Respiratory pattern in the intact pontine-medullary respiratory network in situ
The stereotypical patterns of PN, HN, and cVN activities generated by the intact preparation are shown in . PN burst frequency in these preparations was in the range 0.24–0.47 bursts/s with a relatively constant PN burst duration (1.00 ± 0.14 s, n
= 20 preparations). The activity profiles of different VRC neuron populations and integrated nerve activities are shown in . These patterns resembled those recorded in vivo during generation of a normal three-phase respiratory rhythm (St.-John and Paton 2003
) and exhibited the following characteristics (): 1
) an augmenting shape of PN bursts; 2
) preinspiratory onset of HN bursts (50–100 ms before the onset of PN bursts); and 3
) a prominent epoch of decrementing postinspiratory (post-I) discharge in cVN.
FIG. 3 Representative activity patterns of phrenic (PN), hypoglossal (HN), and central vagus (cVN) nerves recorded from intact (A), medullary (B), and pre-BötC (C) preparations. Each panel shows raw (bottom traces) and integrated (top traces) motor nerve (more ...)
FIG. 4 Neuronal activity patterns within several respiratory compartments recorded in both intact and reduced in situ preparations. Activity patterns shown (top set of traces) within BötC, pre-BötC, and rVRG (indicated at left) are composite (more ...)
Neural activities recorded within the VRC could be clearly subdivided into three phases (): inspiratory (I) that corresponded to PN activity, postinspiratory (pI), and active expiratory (E2), whose timing corresponded to the onset of augmenting expiratory (aug-E) activity recorded in the BötC (below). Extracellular recordings from intact preparations () revealed typical neuronal types with the following characteristic activity patterns: 1) neurons with decrementing post-I, and with augmenting expiratory (aug-E) discharges within BötC; 2) neurons with a preinspiratory onset of activity (pre-I), and with an early-inspiratory (early-I) decrementing discharge within pre-BötC; and 3) neurons with augmenting activity patterns (ramp-I) within rVRG.
Transformations of respiratory pattern with sequential brain stem transections in situ
The spatial organization of pontine-medullary respiratory networks was studied by sequentially reducing the network with a series of rostral to caudal brain stem microtransections starting at the level of the pons or near the pontine-medullary junction. shows an example of the histological appearance of the postfixed brain stem in sagittal view after a series of such transections. These transections allowed us to remove specific circuit components along the brain stem “respiratory column” bilaterally, which included severance of connections across the midline, and to analyze corresponding transformations of neuronal activity and motor output patterns. Vertical dashed lines in indicate several experimental transections and levels, which delineate the rostral extent of the reduced preparations used in this study. The medullary preparations were obtained after transections through the facial nucleus from the pontine-medullary junction at the rostral end to the rostral boundary of BötC at the caudal end. The pre-BötC preparation was obtained after transection at the rostral boundary of pre-BötC, whereas the rVRG preparation was made after transection at the rostral boundary of rVRG ().
On production of a medullary preparation, by transecting at the rostral end of facial nucleus (i.e., pontine-medullary junction; ), the three-phase rhythm was converted into a two-phase inspiratory-expiratory pattern (Figs. and ), which was characterized by a nonramping “square-wave” inspiratory motor profile with the onset of activity synchronized in all nerves and by a lack of post-I discharge in cVN. Burst frequency in medullary preparations was in the range 0.14–0.32 burst/s (n = 20); the inspiratory phase duration was 1.73 ± 0.45 s. The amplitude of inspiratory bursts was reduced by >50% relative to bursts generated by the intact brain stem. Neuronal activities within the BötC, pre-BötC, and rVRG during the two-phase rhythm included decrementing expiratory (in BötC) and inspiratory (in pre-BötC and rVRG) discharges ().
The cycle-to-cycle variability of inspiratory burst frequency in the two-phase rhythm generated by medullary preparations was greater compared with the three-phase rhythm. This variability increased with more caudal transections (reducing the remaining part of FN/RTN; ) because of the emergence of shorter duration “ectopic” bursts interposed between longer duration square-wave bursts (). Interestingly, administration of riluzole (≤10 μM) eliminated the ectopic bursting and stabilized the two-phase rhythm ().
FIG. 5 The 2-phase rhythmic motor pattern generated by medullary preparations. Transection of the brain stem through the rostral facial nucleus (FN) transformed the normal 3-phase pattern with incrementing PN discharges (top trace) into a 2-phase pattern with (more ...)
Transection at the rostral border of pre-BötC (pre-BötC preparation indicated in ; n
= 8) to remove the BötC compartment disrupted rhythmic activity, or resulted in low frequency spontaneous decrementing PN discharges (0.5–1.5 s in duration, 0.04–0.07 bursts/s frequency range). This spontaneous activity could be stabilized at a higher burst frequency (0.13– 0.37 bursts/s range) by elevating the perfusate CO2
concentration (≤10%) and/or extracellular K+
(≤9 mM). In preparations where activity was eliminated by transection, the rhythm could be reactivated by either peripheral chemoreceptor stimulation or stopping the perfusion for ≤1 min. The rhythmic motor pattern was stabilized and maintained by a combination of elevating CO2
and extracellular K+
. The resultant rhythmic activity pattern consisted of decrementing inspiratory bursts synchronized in all motor outputs as shown in . Integrated population activity of pre-BötC and rVRG exhibited similar decrementing inspiratory discharge patterns (). This inspiratory rhythm and its activation/reactivation properties (e.g., with elevation of extracellular K+
) were analogous to those described previously for in vitro slices from the neonatal rat medulla containing the pre-BötC (Del Negro et al. 2001
; Koshiya and Smith 1999
). We called this activity a one-phase inspiratory rhythm because it occurred in the absence of expiratory activity and involved an endogenous bursting mechanism presumably operating within the pre-BötC.
Transection at the rostral boundary of rVRG, which removed the pre-BötC (rVRG preparation in ), eliminated all rhythmic motor activity from the PN, HN, and cVN. After this transection, we failed to find rhythmically active neurons in the rVRG (n
= 8 preparations). Activity could not be restored by chemosensory stimulation with NaCN injections and/or elevated CO2
, and/or elevated extracellular K+
concentrations. This confirmed that pre-BötC circuits are required for the one-phase inspiratory rhythm, as originally shown for the neonatal systems under in vitro conditions (Smith et al. 1991
Probing for persistent sodium (INaP)-dependent rhythmogenic mechanisms
To study a possible contribution of INaP
-dependent mechanisms to the generation of the three-, two-, and one-phase rhythms, we used riluzole (1–20 μ
M), a pharmacological blocker of INaP
(Urbani and Belluzzi 2000
). Riluzole was added to the perfusate in the intact and reduced preparations at concentrations (1–20 μ
M) that were previously shown to attenuate and finally block INaP
at the cellular level and abolish intrinsic bursting activity of the pre-BötC in vitro and in situ (Koizumi and Smith 2002
; Paton et al. 2006
; Rybak et al. 2003b
). , shows effects of riluzole on the frequency and amplitude of PN bursts in the intact pontinemedullary network (n
= 8). The INaP
blocker reduced the PN burst amplitude but did not significantly affect burst frequency ().
FIG. 6 A–C: steady-state dose-dependent effects of riluzole on burst frequency (solid lines) and amplitude (dashed lines) of integrated phrenic (PN) activity recorded in the intact (A), medullary (B), and pre-BötC (C) preparations. PN amplitudes (more ...)
In all medullary preparations generating a two-phase rhythm, low concentrations of riluzole (≤7.5 μM) eliminated ectopic bursting () and stabilized the rhythm at a reduced (~60% of control) burst frequency (n = 7; ); discharge amplitude and frequency were reduced further to ~50% of control with progressive elevation of riluzole concentrations to the maximum tested (20 μM; ).
In contrast, in pre-BötC preparations (n = 7) generating a one-phase rhythm, there was a dose-dependent reduction in discharge frequency, and finally rhythmic activity was terminated at relatively low riluzole concentrations (≤10 μM; ). The discharge amplitude was less sensitive to riluzole but also was attenuated (~50%) before loss of the rhythm (). Recordings of pre-BötC population activity mirrored alterations of motor rhythm and amplitude with INaP blockade and verified complete loss of rhythmic activity in the pre-BötC coincident with the loss of motor output (). Without exception, rhythmic activity in the PN or pre-BötC could not be restored with hypoxic stimulation, elevations of CO2 and/or extracellular K+, or any combination of these stimuli.
Computational modeling of the brain stem respiratory network: model description
A computational model of the spatially distributed brain stem respiratory network was developed to reproduce the above experimental findings and suggest explanations for transformations of the rhythm-generating mechanism with sequential reduction of the network. The schematic of the model is shown in . The model includes the pons and three major medullary compartments: BötC, pre-BötC, and rVRG. Although some respiratory neuron types (e.g., post-I, aug-E) are not localized in particular compartments but rather distributed throughout the VRC, in our model, we assumed for simplicity that each medullary compartment contains only populations of respiratory neuron types that are known to be dominantly present in this compartment. The BötC compartment contains two populations of inhibitory expiratory neurons, the augmenting expiratory (aug-E) and the postinspiratory (post-I), which are both known to provide widely distributed inhibition within the medullary respiratory network during expiration (Ezure 1990
; Ezure and Manabe 1988
; Ezure et al. 2003
; Fedorko and Merrill 1984
; Jiang and Lipski 1990
; Shen et al. 2003
; Tian et al. 1999
). In the model, these populations inhibit neural populations within the pre-BötC and rVRG and each other (). In addition, the BötC compartment contains an excitatory population [conditionally called post-I(e)] that contributes to the post-I component of cVN motor output. We assumed that all BötC neurons [comprising the post-I, post-I(e), and aug-E populations] have intrinsic adapting properties defined by the high-voltage activated calcium (ICaL
) and calcium-dependent potassium (IK,Ca
) currents in these neurons (see APPENDIX
FIG. 7 Schematic of the computational model of the brain stem respiratory network. Model includes interacting neuronal populations within major brain stem respiratory compartments (Pons, BötC, pre-BötC, and rVRG). Spheres represent neuronal populations (more ...)
The pre-BötC compartment includes two neural populations: pre-I and early-I(1) (). The pre-I population is the key excitatory population of pre-BötC that serves as a major source of inspiratory activity in the network. This population projects to the ramp-I population of premotor inspiratory neurons of rVRG and (through a hypoglossal premotor neural population not present in the model) to the hypoglossal motor output (HN). The pre-I population comprises excitatory neurons with INaP
-dependent endogenous bursting properties (see APPENDIX
) and mutual excitatory synaptic connections within the population. At a relatively low level of neuronal excitability or tonic excitatory drive, this population can operate in a bursting mode and intrinsically generate rhythmic bursting activity (Butera et al. 1999a
; Rybak et al. 2003b
; Smith et al. 2000
) that closely reproduces the population rhythmic bursting activity recorded from the pre-BötC in vitro (Johnson et al. 2001
; Koshiya and Smith 1999
). Previous modeling studies have shown that an increase in the average neuronal excitability or in external excitatory drive produces an increase in the burst frequency and, finally, switches population activity to the mode of tonic (asynchronous) spiking in the population. This was confirmed experimentally by an increase of extracellular potassium concentration in medullary slices containing the pre-BötC (Del Negro et al. 2001
; Koshiya and Smith 1999
; Rybak et al. 2003b
). In this model under normal conditions, most neurons of this population operate in a tonic-spiking mode because of high tonic excitatory input and are inhibited by expiratory neurons (post-I, aug-E) during expiration.
The early-I(1) population of pre-BötC is a population of inhibitory interneurons with adapting properties (defined by ICaL
, see APPENDIX
). This population receives excitation from the pre-I population and serves as a major source of inspiratory inhibition (Bianchi et al. 1995
; Ezure 1990
; Segers et al. 1987
). In this model, this population inhibits all expiratory neurons during inspiration ().
The rVRG compartment contains the ramp-I, and early-I(2) populations (). Ramp-I is a population of excitatory premotor inspiratory neurons that project to phrenic motoneurons. Activity of this population defines phrenic motor output (PN) and the inspiratory component of cVN discharge. The major role of the inhibitory early-I(2) population (with adapting neurons containing ICaL
, see APPENDIX
) in the model is in shaping the augmenting patterns of ramp-I neurons (Bianchi et al. 1995
; Richter 1996
; Segers et al. 1987
The behavior of the respiratory CPG depends on a variety of afferent inputs to different respiratory neurons that allow breathing to maintain the appropriate homeostatic levels of O2
and adaptively respond to various metabolic demands. These inputs are modeled as “excitatory drives” that carry state-characterizing information provided by multiple sources distributed within the brain stem (pons, RTN, raphé, NTS), including those considered to be major chemoreceptor sites (sensing CO2
/pH), and/or receiving input from peripheral chemoreceptors (sensing CO2
/pH and low O2) (i.e., RTN, raphé, see Guyenet et al. 2005
; Nattie 1999
; Richerson 2004
). Although currently undefined, these drives seem to have a certain spatial organization that maps specifically on the spatial organization of the brain stem respiratory network. These drives are conditionally represented in the model by three separate sources located in pons, RTN/BötC, and pre-BötC compartments ().
Modeling reorganization of rhythm generating mechanisms after brain stem transections
, shows the performance of the intact model. The activity of each population in is represented by an average spike-frequency histogram of population activity. The post-I population of BötC shows decrementing activity during expiration. This population inhibits all other neuron populations in the model [except post-I(e)] during the first half of expiration (postinspiratory phase). With the progressive reduction of post-I inhibition from the adapting post-I neurons, the aug-E population starts firing later in expiration and forms a late expiratory (E2) phase. At the end of expiration, the pre-I population of pre-BötC is released from inhibition and activates the early-I(1) population that in turn inhibits all expiratory populations within the BötC. As a result, the ramp-I [and early-I(2); ] population of rVRG is released from inhibition (with some delay relative to pre-I) and initiates the next inspiratory phase. During the inspiratory phase, the activity of the early-I(1) population of pre-BötC decreases providing a slow disinhibition of the post-I population of BötC. Once the post-I population starts firing, it inhibits all inspiratory activity completing the inspiratory off-switch. Then the process repeats. In summary, the three-phase respiratory rhythm in the intact model emerges from the mutual inhibitory interactions between early-I(1), post-I, and aug-E populations comprising a three-population ring structure (marked by gray shading in ), with the pre-I excitatory population participating in the onset of inspiration ().
FIG. 8 A–C: key elements and circuits within the intact (A), medullary (B), and pre-BötC (C) models involved in rhythmogenesis (excitatory drives are not shown; excitatory populations, red; inhibitory, blue). A1–C1: activity of selected (more ...)
Motor output patterns () and population activities () in the intact model reproduce all major characteristics of the experimentally recorded three-phase respiratory pattern (for comparison, see Figs. and ), including 1) an augmenting profile of ramp-I and PN inspiratory bursts; 2) a preinspiratory onset of bursts in the pre-I population of pre-BötC and HN (relative to PN); 3) a decrementing activity of post-I neurons in BötC and prominent post-I component in cVN bursts; and 4) an augmenting expiratory activity of BötC aug-E neurons.
To model perturbations caused by transections removing the pons, we removed the pontine excitatory drive (), reducing the intact model to a medullary model. The performance of the medullary model is shown in . Based on experimental evidence that stimulation of the dorsolateral pons (PB/KF region) provides strong activation of post-I neurons (Dutschmann and Herbert 2006
; Rybak et al. 2004a
), we suggested that a major portion of excitatory tonic drive to post-I neurons of BötC comes from the pons. In contrast, the aug-E population in the model is less dependent on pontine drive but receives a major excitatory drive from the RTN and other medullary sources. Thus in our model, removal of the pons reduces the excitability of post-I neurons relative to aug-E neurons so that the post-I population becomes fully inhibited by the aug-E population, which now exhibits a decrementing pattern (defined by ICaL
). Therefore the two-phase rhythm generated in the medullary model is based on an inhibitory half-center circuit of reciprocally interacting populations of adapting aug-E and early-I(1) neurons (see gray shading in and neuronal activities in ). In addition, elimination of pontine drive reduces the excitability and firing frequency of the pre-I and ramp-I populations, reducing the amplitude of all motor outputs (). The medullary model reproduces all major characteristics of the respiratory pattern recorded in the corresponding reduced preparations (Figs. and ): 1
) the loss of post-I activity in the network and cVN; 2
) a reduced amplitude, square-wave-like/slightly decrementing profile of all inspiratory populations and motor bursts; and 3
) synchronized onset of bursts in all motor outputs ().
The pre-BötC model () is characterized by a further reduction in tonic excitatory drive to the pre-I population of pre-BötC and loss of expiratory-related phasic inhibition (). These alterations switch the operating state of the pre-I population, which now generates endogenous bursting activity based on the expression of INaP
and mutual excitatory interactions within the population (Butera et al. 1999b
; Rybak et al. 2003b
; Smith et al. 2000
) (). This pre-BötC activity with a decrementing burst shape now drives the activity of the rVRG and all motor outputs exhibit one-phase (inspiratory) oscillations with a decrementing burst shape (), similar to that recorded from the pre-BötC preparation (Figs. and ).
Testing the dependency of model performance on INaP
To study the role of INaP
and compare model behaviors to experimental data obtained with the INaP
blocker riluzole (), the mean maximal conductance of NaP channels (NaP
) was progressively reduced (to zero) in all pre-BötC (pre-I) neurons. As shown in , a progressive reduction of NaP
in the intact network model causes only a small reduction in the amplitude and frequency of PN bursts. In the medullary model generating the two-phase rhythm, the oscillatory frequency and PN amplitude become more sensitive to INaP
block because after removing pontine excitatory drive the mean level of INaP
inactivation is reduced, enabling some participation of the pre-I population in the expiratory-inspiratory cycle dynamics. The two-phase rhythm, however, persists even at NaP
= 0 (). In the pre-BötC model, the one-phase rhythm is generated solely by endogenous INaP
-dependent bursting activity within the pre-I population of the pre-BötC (). Therefore reducing NaP
progressively decreases PN burst frequency and finally abolishes the rhythm when NaP
becomes less than a critical value (). These modeling results are fully consistent with our experimental data ().
FIG. 9 Effects of reducing INaP on frequency and amplitude of motor output (PN) in the intact (A), medullary (B), and pre-BötC (C) models. Attenuation of INaP by riluzole in experiments is modeled by uniformly reducing the maximum conductance for the (more ...)
presents a more detailed explanation for the differences in the dependence of rhythmogenesis on the pre-BötC INaP-dependent intrinsic mechanism in the intact and reduced models. In the intact model (), strong pontine and RTN excitatory drives depolarize all neurons in the pre-I population and almost completely inactivate the voltage-dependent INaP in these neurons (see trace 4). In addition, these neurons receive strong phasic inhibition from BötC post-I (and aug-E) neurons (see bottom trace in ). Therefore the three-phase rhythm is generated primarily by inhibitory network interactions without critical involvement of the intrinsic INaP-based mechanism (). In the medullary model (), the total drive to the pre-I population is reduced (because of the removal of the pontine portion of total drive), and NaP channels become more active (see elevated hNaP values). Simultaneously, the elimination of pontine drive to post-I neurons reduces phasic inhibition to pre-I neurons (bottom trace). Therefore the INaP-dependent intrinsic mechanism starts contributing to excitability of the pre-I population and to the control of frequency and amplitude of motor outputs. However, network mechanisms (phasic inhibition from aug-E neurons) are still strong enough to maintain rhythm generation even when INaP is completely blocked (). In the pre-BötC model (), excitatory drive is further reduced, which allows full activation of INaP-dependent bursting properties (4th trace) and eliminates phasic inhibition (bottom trace). The intrinsic INaP-dependent bursting properties and mutual excitatory interactions within the pre-I population now completely define rhythmic activity (1-phase rhythm). This rhythm can be abolished by suppression of NaP currents ().
FIG. 10 Dynamics of pre-I population in the intact (A), medullary (B), and pre-BötC (C) models. Top traces are activity raster plots for all 50 neurons in the modeled pre-I population (each lines represents single neuron spiking, dots indicate spikes). (more ...)
represents our simulation of the changes of rhythm and pattern generated in the medullary preparation after transections through the caudal half of FN/RTN (). The medullary model was used, but the RTN drive in the medullary model was reduced by 50%. The reduction of this drive caused ectopic bursts similar to those recorded experimentally (). The raster plot of the pre-BötC pre-I population (top trace) shows that ectopic bursts seen in motor outputs originate from a subpopulation of neurons within the heterogeneous pre-I population. shows that attenuation of INaP
, by reducing the average maximal conductance of NaP channels (NaP
) in the pre-I population, eliminates ectopic bursts and stabilizes the two-phase rhythm. These simulation results are fully consistent with our experimental findings ().
FIG. 11 To simulate changes of rhythm and pattern generated in the medullary preparation after transections through caudal FN/RTN, the RTN drive in the medullary model (see ) was reduced by 50%. A: reduction of RTN drive caused ectopic bursts (indicated (more ...)
Pattern transformations with progressive reduction of Cl−-mediated synaptic inhibition
Our model predicted that the generation of the three- and two-phase rhythmic patterns is based on inhibitory synaptic interactions. Both glycine and GABA, known to be the major inhibitory neurotransmitters in the brain stem respiratory network (Büsselberg et al. 2001
; Ezure et al. 2003
; Haji et al. 2000
; Paton and Richter 1995
; Schreihofer et al. 1999
), involve Cl−
-mediated inhibition. Therefore to test the role of Cl−
-mediated inhibition, we switched the normal perfusate in intact preparations to a solution containing reduced Cl−
concentration (60, 40, or 20% of control concentrations in separate experiments: n
= 6, n
= 7, and n
= 10, respectively; see methods
). In all cases, after switching to the reduced Cl−
perfusate, the three-phase motor output pattern transformed to a two-phase pattern. This transformation developed progressively with time reflecting a slow process of equilibration of the brain extracellular fluid to the reduced Cl−
conditions. This transformation was accompanied by a loss of the post-I component in the cVN activity, a reduction of amplitudes of all motor outputs, and an alteration of the shapes of motor bursts similar to those occurring after brain stem transections (synchronous, square-wave-like inspiratory bursts of PN and cVN; Figs. , and ). Further in time, rhythmic motor output terminated under the low Cl−
conditions (Figs. , and ).
FIG. 12 Transition from 3- to 2-phase rhythmic pattern and subsequent termination of rhythm generation during progressive reduction of Cl−-mediated inhibition in situ. A: simultaneous recordings of raw and continuous histograms (0.25-s bins) of pre-BötC (more ...)
FIG. 13 Composite cycle-triggered histograms of integrated PN, cVN, pre-BötC (pre-I), and BötC (aug-E) population activities showing temporal relationships and profiles of population activities under control conditions (A) and 8 min after switching (more ...)
Recordings of aug-E activity in the BötC (n = 6) and pre-I activity in the pre-BötC (n = 8) during perfusion with low Cl− showed a progressive expansion of pre-I activity into the E phase and aug-E activity into the post-I phase. As a result, the post-I discharge was fully abolished indicating the transition to a two-phase rhythm generating state (). In this state, E-discharge occurred throughout the E phase, alternating with square-wave-like inspiratory activity (see Figs. , and for quantitative analysis of population spiking profiles) similar to that observed in a two-phase rhythm obtained by brain stem transections. With time, this E activity progressively encroached on the inspiratory phase, until both BötC E and pre-BötC population activities became tonic, indicating severe reduction of Cl−-mediated inhibition that fully terminated rhythm generation (Figs. , and ).
, shows the results of separate experiments in which rhythm and pattern transformations were obtained with a perfusate solution containing 20% of control Cl− concentration. With this perfusate, the transition to a two-phase rhythm occurred on average within 4 min, and the termination of rhythmic activities was observed within 30–40 min. In experiments with perfusate solutions containing 60 or 40% of control Cl− concentrations, the sequential pattern transformation described above was consistently observed as well, but the times to the three- to two-phase rhythm transition and to the subsequent termination of rhythmic activity were progressively longer (~1.5 and 2 times longer at 60 and 40% Cl− solutions, respectively). In all cases, rhythmic activity was restored when the perfusate was replaced with the control solution, although, after the 20% Cl− perfusate, we could typically achieve only partial recovery of the normal pattern and discharge amplitudes ().
To simulate the effect of reduction of chloride-based inhibition by lowering extracellular Cl− concentration ([Cl−]out), the chloride reversal potential in the model (ECl = ESynI) was changed from −75 to −60 mV (to be equal to the average neuronal resting potential), which, according to the Nernst equation and with the temperature T = 305 K (used in the experimental preparation), approximately corresponds to a 50% decrease of [Cl−]out. The results of simulation are shown in . Similar to our experimental results (Figs. , and ), changing ECl to −60 mV abolished rhythmic activity in PN and cVN motor outputs and produced sustained activity in the aug-E population of BötC and the pre-I population of pre-BötC (). However, the model did not reproduce the transition from the three-phase to a two-phase rhythm after a smaller reduction of inhibition if the latter was applied uniformly to all neuronal populations in the model. The direct simulation of this transition, as observed in our experiments, is difficult because the reduced inhibition during perfusion of low Cl− solutions may provide different effects on different populations of respiratory neurons and/or has different time courses, which are currently unknown.