The activity of locus coeruleus (LC) neurones (n= 126) was examined in whole‐cell (conventional and amphotericin B‐perforated patch) recordings, and the relationship of this activity to the respiratory discharge recorded on the C4 or C5 phrenic nerve roots was determined at different CO2 concentrations (2 and 8 %; bath pH 7.8 and 7.2) in the in vitro brainstem‐spinal cord preparation of the neonatal rat (1–5 days old). In most neurones (n= 105) ongoing activity was modulated at respiratory frequency. Typically, this consisted of a phase of depolarization and increased discharge frequency synchronous with the phrenic burst, followed by a phase of hyperpolarization and inhibition of discharge (n= 94 of 105). The incidence of respiratory modulation decreased from 91 % on P1 to 57 % on P5. Bath application of the non‐nmDA receptor antagonist 6‐cyano‐7‐nitroquinoxaline‐2,3‐dione (CNQX; 5 μmmu;m) or the nmDA receptor antagonist DL‐2‐amino‐5‐phosphonovaleric acid (APV; 100 μmmu;m) abolished both phases of respiratory modulation. The hyperpolarizing phase alone was abolished by the adrenoceptor antagonists idazoxan (5 μmmu;m) or phentolamine (0.8 μmmu;m). These results indicate that excitatory amino acid pathways are involved in the transmission of both the excitatory and inhibitory components and that the latter involves in addition an α2‐adrenoceptor‐mediated pathway. Increasing the CO2 concentration from 2 to 8 % resulted in a shortening of expiratory duration and weakening or loss of respiratory‐phased inhibition; this was accompanied by depolarization, increased discharge frequency and, in those neurones where they were initially present (60 %), an increase in the frequency of subthreshold membrane potential oscillations. The depolarizing response was retained in the presence of tetrodotoxin (TTX, 0.2–1.0 μmmu;m). These results indicate that in this neonatal preparation LC neurones form part of the synaptically connected brainstem respiratory network, and that the LC constitutes a site of CO2‐ or pH‐dependent chemoreception.

This article presents a model of the neuronal network that generates the rhythm for respiratory movements. It incorporates new data on the synaptic activity and discharge properties of respiratory neurons in the brain stem and on the modulation of their excitability by nonsynaptic intrinsic membrane properties. The model allows a description of the control sequence that generates the rhythm.

1. Using the isolated medulla and spinal cord of the neonatal rat, the response to CO2 -induced changes in superfusate pH was examined in whole cell and perforated patch recordings from ventral medullary neurones which were identified by injection of Lucifer Yellow. The respiratory response to changing the CO2 concentration (from 2 to 8%) consisted of an increase in phrenic burst frequency, which could be accompanied by an increase, decrease or no change in burst amplitude. 2. Five classes of neurone -inspiratory, post-inspiratory, expiratory, respiration-modulated and tonic -were distinguished on the basis of their membrane potential and discharge patterns. Almost all (112 of 123) responded rapidly to 8% CO2 with a sustained change in membrane potential. Depolarizing responses (3-18 mV) occurred in inspiratory, respiration-modulated and 45% of tonic neurones. Hyperpolarizing responses (2-19 mV) occurred in expiratory and post-inspiratory neurones. The remaining tonic neurones were inhibited or showed no response. 3. In representatives of each class of neurone, membrane potential responses to 8 % CO2 were retained when tested in the presence of tetrodotoxin (n = 7), low (0-2 mM) Ca2P-high (5 mM) Mg2+ (n = 23) or Cd2+ (0-2 mM) (n = 3)-containing superfusate, implying that they are mediated by intrinsic membrane or cellular mechanisms.4. Neurones were distributed between 1200 /sm rostral and 400 ,um caudal to obex, and their cell bodies were located between 50 and 700 um below the ventral surface (n = 104).Almost all responsive neurones (n = 78) showed dendritic projections to within 50 ,um of the surface. 5. These experiments indicate that significant numbers of ventral medullary neurones, including respiratory neurones, are intrinsically chemosensitive. The consistency with which these neurones show surface dendritic projections suggests that this sensitivity may arise in part at this level.

Experiments were performed on anesthetized cats to test the theory that the interval between phrenic bursts is comprised of two phases, stage I and stage II of expiration. Evidence that these represent two separate neural phases of the central respiratory rhythm was provided by the extent to which stage duration is controlled individually when tested by superior laryngeal, vagus and carotid sinus nerve stimulation. Membrane potential trajectories of bulbar postinspiratory neurons were used to identify the timing of respiratory phases. Stimulation of the superior laryngeal, vagus and carotid sinus nerves during stage I of expiration prolonged the period of depolarization in postinspiratory neurons without significantly changing the durations of either stage II expiratory or inspiratory inhibition, indicating a fairly selective prolongation of the first stage of expiration. Changes in subglottic pressure, insufflation of smoke into the upper airway, application of water to the larynx or rapid inflation of the lungs produced similar effects. Sustained tetanic stimulation of superior laryngeal and vagus nerves arrested the respiratory rhythm in stage I of expiration. Membrane potentials in postinspiratory, inspiratory and expiratory neurons were indicative of a prolonged postinspiratory period. Thus, such an arrhythmia can be described as a postinspiratory apneic state of the central oscillator. The effects of carotid sinus nerve stimulation reversed when the stimulus was applied during stage II expiration. This was accompanied by corresponding changes in the membrane potential trajectories in postinspiratory neurons. The results manifest a ternary central respiratory cycle with two individually controlled phases occurring between inspiratory bursts.

Membrane potential trajectories of 68 bulbar respiratory neurones from the peri-solitary and peri-ambigual areas of the brain-stem were recorded in anaesthetized cats to explore the synaptic influences of post-inspiratory neurones upon the medullary inspiratory network. A declining wave of inhibitory postsynaptic potentials resembling the discharge of post-inspiratory neurones was seen in both bulbospinal and non-bulbospinal inspiratory neurones, including alpha- and beta-inspiratory, early-inspiratory, late-inspiratory and ramp-inspiratory neurones. Activation of laryngeal and high-threshold pulmonary receptor afferents excited bulbar post-inspiratory neurones, whilst in the case of inspiratory neurones such stimulation produced enhanced postsynaptic inhibition during the same period of the cycle. Activation of post-inspiratory neurones and enhanced post-inspiratory inhibition of inspiratory bulbospinal neurones was accompanied by suppression of the after-discharge of phrenic motoneurones. These results suggest that a population of post-inspiratory neurones exerts a widespread inhibitory function at the lower brain-stem level. Implications of such an inhibitory function for the organization of the respiratory network are discussed in relation to the generation of the respiratory rhythm.