Resetting of sympathetic rhythm by somatic afferents causes post-reflex coordination of sympathetic activity in rat


Corresponding author M. P. Gilbey: Department of Physiology, Royal Free and University College Medical School, University College London, London NW3 2PF, UK. Email:


  • We have proposed previously that graded synchronous activity is produced by periodic inputs acting on weakly coupled or uncoupled oscillators influencing the discharges of a population of cutaneous vasoconstrictor sympathetic postganglionic neurones (PGNs) in anaesthetized rats.

  • Here we investigated the effects of somatic afferent (superficial radial nerve, RaN) stimulation, on the rhythmic discharges of this population. We recorded (1) at the population level from the ventral collector nerve and (2) from single PGNs focally from the caudal ventral artery of the tail.

  • Following RaN stimulation we observed an excitatory response followed by a period of reduced discharge and subsequent rhythmical discharges seemingly phase-locked to the stimulus.

  • We suggest that the rhythmical discharges following the initial excitatory response (conventional reflex) result from a resetting of sympathetic rhythm generators such that rhythmic PGN activity is synchronized transiently. We also demonstrate that a natural mechanical stimulus can produce a similar pattern of response.

  • Our results support the idea that in sympathetic control, resetting of multiple oscillators driving the rhythmic discharges of a population of PGNs may provide a mechanism for producing a sustained and coordinated response to somatic input.

We have proposed previously that a family of oscillators contribute to the bursting pattern of vasomotor sympathetic activity regulating the caudal ventral artery (CVA) of the rat tail (Chang et al. 1999, 2000). This organization, we suggest, allows for linear and non-linear dynamic interactions between sympathetic networks and various inputs and may be important in synchrony coding (see Gilbey, 2001).

Somatic afferent inputs evoke sympathetic responses (somatosympathetic reflexes: e.g. Sato et al. 1965; Coote & Downman, 1966; Coote & Perez-Gonzalez, 1970; Schmidt & Schönfuss, 1970; Morrison & Reis, 1989; Kimura et al. 1996), where sympathetic nerves exhibit intense discharges followed by post-excitatory depression (see Sato et al. 1997, for review). In this study we focused on the impact of somatic input on the synchrony of the rhythmic discharges of postganglionic neurones (PGNs) in the population innervating the CVA. We consider the findings are potentially of great importance to our thinking on sympathetic control as quasi-rhythmic discharges (ongoing bursting activity) are a robust characteristic of most sympathetic outflows (McAllen & Malpas, 1997).

In the rat tail ‘system’ most individual PGNs innervating the CVA exhibit a dominant rhythm in their discharge (≈0.4-1.2 Hz; Johnson & Gilbey, 1996). Discharges of single PGNs can be recorded focally from the surface of the CVA, and population activity from a ventral collector nerve (VCN). Our previous work has shown that the rhythmic discharges of the PGN population can be synchronized by inputs such as central respiratory drive (CRD; Chang et al. 1999) and activity related to lung inflation cycles (Chang et al. 2000). It has also been demonstrated that afferent inputs arising from the aortic nerve can reset the rhythmic discharge of single PGNs (Johnson & Gilbey, 1998). This latter observation indicated to us that somatic afferent inputs might cause a similar synchronization of PGN rhythmic discharges. An alternative possibility was that somatic inputs might act downstream from the point of rhythm generation.

We found that following radial nerve (RaN) stimulation the dominant sympathetic rhythm is reset following the reflex discharge. This was observed at both the single PGN and population level. We also demonstrate that a natural mechanical stimulus can produce a similar pattern of response. We argue that this post-reflex transient synchronization of population discharge may provide a mechanism for producing a coordinated sympathetic response to somatic afferent input. We suggest that a similar mechanism may underlie the sustained bursting configuration of discharges seen in the recordings from cutaneous sympathetic nerves in humans following and during various manoeuvres (e.g. see Delius et al. 1972).

Part of this work has been published previously in abstract form (Staras et al. 1999a,b).


General preparation

Experiments were carried out on 19 Sprague-Dawley rats (220-300 g), anaesthetized with sodium pentobarbitone (60 mg kg−1, i.p.), with supplemental doses of α-chloralose (5-10 mg, i.v.) when required, in accordance with the Animals (Scientific Procedures) Act, 1986. At the end of the experiment, animals were killed by an overdose of sodium pentobarbitone (i.v.). A femoral artery and vein were cannulated to monitor blood pressure and for drug delivery, respectively. The trachea was cannulated and animals were ventilated artificially with oxygen-enriched room air. The urinary bladder was cannulated to allow the free passage of urine. Body (oesophageal) temperature was monitored and maintained at 36.5-37.0 °C. Arterial blood gases and pH were monitored. A solution of sodium bicarbonate (1 m) was administered to counter metabolic acidosis.

In experiments where sympathetic nerve recordings were made (n = 29 from 17 animals), animals were vagotomized and pneumothoracotomized. Phrenic nerve (PN) activity was recorded as a monitor of CRD (Johnson & Gilbey, 1994; Chang et al. 1999). Silver wire electrodes were placed on the left superficia RaN.

In experiments where recordings were made from the VCN (n = 5 from 5 animals), the cauda equina was cut at the L5 level (Smith & Gilbey, 1998). Monophasic activity was recorded differentially from the nerve (see Chang et al. 1999).

For experiments where recordings were made from single PGNs innervating the CVA (n = 19 different PGNs from 12 animals), the exposure of the vessel was performed as described previously (e.g. Johnson & Gilbey, 1994). A glass microelectrode (internal tip diameter, ≈20-80 μm), filled with Krebs solution, was placed on the artery surface. In this way single sympathetic fibres which course over the surface of the artery were recorded (see Johnson & Gilbey, 1994, 1996; Chang et al. 1999). Single units were identified by their characteristic waveform and amplitude.

Data handling

Nerve activity was amplified and filtered (bandpass 300 Hz to 1 kHz). VCN and PN activity were rectified and smoothed with an integrator (time constant: VCN, 0.1 s; PN, 0.2 s). The blood pressure (BP), tracheal pressure (TP), smoothed PN and VCN activity and single PGN activity were also converted into digital signals (sample rate: PN, BP and TP, 100 Hz; PGN, 13.3 kHz) for computer analysis. Transistor-transistor logic (TTL) pulses were generated from single PGN action potentials and the rising phase of PN activity. RaN and sham stimulations were event-marked with a TTL pulse. The BP, TP and PN digital signals and the TTL events for PN, PGN and RaN were sent to a computer for analysis and storage.

Experimental protocol

During a recording period, animals were injected with a neuromuscular blocking agent (gallamine triethiodide, 16 mg kg−1 h−1, i.v.) and the depth of anaesthesia was assessed by the stability of the BP and PN activity. Before and/or after a recording period a blood gas sample (75 μl) was taken. Activation of both Aδ- and C-fibres in RaN was ensured by using a suprathreshold stimulus (0.8-1.5 mA, 0.5 ms pulse; see Adachi et al. 1992 for similar protocol), established in two animals by placing recording electrodes ≈20 mm proximal to the stimulating electrodes to record the evoked C-fibre volley (identity determined by its conduction velocity). Whole-nerve recording experiments (n = 5) were carried out in the absence of CRD to remove the influence of CRD on sympathetic responses. The absence of CRD was achieved by increasing O2 to 50-90 % of the inflow and on some occasions increasing ventilation rate. A control (300 s) and an experimental (300 s) period were performed for each recording, with both experimental-control (n = 2) or control-experimental (n = 3) orders used to ensure no temporal bias. For RaN stimulation with single PGN recordings, animals were recorded in one of two respiratory states: absence of CRD (n = 14 PGNs from 7 animals, see earlier) or normal CRD (n = 4 PGNs from 4 animals). One experimental and control recording period was performed (stimulations every 7 s), and both experimental-control (n = 13 single PGNs) and control- experimental (n = 5 single PGNs) orders were used, to ensure no temporal bias. In four animals the effects of a paw pinch on the discharge pattern of six single units was investigated. These experiments were performed in the absence of CRD. Twelve to eighteen paw pinches were applied at random times within a 300 s period. In a further 300 s recording period no stimulus was given and sham event markers were added after the experiment at the same intervals at which paw pinch was applied in the experimental data set.

Data analysis for whole-nerve activity

The spectral analysis method used to assess rhythmical components in the VCN and PN activity has been described in a previous publication (Chang et al. 1999). Briefly, each 300 s data set of integrated nerve activity (sampled at 100 Hz) was divided into half-overlapped subsections and the linear trend was removed from each (this procedure affects only components < 0.2 Hz). The Welch method was used to generate an average autospectrum from these subsections (size of fast Fourier transformation, 2048). The frequency of the peak with the highest power was taken to represent the dominant frequency of the nerve activity. For examining the relationship between the RaN stimulation (either real or sham) event and the whole-nerve activity, stimulus-triggered averaged waveforms were generated (≈42 triggers).

Data analysis for single PGN experiments

For PGN and PN activity, autocorrelograms were generated to reveal the dominant rhythmicity and the dominant frequency was determined by transforming the event train into an event count time series (counts every 10 ms) and performing spectral analysis on this time series (Rosenberg et al. 1989). For examining the relationship between the RaN stimulation (either real or sham) event and the PGN activity, stimulus-triggered cross-correlograms were generated (≈86 triggers). Rhythmicity in these cross-correlograms was determined by smoothing the correlogram (9-point simple moving average) and measuring the period between the prominent post-stimulus peaks. This allowed a comparison of the frequency of the dominant rhythm and the post-stimulus rhythm to be made. For analysis purposes the dominant rhythm for each PGN was determined from the sham stimulation control data set. Cross-correlograms were also used to examine RaN-PN relationships. Stimulus-triggered raster plots were used to examine the stability of the stimulation events and PGN activity across time. The mean discharge rate for single PGNs in RaN stimulation or sham stimulation conditions was calculated by counting spikes using computer software and dividing the total by the data acquisition period (600 s).

Statistical analysis

All data are presented as medians and 1st and 3rd interquartile ranges (IQR). The Kruskal-Wallis test was used to make multiple comparisons between data sets. Comparisons between paired data sets were performed using the Wilcoxon sign rank test, and for unpaired data sets the Mann-Whitney U test was used. Correlation analysis was performed using the Spearman Rank method. Standard linear regression was used to generate best-fit lines for data sets. P < 0.05 was used as an indicator of statistical significance.


Conditions of animals and influence of RaN stimulation on BP

In the four groups mean BP was not significantly different (note n values are based on the number of experimental runs so dual recordings were counted once only): absence of CRD, whole nerve, 104 mmHg (IQR 87-116 mmHg, n = 5); absence of CRD, single PGN, 83 mmHg (IQR 81-89 mmHg, n = 9); absence of CRD, single PGN with mechanical stimulus, 89 mmHg (IQR 88-89 mmHg, n = 6); single PGN with normal CRD, 99 mmHg (IQR 93- 101 mmHg, n = 4). Arterial PCO2 (Pa,CO2) was significantly lower in the absence of CRD groups than in the normal CRD group (P = 0.05; 31 mmHg, IQR 24-33 mmHg; 34 mmHg, IQR 30-36 mmHg; 29 mmHg, IQR 28-30 mmHg; 42 mmHg, IQR 40-46 mmHg, respectively). Arterial PO2 (Pa,O2) was significantly higher in the absence of CRD groups than in the normal CRD group (P < 0.02; 283 mmHg, IQR 226-302 mmHg; 207 mmHg, IQR 181-232 mmHg; 253 mmHg, IQR 230-327 mmHg; 128 mmHg, IQR 125-131 mmHg, respectively). RaN stimulation evoked a small pressor response (median responses: whole nerve, 7.8 mmHg, IQR 6.0-9.2 mmHg, n = 5; single PGN in the absence of CRD, 4.7 mmHg, IQR 4.1-6.9 mmHg, n = 8; single PGN with normal CRD, 4.8 mmHg, IQR 3.7- 6.1 mmHg, n = 4; no significant difference between groups).

Effect of single shock RaN stimulation on whole-nerve VCN activity

As documented previously (e.g. Chang et al. 1999, 2000), VCN activity (n = 5) showed a dominant rhythmicity revealed by a prominent peak in VCN autospectra (Fig. 1Aa and Ba). The frequency of this rhythm was not influenced significantly by RaN stimulation (RaN stimulation median frequency, 0.73 Hz, IQR 0.73-0.73 Hz; sham RaN frequency, 0.73 Hz, IQR 0.59-0.73 Hz).

Figure 1.

Response of VCN to RaN stimulation and sham stimulation in the absence of CRD

A, RaN stimulation. B, sham stimulation. Aa and Ba, autospectra of VCN activity. Under the two conditions there is a similar broad peak (0.73 Hz) in the relative power density (RPD). Ab and Bb, stimulus-triggered waveform-averaged responses (42 triggers) of smoothed and rectified VCN activity. The filled circle indicates an artefact of the stimulation. ‘1’ indicates the short-latency component of the response. ‘2′ and ‘3′ are burst discharges. Ac and Bc, stimulus-triggered waveform-averaged responses of smoothed and rectified PN activity. Neither RaN stimulation (Ac) nor sham stimulation (Bc) has a significant influence on CRD. For comparison, the averaged waveform of full-inspiratory-related bursts is shown in the middle panel. C, three sample sweeps of smoothed and rectified VCN activity triggered by RaN stimulation.

RaN stimulation evoked a characteristic pattern of VCN activity (Fig. 1Ab and C). The primary response (marked ‘1’ in Fig. 1Ab) had a median latency of 0.372 s (IQR 0.372-0.383 s, n = 5). Such responses were followed, after a period of reduced activity, by further discharge peaks (‘2’ and ‘3’ in Fig. 1Ab). These peaks had median latencies of 1.257 s (IQR 1.189-1.257 s, n = 5) and 2.755 s (IQR 2.699-2.812 s, n = 5). By contrast, sham stimulation-triggered averages were flat (Fig. 1Bb). Three examples of VCN real-time neurograms in response to RaN stimulation are shown in Fig. 1C. RaN stimulation had little effect on PN activity (Fig. 1Ac and Bc).

Effect of single-shock RaN stimulation on single PGN activity

All PGNs recorded (n = 18, 11 animals) showed a rhythmical discharge with frequencies that were unaffected by RaN stimulation and CRD (median frequency in the absence of CRD: RaN stimulation, 0.63 Hz, IQR 0.59-0.73 Hz; sham, 0.61 Hz, IQR 0.59-0.73 Hz, n = 14; normal CRD: RaN stimulation and sham, 0.61 Hz, IQR 0.54-0.72 Hz, n = 4). Since an analysis of variance between groups under both respiratory conditions did not reveal a significant difference, data are considered together.

The dominant rhythm in single PGN discharges was assessed from periodic peaks in autocorrelograms (Fig. 2Aa and Ba). As with the whole-nerve data, RaN stimulation evoked an initial discharge (median latency, 0.420 s, IQR 0.368-0.420 s) followed by greatly reduced activity, seen clearly in the stimulus-triggered cross-correlograms (Fig. 2Ab and Bb). This was followed by rhythmic discharges of similar periodicity to those observed in autocorrelograms (Fig. 2Aa and Ba). Unlike sympathetic discharges, RaN stimulation did not lead to stimulus-locked PN rhythmic discharges (Fig. 2Ac and Bc). The median frequency of PN rhythmic discharges was not significantly different during RaN (0.59 Hz, IQR 0.49-0.68 Hz) and sham (0.61 Hz, IQR 0.54-0.68 Hz) stimulation (n = 4).

Figure 2.

Discharge response of a single PGN to RaN stimulation

A and B, RaN stimulation and sham stimulation, respectively, under normal CRD conditions. Aa and Ba, autocorrelograms reveal the same dominant rhythm under RaN and sham stimulation conditions (0.68 Hz). Ab and Bb, stimulus-PGN cross-correlograms (85 triggers). RaN stimulation evokes an initial discharge followed by a period of reduced firing and then rhythmical bursting. Inset in Ab is the sham stimulation autocorrelogram (Ba) on the same timebase. Ac and Bc, stimulus-PN cross-correlograms show no stimulus-related activity. Ad and Bd, stimulus-triggered (tr) raster plots for RaN stimulation (Ad) reveal clear vertical striations (arrowheads) indicating a consistent temporal response pattern compared to sham stimulation (Bd). Ae and Be, PGN raster plots reordered by the first post-stimulus interevent interval for each stimulation (see Chang et al. 2000). Ae, vertical bands (vertical arrowheads) indicate a constant phase relationship with respect to RaN stimulation. The pre-stimulation diagonal band (diagonal arrowhead) reveals a variable phase relationship with the stimulus. Be, under sham stimulation conditions only diagonal bands (diagonal arrowheads) are seen. Ca, two sample neurograms showing the response of a single PGN to RaN stimulation. Cb, 10 overlapping sweeps of PGN spikes from the unit in Ca.

By computing stimulus-triggered raster plots a robust PGN discharge was observed to be consistently locked to RaN stimulation (Fig. 2Ad) but not to sham stimulation (Fig. 2Bd). Reordered raster plots (where events are ordered by the duration of the first post-stimulus interevent interval; see Chang et al. 2000) were used to examine the distribution of stimulus-PGN discharge phase differences across the whole PGN rhythm cycle (Fig. 2Ae and Be). From this example it can be seen that although prior to RaN stimulation the phase difference between rhythmic PGN discharges and RaN stimulation varied (see diagonal arrowheads and arrows), there was a constant phase difference between PGN firing and RaN stimulation (see vertical arrow heads) following the stimulus. Thus the stimulus-locked rhythmicity appears to arise from a process that is relatively insensitive to the phase of the cycle in which the stimulus is delivered. Figure 2Ca shows two typical responses of a single PGN to RaN stimulation. The single-unit nature of the PGN is shown by overlapping sweeps in Fig. 2Cb.

Paired recordings were made from single PGNs (n = 5, 4 animals) to examine their relative firing patterns with respect to RaN stimulation (Fig. 3). These experiments were performed in the absence of CRD. Cross-correlograms for the units under RaN and sham stimulation conditions are shown in Fig. 3a and B. Joint peristimulus scatter plots (Gerstein & Perkel, 1969) were used to examine the joint firing probability of the two PGNs in relation to a RaN stimulus. Following RaN stimulation (Fig. 3C), activity recorded from pairs was seen to be coordinated in a periodic fashion compared to sham stimulation (Fig. 3D). After several cycles the pattern of rhythmic coincident firing begins to drift (Fig. 3C). The discharges of all PGN pairs behaved in a similar manner.

Figure 3.

Simultaneous recording of two PGNs during RaN stimulation and sham stimulation

Aa and b, stimulus-PGN1 and stimulus-PGN2 cross-correlograms show the same characteristic pattern of response seen in Fig. 2. The insets show autocorrelograms of the dominant rhythms of PGN1 (0.59 Hz) and PGN2 (0.73 Hz). Ba and b, sham stimulus-PGN1 and sham stimulus-PGN2 cross-correlograms show no evidence of patterned activity in relation to sham stimulation events. C and D, joint peri-stimulus scatter plot showing the joint firing probability of the two PGNs in relation to the RaN stimulus (C) and the sham stimulus (D). C, RaN stimulation leads to coincident rhythmical bursting, illustrated by dense patches (arrows), which become less defined as time from the stimulus increases. D, sham stimulation does not lead to periodic patterned activity. In both C and D, weak non-periodic coincident activity clusters along the 45 deg diagonal line (arrowheads); a phenomenon reflecting inconsistent, near-zero lag synchronization, which has been described previously (Chang et al. 1999). RaN st, RaN stimultion; sh st, sham stimulation.

The post-stimulus rhythmicity arises from a resetting of the dominant rhythm

As with the whole-nerve data, RaN stimulation was followed by a stimulus-locked rhythmical pattern in single PGN discharges without a contribution from CRD. We investigated whether such rhythmicity had the same periodicity (frequency) as the dominant rhythm in the discharge of each PGN, since this would suggest that the stimulus-locked rhythm was a consequence of rhythm resetting rather than conventional reflex volleys. Therefore, for each PGN we measured the latency from RaN stimulation to each post-stimulus peak in the cross-correlogram and plotted these values against the period of the dominant rhythm (Fig. 4a). The first and second discharges (‘a’ and ‘b’ in the schematic diagram in Fig. 4a, inset) occurred at a relatively constant latency following the stimulus and there was no significant correlation between the latencies and the dominant period of PGN discharge rhythmicity (see Fig. 4a). By contrast, the third discharge (‘c’ in Fig. 4a) was significantly positively correlated (rs = 0.80, P < 0.0001, n = 18). Furthermore, by plotting the inverse of period c - b (i.e. frequency) against the frequency of the dominant rhythm, these variables were shown to be significantly correlated (rs = 0.95, P < 0.0001, n = 18, Fig. 4B). Pairwise statistical comparisons revealed that the dominant rhythm (median frequency, 0.61 Hz; IQR 0.55-0.73 Hz; n = 18) and this measure of the post-stimulation rhythm (median frequency, 0.64 Hz; IQR 0.56-0.72 Hz; n = 18) were not significantly different (Fig. 4B, inset). We therefore suggest that discharge ‘a’, which has a latency that is not correlated to that of the dominant rhythm period, is a reflex response, and discharge ‘b’ represents the first discharge following the resetting of the dominant rhythm. This idea is consistent with our finding that the frequency of the post-stimulus rhythm (1/(c - b)) is equivalent to the dominant rhythm in each unit.

Figure 4.

Summary of the characteristics of RaN stimulation-evoked single PGN activity

A, scatter plot showing the relationship between the dominant rhythm period for each PGN recorded and the latency of the three main RaN-evoked burst discharges (‘a’, ‘b’ and ‘c’; see schematic diagram, inset). ‘a’ and ‘b’ are not significantly correlated with the dominant rhythm period while ‘c’ is significantly positively correlated. B, scatter plot of the dominant rhythm frequency (Hz) for each PGN against the post-RaN stimulus rhythm (calculated as 1/(c - b)). The dominant rhythm and the post-stimulus rhythm are significantly positively correlated. The histogram in the inset shows the median ± IQR frequency for the dominant rhythm (d) and the post-stimulus rhythm (p-s). The median frequencies of the dominant and post-stimulus rhythm are not significantly different. C, scatter plot of discharge rate (Hz) of each single PGN during RaN stimulation vs. sham stimulation. The discharge rates under these conditions are significantly positively correlated. The histogram in the inset shows the median ± IQR discharge rate under RaN stimulation (st) and sham stimulation (sh) conditions. The median discharge rates are not significantly different. For B and C, PGNs recorded in both the absence of CRD (♦) and normal CRD (⋄) are shown. See Results for more details.

We also measured the discharge activity to assess whether RaN stimulation leads to a change in firing rate compared to the sham stimulation control. Figure 4C shows a scatterplot for the RaN stimulation discharge rate (median, 1.29 Hz; IQR 0.86-1.67 Hz; n = 18) vs. sham stimulation discharge rate (median, 1.18 Hz; IQR 0.77-1.54 Hz; n = 18). Analysis demonstrated that the RaN stimulation and sham stimulation discharge rates were significantly correlated (rs = 0.86, P < 0.0001, n = 18). Furthermore, pairwise statistical comparisons revealed that the RaN stimulation and sham stimulation discharge rates were not significantly different (Fig. 4C, inset).

Effect of a natural mechanical stimulus on single PGN activity

Six single PGNs (from 4 animals) were recorded during a ≈300 s period in which 12-18 pinches were applied to the left forepaw. These were delivered by hand and event-marked. The effect of the stimulus on single PGN activity was very similar to that found using RaN stimulation. An example of the activity of a PGN recorded in the absence of CRD is shown in Fig. 5. The dominant rhythm under mechanical and sham stimulation conditions is shown in Fig. 5Aa and Ba, respectively. As with RaN stimulation, cross-correlograms of post-stimulus discharges showed a characteristic rhythmicity similar to that seen in the sham stimulation autocorrelogram (inset in Fig. 5Ab). In contrast, under sham stimulation conditions cross-correlograms of ‘post-stimulus’ discharges were essentially flat (Fig. 5Bb). The mechanical stimulation raster plots showed clear vertical striations indicating that the mechanical stimulus evoked a consistent pattern of response across trials (Fig. 5Ac and Bc).

Figure 5.

Response of single PGN discharges to a pinch of the forepaw in the absence of CRD

A, mechanical stimulation. B, sham stimulation. Aa and Ba, autocorrelograms showing that the dominant rhythm under the two conditions was the same (0.68 Hz). Ab and Bb, stimulus-PGN cross-correlogram (18 triggers) under both conditions. The inset in Ab is the sham stimulation autocorrelogram (Ba). Ac and Bc, stimulus triggered (tr)-PGN raster plots under both conditions. In response to forepaw pinch (Ac) clear vertical striations can be observed (arrowheads) illustrating the consistency of the response across time.


We have demonstrated that somatic afferent input evokes complex, multiphasic responses in PGNs innervating a defined peripheral target. Averaged VCN responses and cross-correlograms of the responses of single PGNs revealed that RaN stimulation evoked an initial discharge, a subsequent period of reduced activity and then rhythmic discharges. The initial discharge occurred at a latency of ≈400 ms after RaN stimulation. Assuming an additional latency to account for synaptic and conduction delays (estimate of delay between sympathetic chain ganglia and VCN recording site ≈100-230 ms; Johnson & Gilbey, 1994), this is similar to the latencies recorded from cardiac nerves in rats following tibial nerve stimulation (Adachi et al. 1992; Kimura et al. 1996).

Statistical analysis revealed that the frequency of the bursts following the initial reflex volley was not significantly different from the dominant rhythm of ongoing discharges. This is consistent with the view that the stimulus-locked rhythm results from a resetting of the dominant rhythm. An alternative explanation is that these discharges are reflex components. We consider this unlikely for two main reasons. (1) In some single PGNs rhythmical discharges locked to the stimulus were seen > 3 s following the initial reflex. Although the pathway from RaN to CVA may lead to relatively long delays, this latency is far in excess of previous reports of somatosympathetic responses, such as the very late suprapontine reflex in cats (300-350 ms; Sato et al. 1972a). (2) There was a strong correlation between the dominant rhythm and the post-stimulus rhythm (e.g. single PGNs with intrinsically slow dominant rhythms also had slow post-stimulus rhythms and vice versa). If rhythmical discharges arose from reflex responses, one would predict that their frequency would be independent of the dominant rhythm, as is indeed true for the initial reflex volley.

We propose that the stimulus-locked rhythmic discharges, following the initial reflex volley, arise from the resetting of oscillators. This suggestion is consistent with our previous proposal that CVA PGNs are influenced by a population of multiple oscillators, each with rhythms of a different intrinsic frequency (Chang et al. 1999). In this model, somatic afferent stimulation would lead to a resetting of the population of oscillators, so that single PGNs become transiently synchronized. This synchronization explains the prominent post-stimulus bursts seen in whole-nerve activity. In recordings from pairs of PGNs such post-stimulus synchronization following the reflex volley was seen in joint peri-trigger scatter plots. Following the reset, the differences in intrinsic dominant rhythms cause the phase relationships within the population to gradually drift. Thus the post-stimulus synchronization wanes after a few cycles.

Our previous studies have revealed that the activity of single PGNs can be entrained by CRD (Chang et al. 1999) and this raises the possibility that in the present experiments resetting of PGN activity might have arisen from influences on CRD. We have no evidence to support this conjecture. However, in recordings from animals in the absence of CRD we often observed a small burst of activity consistent with the notion that somatic afferent stimulation can influence CRD. Consequently we consider it possible that in freely behaving animals, the influence of the somatic afferent input on CRD might make a much more significant contribution to the resetting behaviour of the sympathetic network. This influence would be particularly striking when the respiratory and sympathetic networks function as a highly coordinated network (see Chang et al. 1999).

The post-stimulus non-reflex synchronization of sympathetic discharges that we report here has not been identified in previous studies examining somatosympathetic responses. Many of these have been concerned with investigating the short-latency reflexes recorded from whole nerves (e.g. Sato et al. 1965; Coote & Downman, 1966; Schmidt & Schönfuss, 1970; Kimura et al. 1996) and while some studies have analysed these responses in single preganglionic or postganglionic fibres/neurones (eg. Sato et al. 1972b; Janig et al. 1972; Dembowsky et al. 1985; Sato et al. 1985) these were usually examined over relatively short peri-stimulus intervals and were not considered in relation to ongoing rhythmical discharges.

The RaN stimulation parameters we used in these experiments were strong (suprathreshold for C-fibre activation) and brief, and led to consistent resetting in the PGNs we examined. We consider such electrical stimulation to be similar to mechanical paw pinch, which appeared to have comparable effects. As paw pinch is a stimulus that might be commonly encountered in a natural environment, where it would produce arousal and alerting behaviour, it provides physiological relevance for the responses we have identified. It would be interesting to investigate what sympathetic pattern might result from a sustained electrical stimulation or a maintained paw pinch, since these would arguably have an even closer resemblance to the afferent inputs associated with an injury.

The influence of somatic inputs on autonomic control presumably provides a means to produce a coordinated response to a change in an organism's external environment. We suggest that inputs that act on a population of oscillators that control the rhythmic discharges of PGNs would provide a mechanism for extending the coordinated response; in most instances for several seconds after a short-lasting stimulus. This would accompany any behavioural changes.

The results presented in this paper taken together with our observations on the effect of respiratory-related inputs on oscillator population synchrony support the idea that the dynamic and graded synchronization of biological oscillators provides a possible mechanism for the generation of sympathetic responses. Such synchronization may contribute to the enhanced burst amplitude of sympathetic activity associated with stress that has been reported in studies on humans (e.g. Delius et al. 1972; Callister et al. 1992; Nordin & Fagius, 1995; Morgan et al. 1996; see McAllen & Malpas, 1997, for review). In functional terms, evidence from the central nervous system has suggested that synchronized burst discharge leads to highly effective synaptic transmission (Lisman, 1997). In the periphery there is evidence indicating that burst discharges in PGNs may influence the efficacy of neuroeffector transmission (Andersson, 1983; Nilsson et al. 1985). Thus synchronization elicited through the actions of somatic inputs may provide an important mechanism for modulating blood flow.


K.S. and the work carried out in this study were supported by Wellcome grant no. 05115; H.-S.C. was supported by Chang Gung Memorial Hospital. We would like to thank Bruce Cotsell for his excellent technical support.