Glycinergic inhibition is essential for co-ordinating cranial and spinal respiratory motor outputs in the neonatal rat

Authors


  • Author's present address M. Dutschmann: Department of Animal Physiology, University of Tübingen, Auf der Morgenstelle 28, 72076 Tübingen, Germany.

Abstract

Eupnoeic breathing in mammals is dependent on the co-ordinated activity of cranial and spinal motor outputs to both ventilate the lungs and adjust respiratory airflow, which they do by regulating upper-airway resistance. We investigated the role of central glycinergic inhibition in the co-ordination of cranial and spinal respiratory motor outflows. We developed an arterially perfused neonatal rat preparation (postnatal age 0–4 days) to assess the effects of blocking glycine receptors with systemically administered strychnine (0.5–1 μM). We recorded respiratory neurones located within the ventrolateral medulla, inspiratory phrenic nerve activity (PNA) and recurrent laryngeal nerve activity (RLNA), as well as dynamic changes in laryngeal resistance. Central recordings of postinspiratory neurones revealed an earlier onset in firing relative to the onset of inspiratory PNA after exposure to strychnine (260 ± 38.9 vs. 129 ± 26.8 ms). After glycine receptor blockade, postinspiratory neurones discharged during the inspiratory phase. Strychnine also evoked a decrease in PNA frequency (from 38.6 ± 4.7 to 30.7 ± 2.8 bursts min−1), but amplitude was unaffected. In control conditions, RLNA comprised inspiratory and postinspiratory discharges; the amplitude of the latter exceeded that of the former. However, after administration of strychnine, the amplitude of inspiratory-related discharge increased (+65.2 ± 15.2%) and exceeded postinspiratory activity. Functionally this change in RLNA caused a paradoxical, inspiratory-related glottal constriction during PNA. We conclude that during the first days of life in the rat, glycine receptors are essential for the formation of the eupnoeic-like breathing pattern as defined by the co-ordinated activity of cranial and spinal motor inspiratory and postinspiratory activities.

The respiratory cycle of mammals can be divided into three phases: inspiration, postinspiration (or stage I expiration) and expiration (or stage II expiration). The respiratory pattern and rhythm are generated by neurones distributed in distinct nuclei that form a network within the pontomedullary brainstem (see Bianchi et al. 1995; Richter, 1996; Richter & Spyer, 2001). This central network activity drives two functionally and anatomically distinct populations of motoneurones. The first is located in the cervical thoracic and lumbar spinal cord (Monteau & Hilaire, 1991; Iscoe, 1998) and controls the contraction of thoracic and abdominal muscles (e.g. the diaphragm and intercostal muscles) to ventilate the lungs. The second group of motoneurones is located in a specific subdivision of the nucleus ambiguus (Bieger & Hopkins, 1987), which is imbedded into the ventral respiratory group of the medulla oblongata. These motoneurones innervate laryngeal muscles and the bronchial tree to adjust respiratory airflow by regulating laryngeal resistance (Harding, 1984; Bartlett, 1986). In particular, the glottis is the valve of the ventilatory system that regulates precisely airflow on a breath-by-breath basis. During inspiration, the vocal fold is dilated (abducted) to decrease airway resistance and constricted (adducted) during the early stage of expiration or postinspiration. This postinspiratory glottic constriction serves multiple functions: it slows expiratory airflow out of the lungs to increase time for efficient gas exchange and maintains functional residual capacity to prevent lung collapse (see Harding, 1984; Bartlett, 1986). The former is most significant in neonatal mammals that have high breathing frequencies and highly compliant lungs and thoracic cage (Harding, 1984).

The co-ordinated activity of both cranial and spinal motor pools is a useful criterion to define the eupnoeic-like breathing pattern (Dutschmann et al. 2000). Although it is largely accepted that the pre-motoneurones of the pontomedullary respiratory network drive both pools of motoneurones, the mechanisms underlying their co-ordination are rarely investigated. An essential paradigm for respiratory pattern and rhythm generation is the function of reciprocal glycinergic inhibition between distinct subsets of respiratory neurones (e.g. Bianchi et al. 1995; Richter, 1996; Ogilvie et al. 1992; Ryback et al. 1997; Smith et al. 2000; Richter & Spyer, 2001). However, over the last decade research concerned with the role of glycinergic inhibition has led to some confusion. In in vitro neonatal preparations, it appeared that glycinergic inhibition was not required for the generation of a respiratory rhythm (Onimaru et al. 1989; Paton & Richter, 1995; Ramirez et al. 1996, 1997; Shao & Feldman, 1997; Brockhaus & Ballanyi, 1998; Rekling & Feldman, 1998; Ballanyi et al. 1999). In contrast, blockade of glycine receptors in mature mammals had severe consequences for both respiratory rhythm generation (Pierefiche et al. 1998) and pattern formation (Hayashi & Lipski, 1992; Büsselberg et al. 2001; Dutschmann & Paton, 2002). This discrepancy led to the hypothesis that glycinergic inhibition became more important during ontogeny (Duffin et al. 1995; Paton et al. 1994; Paton & Richter, 1995; Smith et al. 2000; Richter & Spyer, 2001). In the present study we challenged that question by assessing the functional relevance of glycinergic inhibition in the respiratory network of the newborn rat studied in situ. Our in situ approach permitted cellular, motor nerve and kinesiological analyses in the same preparation, pertaining to the finding that glycine receptors are essential for the respiratory pattern and the co-ordination of cranial and spinal motor outputs.

Methods

Using existing technology based on mature rodents (see Paton, 1996), we developed an in situ, arterially perfused preparation of rats at 0–4 days old (Dutschmann et al. 2000). In this preparation the brainstem is well oxygenated throughout, has a normal pH (Wilson et al. 2001) and generates a eupnoeic pattern of respiratory motor activity (see Dutschmann et al. 2000). All procedures were approved by the University of Bristol Ethics Commitee and the Home Office.

Working heart-brainstem preparation

Rats were anaesthetised deeply in a saturated atmosphere of halothane. In all cases, breathing was depressed markedly and they failed to respond to a noxious pinch applied to the tail or a hind paw. At this time all animals were decerebrated at the precollicular level with all tissue rostral to this level removed by aspiration. Anaesthesia was then discontinued as the animals had been rendered insentient (see Paton, 1996; Paton et al. 1999; Dutschmann et al. 2000). Following decerebration rats were transected below the diaphragm and perfused via a cannula inserted into the descending aorta at a flow rate of 18–25 ml min−1 at 31 °C. The preparations were perfused with a carbogen-gassed Ringer solution (mm: NaCl 125, KCl 5, KH2PO4 1.25, CaCl2 2.5, MgSO4 1.25, NaHCO3 25 and d-glucose 10) containing 1.25 % Ficoll (Sigma) to maintain colloid-osmotic pressure. Rhythmic contractions of respiratory muscles returned within minutes of the onset of re-perfusion. Some preparations were administered vecuronium bromide (3 μg ml−1, added to the perfusate) to block neuromuscular transmission.

Neural recordings

Respiratory motor nerve activity was recorded from the phrenic (PN) and recurrent laryngeal nerves (RLN) simultaneously via suction electrodes connected to headstage pre-amplifiers (Neurolog 100). Activity was amplified and filtered (8 Hz to 3 kHz; Neurolog modules 104 and 125). Intracellular recordings were performed with sharp microelectrodes (80-100 MΩ) filled with potassium methyl sulphate (3 m) and the Ca2+ chelator BAPTA (5 mm; Sigma), which assisted neurone recovery after impalement. Many respiratory neurones that were impaled were rejected because they failed to fulfil the following criteria for neuronal viability. All neurones in this study showed recovery from electrode impalement, a stable membrane potential and action potential overshoot above 0 mV at the start of the recording. Neurones were recorded intracellularly for up to 30 min. Extracellular microelectrodes were also used and filled with 3 m NaCl (5-10 MΩ). Microelectrodes were positioned between 0.5-1 mm rostral to the calamus scriptorius and 1.2-1.5 mm lateral to the midline on the dorsal medullary surface, and were driven in 2 μm steps into the ventral respiratory group using a piezoelectric stepper motor equipped with positional feedback (Burleigh Inchworm). Respiratory neurones were recorded at depths of between 1.2 and 1.8 mm. All data were digitised (1401plus CED) and stored on a computer. Respiratory motor nerve activity was integrated off-line (100 ms time constant). Data analysis was performed off-line using CED Spike 2 software (Cambridge Electronic Design).

Laryngeal resistance measurements

Changes in resistance through the larynx were made by recording sub-glottal pressure (SGP) during constant perfusion with warmed and humidified carbogen gas in the expiratory direction (see Paton et al. 1999, for details). The posterior pharyngeal wall just above the larynx was opened to allow the gas to escape. SGP was recorded via a side-arm tube connected to the gas inlet. SGP measurements were made in non-neuromuscularly blocked preparations of rats that were 2–4 days old. Increases and decreases in SGP were indicative of glottic constriction and dilatation, respectively, thereby giving an index of the dynamic changes in upper-airway resistance during the respiratory cycle.

Protocol and analysis

Glycine receptors were blocked by addition of strychnine (0.5-1 μM; Sigma) to the perfusate. The highest concentration administered was slightly higher than that described to be specific in adult animals (Jonas et al. 1998), but was used since the neonatal glycine receptor isoform has a lower sensitivity to strychnine compared to the adult (Hoch et al. 1989).

Statistical tests of peripheral nerve, SGP and neuronal recordings were always performed by comparing the control discharge with the activity 5 min after application of effective doses of strychnine. Tests on centrally recorded postinspiratory neurones were performed by comparing the onset of spiking relative to the onset of PN discharge of 15 respiratory cycles both before and after the application of strychnine. The analysis of membrane potentials (peak amplitudes of EPSPs or IPSPs) of intracellularly recorded postinspiratory neurones occurred for the same number of cycles and at the same times. Furthermore, we compared frequency and burst duration of PN activity (PNA) by analysing the integral of integrated PNA over 1 min before and after the addition of strychnine. The effect of strychnine on inspiratory and postinspiratory discharges recorded from the RLN were analysed over 30 respiratory cycles during a control period and after drug administration. Measures were made of integral areas of integrated discharge using a custom-written script. In addition, the phase duration of inspiration, postinspiration and expiration was analysed from the RLN. The effects of strychnine on SGP were analysed by comparing the amount of integrated PNA that occurred during the onset-to-peak of the glottal constriction (30 cycles before and after drug administration). To test for a significant difference in the discharge duration between postinspiratory neurones we analysed five periods of discharge for each cell and compared the mean values of individual cells. Absolute values or per cent changes are expressed as the mean ±s.e.m.

Statistical evaluation of experimental data was performed with Systat 8 software (SPSS). To test for significant effects we used a repeated-measure two-way ANOVA with interaction followed by a Fisher least-squares difference post hoc test with Bonferoni adaptation for the effect of strychnine on RLN discharge (mean values for individual preparations). The effect of strychnine on the integral of integrated PNA discharge during glottal constriction, the cellular data and PNA were analysed by using a two-way ANOVA followed by a Tukey's honestly significantly different post hoc test. If not mentioned, different ANOVA tests were performed with equal sample sizes. The dependence of respiratory phase duration on respiratory frequency was analysed with linear correlation. A 95 % confidence limit was taken as being significant.

Results

Influence of strychnine on the discharge pattern of different types of postinspiratory neurones

In the present study a total of 34 postinspiratory neurones (extracellular, n= 15; intracellular, n= 19) were recorded in the ventral respiratory group of neonatal rats. These neurones exhibited a peak firing frequency at the onset of expiration and ceased spiking during inspiration, at which time they became hyperpolarised (as observed from intracellular recordings). However, there was heterogeneity in their discharge pattern, leading to the distinction of two subtypes. The most frequent type showed decrementing spiking activity throughout the expiratory interval (type 1; n= 26, Fig. 1A). A second, smaller group, exhibited a short-duration discharge at the onset of expiration only (type 2; n= 8; Fig. 1B). Thus, we separated these two groups of postinspiratory neurones based on the duration of discharge (Fig. 1C): type-1 neurones discharged for 1133 ± 89 ms, which was significantly longer than type-2 neurones (i.e. 67.5 ± 6.5 ms; P < 0.001; Tukey, unequal sample size).

Figure 1.

Two types of postinspiratory neurones found in arterially perfused neonatal rats

An illustration of the heterogeneity of postinspiratory neurones recorded intracellularly in the ventral respiratory group of neonatal rats. A, type-1 postinspiratory neurones were characterised by an inspiratory-related hyperpolarisation with rebound depolarisation and discharge (arrows) that slowly declined in frequency during the expiratory interval. B, type-2 postinspiratory neurones also exhibited inspiratory-related hyperpolarisation, but this was followed by a brief, high-frequency volley of action potentials that was restricted to early expiration. C, histogram of the discharge duration of all recorded postinspiratory neurones illustrating two separate modes. Please note that the twisted curves of the Gaussian distribution are due to logarithmic scaling of the x-axis. n, number of neurones; P, postnatal age; PNA, phrenic nerve activity; WHBP, working heart-brainstem preparation.

Postinspiratory neurones of both types were exposed to strychnine (n= 14). Six out of 9 type-1 and all type-2 (n= 5) neurones recorded either extra- or intracellularly clearly lost their inspiratory inhibition. In control conditions, firing started 260 ± 38.9 ms after the onset of the inspiratory burst of PNA, but this was advanced significantly after exposure to strychnine (i.e. 129 ± 26.8 ms; Fig. 2A and B; n= 11; Tukey test: P < 0.05). Since PNA burst duration was unchanged following strychnine application (control: 260 ± 38.9 ms vs. 286 ± 72 ms after strychnine; n= 11), postinspiratory neurones started to discharge during the inspiratory phase of PNA, an effect that was never seen in control conditions (see Fig. 2A and B). Furthermore, the inspiratory-related hyperpolarisation of postinspiratory neurones (-5.3 ± 1.1 mV; relative to the level during late expiration) was of shorter duration and converted into a depolarisation (+4.8 ± 1.6 mV, n= 6, Tukey test: P < 0.05, Fig. 2A and B) during PNA. In addition, three of the type-1 postinspiratory neurones recorded extracellularly lost their characteristic discharge pattern and displayed tonic discharge over the entire respiratory cycle after strychnine. Although the hyperpolarisation observed during inspiratory PNA was obviously advanced, a pre-inspiratory inhibition persisted in type-2 neurones (n= 4) and was revealed in type-1 neurones (n= 3) after strychnine application (see arrows in Fig. 2A and B).

Figure 2.

Strychnine application induces firing of postinspiratory neurones during inspiration

Two types of postinspiratory neurones (type-1 (A) and type-2 (B)) found in arterially perfused neonatal rats. A and B, the inspiratory-related hyperpolarisation of a type-1 and type-2 postinspiratory neurone was reduced and converted into depolarisation after application of strychnine (A, 0.5 μM; B, 1 μM). Consequently these neurones fired during the inspiratory phase of PNA. C, example of a type-2 postinspiratory neurone recorded extracellularly that lost its inspiratory modulation and exerted tonic discharge throughout the respiratory cycle after application of strychnine (1 μM). Note that preinspiratory inhibition was revealed in type-1 neurones (n= 3), which tended to persist in type-2 neurones (n= 4, see arrows in A and B). P, postnatal age.

Effect of glycine-receptor blockade on the discharge pattern of spinal and cranial motor outputs

Recurrent laryngeal nerve activity.

RLNA revealed three phases of respiration (n= 11) and was evident from the day of birth (n= 3; see Fig. 3A). The RLN exhibited a burst of central inspiratory activity coincident with PNA. The onset of inspiratory discharge in the phrenic neurogram was delayed relative to that in the RLN by 41.5 ± 6.3 ms (n= 11; t test; P < 0.001). Inspiratory discharge was followed by a decrementing postinspiratory discharge (stage I of expiration), which was larger in amplitude than the inspiratory discharge (Fig. 3A). Finally, the postinspiratory discharge was followed by quiescence indicative of stage II expiration (Fig. 3A). The duration of the postinspiratory discharge was dependent on the baseline frequency of the respiratory cycle, as revealed by linear correlation analysis (see Fig. 4, r= 0.87, degrees of freedom 13, P < 0.001). In preparations with a relatively fast rhythm (1.1-0.8 Hz), postinspiratory activity was short in duration (see Fig. 3A). In contrast, the discharge was more prolonged in preparations with a slower rhythm (0.4-0.5 Hz). There was no correlation between age and frequency in the preparations analysed (r < 0.5). In control conditions, both the peak and integrated postinspiratory discharge exceeded the neural inspiratory activity of the RLN (Fig. 3A and Fig. 5A; n= 9, Fisher test: P < 0.05). After blockade of glycine receptors, the total integrated neural inspiratory activity in the RLN increased significantly by 65.2 ± 15.2 % (Fig. 3B and Fig. 5A; n= 9; Fisher test: P < 0.05) and now exceeded the postinspiratory discharge (Fig. 5A). This was due to an increase in the amplitude of inspiratory discharge, but not its duration (Fig. 3 and Fig. 5B). This change in RLNA pattern was highly significant (Fisher test, interaction: P= 0.002, see Fig. 5A). In the presence of strychnine, a clear decrease in postinspiratory discharge was seen in seven preparations (see Fig. 3B; −28.7 ± 8.2 %, Fisher test: P < 0.05); two other preparations showed a slight increase in postinspiratory activity compared to the control. Furthermore, strychnine reduced the duration of the postinspiratory phase in the RLN (Fig. 5B, −46.1 ± 5.7 %, P < 0.01, Fisher test: n= 9), while the duration of the expiratory phase increased significantly (Fig. 5B, +132.4 ± 29.7 %; Fisher test: P < 0.01).

Figure 3.

Blockade of glycine receptors depresses postinspiratory activity but augments inspiratory discharge in the recurrent laryngeal nerve

A, in the recurrent laryngeal nerve from a neonatal rat that was a few hours old there were both inspiratory-related (i.e. coincident with PNA) and postinspiratory discharges. Note that postinspiratory activity is much greater in amplitude than inspiratory motor discharge in control conditions. B, the influence of systemic application of strychnine (0.5 μM) on PNA and recurrent laryngeal nerve activity (RLNA) is shown. Note the dramatic increase in the amplitude of the inspiratory-related discharge in the RLNA after glycine receptor antagonism. This was accompanied by a reduction in postinspiratory activity. *Expanded view of one cycle illustrating the respiratory phases (I, inspiration; PI, postinspiration; E, expiration) before and after application of strychnine. Note the almost complete absence of postinspiratory activity after glycine receptor blockade. P, postnatal age.

Figure 4.

Duration of respiratory phases in RLNA is correlated with the frequency of phrenic nerve bursts

Diagram illustrating the linear correlation of the duration of the individual respiratory phases (▵ inspiration, ○ postinspiration and ▪ expiration) versus the respiratory cycle length calculated from the interval between two inspiratory bursts of PNA under control conditions. Note that the prolongation of each respiratory phase with increasing cycle length was correlated (postinspiration r= 0.870, degrees of freedom, d.f. 13, P < 0.001; expiration r= 0.856, d.f. 13, P < 0.001; inspiration r= 0.532, d.f., 13, P < 0.05). CL, respiratory cycle length.

Figure 5.

Respiratory frequency and pattern changes induced by glycine receptor blockade in neonatal rats

A, bar graph illustrating the effect of strychnine on the integrals of integrated inspiratory and postinspiratory activities recorded from the recurrent laryngeal nerve (n= 9; contr., control; stry., strychnine). B, bar graph illustrating the influence of 0.5-1 μM strychnine on the duration of the three respiratory phases (inspiration, postinspiration and expiration) as measured from recordings of RNLA. Please note that the brackets over integrated inspiratory and postinspiratory activity reflect a significant difference in the discharge pattern (from PI exceeding I to I exceeding PI discharge) revealed by ANOVA with interaction. *P < 0.05, **P < 0.01. CL, cycle length; E, expiration; I, inspiration; PI, post-inspiration.

Phrenic nerve activity.

Blockade of glycine receptors reduced the frequency of PNA from 38.6 ± 4.7 to 30.7 ± 2.8 bursts min−1 (for example, see Fig. 3; n= 11, Tukey test: P < 0.01). There was no change in PNA duration (286 ± 32 vs. 319 ± 76 ms; n= 11; Tukey test: P= 0.32). Furthermore, in 9 out of 11 preparations PNA was also interrupted by periods of fast (range 2–3 Hz) and short (range 100–160 ms) bursting (data not shown). These periods of fast oscillating PNA lasted for 2–4 s and within the 5 min after application of 0.5-1 μM strychnine occurred 3.66 ± 0.7 times. Due to the short duration of PNA and variable discharge pattern (ramping, decrementing or bell shaped) during control conditions (see PNA Figs 1-6), a consistent change of the pattern of PNA following glycine receptor blockade could not be detected.

Figure 6.

Strychnine shifts glottic constriction into inspiration

A, the dynamic changes in glottal resistance during the respiratory cycle included dilatation during inspiration (decrease in sub-glottal pressure or SGP) and constriction in postinspiration (shaded area; increase in SGP) in a 3-day-old rat (P3). B, after strychnine blockade of glycine receptors glottal constriction occurred earlier during neural inspiration. C, bar diagram comparing the integral of integrated PNA that occurred from the onset to the peak rise in SGP (shaded areas in A and B), which was analysed in control conditions and after exposure to strychnine (n= 8). **P < 0.01.

Effects of glycine receptor blockade on upper-airway resistance

To relate the changes in postinspiratory activity recorded from both postinspiratory neurones and the RLN to function, we performed a kinesiological study on upper-airway resistance. We measured the respiratory modulation of glottal resistance by monitoring changes in SGP during constant gas flow perfusion of the upper airway in the expiratory direction (see Methods). During inspiration, SGP decreased corresponding to the contraction of abductor muscles (i.e. dilatation of the vocal fold), whereas constriction occurred immediately after neural inspiration in the postinspiratory phase, as seen by a sharp rise in SGP (adductor contraction; Fig. 6A). Late expiration (stage II) was characterised by lower levels of SGP compared to those seen during postinspiration (Fig. 6A). Thus, as with the RLN, the inspiratory and postinspiratory respiratory phases were preserved faithfully in the SGP trace of immature rats. Application of strychnine (n= 8) evoked a profound change in the respiratory modulation of the glottis. The activation of glottal adductors, which normally occurred during postinspiration (shaded area of Fig. 6A), now occurred simultaneously with the onset of neural inspiration (shaded area of Fig. 6B). Hence, a paradoxical inspiratory glottal constriction replaced the normal dilatation. This was shown to be highly significant since the proportion of integrated PNA that occurred from the onset to the peak in glottal constriction (shaded area of Fig. 6A and B) increased by 373 ± 128.8 % after exposure to strychnine (Fig. 6C n= 8, Tukey test: P < 0.01). Finally, strychnine did not change the absolute level of SGP, indicating no overall change in the resting tone of the upper-airway muscles.

Effect of high concentrations of strychnine on respiratory motor function

Administration of higher concentrations of strychnine (i.e. 2–5 μM) led to a severe disturbance of respiratory activities, which was characterised by increases in PNA burst frequency from a control value of 32.4 ± 3.23 to 48.8 ± 4.02 bursts min−1 (Fig. 7; n= 5; Tukey test: P < 0.05). This was accompanied by a huge variability in PNA burst-duration including tonic PNA discharge. Overall, this led to an increase of 68 ± 14 % in PNA burst duration following administration of strychnine. High concentrations of strychnine evoked periods of high-frequency PNA bursting superimposed on tonic activity (see Fig. 7B, upper panel). During these periods a persistent rise in SGP was observed with superimposed constrictions in phase with PNA bursting (Fig. 7B, lower panel). This resulted in an unpredictable pattern of upper-airway modulation.

Figure 7.

Effect of a high dose of strychnine (> 2 μM) on PNA and upper-airway resistance

A, the eupnoeic modulation of laryngeal resistance as revealed by changes in SGP (see Fig. 6). B, application of strychnine increased the frequency and duration of inspiratory PNA. These periods of fast and tonic PNA were accompanied by periods of persistent increases in SGP on which was superimposed further laryngeal constrictions that were coincident with PNA bursting. The latter indicates paradoxical glottic constriction during neural inspiration. P, postnatal age.

Discussion

This is the first demonstration that glycinergic inhibition within the brainstem of neonatal rats is essential for the temporal co-ordination of cranial and spinal motor outputs. This temporal co-ordination is likely to be vital for optimal pulmonary ventilation and control of airflow through the larynx. Our data purport the idea that central glycinergic inhibition is essential for the separation and co-ordination of neuronal inspiratory and postinspiratory activities from the onset of postnatal life. Blockade of glycine receptors prevented the inspiratory inhibition of postinspiratory neurones, causing them to fire during the inspiratory phase. We demonstrate that strychnine causes a simultaneous activation of laryngeal abductor and adductors, leading to a paradoxical inspiratory laryngeal adduction (constriction). Laryngeal adduction during inspiration is predicted to severely impede inhalation and cause a major disruption to pulmonary ventilation.

Respiratory activity in the neonatal working heart- brainstem preparation (WHBP)

It is widely accepted that the eupnoeic breathing pattern of mammals comprises three phases - inspiration, postinspiration and expiration (Bianchi et al. 1995; Richter, 1996; Richter & Spyer, 2001). In the present study we defined a eupnoeic-like breathing pattern by recording inspiratory PNA as a reference for the inspiratory phase. In addition, we recorded activity from the RLN or measured the respiratory modulation of laryngeal resistance to discriminate between postinspiration (early expiratory phase) and late expiration. The latter recordings were prerequisite for the definition of a eupnoeic-like breathing pattern (three-phase) in the neonatal WHBP. This was necessary as postinspiratory discharge appears to be weak or absent in the phrenic neurogram of rodents in situ or in vivo (e.g. mouse: Paton, 1996; rat: Schwarzacher et al. 1991; Hayashi et al. 1996; Paton et al. 1999). Furthermore, as reported previously (see Dutschmann et al. 2000), the discharge pattern of inspiratory PNA exhibits variability and is of relatively short duration in neonatal rats, and is therefore not a reliable marker for defining eupnoeic-like respiratory activity. In contrast, the PNA recorded from a WHBP of juvenile and mature rats is of longer duration, exhibiting an incrementing pattern, and serves as an index for the eupnoeic-like respiratory pattern (Paton et al. 1999; St John & Paton, 2000).

Blocking glycine receptors: effect on postinspiratory neurones

Postinspiratory neurones are characterised by a pronounced inspiratory-related inhibition that is mediated by glycine acting on glycine receptors (Haji et al. 1990; Schmid et al. 1991). In the present study we recorded two different types of postinspiratory neurones in the ventral respiratory group of neonatal rats, in accordance with what has been reported previously in adult rat in vivo (Hayashi et al. 1996). The two types were discriminated by a clearly different duration of discharge in the postinspiratory phase: type-1 neurones discharged at least during the first half of the expiratory interval, while type-2 neurones only fired transiently (< 100 ms) in early postinspiration. However, both types showed the same pattern of response in membrane potential during the respiratory cycle in control conditions (Fig. 2). Although we did not functionally or histologically characterise individual postinspiratory neurones as either motor or propriobulbar, the firing pattern of type-2 postinspiratory neurones is similar to that of laryngeal motoneurones (see Barillot, 1990), while that of type-1 postinspiratory neurones reflects the discharge pattern of propriobulbar neurones (Schwarzacher et al. 1995). It should also be recognised that 10 out of 11 postinspiratory neurones recorded in a previous study in the rat were found to be cranial motoneurones (Zheng et al. 1991). Both type-1 and −2 postinspiratory neurones revealed in the present study responded similarly to strychnine by an earlier onset of firing during the inspiratory phase. However, the response of type-2 neurones to strychnine was more consistent than type-1 neurones, which may suggest that laryngeal motoneurones are particularly affected by glycine receptor blockade. Indeed, our peripheral nerve recordings of laryngeal motor outflow strengthen this idea (see below).

Our findings that following strychnine application, postinspiratory neurones lose their inspiratory-related inhibition and start to discharge during the inspiratory phase in neonatal rats is consistent with reports in adult mammals (Büsselberg et al. 2001; Dutschmann & Paton, 2002). The underlying mechanism(s) of this paradoxical inspiratory depolarisation of postinspiratory neurones is unclear. Blockade of glycine receptors either reveals an excitatory synaptic input to postinspiratory neurones, which is normally shunted by glycinergic inhibition in eupnoea, or induces endogenous bursting, as was shown to occur in some types of inspiratory neurones in vitro (Smith et al. 1991; Johnson et al. 1994; Gray et al. 1999; Koshiya & Smith, 1999; Lieske et al. 2000).

Strychnine-mediated effects on postinspiratory neurones: consequences for motor control of the larynx

The postinspiratory activity of laryngeal nerves has a pivotal function in regulating airflow through the larynx. With the central effects of strychnine causing inspiratory depolarisation of postinspiratory neurones, we investigated putative consequences on laryngeal function. Innervation of the larynx is provided by mixed motor nerves (e.g. recurrent laryngeal and superior laryngeal nerves) containing fibres that innervate either the abductor (thyreoarytenoideus, circoarytenoideus and interarytenoideus) or adductor muscles (cricoarytenoideus). Accordingly, two classes of laryngeal motoneurones can be found within the respiratory network (Bryant et al. 1993): inspiratory and postinspiratory laryngeal motoneurones (Barillot et al. 1990). In recordings of the RLN we observed a dramatic change in their discharge pattern in response to strychnine, including a massive increase of inspiratory-related discharge (abductor drive), while postinspiratory discharge (adductor drive) was decreased. This suggests that postinspiratory motor discharge occurred during the inspiratory phase following the blockade of glycine receptors. Indeed, direct measurement of the dynamic changes of laryngeal resistance over the respiratory cycle showed that the normal inspiratory-related laryngeal abduction (dilatation) was replaced by a paradoxical, and pathological, laryngeal adduction (constriction) during inspiration. This would severely compromise attempts at inhalation and therefore disrupt the eupnoeic breathing pattern (for review, see Harding, 1984; Bartlett, 1986). Moreover, since postinspiratory activity is essential for the eupnoeic control of glottal musculature, it can be stated that it is crucial for basic behaviours such as vocalisation and swallowing (Grélot et al. 1992; Sakamoto et al. 1996; Shiba et al. 1999; Gestreau et al. 2000). Moreover, failure to properly co-ordinate inhalation and swallowing would result in fatal aspiration and drowning (Dutschmann & Paton, 2002).

Effects of blocking glycine receptors on the respiratory rhythm

Interestingly, low doses of strychnine in neonatal rats only exerted a mild effect on respiratory rhythm that was characterised by a decrease in the frequency of PNA, with the pattern and duration remaining unchanged. Application of higher doses of strychnine produced major disturbances to the rhythm, which may relate to the non-specific effects of strychnine, such as convulsant discharges, within the respiratory network. This result was surprising to us in two ways. First, since postinspiratory neurones play a key role in the irreversible inspiratory off-switch mechanism (St John & Zhou, 1989, 1990, 1991; Bianchi et al. 1995; Hayashi et al. 1996; Richter, 1996; Rybak et al. 1997; St John, 1998), it would have been expected that the earlier onset of postinspiratory activity would lead to a shortening of PNA burst duration. However, this was observed only periodically in some experiments (see Results). Second, in mature brainstems, reciprocal synaptic inhibition (GABAergic, glycinergic) between respiratory neurones is the core element in many models for describing the formation of the respiratory rhythm (Bianchi et al. 1995; Richter, 1996; Rybak et al. 1997). Thus, a greater disturbance to the respiratory rhythm following application of strychnine would have been anticipated. With the absence of an effect, interpretation of these findings in terms of a role for glycinergic inhibition in respiratory rhythm generation is difficult. The decrease in PNA frequency resulting from strychnine could be due to a change in the operational state of the in situ respiratory network from eupnoeic-like to gasp-like activity. This is supported by (1) a comparable frequency of gasping and eupnoeic inspiratory bursting in neonatal rodents (St John 1998) and (2) a complete absence of effect of strychnine on gasping (St John & Paton, 2002). Whatever the nature of the strychnine-insensitive rhythm might be, an important consideration is that although a rhythm persists, our data raise doubts as to whether this would allow adequate pulmonary ventilation. Increased glottal resistance during inhalation is not consistent with optimal lung inflation.

The role of glycinergic inhibition during postnatal maturation of the respiratory network

Currently, the most widely accepted model for respiratory rhythm formation suggests that maturation of the respiratory network parallels an increasing importance for glycinergic inhibition (Paton et al. 1994; Duffin et al. 1995; Rekling & Feldman, 1998; Smith et al. 2000; Richter & Spyer, 2001). This was based on the assumption that in neonates the respiratory rhythm is generated by a glycine-receptor-independent conditional burster/ pacemaker activity located in the preBötzinger complex, while in mature mammals the rhythm is based on an oscillating network in which glycinergic and GABAergic synaptic inhibition are essential. In our study, the physiological significance of glycinergic inhibition for separating postinspiratory and inspiratory phases from the day of birth was revealed. Since glycinergic inhibition appears to serve the same function in mature mammals (Büsselberg et al. 2001; Dutschmann & Paton, 2002), there is no maturational change in the importance of glycinergic inhibition for the formation of the respiratory pattern. In contrast, the question as to whether glycinergic inhibition is important for respiratory rhythm generation in mammals remains unanswered.

Clinical implications of the study

Disturbance of upper-airway control is a common clinical problem in human infants that leads to upper-airway obstruction (Downey et al. 1993; Gislason et al. 1995). Infants with upper-airway resistance syndrome are at high risk for cot death (Guilleminault et al. 1993), which may relate to our findings. Interestingly, glycine receptors undergo profound changes during early postnatal development: the neonatal isoform is replaced by an adult isoform between 17 and 22 days of life in the rat (Kling et al. 1997). This may be comparable to 3–4 months of age in human infants and coincides with the high-risk time for cot death. A developmental disturbance of glycine receptors may relate to the finding of cot-death victims who died due to upper-airway obstruction but without apparent respiratory arrest or re-breathing (Poets et al. 1999). Whether cot-death victims show mutant forms of glycine receptors in the ventrolateral medulla is unknown.

Conclusions

Glycinergic inhibition within the pontomedullary respiratory network separates the inspiratory and postinspiratory phases in the breathing cycle of neonatal mammals. We conclude that glycine receptors are essential for the formation of the eupnoeic-like breathing pattern as defined by the co-ordinated activity of cranial and spinal respiratory motor outflows required for appropriate ventilation of the lungs in the neonatal rat. Furthermore, since the effect of strychnine is comparable in adult and neonatal rats, we propose that glycinergic inhibition within the respiratory network is functional from the day of birth.

Acknowledgements

We are most appreciative of the financial support from the British Heart Foundation (BS/93003), the Deutsche Forschungsgemeinschaft (Du 338-1/1) and the SFB 430.

Ancillary