Ventilatory response to hypercapnia and hypoxia after extensive lesion of medullary serotonergic neurons in newborn conscious piglets.

Acute inhibition of serotonergic (5-HT) neurons in the medullary raphé (MR) using a 5-HT(1A) receptor agonist had an age-dependent impact on the "CO(2) response" of piglets (33). Our present study explored the effect of chronic 5-HT neuron lesions in the MR and extra-raphé on the ventilatory response to hypercapnia and hypoxia in piglets, with possible implications on the role of 5-HT in the sudden infant death syndrome. We established four experimental groups. Group 1 (n = 11) did not undergo any treatment. Groups 2, 3, and 4 were injected with either vehicle or the neurotoxin 5,7-dihydroxytryptamine in the cisterna magna during the first week of life (group 2, n = 9; group 4, n = 11) or second week of life (group 3, n = 10). Ventilation was recorded in response to 5% CO(2) (all groups) and 12% O(2) (group 2) during wakefulness and sleep up to postnatal day 25. Surprisingly, the piglets did not reveal changes in their CO(2) sensitivity during early postnatal development. Overall, considerable lesions of 5-HT neurons (up to 65% decrease) in the MR and extra-raphé had no impact on the CO(2) response, regardless of injection time. Postlesion raphé plasticity could explain why we observed no effect. 5,7-Dihydroxytryptamine-treated males, however, did present a lower CO(2) response during sleep. Hypoxia significantly altered the frequency during sleep in lesioned piglets. Further studies are necessary to elucidate the role of plasticity, sex, and 5-HT abnormalities in sudden infant death syndrome.

Central chemosensitivity, sleep, and wakefulness.

Neurons in many regions of the lower brain are chemosensitive in vitro. Focal acidification of these same and other regions in vivo can stimulate breathing indicating the presence of chemoreception. Why are there so many sites for central chemoreception? This review evaluates data obtained from unanesthetized rats at three central chemoreceptor sites, the retrotrapezoid nucleus (RTN), the medullary raphé, and the nucleus tractus solitarius (NTS) and extends ideas concerning two hypotheses, which were recently formulated (Nattie, E., 2000. Respir. Physiol. 122, 223-235). (1) The high overall sensitivity of the respiratory control system in the unanesthetized state to small increases in arterial CO(2) relies on an additive or greater effect of these multiple chemoreceptor sites. (2) Chemoreceptor sites can vary in effectiveness dependent on the state of arousal. These ideas fit into a more speculative and general hypothesis that central chemoreceptors are organized in a hierarchical manner as proposed for temperature sensing and thermoregulation (Satinoff, E., 1978. Science 201, 16-22). The presence of a number of chemosensitive sites with varying thresholds, sensitivity, and arousal dependence provides finely tuned control and stability for breathing.

Spontaneous arousals during quiet sleep in piglets: a visual and wavelet-based analysis.

STUDY OBJECTIVES: The objectives of the study were to characterize spontaneous arousals during NREM sleep in piglets and to compare two methods of identifying these events: a "visual" technique using spectral analysis and an automated technique using wavelets. Our goal was to understand the benefits and limits of these methods when applied to sleep in human infants. DESIGN: Arousals were identified by evaluating rapid changes in EEG low frequency activity, blood pressure (BP), and heart rate (HR). A cortical arousal was defined as a rapid decrease in EEG low frequency activity. An autonomic arousal was defined by a transient increase in heart rate or a transient change in mean arterial BP (MAP). SETTING: Laboratory study in sleeping and awake piglets. PARTICIPANTS: Five 1-2 week old piglets. INTERVENTIONS: Chronically instrumented with a femoral arterial line, EEG, EOG, EMG electrodes, and a micro-dialysis probe with its tip located in the rostral ventral medulla. Artificial CSF (aCSF) was dialyzed into the RVM throughout the experiments Measurements: For the visual analysis, the average delta power (0.5-4 Hz) for each 5-second epoch was determined using spectral analysis. MAP and HR were analyzed in 1-second bins. Video images were analyzed for body movements and eye openings. Transient changes in blood pressure, HR, and delta power were then visually identified. For the wavelet analysis, a quantitative, automated technique with a defined "wakefulness threshold" was used to identify rapid decreases in EEG low frequency activity and the rate of change of MAP. RESULTS: Using the visual method, 117 episodes associated with stereotypical hemodynamic, EEG, and behavioral changes (startle) were identified. Seventy five events occurred in isolation or were first in a series of "multiple" events, 41 "multiple" events were defined as events occurring <20 seconds following a previous event. Eighteen events were associated with the termination of apnea. In isolated events or those occurring first in a series, the onset of changes in HR and BP clearly preceded the decrease in EEG amplitude and delta power. Using wavelet analysis, 73 EEG arousals and 115 MAP transients were identified independently; 62% of the EEG events were associated with a transient change in MAP and HR, and in these cases the onset of the hemodynamic events preceded EEG arousals. EEG arousals and MAP transients, however, also occurred alone and not associated with a stereotypical pattern of a startle, changes in MAP and HR and the EEG. CONCLUSIONS: Many of these spontaneous arousals represent integrated EEG, hemodynamic, and behavioral processes similar to arousal phenomena described in adult rats and human infants, but the pattern of spontaneous arousals appears to be more heterogeneous than has been described for arousals induced by exogenous stimuli. Both the visual and wavelet analysis identified these events, but the wavelet technique has the potential advantage of better time resolution and automation of the analysis.

The effects of a GABA(A) agonist in the rostral ventral medulla on sleep and breathing in newborn piglets.

STUDY OBJECTIVES: Abnormalities in the rostral ventral medulla (RVM) in human infants may contribute to the etiology of the sudden infant death syndrome (SIDS) or a subset of SIDS, by interfering with cardiorespiratory and arousal responses to physiological stimuli often encountered during sleep. The purpose of this study was to determine whether inhibition of groups of neurons in the RVM in newborn piglets would alter sleep and/or the sleep-modulation of breathing. We hypothesized that inhibition of neurons in the RVM would produce less wakefulness or increase the low frequency power (delta) during Quiet sleep. DESIGN: Unanesthetized piglets were studied in a whole-body plethysmograph. Artificial cerebral spinal fluid (aCSF) or the GABAA agonist, muscimol, was dialyzed into the RVM for 40 minutes after a control period consisting of aCSF dialysis. Sleep was analyzed using a combination of EEG spectral analysis and behavioral observations. SETTING: N/A. PARTICIPANTS: N/A. INTERVENTIONS: N/A. MEASUREMENTS AND RESULTS: Cardiorespiratory variables varied with state. Dialysis of neither aCSF nor muscimol into the RVM resulted in alterations in resting respiration, BP, HR, or VO2 or their modulation by state. Compared to control dialysis with aCSF, muscimol dialysis caused dramatic effects on sleep architecture. Sleep cycling was abolished in some experiments, whereas in others there were decreases in low-frequency EEG activity or delta power. The animals in which sleep cycling ceased continued in a perpetual state of drowsiness interspersed with periods of wakefulness. CONCLUSIONS: We conclude that dialysis of muscimol into the RVM has little effect on resting breathing, blood pressure, or heart rate or their modulation by state, but interferes with normal sleep architecture. We speculate that abnormalities in the ventral medulla may alter sleep cycling or interfere with arousal mechanisms, thus contributing to the etiology of at least a subset of SIDS.

CO2 dialysis in the medullary raphe of the rat increases ventilation in sleep.

Central chemoreceptors are widespread within the brain stem. We hypothesize that function at different sites varies with arousal state. In unanesthetized rats, we produced focal acidification at single sites by means of microdialysis using artificial cerebrospinal fluid equilibrated with 25% CO2. Tissue acidosis, measured under anesthesia, is equivalent to that observed with 63 Torr end-tidal PCO2 and is limited to 600 microm. Focal acidification of the retrotrapezoid nucleus increased ventilation by 24% only in wakefulness via an increase in tidal volume (Li A, Randall M, and Nattie E. J Appl Physiol 87: 910-919, 1999). In this study of the medullary raphe, the effect of such focal acidification was in sleep (defined by electroencephalographic and electromyographic criteria): ventilation and frequency increased by 15-20% in non-rapid eye movement sleep, and frequency increased by 15% in rapid eye movement sleep. There was no effect in wakefulness. Chemoreception in the medullary raphe appears to be responsive in sleep. Central chemoreceptors at two different locations appear to vary in effectiveness with arousal state.

Muscimol dialysis in the rostral ventral medulla reduced the CO(2) response in awake and sleeping piglets.

Some victims of sudden infant death syndrome have arcuate nucleus abnormalities. The arcuate nucleus may be homologous with ventral medullary structures in the cat known to be involved in the control of breathing and the response to systemic hypercapnia. We refer to putative arcuate homologues in the piglet collectively as the rostral ventral medulla (RVM). We inhibited the RVM in awake and sleeping, chronically instrumented piglets by microdialysis of the GABA(A) receptor agonist muscimol. Muscimol dialysis (10 and 40 mM) had no effect on eupnea but caused a significant reduction in the response to hypercapnia during both wakefulness (34.8 +/- 8.7 and 30.7 +/- 10.1%, respectively) and sleep (36.7 +/- 6.7 and 49.5 +/- 8.9%, respectively). The effect of muscimol on the CO(2) response was entirely via a reduction in tidal volume and appeared to be greater during non-rapid-eye-movement sleep. We conclude that the piglet RVM contains neurons of importance in the response to systemic CO(2) during both wakefulness and non-rapid-eye-movement sleep. We hypothesize that dysfunction of homologous regions in the human infant could lead to impaired ability to respond to hypercapnia, particularly during sleep, which could potentially be involved in the pathogenesis of sudden infant death syndrome.

Lesion or muscimol in the rostral ventral medulla reduces ventilatory output and the CO(2) response in decerebrate piglets.

Developmental abnormalities have been described in the arcuate nucleus of sudden infant death syndrome (SIDS) victims. The arcuate nucleus has putative homologues in chemosensitive areas of the ventral medulla in animals. We refer to some of these areas collectively as the rostral ventral medulla (RVM). In the RVM of decerebrate piglets 2-15 days of age, we studied the effects of electrolytic lesions (n=7) or microdialysis of muscimol (n=15), a GABAA receptor agonist, on ventilatory output and the response to hypercapnia. Lesions caused a 66.7+/-17.3% reduction in eupneic phrenic minute activity (MA) and abolished the response to hypercapnia. Muscimol dialysis caused a 32.4+/-10.4% reduction in MA with a significant downward displacement of the response to hypercapnia with no significant effect on the slope. We conclude that the piglet RVM contains neurons of vital importance in the maintenance of normal breathing and the response to systemic CO(2). We hypothesize that dysfunction of homologous regions in the human infant could lead to impaired ability to respond to hypercapnia and could potentially be involved in the pathogenesis of SIDS.

CO(2) microdialysis in retrotrapezoid nucleus of the rat increases breathing in wakefulness but not in sleep.

Central chemoreceptors are widespread within the brain stem. We suggest that their function at some sites may vary with the state of arousal. In this study, we tested the hypothesis that the function of chemoreceptors in the retrotrapezoid nucleus (RTN) varies with sleep and wakefulness. In unanesthetized rats, we produced focal acidification of the RTN by means of a microdialysis probe (tip containing the semipermeable membrane = 1-mm length, 240-microm diameter, and 45-nl volume). With the use of a dialysate equilibrated with 25% CO(2), the tissue pH change (measured in anesthetized animals) was 1) limited to within 550 microm of the probe and, 2) at the probe tip, was equivalent to that observed with end-tidal PCO(2) of 63 Torr. This focal acidification of the RTN increased ventilation significantly by 24% above baseline, on average, in 13 trials in seven rats only during wakefulness. The effect was entirely due to an increase in tidal volume. During sleep defined by behavioral criteria, ventilation was unaffected, on average, in 10 trials in seven rats. During sleep, the chemoreceptors in the RTN appear to be inactive, or, if active, the respiratory control system either is not responding or is responding with very low gain. Because ventilation is increased during sleep with all central chemoreceptor sites stimulated via systemic CO(2) application, other central chemoreceptor locations must have enhanced effectiveness.

Brain stem lesion size determined by DEAD red or conjugation of neurotoxin to fluorescent beads.

Neurotoxin microinjected into the retrotrapezoid nucleus of anesthetized rats decreases phrenic activity and eliminates the response to CO2. In unanesthetized rats, such treatment has no effect on awake, resting breathing and decreases CO2 sensitivity by 40% (M. Akilesh, M. Kamper, A. Li, and E. E. Nattie. J. Appl. Physiol. 82: 469-479, 1997). One important factor in explaining these disparate results is the actual size of the anatomic lesion. In the present study, we injected ibotenic acid into the retrotrapezoid nucleus of anesthetized rats and evaluated lesion size by using two new approaches: 1) DEAD red, a fluorescent probe that enters impaired cells through leaky membranes and binds to nucleic acids, and 2) conjugation of toxin to fluorescent beads. With the use of DEAD red, the region containing labeled dying cells was 313 +/- 104 nl (n = 4), six times larger than the initial injected volume, and the physiological effects on phrenic amplitude, the CO2 response, and blood pressure began within minutes and were substantial. With conjugated toxin, in theory, neuronal damage would be limited to the region of detectable fluorescence (49 +/- 10 nl; n = 4). Effects on phrenic amplitude, CO2 sensitivity, and blood pressure were absent until approximately 2 h postinjection. Control experiments, with 2 h of in vitro incubation of the neurotoxin-microbead conjugate and injection of the supernatant after centrifugation, showed similar results that suggest release of conjugated neurotoxin. We conclude that DEAD red provides a useful means to monitor neuronal impairment in acute studies in vivo. Conjugation of neurotoxin to microbeads may be less reliable in this regard.

Ventilatory effects of glial dysfunction in a rat brain stem chemoreceptor region.

Glia are thought to be important in brain extracellular fluid ion and pH regulation, but their role in brain stem sites that sense pH and stimulate breathing is unknown. Using a diffusion pipette, we administered the glial toxin, fluorocitrate (FC; 1 mM) into one such brain stem region, the retrotrapezoid nucleus (RTN) for 45-60 min. This dose and time period were chosen so that the effects of FC would be largely reversible. Within minutes, tissue pH decreased, and respiratory output increased. Both recovered almost completely after cessation of FC administration. The response to systemic CO2 stimulation was unaffected by FC treatment compared with that following control diffusion. Anatomic analysis showed, at the center of FC administration, some small (mean diameter = 5.1 micrometer) cells that stained for DEAD Red, a marker for altered cell membrane permeability, and some fragmented glia (glial fibrillary acidic protein immunohistochemistry). The average RTN tissue volume that contained such DEAD Red-positive cells was 271 nl, approximately 23% of the volume of one RTN region. Reversible disruption of glia in the RTN, a region known to contain central chemoreception, results in an acidic local pH and in stimulation of respiratory output.

RTN TRH causes prolonged respiratory stimulation.

We injected thyrotropin-releasing hormone (TRH; 10 nl; 0.25, 0.5, 1.0, or 10 mM), its inactive free acid form (TRHOH; 1 mM), or a metabolite with low TRH-receptor binding affinity, histidine-proline diketopiperazine (cHP; 1 mM), into the retrotrapezoid nucleus of anesthetized rats. Injection location was verified by anatomic analysis. Lower doses (0.25-0.5 mM) significantly increased both the product of integrated phrenic amplitude and frequency (Phr . f) and f for 20-30 min compared with artificial cerebrospinal fluid control injections. Higher doses (1. 0-10 mM) produced greater and long-lasting stimulation of Phr . f, Phr, and f and of blood pressure. This stimulation reached values 150% of baseline and durations of 270 min after a single injection. TRHOH (1 mM ) or cHP (1 mM) had no effect on Phr but increased f, as did 1 mM TRH. We conclude that TRH has a very powerful stimulatory effect in the retrotrapezoid nucleus region on Phr . f, with the Phr response seemingly specific for TRH receptors. Similar responses of f to TRHOH and cHP suggest it may be nonspecific.

Focal central chemoreceptor sensitivity in the RTN studied with a CO2 diffusion pipette in vivo.

We describe and use a CO2 diffusion pipette to produce a quickly reversible focal acidosis in the retrotrapezoid nucleus region of the rat brain stem. No tissue injection is made. Instead, artificial cerebrospinal fluid (aCSF) equilibrated with CO2 circulates within the micropipette, providing a source for continued CO2 diffusion into the tissue from the pipette tip. Tissue pH electrodes show the acidosis is limited to 500 micron from the tip. In controls (aCSF equilibrated with air), 1-min pipette perfusions increased tissue pH slightly and decreased phrenic nerve amplitude. In moderate- and high-CO2 groups (aCSF equilibrated with 50 or 100% CO2), 1-min perfusions significantly decreased tissue pH and increased phrenic nerve amplitude in a dose-dependent manner. The responses developed and reversed within minutes. Compared with our prior use of medullary acetazolamide injections to produce a focal acidosis, in this approach the acidosis 1) arises and reverses quickly and 2) its intensity can be varied. This allows study of sensitivity and mechanism. We conclude from this initial experiment that retrotrapezoid nucleus region chemoreceptors operate within the normal physiological range of CO2-induced tissue pH changes.

Effects of unilateral lesions of retrotrapezoid nucleus on breathing in awake rats.

In anesthetized rats, unilateral retrotrapezoid nucleus (RTN) lesions markedly decreased baseline phrenic activity and the response to CO2 (E. E. Nattie and A. Li. Respir. Physiol. 97:63-77, 1994). Here we evaluate the effects of such lesions on resting breathing and on the response to hypercapnia and hypoxia in unanesthetized awake rats. We made unilateral injections [24 +/- 7 (SE) nl] of ibotenic acid (IA; 50 mM), an excitatory amino acid neurotoxin, in the RTN region (n = 7) located by stereotaxic coordinates and by field potentials induced by facial nerve stimulation. Controls (n = 6) received RTN injections (80 +/- 30 nl) of mock cerebrospinal fluid. A second control consisted of four animals with IA injections (24 +/- 12 nl) outside the RTN region. Injected fluorescent beads allowed anatomic identification of lesion location. Using whole body plethysmography, we measured ventilation in the awake state during room air, 7% CO2 in air, and 10% O2 breathing before and for 3 wk after the RTN injections. There was no statistically significant effect of the IA injections on resting room air breathing in the lesion group compared with the control groups. We observed no apnea. The response to 7% CO2 in the lesion group compared with the control groups was significantly decreased, by 39% on average, for the final portion of the 3-wk study period. There was no lesion effect on the ventilatory response to 10% O2. In this unanesthetized model, other areas suppressed by anesthesia, e.g., the reticular activating system, hypothalamus, and perhaps the contralateral RTN, may provide tonic input to the respiratory centers that counters the loss of RTN activity.

Central chemoreception in the region of the ventral respiratory group in the rat.

We injected acetazolamide (AZ; 5 x 10(-6) M, 1 nl) into the region of the ventral respiratory group (VRG) of anesthetized paralyzed ventilated rats. Control injections (mock cerebrospinal fluid, n = 6, or the inactive AZ analogue 2-acetylamino-1,3, 4-thiadiazole-5-sulfon-t-butylamide, n = 6) did not increase the integrated phrenic neurogram [phrenic nerve amplitude (PNA)]. The AZ injections produced a focal region of tissue acidosis with a radius < 300-400 microns and are used as a probe for sites of central chemosensitivity. Injection location is determined by anatomic analysis. Of 22 VRG injections of AZ, 14 increased the amplitude of the PNA over 15-90 min; 8 had no effect. In 17 cases, we measured medullary tissue pH at the injection center and/or at a distant site and reaffirmed the size of the acidotic region produced by such small AZ injections. Of injections with pH electrodes within 300-400 microns of the injection center, all responders showed an acid pH; three nonresponders showed an acid pH, and one an alkaline pH. In a subgroup of five rats, at VRG sites with known respiratory effects identified by prior glutamate injection (10 nl, 100 mM), all subsequent AZ injections produced a PNA response. Simultaneous measurement of PNA and tissue pH responses at the injection center of eight rats did not show a uniform correlation in time; initially, both changed with a similar time course, but PNA recovered more quickly. We conclude that 1) the region of the VRG contains sites of ventilatory chemoreception, 2) ineffective AZ injections do produce a tissue acidosis but at sites with minimal impact on breathing, and 3) tissue pH does not uniquely represent the chemoreceptor stimulus.

Phrenic response to hypercapnia in the unanesthetized, decerbrate, newborn rat.

We developed a decerebrate, vagotomized, newborn rat preparation to investigate brainstem respiratory control mechanisms without the influence of anesthesia, supra-pontine structures, or vagally mediated feedback mechanisms. We measured the changes in phrenic nerve electrical activity in response to breathing 3% and 5% CO2 in unanesthetized, vagotomized, decerebrate newborn rats from 0 to 10 days of age and compared them with the changes in anesthetized, vagotomized, newborn rats and adult, vagotomized, decerebrate or anesthetized, animals. Phrenic nerve activity was irregular in the young newborn rats and became more regular between 7 and 10 days of age. T1 and T1/Ttot increased with age but increasing age had no influence on the response to CO2. The response to CO2 was dominated by increases in phrenic amplitude, minute activity, and inspiratory slope with no change in timing variables. These responses are similar to those that have been reported previously in vagally intact animals, suggesting that vagal feedback contributes little to the response to hypercapnia in the newborn rat. In summary, decerebrate newborn rats consistently respond to hypercapnia by increasing inspiratory drive similar to conscious animals.

Evidence for central chemoreception in the midline raphé.

We injected acetazolamide (AZ; 5 x 10(-6) M; 1 nl; n = 14), its inactive analogue 2-acetylamino-1,3,4-thiadiazole-5-sulfon-t-butylamide (5 x 10(-5) M; n = 6), or mock cerebrospinal fluid (n = 5) into the caudal raphé in the midline brain stem of anesthetized paralyzed ventilated rats. These AZ injections have been shown to produce a focal region of tissue acidosis with a radius < 350 microns and are used as a probe for sites of central chemosensitivity. Compared with control injections, AZ injection into the raphé, as demonstrated by anatomic analysis of injection location, significantly increased the amplitude of the integrated phrenic neurogram over 10-40 min. Not all raphé injections produced such a response. AZ injections identified as responders (n = 8 of 14) increased integrated phrenic amplitude 43.3 +/- 10.7% (SE) of baseline 20 min after the injection. We conclude that the midline caudal raphé contains sites of ventilatory chemoreception.

Prolonged stimulation of respiration by brain stem metabotropic glutamate receptors.

Stimulation of metabotropic glutamate receptors (mGluRs) in the retrotrapezoid nucleus (RTN) of chloralose-urethan-anesthetized rats by the mGluR agonist (1S,3R)-aminocyclopentanedicarboxylic acid [(1S,3R)-ACPD; 10 nl, 1 mM] increases integrated phrenic nerve amplitude (PNA) for > 60 min. Here we ask if the mGluR antagonist (+-)-alpha-methyl-4-carboxyphenylglycine [(+-)-MCPG] can block this effect. Using multibarreled micropipettes, we first identified RTN sites that affect respiration by noting short-lived stimulation of PNA produced by brief-duration injection of glutamate (10 nl, 100 mM), a presumed ionotropic receptor response. We then injected the active or inactive isomer of (+-)-MCPG (10 nl, 10 mM) followed by (1S,3R)-ACPD or by a long-duration (60-s) injection of glutamate (10 nl, 100 mM) at the same site. Each injection location was verified anatomically. The active antagonist (+-)-MCPG itself had no significant effect on PNA, but it blocked 1) subsequent (1S,3R)-ACPD-induced PNA stimulation for 99 +/- 11 (SE) min and 2) the PNA response to 60-s glutamate injection for 66 +/- 6 min. The inactive form (-)-MCPG had no effect on (1S,3R)-ACPD-induced PNA stimulation. We conclude that 1) RTN mGluRs may be involved in respiratory control, 2) RTN mGluRs are not active in eupnea, and 3) stimulation of RTN mGluRs may require prolonged glutamate release.

Rat retrotrapezoid nucleus iono- and metabotropic glutamate receptors and the control of breathing.

We injected 10 nl (unilateral) of glutamate receptor antagonists or agonists into the region of the retrotrapezoid nucleus and measured the phrenic nerve and blood pressure responses. The rats were chloralose-urethan anesthetized, paralyzed, vagotomized, and ventilated, and each injection location was verified anatomically. Integrated phrenic amplitude was most reliably affected. The N-methyl-D-aspartic acid (NMDA) antagonists 2-amino-5-phosphonopentanoic acid and 6-cyano-7-nitroquinoxaline-2,3-dione (which affects both NMDA and non-NMDA receptors) both decreased baseline eucapnic phrenic amplitude and the CO2 response. Glutamate increased phrenic amplitude in a dose-dependent manner, an effect blocked by prior injection of the NMDA and non-NMDA antagonists at the same site. The response duration depended on the duration of the glutamate injection: responses to 3-s injections lasted a few minutes, and responses to 60-s injections lasted for > 30 min. The long-lasting effect was reproduced by injection of the metabotropic agonist 1(S),3(R)-aminocyclopentanedicarboxylic acid at 0.01-0.02 times the glutamate dose. We conclude that the rat retrotrapezoid nucleus has an endogenous source of glutamate that maintains eucapnic phrenic output and allows expression of the CO2 response. NMDA and possibly non-NMDA receptors are involved. Their stimulation increases phrenic output via ionotropic and metabotropic receptor processes with the latter resulting in long-lasting phrenic stimulation.