Contribution of renal nerves to renal blood flow variability during ...

Contribution of renal nerves to renal blood flow variability during hemorrhage SIMON C. MALPAS,1,2 ROGER G. EVANS,3 GEOFF A. HEAD,1 AND ELENA V. LUKOSHKOVA1 Medical Research Institute, Prahran 3181; 3Department of Physiology, Monash University, Clayton 3168, Victoria, Australia; and 2Department of Physiology, University of Auckland, Auckland, New Zealand

1Baker

Malpas, Simon C., Roger G. Evans, Geoff A. Head, and Elena V. Lukoshkova. Contribution of renal nerves to renal blood flow variability during hemorrhage. Am. J. Physiol. 274 (Regulatory Integrative Comp. Physiol. 43): R1283–R1294, 1998.—We have examined the role of the renal sympathetic nerves in the renal blood flow (RBF) response to hemorrhage in seven conscious rabbits. Hemorrhage was produced by blood withdrawal at 1.35 ml · min21 · kg21 for 20 min while RBF and renal sympathetic nerve activity (RSNA) were simultaneously measured. Hemorrhage was associated with a gradual increase in RSNA and decrease in RBF from the 4th min. In seven denervated animals, the resting RBF before hemorrhage was significantly greater (48 6 1 vs. 31 6 1 ml/min intact), and the decrease in RBF did not occur until arterial pressure also began to fall (8th min); however, the overall percentage change in RBF by 20 min of blood withdrawal was similar. Spectral analysis was used to identify the nature of the oscillations in each variable. Before hemorrhage, a rhythm at ,0.3 Hz was observed in RSNA, although not in RBF, whose spectrogram was composed mostly of lower-frequency (,0.25 Hz) components. The denervated group of rabbits had similar frequency spectrums for RBF before hemorrhage. RSNA played a role in dampening the effect of oscillations in arterial pressure on RBF as the transfer gain between mean arterial pressure (MAP) and RBF for frequencies .0.25 Hz was significantly less in intact than denervated rabbits (0.83 6 0.12 vs. 1.19 6 0.10 ml · min21 · mmHg21 ). Furthermore, the coherence between MAP and RBF was also significantly higher in denervated rabbits, suggesting tighter coupling between the two variables in the absence of RSNA. Before the onset of significant decreases in arterial pressure (up to 10 min), there was an increase in the strength of oscillations centered around 0.3 Hz in RSNA. These were accompanied by increases in the spectral power of RBF at the same frequency. As arterial pressure fell in both groups of animals, the dominant rhythm to emerge in RBF was centered between 0.15 and 0.20 Hz and was present in intact and denervated rabbits. It is speculated that this is myogenic in origin. We conclude that RSNA can induce oscillations in RBF at 0.3 Hz, plays a significant role in altering the effect of oscillations in arterial pressure on RBF, and mediates a proportion of renal vasoconstriction during hemorrhage in conscious rabbits. conscious rabbit; sympathetic nervous system; spectral analysis; renal denervation

one variable that exhibits oscillations at a number of frequencies, components of which are due to renal nerve activity (22). We previously observed that RBF would respond to changes in the power of higherfrequency components of renal sympathetic nerve activity (RSNA), for example, associated with respiration and with vasoconstriction but not oscillations. However, increases in the power of sympathetic nerve activity variability at the lower frequencies could cause oscillations in RBF (22). Of particular interest in the present study is the possibility that stimuli that increase or decrease RSNA could produce or alter the rhythms present in RBF. Thus the nerve activity could dampen or enhance the effect of oscillations in arterial pressure on the renal vasculature. These possibilities may reflect new mechanisms by which sympathetic nerve activity regulates blood flow, thus controlling arterial pressure. Hemorrhage is one stimulus well established to cause profound changes in renal function (1). Before a significant decrease in arterial pressure, RSNA increases, although this is not thought to alter RBF, the effect being preferential for renin and sodium excretion in dogs (37). Subsequently, as arterial pressure begins to fall and the increase in RSNA is very large, RBF is reduced (45). Although a component of this renal response may be mediated via the nerves, it is clear that a number of circulating substances, such as angiotensin II and norepinephrine, also play a role (41). In this study, we have investigated the role of the renal nerves in the dynamic response of RBF during hemorrhage. We measured the various rhythms present in RBF, arterial pressure, and RSNA during the different phases of hemorrhage and also examined the interactions between these rhythms in conscious rabbits. In particular, we were interested in whether the nerve activity induces oscillations in RBF, which in turn may be responsible for some of the arterial pressure variability previously observed (23). To confirm the specific contribution of the renal nerves in these responses, experiments were also conducted in a separate group of rabbits after bilateral renal denervation. METHODS

IT IS NOW ESTABLISHED that many cardiovascular param-

eters display oscillations at a number of frequencies (2). Although their origin is a matter of current research, it is becoming clear that the interaction of these rhythms may, in some circumstances, provide important information about circulatory control mechanisms, particularly those mediated by the autonomic nervous system (6, 9). Recent research suggests that renal blood flow (RBF) is

Animal preparation. Experiments were performed on rabbits bred at the Baker Medical Research Institute, in accordance with the Australian Code of Practice for the Care and Use of Animals for Scientific Purposes. Seven rabbits (weight 2.5–3.0, mean 2.6 kg) underwent surgery, at least 7 days before the experiment, to implant a recording electrode around the left renal sympathetic nerve and a transit-time flow probe around the left renal artery. Under halothane

0363-6119/98 $5.00 Copyright r 1998 the American Physiological Society

R1283

R1284

RENAL NERVES AND HEMORRHAGE

anesthesia, the left renal nerve was exposed by a retroperitoneal approach. With the use of an operating microscope, an intact nerve was fed through a coiled electrode (11), and a transit-time flow probe (Type 2SB Transonic Systems) was placed around the renal artery. Both devices were held in place and insulated from the surrounding tissue by Wacker Sil-Gel (Wacker-Chemie, Munich, Germany). The other end of the electrode and flow probe were tunneled under the skin for later retrieval on the day of the experiment. In a separate group of seven animals (weight 2.4–2.8, mean 2.7 kg), instead of implantation of a renal electrode, bilateral renal denervation was performed, and a flow probe was implanted around the left renal artery. All animals also had a Silastic catheter (0.86 mm ID) inserted into the superior vena cava via the right jugular vein during this surgery for withdrawal of blood. This was filled with heparin (1,000 IU/ml) with the other end sealed and placed under the skin for retrieval on the day of the experiment. Experimental preparation. On the day of the experiment, catheters were inserted in a central ear artery for measurement of arterial pressure. The ends of the renal electrode, flow probe, and venous catheter were retrieved from under the skin under local anesthesia. Sympathetic nerve activity was amplified, filtered between 50 and 5,000 Hz, full wave rectified, and integrated using a low pass filter with a time constant of 20 ms. This integrated RSNA signal, the RBF waveform, and arterial pressure were continuously recorded throughout the experiment and were sampled at 1,000 Hz using an analog-to-digital data acquisition card (National Instruments). This was the input signal for all subsequent computer analysis. Calibrated signals were displayed on a computer screen and saved to disk using a program written in the LabVIEW graphical programming language (National Instruments). After completion of the preparatory procedures, rabbits were left for 60 min before commencement of the experimental protocol. Experimental protocol. Animals were given 1,000 IU in 2 ml of heparin via the chronically indwelling superior vena cava catheter. After a 20-min period of control recording, blood was withdrawn using a constant withdrawal pump at a rate of 1.35 ml · kg21 · min21 for 20 min. This rate is equivalent to 3% of blood volume per minute (12). Blood samples were also taken for analysis of plasma renin activity during the 20-min prestimulus control period and at the end of the 20-min period of each stimulus. One-milliliter arterial blood samples were collected into chilled tubes containing 100 µl 2,3dimercaptopropanol-EDTA (Sigma Chemical, St Louis, MO). The rate at which angiotensin I was generated in nanograms per milliliter per hour was measured via radioimmumoassay (39), and free angiotensin I was separated using 20% polyethylene glycol. The disequilibrium angiotensin I radioimmumoassay had a detection limit of 2 pg of angiotensin I (interassay coefficient of variation 5 16.7%). Steady-state data analysis. Data were collected for analysis by two techniques, steady-state and spectral analysis. The term steady state refers to the average change in each parameter in the control period and during hemorrhage. With regard to RSNA, the activity was measured in three ways; simply, the average RSNA voltage per 2-s period was calculated from the integrated neurogram, and this is termed total RSNA. Additionally, each burst of RSNA, from the integrated neurogram, had its amplitude and interval until the next burst measured. The discharges of sympathetic activity were detected on-line using a dedicated program written in the LabVIEW graphical programming language (National Instruments). The program uses a series of fast and slow filters to detect changes in the voltage of the signal from increasing to

decreasing levels. Providing these changes were above an operator-defined threshold, they were classified as discharges in RSNA. The threshold level was usually set to 10–15% of the average maximum discharge height, as discussed previously (33). The threshold and other settings for the sympathetic discharge detection were set at the beginning of the experiment and not subsequently altered. The interval between discharges was expressed as the average number of discharges occurring per second. This frequency of RSNA pertains only to the frequency of discharge .2 Hz, i.e., cardiac and above, and is not based on spectral analysis, as are subsequent results, but rather uses simple burst detection and time domain analysis. RSNA voltages vary between animals due to the physical nature of the nerve recording and electrode system; therefore, these parameters (excluding the frequency of discharges) were normalized to control values measured in each rabbit during the prestimulus period. Spectral analysis. The fluctuations in mean arterial pressure (MAP), heart rate (HR), RSNA, and RBF were investigated in the frequency domain using spectral analysis techniques. Although these variables display oscillations at frequencies .2 Hz, such as with each cardiac cycle, this has already been described in some detail (22) and was not the focus of the present study. Analysis of oscillations at frequencies ,2 Hz was performed using the following steps. Initially, all beat-to-beat signals were replayed on screen for visual inspection. Data segments with artifacts due to obstruction of the arterial catheter or animal movement were eliminated. The 20-min hemorrhage period was divided into sections of 5 min for analysis along with the 20-min prehemorrhage control period. Before calculation of power spectral density, each segment was subjected to detrending to remove the underlying mean value and slow changes in the parameters, windowed with a tapered cosine function, and padded with zeroes up to 1,024 points. The auto- and cross-power spectral density was calculated using the fast-Fourier algorithm (LabVIEW; National Instruments). Because this algorithm requires equidistantly sampled data, the beat-to-beat data segments were resampled at 10.24 Hz. Then these data sections were partitioned into segments of 50 s (512 points) length overlapping by 25 s. To investigate to what extent fluctuations of MAP influenced fluctuations in RBF, the transfer function between MAP and RBF was calculated from the averaged spectra. For the average spectrum, the magnitude and phase of the transfer function was calculated as the quotient of the input and cross spectrum. The actual units of gain cannot be used to assess the effect of changes in arterial pressure on RBF in isolation; rather a comparison of the gain values between animals or with a stimulus was deemed appropriate. However, in general, the larger the gain value the greater the influence of the input, MAP, on the output, RBF. The phase of the transfer function indicates the temporal relationship between the signals in the frequency domain, that is, that oscillations in one parameter may induce a similar frequency oscillation in another parameter delayed in time. Thus, by calculating the phase angle, it is possible to determine if one rhythm is preceding or succeeding another rhythm. In addition, the coherence function was computed. The coherence can have a value between 0 and 1 and is a frequency domain estimate of the correlation coefficient indicating the degree that variance in one variable can be explained by the variance in the other. It is important to note that, unlike our previous study (22) in which RBF was measured by doppler flowmetry, we measured absolute blood flow by transit time flowmetry. This meant that we could compare RBF and absolute power


or calculated transfer gain and coherence between groups of animals. The spectra in the range below 2 Hz were divided into three different domains as previously described (22): 1) a respirationrelated frequency (high) band containing oscillations linked to the respiratory cycle. The range of this band (0.75–1.4 Hz) was relatively wide in order to cover the variable nature of breathing frequencies in conscious rabbits (26; respiration rate was not specifically measured); 2) a mid-frequency range (mid: 0.25–0.45 Hz) containing oscillations thought to be associated with sympathetic activity (3, 22, 43); and 3) a low-frequency range (low: 0.04–0.25 Hz). Statistical analysis. With regard to the steady-state data, the influence of RSNA on the levels of each variable during hemorrhage was tested by repeated-measures analysis of variance, the factors being neural status (intact or denervated) and time (1-min averages). We tested the hypotheses that the resting levels of each variable were different in the two groups of animals (effect of neural status) and that the responses to hemorrhage were different in the two groups of animals (neural status 3 time interaction). P values were conservatively adjusted by the Greenhous-Geisser correction to protect against the increased risk of type 1 error due to compound asymmetry (28). The magnitude of the power in the high, mid, and low frequencies before hemorrhage was compared in the intact and denervated groups by a split-plot ANOVA, as described previously (25). The total sums of squares (SS) were partitioned into between groups (state of innervation) with further partitioning of the SS of each group into between animals SS and between frequencies. The significance of the difference between groups was assessed from the variance ratio F 5 between groups mean square/ within groups mean square. The latter was calculated as the sum of the between animals SS from each group (animals 3 groups interaction) and the sum of the animals 3 frequencies. The latter were used to calculate the average treatment regimen SE of the mean (indicating variation between animals). Within each group, the between frequencies were partitioned into orthogonal contrasts (low vs. mid plus high; mid vs. high), each with one degree of freedom. For these contrasts, the within-animal residual mean square was used as the denominator, as it indicates the within-animal variance. RESULTS

Steady-state changes during hemorrhage. Hemorrhage was associated with a profound increase in RSNA (Fig. 1). After 4 min of blood withdrawal, the increase was 12 6 11% (between animal SE) above control levels, at 9 min it was 22 6 7%, and, by the end of hemorrhage, RSNA had reached 201 6 87% of control levels. RSNA was also divided into amplitude and frequency components reflecting more closely the factors influencing the bursting pattern of discharges. During hemorrhage, there was a significant increase in the amplitude of sympathetic discharges that mirrored the changes in the total level of RSNA. There were no significant changes in the rate at which discharges occurred. Arterial pressure did not decrease significantly until the 8th to 11th min of blood withdrawal and reached 60 6 4 mmHg after 20 min (control 81 6 2 mmHg). HR initially increased during the first 4–8 min of hemorrhage and then decreased after 12–14 min. RBF decreased as RSNA was increasing, i.e., beginning 4 min after commencing blood withdrawal, reaching a

R1285

minimum of 10 6 5 ml/min at 20 min of blood withdrawal (control 5 31 6 1 ml/min). Plasma renin activity was significantly increased after 20 min of blood withdrawal (10.4 6 1.4 control vs. 17.9 6 2.5 ng · ml21 · h21, P , 0.01). The denervated group of rabbits had similar responses in HR and blood pressure to the withdrawal of blood as the intact animals (Fig. 1). Plasma renin activity, although significantly less in the denervated group of rabbits in the prehemorrhage control period (5.2 6 1.4 ng · ml21 · h21 ), also was significantly increased after 20 min of hemorrhage (16.0 6 3.4 ng · ml21 · h21, P 5 0.03). The percentage increase in renin activity with hemorrhage was significantly greater in the denervated animals compared with the intact animals. RBF also decreased in these animals; however, the decrease did not commence until 8–9 min from the beginning of hemorrhage and corresponded to the start of the decrease in arterial pressure. The time until a 10% reduction in RBF was significantly longer than in the intact group (P , 0.05, t-test). The decrease in RBF reached 19 6 4 ml/min after 20 min of blood withdrawal. Although resting RBF was greater in denervated rabbits (48 6 1 vs. 31 6 1 ml/min intact) and RBF began to decrease at a later time relative to the intact group, the percentage decrease in RBF by the end of the 20 min of hemorrhage was similar in both groups (32% intact, 37% denervated). The rate of change in RBF between the two groups of rabbits was not significantly different. Interactions between arterial pressure and RBF: Comparison between intact and denervated rabbits. The mean spectral components for RSNA, MAP, RBF, and HR from the group of intact rabbits before commencing hemorrhage are shown in Fig. 2, and an original neurogram of the variables is shown in Fig. 3. A peak at ,0.3 Hz was observed in RSNA, although not in the other variables, with the spectrogram being composed of mostly lower-frequency (,0.25 Hz) components. The denervated group of rabbits had similar frequency spectrums for RBF, MAP, and HR (not shown). Calculating the percentage of the total spectral power that can be accounted for by each of the three defined frequency bands (low 0.04–0.25 Hz, mid 0.25–0.45 Hz, and high 0.75–1.4 Hz) allows comparison between intact and denervated groups (Fig. 4). The low-frequency component comprised the greatest proportion of spectral power in each group (31 6 2% for RBF in intact animals; 30 6 7 denervated). There were also no significant differences between groups in the proportions for MAP, HR, and RBF for each of the three frequency bands. Thus, although resting levels of RSNA contain oscillations at distinct frequencies, these do not exert effects on RBF at a similar frequency, as the spectral components were not significantly different between the intact and denervated group of rabbits. However, as outlined above, the resting RBF was considerably greater in the denervated group of rabbits. It is possible that RSNA may exert its effects evenly across a wide band of frequencies, which the calculation of the percentage of the total power in the

R1286


Fig. 1. Levels of renal blood flow (RBF), renal vasculature conductance, mean arterial pressure (MAP), heart rate (HR), and renal sympathetic nerve activity (RSNA; calculated as 2-s average from the integrated neurogram and as the amplitude and frequency of discharges, see METHODS for description) throughout the 20-min control period and the 20-min period of constant withdrawal of blood at a rate of 1.35 ml · kg21 · min21 (the beginning of hemorrhage is indicated by the vertical dotted line). Lines represent 1-min averages of the pulsatile signals. Solid line, renal nerve intact group; broken line, group of denervated rabbits. It should be noted that the RSNA frequency shown was not calculated using spectral analysis means as was the data in subsequent figures but was derived from simple time domain analysis using detection of the individual discharges in RSNA and calculating the time interval between them. Thus this frequency reflects the ongoing rate of discharges.

three frequency bands would not display. Such a hypothesis is supported by the calculation of the transfer gain between MAP and RBF (Fig. 5). As described in METHODS, this is an index of the degree to which MAP as an input affects RBF as an output. In the intact group, the gain was between 0.6 and 1.0 ml · min21 · mmHg21 (mean 6 SD 0.83 6 0.12) for frequencies .0.25 Hz (when the coherence function was .0.5). In the denervated group, the calculated gain was significantly greater across all frequencies, between 0.9 and 1.4

ml · min21 · mmHg21 (mean 6 SD, 1.19 6 0.10, P , 0.001). This gain was calculated using the spectrums obtained from detrended data in which the mean RBF and MAP were removed, leaving the oscillatory portion of the signal. The coherence between MAP and RBF was also significantly greater in denervated rabbits, suggesting tighter coupling between the two variables in the absence of RSNA. Overall, these results indicate that nerve activity maintains a constant level of vasoconstriction without inducing oscillations in RBF at


R1287

Fig. 2. Mean spectral components for RSNA, MAP, RBF, and HR from the group of 7 renal nerve intact rabbits before commencement of hemorrhage. Note the peak at ,0.3 Hz in RSNA but not in the other variables, the spectrogram being composed of mostly lowerfrequency (,0.25 Hz) components. The group of denervated rabbits had similar frequency spectrums for RBF, MAP, and HR (refer to Fig. 3 for comparisons).

certain frequencies. Although there were oscillations in RSNA at 0.3 Hz under control conditions, these were not pronounced enough to induce similar frequency oscillations in RBF. The phase angle between MAP and RBF was steady across all frequencies .0.25 Hz, averaging 0.067 6 0.1 (SD) radians (equivalent to 5–40 ms, depending on the frequency). This indicates that RBF oscillates essentially in the same time as MAP across frequency bands .0.25 Hz. The phase relationship was not significantly different between the intact and denervated groups of rabbits. Transfer gain in the low-frequency band (where coherence was less but still .0.3) was considerably less

Fig. 3. Recording of the original RSNA neurogram and its rectified and short-term (20 ms) integration (Int RSNA), arterial pressure (AP), and RBF from one conscious rabbit under control conditions. Tick marks under the integrated RSNA signal represent discharges detected by the data acquisition program.

than at higher frequencies, varying between 0.2 and 0.45 ml · min21 · mmHg21. This suggests that lowerfrequency oscillations in MAP have less influence on RBF variability compared with the high-frequency oscillations. Hemorrhage, interactions between arterial pressure and RBF, and the influence of renal nerves. Figure 6 illustrates the changes in spectral power for RBF, RSNA, and MAP from one rabbit during the 20-min period of hemorrhage. The initial phase of hemorrhage (0–5 min), when total sympathetic nerve activity was beginning to increase and RBF to fall, was associated with pronounced increases in the power of oscillations centered at 0.3 Hz in RSNA. This was associated with the emergence of oscillations at this frequency in RBF, changes that were absent in the denervated group of animals. An oscillation at 0.3 Hz also increased in power in MAP (Table 1). However, during the 5–15 min of hemorrhage, this oscillation in RBF was not maintained and was replaced by increased power of lowfrequency oscillations between 0.1 and 0.2 Hz in both groups of animals. In the last 5 min of hemorrhage, however, when the reduction in mean RBF was greatest, the RBF variability was almost zero. This was despite sustained variability in MAP and RSNA. To make comparisons between groups of animals over the period of hemorrhage, the percentage of the total spectral power was plotted against time (Fig. 7). To avoid misinterpretation of the changes in spectral power using this index, particularly when overall variability may be reduced, the absolute power over the different time periods is also shown in Table 1.

R1288


Fig. 4. Percentage of total spectral power before hemorrhage that can be accounted for in three defined frequency bands (low 0.04–0.25 Hz, mid 0.25–0.45 Hz, and high 0.75–1.4 Hz). Filled bars, intact group; open bars, denervated group. Values are means 1 SE. There were no differences between groups in the proportions.

Low frequency (0.04–0.25 Hz). Hemorrhage was associated with pronounced significant increases in power of oscillations in this frequency band for RBF, MAP, and RSNA (Fig. 7 and Table 1). Spectral peaks were seen centered between 0.15 and 0.20 Hz in RBF and sometimes overlapping with the 0.3-Hz peak described above (Fig. 6). There were no differences between the intact and renal denervated groups of rabbits in their responses to hemorrhage at this frequency despite the large increases in power at this band in RSNA. This suggests that the origin of this band in RBF and MAP was not RSNA directly. In the 15–20 min of hemorrhage, although the percentage of the total power for RBF oscillations continued to increase, the absolute power had declined. This period corresponds to the lowest RBF levels and indicates a reduction in overall variability during low RBF flow across all frequencies (Table 1). Mid frequency (0.25–0.45 Hz). When the RBF response to all periods of hemorrhage was compared between groups for this frequency band, there were significant differences between intact and denervated groups. This was due to differences in the 0- to 10-min period of hemorrhage in which the 0.3-Hz peak emerged in RBF, which was mediated by RSNA as it did not occur in denervated rabbits. However, in the 10- to 20-min period of hemorrhage, the RBF power in both groups decreased both as a percentage of the total power and in absolute power (Fig. 7 and Table 1). With regard to MAP, both intact and denervated rabbits had significant increases in power during hemorrhage, reaching maximum during the 5- to 10-min period of hemorrhage (Table 1). High frequency (0.75–1.40 Hz). There were significant decreases in the high-frequency component of RBF

oscillation as a percentage of the total power and absolute power in both intact and denervated rabbits, with the overall response not being different between the groups (Fig. 7 and Table 1). Absolute power in RSNA and MAP was significantly increased during hemorrhage (Table 1). Interactions between MAP and RBF during hemorrhage. As indicated above, the transfer gain between MAP and RBF was significantly greater in the denervated group of rabbits across all frequencies, suggesting that RSNA plays a tonic role in dampening the effects of oscillations in MAP on RBF. Figure 8 shows the changes in gain during hemorrhage for each of the three frequency bands. Although in the first 10 min of hemorrhage gain was steady in both groups, the second 10 min was associated with a significant reduction in gain across all frequencies. It is this period that corresponded to the largest reduction in MAP and RBF. Thus, in this period of hemorrhage, the oscillations in MAP were not being transmitted to the same extent to RBF, i.e., increased dampening. This observation is also clear from the reduction in overall RBF variability (Fig. 6 and Table 1) despite maintained variability in MAP. DISCUSSION

The aim of our study was to examine how the various rhythms in RBF are affected by hemorrhage and to what extent these oscillations are driven by changes in RSNA and/or arterial pressure. There are several findings of interest. First, although there was little effect of the actual oscillations in RSNA on RBF under control conditions, the lower resting RBF in intact vs. denervated rabbits and the reduced effect of any fluctuations


Fig. 5. Mean gain, coherence, and phase between MAP and RBF in the period before commencement of hemorrhage. Solid line, intact group; broken line, renal denervated group. Gain and coherence were significantly greater (P , 0.0001) across all frequencies in the denervated group. At frequencies .0.2 Hz, phase was close to zero in both groups of animals, indicating that oscillations in RBF occur at the same time as oscillations in MAP.

in arterial pressure on RBF (lower transfer gain values) compared with denervated rabbits suggest that the nerves have a tonic vasoconstrictive action and serve to alter the effect of oscillations in arterial pressure under control conditions. Second, we observed a distinct rhythm every 3 s in RSNA, a rhythm not present in RBF under control conditions. Although with the onset of hemorrhage and increases in the strength of oscillations in RSNA at this frequency, a similar rhythm was induced in RBF. These changes must be due to the renal nerves, as they were not present in denervated rabbits. We observed that there was an increase in the size of the discharges rather than their rate of firing with

R1289

hemorrhage. This reflects the recruitment of more nerve fibers (38) while the rate reflects the inherent generation of bursts by the central nervous system and their modulation by inputs from baroreceptors (30, 32). Thus these two components are likely to reflect different central nervous system controlling processes. We have previous evidence that afferent stimuli alter these two components of RSNA in a differential manner, supporting this hypothesis (31). It should be noted that we are yet to establish a functional effect of this patterning of RSNA, although changes in amplitude may reflect the recruitment of fibers innervating different sites within the kidney vasculature. The analysis of the frequency characteristics of a number of cardiovascular variables, such as HR and arterial pressure, is increasingly being used to study the role of sympathetic activity in the regulation of the cardiovascular system. In particular, changes in the spectral power are often being measured during maneuvers designed to alter sympathetic activity, such as standing and hand grip, and changes in the power associated with a variety of pathological conditions (29, 46). However, the difficulty arises in human studies from a lack of appropriate control because the denervated state cannot be assessed. Therefore, the possibility that the rhythms are nonneural in origin cannot be discounted. Such a rhythm was observed in this study between 0.15 and 0.20 Hz (5–7 s periodicity). Because this frequency was close to the rhythm of neural origin at 0.3 Hz and because its power was increased during hemorrhage, as RSNA was also increased, it is possible that it may be mistakenly considered to be of neural origin. However, its presence in RBF in the renal denervated group of rabbits indicates that this is not the case and that its presence in RSNA is the result of other mechanisms, possibly from arterial baroreflexes or autoregulation. Most of the oscillations in RBF were in this low-frequency band even under resting conditions, and, as such, they appear to be an integral part of RBF variability in the conscious animal whose strength and occurance may be increased during stimuli. The frequency of this rhythm suggests that they are not the result of tubuloglomerular feedback, whose frequency is generally recorded between 1 and 3 cycles/min (27), although dependent on nephron length (20). Our data segments of 50 s were too short to capture such events. Oscillations between 0.1 and 0.2 Hz have been observed in isolated blood vessels, suggesting that they are an intrinsic myogenic property of the smooth muscle in the vessel wall (36, 40). There are indications that nitric oxide plays a permissive role in their generation and that the presence of an endothelium is essential for their occurrence (16). Furthermore, their strength can be increased by a range of vasoconstrictors, such as norepinephrine, vasopressin, and angiotensin II (7, 13). In the present study, we found that the power of this frequency was increased in the last 10 min of hemorrhage corresponding to the onset of the reduction in arterial pressure. This hypotensive period is well established to be associated with increases in circulating levels of norepinephrine and angiotensin II (8, 41). We

R1290


Fig. 6. Changes in spectral power for one rabbit during the 20-min period of hemorrhage for RBF, RSNA, and MAP. Initial phase of nonhypotensive hemorrhage (0–10 min) was associated with pronounced increases in oscillations centered at 0.3 Hz in RSNA and the appearance of oscillations in RBF at this frequency (note the different y-axis scales between some periods). See Fig. 7 for mean values and comparison between intact and denervated groups.

suggest that in vivo, under control conditions, the blood vessels of the kidney are under some degree of tonic myogenic activity at this frequency as a consequence of the normal circulating levels of vasoconstrictors and

that stimuli that evoke physiological increases in the level of these compounds can increase the power of these oscillations. The purpose of these oscillations is unknown; however, we have already postulated that

R1291


Table 1. Changes in absolute spectral power in intact rabbits during the different periods of hemorrhage Hemorrhage, min Control

RBF Low frequency Mid frequency High frequency RSNA Low frequency Mid frequency High frequency MAP Low frequency Mid frequency High frequency

0–5

5–10

10–15

15–20

1.15 6 0.44 0.76 6 0.34 0.45 6 0.21

1.96 6 0.77 1.38 6 0.51 0.69 6 0.31

1.81 6 0.52 1.13 6 0.37 0.52 6 0.20

1.94 6 0.75 0.77 6 0.58 0.23 6 0.14

0.65 6 0.28 0.13 6 0.06 0.03 6 0.02

7.75 6 1.90 6.71 6 1.29 3.93 6 1.00

18.32 6 4.50 19.09 6 5.49 12.73 6 5.69

37.97 6 15.40 31.75 6 8.50 10.95 6 3.70

62.24 6 21.65 41.09 6 13.35 16.29 6 6.34

145.29 6 57.59 42.27 6 13.58 14.18 6 4.66

2.30 6 0.50 0.79 6 0.15 0.35 6 0.07

4.82 6 0.81 2.47 6 0.88 0.63 6 0.20

15.67 6 6.36 8.61 6 5.33 0.73 6 0.22

18.35 6 7.39 3.55 6 1.19 0.92 6 0.36

14.46 6 2.50 3.22 6 1.78 0.71 6 0.41

Values are means 6 SE. RBF, renal blood flow; RSNA, renal sympathetic nerve activity; MAP, mean arterial pressure; low frequency, 0.04–0.25 Hz; mid frequency, 0.25–0.45 Hz; high frequency, 0.75–1.40 Hz. Each frequency band for each variable was significantly changed during hemorrhage.

the dynamic response of juxtaglomerular cells, tubular cells, or vascular smooth muscle cells to changes in the frequency components of RSNA may vary (22). For instance, in response to an increase in low-frequency

variations in RSNA, the juxtaglomerular cells could release renin, whereas the renal vascular smooth muscle cells could respond with alternating constriction and relaxation without altering the mean level of RBF.

Fig. 7. Changes in spectral composition of oscillations in RBF (A, C, and E) and RSNA (B, D, and F) during the phases of hemorrhage. Data were divided into the prehemorrhage control period and averages from 5-min stages throughout hemorrhage. r, Renal nerve intact rabbits; s, denervated rabbits. Data were calculated as the percentage of the total power comprised by each frequency band. * Significant difference between the intact and denervated groups of rabbits in response to hemorrhage, i.e., there was an increase in the power of RBF oscillations between 0.25 and 0.4 Hz during the first 10 min of hemorrhage, which did not occur in denervated rabbits. ** Both intact and denervated groups underwent significant changes from control during hemorrhage.

R1292


Fig. 8. Changes in gain, phase, and coherence between MAP and RBF during the phases of hemorrhage in intact rabbits (A) and denervated rabbits (B). r, Changes in the low-frequency band (0.04–0.25 Hz); s, midfrequency band (0.25–0.45 Hz); m, highfrequency band (0.75–1.4 Hz). There was a significant decline across all frequency bands for both groups of animals during hemorrhage, as indicated by asterisk (*).

In our previous study, we identified a 0.3-Hz oscillation in RSNA that was, under control conditions, not strong enough to be transmitted to RBF (22). However, when total RSNA levels were increased during hypoxia, this increased the power of these oscillations in RSNA and consequently drove oscillations in RBF. We have confirmed these observations by showing an increase in power at 0.3 Hz during the initial phase (0–10 min) of hemorrhage, which produced oscillations in RBF at this same frequency. However, our previous study was limited in that the absolute RBF was not measured, and thus direct comparisons between intact and denervated groups in transfer gains could not be made. Although not the focus in the present study, we also found a 0.3-Hz rhythm in arterial pressure. This was submerged by the dominant lower-frequency power under control conditions, although Table 1 and Fig. 6 show clear increases in the occurrence of this rhythm during hemorrhage. There have been few studies showing the direct coupling between sympathetic activity

and arterial pressure in this frequency band although it was previously observed by Brown et al. (3), most previous studies inferring that a change in the power of arterial pressure at this frequency was due to sympathetic activity without having directly measured the nerve activity (5, 14, 24). It should be noted that RSNA displays a number of frequencies .0.7 Hz, for example, those associated with respiration or the cardiac cycle. Indeed, .75% of the spectral power of RSNA is contained in these frequencies (22). However, this does not mean that the lower-frequency rhythms in the present study are of mere esoteric value because, unlike frequencies .0.7 Hz, they are slow enough to directly induce a rhythm of vasoconstriction and vasodilation in the smooth muscle of the vessels that the nerves innervate. In vitro work suggests the smooth muscle cannot respond any faster with neuroeffector delays at vasculature smooth muscle between 1 and 3 s (19, 42), and it is likely that oscillations above this frequency contribute to the overall level of vasoconstrictive tone within the vessels. The results of our study suggest that the nerves play a tonic role in setting the resting RBF level and in altering the transmission of naturally occurring oscillations in arterial pressure. If only arterial pressure affected RBF, we would have expected a high coherence between the two signals. Therefore, the relatively low coherence in the innervated kidneys suggests the presence of multiple inputs (21). Thus, in the absence of RSNA, there is greater coupling between arterial pressure and RBF, and resting levels of RSNA under control conditions alter this coupling. Our previous results (22) provided the a priori information needed to make such conclusions, where we found that changes in the presence of oscillations in RBF could be ascribed to changes in the power of oscillations in RSNA. Overall, this indicates that small changes in RNSA can affect renal hemodynamics, a hypothesis that we have earlier evidence supporting (34). Certainly, the observation that the small increase in RSNA during the initial period (0–5 min, before blood pressure started to fall) of hemorrhage was associated with a decrease in RBF additionally supports this hypothesis. However, it should be noted that this is at odds with a number of earlier studies in which it has been considered that RBF is affected little by mild changes in RSNA and, with relevance to the present study, that the increases in RSNA accompanying the initial stages of hemorrhage are preferential for altering sodium excretion and renin release (1). Morita (35) and Nelson and Osborn (37) had observed increases in RSNA between 60 and 90% that did not alter RBF. However, these observations were based on experiments performed in the conscious dog, a species in which it is suggested that RBF is less sensitive to changes in RSNA than the rat and rabbit (15, 18, 31). It is also relevant to note that we did not continue hemorrhage until the decompensatory phase of hemorrhage, characterized by a decline in RSNA and a rapid fall in arterial pressure (4, 17). Our primary end point for ceasing hemorrhage was when RBF had declined below 7 ml/min, as we were


concerned for the accuracy of the Transonic flow probes under conditions of extremely low flow. We also observed a fall in transfer gain between MAP and RBF in the last 10 min of hemorrhage across all frequencies in both intact and denervated animals. The reason for this is unknown but may indicate a reduced capacity to respond to changes in arterial pressure during low flow conditions. This time was associated with the largest reduction in arterial pressure, probably the greatest changes in hormonal levels and the highest RSNA. These may combine to ensure that the blood vessels behave as relatively rigid structures to all changes in arterial pressure. Limitations. We found resting RBF to be lower in intact rabbits vs. denervated rabbits, which we attributed predominantly to resting levels of renal nerve activity. However, we cannot discount the possibility that the significantly greater resting plasma renin levels in intact rabbits compared with denervated rabbits also played a role in setting the level of RBF. Certainly angiotensin II levels are well established to regulate RBF levels, as blockade of the renin angiotensin system leads to increased RBF (10, 44). The significantly enhanced increase in plasma renin levels in the denervated animals during hemorrhage compared with intact animals may account to some extent for the similarity in the steady-state RBF response to hemorrhage. Whatever the case, these alterations in the baseline level of renin were unlikely to constrain the dynamic response of the renal vasculature to frequency changes in RSNA and the measured transfer gain between MAP and RBF. It is also important to note that our analysis techniques made a number of assumptions on the linearity and stationarity of the data. This is particularly relevant when performing spectral analysis during hemorrhage and RBF is falling. The control of RBF will contain nonlinear elements that were not assessed by our analysis (21). We did take several steps to ensure that this effect on our results was minimized. First, we selected only periods of noise-free RSNA signals, as movement of the animal can induce a number of artifacts. Second, all of the data were detrended before spectral analysis. Third, we did not continue the hemorrhage until the RBF was extremely low and thus likely to be highly nonlinear. We suggest therefore that the interpretation of our results is confined to the linear control of RBF and cannot describe the nonlinear facets of the signal. Perspectives We suggest that, under control conditions, tonic renal nerve activity may smooth the effect of changes in arterial pressure on RBF. This would ensure a relatively stable flow within the renal microvasculature and therefore steady glomerular filtration rate, sodium excretion, and renin release in the short term. However, during stimuli that sustainably increase RSNA, this mechanism is altered to allow oscillations at a variety of frequencies to directly affect RBF variability. Although the origin of some of these oscillations at certain frequencies can be attributed to the renal

R1293

nerves, their functional importance for the kidney remains to be established. It is possible that certain effector mechanisms, such as renin release, are frequency modulated, responding to increases in the strength of oscillations at a particular frequency, but that other frequencies are selective for regulating sodium excretion. We conclude that hemorrhage is associated with profound changes in the regulation of RBF. Before decreases in arterial pressure occurred, oscillations in renal nerve activity with a period of 3–4 s increased in strength and were transmitted to RBF. However, subsequent decreases in arterial pressure were associated with decreases in the strength of these oscillations and the emergence of nonneural oscillations with a period of 5–7 s. These results suggest that RSNA mediates a proportion of the renal vasoconstriction during hemorrhage and plays a significant role in affecting the transmission of oscillations in arterial pressure to RBF. We are grateful to Amany Shweta for performing the renin assay and to Dr. Ben Janssen for constructive comments. This work was supported by a National Heart Foundation of Australia project grant and by grants from the National Health and Medical Research Council of Australia, the Auckland Medical Research Foundation, the New Zealand Neurological Foundation, and the Lottery Grants Board of New Zealand. Copies of the data acquisition and analysis system for sympathetic activity are available from S. Malpas ([email protected]). Present address for E. V. Lukoshkova: National Cardiology Research Center, 121552 Moscow, Russia. Address for reprint requests: S. C. Malpas, Dept. of Physiology, Univ. of Auckland Medical School, Private Bag 92019, Auckland, New Zealand. Received 18 August 1997; accepted in final form 16 January 1998. REFERENCES 1. Anderson, W. P., and G. Szenasi. Regional blood flow—renal. In: Blood Loss and Shock, edited by N. Secher, J. Pawelczyk, and J. Ludbrook. London: Arnold, 1994, p. 121–131. 2. Berger, R. D., J. P. Saul, and R. J. Cohen. Transfer function analysis of autonomic regulation. I. Canine atrial rate response. Am. J. Physiol. 256 (Heart Circ. Physiol. 25): H142–H152, 1989. 3. Brown, D. R., L. V. Brown, A. Patwardhan, and D. C. Randall. Sympathetic activity and blood pressure are tightly coupled at 0.4 Hz in conscious rats. Am. J. Physiol. 267 (Regulatory Integrative Comp. Physiol. 36): R1378–R1384, 1994. 4. Burke, S. L., and P. K. Dorward. Influence of endogenous opiates and cardiac afferents on renal nerve activity during haemorrhage in conscious rabbits. J. Physiol. Paris 402: 9–27, 1988. 5. Cerutti, C., C. Barres, M. P. Gustin, C. Julien, M. Lo, C. Paultre, M. Vincent, and J. Sassard. Sympathectomy, sinoaortic denervation and spectral powers of blood pressure and heart rate in lyon rats. Comp. Anal. Cardiovasc. Sig. 13: 243–256, 1995. 6. Cerutti, C., C. Barres, and C. Paultre. Baroreflex modulation of blood pressure and heart rate variabilities in rats: assessment by spectral analysis. Am. J. Physiol. 266 (Heart Circ. Physiol. 35): H1993–H2000, 1994. 7. Colantuoni, A., S. Bertuglia, and M. Intaglietta. The effects of alpha- or beta-adrenergic receptor agonists and antagonists and calcium entry blockers on the spontaneous vasomotion. Microvasc. Res. 28: 143–158, 1984. 8. Courneya, C. A., and P. I. Korner. Neurohumoral mechanisms and the role of arterial baroreceptors in the reno-vascular response to haemorrhage in rabbits. J. Physiol. Paris 437: 393–407, 1991. 9. Daffonchio, A., C. Franzelli, A. Radaelli, P. Castiglioni, M. Dirienzo, G. Mancia, and A. U. Ferrari. Sympathectomy and

R1294

10. 11.

12.

13. 14.

15.

16. 17. 18.

19. 20.

21. 22.

23.

24. 25.

26.

27. 28.


cardiovascular spectral components in conscious normotensive rats. Hypertension 25: 1287–1293, 1995. Denton, K. M., and W. P. Anderson. Role of angiotensin II in renal wrap hypertension. Hypertension 7: 893–898, 1985. Dorward, P. K., W. Riedel, S. L. Burke, J. Gipps, and P. I. Korner. The renal sympathetic baroreflex in the rabbit. Arterial and cardiac baroreceptor influences, resetting, and effect of anesthesia. Circ. Res. 57: 618–633, 1985. Evans, R. G., I. P. Hayes, and J. Ludbrook. Does the haemodynamic response to acute central hypovolaemia depend on the rate of fall of cardiac output? Clin. Exp. Pharmacol. Physiol. 19: 657–661, 1992. Fujii, K., D. D. Heistad, and F. M. Faraci. Vasomotion of basilar arteries in vivo. Am. J. Physiol. 258 (Heart Circ. Physiol. 27): H1829–H1834, 1990. Gonzalez, J. G., J. J. C. Valeriano, and M. F. Rodriguez. Autonomic mediation of short-term cardiovascular oscillations after acute hemorrhage in conscious rats. J. Auton. Nerv. Syst. 55: 123–130, 1995. Grady, H. C., and E. M. A. Bullivant. Renal blood flow varies during normal activity in conscious unrestrained rats. Am. J. Physiol. 262 (Regulatory Integrative Comp. Physiol. 31): R926– R932, 1992. Gustafsson, H., M. J. Mulvany, and H. Nilsson. Rhythmic contraction of isolated small arteries from rat: influence of the endothelium. Acta Physiol. Scand. 148: 153–163, 1993. Hasser, E. M., and J. C. Schadt. Sympathoinhibition and its reversal by naloxone during hemorrhage. Am. J. Physiol. 262 (Regulatory Integrative Comp. Physiol. 31): R444–R451, 1992. Hermansson, K., M. Larson, O. Kallskog, and M. Wolgast. Influence of renal nerve activity on arterioloar resistance ultrafiltration dynamics and fluid reabsorption. Pflu¨gers Arch. 389: 85–90, 1981. Hirst, G. D., and F. R. Edwards. Sympathetic neuroeffector transmission in arteries and arterioles. Physiol. Rev. 69: 546– 604, 1989. Holstein-Rathlou, N., A. Wagner, and D. Marsh. Tubuloglomerular feedback dynamics and renal blood flow autoregulation in rats. Am. J. Physiol. 260 (Renal Fluid Electrolyte Physiol. 29): F53–F68, 1991. Holstein-Rathlou, N.-H., and D. J. Marsh. Renal blood flow regulation and arterial pressure fluctuations: A case study in nonlinear dynamics. Physiol. Rev. 74: 637–681, 1994. Janssen, B. J. A., S. C. Malpas, S. L. Burke, and G. A. Head. Frequency-dependent modulation of renal blood flow by renal nerve activity in conscious rabbits. Am. J. Physiol. 273 (Regulatory Integrative Comp. Physiol. 42): R597–R608, 1997. Janssen, B. J. A., J. Oosting, D. W. Slaaf, P. B. Persson, and H. A. J. Struijkerboudier. Hemodynamic basis of oscillations in systemic arterial pressure in conscious rats. Am. J. Physiol. 269 (Heart Circ. Physiol. 38): H62–H71, 1995. Julien, C., Z. Q. Zhang, C. Cerutti, and C. Barres. Hemodynamic analysis of arterial pressure oscillations in conscious rats. J. Auton. Nerv. Syst. 50: 239–252, 1995. Korner, P. I., E. Badoer, and G. A. Head. Cardiovascular role of the major noradrenergic cell groups in the rabbit: analysis based on 6-hydroxydopamine-induced transmitter release. Brain Res. 435: 258–272, 1987. Korner, P. I., and A. W. T. Edwards. The immediate effects of acute hypoxia on the heart rate, arterial pressure, cardiac output and ventilation of the unanaesthetized rabbit. Q. J. Exp. Physiol. 45: 113–122, 1960. Leyssac, P. P., and L. Baumbach. An oscillating intratubular pressure response to alterations in Henle loop flow in the rate kidney. Acta Physiol. Scand. 126: 271–277, 1983. Ludbrook, J. Repeated measurements and multiple comparisons in cardiovascular research. Cardiovasc. Res. 28: 303–311, 1994.

29. Malliani, A., M. Pagani, and F. Lombardi. Physiology and clinical implications of variability of cardiovascular parameters with focus on heart rate and blood pressure. Am. J. Cardiol. 73: C3–C9, 1994. 30. Malpas, S. C. A new model for the generation of sympathetic nerve activity. Clin. Exp. Pharmacol. Physiol. 22: 11–15, 1995. 31. Malpas, S. C., R. D. Bendle, G. A. Head, and J. H. Ricketts. Frequency and amplitude of sympathetic discharges by baroreflexes during hypoxia in conscious rabbits. Am. J. Physiol. 271 (Heart Circ. Physiol. 40): H2563–H2574, 1996. 32. Malpas, S. C., and I. Ninomiya. The amplitude and periodicity of synchronized renal sympathetic nerve discharges in anesthetized cats: differential effect of baroreceptor activity. J. Auton. Nerv. Syst. 40: 189–198, 1992. 33. Malpas, S. C., and I. Ninomiya. A new approach to analysis of synchronized sympathetic nerve activity. Am. J. Physiol. 263 (Heart Circ. Physiol. 32): H1311–H1317, 1992. 34. Malpas, S. C., A. Shweta, W. P. Anderson, and G. A. Head. Functional response to graded increases in renal nerve activity during hypoxia in conscious rabbits. Am. J. Physiol. 271 (Regulatory Integrative Comp. Physiol. 40): R1489–R1499, 1996. 35. Morita, H. Dissociation between change in renal nerve activity and renal vascular resistance in conscious dogs. Jpn. J. Physiol. 36: 585–593, 1986. 36. Mulvany, M. J., C. Aalkjaer, and J. Christensen. Changes in noradrenaline sensitivity and morphology of arterial resistance vessels during development of high blood pressure in spontaneously hypertensive rats. Hypertension 2: 664–671, 1980. 37. Nelson, L. D., and J. L. Osborn. Neurogenic control of renal function in response to graded nonhypotensive hemorrhage in conscious dogs. Am. J. Physiol. 264 (Regulatory Integrative Comp. Physiol. 33): R661–R667, 1993. 38. Ninomiya, I., S. C. Malpas, K. Matsukawa, T. Shindo, and T. Akiyama. The amplitude of synchronized cardiac sympathetic nerve activity reflects the number of activated pre- and postganglionic fibers in anesthetized cats. J. Auton. Nerv. Syst. 45: 139–147, 1993. 39. Oliver, J. R., P. L. Korner, R. L. Woods, and J. L. Zhu. Reflex release of vasopressin and renin in hemorrhage is enhanced by autonomic blockade. Am. J. Physiol. 258 (Heart Circ. Physiol. 27): H221–H228, 1990. 40. Ross, G., E. Stinson, J. Schroeder, and R. Ginsburg. Spontaneous phasic activity of isolated human coronary arteries. Cardiovasc. Res. 14: 613–618, 1980. 41. Schadt, J. C., and J. Ludbrook. Hemodynamic and neurohumoral responses to acute hypovolemia in conscious mammals. Am. J. Physiol. 260 (Heart Circ. Physiol. 29): H305–H318, 1991. 42. Somlyo, A. P., and A. V. Somlyo. Flash photolysis studies of excitation-contraction coupling, regulation, and contraction in smooth muscle. Annu. Rev. Physiol. 52: 857–874, 1990. 43. Stauss, H. M., and K. C. Kregel. Frequency response characteristic of sympathetic-mediated vasomotor waves in conscious rats. Am. J. Physiol. 271 (Heart Circ. Physiol. 40): H1416–H1422, 1996. 44. Takishita, S., H. Muratani, S. Sesoko, H. Teruya, M. Tozawa, K. Fukiyama, and Y. Inada. Short-term effects of angiotensin II blockade on renal blood flow and sympathetic activity in awake rats. Hypertension 24: 445–450, 1994. 45. Walton, R. P., J. A. Richardson, J. E. P. Walton, and W. L. Thompson. Sympathetic influences during hemorrhagic hypotension. Am. J. Physiol. 197: 223–230, 1959. 46. Weise, F., G. M. London, A. P. Guerin, B. M. Pannier, and J. L. Elghozi. Effect of head-down tilt on cardiovascular control in healthy subjects: a spectral analytic approach. Clin. Sci. (Colch.) 88: 87–93, 1995.