Learning to breathe: control of the inspiratory–expiratory phase transition shifts from sensory- to central-dominated during postnatal development in rats


  • Mathias Dutschmann,

    1. Institute for Membrane and Systems Biology, University of Leeds, Leeds LS2 9JT, UK
    2. Berstein Center for Computational Neuroscience, Göttingen, Germany
    Search for more papers by this author
  • Michael Mörschel,

    1. Department of Neurophysiology and Sensory Physiology, Faculty of Medicine, Georg August University of Göttingen, Göttingen 37073, Germany
    2. Berstein Center for Computational Neuroscience, Göttingen, Germany
    Search for more papers by this author
  • Ilya A. Rybak,

    1. Department of Neurobiology and Anatomy, Drexel University College of Medicine, Philadelphia, PA 19129, USA
    Search for more papers by this author
  • Thomas E. Dick

    1. Department of Medicine, Case Western Reserve University, Cleveland, OH 44106, USA
    Search for more papers by this author

  • M. Dutschmann and M. Mörschel contributed equally to this work.

Corresponding author M. Dutschmann: Institute for Membrane and Systems Biology, University of Leeds, Leeds LS2 9JT, UK.  Email: m.dutschmann@leeds.ac.uk


The hallmark of the dynamic regulation of the transitions between inspiration and expiration is the timing of the inspiratory off-switch (IOS) mechanisms. IOS is mediated by pulmonary vagal afferent feedback (Breuer–Hering reflex) and by central interactions involving the Kölliker–Fuse nuclei (KFn). We hypothesized that the balance between these two mechanisms controlling IOS may change during postnatal development. We tested this hypothesis by comparing neural responses to repetitive rhythmic vagal stimulation, at a stimulation frequency that paces baseline breathing, using in situ perfused brainstem preparations of rats at different postnatal ages. At ages < P15 (P, postnatal days), phrenic nerve activity (PNA) was immediately paced and entrained to the afferent input and this pattern remained unchanged by repetitive stimulations, indicating that vagal input stereotypically dominated the control of IOS. In contrast, PNA entrainment at > P15 was initially insignificant, but increased after repetitive vagal stimulation or lung inflation. This progressive adaption of PNA to the pattern of the sensory input was accompanied by the emergence of anticipatory centrally mediated IOS preceding the stimulus trains. The anticipatory IOS was blocked by bilateral microinjections of NMDA receptor antagonists into the KFn and PNA was immediately paced and entrained, as it was seen at ages < P15. We conclude that as postnatal maturation advances, synaptic mechanisms involving NMDA receptors in the KFn can override the vagally evoked IOS after ‘training’ using repetitive stimulation trials. The anticipatory IOS may imply a hitherto undescribed form of pattern learning and recall in convergent sensory and central synaptic pathways that mediate IOS.


Breuer–Hering reflex


central pattern generator


inspiratory off-switch


Kölliker–Fuse nuclei


NMDA receptors


nucleus tractus solitarii


phrenic nerve activity


pulmonary stretch receptors


The postnatal developmental process by which animals acquire the fully matured respiratory control is poorly understood. Breathing is produced and controlled by a central pattern generator (CPG). CPGs are neuronal networks that are capable of generating neural oscillations without rhythmic afferent feedback (see Yuste et al. 2005), but are nonetheless typically modulated by afferent feedback, which adjusts the motor output pattern in accordance with the demands of external and internal conditions. In mammals, the respiratory CPG is controlled by sensory signals about lung volume arising from pulmonary stretch receptors (PSRs). It is widely believed that afferent information from PSRs influences the inspiratory-to-expiratory phase transition and contributes to the Breuer–Hering reflex (BHR), thereby providing the crucial afferent signal for the inspiratory off-switch (IOS, Breuer, 1868; Hering, 1868; reviewed in Kubin et al. 2006). However, this view is complicated by evidence that the role of PSR afferents in the control of inspiratory duration changes during the course of development. For example, in neonates the influence of PSR afferents upon IOS is greater than in adults, in which the influence of central mechanisms appears to dominate (Mortola et al. 1984; Trippenbach, 1994). It is unknown when the shift in balance between sensory and central mechanisms controlling IOS occurs during postnatal development, and what neural mechanisms are involved in this transition – questions that represent fundamental unaddressed challenges in developmental respiratory physiology.

In the BHR, afferent feedback from PSRs activates neural circuitry within the medullary respiratory network involved in the control of IOS (Hayashi et al. 1996; Haji et al. 2002; Cohen & Shaw, 2004; Krolo et al. 2005). In addition, neural processes within the pontine Kölliker–Fuse nuclei (KFn) interact with the same circuitry to influence IOS (Oku & Dick, 1992; Haji et al. 2002; Okazaki et al. 2002; Cohen & Shaw, 2004; Alheid et al. 2004; Rybak et al. 2004, 2007, 2008; St-John & Paton, 2004). For example, in animals deprived of afferent feedback from PSRs following vagotomy, lesions or pharmacological suppression of neurotransmission within the KFn cause apneusis (pathologically prolonged inspiration) (Marckwald, 1887; Richter 1982; Morrison et al. 1994; St-John, 1998; Dutschmann & Herbert, 2006; Smith et al. 2007), whereas electrical stimulation within the KFn can terminate inspiration, reset phase, and trigger IOS (Oku & Dick, 1992; Chamberlin & Saper, 1994; Dutschmann & Herbert, 1996, 2006; Okazaki et al. 2002). The effects of the KFn on IOS are associated with corresponding changes in postinspiratory activity. For example, apneusis caused by lesions of the KFn is associated with loss of postinspiratory motor activity, whereas KFn stimulation leads to prolonged postinspiratory motor activity (Dutschmann & Herbert, 2006). Thus, the PSR vagal input and the pontine KFn interact to control the IOS.

The pontine and medullary nuclei involved have dense anatomical interconnections (Smith et al. 1989; Ellenberger & Feldman, 1990; Herbert et al. 1990; Nunez-Abades et al. 1993; Dobbins & Feldman, 1994; Gaytan et al. 1997) representing a potential neural substrate for the alterations of IOS that are seen with manipulations of the KFn (Dutschmann et al. 2004, 2008). The activity of pontine respiratory-associated neurones is dependent upon vagal feedback (Ezure et al. 2002; Ezure, 2004; Ezure & Tanaka, 2006; Dick et al. 2008; Segers et al. 2008), but also correlates with the activities of medullary respiratory neurones (Segers et al. 2008). The IOS-modifying mechanisms within the KFn are thought to involve NMDA receptor-mediated glutamatergic neurotransmission (Pierrefiche et al. 1992; Ling et al. 1994; Fung et al. 1994; Bonham, 1995; St-John, 1998), consistent with findings that NMDA receptors (NMDA-Rs) are densely expressed within the KFn (Monaghan & Cotman, 1985; Guthmann & Herbert, 1999).

The KFn can influence the IOS in a short-time domain. BHR feedback evokes experience-dependent spatiotemporal (Hebbian) synaptic plasticity in respiratory control circuits that depend on neural activity within the neural circuits concerned with the mediation IOS (Siniaia et al. 2000; Song & Poon, 2004; Poon & Young, 2006). Such plasticity contributes to the habituation to PSR-derived input, thereby diminishing the manifestation and influence of the BHR during vagal stimulation and causing rebound increases in respiratory frequency after vagal stimulation (Siniaia et al. 2000). Habituation to vagal feedback is one factor that might explain why the contribution of BHR feedback to IOS is weak in adult animals (Kubin et al. 2006). In contrast, we examined development and tested the hypothesis that the role of vagal feedback in shaping the breathing pattern diminishes as a result of the maturation of the central respiratory control system, in particular the KFn.

To determine the developmental time course and mechanisms involved in the postnatal shift of balance between the influences of the vagally mediated BHR and KFn mechanisms upon IOS, we used in situ perfused brainstem preparations of young rats aged between postnatal days (P) 4 and P28. The in situ preparation is advantageous because it generates eupnoea-like, three-phase motor patterns in the absence of anaesthesia and afferent PSR-derived feedback, and it excludes the potentially confounding influences of higher motor centres and variations in blood gas levels. We found that repetitive BHR-like feedback in neonatal (< P15) animals produced highly stereotyped response patterns, whereas juvenile (> P15) animals exhibited progressive adaptation of breathing parameters that ultimately led to anticipatory IOS, preceding the onset of vagal stimulation. The emergence of anticipatory IOS was abolished after blockade of NMDA-R-mediated neurotransmission within the KFn. Moreover, after NMDA-R blockade mature animals showed immediate and robust entrainment of breathing to the vagal IOS signal, similar to that observed in neonates. Together, these findings indicate that KFn-mediated and NMDA-R-dependent synaptic mechanisms for IOS become increasingly important with ongoing postnatal maturation.


All experimental procedures were performed in accordance with European Community and National Institutes of Health guidelines for the care and use of laboratory animals. The ethical committee of the Georg August University of Göttingen approved the study.

Perfused brainstem preparation

The experiments were performed using the arterially perfused brainstem preparation (Paton, 1996). Rats of different postnatal stages between P4 and P28 were anaesthetized deeply in an atmosphere saturated with isoflurane (1-chloro-2,2,2-trifluoroethyl-difluoromethylether; Abbott, Wiesbaden, Germany). Once the animal failed to respond to noxious pinch to the tail or a hind paw, the whole animal was transected below the diaphragm, and the rostral half was transferred into ice-cold (5°C) artificial cerebrospinal fluid (aCSF) gassed with carbogen (95% O2 and 5% CO2), decerebrated at the precollicular level and cerebellectomized. The lungs were removed. The left phrenic nerve was isolated and cut at the level of the diaphragm. The descending aorta was isolated from the ventral surface of the spinal column. These initial procedures required approximately 5–10 min. The preparation was then transferred to a recording chamber. The descending aorta was cannulated and perfused using a peristaltic pump (Watson-Marlow, Wilmington, MA, USA) with carbogen-gassed aCSF at 31°C containing Ficoll (1.25%; Sigma-Aldrich, Steinheim, Germany) to maintain colloid-osmotic pressure. The perfusate contained (in mm): NaCl 125, KCl 3, KH2PO4 1.25, CaCl2 2.5, MgSO4 1.25, NaHCO3 25, d-glucose 10 and 1.25% Ficoll. The osmolarity of the perfusate was 300 ± 10 mosmol l−1 and the pH was 7.35 ± 0.05 when gassed with carbogen. The perfusate was filtered and passed through bubble traps to remove gas bubbles. The perfusate leaking from the preparation was collected and recirculated after reoxygenation. Rhythmic contractions of respiratory muscles returned within 3–5 min after the onset of reperfusion. Respiratory-related movements were abolished by vecuronium bromide (0.3 μg ml−1; Inresa, Freiburg, Germany). In contrast to an originally published protocol (Paton, 1996), the heart was removed to abolish the electrical and mechanical artefacts associated with the cardiac excitation and contraction.

Nerve recording

As an index of respiratory activity we recorded the phrenic nerve activity (PNA) via suction electrodes. Activity was amplified (differential amplifier; Tektronix, Beaverton, OR, USA), digitized (PowerLab/8SP ADInstruments, Sydney, Australia) and stored on a computer using Chart v5.0/s software (ADInstruments). In each preparation, PNA was used to fine-tune the perfusion in order to obtain a ramping envelope of the integrated PNA. Consequently, flow rates (5–22 ml min−1) and perfusion pressures (40–70 mmHg) varied with age of the rats.

Lung inflation

To demonstrate the physiological response to repetitive activation of PSRs, we performed an initial set of experiments (n= 3) in which the lungs of the perfused brainstem preparation were left intact. The trachea was intubated and the lungs were inflated (1–2 ml tidal volume) at 60 inflations min−1, which was the same frequency as the short trains of vagal stimulation. To determine the tidal volume required to reach threshold for IOS, we increased the tidal volume progressively until 3–5 lung ventilations triggered a transient apnoea (see Fig. 2A). This tidal volume was used for the subsequent experimental protocol that involved repetitive trials of lung inflation.

Figure 2.

Response of phrenic nerve activity (PNA) to consecutive trials of repeated lung inflation in the perfused lung–brainstem preparation from a juvenile rat (P19)
A, upper trace: initial response pattern of integrated phrenic nerve activity (∫PNA) to repetitive episodes of lung inflation (2 ml, 60 min−1; see scale bar under lower traces). Lower traces (1.–3.): high-speed traces of ∫PNA (black) corresponding to the 20 s intervals indicated below upper trace, with lung inflation (grey) superimposed. During the episode of repetitive lung inflation, PNA ceased after lung inflation began. Thus, PNA was entrained to the interval between lung inflations. The burst frequency (FR) of PNA increased to match that of the pump and the peak amplitude of ∫PNA decreased. B, upper trace: response of PNA to lung inflation during the 7th lung inflation trial. Lower traces (1.–3.): traces of lung inflation (grey) superimposed on ∫PNA (black) show the dissociation between lung inflation and the termination of PNA; specifically the apnoeic period at the beginning of the lung inflation trial was absent and the IOS occurred at a lower lung volume than in A. C, progressive decrease (ANCOVA P < 0.01) of lung volume required to terminate PNA during the seven lung inflation trials.

Electrical stimulation of vagal afferents

Experiments were initiated 30–45 min after the perfusion of the preparation with aCSF began. At this time the pattern of PNA had stabilized and was characterized by an augmenting ramp of the integrated burst and a frequency of 20–30 bursts min−1. We introduced fictive input that mimicked feedback from PSRs by stimulating the central end of the left vagal nerve using a programmable stimulus generator (Master 8, A.M.P.I., Jerusalem, Israel). The nerve was placed on an uninsulated end of a teflon-insulated silver wire and was isolated from neighbouring tissue by encasing the electrode tip and nerve in paraffin wax that had a low melting point. The indifferent electrode was placed nearby in the surrounding tissue. Dynamic adaptation of the respiratory network activity to fictive PSR feedback was investigated in several ways. First, to mimic repetitive rhythmic feedback of PSRs, we applied short stimulus trains (50 μs stimulus duration, 20 Hz stimulus frequency, 200–300 ms train duration, 0.5–2 mA intensity) for 1 min at a frequency of 1 Hz. This stimulus protocol was chosen because it reflected a breathing frequency of 60 breaths min−1, a physiological value for rats. For clarity, we will refer to the application of 1 min of short stimulus trains as a ‘stimulus trial’. Second, to analyse vagally induced plasticity phenomena, we repeated these 1 min stimulus trials 10 times (or 7 times after NMDA receptor (NMDA-R) blockade), with a 2 min interval between trials. The stimulus threshold was determined by increasing the stimulus intensity of single stimulation trains applied manually during the onset of PNA until a stimulus triggered a premature termination of PNA. Later the stimulus intensity was adjusted to 1.5 times threshold and kept constant during the subsequent experimental protocol which involved repetitive trials of vagal stimulation. The stimulus intensity required to trigger IOS varied but did not correlate with postnatal ages and rather reflected the quality of the contact of the stimulation electrode with the nerve after insulation.

Local and systemic NMDA receptor blockade

In 13 experiments in rats between P16 and P21, NMDA-Rs within the dorsolateral pons were blocked by microinjecting (2R)-amino-5-phosphonovaleric acid (AP5), a selective NMDA-R antagonist (ICN Biomedicals, Costa Mesa, CA, USA). Injections were made at stereotactic co-ordinates: 0.1–1 mm caudal to the inferior colliculi and 2–3 mm lateral to the midline. Individual barrels of a triple-barrelled pipette were filled with 10 mm glutamate (ICN Biomedicals), 10 mm AP5 and 2% Pontamine Sky Blue (Sigma-Aldrich). The injected volume ranged from 20 to 50 nl and was monitored by the movement of the liquid meniscus examined by a microscope equipped with a reticule. Injection sites were first characterized by using glutamate injections to evoke respiratory modulation (e.g. a brief apnoea). After the characterization, we injected AP5 at the same locus, followed by Pontamine Sky Blue for later histological verification. The pipette was then immediately moved to the same co-ordinates on the contralateral side, and AP5 and Pontamine Sky Blue were injected sequentially. To minimize elapsed time and washout of AP5, glutamate injections were not performed on the contralateral side. Even unilateral AP5 injections caused a prolongation of time of inspiration (TI) and a decrease in PNA burst frequency. Bilateral NMDA blockade transformed the breathing pattern to apneusis as described previously (Dutschmann & Herbert, 2006). Following the stimulation protocols, the brainstem was removed and fixed in 4% paraformaldehyde for histological analysis.

In additional experiments (n= 5) also performed in preparations from rats between P16 and P21, we administered MK-801 (dizocilpine; ICN Biomedicals) systemically via the perfusate to induce global NMDA-R blockade. The concentration of MK-801 was progressively increased in 2.5 μm increments until a cumulative concentration of 20 μm was reached. We considered this MK-801 concentration to be effective because the TI was doubled. As stated earlier, only seven stimulus trials were performed due to potential washout of AP5 or progressive poisoning of the brainstem with MK-801 (see Table 2).

Table 2.  Summary of changes in respiratory variables after NMDA receptor blockade
A. Respiratory phase durations before vagal stimulation
Adolescent (P22–P28) ControlNMDA-R blockANCOVA
1st trial3rd trial7th trial
  1. Summary of changes after systemic (MK-801 in perfusate) or local (microinjection of AP5 into the KF) NMDA receptor blockade before (A), during (B) and after (C) consecutive vagal stimulation trials. *P < 0.05, **P < 0.01, ***P < 0.001, two-tailed Student's paired t tests. Definitions as in Table 1.

AP5 into KF T I 0.82 ± 0.092.84 ± 0.17 ***2.10 ± 0.182.21 ± 0.17 P < 0.01**
T E 1.72 ± 0.084.01 ± 0.45***3.83 ± 0.334.11 ± 0.34 P= 0.73
T TOT 2.54 ± 0.036.85 ± 0.5 ***5.92 ± 0.446.33 ± 0.40 P= 0.22
MK-801 syst. T I 0.84 ± 0.151.40 ± 0.2*1.56 ± 0.031.59 ± 0.40 P < 0.01**
T E 1.49 ± 0.813.81 ± 0.33**5.10 ± 0.486.99 ± 1.01 P < 0.01**
T TOT 2.3 ± 1.065.19 ± 0.34**6.66 ± 0.498.58 ± 1.02 P < 0.01**
B. Respiratory phase durations during vagal stimulation
Adolescent (P22–P28)Before Stim. 1st trial3rd trial7th trialANCOVA
AP5 into KFSee A T I 0.53 ± 0.02***0.56 ± 0.020.59 ± 0.20 P= 0.21
T E 0.85 ± 0.40***0.83 ± 0.400.78 ± 0.03 P= 0.24
T TOT 1.38 ± 0.01***1.38 ± 0.051.37 ± 0.04 P= 0.61
MK-801 syst.See A T I 0.76 ± 0.05**0.96 ± 0.050.97 ± 0.06 P < 0.01**
T E 3.76 ± 0.214.99 ± 0.415.38 ± 0.60 P < 0.01**
T TOT 3.53 ± 0.24*5.95 ± 0.416.36 ± 0.62 P < 0.01**
C. Respiratory phase durations after vagal stimulation
Adolescent (P22–P28)Before Stim. 1st trial3rd trial7th trialANCOVA
AP5 into KFSee A T i 1.72 ± 0.111.78 ± 0.131.78 ± 0.14 P= 0.80
T E 2.53 ± 0.223.67 ± 0.323.51 ± 0.30 P= 0.42
T TOT 4.26 ± 0.305.45 ± 0.405.30 ± 0.38 P= 0.68
MK-801 syst.See A T I 1.43 ± 0.021.59 ± 0.041.66 ± 0.04 P < 0.01**
T E 4.09 ± 0.385.27 ± 0.507.02 ± 0.95 P < 0.01**
T TOT 5.53 ± 0.396.86 ± 0.528.68 ± 0.96 P < 0.01**

Histological analysis

After fixation of brainstems for several days in 4% paraformaldehyde–20% glucose, serial 50 μm cryosections were prepared using a freezing microtome (Reichert-Jung, Vienna, Austria) and stained with neutral red, and the locations of the microinjections were documented on schematic drawings of coronal sections showing the parabrachial complex of the dorsolateral pons.

Data analyses

To investigate developmental changes of the vagally evoked entrainment of PNA, we pooled preparations derived from animals of different postnatal stages into four age groups: neonatal (P4–P8), intermediate (P9–P15), juvenile (P16–P21) and adolescent (P22–P28). We determined total respiratory cycle length (TTOT) and the durations of the inspiration (TI) and expiration (TE) from the integrated PNA traces. Breathing variables (TTOT, TI, TE) were analysed in 1 min periods before (pre-stimulation), during (stimulation), and after (post-stimulation) the repetitive vagal stimuli. Progressive changes of the mean respiratory parameters were tested statistically by regression analysis (ANCOVA; Systat Software, Inc., Richmond, CA, USA). We also calculated the time elapsed until the post-stimulus respiratory activity returned to the pre-stimulus level. The dynamic adaptation of TTOT is illustrated and summarized in Poincaré plots (Fig. 7). With the exception of the Poincaré plots, data are presented as mean ±s.e.m.

Figure 7.

Variability of TTOT during repetitive vagal stimulation trials illustrated in Poincaré plots
Poincaré plots of the pooled TTOT during consecutive trials of vagal stimulation for each age group (Neonatal, top; Intermediate, second; Juvenile, third; and Adolescent, fourth; after AP5 in the dorsolateral pons (KFn), fifth; and systemic MK-801, bottom). The Poincaré plots in the adolescent and juvenile age groups reveal a progressive increase in 1:1 entrainment during stimulation trials because the points became a tighter cluster from trial 1 (left column of Poincaré plots) to trial 10 (right column of Poincaré plots). In contrast, the Poincaré plots from the neonatal and intermediate age groups did not change from the first to the tenth stimulus trial and consistently showed clusters of points that indicated 1:1 entrainment. After blocking NMDA receptor transmission in the KFn with AP5 microinjections in the adolescent age group, the entrainment pattern reverted to the neonatal stage and 1:1 entrainment is evident from the first to the seventh stimulus trial. However, after blocking NMDA receptors systemically with MK-801 in the perfusate, the pattern simply did not entrain. For the adolescent and juvenile age groups, the temporal relationship between the onset of a stimulus train and PNA (right-hand column, see Fig. 1) changed progressively from trial to trial. As the 1:1 entrainment increased there was a progressive shift from the vagal stimulus eliciting the inspiratory off-switch (sensory IOS) to an anticipatory IOS (central) occurring just prior to the onset of vagal stimulus. In contrast, in the neonatal and intermediate age groups the IOS was sensory-mediated throughout each stimulation trial. The same constant sensory-mediated IOS during all stimulation trials was observed after NMDA receptor blockade in the KF.

To investigate long-term changes of respiratory variables in response to either tonic or rhythmic vagal stimulation, we compared the control activity between the first stimulus trial with the post-stimulus activity following the last stimulation trial. Statistical significance was tested by two-tailed paired t tests (Excel; Microsoft Inc., Bellevue, WA, USA). Probability values < 0.05 were considered significant.

Analyses of the stimulus phase relations

The repetitive vagal stimuli were not phase-triggered, thus stimulus trains could occur during any phase of the respiratory cycle. To assess the temporal relationship between PNA and vagal stimulation, we used the termination of PNA as a reference point and defined it as T= 0. We calculated the latency from offset of PNA (T= 0) to the onset of the individual stimulus train (see Fig. 1). A negative number for the latency indicates the onset of the stimulus train during PNA, whereas a positive number indicates that the PNA terminated spontaneously prior to the stimulus (Fig. 1). For statistical analysis, we calculated the mean latency for each of the short (200–300 ms) stimulus trains (n= 60) during the 1 min stimulus trial, and compared the mean across stimulus trials (n= 10) within a specific age group and among the different developmental stages. Significance of progressive changes in the PNA–stimulus relation was analysed with ANCOVA (Systat Software, Inc.).

Figure 1.

Analysis of relationship between stimulus onset and respiratory phase
Illustration of analysis quantifying the time interval between the onset of a vagal stimulus train and the inspiratory off-switch (IOS, thick grey vertical bar), which is also the phase transition between inspiration and expiration.


Entrainment of phrenic nerve activity (PNA) to lung inflation

Entrainment We determined the pattern of PNA during lung ventilation in arterially perfused in situ preparations of juvenile rats (P16–P21) to determine the influence of lung inflation upon ventilatory motor pattern. Lung inflation was not ‘cycle-triggered’ and was independent of PNA to permit analysis of the adaptation of the spontaneous motor pattern to the fixed pattern of lung inflation. During seven repetitive trials delivered at 2 min intervals, the lungs were inflated at a rate of 60 min−1 for 1 min periods by means of a respirator (Fig. 2A). The onset of lung ventilation evoked a brief cessation of PNA activity for 15.1 ± 3.4 s, followed by entrainment of PNA to the lung ventilation pattern. Repetitive lung inflation produced a stable 1:1 entrainment to the lung inflations in all cases (n= 3); in one case, PNA entrained 1:1 to lung ventilation immediately after transient apnoea (Fig. 2A), in the other two (not shown), PNA entrained briefly (10.7 ± 2.1 s) at 1:2 or 1:3 before entraining 1:1 with lung inflation. The entrainment pattern was such that lung inflation terminated PNA and both TI and TE were shortened. After the 1 min lung inflation trial, respiratory frequency (FR) decreased gradually.

Decreasing lung volume required to trigger IOS During repetitive inflation trials, the lung volume required to trigger the termination of PNA decreased progressively (Fig. 2B and C). This was apparent even in the raw signal, because during the last trial, cessation of PNA almost coincided with the onset of lung inflation, and peak integrated PNA did not coincide with lung inflation (Fig. 2B). For the group of juvenile rat preparations, the decrease in lung volume required to trigger IOS decreased significantly across the repetitive inflation trials (Fig. 2C). The decreased lung volume required to trigger IOS during last lung inflation trials indicated that IOS was initiated before sufficient lung inflation and PSR input could have triggered IOS, as observed during the initial trials.

Persistance of the increase in FR following lung inflation After each lung inflation trial, FR tended to exceed the immediate baseline FR (1st trial, 16.2 ± 2.8 vs. 21.6 ± 6.1 bursts min−1, compare beginning and end of Fig. 2A; 7th trial, 24.3 ± 5.3 vs. 30.9 ± 7.4 bursts min−1, compare beginning and end of Fig. 2B). The accumulated effect resulted in FR increasing significantly with repetitive inflation trials (16.2 ± 2.8 vs. 30.9 ± 7.4 bursts min−1, P < 0.05, compare beginning of Fig. 2A with end of 2B). These data suggest that the neural network controlling rhythm is malleable by sensory input, and retains the evoked effect even after sensory input has ceased.

The response of PNA to rhythmic vagal stimulation

To facilitate understanding the observed phenomenon in IOS during repetitive lung inflation trials (Fig. 2), we adopted rhythmic stimulation of the vagal nerve. This permitted us to deliver afferent input that mimicked phasic feedback from slowly adapting PSRs while more precisely measuring changes in the timing of the vagally mediated inspiratory off-switch (IOS).

In a juvenile (P19) rat preparation, vagal stimulation at 1.5 times threshold terminated PNA but was insufficient to produce the prolonged apnoea observed with lung inflation, indicating that the stimulus intensity selected was not greater than that evoked by lung inflation (Fig. 3A). In the first trial (Fig. 3A), PNA became weakly entrained to the IOS signal provided by vagal stimulation (asterisks denote skipped intervals Fig. 3A). With the 10th stimulus trial, 1:1 entrainment with PNA during the stimulus-off period was well established (few asterisks in Fig. 3B). As predicted by the lung inflation trials, PNA appeared to be terminating prior to the stimulus by the end of the vagal stimulation trials (Fig. 3B‘Zoom’), indicating that repeated stimulus trials not only entrained the pattern more consistently but caused IOS to precede rather than follow the fictive PSR input.

Figure 3.

Response of PNA to repetitive vagal stimulation trials in the perfused brainstem preparation of a juvenile rat (P19)
A, upper trace: PNA response pattern to the first of ten repetitive vagal stimulus trials (bar). Lower traces (1.–3. and ‘Zoom’): high-speed traces of PNA (black) with the stimulus artefact (grey) of vagal stimulus superimposed on the PNA trace. The initial response was characterized by weak entrainment because PNA bursts skipped intervals between vagal stimuli (asterisks over traces) and because PNA peak amplitude decreased during the stimulus trial. Entrainment, i.e. PNA during the inter-stimulus interval, did not occur even by the end of the first stimulus (trace 3.). During a brief period of entrainment in the first stimulus trial (Zoom), vagal stimuli terminated the last three bursts of PNA. B, upper trace: response of PNA to vagal stimulation during the 10th vagal stimulus trial. Lower traces (1.–3.): the high-speed traces revealed an increased quality of entrainment after nine repetitive trains of vagal stimulation. In trace 2, each burst of PNA occurred during each interval between vagal stimuli. In the highest speed trace (Zoom), arrows point to bursts of PNA that ceased spontaneously just before the onset of the vagal stimuli.

Developmental changes in PNA response to repetitive activation of vagal afferents

To ascertain in what manner, and with what time course, postnatal development alters the ability of repetitive vagal afferent input to influence PNA response patterns and shift the timing of IOS, we tested such responses in perfused brainstem preparations from rats of four different postnatal/developmental stages: neonatal (P4–P8), intermediate (P9–P15), juvenile (P16–P21) and adolescent (P22–P28).

Baseline respiratory rhythm slowed with increasing ages until P16, owing to increases in both TI and TE. Baseline FR did not change further in preparations from juvenile vs. those of adolescent rats (Table 1). These data indicate that development of the rhythm-generating mechanism stabilizes by P16.

Table 1.  Summary of developmental changes in respiratory variables
A. Respiratory phase durations before vagal stimulation
   1st trial5th trial10th trialANCOVA
  1. Summary of progressive changes in respiratory cycle length (TTOT), duration of inspiration (TI) and expiration (TE) before (A), during (B) and after (C) consecutive vagal stimulation trials. *P < 0.05, **P < 0.01, ***P < 0.001, two tailed Student's paired t tests.

Adolescent (P22–P28)  T I 0.57 ± 0.010.56 ± 0.010.54 ± 0.01 P= 0.52
  T E 2.62 ± 0.052.38 ± 0.052.24 ± 0.05   P < 0.001**
  T TOT 3.19 ± 0.052.95 ± 0.052.78 ± 0.05   P < 0.001**
Juvenile (P16–P21)  T I 0.57 ± 0.030.56 ± 0.020.55 ± 0.02 P= 0.99
  T E 2.53 ± 0.432.12 ± 0.291,74 ± 0.24 P= 0.72
  T TOT 3.08 ± 0.432.68 ± 0.272.30 ± 0.22 P= 0.62
Intermediate (P9–P15)  T I 0.50 ± 0.050.52 ± 0.050.53 ± 0.07 P= 0.99
  T E 1.76 ± 0.201.49 ± 0.151.37 ± 0.2  P= 0.94
  T TOT 2.25 ± 0.192.03 ± 0.191.90 ± 0.27 P= 0.86
Neonate (P4–P8)  T I 0.43 ± 0.010.38 ± 0.01 0.39 ± 0.013 P= 0.90
  T E 1.96 ± 0.161.63 ± 0.261.68 ± 0.44 P= 0.95
  T TOT 2.40 ± 0.43 2.0 ± 0.292.10 ± 0.45 P= 0.92
B. Respiratory phase durations during vagal stimulation
 Before 1st trial5th trial10th trialANCOVA
Adolescent (P22–P28)See A T I 0.32 ± 0.01*0.31 ± 0.010.29 ± 0.01   P < 0.001***
T E 1.12 ± 0.31*0.83 ± 0.020.83 ± 0.22   P < 0.001***
T TOT 1.44 ± 0.34*1.14 ± 0.021.12 ± 0.21   P < 0.001***
Juvenile (P16–P21)See A T I 0.26 ± 0.01*0.24 ± 0.020.24 ± 0.01P= 0.796
T E 1.12 ± 0.07*0.84 ± 0.020.83 ± 0.02   P < 0.001***
T TOT 1.38 ± 0.07*1.09 ± 0.011.06 ± 0.02   P < 0.001***
Intermediate (P9–P15)See A T I   0.22 ± 0.014**0.23 ± 0.020.21 ± 0.01 P= 0.87
T E  0.86 ± 0.021*0.83 ± 0.020.88 ± 0.06 P= 0.67
T TOT  1.08 ± 0.01**1.06 ± 0.021.08 ± 0.05 P= 0.88
Neonate (P4–P8)See A T i   0.22 ± 0.011**0.22 ± 0.020.24 ± 0.02 P= 0.99
T E  0.86 ± 0.12**0.90 ± 0.030.84 ± 0.05 P= 0.99
T TOT   1.08 ± 0.013**1.10 ± 0.201.09 ± 0.03 P= 0.98
C. Respiratory phase durations after vagal stimulation
 Before 1st trial5th trial10th trialANCOVA
Adolescent (P22–P28)See A T I 0.51 ± 0.010.49 ± 0.01 0.47 ± 0.01*   P < 0.001***
T E 2.48 ± 0.052.33 ± 0.05 2.25 ± 0.05*   P < 0.001***
T TOT 2.99 ± 0.052.82 ± 0.05 2.72 ± 0.05*   P < 0.001***
Juvenile (P16–P21)See A T I 0.54 ± 0.030.50 ± 0.03 0.51 ± 0.03* P= 0.99
T E 1.90 ± 0.2 1.53 ± 0.13 1.36 ± 0.17* P= 0.23
T TOT 2.44 ± 0.192.03 ± 0.13  1.87 ± 0.18** P= 0.11
Intermediate (P9–P15)See A T I 0.49 ± 0.060.49 ± 0.060.51 ± 0.08 P= 0.99
T E 1.50 ± 0.181.40 ± 0.161.25 ± 0.25 P= 0.99
T TOT 1.99 ± 0.211.88 ± 0.211.77 ± 0.26 P= 0.99
Neonate (P4–P8)See A T I 0.36 ± 0.020.35 ± 0.010.39 ± 0.01 P= 0.82
T E 1.62 ± 0.301.53 ± 0.271.51 ± 0.17 P= 0.99
T TOT 1.99 ± 0.301.88 ± 0.281.87 ± 0.19 P= 0.99

Postnatal age-dependent responses to the first stimulus trial The first vagal afferent stimulus trial evoked decreases in both TI and TE in all age groups (Table 1), but only led to a decrease in PNA peak amplitude in preparations from juvenile and adolescent rats (25.1 ± 12.6% and 31.6 ± 11.3%, respectively). After the first stimulus trial, the post-stimulus pattern was characterized by subtle and non-significant changes in TI and TE and in each age group (Table 1).

Postnatal development-associated increases in malleability of PNA response patterns during repetitive stimulation trials Our tests of repetitive stimulation in perfused brainstem preparations of rats of different postnatal ages revealed progressive changes in respiratory parameters in all age groups from the first stimulus trial to the 5th and 10th stimulus trials (Table 1 and Fig. 7). In response to the initial stimulus trial, perfused brainstem preparations of rats from all ages studied exhibited increased FR, and entrainment of PNA in a 1:1 ratio to the interval between stimulus trains and to the 1 Hz frequency of the stimulus trains (Table 1; Figs 3A and 4A). Preparations from juvenile and adolescent rats exhibited progressive decreases in TTOT, TI and TE throughout the 10 trials of rhythmic vagal stimulation. TTOT progressively decreased to a value of approximately 1 s (Figs 3 and 7; see ANCOVA analysis, Table 1). These responses contrasted markedly with the response patterns observed in the neonatal and intermediate age groups, which after the first trial were stereotyped, remaining unchanged during the ensuing repetitive stimulation trials (Fig. 4B), and exhibited entrainment of PNA with activity during the ‘stimulus-off’ intervals.

Figure 4.

Response of PNA to repetitive vagal stimulation trials in the perfused brainstem preparation of an intermediate-aged rat (P10)
A, upper trace: PNA response pattern to the first of ten repetitive vagal stimulus trials in an in situ preparation from an animal < P15. Lower traces (1.–3.): high-speed traces of PNA (black) with the artefact (grey) of the vagal stimulus superimposed on the PNA trace. Even the initial response was characterized by an immediate high quality entrainment of PNA bursts to the interval between the short (200–300 ms) trains of vagal stimuli. Further, the peak amplitude of PNA remained and was similar to baseline. B, upper trace: response of PNA to vagal stimuli during the 10th stimulus trial. Lower traces (1.–3. and Zoom): high-speed traces revealed a 1:1 entrainment pattern that was similar after repetitive trials to that in the first trial. In Zoom, arrows point to individual vagal stimuli trains that occurred during PNA and activated IOS.

Influence of postnatal development on timing and relation of individual vagal stimulus trains with the PNA off-switch Because vagal stimulation was not triggered by PNA in our model system, we investigated the spontaneous adaptation of the respiratory network activity to an independent pattern of fictive PSR feedback. As illustrated in Fig. 1, the analysis assumed that if a vagal stimulus train began during a PNA burst then it would terminate the burst, and this would reflect an afferent-evoked inspiratory off-switch; in contrast, if a train began after a PNA burst then central mechanisms must have terminated the burst and this would represent a centrally evoked IOS.

The temporal relationship between stimulus trains and the IOS varied with development. Vagal stimulation evoked cessation of PNA and IOS consistently in the neonatal- and intermediate-aged rat preparations (Fig. 7 right-hand column), but in those of juvenile and adolescent rats the IOS shifted progressively from afferent-evoked to centrally mediated during the course of the 10 stimulation trials (Fig. 7 right-hand column). This shift occurred during vagal stimulation, as FR increased and PNA was entrained to the intervals between vagal stimulation. Accordingly, the shift from afferent- to central-mediated IOS correlated with increasing periods of 1:1 entrainment of PNA to the vagal stimulation (Fig. 7).

Our analyses of respiratory variables during the recovery period following each stimulus trial are shown in Table 1C. Only the most mature (adolescent) age group exhibited progressive changes in stimulus rebound of the breathing pattern, seen as progressive shortening of TTOT after each stimulus trial. Other age groups showed no such changes. Nevertheless, preparations from all age groups except neonates did exhibit significant changes in the post-stimulus interval required for PNA burst duration to return to the mean pre-stimulation TTOT. Accordingly, in the intermediate age group, the recovery interval increased progressively from 15.2 ± 1.8 to 31.1 ± 3.8 s (ANCOVA, P < 0.001) between the 1st and 10th stimulation trials; in juveniles, from 17.6 ± 1.0 to 30.0 ± 3.1 s (ANCOVA, P < 0.001); and in the adolescent group, from 16.3 ± 1.5 to 25.6 ± 2.7 s (ANCOVA, P < 0.05). No such changes occurred following the stimulus trials in neonatal preparations, indicating that they were incapable of altering their pattern and their post-stimulus respiratory behaviours, contrasting with the more mature preparations, in which the control networks were malleable and their properties depended on afferent inputs.

Systemic or local NMDA-R blockade prevents the emergence of anticipatory IOS by repetitive vagal stimulation

The anticipatory termination of inspiration prior to the onset of the stimulus, observed in rats aged > P16, indicated that the repetitive vagal stimulation caused a progressive switch from afferent to central initiation of the IOS. Because central mediation of the IOS may depend on NMDA-Rs within the KFn, we microinjected AP5, a selective NMDA-receptor antagonist, bilaterally into the KFn.

Microinjecting AP5 (n= 5) into the KFn increased TI and caused an apneustic breathing pattern (Fig. 5A, Table 2C). Applying rhythmic vagal stimulation immediately attenuated the AP5-evoked apneusis, whereupon PNA entrained predominantly in a 1:1 ratio with the stimulation with activity during the stimulus-off period (Table 2A, Fig. 4B, Fig. 7). However, the mean TI was still prolonged in comparison with that observed without NMDA-R blockade (0.53 ± 0.02 vs. 0.32 ± 0.01 s, P < 0.01, Tables 1 and 2), and the progressive increases in TTOT,TI and TE that were evoked by repetitive stimulus trials in NMDA-R-intact rat preparations were prevented (n= 7, Table 2). In the AP5-treated preparations, the adaptation of TTOT to 1 Hz vagal stimulation was similar to that seen in the neonatal and intermediate age groups (Fig. 7). Accordingly, after NMDA-receptor blockade in KFn, PNA failed to cease spontaneously, and was terminated only upon vagal stimulation. Thus, the progressive shift from afferent- to central-mediated IOS mechanisms was prevented by NMDA-R blockade (Fig. 7; right column).

Figure 5.

Response of PNA to repetitive vagal stimulation trials after AP5 microinjection into the Kölliker–Fuse nucleus in the perfused brainstem preparation from an adolescent rat (P26)
A, traces: baseline PNA (left) became apneustic (right) after bilateral injections of the selective NMDA receptor antagonist AP5 into the pontine Kölliker–Fuse nuclei. B, upper trace: response of PNA to vagal stimuli during the 1st stimulus trial (bar) after NMDA receptor blockade; lower traces (1.–3. and Zoom): high-speed traces of PNA (black) with the artefact from vagal stimulus trains (grey) superimposed. Vagal stimulation activated the IOS and PNA entrained immediately to burst between stimulus trains. Initially, the entrainment pattern was 1:1 (1.), which broke down (2.), and then returned to 1:1 entrainment (3.). In Zoom, arrows point to bursts of PNA that terminated in response to the vagal stimuli. C, response of PNA to vagal stimulation during the 7th consecutive stimulus trial. Lower traces (1.–3. and Zoom): high-speed traces as in B. The 1:1 entrainment pattern was not maintained (see trace 2.). In Zoom, arrows point to bursts of PNA that terminated in response to the vagal stimuli; similar to that in B.

We analysed the rostro-caudal distribution of bilateral injection sites histologically following the experiments, and found that the influence of a given microinjection upon TI was related to its neuroanatomic location. The five injections that were placed bilaterally within the intermediate KF caused an increase of TI (< 200%) (Fig. 6, black circles). Using this efficacy as a criterion for analysis, we analysed the PNA response pattern during repetitive vagal stimulus trials only in the data from these preparations (Fig. 6, black circles). We did not analyse the response patterns to vagal stimulation from preparations that had less effective (< 50–80% prolongation of TI, grey circles; Fig. 6) or ineffective (< 20% prolongation of TI, white circles; Fig. 6) AP5 injections.

Figure 6.

Histological verification of the AP5 injection sites in the dorsolateral pons
Line drawings illustrate the anatomical location of the injection sites (circles) in the parabrachial/Kölliker-Fuse (PB/KF) complex in the dorsolateral pons, from rostral (top) to caudal (bottom). The numbers in the circles allow matching of the corresponding pairs of ipsi- (left) and contralateral (right) injection sites for AP5. Injection sites (black circles 1–5) were effective in producing apneusis. Injection sites (grey and white circles 6–13) weakly or non-significantly (respectively) elicited apneusis. Abbreviations: Dll; dorsal nucleus of the lateral lemniscus; KF, Kölliker–Fuse nucleus; ll, lateral lemniscus; Mo5, motor trigeminal nucleus; Pr5, principal sensory trigeminal nucleus; scp, superior cerebellar peduncle. Subnuclei of the parabrachial complex: el, external lateral parabrachial nucleus (inner and outer section, eli, elo); exl, extreme lateral parabrachial nucleus; il, internal lateral parabrachial nucleus; m, medial parabrachial nucleus; s, superior lateral parabrachial nucleus; v, ventral lateral parabrachial nucleus; w, waist area of the parabrachial nucleus.

In order to block NMDA receptors in the entire ponto-medullary respiratory network we administered MK-801 (dizocilpine), an NMDA channel blocker that crosses the blood–brain barrier, by adding it to the perfusate. Such systemic blockade of NMDA receptors caused increased TI and TE, similar to the effects of AP5 microinjections (Table 2, Supplemental Fig. 1A, available online only). Although vagal stimulation reduced TI and TTOT initially, 1:1 entrainment was not observed after MK-801 (Supplemental Fig. 1B, available online only). If the PNA entrained to the vagal stimulus, then the entrainment pattern was in a ratio from 1:3 to 1:6 rather than 1:1. During repetitive vagal stimulus trials, the response pattern deteriorated progressively such that after the fifth or sixth stimulus trial, PNA was not responsive to vagal stimulation (Supplemental Fig. 1C, available online only, Table 2). These data indicate that blockade of NMDA-Rs in the ponto-medullary neuraxis not only affected the IOS but also blocked transduction of the vagal sensory input into the pattern generator.


Our study revealed that during repeated stimulus trials consisting of either lung inflation or vagal stimulation, the incidence of IOS shifted from an afferent input-evoked mode to one that coincided with the very early onset of lung inflation, or occurred prior to the onset of the vagal stimulus in animals > P15. The latter suggests that IOS after repetitive stimulation trials was mediated by central mechanisms although breathing still followed the pattern dictated by the 1 Hz frequency of vagal stimulation or lung inflation. The emergence of the anticipatory IOS was prevented by pharmacological blockade of NMDA-Rs within the pontine KFn. In contrast, a shift from peripheral- to central-mediated IOS timing was not observed in preparations from animals aged < P15, including preparations from neonatal animals. Repeated vagal stimulation at the neonatal stages revealed a highly stereotyped and constant entrainment pattern, an afferent-evoked IOS.

Technical considerations

Over the last 10 years, the perfused brainstem preparation has provided a rhythmic respiratory pattern that reflects that observed in decerebrate in vivo preparations. However, significant differences compared to in vivo experiments include the open loop circulation, a much lower temperature and constant supply of oxygen via the perfusate, even when breathing activity is suppressed experimentally. Thus, changes in the breathing pattern as described in the present study have no influence on the concentration of blood gasses unlike that expected in vivo. Furthermore, electrical stimulation of the vagus used in our study was limited in its intensity and duration to correspond to the physiological activation of PSRs. However, other vagal afferents could have been activated by the electrical pulse even though the magnitude of the pulse was adjusted (short-duration and low current) to preferentially stimulate the large, myelinated PSR afferents. Further, electrical stimulation activated all afferents synchronously. However, the response to the applied stimulation was fully consistent with what is known regarding the effects of PSR inputs on phrenic nerve activity (Karczewski et al. 1980; Budzińska et al. 1981; Widdicombe, 2006). Moreover, the initial experiments using lung inflation support the shift from afferent (vagal stimulation) to central (KFn) mediation of IOS. Specifically in these experiments, PNA was terminated at progressively smaller lung volumes (rising slope of the ventilator trace) in repeated inflation trials, so that after the sixth inflation trial the lung volume was so small that it became subthreshold for activation of all but the lowest threshold PSRs, which would be insufficient to initiate inspiratory termination. Thus, despite technical limitations associated with electrical vagal stimulation, the latter defined clearly the onset of a suprathreshold stimulus and confirm the observation that rhythm was malleable and depended on the vagal feedback.

The experimental protocol involving repetitive activation of vagal afferents allowed us to reveal developmental changes in response to activation of vagal afferents (for review see Poon & Young, 2006). Repetitive stimulation trials are a common approach to investigate different types of Hebbian plasticity in various brain/neural systems including the respiratory control system (Thompson et al. 1972; Poon & Young, 2006). The timing of vagal stimulation was independent of naturally occurring IOS. Anticipatory IOS occurred even though phrenic bursts were entrained during intervals between vagal stimulation trains. Of course, dissociation between PNA and vagal activity does not occur in living animals where PNA causally drives inflation of the lung, linking vagal activity with PNA. Nevertheless the input served as BHR-like feedback in our preparation. The onset of vagal stimulation trains or the rising slope of lung inflation coincided with the last third of PNA in preparations from rats < P15 and in the first trials in preparations from rats > P15. The last third of the phrenic activity represented a regime in which PSR stimulation corresponded to the afferent-mediated IOS.

Central mechanisms involved in inspiratory off-switch (IOS)

The afferent fibres from slowly-adapting PSRs that mediate the BHR project to the ventrolateral nucleus tractus solitarii (nTS) (Berger & Dick, 1987; Kubin et al. 2006). The ventrolateral nTS has reciprocal connectivity with the medullary ventral respiratory column (VRC) and with the pontine KFn (Ellenberger & Feldman, 1990; Herbert et al. 1990; Dobbins & Feldman, 1994; Hayashi et al. 1996; Ezure et al. 2002). The KFn and VRC are reciprocally connected as well (Segers et al. 2008). Because both nTS and KFn are involved in mediating IOS, and IOS is executed in the VRC, the nTS and KFn form the neuraxis that modulates IOS (Dutschmann et al. 2008).

We found here that blocking NMDA-Rs either locally (KFn) or systemically altered the progressive changes in IOS timing observed with repetitive simulation of BHR feedback. This is consistent with previous findings that NMDA-Rs in KFn play an essential role in its function in the control of breathing (Bianchi et al. 1995; Bonham et al. 1995; St-John, 1998). In addition, NMDA receptors are involved in the primary relay mechanisms of BHR feedback within the nTS (Bonham et al. 1995; Miyazaki et al. 1999; Wassermann et al. 2000). The distribution of NMDA-Rs in both the pons and medulla explains the difference between the pattern after local versus systemic application of MK-801 and the severely reduced responsiveness of the entire ponto-medullary network to vagal stimulation systemic blockade of NMDA receptors with MK-801 (online Supplemental Fig. 1). The central (ponto-medullary) and peripheral pathways interact and appear to perform similar functions in the control of IOS. For example, the apneusis evoked by blocking NMDA-Rs in the KFn can be compensated by stimulating the vagi in juvenile rats. Also, we found that entrainment of PNA evoked by afferent feedback was more consistent after blockade of NMDA-Rs within the KFn, suggesting that connections of the KFn to nTS may contribute to controlling the gain of synaptic inputs at the level of the nTS integration and relay of vagal afferent information. Specifically, NMDA-R blockade in the KFn may decrease descending modulation of the nTS and relieve gating of afferent input resulting in an increase in the synaptic gain of PSR inputs, which prevents an anticipated habituation of the BHR at the juvenile stage (Dutschmann et al. 2008). However, changes in IOS that occurred in vagi-intact cats with chronic lesions of the dorsolateral pons included increased tidal volume, TI and TE for breathing at rest (Gautier & Bertrand, 1975), indicating that the remaining respiratory CPG was less sensitive to PSR input. Nevertheless with the experimental protocols used in the present study the baseline breathing frequency was paced and the gain for PSR inputs seems to be increased under the condition of high synaptic drive.

Postnatal development of respiratory control involves the respiration-stabilizing and lung volume-protective Breuer–Hering reflex

Synaptic mechanisms for plasticity within the central respiratory network may develop with postnatal maturation (Dutschmann et al. 2004, 2008). Even though in situ preparations from newborn rats can generate a three-phase breathing pattern in the absence of PSR sensory feedback for several hours (Dutschmann et al. 2000), experiments on neonates suggest that breathing activity in vivo strongly depends on vagal feedback from PSRs (Mortola et al. 1984; Fedorko et al. 1988; Rabbette & Stocks, 1998).

In neonates acute vagotomy causes transient cessation of breathing and the respiratory rhythm remains unstable after it re-emerges; thus, phasic feedback from the lungs is an essential factor that stabilizes neonatal respiratory network activity. The immediate entrainment of PNA by vagal stimulation demonstrated here suggests that the respiratory pattern is highly responsive to afferent inputs. Further, the stereotypical response during repetitive vagal stimulation indicates little or no plasticity in the central processing of the sensory feedback.

Habituation of the BHR, as seen in adult rats (Siniaia et al. 2000), could be life-threatening in neonates, and accordingly the BHR and other respiratory-related reflexes usually do not habituate at ages < P15 (Dutschmann et al. 2004; Mörschel & Dutschmann, 2005). Such reflex habituation is associated with long-term synaptic depression (LTD). For example, in mice that were deficient in an NMDA receptor subunit, an anomalous form of LTD occurred in the nTS, leading to life-threatening respiratory depression and death (Poon et al. 2000). The present study revealed that a progressive shift occurs from stimulus-evoked to anticipatory IOS during a repetitive series of vagal stimulus trials in juvenile (> P15), but not in younger, less-developed rats. This finding raises the possibility that the respiratory network response to pulmonary afferent input may habituate at the level of the nTS after repetitive activation in juveniles as a prerequisite to releasing NMDA-R-mediated synaptic mechanisms in the KFn that can mediate the anticipatory IOS.

In summary, we conclude that as postnatal development advances, both the role of the BHR and the contribution of vagal feedback to IOS diminish, particularly during breathing at rest. In the present study we focused on the role of the BHR-mediating PSRs in IOS control. However, neonates also have vagally mediated inflation reflexes that protect lung volume (Hannam et al. 2000). In contrast, the role of dorsolateral pons in IOS progressively increases with the development. We believe that this mechanism has been underappreciated in its physiological importance.

Postnatal emergence of NMDA-receptor-dependent, BHR-associated pattern learning in the Kölliker–Fuse nucleus

The main finding of the present study is that during postnatal development, a KFn-mediated, NMDA-receptor-dependent learning mechanism may emerge, and this ‘learning’ is associated with repetitive activation of vagal fibres and processing of BHR-associated synaptic inputs. The definition of pattern learning in this context involves the fact that anticipatory IOS occurs only after repetitive lung inflation or vagal stimulation trials. Thus, a specific pattern of sensory feedback is presented and trained and this pattern is learned by the intrinsic synaptic pathways of the ponto-medullary CPG and is recalled during the presence of sensory feedback, as is demonstrated by the anticipatory IOS at later stages of the stimulation trials.

These findings may correlate with recent findings of developmental changes in the characteristics of inhibitory, excitatory but in particular NMDA currents in the KFn (Kron et al. 2007a,b, 2008). We speculate that the observed up-regulation of the NR2D subunit of the NMDA-R in KFn during postnatal development (Kron et al. 2008) could correspond to synaptic strengthening and learning required for anticipatory IOS. The presence of the NR2D subunit in the heteromeric NMDA-R lowers conductance, lengthens deactivation kinetics, diminishes desensitization (Cull-Candy et al. 2001), reduces magnesium block (Arvanian et al. 2004) and permits rapid membrane depolarization and calcium influx following NMDA-R-mediated glutamatergic neurotransmission, without AMPA/kainate receptor-dependent pre-depolarization (Clarke & Johnson, 2006). These mechanisms increase synaptic efficacy. Thus, systemic or local blockade of NMDA-Rs within the KFn can transform the respiratory pattern, as seen here, even without prior pharmacological manipulation of other glutamate receptor types (Bianchi et al. 1995; Bonham, 1995; St-John, 1998). Furthermore, the long deactivation kinetics of NMDA currents provides a background current for a longer time allowing for activity-dependent forms of synaptic plasticity (Vicini & Rumbaugh, 2000) as observed in the present study.


We conclude that during postnatal development, a profound maturation of synaptic mechanisms related to Hebbian plasticity and specific forms of pattern learning occur in the neural circuits involved in the generation and adaptation of the respiratory motor pattern. Breathing is a vital motor behaviour that needs to be functional from birth, but the postnatal changes observed here are consistent with the previously recognized postnatal maturation of the CPG for breathing. We suggest that the pontine influences on the motor pattern increase with the progressive increase in the behavioural repertoire during postnatal maturation. Thus, maturation of the ponto-medullary respiratory control network allows a rapid adaptation and co-ordination of the breathing pattern during exercise or vocalization.

Nevertheless, the redundant and interconnected afferent- and central-mediated IOS mechanisms remain significant physiological mechanisms that contribute to the variable shape of the motor pattern. For example, the pathological changes in the breathing pattern of patients suffering from Rett syndrome may relate to the chaotic fluctuation in the postinspiratory motor output to the upper airways as identified in the breathing pattern of an experimental model of Rett syndrome (Stettner et al. 2007). A detailed analysis revealed that indeed both afferent and central pathways for IOS show pathological changes as a prerequisite for the transformation of the normal and stable breathing pattern to a highly pathological state.



This study was supported by the Bernstein Center for Computational Neurosciences (BCCN, 01GQ0432, M.D.) and NHLBI (Cluster Grant R33 HL087377, M.D., I.A.R. and T.E.D.). I.A.R. was also supported by CRCNS NIH grant R01 NS057815. We thank A. Bischoff for excellent technical assistance. We thank Dr Jeffrey Tatro for his careful reading and editing of the manuscript.