Intracellular electrophysiological and morphological study of the medullary inspiratory neurons of the decerebrate rat

Phrenic motoneurons: output elements of a highly organized intraspinal network

pontomedullary respiratory network generates the respiratory pattern and relays it to bulbar and spinal respiratory motor outputs. The phrenic motor system controlling diaphragm contraction receives and processes descending commands to produce orderly, synchronous, and cycle-to-cycle-reproducible spatiotemporal firing. Multiple investigators have studied phrenic motoneurons (PhMNs) in an attempt to shed light on local mechanisms underlying phrenic pattern formation. I and colleagues (Marchenko V, Ghali MG, Rogers RF. Am J Physiol Regul Integr Comp Physiol 308: R916–R926, 2015.) recorded PhMNs in unanesthetized, decerebrate rats and related their activity to simultaneous phrenic nerve (PhN) activity by creating a time-frequency representation of PhMN-PhN power and coherence. On the basis of their temporal firing patterns and relationship to PhN activity, we categorized PhMNs into three classes, each of which emerges as a result of intrinsic biophysical and network properties and organizes the orderly contraction of diaphragm motor fibers. For example, early inspiratory diaphragmatic activation by the early coherent burst generated by high-frequency PhMNs may be necessary to prime it to overcome its initial inertia. We have also demonstrated the existence of a prominent role for local intraspinal inhibitory mechanisms in shaping phrenic pattern formation. The objective of this review is to relate and synthesize recent findings with those of previous studies with the aim of demonstrating that the phrenic nucleus is a region of active local processing, rather than a passive relay of descending inputs.

The Kölliker-Fuse nucleus gates the postinspiratory phase of the respiratory cycle to control inspiratory off-switch and upper airway resistance in rat

Lesion or pharmacological manipulation of the dorsolateral pons can transform the breathing pattern to apneusis (pathological prolonged inspiration). Apneusis reflects a disturbed inspiratory off-switch mechanism (IOS) leading to a delayed phase transition from inspiration to expiration. Under intact conditions the IOS is irreversibly mediated via activation of postinspiratory (PI) neurons within the respiratory network. In parallel, populations of laryngeal premotoneurons manifest the IOS by a brief glottal constriction during the PI phase. We investigated effects of pontine excitation (glutamate injection) or temporary lesion after injection of a GABA-receptor agonist (isoguvacine) on the strength of PI-pool activity determined from respiratory motor outputs or kinesiological measurements of laryngeal resistance in a perfused brainstem preparation. Glutamate microinjections into distinct parts of the pontine Kölliker-Fuse nucleus (KF) evoked a tonic excitation of PI-motor activity or sustained laryngeal constriction accompanied by prolongation of the expiratory phase. Subsequent isoguvacine microinjections at the same loci abolished PI-motor or laryngeal constrictor activity, triggered apneusis and established a variable and decreased breathing frequency. In summary, we revealed that excitation or inhibition of defined areas within the KF activated and blocked PI activity and, consequently, IOS. Therefore, we conclude, first, that descending KF inputs are essential to gate PI activity required for a proper pattern formation and phase control within the respiratory network, at least during absence of pulmonary stretch receptor activity and, secondly, that the KF contains large numbers of laryngeal PI premotor neurons that might have a key role in the regulation of upper airway resistance during reflex control and vocalization.

Respiratory cycle timing and fast inspiratory discharge rhythms in the adult decerebrate rat

In supracollicular decerebrate paralyzed adult rats, neural respiration was monitored by bilateral phrenic recordings. In the study of respiratory cycle timing, the effects of vagal afferent input (lung inflation) on respiratory phase durations resembled those seen in decerebrate cats. 1) Withholding lung inflation during neural inspiration (I) produced lengthening of I phase duration by 46% (mean, n = 11). 2) Maintaining lung inflation during neural expiration (E) produced lengthening of E phase duration by 112% (mean, n = 4). In the study of fast rhythms in inspiratory discharges, phrenic nerve autospectra and bilateral (left-right) phrenic coherences in 16 rats revealed two types of fast rhythm: 1) high-frequency oscillation (HFO), which had significant coherence peaks (n = 9, range 106-160 Hz, mean 132 Hz); and 2) medium-frequency oscillation (MFO), which had autospectral peaks but no distinct coherence peaks (n = 11, range 46-96 Hz, mean 66 Hz). These rhythms resembled MFOs and HFOs in the decerebrate cat, but the modal frequency range was about twice as large. In addition, these frequency values differed markedly from the 20-40 Hz of the rhythms found in earlier studies in neonatal in vitro preparations; the difference may be due to developmental immaturity.

Glutamic acid decarboxylase-immunoreactivity of bulbar respiratory neurons identified by intracellular recording and labeling in rats

To distinguish the GABAergic neuron in the ventral respiratory group (VRG) of rats, immunohistochemical staining of glutamic acid decarboxylase (GAD) was performed in neurons that had been individually identified by in vivo intracellular recording and labeling with neurobiotin. A total of five types of respiratory neurons were identified and labeled; augmenting inspiratory (aug-I, n=12), decrementing or early inspiratory (early-I, n=3), inspiration-expiration phase spanning or late inspiratory (late-I, n=3), decrementing expiratory or postinspiratory (PI, n=8), and augmenting or stage 2 expiratory (E2, n=3). In addition, expiration-inspiration phase-spanning or pre-inspiratory neurons (pre-I, n=2) were recorded, but not labeled. The membrane potential trajectory of each neuron type resembled that previously described in cat, suggesting a comparable neuronal organization between the two species. According to the axonal arborization, those labeled neurons were further classified as propriobulbar (6 aug-I, all early-I, all late-I, and 3 PI), bulbospinal (2 aug-I and all E2) and cranial-motor neurons (4 aug-I and 5 PI). GAD-immunoreactivity was consistently detected in the propriobulbar neurons, while it was not seen in cranial-motor and bulbospinal neurons. In addition, GAD-immunoreactive varicosities were found surrounding the somatic and dendritic surface of all labeled neurons. The present results illustrate that the propriobulbar types of early-I, aug-I, late-I and PI neurons are GABAergic inhibitory neurons and virtually all types of respiratory neurons receive GABAergic inputs in the rat's VRG.

Effects of serotonin on crossed phrenic nerve activity in cervical spinal cord hemisected rats

The present study investigates the effect of 5-hydroxytryptophan (5-HTP), a serotonin precursor, on crossed phrenic nerve activity (CPNA) in rats subjected to a left C2 spinal cord hemisection. Electrophysiological experiments were conducted on anesthetized, vagotomized, paralyzed, and artificially ventilated rats to assess phrenic nerve activity. The left phrenic nerve lost rhythmic activity due to the disruption of the bulbospinal respiratory pathways following spinal cord hemisection. Activity was induced in the left phrenic nerve (CPNA) by temporary asphyxia. 5-HTP administration increased CPNA during asphyxia in the left phrenic nerve in a dose-dependent fashion. Specifically, in a group of eight animals, application of 5-HTP at 0.5, 1.0, and 2.0 mg/kg significantly increased CPNA by 102.2+/-18.5%, 200.8+/-58.1%, and 615.0+/-356.9% compared with predrug control values, respectively. 5-HTP-induced increases in CPNA were reversed by methysergide (2-6 mg/kg, i.v.), a serotonin receptor antagonist. The results suggest that serotonin is involved in the modulation of crossed phrenic nerve activity following spinal cord injury.

Maturation of the mammalian respiratory system

In this review, the maturational changes occurring in the mammalian respiratory network from fetal to adult ages are analyzed. Most of the data presented were obtained on rodents using in vitro approaches. In gestational day 18 (E18) fetuses, this network functions but is not yet able to sustain a stable respiratory activity, and most of the neonatal modulatory processes are not yet efficient. Respiratory motoneurons undergo relatively little cell death, and even if not yet fully mature at E18, they are capable of firing sustained bursts of potentials. Endogenous serotonin exerts a potent facilitation on the network and appears to be necessary for the respiratory rhythm to be expressed. In E20 fetuses and neonates, the respiratory activity has become quite stable. Inhibitory processes are not yet necessary for respiratory rhythmogenesis, and the rostral ventrolateral medulla (RVLM) contains inspiratory bursting pacemaker neurons that seem to constitute the kernel of the network. The activity of the network depends on CO2 and pH levels, via cholinergic relays, as well as being modulated at both the RVLM and motoneuronal levels by endogenous serotonin, substance P, and catecholamine mechanisms. In adults, the inhibitory processes become more important, but the RVLM is still a crucial area. The neonatal modulatory processes are likely to continue during adulthood, but they are difficult to investigate in vivo. In conclusion, 1) serotonin, which greatly facilitates the activity of the respiratory network at all developmental ages, may at least partly define its maturation; 2) the RVLM bursting pacemaker neurons may be the kernel of the network from E20 to adulthood, but their existence and their role in vivo need to be further confirmed in both neonatal and adult mammals.

Brainstem neurons projecting to the rostral ventral respiratory group (VRG) in the medulla oblongata of the rat revealed by co-application of NMDA and biocytin

Groups of neurons in the medulla and pons are essential for the rhythm generation, pattern formation and modulation of respiration. The rostral Ventral Respiratory Group (rVRG) is thought to be a crucial area for rhythm generation. Here we co-applied biocytin and NMDA in the rVRG to label retrogradely brainstem neurons reciprocally connected to a population of inspiratory neurons in the rat rVRG. The procedure excited rVRG neurons in multi-unit recordings and led to a Golgi-like labelling of distant cells presumably excited by efferents from the rVRG. Injection of biocytin without NMDA did not label neurons in distant structures. Several brainstem ipsi- and contralateral structures were found to project to the rVRG, but three major respiratory-related structures, the nucleus of the solitary tract (NTS), the parabrachialis medialis and Kölliker-Fuse nuclei (PB/KF) and the caudal VRG, which are known to project bilaterally to the rVRG, were exclusively labelled ipsilaterally, suggesting an ipsilateral excitation of these structures by the rVRG. The pathways of efferent axons from labelled neurons in the rVRG were traced rostrally towards the pons and caudally to the spinal cord. Terminal axonal arborizations were seen in the same regions where retrogradely filled neurons were found as well as in a few other motor nuclei (the dorsal vagal motor nucleus and XII nucleus). Moreover, in the NTS and the PB/KF, efferent terminal varicosities were seen closely apposed to the soma and proximal dendrites of labelled neurons, suggesting monosynaptic connections between the rVRG and these nuclei.

Effect of cardiopulmonary C fibre activation on the firing activity of ventral respiratory group neurones in the rat

1. Cardiopulmonary C fibre receptor stimulation elicits apnoea and rapid shallow breathing, but the effects on the firing activity of central respiratory neurones are not well understood. This study examined the responses of ventral respiratory group neurones: decrementing expiratory (Edec), augmenting expiratory (Eaug), and inspiratory (I) neurones during cardiopulmonary C fibre receptor-evoked apnoea and rapid shallow breathing. 2. Extracellular neuronal activity, phrenic nerve activity and arterial pressure were recorded in urethane-anaesthetized rats. Cardiopulmonary C fibre receptors were stimulated by right atrial injections of phenylbiguanide. Neurones were tested for antidromic activation from the contra- and ipsilateral ventral respiratory group (VRG), spinal cord and cervical vagus nerve. 3. Edec neurones discharged tonically during cardiopulmonary C fibre-evoked apnoea and rapid shallow breathing, displaying increased burst durations, number of impulses per burst, and mean impulse frequencies. Edec neurones recovered either with the phrenic nerve activity (25 s) or much later (3 min). 4. By contrast, the firing activity of Eaug and most I neurones was decreased, featuring decreased burst durations and number of impulses per burst and increased interburst intervals. Eaug activity recovered in approximately 3 min and inspiratory activity in approximately 1 min. 5. The results indicate that cardiopulmonary C fibre receptor stimulation causes tonic firing of Edec neurones and decreases in Eaug and I neuronal activity coincident with apnoea or rapid shallow breathing.

Medullary post-inspiratory neuronal and diaphragm post-inspiratory activities during spontaneous augmented breaths in anesthetized rats

To test the hypothesis that during augmented breaths the prolonged expiratory time is correlated with the activity of medullary post-inspiratory neurons, changes in diaphragm electromyogram and post-inspiratory neuronal activities were examined in pentobarbital anesthetized rats. During augmented breaths, the mean total expiratory time significantly increased (from 0.57 +/- 0.02 to 1.20 +/- 0.07 s, P < 0.01), but the mean duration of post-inspiratory inspiratory activity (PIIA) was significantly decreased (from 0.08 +/- 0.01 to 0.02 +/- 0.003 s, P < 0.01) compared to control breaths. The mean time which the diaphragm was electrically silent was observed to be significantly prolonged during augmented breaths (from 0.48 +/- 0.02 to 1.19 +/- 0.07 s, P < 0.01). During significant prolongations of expiratory time in augmented breaths, most of the recorded neurons (9/10) in the medulla showed sustained discharges. The mean duration of discharges was longer than that of PIIA (0.02 +/- 0.003 vs. 0.50 +/- 0.10 s, P < 0.01). The mean firing frequency/breath significantly increased compared to control breaths (9.45 +/- 1.69 vs. 37.6 +/- 7.13 imp/breath, P < 0.01). These results suggested that during augmented breaths, the prolonged expiratory period may not be mediated by persisting post-inspiratory neuronal activity in the medulla.

The morphology and connections of neurons in the gasping centre of adult rats

Neuronal activities in the intermediate reticular nucleus and adjacent lateral tegmental field are critical for the neurogenesis of the ventilatory pattern of gasping. We report herein the anatomical features of these neurons, their axonal projections and the location of neurons providing afferent inputs. These neuroanatomical evaluations were performed by iontophoretic injection of the tracer Neurobiotin into the region of the intermediate reticular nucleus of the rat. At the site of injection, neurons having soma of 30-50 microns were filled. Labelled axons and terminals were observed in ipsilateral regions which contain neurons having established functions in the control of ventilatory activity. These regions include the nucleus ambiguous and motor nuclei of the hypoglossal and facial nerves. In addition, axonal projections extended to the contralateral region of the intermediate reticular nucleus. From this contralateral region, retrograde tracing revealed projections to the site of injection. Similarly, many ipsilateral regions which received axonal terminals from the region of the intermediate reticular nucleus had reciprocal projections to this region. These anatomical results support the physiological observation that the neurogenesis of gasping involves a synchronized activation of diverse components of the brainstem ventilatory control system.

Pontomedullary efferent projections of the ventral respiratory neuronal subsets of the rat

The pontomedullary trajectories of projections efferent from the ventral respiratory cell group were anterogradely labelled after discrete injections of Fluoro Ruby into three morphophysiologically identified subdivisions (Bötzinger complex, rostral inspiratory, and caudal expiratory cell groups). The anterogradely labelled varicosities were located in a variety of areas involved in cardiorespiratory function: other subdivisions of the ventral respiratory cell group, the parabrachial (medial, central, and external lateral), Kölliker-Fuse, and lateral paragigantocellular nuclei, A5, and perifacial areas. Although the target areas were similar for the three studied subdivisions, some differences of the location and densities of labelled varicosities were found. Anterogradely labelled fibre bundles were found bilaterally after all of the tracer injections. Three caudally efferent bundles passed through the ventral respiratory cell group, dorsal medullary, and paramedian reticular nuclei. A labelled fibre bundle also took an ascending route through the ventral respiratory cell group: it surrounded the facial nucleus, and then followed two different pathways, one coursing towards forebrain areas and the other to the parabrachial and Kölliker-Fuse complex. Bundles of efferent axons decussated mainly at medullary levels and to a lesser extent in the pons. In the contralateral medulla and pons these labelled fibre bundles followed pathways similar to those observed ipsilaterally. The three ventral respiratory neuronal subsets sent axonal projections through similar tracts, but within them they were topographically organized. The present data are discussed with respect to the circuitry involved in the mechanisms of cardiorespiratory and other visceral functions.

Bilateral connections from ventral group inspiratory neurons to phrenic motoneurons in the rat determined by cross-correlation

We examined the functional connections from inspiratory neurons in the ventrolateral medulla to phrenic motoneurons in 11 Sprague-Dawley rats with intact vagi anaesthetized with sodium pentobarbital and paralysed with pancuronium bromide. Cross-correlation histograms were computed between the extracellular activity of 70 of these neurons and the discharge of the phrenic nerves; 38 contralateral only, 10 ipsilateral only and 22 bilaterally. A total of 22 peaks were detected in the cross-correlation histograms. These were classified as indicating common activation or synaptic connection according to their latencies to onset, and those suggesting connections as monosynaptic or paucisynaptic according to their half-amplitude widths. Nine peaks suggesting monosynaptic excitation of phrenic motoneurons were detected for 7 ventral group inspiratory neurons; 4 with contralateral connections and not tested for ipsilateral connections, 2 with bilateral connections, and 1 with a monosynaptic contralateral connection and a paucisynaptic ipsilateral connection. Four peaks suggesting paucisynaptic excitation of phrenic motoneurons were detected for 3 ventral group inspiratory neurons; 1 with bilateral connections, 1 with a contralateral connection only, and 1 (previously mentioned) with a paucisynaptic ipsilateral connection and a monosynaptic contralateral excitation. The remaining 9 peaks were classified as due to a common activation of the phrenic motoneurons and the ventral group inspiratory neurons. Cross-correlation histograms were also computed between the left and right phrenic nerve discharges in 5 rats and all displayed central broad peaks indicative of common activation, possibly due to excitation from bilaterally projecting medullary inspiratory neurons. We concluded that there is a substantial bilateral excitation of phrenic motoneurons by inspiratory neurons in the ventral medullary group of the rat.

Brain stem projections by axonal collaterals to the rostral and caudal ventral respiratory group in the rat

The location of neurons projecting by axonal collaterals to the rostral and caudal ventral respiratory group (VRG) regions was determined after discrete injections of Fast blue and Diamidino yellow into the physiologically identified rostral inspiratory VRG and the caudal expiratory VRG areas, respectively. In contrast with single fluorochrome labeled neurons found throughout the rostro-caudal extent of the medulla and pons (in a variety of areas known to have cardiorespiratory function), double-labeled neurons were located in discrete ponto-medullary regions. The largest number of the double-labeled neurons was counted within the peripheral facial area, lateral paragigantocellular nucleus, and the VRG region, ipsi- and contralaterally to the injected side. Only a few double-labeled neurons were found within the ventrolateral and intermediate subnuclei of the solitary tract, medial parabrachial, and Kölliker-Fuse nuclei. The possible physiological implications of this neuronal network have also been emphasized.

Morphological study of long axonal projections of ventral medullary inspiratory neurons in the rat

The aim of this study was to examine medullary and spinal axonal projections of inspiratory bulbospinal neurons of the rostral ventral respiratory group (VRG) in the rat. A direct visualization of long (9.8-33 mm) axonal branches, including those projecting to the contralateral side of the medulla oblongata and the spinal cord, was possible due to intracellular labeling with neurobiotin and long survival times (up to 22 h) after injections. Seven of the nine labeled neurons had bilateral descending axons, which were located in discrete regions of the spinal white matter; ipsilateral axons in the lateral and dorsolateral funiculus, contralateral in the ventral and ventromedial funiculus. The collaterals issued by these axons at the mid-cervical level formed close appositions with dendrites of phrenic motoneurons, which had also been labeled with neurobiotin. None of these collaterals crossed the midline. The significance of this finding is discussed in relation to the crossed-phrenic phenomenon. Additional spinal collaterals were identified in the C1 and T1 segments. Within the medulla, collaterals with multiple varicosities were identified in the lateral tegmental field and in the dorsomedial medulla (in the hypoglossal nucleus and in the nucleus of the solitary tract). These results demonstrate that inspiratory VRG neurons in the rat have some features which have not been previously described in the cat, including frequent bilateral spinal projection and projection to the nucleus of the solitary tract. In addition, this study shows that intracellular labeling with neurobiotin offers an effective way of tracing long axonal projections, supplementing results previously obtainable only with antidromic mapping, and providing morphological details which could not be observed in previous studies using labeling with horseradish peroxidase.

Correlation analysis of respiratory neuron activity in ventrolateral medulla of brainstem-spinal cord preparation isolated from newborn rat

Cross-correlation analysis was used to study functional connections between one inspiratory (I) neuron and another, and between one pre-inspiratory (Pre-I) neuron and another, in 54 brainstem-spinal cord preparations isolated from newborn rats. Pre-I neurons usually fired in the pre- and post-inspiratory phases. Neurons were recorded extracellularly with pairs of microelectrodes placed on the same or opposite sides of the brainstem. Fourteen pairs of Pre-I neurons recorded bilaterally in the rostral ventrolateral medulla (RVL), 14 pairs of ipsilateral Pre-I neurons in the RVL, 14 pairs of bilateral I neurons in the RVL and 12 pairs of ipsilateral I neurons in the ventrolateral medulla were studied. Cross-correlation histograms (CCHs) were computed. Significantly high peak bin counts were detected in 24 of 54 pairs. Peaks on one side of the origin of the CCHs were observed for one pair of ipsilateral Pre-I neurons, four pairs of bilateral I neurons and five pairs of ipsilateral I neurons. These findings suggest mono- or oligo-synaptic excitatory connections between paired neurons or shared inputs. Only one trough suggesting an oligo-synaptic inhibitory connection was evident in a CCH obtained from the pair of bilateral I neurons. This CCH revealed the peak and the trough on opposite sides of the origin, which was consistent with reciprocal excitatory and inhibitory connections between recorded neurons. Peaks on both sides of the origin were observed for three pairs of bilateral I neurons. From auto-correlation analysis and the latencies of these peaks, two of the three CCHs were consistent with reciprocal excitatory connections between recorded neurons, whereas the other CCH suggests shared inputs. Peaks at the origin were observed for two pairs of ipsilateral Pre-I neurons, four pairs of bilateral I neurons and five pairs of ipsilateral I neurons. These results suggest shared inputs. For Pre-I neurons recorded in opposite sides, no significant bin counts were detected. Peaks on one side were detected for three pairs. Present results suggest short-term synchronisation of I neurons, and of Pre-I neurons via excitatory coupling, and the likelihood of comparatively strong interaction between I neurons, which may be important in maintaining the I burst.

Central respiratory modulation of facial motoneurons in rats

Facial motoneurons (FMN) were recorded intracellularly in Sprague-Dawley rats anesthetized with halothane. The animals were vagotomized, paralyzed, and artificially ventilated. The average membrane potential of the cells was 62.6 +/- 1.9 mV and their input impedance ranged from 5 to 30 M omega (9.8 +/- 1.1 M omega, n = 38). The membrane potential of most FMNs varied throughout the central respiratory cycle and four distinct patterns were detected. Type I (post-inspiratory) cells (21/44) showed a two-phase Cl(-)-mediated hyperpolarization during the respiratory cycle, one during central inspiration and the second during late expiration. Type II cells (early inspiratory, n = 10) showed early inspiratory depolarization. Type III (n = 2, stage-2 expiratory) cells displayed late expiratory depolarization and one cell (type IV or throughout inspiratory) exhibited expiratory Cl(-)-mediated hyperpolarization. The remaining 10 cells showed no detectable respiratory modulation. The results reflect the heterogeneity of the central respiratory modulation of FMNs and suggest that these cells receive both excitatory and inhibitory inputs from elements of the central respiratory pattern generating network.

Possible mutual excitatory couplings between inspiratory neurons in caudal ventrolateral medulla of brainstem-spinal cord preparation isolated from newborn rat

In in vitro brainstem-spinal cord preparations, projection of inspiratory neurons in the caudal ventrolateral medulla (CVL) was examined electrophysiologically, and connectivity between bilateral inspiratory neurons in the CVL was analyzed by pulse-cross correlation (PCC) analysis. CVL inspiratory neurons were found to project to the contralateral CVL and/or mainly ipsilateral spinal cord. PCC analysis revealed significant peaks with different latency on both sides of time zero in 3 of 8 pairs. Results were consistent with mono- or oligo-synaptic excitatory connections between bilateral inspiratory neurons in the CVL.

Medullary expiratory neurons in the decerebrate rat: an intracellular study

Intracellular recordings and labelings with horseradish peroxidase (HRP) of expiratory (E) neurons were performed in decerebrate, paralyzed, and ventilated rats. A total of 37 neurons were recorded, from which 4 cells and 1 axon were labeled. They were located in two regions of the ventrolateral medulla. One was in the rostral portion of the nucleus ambiguus just caudal to the facial nucleus, and the other in the nucleus retroambiguus at the level of the caudal medulla. These expiratory neurons had rhythmical changes in membrane potential similar to those reported in cat, i.e., a depolarization in the intervals between phrenic bursts which evolved in an augmenting (E-aug, n = 15), or bell-shaped or 'plateau' (E-all, n = 22) pattern until a rapid hyperpolarization at the start of inspiration. Both types were hyperpolarized during inspiration by chloride-dependent, inhibitory postsynaptic potentials (IPSPs) which were demonstrated in 17 neurons (10 E-aug and 7 E-all) from which reversal was obtained. Such IPSPs also existed during post-inspiration (stage I of expiration) in 4 of the 10 augmenting E neurons. They were identified by antidromic stimulation or HRP labeling, or both, as bulbospinal neurons (n = 2), cranial motoneurons (n = 4), or not antidromically activated (NAA) neurons (n = 31). All the identified bulbospinal neurons and the motoneurons exhibited an E-all pattern. The expiratory neurons of the caudal medulla had various projections as demonstrated with HRP labeling: one bulbospinal neuron with ipsilateral axon giving off intramedullary collaterals, and NAA neurons with rostral medullary projections or with axons crossing the midline.(ABSTRACT TRUNCATED AT 250 WORDS)