We examined the ability of different frequencies in sympathetic nerve activity (SNA) to induce oscillations in renal blood flow (RBF). In anesthetized rabbits the renal nerves were stimulated using modulated sine patterns (base frequency 5 Hz, 5-ms duration pulses) that varied in amplitude between 0 and 10 V at a frequency between 0.04 and 1.0 Hz. The strengths of the induced oscillations in RBF were calculated using spectral analysis. Although faster rhythms in simulated SNA >0.6 Hz contributed to the level of vascular tone, 95% of the power in the frequency response curve was below this frequency, indicating a low-pass filtering/integrating characteristic of the vasculature. Frequencies <0.6 Hz were associated with increasing ability to induce oscillations in RBF. The ability of an SNA rhythm at 0.6 Hz to induce a rhythm in RBF was 21 times less than that at 0.25 Hz. At 0.16 Hz there was a distinct peak in the frequency response curve, indicating the vasculature was more sensitive in this frequency band to sympathetic stimulation. Blockade of endogenous nitric oxide byN G-nitro-l-arginine methyl ester (l-NAME; 20 mg/kg) did not alter resting RBF levels nor was the low-pass filtering/integrating characteristic of the vasculature to nerve stimulation changed (i.e., the curve was not shifted left or right); however, there was a selective increase in the sensitivity to stimulation at 0.16 Hz, i.e., larger oscillations in RBF were evoked. These results indicate an ability of SNA to induce resonant oscillations in the renal vasculature and that there may be active and passive modulators of these responses. Naturally occurring oscillations in SNA <0.6 Hz are likely to contribute to the dynamic control of RBF, ensuring it responds rapidly and with high gain to the stimuli of daily life, while filtering out the faster oscillations ensures stable glomerular filtration.
- sympathetic nervous system
- spectral analysis
- nitric oxide
sympathetic nerve activity (SNA) is well established to contain oscillations at a number of frequencies (16). The most obvious of these is related to the cardiac cycle and represents the synchronized activation of thousands of individual axons. Additional fast rhythms have been observed at 10 Hz and are also related to the respiratory cycle (1, 3). Recently it has emerged that a rhythm between 0.1 and 0.4 Hz, depending on the species, also exists (2, 18). Although this rhythm occupies a small percentage of the total power in the oscillations (between 15 and 20%; Ref. 14), it differs from the faster rhythms present by directly inducing oscillations in vascular tone. With regard to the renal vasculature in rabbits, a rhythm between 0.2 and 0.3 Hz can be observed in renal blood flow (RBF) under normal resting conditions (18). During stimuli resulting in increased sympathetic drive, such as hypoxia and hemorrhage, the strength of this rhythm also increases. This rhythm appears to be a widespread feature of blood flow through a number of organs as it is also observed in blood pressure (2, 15, 30). Understanding the origin and effect of this rhythm is likely to be of considerable clinical importance, inasmuch as previous studies showed decreased blood pressure and heart rate variability at this frequency to be associated with increased risk of cardiovascular mortality (7).
Previously it was hypothesized that the faster rhythms (>1 Hz) in SNA contribute to the overall tone of the vasculature but are unable to induce oscillations in the vasculature (14, 29) because the neuroeffector delays at vasculature smooth muscle are sluggish, in the order of 2–3 s (10, 28). Thus it was proposed that the vasculature acts like a low-pass filter to the different patterns of SNA (31). The nature of this filter has yet to be identified but has several important implications with regard to end organ function. First, it is possible that different end organ functions are tuned to the different frequencies of SNA (23). For example, within the kidney certain frequencies may encode for renin release, whereas others regulate sodium excretion or RBF as proposed by Janssen et al. (14). To define the nature of the frequencies and their effect may allow targeting of therapeutic treatments or stimuli to selectively alter function. Second, it may be possible that the vasculature is sensitive to select frequency bands, in other words, that it may “listen” to the various frequencies present in the naturally occurring SNA. It is possible that hormones and intrinsic pathways may alter this listening ability. Such a faculty would imply a specialized frequency response mechanism within the vasculature. Furthermore, definition of the frequency response function for the vasculature using methodology that keeps the integrated responses intact may provide the framework for exploring the mechanical properties of the vasculature in response to disease or therapeutic treatment.
It has been proposed that nitric oxide is an important mechanism in buffering the variability in blood pressure between 0.2 and 0.6 Hz (22). This frequency band is due predominantly to SNA (2, 18), therefore it is likely that the two systems are in some balance. This balance may occur at the level of the vasculature and alter the ability of SNA to produce vasoconstriction. Therefore nitric oxide may be a factor in setting the low-pass filtering characteristics of the vasculature to patterns of SNA. In the present study, using patterned electrical stimulation of the renal nerves in anesthetized rabbits, we defined the frequency response characteristics of the renal vasculature and tested the hypothesis that blockade of endogenous nitric oxide alters the magnitude of the frequency response.
Animal preparation. New Zealand White rabbits weighing between 2.2 and 3.2 kg were anesthetized with urethan (1.4 gm/kg iv) and infusion of Saffan (alphaxalone 0.9%, 1 ml/h). The renal artery and nerves were exposed via retroperitoneal incision. A transit time flow probe (type 2SB Transonic Systems, Ithaca, NY) was placed around the renal artery. The renal nerves were identified using a surgical microscope and placed across a pair of hook electrodes. The central end of the nerves was either cut or crushed. Blood pressure was measured via a femoral artery catheter.
Electrical stimulation protocol. Patterned electrical stimulation of the renal nerves was produced using purpose-written software in the LabVIEW graphical programing language (National Instruments) coupled to a LabPC+ data acquisition board (National Instruments). Stimulation was at 5 Hz with 5-ms duration pulses, i.e., 5 ms on 195 ms off. To produce a range of frequencies between 0.04 and 1.0 Hz, the voltage of each of these pulses was varied in a sine fashion between 0 and 10 V at a predetermined frequency. It is important to consider whether the chosen input stimuli, varying between 0 and 10 V, is representative of the vasculature’s ability to respond. Therefore in preliminary experiments in six rabbits the renal nerves were stimulated at different constant voltages ranging from 2 to 10 V (5-ms pulses at 5 Hz) for 2-min periods followed by 2 min of recovery. The stepped RBF response to these different voltage levels from one rabbit is shown in Fig. 1 and clearly indicates that patterned stimulation between 0 and 10 V would not be outside the responsiveness of the vasculature, i.e., the stimulation pulses up to 10 V would not be clipped at 8 V or less. This validation is further discussed inresults. The patterned stimulation protocol was termed modulated sine, and an example of the output is shown in Fig. 2. The form of modulated sine stimulation at a range of frequencies between 0.04 and 1.0 Hz was applied to the renal nerves for periods of 7 min. The order of the stimuli was randomized and interspaced with a 2-min recovery period before the next stimulation. It is important to note that the only difference between each modulated sine pattern is the frequency (e.g., 0.04, 0.08 Hz, etc.) and the power of each of these patterns applied to the nerves was the same. Thus the mean change in RBF could be compared between frequencies. After a full sequence of modulated sine stimulation, six rabbits were given a bolus ofN G-nitro-l-arginine methyl ester (l-NAME; 20 mg/kg iv) followed by a constant infusion of 5 mg ⋅ kg−1 ⋅ h−1to block endogenous nitric oxide. A further group of five rabbits underwent the same protocol, however instead of receivingl-NAME they received a saline infusion only.
Data acquisition and analysis. The programmable stimulator described above also performed continuous sampling of RBF, the stimulator output and arterial pressure at 500 Hz. The arterial pressure waveform was used to calculate the heart rate and collated with RBF to form a file containing the mean values per heart beat. This beat-to-beat data then underwent spectral analysis to determine the spectral power of the induced oscillations in RBF (18). Initially all beat-to-beat signals were replayed on screen for visual inspection. Data segments with artifacts or nonstationarity were eliminated. The beat-to-beat data segments were resampled at 10.24 Hz and were partitioned into segments of 50-s (512 points) length overlapping by 25 s. Each section underwent linear detrending to remove the underlying mean value and was windowed with a tapered cosine function. The spectral density was calculated using the fast Fourier transform algorithm (LabVIEW; National Instruments), and the final spectra were estimated from ensemble averages of the spectral and cross spectral densities for each 50-s data segment. The phase of the transfer function between nerve stimulation and RBF was calculated as the quotient of cross spectrum and the input power spectrum. However, because stimulation induces a decrease in RBF, i.e., these signals are out of phase, the phase shift may be calculated not from zero but from 1π. Thus the true phase delay expressed as fractions of π(φ1) is 1 minus the measured phase delay. This true phase delay was used to calculate the actual time delay between the stimulation and the RBF response and is shown by Equation 1 whereTd is the time delay in seconds, andf is the frequency of stimulation.
The effect of stimulating at three frequencies (0.04, 0.16, and 0.6 Hz) in one rabbit is shown in Fig. 3. Although each was associated with reductions in mean RBF, only stimulation at 0.04 and 0.16 Hz produced an oscillation in RBF. The mean change in RBF from resting level and during each stimuli was also determined. The absolute changes were normalized against the mean fall during 0.04 Hz stimulation. Given that the strength of the stimuli were the same, one might have expected a similar decrease in RBF with each stimuli; however stimulation at 0.04 and 0.6 Hz were associated with significantly greater reductions in RBF compared with the surrounding stimulation frequencies (Fig. 3).
As described in methods, the amplitudes of the oscillations in RBF were measured using spectral analysis. A series of spectrograms showing the different modulated sine stimulations in a single rabbit is shown in Fig.4. With increases in the modulated sine frequency, the induced oscillation in RBF became smaller. It should be remembered that when spectral analysis was applied to the electrical stimulation signal, the power of the applied sinusoidal stimulation was the same for each frequency.
In methods (Fig. 1), we validated the use of stimulation up to 10 V; this was confirmed again by applying sinusoidal stimulation varying between 0 and 5 V or 0 and 10 V (5-ms pulses at 5 Hz). When the amplitude of the induced oscillations in RBF were calculated using spectral analysis, the power of the oscillation at any frequency was always greater when the stimulus varied between 0 and 10 V. A representative example is shown in Fig.5.
Because the absolute spectral power was different between animals, the amplitude of the oscillations in RBF was normalized by calculating the power as a percentage of the power of the oscillation at 0.04 Hz (the frequency of greatest power in each animal). With regard to the mean frequency response relationship, 95% of the curve was completed at frequencies <0.6 Hz (Fig. 6). In terms of relative power, the amplitude of the power at 0.6 Hz was 21 times lower than at 0.25 Hz. The decay function was, however, not smooth, with a distinct peak at 0.16 Hz.
l-NAME. Blockade of endogenous nitric oxide withl-NAME produced a significant decrease in heart rate (308 ± 20 to 283 ± 15 beats/min), although no change in mean arterial pressure (71 ± 2 mmHg). The resting RBF before any electrical stimulation was 25 ± 4 ml/min and was not altered by l-NAME. Frequency response curves from four rabbits to the modulated sine stimulation and the group mean are shown in Fig. 6. There was a significant difference from control only at 0.16 Hz where the effect of stimulating at this frequency lead to significantly stronger oscillations in RBF. At 0.16 Hz, the induced oscillations in RBF were on average 53 ± 18% stronger afterl-NAME for the same stimulation input. The curve was not, however, shifted to the left or right, i.e., 95% of the frequency response curve was still completed by 0.6 Hz. In the five rabbits that received saline only instead ofl-NAME, the frequency response curve was unchanged from control. The mean reductions in RBF with each stimuli shown in Fig. 3 were not significantly different from the control responses, i.e., the greater reductions in RBF occurred with stimulation at 0.04 and 0.6 Hz.
It is important to note that our preparation leaves the effect of arterial pressure intact. This has the potential to confound the response to SNA stimulation, as arterial pressure may already contain significant variability in similar frequency bands to the electrical stimulation. We therefore conducted spectral analysis on the arterial pressure signal under control conditions and during each of the stimulation periods. Although the arterial pressure spectrum contained peaks associated with the respiratory cycle, we found little variability in our anesthetized rabbits between 0.05 and 1.0 Hz. Importantly the sinusoidal stimulation did not alter mean arterial pressure. After l-NAME the blood pressure spectrum was unchanged.
Phase delay between stimulation and the RBF response. The calculated phase angle up to 0.5 Hz between the nerve stimulation as the input and RBF as the output is shown in Fig. 7. As stated inmethods, because an increase in simulated SNA induces a decrease in RBF these signals are therefore out of phase. At 0.04 Hz under control conditions, the lag between SNA and RBF is equivalent to 3.7 ± 0.6 s. It was clear from this plot of the true time delay that there was a component of the transfer function that was not dependent on the frequency of stimulation. For the group of animals this equaled 1.14 s and was termed a constant lag. Blockade of nitric oxide did not alter the phase relationship
In the present study we applied modulated sine stimulation between 0.04 and 1 Hz to the renal nerves and measured the amplitude of the resultant oscillations in RBF. There are several important features in the responses obtained; first, although faster rhythms in simulated SNA >0.6 Hz contributed to the level of vascular tone, 95% of the power in the frequency response curve was below this frequency, indicating a low-pass filtering/integrating characteristic of the vasculature. Second, there was evidence of increased sensitivity of the vasculature to stimulation at 0.16 Hz. This sensitivity was increased after blockade of nitric oxide, although the rest of the integrating response was otherwise unaltered. Third, the average reduction in RBF from mean levels was not the same with the different frequencies of sinusoidal stimuli, with larger decreases occurring at 0.04 and 0.6 Hz than stimuli at surrounding frequencies.
Resonance. On first consideration of the data, one could conclude that the ability of SNA to induce stronger oscillations in RBF at 0.16 Hz than at 0.12 or 0.25 Hz was because the vasculature was tuned to this frequency band, that is that some undefined mechanism confers a frequency-dependent relationship between SNA and the vasculature. Such a possibility could provide a functional purpose for the naturally occurring frequencies present in SNA. We suggest that the results can be explained by SNA revealing resonance in the vasculature, where the input stimulus, in this case SNA, contains the frequency components necessary to induce resonant oscillations in the vasculature in a manner suggestive of phase locking and entrainment. The autoregulatory phenomena of the renal vasculature has been shown to exhibit resonance between 0.1 and 0.25 Hz (4, 12, 26), possible where oscillations in arterial pressure reveal resonance in the renal vasculature. Before our study it was not known whether the responsiveness of the renal vascular bed to changes in perfusion pressure is different from that resulting with changes in SNA. The ability of the renal nerves to evoke resonance has not previously been considered, as it was thought that the vasculature was insensitive to small changes in SNA (6). However the present work combined with our earlier work suggests that the vasculature maintains a high degree of sensitivity to changes in SNA (17, 18). The ability to evoke resonance should not be considered an esoteric phenomenon, as it has the ability to explain, using the simple experimental protocols outlined in the present study, the properties of the vasculature and of some of the naturally occurring oscillations observed in blood flow, thus opening a new avenue for examining control of renal hemodynamics. Previous approaches to determine such features have generally isolated the kidney neurally from the circulation and then applied discrete frequencies and amplitudes of renal perfusion pressure via a pump (13).
During hemorrhage it was observed that a low-frequency oscillation between 0.15 and 0.2 Hz emerges in RBF (18). This was not due to SNA, as it also occurred in denervated animals. Its emergence in arterial pressure suggested that it was a widespread feature of the cardiovascular system. The present study offers an explanation for these oscillations, where the natural resonance of the vasculature is in this frequency band but that changes in circulating hormones, such as ANG II, or endogenous mediators, such as nitric oxide, alter the amplitude of this resonance in response to changes in blood pressure. In the present study we showed that administration ofl-NAME selectively increased the amplitude of the resonance frequency. Thus the presence of NO has an important buffering effect on the ability of the vasculature to oscillate. 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 (21, 25), and the presence of an endothelium is essential for their occurrence (9).
Many mechanical structures have the ability to exhibit a resonance frequency in response to stimuli, providing those stimuli contain the appropriate frequencies. Thus the rhythmical inputs of arterial pressure and SNA will provoke resonance in the renal vasculature. Nitric oxide or other circulating hormones that exhibit much slower rhythmicities cannot induce resonance by themselves but may actively modulate the ability to resonate. Thus the calculation of the frequency response curve as shown in Fig. 6, with its resonance peak, has the potential to provide extensive information on the properties of the vasculature and the effect of hormones on these properties. Such information can be extracted through a dynamic systems modeling approach from which parameters such as mass, elasticity, and damping can be determined, using appropriate active and passive mechanical analogies. Previous approaches to determining such information have generally had to rely on an isolated vessel or organ approach (5, 8,26). Clearly, the observed frequency response is the result of complex interactions between the characteristics of the neural-muscle coupling, the characteristics of the second messenger pathways in the smooth muscle (the excitation-contraction coupling), the interaction with the intrinsic regulatory systems of the kidney (tubuloglomerular feedback and the myogenic response), and the mechanical properties of the tissues in question. Such approaches remove many of the factors that may influence the dynamic control of vessel tone, including SNA and a multitude of hormones. We suggest that our simpler methodology provides such information while still leaving the integrated control systems intact; furthermore, it provides the foundation for study of the changes in the control of vasculature tone in cardiovascular diseases such as hypertension.
A low-pass filter/integrator. Our initial aim in conducting the present study was to determine the frequency response of the renal vasculature, that is, to determine the extent of its responsiveness to different frequencies of SNA. Previous studies have shown, while measuring the SNA and blood flow in conscious rabbits and rats, that the higher frequencies present in SNA signals associated with the cardiac cycle and respiration could not directly induce rhythms in blood flow, but rather that this activity contributed to the steady tone under which the vasculature was held (14, 29, 31). Although the slower rhythms in SNA between 0.1 and 0.4 Hz were only a small percentage of the total power in SNA (∼15%), their importance with regard to circulatory control was underlined by their ability to directly induce a rhythm of vasoconstriction and vasodilation within the vasculature they innervated. This means that the vasculature must act as a low-pass filter/integrator to the various rhythms in SNA. Rosenbaum and Race (24) proposed a corner frequency of 0.017 Hz for the hindlimb vasculature of the dog, which is far too slow to allow any naturally occurring oscillations in SNA to affect blood flow. However their methodological approach of measuring changes in perfusion pressure, rather than blood flow, in response to sinusoidal nerve stimulation may have affected their results; alternatively the hindlimb vasculature may have very different mechanisms for responding to oscillations in SNA. However Stauss and colleagues (31) showed in the rat that stimulation frequencies <0.5 Hz could induce oscillations in mesenteric blood flow, but frequencies >1 Hz could not. Our aim was to define this frequency response curve accurately and to assess whether the response is regulated in part by nitric oxide. Our results show that in the rabbit, rhythms in SNA >0.6 Hz do not induce rhythms in RBF but produce a steady degree of vasoconstriction. At this frequency the ability of a SNA rhythm to induce a rhythm in RBF is 21 times less than that at 0.25 Hz. Given nitric oxide’s ability to buffer blood pressure oscillations between 0.2 and 0.6 Hz (22), one might have expected that it could interact with the vasculature to regulate the corner frequency of responses to SNA. However, neither the frequency response curve nor the phase plots were shifted.
Different reductions in renal blood flow with different stimuli. We were surprised to observe that despite the strength of stimuli being equal at each frequency, the mean reduction in RBF was not similar for all stimuli. Sinusoidal stimulation at 0.04 and 0.6 Hz was associated with larger reductions in RBF than surrounding frequencies. It is possible that 0.04 Hz is also a resonant frequency, as oscillations in tubular pressure at 0.03–0.05 Hz have been shown to influence RBF (11, 32). The large constriction associated with stimulation at 0.6 Hz is more difficult to explain, as the stimuli did not evoke oscillations in RBF but rather a steady increase in vasoconstrictor tone (Fig. 2). It is possible that this frequency evokes a different neurotransmitter response; for example, neuropeptide Y has been shown to be increased by low-frequency stimulation <2 Hz but higher frequencies evoke both neuropeptide Y and norepinephrine (23). Also stimulation at low frequencies has been shown to be highly calcium dependent but less so during higher frequency stimulation (27). The frequencies applied have generally been well above those of interest in the present study.
Limitations. It could be argued that stimulating the renal nerves electrically is artificial and that it has little place in delineating the role of SNA in the regulation of blood flow. Indeed our own previous work (19, 20) criticized the approach, as it can activate 100% of the nerve fibers in the bundle and thus is hardly a natural physiological form of stimulus. However there are several points that make its use valid for studying its effect on RBF under some conditions: first, it provides a pure stimulus, without the conflicting input of altered blood pressure that so often accompanies activation of the sympathetic nervous system; second, it provides a method of introducing step changes in an input, in which these changes contain known frequencies. Naturally occurring SNA contains activity in multiple frequency bands, making it difficult to determine precisely the corner frequency of the vasculature’s response to SNA without resorting to mathematical modelling. Finally, we used modulated sine stimulation, where the amplitude of the stimuli varied between 0 and 10 V. In each animal before the commencement of this stimulus, we applied a period of constant frequency/amplitude stimuli at voltage steps between 0 and 10 V to find the voltage that gave the maximal reduction in RBF. In all cases this was between 8 and 10 V. Thus we are confident that the stimuli applied were not being clipped at the higher voltages and therefore providing nonsinusoidal stimuli to the nerves.
Why should the vasculature have a low-pass filtering/integrating characteristic? With regard to the renal vasculature, the maintenance of adequate glomerular filtration rate and thus urine flow requires RBF to remain relatively stable. Oscillations in RBF caused by SNA rhythmicities >0.5 Hz are likely to diminish this ability. Conversely a system that has no variability, and in which the inputs only adjust the mean level of RBF, is one with reduced controllability. Therefore the low-frequency oscillations in SNA are likely to assist in the dynamic control of RBF, ensuring it responds rapidly and with sufficiently high gain to the stimuli of daily life. The ability to measure the frequency response curve and the resonant characteristics of the vasculature using the simple methodology presented in our study is likely to considerably enhance our ability to examine the properties of the vasculature and the control mechanisms under disease and different therapeutic treatments.
This work was supported by the Auckland Medical Research Foundation, the New Zealand Neurological Foundation, and the Lottery Grants Board of New Zealand. T. A. Hore was the recipient of a Kidney Foundation Somerville-Hansen Memorial studentship.
Address for reprint requests and other correspondence: S. C. Malpas, Dept. of Physiology, Univ. of Auckland Medical School, Private Bag 92019, Auckland, New Zealand (E-mail:).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.
- Copyright © 1999 the American Physiological Society