1. Top of page
  2. Abstract
  6. Acknowledgements
  • 1
    Blockade of NMDA receptors by dizocilpine impairs the inspiratory off-switch (IOS) of central origin but not the IOS evoked by stimulation of sensory afferents. To investigate whether this difference was due to the effects of different patterns of synaptic interactions on respiratory neurones, we stimulated electrically the superior laryngeal nerve (SLN) or vagus nerve in decerebrate cats before and after i.v. administration of dizocilpine, whilst recording intracellularly.
  • 2
    Phrenic nerve responses to ipsilateral SLN or vagal stimulation were: at mid-inspiration, a transient inhibition often followed by a brief burst of activity; at late inspiration, an IOS; and at mid-expiration, a late burst of activity.
  • 3
    In all neurones (n= 16), SLN stimulation at mid-inspiration evoked an early EPSP during phase 1 (latency to the arrest of phrenic nerve activity), followed by an IPSP in inspiratory (I) neurones (n= 8) and by a wave of EPSPs in post-inspiratory (PI) neurones (n= 8) during phase 2 (inhibition of phrenic activity). An EPSP in I neurones and an IPSP in PI neurones occurred during phase 3 (brief phrenic burst) following phase 2.
  • 4
    Evoked IOS was associated with a fast (phase 1) activation of PI neurones, whereas during spontaneous IOS, a progressive (30-50 ms) depolarization of PI neurones preceded the arrest of phrenic activity.
  • 5
    Phase 3 PSPs were similar to those occurring during the burst of activity seen at the start of spontaneous inspiration.
  • 6
    Dizocilpine did not suppress the evoked phrenic inhibition and the late burst of activity. The shapes and timing of the evoked PSPs and the changes in membrane potential in I and PI neurones during the phase transition were not altered.
  • 7
    We hypothesize that afferent sensory pathways not requiring NMDA receptors (1) terminate inspiration through a premature activation of PI neurones, and (2) evoke a late burst of phrenic activity which might be the first stage of the inspiratory on-switch.

Transition between phases in the respiratory cycle is an essential process of respiratory rhythm generation. Previous studies have shown that inspiratory termination is normally promoted by two pathways projecting onto the respiratory central rhythm generator, the pontine pneumotaxic centre and vagal afferents from pulmonary stretch receptors (reviews in Bianchi, Denavit-Saubié & Champagnat, 1995; von Euler, 1986). In vagotomized cats the spontaneously occurring inspiratory off-switch (IOS) can be delayed pharmacologically by dizocilpine, an antagonist of N-Methyl-d-aspartate (NMDA) receptors, resulting in an apneustic respiration (Foutz, Champagnat & Denavit-Saubié, 1988, 1989; Pierrefiche, Foutz, Champagnat & Denavit-Saubié, 1992). However, blockade of NMDA receptors does not suppress IOS induced by stimulation of peripheral sensory afferents (Foutz et al. 1989; Karius, Ling & Speck, 1991). The present study investigated whether the different sensitivity to NMDA receptor blockade of IOS of central or peripheral origin was due to different patterns of synaptic interactions on the target respiratory neurones of the ventral respiratory group (VRG). We recorded synaptic responses of respiratory neurones to stimulation of the superior laryngeal nerve (SLN) and vagus nerve, before and after induction of an apneustic respiratory pattern by dizocilpine. The effects of SLN stimulation on the respiratory rhythm have been extensively described (Berger, 1977; Iscoe, Feldman & Cohen, 1979; McCrimmon, Speck & Feldman, 1987), and it was shown in electrophysiological and anatomical studies that laryngeal afferents in cats terminate in the interstitial, medial, and ventrolateral subnuclei of the nucleus tractus solitarius (Kalia & Mesulam, 1980; Lucier, Egizii & Dostrovski, 1986; Bellingham & Lipski, 1992; Takagi, Umezaki & Shin, 1995). Stimulation of SLN afferents affects a wide population of inspiratory (I), post-inspiratory (PI) and expiratory neurones located in the VRG and its rostral-most part, the Bötzinger Complex (Czyzyk-Krzeska & Lawson, 1991; Jiang & Lipski, 1992), but the pathways through which these afferents innervate respiratory neurones remain to be determined. Some relay neurones of the SLN were found in the retrofacial nucleus, but these neurones do not seem to project to propriobulbar or bulbospinal respiratory neurones (Ezure, Oku & Tanaka, 1993). Within the respiratory neuronal network, PI neurones are of great interest because they can be activated by peripheral afferent stimulation (Remmers, Richter, Ballantyne, Bainton & Klein, 1986) and because their activity is profoundly depressed after pharmacological blockade of the spontaneous IOS mechanism (Pierrefiche et al. 1992; Haji, Pierrefiche, Takeda, Foutz, Champagnat & Denavit-Saubié, 1996b). Inspiratory (I) neurones are also important because they are the target of the IOS mechanism (Bianchi et al. 1995). Therefore we analysed the postsynaptic events occurring in I and PI neurones during the peripherally evoked IOS in eupneic and apneustic respiration induced by dizocilpine.


  1. Top of page
  2. Abstract
  6. Acknowledgements

Surgical procedures

Experiments were performed on sixteen adult cats of either sex, weighing 2.5-3.8 kg, in accordance with the regulations of the French Ministry of Agriculture. The animals were anaesthetized with halothane (2.0-2.5 % in 50 % oxygen during induction and 1.5-1.8 % during surgery, ensuring a total absence of nociceptive reflexes and stable heart rate and blood pressure). Catheters were placed in the femoral vein, the femoral artery and the urethra. To secure haemostasis during and after decerebration, the external carotid arteries were tied bilaterally distal to the lingual artery branch. The head of the animal was then mounted on a stereotaxic frame and the mid-collicular decerebration was performed according to the procedures described by Kirsten & St John (1978). A C2-C3 laminectomy and an occipital craniotomy were performed. Phrenic nerves, cervical vagus nerves and superior laryngeal nerves (SLNs) were exposed by a dorsal approach and cut bilaterally. A wide bilateral pneumothorax was performed to minimize the movements of the brainstem associated with ventilation. Then halothane anaesthesia was discontinued.

After decerebration, the animals were paralysed with pancuronium bromide (0.3 mg kg−1 initially and 0.1 mg kg−1 h−1i.v.) and the lungs were artificially ventilated with oxygen-enriched air (Fi,O2= 0.3). An end-expiratory pressure of 1 cmH2O was applied to prevent collapse of the lungs. Tracheal pressure was kept below 8 cmH2O at maximal lung inflation. The end-tidal concentration of CO2 was continuously monitored (Datex Capnomac, Helsinki, Finland) and kept at 4-5 % by regulating the rate of ventilation with a fixed tidal volume of 10 ml kg−1. A glucose-lactate Ringer solution was infused i.v. at the rate of 3-5 ml kg−1 h−1. Rectal temperature was maintained at 37-38°C by external heating. Mean arterial blood pressure was higher than 100 mmHg in all animals. At the end of the experiments the animals were killed with an overdose of pentobarbitone (120 mg kg−1i.v.).

Recording and stimulating procedures

Respiratory activity was recorded from the central cut end of the phrenic nerve, positioned on bipolar silver electrodes. The amplified phrenic activity (× 2000-10000) was filtered (3-3000 Hz), rectified and integrated with a leaky integrator (time constant, 0.1 s). Respiratory neurones were sought in the ventral respiratory area, extending 2.5-4.3 mm lateral to the mid-line, 0-3 mm rostral to the obex, and 2.6-4.2 mm below the dorsal surface of the medulla oblongata. Membrane potentials were recorded with sharp glass micropipettes filled with 2 M potassium citrate, with a resistance of 20-40 MΩ as measured in brain tissue. The type of respiratory neurone was identified by the pattern of membrane potential fluctuations in relation to phrenic nerve activity. To identify the axonal projections of recorded neurones, the central ends of the ipsilateral vagus nerve and SLN were stimulated with a 0.1 ms pulse of 0.1-0.3 mA intensity, using bipolar silver electrodes (Haji, Pierrefiche, Foutz, Champagnat, Denavit-Saubié & Takeda, 1996a). The spinal axons were identified by stimulation with an array of five concentric stimulating electrodes inserted into the ventrolateral part of the C2-C3 spinal cord, using 0.2 ms pulses of 0.3-0.5 mA.

Postsynaptic potentials (PSPs) were induced by a single shock stimulation to the ipsilateral SLN and vagus nerve. The timing of stimuli delivered during the respiratory cycle is shown in Fig. 1C. Stimulus intensities (0.3-1.0 mA, 0.1 ms) were set at 1.5 times the threshold to evoke a transient inhibition in the phrenic nerve discharge (evoked inhibition) when the stimulus was applied at mid-inspiration (500-600 ms after the onset of inspiration). When this stimulus was applied at late inspiration (1000-1200 ms after the onset of inspiration and at least 100 ms before the timing of spontaneous off-switch measured on unstimulated inspiratory phases), it evoked the phase transition from inspiration to expiration (evoked IOS). In some experiments, dizocilpine-induced changes in the PSPs were measured by current clamping the membrane potentials to the pre-drug potentials (8100-1, Dagan, Minneapolis, MN, USA). Phrenic, bulbar neuronal and stimulation signals were displayed on a chart recorder and stored on magnetic tape. Stored data were played back for off-line analysis using signal processing software, Wave Master (Canopus, Kobe, Japan).


Figure 1. PSPs evoked in two I neurones during transient phrenic inhibition (A) and IOS (B) elicited by SLN stimulation

The superior laryngeal nerve (SLN, triangles) was stimulated 500 ms after the onset of inspiration to evoke a transient phrenic inhibition (A) and 1 s after the onset of inspiration to evoke the phase transition (B), before (Aa, Ba) and after (Ab, Bb) dizocilpine injection (0.1 mg kg−1i.v.). Each trace was the average of 5 consecutive respiratory cycles. The value (mV) of the reference membrane potential (dashed horizontal line) is indicated on the left. In Bb the potential was re-adjusted to the same potential level as in Ba by current clamping. PN, phrenic neurogram. In this and the following figures, phase 1 (latency to onset of phrenic inhibition) and phase 2 (phrenic inhibition) of the response are indicated by vertical dashed lines. Both cells are laryngeal motoneurones. C, phrenic neurogram showing full respiratory cycles during eupnoea (left) and after dizocilpine (right). Arrows indicate the stimulations.

Download figure to PowerPoint

Drug administration and data analysis

Recordings of neuronal activity started at least 4 h after discontinuation of halothane anaesthesia. Dizocilpine (Research Biochemicals International) was slowly administered intravenously over a period of 5 min, at doses of 0.1-0.2 mg kg−1. These doses produced consistent effects both on bulbar neuronal and phrenic activity. Since dizocilpine has a long half-life (Hucker, Hutt, White, Arison & Zacchei, 1983) and its effects on the respiratory function do not fully recover in the course of an experiment (Foutz et al. 1989), at most one continuous recording of membrane potential before and after dizocilpine injection was achieved in every animal (16 neurones in 16 cats). To reduce the length of the recording and the risk of dislodging the electrode from the cell, one type of stimulation (applied to the SLN) was chosen as the main procedure in the analysis of evoked responses. All recorded respiratory neurones gave membrane potentials more negative than -50 mV, and spontaneous action potentials with overshoot and after-hyperpolarization.

Evoked responses in the phrenic nerve and bulbar respiratory neurones were averaged over five consecutive respiratory cycles. For the phrenic nerve responses, the latency to onset and duration of the evoked inhibition and late excitation, and the latency to onset of the evoked IOS were measured in the averaged phrenic neurogram. IPSPs in I and PI neurones were identified as such if they were hyperpolarizing, if they involved an arrest of action potential firing, and if they decreased in size with hyperpolarizing current injection. EPSPs were identified by showing depolarization of the membrane accompanied with a decrease in input resistance and an increase of action potential firing. The membrane potential just before the stimulation was used as a reference to measure the peak amplitudes of the PSPs. To distinguish the transient from the irreversible phase of the membrane potential change during the IOS process, the digitized membrane potential trace during the evoked transient phrenic inhibition was subtracted from the membrane potential during the evoked IOS before and after dizocilpine administration. This procedure reduced or abolished the transient change in membrane potential associated with the evoked PSPs. Quantitative data were expressed as means ±s.e.m. Differences in the mean values before and after dizocilpine administration were evaluated by Student's paired t tests, with significance level at P < 0.05.


  1. Top of page
  2. Abstract
  6. Acknowledgements

Effects of dizocilpine on phrenic nerve responses to peripheral stimulation

When either the SLN or the vagus nerve was stimulated at mid-inspiration during eupnoea (n= 16 animals), a transient inhibition was evoked in the phrenic neurogram (PN, lower traces in Figs 1 and 2). The period between the stimulus and the start of inhibition (phase 1) and the period of inhibition (phase 2) of this transient inhibition did not differ significantly after SLN stimulation (n= 16 animals, Table 1A), and after vagal stimulation (7.6 ± 1.1 ms and 22.6 ± 3.7 ms, respectively, n= 9 animals). Administration of dizocilpine caused an apneustic respiration, characterized by a prolonged inspiration with a plateau-like discharge, as described previously (Foutz et al. 1989). The evoked transient phrenic inhibition persisted after dizocilpine. After SLN stimulation, it had the same latency and, on average, an increased duration (n= 16, P < 0.05, Table 1A and Fig. 1A).

Table 1.  Responses of the phrenic nerve and of inspiratory neurones to stimulations of the SLN applied at mid-inspiration (A) and late inspiration (B)
 Duration (ms)Amplitude (mV)
  1. * P < 0.05, versus control.

A. Stimulation at mid-inspiration
   Phrenic nerve (n= 16)    
      Phase I8.6 ± 0.58.4 ± 0.9
      Phase 229.7 ± 5.143.8 ± 7.4 *
   I neurones (n= 8)    
      Latency to EPSP4.3 ± 0.85.8 ± 0.8
      EPSP6.8 ± 0.56.9 ± 1.14.8 ± 0.44.0 ± 1.0
      IPSP36.5 ± 6.529.9 ± 7.54.9 ± 0.34.3 ± 0.5
B. Stimulation at late inspiration
   Phrenic nerve (n= 9)    
      Latency to IOS9.8 ± 1.69.7 ± 1.4
   I neurones (n= 5)    
      Latency to EPSP4.5 ± 0.84.4 ± 0.6
      EPSP6.7 ± 1.010 ± 1.92.4 ± 0.73.1 ± 0.7

When SLN stimulation was applied at late inspiration (Fig. 1B, Table 1B), it evoked an irreversible termination of the inspiratory phase (evoked IOS, n= 9) with an onset latency which was not significantly different from the latency of the transient inhibition in mid-inspiration, and it was unchanged by dizocilpine (n= 9).

The transient phrenic inhibition (phase 2) in mid-inspiration was often followed by a burst of phrenic activity lasting 15-25 ms (phase 3). SLN stimulation produced this burst in 9 of 16 animals and vagal stimulation in 6 of 9 animals. Similar bursts (with durations of 17 ± 3 ms after SLN stimulation, 21 ± 2 ms and 16 ± 2 ms after vagal stimulation) also occurred when the stimulus was applied during late inspiration and evoked an IOS, and when it was applied at mid-expiration (Figs 3C and 4-6). These late bursts of activity persisted after dizocilpine administration (Fig. 5). On average, the onset latency of phase 3 bursts was 24 ± 1 ms after SLN stimulation and 21 ± 2 ms after vagal stimulation. This latency showed a rather large variability between animals. However, in each animal, these bursts occurred in inspiration and expiration with near identical latencies (they differed by 0-3 ms between inspiration and expiration; Fig. 4).


Figure 5. Effect of dizocilpine on PSPs in an I neurone and late burst of phrenic nerve discharge (phase 3) evoked by vagal stimulation

The stimulus (X, triangle) was applied at mid-expiration before (A) and after (B) dizocilpine administration. Top traces, laryngeal motoneurone; lower traces, phrenic neurogram (PN).

Download figure to PowerPoint


Figure 4. SLN stimulation applied during inspiration and expiration evoked PSPs in a PI neurone and a late burst of phrenic nerve discharge

A, SLN stimulation applied 500 ms after the onset of inspiration evoked an EPSP/IPSP sequence in a non-antidromically activated PI neurone (top trace) and a transient inhibition of phrenic discharge (PN, lower trace) terminated by a burst of activity (truncated). B, the same stimulus applied at mid-expiration evoked a phrenic burst with the same onset latency as during inspiration. The three phases of the phrenic response are indicated.

Download figure to PowerPoint

Effect of dizocilpine on respiratory neuronal responses to peripheral stimulation

Eight I neurones of the ‘ramp’ type (including four laryngeal motoneurones) and eight PI neurones (including five laryngeal motoneurones) were tested with SLN stimulation during eupneic respiration. The evoked PSPs did not differ between laryngeal motoneurones and non-laryngeal neurones, as was described previously (Haji et al. 1996a). SLN or vagal stimulation at different times of the respiratory cycle produced three-phase neuronal responses. These responses persisted after administration of dizocilpine. Quantitative measurements were made on SLN-evoked PSPs. Vagally evoked PSPs are shown in Figs 5 and 6.

Inspiratory neurones.

SLN stimulation applied during inspiration evoked a polyphasic PSP in all I neurones (Fig. 1Aa). The first PSP to occur (phase 1 EPSP) was a short lasting EPSP (Table 1A). It was followed by a longer-lasting IPSP, concomitant with the transient inhibition of phrenic nerve activity (phase 2). The late burst of phrenic activity (phase 3) was accompanied by an EPSP which started at the beginning of the phrenic burst and terminated the on-going hyperpolarization resulting from the phase 2 IPSP (as shown in Figs 3C and 5A).

When SLN stimulation evoked an IOS (n= 5), the first event was a phase 1 early EPSP which had a similar onset latency as when the stimulus evoked a reversible inhibition (Table 1B). This EPSP was followed by a hyperpolarization as the phrenic nerve remained silent (Fig. 1Ba).

After apneusis was induced by dizocilpine, the latency to onset and duration of each PSP were not significantly changed, and changes in their amplitude were inconsistent. The amplitude of the phase 1 early EPSP was increased by dizocilpine in one cell (Fig. 1A), decreased in another, and unchanged in the others (Fig. 1B). Table 1 shows that on average, dizocilpine did not change significantly the amplitudes of the phase 1 EPSP and phase 2 IPSP during evoked inhibitions (n= 8) and evoked IOS (n= 5). Because of the membrane depolarization during inspiration, there was a significant decrease in the amplitude of the phase 1 EPSP between mid-inspiration and late inspiration in eupnoea (P < 0.01; compare amplitudes in Table 1A and B). Both the depolarization during inspiration (see Haji et al. 1996b) and this decrease in the EPSP were significantly reduced by dizocilpine.

Depending on the timing of stimulation in the respiratory cycle, at the end of the phase 3 phrenic burst the membrane potential returned to the baseline value at mid-inspiration, hyperpolarized when the stimulus evoked an IOS, or returned to the hyperpolarized potential of expiration.

Post-inspiratory neurones.

SLN stimulation applied at mid-inspiration elicited in PI neurones an EPSP (Fig. 2A) which had an onset latency (5.6 ± 0.7 ms, n= 8) and a peak amplitude (4.8 ± 1.1 mV) which were not significantly different from phase 1 EPSP in I neurones (Table 1A), but had a longer duration (36.2 ± 7.5 ms). This EPSP thus started during phase 1 and peaked during phase 2. When the phrenic silencing was followed by a late burst of activity (phase 3), an IPSP occurred concomitantly in PI neurones (see Fig. 4).


Figure 2. EPSPs evoked in two PI neurones during transient phrenic inhibition (A) and IOS (B) elicited by SLN stimulation

The SLN was stimulated (triangles) with the same timing as in Fig. 1 before (Aa, Ba) and after (Ab, Bb) dizocilpine injection (0.2 mg kg−1i.v.). In both cells the membrane potential was re-adjusted to the pre-dizocilpine level by current clamping. Each trace was the average of 5 consecutive respiratory cycles. Dashed lines show the reference membrane potentials. PN, phrenic neurogram. A, laryngeal motoneurone; B, non-antidromically activated PI neurone.

Download figure to PowerPoint

During evoked IOS, a wave of EPSPs started in PI neurones during phase 1 (Fig. 2B) with the same latency to onset (5.0 ± 0.7 ms) and peak amplitude (5.1 ± 0.9 mV) as when the stimulus, applied at late inspiration, evoked a transient inhibition of phrenic activity. The initial EPSP was followed by a sustained depolarization and firing of action potentials at the time when phrenic activity failed to resume as it did during transient inhibition, and expiration became irreversible.

Dizocilpine did not significantly affect the time course of the evoked EPSP (Fig. 2B), nor its amplitude (5.0 ± 1.4 mV during the evoked inhibition and 5.2 ± 1.0 mV during evoked IOS; n= 6, including 2 non-antidromically activated neurones). During evoked IOS, the phase 2 action potential firing of PI neurones was suppressed by dizocilpine (Fig. 2Bb), in keeping with the decrease of membrane depolarization throughout post-inspiration (Haji et al. 1996b).

Premature discharge of PI neurones during evoked IOS

During evoked IOS, the onset of the EPSP in PI neurones preceded by a few milliseconds the silencing of the phrenic nerve discharge (Fig. 3A). Such a fast-rising EPSP was not seen in PI neurones during spontaneous (non-evoked) IOS, where a progressive depolarization started 35-40 ms before the arrest of phrenic activity (Fig. 3B). In I neurones, the spontaneous phase transition did not start by a fast-rising PSP as during evoked IOS. Instead, a progressive hyperpolarization started 30-35 ms before the arrest of phrenic activity, and paralleled the progressive depolarization of PI neurons (Fig. 3D).


Figure 3. Evoked (left) and spontaneous (right) inspiratory off-switch in PI and I neurones

SLN stimulation evokes an EPSP in a non-antidromically activated PI neurone before the arrest of phrenic activity (A), whereas during spontaneous phase transition (B) a slow depolarization of the PI neurone started 35 ms before the arrest of phrenic activity (boxed area). In an inspiratory laryngeal motoneurone (C) the stimulus evoked an IPSP concomitant with phrenic arrest, whereas during the spontaneous phase transition (D) the neurone started to hyperpolarize 35 ms before phrenic arrest (boxed area). Note in C the late burst of phrenic activity and the EPSP in the I neurone.

Download figure to PowerPoint

Reversible and irreversible off-switch

To distinguish transient from irreversible phases of the neurone response, we subtracted the membrane potential during evoked phrenic inhibition from the membrane potential during evoked IOS (not illustrated, see Methods). We found that during evoked IOS, the hyperpolarization of membrane potential in I neurones started 30-50 ms after the onset of the peripherally evoked phase 1 EPSPs, with an average onset latency of 39.9 ± 7.9 ms (n= 5), while PI neurones depolarized with a latency of 37.1 ± 9.5 ms (n= 4). Dizocilpine did not significantly change the timing of membrane potential hyperpolarization related to evoked IOS (onset latency 40.1 ± 8.4 ms in I neurones and 36.8 ± 9.2 ms in PI neurones).

Respiratory neuronal responses to peripheral stimulation during expiration: transient and full ‘on-switch’

In mid-expiration, when SLN stimulation evoked a phase 3 burst of phrenic activity (n= 5), this was accompanied in four of five cells by an IPSP in PI neurones (Fig. 4B) or an EPSP in I neurones (also seen after vagal stimulation; Fig. 5A). The remaining cell did not respond at all during the phrenic burst. In three additional cells, phase 3 PSPs occurred with a low amplitude in the neuronal recording, although the burst discharge was not visible on the phrenic nerve recording.

The depolarization of I neurones and hyperpolarization of PI neurones at the spontaneous onset of inspiration was in most animals (12/16) associated with a transient burst of phrenic activity resembling phase 3 described in the present study. This phrenic burst was not affected by dizocilpine (not shown, but see Fig. 4 in Haji et al. 1996b). The membrane potential of I and PI neurones was also recorded in one animal in which vagal stimulation reproducibly triggered an inspiratory ‘on-switch’ before and after dizocilpine administration (Fig. 6). In this animal the evoked phase 3 phrenic burst starting inspiration was comparable with the burst of phrenic activity preceding the ‘ramp’ discharge and accompanied by hyperpolarization of PI neurones and depolarization of I neurones observed during spontaneous inspiratory on-switch. Phase 1 and phase 2 PSPs evoked by the stimulus have no equivalent during spontaneous on-switch. In contrast, phase 3 responses appear to be a higher-order network-generated event which have an equivalent during spontaneous inspiratory on-switch. In our experimental conditions, the short burst of phrenic activity elicited by SLN stimulation at mid-expiration was never followed by a resumption of inspiration.


Figure 6. PSPs in PI and I neurones during on-switch of inspiration evoked by vagal stimulation

In both neurones the stimulus was applied in expiration after dizocilpine administration. It evoked an EPSP in a post-inspiratory laryngeal motoneurone (A) and concomitantly an IPSP in an inspiratory laryngeal motoneurone (B). The late burst of phrenic activity (phase 3) was followed by a decrease of activity, then by resumption of discharge as the PI neurone hyperpolarized and the I neurone depolarized.

Download figure to PowerPoint


  1. Top of page
  2. Abstract
  6. Acknowledgements

The present results suggest that synaptic responses to stimulation of the SLN or vagus nerve exhibit three successive phases. The early excitatory PSP (phase 1) was independent of the type of neurone, I or PI, whereas during phase 2 (phrenic inhibition) and phase 3 (late phrenic burst), the PSPs had opposite directions in I and PI neurones. These responses were not significantly affected by dizocilpine, which indicates that the pathways involved in the generation of these three successive phases might be distinct from the pathways of the centrally controlled IOS, which have been shown to act through activation of NMDA receptors (Pierrefiche et al. 1992). Furthermore, the dizocilpine-insensitive late burst of phrenic activity (phase 3) is analysed and proposed to be the first stage of inspiratory on-switch.

Different phases of postsynaptic responses evoked by SLN stimulation on respiratory neurones

Stimulation of SLN afferents evoked a rather uniform PSP response in I and PI neurones, always starting with an early and short lasting phase 1 EPSP. This EPSP was evoked in both inspiratory and expiratory phases with latencies which were not significantly different in I and PI neurones, and may represent a relatively direct oligosynaptic projection of SLN afferents on both types of respiratory neurones. Our data are therefore in agreement with other groups which observed this early excitation in I and PI neurones and on the whole phrenic nerve activity contralateral to the stimulation (Berger & Mitchell, 1976; Iscoe et al. 1979; Sica, Cohen, Donnelly & Zhang, 1984; Bellingham, Lipski & Voss, 1989; Donnelly, Sica, Cohen & Zhang, 1989; Jiang & Lipski, 1992). In contrast, the early PSP response is inhibitory in expiratory neurones of the VRG and Bötzinger Complex (Czyzyk-Krzeska & Lawson, 1991; Jiang & Lipski, 1992). Phase 2 IPSPs started approximately 10 ms after the stimulus in I neurones while EPSPs continued in PI neurones in accordance with previous observations (Czyzyk-Krzeska & Lawson, 1991) and reciprocal inhibition between the different types of respiratory neurones (Bianchi et al. 1995). During the inspiratory phase, while the initial EPSP reached its peak in PI neurones, a phase 2 IPSP occurred in I neurones and the phrenic nerve ceased to discharge. These PSPs probably originated from distinct groups of respiratory neurones producing excitation and inhibition in response to early excitation by SLN afferents (Czyzyk-Krzeska & Lawson, 1991).

Peripherally evoked on-switch processes

The period of phrenic silence evoked by SLN or vagal stimulation was often followed by a short burst of activity. Its onset coincided with the onset of an IPSP in PI neurones and an EPSP in I neurones (phase 3). During inspiration, this burst was often followed by the resumption of phrenic activity. During expiration the same stimulus evoked the same phase 3 PSP sequence and transient burst of phrenic activity with a similar latency to that during inspiration, after which the evoked phrenic burst could also be followed by a re-setting of inspiration. A similar short burst of activity initiates the phrenic discharge during the ‘spontaneous’ on-switch and, like the evoked phrenic burst, is unaffected by dizocilpine. This suggests that the resumption of inspiratory activity when the transient phrenic inhibition fails to evolve into an IOS might be due to activation of an on-switch process triggered by the stimulus after an initial phase 1-2 reversible off-switch. Recent work by Oku & Dick (1992) and Oku, Dick & Cherniack (1993) indeed supports the idea that phase 3 burst of phrenic nerve discharge previously observed by several authors (Berger & Mitchell, 1976; Iscoe et al. 1979; McCrimmon et al. 1987) might be viewed as an abortive ‘on-switch’. Although SLN stimulation during the expiratory phase generally prolongs the phase and delays the next inspiration, a phase-resetting characteristic of SLN stimulation has been demonstrated in cats with small pontine lesions (Oku & Dick, 1992) and in a few non-lesioned animals (Oku et al. 1993). In these animals, SLN stimulation applied during late expiration consistently terminated expiration and prematurely advanced the onset of the next inspiration. Oku & Dick (1995) proposed a model explaining the paradoxical phase-resetting response to SLN stimulation by an activation by SLN afferents of E-decrementing and I-decrementing (i.e. early-I) neurones. The present recordings of I and PI neurones support this hypothesis, although we did not record early-I neurones. Considering synaptic relationships between I, PI and early-I neurones (Bianchi et al. 1995), it is likely that the phase 3 IPSP observed in PI neurones during the burst of phrenic activity might originate from early-I neurones (Ezure, Manabe & Otake, 1989).

Taken together, the data show that stimulation of peripheral afferents evoked rather uniform polyphasic responses of the respiratory neurones and of the phrenic nerve throughout the respiratory cycle. How the respiratory network evolves after these responses might depend essentially on the state of the network at the time when the stimulus is delivered, leading to either resumption of phrenic inspiratory activity, IOS, resumption of expiration, or on-switch.

Respiratory-related discharge of PI neurones produces ‘irreversible’ off-switch

The discharge of PI neurones defines one of the two phases of expiration (Champagnat, Denavit-Saubié, Moyanova & Rondouin, 1982; Richter, 1982; Richter, Ballantyne & Remmers, 1986; St John & Zhou, 1989; Pierrefiche et al. 1992; Oku et al. 1993). Although many neurones with a post-inspiratory discharge are undoubtedly motoneurones (Zheng, Barillot & Bianchi, 1991), indirect evidence based on the patterns of chloride-mediated IPSPs and glycinergic postsynaptic inhibition in other neurones (Richter et al. 1986; Schmid, Foutz & Denavit-Saubié, 1996) supports the idea that PI neurones exert a widespread inhibition on other respiratory neurones, particularly I neurones (Richter et al. 1986). Thus the IPSP observed on I neurones after the initial EPSP might originate from PI neurones. Furthermore, the timing of PI discharge just after inspiratory arrest, suggests that these neurones carry out the ‘irreversible’ phase of the IOS process, preventing the re-activation of phrenic discharge (Richter et al. 1986). However, intervention of other neurones (i.e. late-inspiratory) in the evoked IOS cannot be excluded.

SLN stimulation evokes a premature discharge of PI neurones

During spontaneous IOS, PI neurones started to depolarize 30-50 ms before the arrest of the phrenic activity, giving rise to an ‘off-switch potential’ (Fig. 3B, the slow slope in the window), whereas SLN stimulation evoked a massive, fast and synchronous synaptic activation of PI neurones which preceded the transient phrenic inhibition or the IOS. Thus, stimulation of peripheral afferents changed the normally slow respiratory-related depolarization of PI neurones into a fast and steep depolarization reaching quickly the threshold for action potentials. This is in keeping with previous works which showed that SLN stimulation activates PI neurones and tyroarytenoid motoneurones on a similar time scale (Remmers et al. 1986; St John & Zhou 1989), and that a brief activation of the tyroarytenoid nerve precedes the reactivation of the phrenic nerve when reversible phrenic inhibition is triggered by a short train of stimuli applied to the SLN (Oku et al. 1993). A strong activation of the discharge of PI neurones by various peripheral inputs during the post-inspiratory phase might be the mechanism through which these inputs modulate the respiratory rhythm (Remmers et al. 1986; Lawson, Richter, Ballantyne & Lalley, 1989). Therefore, we conclude that in normal conditions the IOS process starts by a slow ‘off-switch potential’ whereas during stimulation, synchronized activation of the inputs triggers the discharge of PI neurones which prematurely block inspiratory discharges within a few milliseconds, well before the onset of the ‘off-switch potential’. SLN stimulation activated PI motoneurones and non-antidromically responding PI neurones in a similar fashion. This is in keeping with previous results which did not reveal any difference in the pre-synaptic drive (cycle-related changes in membrane resistance or excitability) of motoneurones and presumed propriobulbar neurones with same patterns of cycle-related discharge (Pierrefiche et al. 1992; Haji et al. 1996b).

Evoked responses are not mediated by NMDA receptors

Dizocilpine administration decreases respiratory-related EPSPs and IPSPs throughout the respiratory cycle (Feldman, Windhorst, Anders & Richter, 1992, Haji et al. 1996b). This might be the main reason why the transient inhibition of phrenic activity evoked by SLN stimulation was prolonged after dizocilpine, because the EPSPs and IPSPs evoked in I and PI neurones showed no significant change. This ineffectiveness of systemically administered dizocilpine on the peripherally evoked PSPs supports the idea that NMDA receptors are not significantly involved in the pathways between the primary SLN afferents and the target I or PI neurones. Indeed our microiontophoretic study (Haji et al. 1996a) has provided evidence that the EPSPs evoked by peripheral afferent stimulation are mediated by glutamatergic synapses acting through non-NMDA receptors on the target neurones, and the IPSPs are mediated by GABAA receptors. The circuits and/or synapses of the evoked PSPs might thus not be directly involved in the spontaneous IOS, the occurrence of which is delayed by dizocilpine (Pierrefiche et al. 1992; Feldman et al. 1992; Haji et al. 1996b). Moreover, relay neurones in the NTS responsible for the short-latency excitation of phrenic motor output and subsequent IOS also involve activation of non-NMDA receptors (Karius, Ling & Speck, 1993, 1994; Karius & Speck, 1995). These observations suggest that the most likely target of dizocilpine action might be a presynaptic source or modulatory control of the respiratory neuronal network but not of the pathways responsible for the peripherally evoked PSPs. Such presynaptic control might originate from the pons (Ling, Karius & Speck, 1994), within a population of early-I neurones (Dick, Bellingham & Richter, 1994), whose accommodation of discharge might be blocked by dizocilpine (Pierrefiche et al. 1992; Pierrefiche, Champagnat & Richter, 1995).

We had previously shown that dizocilpine, while increasing inspiratory time, decreases the respiratory-related discharge and fluctuations of the membrane potential of PI neurones, but not their progressive depolarization preceding a naturally occurring IOS (Pierrefiche et al. 1992; Haji et al. 1996b). The present results show that NMDA receptor blockade does not impair their activation by peripheral inputs.


The present results are consistent with the hypothesis that stimulation of SLN afferents interrupts the inspiratory discharge by producing a synchronized, premature excitation of PI neurones. Dizocilpine depresses the respiratory-related potentials but not the evoked potentials of PI neurones, which might allow afferent inputs from peripheral afferents to terminate inspiration through activation of these neurones. Peripheral stimulation also triggered a long-latency brief excitation of phrenic nerve discharge and related PSPs on respiratory neurones, which may represent the first step of the inspiratory on-switch process, and was unaffected by NMDA receptor blockade.


  1. Top of page
  2. Abstract
  6. Acknowledgements

This work was supported by CNRS, DRET 95/091 (France) and a Grant-in-Aid for Scientific Research from the Ministry of Education, Science, Sports and Culture of Japan (no. 05680672) and a Grant for International Scientific Research Program (no. 07044238) of Japan.