Respiration Physiology 122 (2000) 237 246 www.elsevier.

com/locate/resphysiol

Functional synaptic connections among respiratory neurons

James Dufn *, Guo-Feng Tian, John H. Peever
Department of Physiology and Department of Anaesthesia, Uni6ersity of Toronto, Toronto, Ont., Canada, M5S 1A8 Accepted 23 February 2000

Abstract This presentation focuses on the application of methods to determine functional connections between neurons in the respiratory network of adult decerebrate rats. We employ a general network investigation paradigm that rst examines the intracellular recordings of a respiratory neuron and then determines which neurons synapse with it to produce the observed membrane potential changes. It is used to pursue the source of respiratory excitation and inhibition from its arrival at phrenic motoneurons to respiratory neurons in the medulla, and then examine some of the interactions among these neurons that shape their patterns of activity. Findings include a demonstration that phrenic motoneuron activity is determined by excitation from medullary inspiratory premotor neurons and inhibition by Botzinger complex expiratory neurons, and that the latter neurons inhibit both medullary inspiratory premotor neurons and themselves. We conclude that these functional interconnections explain the activity patterns of some respiratory neurons, but the connections between neurons thought to be involved in rhythm generation remain to be demonstrated in adult rats. 2000 Elsevier Science B.V. All rights reserved.
Keywords: Control of breathing, respiratory network; Mammals, cat; Network, neuronal, respiration

1. Introduction The goal of determining how respiratory rhythm is generated, shaped and transmitted to respiratory muscles has been pursued in a great number of ways. These range from anatomical tracing to complex signal analysis, and use preparations ranging from the whole animal to tissue slices. One approach is to construct the functional interconnections among neurons of the central

* Corresponding author. Tel.: +1-416-9786379; fax: +1416-9784940. E-mail address: j.dufn@utoronto.ca (J. Dufn).

respiratory system in order to deduce the system operation from a network perspective. We describe here, recent ndings from our laboratory using this approach in adult decerebrate rats. The respiratory rhythm generator has been the subject of numerous investigations as recent reviews attest (Bianchi et al., 1995; Dufn et al., 1995; Onimaru et al., 1997; Ramirez et al., 1997; Rekling and Feldman, 1998; St. John, 1998; Ballanyi et al., 1999; Feldman and McCrimmon, 1999; Hilaire and Duron, 1999). Some investigators name a region spanning the ventral respiratory group (VRG) and Botzinger complex, the pre-Botzinger complex, as containing the kernel for rhythm generation (Smith et al., 1991). It is

0034-5687/00/$ - see front matter 2000 Elsevier Science B.V. All rights reserved. PII: S 0 0 3 4 - 5 6 8 7 ( 0 0 ) 0 0 1 6 2 - 6

238

J. Dufn et al. / Respiration Physiology 122 (2000) 237246

generally agreed (Dufn et al., 1995) that the essential area for rhythm generation lies in the rostral ventrolateral medulla (RVLM). Recent experiments in cats using kainic acid lesions (Hsieh et al., 1998) as well as a novel calcium channel blocker (Ramirez et al., 1998) also implicate the RVLM as key to the generation of respiratory rhythm. It is known to contain early-burst inspiratory neurons with a decrementing pattern of discharge (I-DEC) and expiratory neurons of the Botzinger complex (E-BOT) neurons (Ezure, 1990). By consensus, these neurons as well as early expiratory (post-inspiratory) neurons with a decrementing pattern of discharge (E-DEC) are thought to be the essential elements of respiratory rhythm generation in adult animals. In early models of respiratory rhythm generation, originating mainly from experiments in adult cats, it was hypothesised that an oscillator was formed by mutual inhibition between inspiratory and expiratory neuron populations. Many recent network models have been proposed based on this concept, choosing various combinations of neurons for the mutual inhibition. In two-phase models the respiratory cycle is divided into inspiration and expiration, and mutual inhibition is between I-DEC and E-DEC neurons (Ezure, 1990) or IDEC and E-BOT neurons (Dufn, 1991). In the three-phase model expiration is subdivided into early (stage E1) and late (stage E2) phases, and mutual inhibition is between I-DEC, E-DEC and E-BOT (Richter et al., 1992). Simulations of these models range from the simple to the complex (Dufn, 1991; Ogilvie et al., 1992; Balis et al., 1994; Gottschalk et al., 1994; Bianchi et al., 1995; Dufn et al., 1995; Smith, 1995; Ryback et al., 1997; Matsugu et al., 1998). The connections from some of the key neurons of these models to other respiratory neurons have been studied in cats; see reviews (Ezure, 1990; Bianchi et al., 1995), but information about interconnections among the key neurons themselves is scarce. Intracellular recordings may suggest such interconnections, because the ring of one type of neuron coincides with membrane potential changes in another (Richter et al., 1992), but such inference by coincidence does not constitute demonstration of a connection. Cross-correlation

experiments in cats have conrmed some hypothesised connections (Lindsey et al., 1989b) as have spike-triggered averages of intracellular potentials. For example, Ezure (1990) provided evidence for an inhibitory connection from I-DEC to E-BOT neurons using spike-triggered averaging, and we (Dufn and Douse, 1993) obtained crosscorrelation evidence for an inhibitory connection from E-BOT to I-DEC, thereby verifying a connection hypothesised for a number of models. We also used cross-correlation to show that E-BOT neurons inhibited each other (Dufn and van Alphen, 1995b) conrming earlier work (Lindsey et al., 1989a). Much of the basic adult respiratory neurophysiology incorporated in the models was discovered from experiments in adult cats. Now the adult decerebrate rat has become a common preparation (Bianchi et al., 1995). Rats offer an advantage in that they are well studied and complementary knowledge from other elds is available. In addition, rats can also be studied using in-vitro preparations of neonatal rats. Invitro preparations (Ballanyi et al., 1999) not only offer opportunities to study the effects of changes in the cellular environment, but also the ability to utilise dynamic imaging of uorescent dyes in living tissue (Koshiya and Smith, 1999). Thus, electrophysiological and anatomical methods can be used to complement and reinforce each other. In addition, by comparing ndings from neonatal in-vitro preparations with those from adults (taking into account preparation differences) developmental changes can be deduced. In adult rats a network involving inhibitory synapses appears to generate respiratory rhythm, because rhythm stops if inhibitory synapses are blocked (Hayashi and Lipski, 1992; Paton et al., 1994). However, there are differences between species because recent experiments showed that inhibition is not essential for the generation of respiratory rhythm in adult mice (Ramirez et al., 1996). Not only does the species make a difference, but also the age. Blocking inhibitory synapses in neonatal rat in-vitro preparations does not stop rhythm, and rhythm here is hypothesised to originate in the activity of pacemaker cells (Smith et al., 1991; Onimaru et al., 1997;

J. Dufn et al. / Respiration Physiology 122 (2000) 237246

239

Rekling and Feldman, 1998). In adult rats either side of the brainstem can generate respiratory rhythm, so that a medullary midline transection shows independent rhythms (Peever et al., 1998), whereas rhythm stops if this is attempted in the neonatal in-vitro preparation (McLean and Remmers, 1994; Peever et al., unpublished observations). As suggested in an earlier review (Dufn et al., 1995), evidence is accumulating that a synthesis of both pacemaker and network elements may account for respiratory rhythm generation. Each element may dominate under different circumstances such as hypoxic gasping (Ramirez et al., 1998; St. John, 1998), or at different times in development (Paton et al., 1994; Ramirez et al., 1996; Hilaire and Duron, 1999) or in different species (Ramirez et al., 1996). Models of this synthesis have, therefore, been proposed (Smith, 1995; Matsugu et al., 1998) as well as models of rhythm generation in the in-vitro preparations (Butera et al., 1999). It appears then that network interactions play a major role in the generation and shaping of the respiratory drive output. But, since most of the evidence for interconnections that are used in network model simulations has come from experiments in adult cats, we felt that the functional connections among respiratory neurons should be determined in adult rats as well, and set about this task about 5 years ago. Comparison of ndings from rats with those from cats has already found important differences. We describe here the physiological methods used to discover functional connections among respiratory neurons, and some of the information it has produced when used to examine central respiratory control. While the ndings described are conned to rats, hindsight from earlier adult cat experiments was used to predict connections in adult rats.

2. Methods One general network investigation paradigm employed is to rst examine the intracellular recordings of a respiratory neuron and then determine which neurons synapse with it to produce

the observed membrane potential changes. It is assumed that the synapsing neurons discharge patterns will determine the entire membrane potential trajectory throughout the respiratory cycle. This assumption neglects intrinsic neuron properties that can also affect activity, but is a useful starting position. Where do these synaptic inputs originate? To answer this question, it is not sufcient to nd neurons with discharge patterns that are coincident with the synaptic input. Nor is it sufcient to show the anatomical projections of these neurons are appropriate. The proposed connection must be shown to be functional. Such functionality can be established using either cross-correlation or spike-triggered averaging. Both are electrophysiological in nature and require microelectrode recording and sophisticated analysis to detect synchronisation of activity that is interpreted as evidence for synaptic connections. The rst is cross-correlation of extracellularly recorded trains of action potentials (Kirkwood, 1979). These are converted to trains of uniform pulses using time- and amplitude-window gating, and event cross-correlation is used to detect synchronisation of events on a short time scale. When combined with information like the expected transmission latency, cross-correlation can provide convincing evidence for synaptic connections. The second method is known as spike-triggered averaging (Kirkwood and Sears, 1973). The extracellularly recorded action potentials (spikes) of a putative driver neuron are used to trigger an average of the intracellularly recorded membrane potentials of a putative receiver neuron. Again, short-time-scale events in the average, combined with expected transmission latency provide evidence for synaptic connections. Each method has its own strengths. Cross-correlation is easier technically but requires stable extracellular recordings over considerable time and is conned to neuronal activities that overlap in time. Spike-triggered averaging is technically more difcult because it requires stable intracellular recordings, but is considerably quicker and not conned to neurons whose activities overlap in time. Application of the general network investigation paradigm has been proceeding in our labora-

240

J. Dufn et al. / Respiration Physiology 122 (2000) 237246

tory for rats. Fig. 1 shows some basic starting assumptions about the network interconnections of the neurons that generate respiratory rhythm, shape it and relay it to output motoneurons. The pathways that are known to be functional are shown as solid lines, and those yet-to-be-demonstrated are shown dotted. Our investigations began with the phrenic motoneurons and proceeded centrally. Working backwards up the chain of interconnected neurons, should yield an idea of how the respiratory pattern results for each neuron in the chain and eventually lead to insights about respiratory rhythm generation.

3. Findings

3.1. Synaptic connections to phrenic motoneurons

Fig. 2 shows the membrane potential trajectory for a phrenic motoneuron, both before and after chloride ion iontophoresis to reverse inhibitory post-synaptic potentials, as well as the activity of neurons thought to provide synaptic input. The intracellular recordings show that the membrane potential trajectories display three stages in the respiratory cycle; increasing depolarisation during

Fig. 2. The membrane potential trajectory for a phrenic motoneuron, both before and after chloride ion iontophoresis to reverse inhibitory post-synaptic potentials, as well as the activity of the neurons thought to provide synaptic input. The latter recordings were obtained from different rats and scaled to t the respiratory cycle of the phrenic nerve.

Fig. 1. Functional connections among respiratory neurons in the rat. Known are solid lines and unknown are dotted lines. Arrows represent excitatory synaptic connections and lled circles represent inhibitory synaptic connections.

inspiration, declining depolarisation during the rst stage of expiration and nally a hyperpolarization during the second stage of expiration. The reversed inhibitory post-synaptic potentials show that the hyperpolarisation is due to inhibitory synaptic input. In addition to detecting connections from one neuron to another, cross-correlation can also be used to detect the presence of synaptic inputs to two different neurons or nerves that originate from a common source. For example, if the discharge of the left and right phrenic nerves is cross-correlated, the resulting histograms display a central peak (Fig. 3A). This peak is interpreted as evidence that the source of the synaptic excitation is common to both left and right phrenic motoneuron pools. The conclusion is that both left and right phrenic nerves are driven by the

J. Dufn et al. / Respiration Physiology 122 (2000) 237246

241

same VRG inspiratory premotor neurons (Dufn and van Alphen, 1995a). Inspiratory neurons (Fig. 3) with the appropriate augmenting discharge pattern (I-AUG) for the observed phrenic excitation are found in the medulla (VRG). Their extracellularly recorded action potentials were cross-correlated with both left and right phrenic nerve discharge. Both the resulting cross-correlation histograms (Fig. 3C and D) display narrow peaks at a short latency appropriate for the transmission time between medulla and spinal cord. The peaks were interpreted as evidence for the excitation of left and right phrenic motoneurons by the same VRG inspiratory premotor neuron (Dufn and van Alphen, 1995a; Tian and Dufn, 1996). These ndings indicate that in rats most VRG inspiratory premotor neurons project bilaterally down the spinal cord, and monosynaptically excite phrenic motoneurons. The connections are in contrast to those in cats, where such projections are mostly unilaterally and

crossed, and few VRG inspiratory neurons monosynaptically excite phrenic motoneurons (Bianchi et al., 1995). Expiratory neurons (Fig. 3) with the appropriate augmenting discharge pattern (E-AUG) for the phrenic inhibition are found in the Botzinger complex of the medulla. They were used to trigger averages of the membrane potentials recorded intracellularly from phrenic motoneurons. The resulting averages showed inhibitory post-synaptic potentials at latencies appropriate for transmission time (Fig. 3B). Such ndings are interpreted as evidence for the inhibition of phrenic motoneurons by Botzinger complex expiratory neurons (Tian et al., 1998). Thus, the paradigm has been used to show that the phrenic motoneurons membrane potential trajectories result from synaptic excitation provided by VRG inspiratory premotor neurons and synaptic inhibition from Botzinger complex ex piratory neurons. Further explanations in terms

Fig. 3. Demonstration of functional connections to phrenic motoneurons. (A) Cross-correlation of the left phrenic nerve discharge with that of the right. The histogram displays a central peak interpreted as evidence for a common source of excitation. (B) Averaging of a phrenic motoneuron membrane potential triggered by the action potentials of a Botzinger complex expiratory neuron. The average shows an inhibitory post-synaptic potential at a short latency interpreted as evidence for a monosynaptic inhibition of the phrenic motoneuron by the Botzinger complex expiratory neuron. (C, D) Cross-correlation of a single VRG inspiratory neuron activity with both left and right phrenic nerve discharges. The histograms display narrow peaks at short latencies interpreted as evidence for a monosynaptic excitation of both left and right phrenic motoneurons by the VRG inspiratory neuron.

242

J. Dufn et al. / Respiration Physiology 122 (2000) 237246

3.2. Synaptic connections to VRG inspiratory premotor neurons

Starting at the motoneuron output, then working centrally up the chain of interconnected neurons, the next neurons to be subject to the investigation paradigm are the VRG inspiratory premotor neurons. Fig. 4 shows the membrane potential trajectory for a bulbospinal VRG inspiratory neuron, both before and after chloride ion iontophoresis to reverse inhibitory postsynaptic potentials, as well as the activity of neurons thought to provide synaptic input. The membrane potential trajectory of the VRG inspiratory neuron is composed of a depolarisation during inspiration, a declining depolarisation during the rst stage of expiration and a hyperpolarization during the second stage of expiration, similar to the phrenic motoneurons. However, intracellular iontophoresis of chloride ions to reverse inhibitory synaptic inputs reveals a more complex situation. The inspiratory depolarisation results from a combination of an excitation and a declining inhibition. The rst stage of expiration is also a combination of excitation and a declining inhibition, and the second stage of expiration solely the result of inhibitory input. Neurons with the appropriate discharge patterns for the inspiratory excitation have yet to be identied. However, their distribution to the population of VRG inspiratory neurons has been shown to be widespread using cross-correlation (Tian and Dufn, 1997). When the action potential discharges from pairs of VRG inspiratory premotor neurons are cross-correlated, the resulting histograms display central peaks interpreted as evidence for common excitation (Fig. 5A), and spike-triggered averaging conrms this conclusion (Fig. 5B). One source of this excitation may be other VRG inspiratory premotor neurons. Crosscorrelation studies of VRG inspiratory neurons in brainstem spinal cord preparations show that these neurons may excite each other (Kashiwagi et al., 1993; Onimaru et al., 1993), however, such connections have not been detected in adult rats (Tian and Dufn, 1997). Neurons with the appropriate discharge patterns for the inhibitory inputs are found in the

Fig. 4. The membrane potential trajectory for a bulbospinal VRG inspiratory neuron, both before and after chloride ion iontophoresis to reverse inhibitory post-synaptic potentials, as well as the activity of the neurons thought to provide synaptic input. The latter recordings were obtained from different rats and scaled to t the respiratory cycle of the phrenic nerve.

of intrinsic phrenic motoneuron properties can be added as necessary to provide a more detailed description, but the demonstrated functional synaptic connections are sufcient to explain the major aspects of their behaviour. The above ndings were made using adult rat preparations. However, if neonatal rat in-vitro brainstem spinal cord preparations are used some of the results are different. Cross-correlation histograms between left and right phrenic nerves in these preparations show no evidence for common excitation (personal observation). This nding suggests that the connection from the VRG inspiratory premotor neurons to the phrenic motoneurons has not fully developed in the neonate.

J. Dufn et al. / Respiration Physiology 122 (2000) 237246

243

RVLM. Early-burst inspiratory neurons, I-DEC, are active during inspiration and have the appropriate decrementing discharge pattern. Similarly, E-DEC neurons are active during the rst stage of expiration with the appropriate decrementing discharge pattern, and Botzinger complex expiratory neurons, E-BOT, are active during the second stage of expiration, also with the E-AUG. While the I-DEC and E-DEC neurons have

Fig. 6. The membrane potential trajectory for a Botzinger complex expiratory neuron, both before and after chloride ion iontophoresis to reverse inhibitory post-synaptic potentials, as well as the activity of the neurons thought to provide synaptic input. The latter recordings were obtained from different rats and scaled to t the respiratory cycle of the phrenic nerve.

Fig. 5. Demonstration of functional connections to VRG inspiratory neurons. (A) Cross-correlation of two VRG inspiratory bulbospinal neurons. The histogram displays a central peak interpreted as evidence for a common source of excitation. (B) Averaging of a VRG inspiratory bulbospinal neuron membrane potential triggered by the action potentials of a nearby VRG inspiratory bulbospinal neuron. The VRG displays a central peak interpreted as evidence for a common source of excitation. (C) Averaging of a VRG inspiratory bulbospinal neuron membrane potential triggered by the action potentials of a Botzinger complex expiratory neuron. The average shows an inhibitory post-synaptic potential at a short latency interpreted as evidence for a monosynaptic inhibition of the VRG inspiratory neuron by the Botzinger complex expiratory neuron.

been shown to inhibit VRG inspiratory neurons in cats (Ezure, 1990), these connections have yet to be demonstrated in rats. However, the connection from Botzinger complex expiratory neurons has been shown to exist in rats using spike-triggered averaging (Fig. 5C), so one of the hypothesised inhibitory input connections has been veried for the VRG inspiratory neurons (Tian et al., 1999a).

3.3. Synaptic connections to Botzinger complex expiratory neurons

Again, working centrally up the chain of interconnected neurons, the next neurons in the chain to be investigated are the Botzinger complex ex piratory neurons. Fig. 6 shows the membrane potential trajectory for an E-BOT expiratory neu-

244

J. Dufn et al. / Respiration Physiology 122 (2000) 237246

ron, both before and after chloride ion iontophoresis to reverse inhibitory post-synaptic potentials, as well as the activity of neurons thought to provide synaptic input. The membrane potential trajectory of the Botzinger complex expiratory neuron is composed of depolarisations during inspiration and expiration. However, reversal of inhibitory synaptic inputs reveals a more complex situation. A declining inhibition is present during inspiration and the rst stage of expiration. Whether the second stage of expiration is solely a depolarisation or a combined depolarisation and hyperpolarisation cannot be determined. The latter

possibility is likely because, somewhat unexpectedly, Botzinger complex neurons have been shown to inhibit each other using cross-correlation in both rats (Tian et al., 1999b) and cats (Dufn and van Alphen, 1995b). The cross-correlation histograms implicating this conclusion are shown in Fig. 7. Other connections have not been investigated.

4. Conclusion This is the extent to which the exploration of the functional synaptic connections has advanced in rats. It has almost progressed as far as that of cats, before rat and neonatal in-vitro preparations began to predominate in respiratory neurophysiological research. The next steps in nding the connections from the neurons responsible for the VRG inspiratory neurons membrane potential trajectories and those of the Botzinger complex expiratory neurons will expand the search further to the I-DEC and E-DEC neurons. Then they too must be examined in the same way. In the meantime, a number of questions remain apart from those connections hypothesised in this network paradigm approach. Suitable candidates for the excitation of VRG inspiratory neurons and Botzinger complex expiratory neurons are lacking. While they could be central chemoreceptors responding to [H+ ] or reticular formation neurons responding to the state of alertness, they have yet to be identied with any certainty. The neurons characterised as pacemaker cells in the neonate have yet to be discovered in the adult. They may be too difcult to detect and identify in intact adult preparations, or have been transformed into neurons with different characteristics, perhaps into I-DEC neurons. These specic questions, and indeed the overall one of how respiratory rhythm is generated, are likely to be answered only by combining the ndings from experiments like these, seeking functional synaptic connections between neurons, with those from experiments seeking the intrinsic properties and transmitter pharmacology of respiratory neurons. Such ndings can only be obtained by using a combination of investigational tech-

Fig. 7. Demonstration of functional connections to Botzinger complex expiratory neurons. (A) Post-stimulus histogram of the activity of a Botzinger complex expiratory neuron follow ing application of stimuli ( B50 mA, 0.2 ms pulses) to Botzinger complex expiratory neuron axons at the C1C2 border. The histogram displays a trough at short latency interpreted as evidence for inhibition of the Botzinger complex expiratory neuron by another. (B) Cross-correlation of two Botzinger complex expiratory neurons. The histogram displays narrow troughs at short latencies either side of zero time interpreted as evidence for mutual inhibition.

J. Dufn et al. / Respiration Physiology 122 (2000) 237246

245

niques utilising preparations. Acknowledgements

both

intact

and

reduced

This research was supported by a grant from the Canadian Medical Research Council. We thank the members of the Respiratory Research group in the Department of Physiology at the University of Toronto for helpful discussion. References
Balis, U.J., Morris, K.F., Koleski, J., Lindsey, B.G., 1994. Simulations of a ventrolateral medullary neural network for respiratory rhythmogenesis inferred from spike train cross- correlation. Biol. Cyber. 70, 311327. Ballanyi, K., Onimaru, H., Homma, I., 1999. Respiratory network function in the isolated brainstem-spinal cord of newborn rats. Prog. Neurobiol. 59, 583634. Bianchi, A.L., Denavit-Saubie, M., Champagnat, J., 1995. Central control of breathing in mammals: neuronal circuitry, membrane properties, and neurotransmitters. Physiol. Rev. 75, 1 45. Butera, R.J., Rinzel, J., Smith, J.C., 1999. Models of respiratory rhythm generation in the pre-Botzinger complex. II. Populations of coupled pacemaker neurons. J. Neurophysiol. 82, 398 415. Dufn, J., 1991. A model of respiratory rhythm generation. NeuroReport 2, 623 626. Dufn, J., Douse, M.A., 1993. Botzinger expiratory neurones inhibit propriobulbar decrementing inspiratory neurones. NeuroReport 4, 12151218. Dufn, J., van Alphen, J., 1995a. Bilateral connections from ventral group inspiratory neurons to phrenic motoneurons in the rat determined by cross-correlation. Brain Res. 694, 55 60. Dufn, J., van Alphen, J., 1995b. Cross-correlation of augmenting expiratory neurons of the Botzinger complex in the cat. Exp. Brain Res. 103, 251255. Dufn, J., Ezure, K., Lipski, J., 1995. Breathing rhythm generation: focus on the rostral ventrolateral medulla. News Physiol. Sci. 10, 133140. Ezure, K., 1990. Synaptic connections between medullary respiratory neurons and considerations on the genesis of respiratory rhythm. Prog. Neurobiol. 35, 429450. Feldman, J.L., McCrimmon, D.R., 1999. Neural control of breathing. In: Zigmond, M.J., Bloom, F.E., Landis, S.C., Roberts, J.L., Squire, L.R. (Eds.), Fundamental Neuroscience. Academic Press, New York, NY, pp. 10631090. Gottschalk, A., Oglvie, M.D., Richter, D.W., Pack, A.I., 1994. Computational aspects of the respiratory pattern generator. Neural Comput. 6, 5668.

Hayashi, F., Lipski, J., 1992. The role of inhibitory amino acids in control of respiratory motor output in an arterially perfused rat. Respir. Physiol. 89, 47 63. Hilaire, G., Duron, B., 1999. Maturation of the mammalian respiratory system. Physiol. Rev. 79, 325 360. Hsieh, J.H., Chang, Y.C., Su, C.K., Hwang, J.C., Yen, C.T., Chai, C.Y., 1998. A single minute lesion around the ventral respiratory group in medulla produces fatal apnea in cats. J. Auton. Nerv. Syst. 73, 7 18. Kashiwagi, M., Onimaru, H., Homma, I., 1993. Correlation analysis of respiratory neuron activity in ventrolateral medulla of brainstem-spinal cord preparation isolated from newborn rat. Exp. Brain Res. 95, 277 290. Kirkwood, P.A., 1979. On the use and interpretation of crosscorrelation measurements in the mammalian central nervous system. J. Neurosci. Meth. 1, 107 132. Kirkwood, P.A., Sears, T.A., 1973. Interaction between the monosynaptic EPSP and the central respiratory drive potential of expiratory motoneurones in the cat. J. Physiol. Lond. 232, 38 40. Koshiya, N., Smith, J.C., 1999. Neuronal pacemaker for breathing visualized in vitro. Nature 400, 360 363. Lindsey, B.G., Segers, L.S., Shannon, R., 1989a. Discharge patterns of rostrolateral medullary expiratory neurons in the cat: regulation by concurrent network processes. J. Neurophysiol. 61, 1185 1196. Lindsey, B.G., Shannon, R., Gerstein, G.L., 1989b. Gravitational representation of simultaneously recorded brainstem respiratory neuron spike trains. Brain Res. 483, 373 378. Matsugu, M., Dufn, J., Poon, C.S., 1998. Entrainment, instability, quasi-periodicity, and chaos in a compound neural oscillator. J. Comput. Neurosci. 5, 35 51. McLean, H.A., Remmers, J.E., 1994. Respiratory motor output of the sectioned medulla of the neonatal rat. Respir. Physiol. 96, 49 60. Ogilvie, M.D., Gottschalk, A., Anders, K., Richter, D.W., Pack, A.I., 1992. A network model of respiratory rhythmogenesis. Am. J. Physiol. 263, R962 R975. Onimaru, H., Kashiwagi, M., Arata, A., Homma, I., 1993. Possible mutual excitatory couplings between inspiratory neurons in caudal ventrolateral medulla of brainstemspinal cord preparation isolated from newborn rat. Neurosci. Lett. 150, 203 206. Onimaru, H., Arata, A., Homma, I., 1997. Neuronal mechanisms of respiratory rhythm generation: an approach using in vitro preparation. Jpn. J. Physiol. 47, 385 403. Paton, J.F.R., Ramirez, J.M., Richter, D.W., 1994. Mechanisms of respiratory rhythm generation change profoundly during early life in mice and rats. Neurosci. Lett. 170, 167 170. Peever, J.H., Tian, G.F., Dufn, J., 1998. Bilaterally independent respiratory rhythms in the decerebrate rat. Neurosci. Lett. 247, 41 44. Ramirez, J.M., Quellmalz, U.J., Richter, D.W., 1996. Postnatal changes in the mammalian respiratory network as revealed by the transverse brainstem slice of mice. J. Physiol. Lond. 491, 799 812.

246

J. Dufn et al. / Respiration Physiology 122 (2000) 237246 region that may generate respiratory rhythm in mammals. Science 254, 726 729. St. John, W.M., 1998. Neurogenesis of patterns of automatic ventilatory activity. Prog. Neurobiol. 56, 97 117. Tian, G.F., Dufn, J., 1996. Spinal connections of ventralgroup bulbospinal inspiratory neurons studied with crosscorrelation in the decerebrate rat. Exp. Brain Res. 111, 178 186. Tian, G.F., Dufn, J., 1997. Synchronization of ventral-group, bulbospinal inspiratory neurons in the decerebrate rat. Exp. Brain Res. 117, 479 487. Tian, G.F., Peever, J.H., Dufn, J., 1998. Botzinger-complex expiratory neurons monosynaptically inhibit phrenic motoneurons in the decerebrate rat. Exp. Brain Res. 122, 149 156. Tian, G.-F., Peever, J.H., Dufn, J., 1999a. Botzinger-com plex, bulbospinal expiratory neurons monosynaptically inhibit ventral-group respiratory neurons in the decerebrate rat. Exp. Brain Res. 124, 173 180. Tian, G.-F., Peever, J.H., Dufn, J., 1999b. Mutual inhibition between Botzinger-complex bulbospinal expiratory neurons detected with cross-correlation in the decerebrate rat. Exp. Brain Res. 125, 440 446.

Ramirez, J.M., Telgkamp, P., Elsen, F.P., Quellmalz, U.J., Richter, D.W., 1997. Respiratory rhythm generation in mammals: synaptic and membrane properties. Respir. Physiol. 110, 71 85. Ramirez, J.M., Schwarzacher, S.W., Pierreche, O., Olivera, B.M., Richter, D.W., 1998. Selective lesioning of the cat pre-Botzinger complex in vivo eliminates breathing but not gasping. J. Physiol. Lond. 507, 895907. Rekling, J.C., Feldman, J.L., 1998. PreBotzinger complex and pacemaker neurons: hypothesized site and kernel for respiratory rhythm generation. Ann. Rev. Physiol. 60, 385 405. Richter, D.W., Ballanyi, K., Schwarzacher, S., Schwarzacher, S.W., 1992. Mechanisms of respiratory rhythm generation. Curr. Opin. Neurobiol. 2, 788793. Ryback, I.A., Paton, J.F.R., Schwaber, J.S., 1997. Modeling neural mechanisms for genesis of respiratory rhythm and pattern. II. Network models of the central respiratory pattern generator. J. Neurophysiol. 77, 20072026. Smith, J.C., 1995. New computational models of the respiratory oscillator in mammals. Adv. Exp. Med. Biol. 393, 713. Smith, J.C., Ellenberger, H.H., Ballanyi, K., Richter, D.W., Feldman, J.L., 1991. Pre-Botzinger complex: a brainstem