Regulatory, Integrative and Comparative Physiology

Responses of mesenteric and renal blood flow dynamics to acute denervation in anesthetized rats

Isam Abu-Amarah, David O. Ajikobi, Hélène Bachelard, William A. Cupples, Fred C. Salevsky


Previous studies have shown that renal autoregulation dynamically stabilizes renal blood flow (RBF). The role of renal nerves, particularly of a baroreflex component, in dynamic regulation of RBF remains unclear. The relative roles of autoregulation and mesenteric nerves in dynamic regulation of blood flow in the superior mesenteric artery (MBF) are similarly unclear. In this study, transfer function analysis was used to identify autoregulatory and baroreflex components in the dynamic regulation of RBF and MBF in Wistar rats and young spontaneously hypertensive rats (SHR) anesthetized with isoflurane or halothane. Wistar rats showed effective dynamic autoregulation of both MBF and RBF, as did SHR. Autoregulation was faster in the kidney (0.22 ± 0.01 Hz) than in the gut (0.13 ± 0.01 Hz). In the mesenteric, but not the renal bed, the admittance phase was significantly negative between 0.25 and 0.7 Hz, and the negative phase was abrogated by mesenteric denervation, indicating the presence of an arterial baroreflex. The baroreflex was faster than autoregulation in either bed. The presence of sympathetic effects unrelated to blood pressure was inferred in both vascular beds and appeared to be stronger in the SHR than in the Wistar rats. It is concluded that a physiologically significant baroreflex operates on the mesenteric, but not the renal circulation and that blood flow in both beds is effectively stabilized by autoregulation.

  • transfer function
  • Wistar rats
  • spontaneously hypertensive rats
  • autoregulation
  • baroreflex

in rats, arterial baroreflexes operate largely on peripheral resistance within the frequency band from 0.25 to 0.7 Hz (5,20). Efferent renal sympathetic nerve activity is coherent with blood pressure (PA) in a band centered on ∼0.4 Hz (3) and predicts the observed discrete rhythm in PA at 0.4 Hz (4). Despite this, baroreflexes appear to have relatively small effects on renal blood flow (RBF) in several species (12, 30). This is perhaps not too surprising, since operation of baroreflexes to defend PA must introduce fluctuation of blood flow of the bed being operated on. The kidney has a very strong requirement for stability of RBF that follows both from the necessity for high and stable glomerular filtration rate to permit tubular regulation of sodium excretion and from the sensitivity of the glomerulus to hypertensive injury. In denervated kidneys, it is autoregulation that provides this stability, preventing PA fluctuations slower than ∼0.1 Hz from being transmitted to RBF (19). Thus one would predict that autoregulation should dominate dynamic regulation of RBF in innervated as in denervated kidneys.

Considerably less is known of the dynamic regulation of blood flow to the intestine. As in the kidney, mesenteric sympathetic nerve activity contains a large amount of information related to baroreflexes, although there is also a substantial somatosensory component (37). Blood flow through the superior mesenteric artery (MBF) is subject to autoregulation (26), although both the efficiency and the mechanism of autoregulation in this bed are subject to debate. Both myogenic and washout mechanisms have been proposed to explain experimental data (reviewed in Ref. 21). The requirement for stability of blood flow in this bed is not as strict as in the kidney because organ function is not as tightly linked to blood flow and because much larger structural resistance protects the capillaries from hypertensive damage. Furthermore, the gut is upstream from the portal circulation, a major capacitance bed that is subject to control by sympathetic inputs, in particular baroreflexes (16, 36). Thus it is not clear what control mechanisms dominate dynamic regulation of MBF. Because this bed receives a large fraction of cardiac output and because of the large downstream capacitance, it is likely that baroreflexes contribute substantially to dynamic regulation of MBF.

Experiments were performed in normotensive rats to compare dynamic regulation of MBF and RBF; additional experiments were performed in spontaneously hypertensive rats (SHR) at a time (6–7 wk of age) when efferent renal sympathetic nerve activity is substantially elevated (12, 27). The relative contributions of autoregulation and baroreflexes were assessed from changes in the pressure-flow transfer functions induced by acute denervation of the vascular bed being studied.


All experiments received approval of the SMBD-Jewish General Hospital Animal Care Committee and the McGill University Animal Care Officer and were conducted under the guidelines promulgated by the Canadian Council on Animal Care. Adult, male Wistar rats (288 ± 57 g, mean ± SD) and 6- to 7-wk-old male SHR (160 ± 20 g) were obtained from Charles River (Canada). Rats had free access to food and water until the time of the experiment.

Anesthesia was induced by isoflurane or halothane at a concentration of 5% in inspired gas (40% O2-60% air). The animal was transferred to a servo-controlled, heated table to maintain body temperature at 37°C, intubated, and ventilated by a constant volume, small-animal respirator according to the Harvard nomogram (expts 1,2, and5) or by a pressure-controlled respirator operating in respiratory-assist mode (expts 3, 4, and6). After induction, the anesthetic concentration was reduced to ∼2.5% (isoflurane) or ∼1.5% (halothane). During the postsurgical equilibration period, inspired anesthetic concentration was titrated to the minimal concentration that precluded a PA response when the tail or pinna was pinched. The tail was pinched periodically throughout the experiment to ensure adequate anesthesia. Anesthetics were delivered by using calibrated vaporizers in a Boyle circuit modified for small animals.

Cannulas were placed in a femoral artery (PE-90 with narrowed tip) and vein (PE-50); the venous line contained a Silastic insert to allow drug infusion. In experiments 1,2, and5, the animal was paralyzed with pancuronium bromide (Pavulon, Organon, 1 mg/kg + 1 mg ⋅ kg−1 ⋅ h−1) after placement of the arterial cannula to permit monitoring of PA. A constant infusion delivered 1% of body weight per hour; this infusion continued throughout the experiment and contained 2% charcoal-washed BSA in normal saline. A left subcostal flank incision was used in all experiments to approach the renal or superior mesenteric artery. Inexperiments 4 and5, the left kidney was immobilized in a plastic cup and covered with fat to limit cooling and drying. The flow probe was placed after the minimal possible dissection, with care taken to preserve the nerves supplying the vascular bed.

Six experiments were performed. Four experiments examined the dynamics of MBF in Wistar rats under isoflurane (expts 1 and 2,n = 6 each) or halothane anesthesia (expt 3,n = 7) and in SHR under isoflurane anesthesia (expt 4,n = 8). RBF dynamics were evaluated in Wistar rats (expt 5,n = 7) and in SHR (expt 6,n = 6), both under isoflurane anesthesia. Experiment 1 was a time control designed to confirm that the observed changes in the PA-MBF transfer function were caused by denervation. In the other five experiments, the vascular bed being examined was denervated after an initial control period.

After a 1-h equilibration, PA and RBF or MBF were recorded continuously for ≥20 min. Anesthesia was then returned to a surgical plane, and the flow probe was dismounted. Inexperiment 1, the flow probe was remounted on the superior mesenteric artery after the amount of time typically required for denervation. In experiments 2–4, the superior mesenteric artery and vein were stripped from the aorta to the first arterial bifurcation, and the region was painted with 1% toluidine blue in saline, which stains the nerve fibers (25). All stained fibers were separated, and the region was painted with 5% phenol in ethanol. In experiments 5 and 6, the renal artery and vein were stripped from the aorta to the hilus of the kidney by using the same procedure. After denervation of the bed, the flow probe was remounted, and inspired anesthetic concentration was returned to the level used in the first experimental period. After a 30- to 40-min equilibration, a second record of ≥20 min was acquired.

Femoral arterial pressure was measured by a pressure transducer (HP 1290C) driven by a Stemtech GPA-1 amplifier. Blood flows were measured by a Transonic Systems T106 transit time ultrasound flowmeter (R1 probe). Blood pressure and flowmeter output were recorded on magnetic tape (Vetter 420-K) for off-line analysis. Pressure and flow signals were filtered at 22 Hz and sampled at 48 Hz using 12-bit analog-to-digital conversion. The two data streams were low-pass filtered and subsampled to 3 Hz.

Power spectra and transfer functions based on the fast Fourier transform were computed by using standard algorithms as previously described (1). Data segments of 1,024 s were first subjected to trend removal by using a 500-point rectangular window with 0.006-Hz cutoff. Fast Fourier transforms were computed on 512-point segments shaped by the Hann window with 50% overlap. Transfer functions (magnitude and phase angle) were calculated separately and employed 128 (MBF)- or 256 (RBF)-point segments, the Hann window, and 69% overlap. Fractional admittance gain was calculated as magnitude divided by conductance. Thus gain of >1 means that PAfluctuations are actually amplified into blood flow as expected from passive, compliant vessels or from a baroreflex operating on blood flow to stabilize PA; gain of 1 means that the vasculature behaves as a stiff tube; and gain of <1 means that flow is actively being stabilized, for instance by autoregulation. By convention, a positive admittance phase means that output (blood flow) leads input (PA) as expected in a passive or autoregulating system (18). A baroreflex will produce negative admittance phase because flow follows pressure after a delay governed by delays in nervous transmission and the (de)activation kinetics of vascular smooth muscle.

As noted by Holstein-Rathlou et al. (19), PA is undoubtedly the input to the system under study. However, either blood flow or “instantaneous” resistance may be considered to be the system’s output. Pressure-resistance transfer functions often provide information complementary to pressure-flow transfer functions (19, 23). Instantaneous resistance is calculated as pressure/flow at each 0.33-s sampling interval. Fractional resistance gain is expected to be <1 when the system is behaving passively with respect to PA and ≥1 in regions where regulatory systems operate. Resistance phase approaches −π rad when the system is pressure passive and increases toward 0 rad below the operating frequency of autoregulation (19, 23). In contrast, a baroreflex will have positive resistance phase over the band in which it operates. An obvious caveat is that both active and capacitative changes may contribute to resistance at high frequencies. There are conceptual and experimental reasons to believe that capacitative events in the kidney are small and faster than 1 Hz (5, 19). In the gut, the situation is not as clear cut. After denervation, however, admittance phase in the PA-MBF transfer function was slightly positive (0.12 ± 0.03 rad,n = 21) and inversely related to frequency between 0.25 and 0.7 Hz, suggesting that capacitative changes are small in this region of the spectrum. These observations suggest that resistance changes in the range from 0.01 to 1 Hz are dominated by active events in both vascular beds.

Results are presented as means ± SE of original data. Effects of denervation were assessed by comparing average gain and phase before and after denervation across a frequency band faster than the operating frequency of autoregulation. This band was defined to extend from 0.25 to 0.7 Hz and is referred to as the high-frequency (HF) band. For statistical testing, spectral powers were subjected to logarithmic transformation. Statistical testing was performed by paired or independent t-tests as and where appropriate. P ≤ 0.05 was considered to indicate a significant difference.


Body weights, hematocrit, PA, and blood flows are reported in Table 1. Hematocrits were similar among the six experiments and suggest that the animals were euvolemic. As expected, rats anesthetized with isoflurane had higher PA than those anesthetized with halothane (1). In the time-control experiment, neither PA nor MBF varied fromperiods 1 to2 (Table 1). Mesenteric denervation reduced PA significantly inexperiments 2 and4 and marginally inexperiment 3(P < 0.1), whereas PA was unchanged inexperiment 1 (time control). Renal denervation in experiments 5 and6 had no effect on PA. Blood flow was not significantly altered after denervation in any experiment.

View this table:
Table 1.

Status of animals before and after denervation

In the MBF time-control experiment, neither PA or MBF dynamics changed with time, as shown in Fig. 1. Noteworthy features of the power spectra from this experiment include a discrete oscillation of both PA and MBF at 0.45 ± 0.03 Hz and a broad peak in the MBF spectrum at 0.1–0.15 Hz. The PA-MBF transfer function, shown in Fig. 1, was similar in the two periods. Fractional admittance gain declined from ≥1 above 0.25 Hz to <1 below 0.1 Hz, and the decline was associated with a broad peak in admittance phase. In the HF band, gain was 1.27 ± 0.10 and 1.11 ± 0.11 in the first and second periods, respectively, whereas phase was significantly negative, being −1.29 ± 0.14 and −1.09 ± 0.14 rad in the first and second periods, respectively. Resistance gain and phase are reported in Table2. In both periods, gain was ≥1 and phase was positive in the HF band; neither variable differed significantly between periods.

Fig. 1.

Dynamics from expt 1 [blood flow in superior mesenteric artery (MBF), time control] are presented. In this and subsequent figures, power spectra of blood pressure (PA,A) and MBF (B) are shown attop, and PA-blood flow transfer function, fractional admittance gain (C) and admittance phase (D), is shown below. A: there was a pronounced oscillation of PA centered at 0.45 ± 0.03 Hz that was present in both periods.B: this oscillation plus a broad, autonomous peak associated with autoregulation and centered at 0.1–0.15 Hz are visible in MBF.C: fractional admittance gain was ≥1 in high-frequency (HF) band in both periods. Below 0.2 Hz, gain declined to <1 in both periods. D: admittance phase was negative in HF band in both periods and became positive below 0.2 Hz. Here and in subsequent figures unity gain and 0 phase angle are highlighted by dash-dot lines. Values are means ± SE; n = 6.

View this table:
Table 2.

Effects of denervation on pressure-resistance transfer functions

The results acquired in experiment 2(Wistar, MBF, isoflurane) are shown in Fig.2. MBF dynamics before mesenteric denervation were very similar to those from the time-control study. There was a peak in PA spectral power at 0.47 ± 0.01 Hz; MBF power showed this feature plus a broad autonomous peak at 0.1–0.15 Hz. Fractional admittance gain declined from 1.04 ± 0.16 in the HF band to <1 below 0.1 Hz, and this decline was associated with a broad peak in admittance phase. In the HF band, admittance phase was significantly negative (−1.32 ± 0.23 rad). Mesenteric denervation reduced integrated PA spectral power at ≥0.25 Hz (P < 0.05) due to removal of the 0.47-Hz peak. The equivalent effect on MBF power was not significant. The reduction of gain and phase peak below 0.15 Hz were not affected by denervation. However, in the HF band, admittance gain was increased to 1.79 ± 0.07 (P < 0.05) and phase to 0.16 ± 0.07 rad (P < 0.01). The PA-resistance transfer function is shown in Fig. 2, E andF. Resistance gain and phase in the HF band are also reported in Table 2. Resistance gain in the HF band was not different from 1 either before or after denervation. Before denervation, resistance phase was positive above 0.2 Hz (i.e., resistance led PA), whereas after denervation, resistance phase was strongly negative and followed PA in this region of the spectrum. In the HF band, this change was highly significant (P ≤ 0.01).

Fig. 2.

Dynamics from expt 2 (MBF, Wistar, isoflurane) are presented. A: after mesenteric denervation, pronounced oscillation of PA at 0.47 ± 0.01 Hz was abrogated. B: equivalent reduction in MBF was obscured by variability. Autonomous oscillation of MBF at 0.1–0.15 Hz was not affected by denervation.C: there was low admittance gain in HF band before denervation and reduction of gain to <1 below 0.1 Hz. Denervation increased admittance gain (P < 0.05) in HF band but did not affect gain reduction below 0.1 Hz.D: admittance phase was negative in HF band and became positive at ≤0.1 Hz. After denervation, phase angle in HF band became slightly positive (P < 0.01 vs. innervated), whereas lower frequency rise of admittance phase was not affected. PA-resistance transfer function (E andF) showed resistance gain ≥1 and positive resistance phase in HF band before denervation, indicating that strong resistance adjustments preceded PA fluctuations in this band.n = 6.

Experiment 3 (Wistar, MBF, halothane) was performed with halothane because this anesthetic is known to inhibit sympathetic activity at several sites (7, 33), whereas sympathetic nerve activity is relatively preserved under isoflurane anesthesia, which was used for all other experiments. This inhibition is reflected in the lack of PAspectral power above 0.25 Hz in experiment 3 compared with experiment 1 (P < 0.05). In particular, no discrete PAoscillation at or about 0.45 Hz was visible, and mesenteric denervation had no effect on PA or MBF spectral power in the region ≥0.25 Hz (Fig. 3). Fractional admittance gain declined from 1.61 ± 0.26 in the HF band to <1 below 0.1 Hz, and this decline was associated with a broad peak in admittance phase. In the HF band, admittance phase was significantly negative (−1.09 ± 0.29 rad). Mesenteric denervation did not affect the reduction of gain or the phase peak below 0.15 Hz, nor did it affect gain in the HF band (1.56 ± 0.19). However, in this band admittance phase was increased by denervation to 0.09 ± 0.06 rad (P < 0.01). Resistance gain and phase in the HF band are reported in Table 2. Resistance gain tended to decline after denervation (P < 0.1), and denervation changed resistance phase from positive to negative (P < 0.01).

Six- to seven-week-old SHR were employed in experiment 4 (SHR, MBF, isoflurane) because they display increased sympathetic activity (27), particularly during the period from 6 to 8 wk, during which they become hypertensive (12). Dynamics of PA and MBF from this experiment are shown in Fig. 4. No discrete 0.45 Hz PA oscillation was evident, although there was a prominent shoulder in the PA spectrum (cf. Figs.3 A and4 A). Mesenteric denervation significantly increased PA and MBF integrated spectral power at ≥0.25 Hz (both P < 0.01). Fractional admittance gain declined from 1.21 ± 0.09 in the HF band to <1 below 0.1 Hz, and this decline was associated with a broad peak in admittance phase. In the HF band, admittance phase was <0, being −0.31 ± 0.13 rad (P = 0.05). Mesenteric denervation did not affect the reduction of gain or the phase peak below 0.15 Hz, but increased admittance gain in the HF band to 1.81 ± 0.11 (P < 0.01) and admittance phase to 0.15 ± 0.02 (P< 0.05). Resistance gain and phase in the HF band are reported in Table 2. Resistance gain was <1 both before and after denervation, whereas resistance phase was negative and unchanged by denervation. Before denervation, resistance phase in the HF band varied from 1.37 to −4.05 rad in individual animals.

Fig. 3.

Dynamics from expt 3 (MBF, Wistar, halothane) are presented. There was no discrete oscillation of PA(A) or MBF (B) at 0.45 Hz, nor did mesenteric denervation affect PA or MBF spectral power. C: fractional admittance gain in HF band was unchanged after denervation.D: admittance phase in HF band was negative before denervation and not different from 0 after denervation (P < 0.01). Denervation did not affect low-frequency reduction of gain and rise of phase.n = 7.

Fig. 4.

Dynamics from expt 4 (MBF, SHR, isoflurane) are presented. Unlike Wistar rats presented in Figs. 1, 2, and 5, these young SHR did not display a discrete 0.45-Hz oscillation of PA(A) or MBF (B). Mesenteric denervation increased integrated spectral power above 0.25 Hz of PA and MBF (bothP < 0.01). Mesenteric denervation increased admittance gain in HF band (C; P< 0.01) and increased admittance phase (D;P < 0.05) in HF band. Denervation did not affect low-frequency reduction of gain to <1 or associated phase peak. n = 8.

RBF dynamics were studied before and after renal denervation in Wistar rats (expt 5) and young SHR (expt 6) anesthetized with isoflurane. Results from experiment 5are presented in Fig. 5. There was the expected discrete peak in PAspectral power centered at 0.45 ± 0.02 Hz, and this peak was unaltered after renal denervation, as was integrated PA spectral power at ≥0.25 Hz. Fractional admittance gain displayed a prominent resonance peak at 0.21 ± 0.02 Hz, below which it declined to <1 associated with a broad phase peak. This autoregulatory signature was not altered after denervation. In the HF band, denervation increased admittance gain from 1.01 ± 0.12 to 1.60 ± 0.17 (P< 0.05). Admittance phase in this band was 0.08 ± 0.09 rad before denervation and was unchanged at 0.23 ± 0.05 rad after denervation. The PA-resistance transfer function from this experiment is also shown in Fig. 5,E andF. In the HF band, resistance gain was considerably <1 both before and after denervation (Table 2), indicating that PA fluctuations in this band were associated with only small resistance changes. Resistance phase was strongly negative above 0.3 Hz before and after denervation. The apparent difference (P < 0.1) arises from one animal that showed positive resistance phase during the innervated period.

Fig. 5.

Dynamics from expt 5 (RBF, Wistar, isoflurane) are presented. There was a discrete oscillation of PA(A) and RBF (B) at 0.45 ± 0.02 Hz that was not affected by renal denervation. Denervation increased admittance gain in HF band (C;P < 0.05) but did not affect admittance phase (D) in this band. Renal denervation did not affect resonance peak in admittance gain at 0.2 Hz and subsequent decline to <1 or associated peak of admittance phase. PA-resistance transfer function is shown in E andF. In HF band, resistance gain was considerably <1 and resistance phase was strongly negative, indicating that in this region of spectrum PA fluctuations preceded small resistance changes. n = 7.

Results from young SHR (expt 6) are presented in Fig. 6. As in the other SHR experiment (expt 4), the PA and RBF spectra conspicuously lacked a discrete peak at ∼0.45 Hz, although there were prominent shoulders in both. Denervation increased PA and RBF integrated spectral power at ≥0.25 Hz (both P < 0.01). Renal denervation increased admittance gain in the HF band from 1.39 ± 0.07 to 1.86 ± 0.07 (P < 0.05) but did not affect admittance phase in this region of the spectrum (0.14 ± 0.04 rad before and 0.22 ± 0.01 rad after denervation). Resistance gain and phase in the HF band are reported in Table 2. Resistance gain was <1 both before and after denervation, whereas resistance phase was strongly negative and unchanged by denervation.

Fig. 6.

Dynamics from expt 6 (RBF, SHR, isoflurane) are presented. Unlike Wistar rats presented in Figs. 1, 2, and 5 these young SHR, like those shown in Fig. 4, did not display any discrete 0.45 Hz oscillation of PA(A) or RBF (B). Renal denervation increased integrated spectral power at ≥0.25 Hz in PA and RBF (bothP < 0.01). Denervation increased admittance gain in HF band (C;P < 0.05) but did not affect admittance phase (D) in same band. Renal denervation enhanced resonance peak of admittance gain at 0.2 Hz but did not affect subsequent decline of gain or associated peak of phase. n = 6.

Examination of pooled results revealed that before denervation SHR (expts 4 and6) had considerably greater total PA spectral power than did Wistar rats anesthetized with isoflurane (expts 2 and 5), integrated total power being 2.57 ± 0.40 mmHg2/Hz in Wistar rats and 12.40 ± 2.74 mmHg2/Hz in SHR (P ≤ 0.01). In both strains of rats, there was also a significant difference between MBF and RBF with respect to autoregulation. The resonance peak in gain of autoregulation can be used as an index of the system’s natural frequency (18). The resonance peak of mesenteric autoregulation occurred at 0.13 ± 0.01 after denervation in each of experiments 2–4. These values were taken from 256-point transfer functions. In contrast, the resonance peak in gain of renal autoregulation occurred at 0.21 ± 0.02 and 0.23 ± 0.01 Hz inexperiments 5 and6, respectively, after denervation. When the values from the three MBF and two RBF experiments were pooled, there was a highly significant difference between the two beds (P ≪ 0.01).


These experiments were performed to assess the dynamic regulation of blood flow to two major vascular beds, the kidney and the gut. These beds have in common that they receive large fractions of cardiac output and that their arteries originate close together on the abdominal aorta. However, they have widely disparate structure and physiology. Consequently, one might expect them to respond differently to spontaneous PA fluctuations and to contribute differently to overall cardiovascular regulation. The results show strong autoregulation of blood flow in both beds as well as significant, though different, centrally mediated (i.e., sympathetic) dynamic control elements.

Two strains of rat and two anesthetics were used to generate three levels of sympathetic activity. SHR have elevated renal sympathetic nerve activity compared with Wistar rats (12), especially during the period from 6 to 8 wk of age during which they become markedly hypertensive (27). Sympathetic vasomotor activity generates most PA spectral power at frequencies >0.1 Hz (5, 31) and is largely related to operation of arterial baroreflexes (5, 20), consistent with fivefold higher integrated PA spectral power at ≥0.25 Hz in young SHR than in Wistar rats. Halothane is known to inhibit sympathetic output at several sites (7, 33), and we have previously observed that changing from isoflurane to halothane anesthesia within an experiment reduced PA spectral power at frequencies >0.1 Hz (10). Similarly, in the present study, there was less integrated PAspectral power above 0.25 Hz (P < 0.05) in rats anesthetized with halothane (expt 3) than in identically prepared rats anesthetized with isoflurane (expt 1). In addition, a time-control study (expt 1) was included to ensure that observed changes in MBF dynamics resulted from denervation. It was not necessary to include a similar control for the effects of denervation on RBF dynamics because the postdenervation renal transfer functions are well described for both normotensive rats (1, 10, 18, 19, 23, 32) and SHR (6, 11).

Mesenteric, but not renal, denervation significantly reduced PA. This response was not associated with significant changes in MBF as shown in Table 1 or with any consistent change in heart rate (data not shown). Most likely, it was a consequence of the well-known regulation of portal capacitance by sympathetic inputs (16, 36). Thus removal of adrenergic tone by mesenteric denervation increased portal capacitance, reducing cardiac preload. Consistent with this interpretation is the observation that PA was reduced by −22 ± 3 mmHg in SHR (P < 0.01), by −14 ± 5 mmHg in Wistar rats anesthetized with isoflurane (P < 0.05), and by only −3.3 ± 1.5 mmHg in Wistar rats anesthetized with halothane (P < 0.1). Thus the reduction of PA varied in parallel to the predicted sympathetic activity.

Transfer functions from both vascular beds, acquired after denervation, show only one active system. This was undoubtedly autoregulation. Before denervation, particularly in the gut, two or more active systems were apparent. The additional control systems presumably reflect ongoing sympathetic activity. It is relatively simple to predict that an arterial baroreflex operating on a vascular bed to stabilize PA will have negative admittance phase because flow follows pressure after a delay governed by transmission time and that admittance gain should be high because flow is altered to regulate pressure. However, neural inputs unrelated to PA are not defined in the transfer function, although they may be expected to affect the observed transfer function in both PA-independent and -dependent fashions. Because there is no characteristic temporal relationship between “other” sympathetic nerve activity and PA, phase may be expected to trend to 0. Flow fluctuations unrelated to PA would be observed as increased gain. Sympathetic activity may also reduce arterial compliance and thus reduce the passive, PA-induced fluctuation of blood flow (2). This would be seen as reduction of admittance gain. Thus there is no clear interpretation as to how such inputs would affect the observed admittance gain. In this situation admittance phase is therefore more informative than admittance gain.


Transfer functions from all six experiments show the presence of dynamic autoregulation (1, 6, 10, 11, 18, 19, 32). This signature was present before and after denervation, indicating that autoregulation was not overridden by central mechanisms. The autoregulatory signatures in the two beds (e.g., Figs. 2 and 5) are very similar to that predicted for a myogenic system (18), suggesting that dynamic autoregulation was dominated by a myogenic mechanism in both renal and mesenteric vascular beds. With respect to the kidney, this conclusion is consistent with the results of previous studies, performed in rats (1, 11) and dogs (23), in which tubuloglomerular feedback was explicitly excluded by physical or pharmacological blockade and in isolated perfused hydronephrotic kidneys that lack tubuloglomerular feedback (9).

In both beds, autoregulation substantially stabilized blood flow in the face of PA fluctuation. Pooled results indicate that, after denervation, fractional admittance gain at 0.01 Hz was 0.29 ± 0.04 in the gut and 0.28 ± 0.04 in the kidney. Thus dynamic autoregulation was as efficient in the gut as in the kidney. Although efficient renal autoregulation was expected, the gut is reputed to show weak autoregulation (e.g., Ref. 17).

We are not aware of previous studies that address dynamic autoregulation by the mesenteric circulation. With respect to steady-state autoregulation (i.e., MBF responses to imposed and sustained PA steps), data in the literature are rather variable. Some reports show substantial autoregulatory efficiency (e.g., Refs. 29, 34), whereas others show considerably less efficient autoregulation (e.g., Refs. 15, 21). A number of factors could contribute to this divergence, including the presence or absence of chyme in the lumen (34) and the species studied (15, 26, 29). We cannot explicitly exclude food as a variable, since our rats were studied during the day, their inactive period, although the lumen of the small bowel typically contained chyme. Arguing by analogy to autoregulation of RBF, we would suggest that species differences are inconsequential. Steady-state autoregulation of RBF is similar in dogs (17) and rats (8), and RBF dynamics are very similar in dogs (23) and rats (19).

There was one major difference between the two beds. Autoregulation was significantly slower in the gut (0.13 ± 0.01 Hz) than in the kidney (0.22 ± 0.01 Hz). The data do not address the mechanism(s) behind this difference. There are, however, several important functional consequences. First, the gut and two kidneys together receive a large portion of cardiac output. If they were to autoregulate at the same frequency, there would be considerable potential for phase locking, which could result in destabilizing oscillation of PA and of blood flows to all three organs. In this light, it is noteworthy that the operating frequencies in the two beds are linked to each other in a ratio of ∼5 to 3, thus minimizing the likelihood of phase locking (13). Second, in conscious animals, frequent sudden falls of PA are surprisingly common and are often associated with onset of activity (5, 31). Because the kidney responds faster than the gut, it may well be better protected from such rapid transients.

Sympathetic influences.

Before denervation both the kidney and the gut displayed active events at ≥0.25 Hz, indicating the presence of significant sympathetic inputs in this region of the spectrum. These events were specific to the different vascular beds. In the gut, the admittance phase was significantly negative, whereas resistance phase was >0 rad, implying the presence of an active controller related to PA and acting on resistance, i.e., a baroreflex. This result was very clear in the Wistar rats but less clear in the young SHR. It is not possible to determine from the present data whether a functional baroreflex in these SHR was being swamped by unmodulated sympathetic activity or whether the baroreflex itself was impaired. The kidneys of Wistar rats and SHR did not show the signature of a baroreflex. Instead, admittance phase was 0 or slightly positive, whereas resistance phase was strongly negative, both characteristic of pressure-passive behavior. Furthermore, mesenteric denervation in experiment 2 abrogated the 0.47-Hz PA rhythm, whereas renal denervation in experiment 5 did not. The simplest interpretation of these findings is that there is a physiologically significant baroreflex operating on MBF, but not RBF, to stabilize PA. Because mesenteric denervation abrogated the 0.47-Hz PA rhythm, it is tempting to surmise that the mesenteric circulation is a predominant site of baroreflex operation.

The renal circulation of the young SHR used inexperiment 6 behaved very similarly to that of the Wistar rats. In both experiments, denervation increased admittance gain in the HF band but did not alter admittance phase. Very similar results have recently been reported from conscious rabbits studied intact and after chronic renal denervation (28) and from conscious dogs before and during ganglionic blockade (23). In neither case did the PA-RBF transfer function before denervation show evidence of a baroreflex. It is perhaps not surprising that baroreflexes operate preferentially on the mesenteric and not the renal circulation. The mesenteric bed is in series with the hepatic portal circulation, which is a major capacitance bed (36). Additionally, the capacitance of the portal circulation is subject to considerable sympathetic and, in particular, baroreflex regulation (16, 36). In contrast, the kidney is a low-capacitance bed (8, 19), so that rapid fluctuations of RBF would be transmitted relatively unfiltered to venous return.

There appeared to be sympathetic influence on blood flow that was unrelated to PA in both renal and mesenteric circulations. This is readily seen for the kidney, in which, with intact innervation, admittance phase in the HF band was ∼0, a value not consistent with operation of a physiologically significant baroreflex (see above) and unchanged by denervation. However, denervation significantly increased admittance gain in this band to values consistent with passive dynamics (i.e., gain ∼1.6–2). Thus one can infer that neural inputs unrelated to PA affected admittance gain in the HF band. Similarly, in the mesenteric circulation, admittance gain in the HF band before denervation did not appear to be consistent with operation of only a baroreflex. As in the kidney, denervation increased admittance gain in the HF band consistent with passive dynamics. Again, one can infer that neural inputs unrelated to PA affected admittance gain. A parsimonious explanation of this effect would be that sympathetic nerve activity reduces arterial and arteriolar compliance, thus limiting passive, PA-induced fluctuation of blood flow (2).

The young SHR employed in experiments 4 and 6 displayed several significant differences compared with the Wistar rats used in other experiments. They had elevated PA spectral power at ≥0.25 Hz, and this power appeared to be modulated in a different manner: no discrete 0.45-Hz oscillation of PAcould be identified. Unlike the Wistar rats, denervation of either bed increased PA and blood flow spectral power (Figs. 4 and 6). In the HF band, admittance phase was only modestly negative (Fig. 4), whereas resistance phase was highly variable (Table 2). In fact, only two of the eight rats studied showed clear baroreflex signatures in the PA-MBF and PA-resistance transfer functions. In the absence of nerve traffic analysis, no definitive conclusions can be drawn. However, the overall pattern is suggestive of increased efferent sympathetic nerve activity that is not effectively modulated by baroreflexes. The data suggest that elevated sympathetic activity reduced compliance and that its removal permitted a strong increase in PA-passive blood flow fluctuation which was then transmitted to PA. It is known that both beds are sufficiently large, in terms of blood flow, so that blood flow fluctuations can generate corresponding PA fluctuations (22, 35).


Positive controls to ensure intact innervation were not performed; instead, a very extensive denervation procedure was employed. A much less aggressive procedure is normally used in this laboratory and reliably reproduces the features of the PA-RBF transfer function seen after denervation (1, 10; unpublished results). We are thus confident of the effectiveness of the denervation procedure employed.

Spontaneous PA fluctuation was used as the input signal to examine dynamic regulation of RBF and MBF. Because the amount of PAfluctuation is reduced in anesthetized as opposed to conscious rats and is certainly less than when PA is externally forced, the apparent lack of an arterial baroreflex affecting RBF dynamics could reflect insufficient input power. This explanation is unlikely because strong stimuli caused only minor baroreflex modulation of RBF in steady-state experiments (30) and because the input power was similar in the MBF experiments, which did disclose an arterial baroreflex, and the RBF experiment, which did not. The identification of an arterial baroreflex rests on the predicted transfer function, the central origin of the negative admittance phase (and positive resistance phase), the known origin of the 0.47-Hz rhythm in PA, and its generation in the mesenteric circulation.

Because the control systems studied were operating in closed-loop mode, it is possible that the open-loop analysis used may not capture their dynamics accurately (24). However, we would note the following:1) the analysis captures autoregulation dynamics in both beds (observed dynamics are consistent with model-based predictions and with a large body of open- and closed-loop studies); 2) the observed central effects on RBF dynamics are consistent with what is known from steady-state studies; and3) conclusions drawn from the PA-MBF transfer function are consistent with inferences drawn from the power spectra, which are not subject to this constraint.


These experiments assessed three major events that actively alter vascular conductance in response to PA fluctuations. The arterial baroreflex operates on vascular admittance to stabilize PA, whereas renal and mesenteric autoregulation operate on admittance to stabilize regional blood flows. Thus the baroreflex must be separated from the autoregulatory systems to avoid potentially destructive interactions among them. This is accomplished by the baroreflex operating faster than autoregulation, i.e., separation in time. Equally, if autoregulation in two major beds (intestine and 2 kidneys) were to operate at the same frequency, there would be considerable potential for entrainment and subsequent resonance in MBF and RBF and in PA. Overall, the faster autoregulation in the kidney will tend to minimize transmission of baroreflex-induced PA fluctuations to the glomerulus, whereas the slower autoregulation in the gut will tend to minimize interactions between the baroreflex and autoregulation in this bed.


This work was funded by grants to W. A. Cupples from the Heart and Stroke Foundation of Quebec, Kidney Foundation of Canada, and Medical Research Council of Canada.


  • Address for reprint requests: W. A. Cupples, Lady Davis Institute, SMBD-Jewish General Hospital, 3755 Cote St.-Catherine Rd., Montreal, QC, Canada H3T 1E2.

  • 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.


View Abstract