997), and enhanced baroreflex sensitivity of sympathetic nerve activity (Pelletier & Shepherd, 1975; Iriki et al. 1977; Malpas et al. 1996). Of some interest is also the fact that the most consistent observation seen in the above mentioned studies (Bristow et al. 1971; Pelletier & Shepherd, 1975; Iriki et al. 1977; Eckberg et al. 1982; Pisarri & Kendrick, 1984; Malpas et al. 1996; Sagawa et al. 1997; Iellamo et al. 1999; Cui et al. 2001; Kamiya et al. 2001; Halliwill & Minson, 2002; Ichinose et al. 2002, 2004; Halliwill et al. 2003; Cooper et al. 2005), namely that cardiac baroreceptor control during handgrip is reset to higher values with an unchanged sensitivity (Iellamo et al. 1999; Cui et al. 2001; Ichinose et al. 2002), was also observed in our study.

Within a group of investigators, using the same methodology, results appear, however, reproducible (Halliwill & Minson, 2002; Halliwill et al. 2003). We believe therefore that the most relevant comparisons in our study reside in the analysis of changes in baroreflex control between interventions. We are not aware of a previous study on baroreflex control during separate and concomitant metaboreflex and peripheral chemoreflex activation.

Sympathetic baroreflex function during post-exercise in hypoxia

The larger baroreflex sensitivity of MSNA during hypoxia alone disappeared during metaboreflex activation in hypoxia, and was replaced by the unchanged baroreflex sensitivity observed during metaboreflex activation alone. This occurred also for the operating point of baroreceptor control of MSNA.

Muscle metaboreceptors regulate sympathetic activation during exercise (Mark et al. 1985). This reflex is activated by lactic acid, phosphate, K+, H+, adenosine, prostaglandins, and bradykinin released from exercising skeletal muscle (Kaufman et al. 1983; Smith et al. 2006). These ischaemic metabolites stimulate groups III and IV chemosensitive afferents (Kaufman et al. 1983; Smith et al. 2006). These afferents project to the dorsal horn of the spinal cord. Moreover, it was reported that neurons of the cervical spinal cord project to the cuneate nucleus, the nucleus of the solitary tract and the lateral reticular nucleus, as well as the caudal and rostral ventrolateral medulla within the medulla oblongata (Potts & Mitchell, 1998; Potts et al. 2002). Indeed, the nucleus tractus solitarii and the rostral ventral medulla may represent a central site of integration for the exercise pressor reflex. These findings support the existence of a spinomedullary pathway transmitting metaboreflex afferents information to the brainstem, where they induce an increase in sympathetic nerve activity to muscle circulation to both non-exercising and exercising limbs (Smith et al. 2006). However, this produces vasoconstriction only in the non-exercising limbs, because local vasomotor regulatory mechanisms prevail over sympathetic vasoconstriction in the exercising limbs (Rotto & Kaufman, 1988; Sinoway et al. 1994). The resulting increased perfusion pressure corrects the deficit in blood flow in the exercising limb, where circulation is hampered by the increased intramuscular pressure (Rotto & Kaufman, 1988; Sinoway et al. 1994). There is also evidence from animal studies (Potts & Mitchell, 1998) that afferent fibres from baroreceptors and skeletal muscle receptors synapse in the nucleus tractus solitarii. One possible mechanism that could explain the resetting of the operating point of the baroreceptors to a higher pressure, is that activated skeletal muscle afferent fibre could inhibit the firing of barosensitive nucleus tractus solitarius neurons (Potts & Mitchell, 1998; Potts et al. 2002). The same pathway could also be implicated in the inhibitory interaction observed between metaboreflex and chemoreflex on baroreflex control of MSNA during hypoxic PEI. Moreover, our observation that the metaboreceptors prevailed over the chemoreceptors in the regulation of sympathetic baroreflex function during post-exercise in hypoxia, could be explained by the fact that hypoxia failed to raise MSNA in our study, in contrast to metaboreceptor activation. This could be due to hyperventilation, which inhibits sympathetic outflow (Somers et al. 1989a,b), and also as a result of central chemoreflex inhibition in response to hypocapnia (Somers et al. 1989a).

Cardiac baroreflex function during post-exercise in hypoxia

Two of our previous studies have suggested that metaboreceptors did not play an important role in mean heart rate during exercise in hypoxia (Houssiere et al. 2005, 2006a). This was especially evident during concomitant chemoreflex inhibition and metaboreflex activation (Houssiere et al. 2006a). Under this latter circumstance, metaboreceptor activation suppressed the sympatholytic and hypotensive effects of peripheral chemoreceptor inhibition, while the bradycardic effects of peripheral chemoreflex inhibition remained unaffected (Houssiere et al. 2006a). Moreover, during hypoxic PEI, exercise per se also did not potentiate the effects of hypoxia on heart rate (Hanada et al. 2003).

Alternatively, the fact that heart rate returned to baseline levels once chemoreflex activation was stopped, and metaboreflex activation maintained, could have been due to a rise in cardiac baroreflex sensitivity (Houssiere et al. 2005). Because of the blood pressure increase in response to metaboreflex activation, this would have favoured a prompt reduction in heart rate upon cessation of chemoreflex activation (Houssiere et al. 2005). This was, however, not observed in our study, as arterial baroreflex control of R–R interval was reduced during hypoxia, and this reduction persisted during post-handgrip exercise in hypoxia.

Effects of hypoxia on heart rate and sympathetic activity are complex. Selective stimulation of the peripheral chemoreceptors is accompanied by a reduction in heart rate if breathing ceases (Uther et al. 1970; Pisarri & Kendrick, 1984; Somers et al. 1989a,b). This is due to an increase in cardiac vagal activity (Pisarri & Kendrick, 1984; Somers et al. 1989a,b). However, with hyperventilation, an opposite response is observed because heart rate increases (Uther et al. 1970; Somers et al. 1989a,b). Carotid chemoreceptors increase sympathetic activity (Kara et al. 2003; Somers et al. 1989a,b), while, in contrast, aortic chemoreceptors raise heart rate (Pisarri & Kendrick, 1984). The overall effect of peripheral chemoreflex activation, in the presence of hyperventilation, is, however, an increase in heart rate, as a result of an increase in cardiac sympathetic outflow (Somers et al. 1989a; Kara et al. 2003). Moreover, hyperventilation limits the rise in sympathetic activity during peripheral chemoreflex activation through pulmonary vagal activation (Somers et al. 1989a). As a result, increases in sympathetic activity become most manifest when breathing is interrupted in the presence of hypoxia.

Projection of the peripheral chemoreceptors and baroreceptors coincide within the medulla, where multiple interactions may occur (Miura & Reis, 1972), namely within the paramedian reticular nuclei and nucleus tractus solitarii (Miura & Reis, 1972). These overlaps may explain why modifications in the baroreflex control of the R–R interval observed during hypoxia persisted also during post-exercise in hypoxia. Moreover, the lack of effects on heart rate of metaboreflex activation (Rowell & O'Leary, 1990; Houssiere et al. 2005, 2006a) may further explain the limited effects of this muscle reflex on baroreflex regulation during simultaneous chemoreflex and metaboreflex activation. This contrasts with mechanoreceptor activation, which increases heart rate (Coote et al. 1971; Rowell & O'Leary, 1990; Gladwell & Coote, 2002). Indeed, joint movements and muscle tension result in group III mechanosensitive afferent fibre projection, which raise heart rate and sympathetic activity (Rowell & O'Leary, 1990).

The differential responses observed in heart rate and sympathetic baroreflex control during normoxic PEI, resting hypoxia and hypoxic PEI could be due to a non-homogenous regulation of peripheral and cardiac sympathetic activity, or to the fact that changes in both sympathetic and vagal activity affected heart rate control, while changes in peripheral vasomotor tone control involved only the sympathetic system. Indeed, there is evidence that a rise in cardiac vagal activity overwhelms the increase in cardiac sympathetic activity during metaboreflex activation, and that this mechanism explains a return to baseline heart rate level during PEI (Iellamo et al. 1999).

Limitations

We did not maintain isocapnia during the interventions. Hyperventilation produced by both exercise and hypoxia decreased end-tidal CO2 in our subjects. Hypocapnia is known to reduce the ventilatory and MSNA responses during peripheral chemoreflex activation, mainly as a result of central chemoreceptor inhibition (Somers et al. 1989a). In our study, multivariate analysis revealed that progressively larger reductions in end-tidal CO2 from normoxia, to post-ischaemic exercise, to hypoxia and to post-ischaemic exercise in hypoxia participated in the progressive reduction in cardiac baroreflex sensitivity, because modifications in cardiac baroreflex sensitivity became non-significant after controlling for changes in end-tidal CO2. However, changes in sympathetic baroreflex sensitivity and in the operating points were independent of end-tidal CO2. This important observation further highlights that there is a differential regulation of cardiac and sympathetic baroreflex sensitivity in normal humans. Last, modifications in baroreflex control we observed during PEI, may not apply to active exercise and other types of exercise such as dynamic exercise. This latter type of exercise produces different haemodynamic responses than isometric exercise (Rowell & O'Leary, 1990), and this is also accompanied by a different pattern of metaboreflex and mecanoreflex activation (Rowell & O'Leary, 1990). Further studies will also be needed to address this issue.

In conclusion, our study reveals that the metaboreceptor and chemoreceptors exert differential effects on sympathetic and cardiac baroreflex regulation during post-exercise muscle reflex activation in hypoxia. Under this condition, metaboreceptor activation is the major determinant of sympathetic baroreflex sensitivity and higher blood pressure operating points, while the chemoreflex is mainly responsible for the shorter R–R interval operating point.

Appendix

Acknowledgements

This study was supported and funded by the Erasme Foundation, Belgium (M.G., O.X., J-F.A.), the Lambertine-Lacroix and Saucez-Van Poucke Foundation (P. v.d.B.), the Belgian National Fund for Research (R.N., P.v.d.B.) the Foundation for Cardiac Surgery, Belgium (R.N., P.v.d.B.) and the Foundation David et Alice Van Buuren (M.G., S.B., J.F.A., P.v.d.B.).

Ancillary