Hypoxic response of hypoglossal motoneurones in the in vivo cat



  • 1In current and voltage clamp, the effects of hypoxia were studied on resting and synaptic properties of hypoglossal motoneurones in barbiturate-anaesthetized adult cats.
  • 2Twenty-nine hypoglossal motoneurones with a mean membrane potential of −55 mV responded rapidly to acute hypoxia with a persistent membrane depolarization of about +17 mV. This depolarization correlated with the development of a persistent inward current of 0.3 nA at holding potentials close to resting membrane potential.
  • 3Superior laryngeal nerve (SLN) stimulation-evoked EPSUPs were reduced in amplitude by, on average, 46%, while IPSUP amplitude was reduced by 31 %. SLN stimulation-evoked EPSCs were reduced by 50–70%.
  • 4Extracellular application of adenosine (10 mm) hyperpolarized hypoglossal motoneurones by, on average, 5.6 mV, from a control value of –62 mV. SLN stimulation-evoked EPSUPs decreased by 18% and IPSUPs decreased by 46% during adenosine application.
  • 5Extracellular application of the KATP channel blocker glibenclamide led to a blockade of a persistent outward current and a significant increase of SLN stimulation-evoked EPSCs.
  • 6We conclude that hypoglossal motoneurones have a very low tolerance to hypoxia. They appear to be under metabolic stress even in normoxia and their capacity to activate protective potassium currents is limited when compared with other brainstem neurones. This may help to explain the rapid disturbance of hypoglossal function during energy depletion.

The mammalian CNS shows regional and developmental differences in the sensitivity to hypoxia (Fujiwara, Higashi, Shimoji & Yoshimura, 1987; Crepel, Krnjevic & Ben-Ari, 1992; Doutheil, Ballanyi & Richter, 1992; Ballanyi, Kuwana, Völker, Morawietz & Richter, 1992; Donnelly, Jiang & Haddad, 1992; Luhmann, Kral & Heinemann, 1993; Haddad & Jiang, 1993a; O'Reilly, Jiang & Haddad, 1995; Richter & Ballanyi, 1996). Cortical and hippocampal neurones, for example, respond with an initial membrane hyperpolarization to hypoxia, but then depolarize when hypoxia lasts longer than 5 min (Krnjevic & Leblond, 1989; Leblond & Krnjevic, 1989). Respiratory neurones within the lower brainstem behave similarly, although the initial membrane hyperpolarization appears less pronounced (Richter, Bischoff, Anders, Bellingham & Windhorst, 1991).

In contrast to such biphasic hypoxic reactions, dorsal vagal and hypoglossal motoneurones show only a monophasic hypoxic response. Vagal neurones respond with a persistent membrane hyperpolarization (Doutheil et al. 1992; Trapp & Ballanyi, 1995; Ballanyi, Doutheil & Brockhaus, 1996), while hypoglossal neurones react with a rapid and pronounced membrane depolarization (Haddad & Donnelly, 1989, 1990; Donnelly et al. 1992; Doutheil et al. 1992; Haddad & Jiang, 1993b).

Much of the apparent differences in the hypoxic responses of neurones might be due to different activity states of these groups of neurone when they are analysed under in vivo or under in vitro conditions. From the analysis of in vivo supontaneously active respiratory neurones it is known that the hypoxic response is a combination of various complex reactions: there is an early blockade of excitatory (EPSCs) and inhibitory postsynaptic currents (IPSCs) and reversal of IPSC polarity, which in periodically inhibited neurones leads to the development of relative membrane depolarization. In addition there is also a simultaneous activation of persistent potassium conductances which counteract membrane depolarization (Richter et al. 1991). However, when studied in vitro, many respiratory neurones behave differently to those studied in vivo in that they show persistent membrane hyperpolarization and blockade of synaptic drive potentials in periods when the network is still rhythmically active (Ballanyi, Völker & Bichter, 1994; Volker, Ballanyi & Richter, 1995; Richter & Ballanyi, 1996). Hence, one might susupect that the monophasic hypoxic response of hypoglossal neurones reflects an in vitro phenomenon, which might differ from the response observed in vivo, when neurones receive stronger synaptic inputs (Hwang, Bartlett & St John, 1983; Withington-Wray, Mifflin & Supyer, 1988).

The present study was therefore designed to assess whether the effects of hypoxia on in vivo hypoglossal motoneurones are similar to those seen in medullary respiratory neurones. We supecifically determined the hypoxic responses of persistent currents and membrane potential as well as postsynaptic currents evoked by superior laryngeal nerve stimulation. The functional significance of such in vivo studies relates to possible hypoxic disturbances of phonation, mastication and deglutition, because hypoglossal motoneurones innervate intrinsic and extrinsic muscles of the tongue and maintain airway patency by adjusting motor activities to respiration (Miller, 1982; Harding, 1984; Grélot, Barillot & Bianchi, 1990). A particular problem of hypoxic disturbances of respiration-related synaptic control of hypoglossal motoneurones is considered to be obstructive apnoea (Withington-Wray et al. 1988; Säriås, Cormier, Desmeules & La Forge, 1989; Richard & Harper, 1991).



Experiments were performed on adult cats of either sex weighing 2.5–3.5 kg. Care and use of animals were in accordance with the guiding principles of the German and British Physiological Societies. Animals were anaesthetized with sodium pentobarbitone (Nembutal, Sanofi) at 40 mg kg−1i.p. as an initial dose. Supplementary doses of Nembutal were given i.v. (1.3–2.5 mg kg−1) whenever unprovoked changes in heart rate or blood pressure were observed or if phrenic nerve burst discharges increased in frequency. Additional doses of anaesthetic were also given if a mild nocioceptive stimulus applied to the paw elicited increases in central respiratory activity or arterial blood pressure. Animals were pre-medicated with atropine sulphate (B. Braun AG, Melsüngen, Germany; 0.1–0.2 mg kg−1) to block mucus secretion and dexamethasone (Fortecortin Mono, Merck, 0.2 mg kg−1, i.m.) to prevent brain oedema. At the end of the experiment the animals were killed with an anaesthetic overdose (i.v.).

Catheters were inserted into a femoral artery to monitor arterial blood pressure and both femoral veins for administration of drugs. If necessary, arterial blood pressure was maintained above 100 mmHg by infusing standard Ringer solution containing adrenaline (Suprarenin, Hoechst AG, 40 mg ml−1) and glucose (27 mg ml−1). Body temperature was maintained between 36 and 38 °C by external heating. The trachea was cannulated below the larynx for artificial ventilation with positive pressure using oxygen-enriched air (40–50 vol.% O2). Animals were paralysed with gallamine triethiodide (Flaxedil, Bhone-Poulenc Rorer; initial dose 10 mg kg −1i.v. followed by 5 mg kg −1 h−1). A bilateral pneumo thorax was made to avoid respiratory movements of the thorax and thus increase the stability of brain tissue for long-term intracellular recordings. Atelectasis of lungs was prevented by applying a positive pressure of 1–2 cmH2O to the expiratory line. End-tidal CO2 was monitored (DATEX normocap, Hoyer AG, Bremen, Germany) and maintained at 30–40 Torr by adjusting the ventilatory rate. Inspired oxygen content was monitored continuously (DATEX) as a percentage of the inspired gas mixture delivered to the animal. Inspiratory and end-expiratory pressures were monitored to identify and treat airway obstruction by mucus. The head of the animal was fixed in a stereotaxic frame. Both phrenic nerves were prepared from a dorsal approach, severed peripherally, desheathed and placed on bipolar hook electrodes. The hypoglossal and superior laryngeal nerves were isolated ipsilaterally to the side of medullary recordings and placed on bipolar stimulating electrodes. An occipital craniotomy exposed the dorsal surface of the brainstem. Dural and arachnoidal membranes were removed and the pial membrane was opened at the point of microelectrode insertion. A circular-shaped perforated ‘pressure foot’ was placed over the recording site and a second ‘pressure foot’ was placed close to the site of recording to further reduce arterial pulse pressure-related and ventilation-related movements of brain tissue. The activity of both phrenic nerves was amplified (x 10000), filtered (1–3 kHz) and displayed on an oscilloscope in both original and integrated forms (τ= 100ms). Phrenic nerve activity was taken as an index of central respiratory activity.

Hypoglossal motoneurones were localized at positions 0–1 mm caudal, 0.5–1.5 mm lateral to the midline, at a depth of 1.7–2.0 mm beneath the dorsal surface of the medulla. They were identified by their antidromic response to stimulation of the ipsilateral hypoglossal nerve (o.1 ms pulses, 1 Hz, 10–50 V) and tested for their response to electrical stimulation of superior laryngeal nerve (SLN, 1–3 Hz, 0.7 ms pulses, 5 V). Intracellular recordings were made with glass microelectrodes (tips broken back to 0.5–1.0 μm outer diameter) that were filled with 1.0–2.0 m K-CH3SO4 containing 5 mm BAPTA. These electrodes had resistances ranging from 20 to 60 MΩ as measured in brain tissue. Membrane potentials of motoneurones were measured in single electrode current clamp mode (SEC-05L amplifier; npi electronics GmbH) and displayed on an oscilloscope. Single electrode voltage clamp measurements were performed with the same amplifier at 20–30 kHz switching frequency and a 25% duty cycle (Bichter, Pierrefiche, Lalley & Polder, 1996). In order to ensure correct operation during single electrode voltage clamp measurements, we monitored the electrode potentials during current pulse injections on an oscilloscope. Holding potentials were set within the range of membrane potential as measured in current clamp. All signals were displayed on a chart recorder (Gould TA2000), and on a Macintosh computer using MacLab software, and stored on videotape for offline analysis.

Extracellular drug application

Solutions of adenosine (10 mm in Binger solution) and gliben clamide (5 mm in Binger solution) were applied to the dorsal medullary surface by perfusing the central opening (diameter 0.5 mm, height 1.5 mm) of the perforated ‘pressure foot’ through which microelectrodes were inserted into the medulk (Richter et al. 1996; Pierrefiche, Bischoff & Richter, 1996). The latency of diffusion of applied substances to hypoglossal neurones ranged between 4 and 10 min. Changes in single neuronal activity were not accompanied by measurable changes in respiratory or cardiovascular activity.

Protocols and data analysis

Neurones were selected for further investigation if they showed a stable membrane potential without inactivation of action potentials and if they were antidromically activated by hypoglossal nerve stimulation. Desupite these criteria several apparently stable current clamp recordings from hypoglossal motoneurones were disregarded when voltage clamp measurements revealed a substantial leak current. Neurones were tested during mild and/or severe acute hypoxia (12 and 6 vol.% O2, respectively), by mixing a gas mixture of 95% NO2 and 5%2 to room air. Oxygen was then re-supplied once a response had been fully characterized. Membrane potentials were calculated as the mean of voltage fluctuations within the time of a respiratory cycle taken before, during and after hypoxia tests. Ten to twenty single responses to SLN stimulation were averaged before, during and after each hypoxic text. The amplitudes of synaptically evoked events were measured with reference to the level of the membrane potential immediately preceding the stimulus. Values are given as means ± standard deviation for the pool of tested neurones. Student's paired t tests were performed between control and test groups and significant changes were assumed with P < 0.05.


The study is based on the analysis of twenty-nine hypoglossal motoneurones with a membrane potential of –55 ± 0.8mV (n= 29) as determined as the mean of voltage fluctuations during the respiratory cycle. Some hypoglossal motoneurones showed respiratory modulation of their activity similar to that described previously (Withington-Wray et al. 1988). Fifteen hypoglossal motoneurones displayed augmenting inspiratory depolarization with, or without, discharges in phase with phrenic nerve activity, and four other hypoglossal motoneurones showed a rapid onset of a burst discharge during postinspiration, while they were synaptically inhibited during inspiration (see Fig. 1). The remainder of hypoglossal motoneurones (n= 10) had a relatively constant membrane potential throughout respiratory phases.

Figure 1.

Illustration of the three different types of hypoglossal motoneurones recorded in the present study

A, inspiratory-modulated motoneurone responding to hypoglossal nerve stimulation (dots). B, a neurone exhibiting a postinspiratory pattern of discharge that responds to SLN stimulation (dots). C, a motoneurone which responded to SLN stimulation (dots) but without respiratory-related membrane potential changes. Vm, membrane potential; PN, phrenic nerve activity.

Hypoxic response of resting potential and persistent currents

Ventilation of animals with hypoxic gas mixtures elicited a typical hypoxic response of phrenic nerve activity. The response was biphasic consisting of an initial increase in frequency and amplitude of inspiratory bursts, followed by a decline of respiratory activity when hypoxia persisted for longer than 3–4 min. Twenty-six hypoglossal motoneurones responded very quickly to hypoxia with a membrane depolarization which started after a short latency of 48 ± 28 s. During this time arterial oxygen levels, as monitored from the inspired gas mixture, had just reached minimal levels (6–12%). Maximal membrane depolarization of 17 ± 0.25mV (n=11, P < 0.05) was evident after 66 ± 29 s (n= 13), when phrenic nerve activity was still enhanced. Membrane depolarization of hypoglossal motoneurones was accompanied by an initial increase in discharge frequency of action potentials. However, the amplitudes of action potentials, and after-hyperpolarization decreased and action potential inactivation occurred quickly.

Respiration-related synaptic drive potentials decreased at the same time (Fig. 2). The membrane potential returned to control levels after a latency of 100 ± 37 s in nine hypoglossal motoneurones, once normal oxygen supply was re-established (Fig. 2A). In three hypoglossal motoneurones, however, membrane potential stabilized at a slightly depolarized level. The remaining nine hypoglossal motoneurones remained strongly depolarized or even continued to depolarize further without sign of mechanical instability of recordings. Nevertheless, we disregarded these measurements.

Figure 2.

Effect of mild (A) and severe (B) hypoxia on the membrane potential of an inspiratory-related hypoglossal motoneurone

Recordings of membrane potential (Vm) and phrenic nerve activity (PN) were taken before hypoxia test (Control), at the maximum effect of hypoxia (middle panels) and during recovery (right panels). The neurone received continuously evoked inputs on stimulation of the ipsilateral SLN (artifacts observed throughout the respiratory cycle). Note that during severe hypoxia (B) depolarization brings the neurone to an excitable state so that every SLN stimulus evokes an action potential. Recovery took place after 85 s in A and more than 140 s in B.

Voltage clamp measurements at holding potentials close to resting membrane potential showed that hypoxic membrane depolarization originated from a net inward current of 0.3 ± 0.6 nA (18 measurements in 7 cells; 9 measurements in inspiration and 9 in expiration). Spontaneous synaptic noise currents remained relatively unchanged. The hypoxia evoked inward current decreased during recovery from hypoxia in those cases tested (Fig. 3).

Figure 3.

Voltage clamp measurement of the effect of hypoxia on a hypoglossal motoneurone

The voltage trace (upper trace), the current recorded (middle trace) and the phrenic nerve activity (PN) are illustrated, before hypoxia (Control), at the end of hypoxia test when ventilation was resumed (middle panel) and after recovery (right panel). All cells tested in this way showed an inward current that was maintained throughout the respiratory cycle during hypoxia test. In the upper two traces the rapid deflections are SLN stimulus artifacts (small deflections) and evoked action potentials. Vh, holding potential.

Hypoxic effects on stimulation-evoked postsynaptic responses

Under control conditions, SLN stimulation evoked a sequence of EPSUPs and IPSUPs, which were qualitatively similar independent of whether they were evoked during inspiration or expiration (Figs 4A and 5A). Quantitative differences, however, were apparent in those hypoglossal motoneurones which received respiratory drive potentials.

Figure 4.

Effect of hypoxia on evoked postsynaptic potentials obtained by stimulation of the ipsilateral SLN

A, hypoxia decreased both evoked EPSUP and IPSUP during inspiration and expiration. No differences were seen between inspiration and expiration. Evoked responses were obtained at –44 mV membrane potential during control in both phases and at –30 mV during hypoxia test. B, a short burst of SLN stimuli initiated an IPSUP which was clearly reduced during hypoxia (membrane potential was –56 mV during control and –48 mV during hypoxia). All traces are aligned.

Figure 5.

Effect of hypoxia on postsynaptic evoked currents

A, postsynaptic events evoked on SLN stimulation during inspiration and expiration recorded in current clamp. B, decrease in amplitude of the sequence of EPSC and IPSC from the same neurone evoked by SLN stimulation at holding potential (Vh of –70 mV during inspiration and expiration. All traces are aligned. Im, membrane current.

In such cases, synaptic potentials were reduced during periods of synaptic inhibition (Fig. 5A).

When neurones were depolarized during hypoxia, stimulation-evoked synaptic potentials were reduced in amplitude. EPSUPs appeared to be more sensitive to hypoxia as they decreased by, on average, –46% (0.8 ± 0.7 mV; n= 16, P < 0.05) from a control of 2.0 ± 0.8 mV (Fig. 4A) or disappeared completely. IPSUPs decreased by –30% (0.6± 0.7mV; n=16, P < 0.05) from a control of 1.8 ± 1.7 mV (Fig. 4A and B), but sometimes were also abolished.

In voltage clamp, SLN stimulation-evoked responses revealed equivalent sequences of EPSCs and IPSCs (Fig. 5B). During hypoxia, the magnitude of SLN stimulation-evoked postsynaptic currents were also reduced. EPSCs declined by 50% (0.1 ± 0.1 nA; n= 10, P < 0.05) from a control value of 0.20 ± 0.17 nA, and IPSCs decreased by 49% (0.07 ± 0.07 nA; n= 9, P < 0.05) from a control value of 0.14 ± 0.05 nA. In some cases, both EPSCs and IPSCs were abolished (Fig. 5B).

Responses to applied adenosine

As extracellular levels of adenosine are known to rise during hypoxia, the possibility that this could be responsible for the changes in membrane potential and current as well as synaptic responses has been investigated. The application of adenosine to the dorsal surface of the medulla affected hypoglossal motoneurones after a latency of 4 min. Adenosine induced a membrane hyperpolarization of 5.6 ± 4.4 mV (n= 4; P < 0.05) from a mean potential of 62 ± 7.5 mV as seen in four of six tests (Fig. 6A). Also SLN stimulation-evoked postsynaptic potentials were affected.

Figure 6.

Effect of extracellular application of adenosine on membrane potential (A) and on the postsynaptic potentials evoked by SLN stimulation (B)

A, after 5 min diffusion, adenosine induced a membrane hyperpolarization throughout the respiratory cycle. Recovery was obtained 6 min after washout. B, superimposition of SLN-evoked responses obtained during inspiration in the control situation at –64 mV and after adenosine administration at –69 mV in the cell illustrated in A. Adenosine reduced both EPSUP and IPSUP. All traces are aligned.

EPSUP amplitudes diminished by 18 ± 8% (a reduction of 0.4 ± 0.25mV; n= 4, P < 0.05 from control) (Fig. 6B), while IPSUP amplitudes decreased by 46 ± 31% (a reduction of 1.6 ± 2 mV; n= 8, P < 0.05 from control) in parallel with membrane hyperpolarization. Recovery occurred after a delay of 3–6 min following washout of adenosine from the ‘pressure-foot’.

Blockade of ATP-regulated potassium currents

Previous studies have shown that several types of brainstem neurones reveal a persistent (Pierrefiche et al. 1996) or spontaneous (Trapp, Ballanyi & Bichter, 1994) activation of ATP-regulated channels even under normoxic conditions. We therefore sought to determine whether KATP channels were also spontaneously active in hypoglossal motoneurones in vivo. Measurements were made at holding potentials close to the resting membrane potential. Extracellular application of the specific KATP channel blocker glibenclamide (5 mm) within the local bath on the brainstem surface induced a steady inward current ranging from 0.04 to 0.12 nA after 5 min delay. Glibenclamide, also led to a 2.5-fold increase of SLN stimulus-evoked EPSCs from a mean of 0.1 ± 0.09 nA to 0.25 ± 0.15 nA (n= 6, P < 0.05) (Fig. 7). The effect on IPSC amplitudes (n= 6), however, was inconsistent and failed to reach statistical significance.

Figure 7.

Effect of extracellular application of glibenclamide on postsynaptic evoked currents

Glibenclamide (5 mm), after a diffusion time of about 5 min, significantly increased the amplitude of SLN-evoked EPSCs, while IPSCs were only modified to a small extent. All traces are aligned.


There is an abundant literature on in vitro effects of hypoxia on hypoglossal motoneurones (for review see: Haddad & Jiang, 1993a; Richter & Ballanyi, 1996), but this is the first analysis of the hypoxic response of hypoglossal motoneurones studied at the cellular level in mature cats in vivo. The analysis concentrated on the effects of hypoxia on resting membrane currents and afferent nerve stimulation-evoked postsynaptic currents. The data indicate that even in normoxic conditions hypoglossal motoneurones may be controlled by metabolic factors such as ATP and adenosine, while their compensatory mechanisms involving ion channels are weak and insufficient for effective protection from hypoxia.

Pronounced sensitivity of hypoglossal motoneurones to hypoxia

The sensitivity of hypoglossal motoneurones to hypoxia is described to be significantly stronger than in other brain stem neurones (Haddad & Donnelly, 1990; Haddad & Jiang, 1993a,b) such as, for example, in vagal motoneurones (Donnelly et al. 1992; Doutheil et al. 1992; Ballanyi et al. 1996) or respiratory neurones (Richter et al. 1991; Richter & Ballanyi, 1996). Such high sensitivity to hypoxia has been confirmed by the present in vivo experiments. The hypoxic response of in vivo recorded hypoglossal motoneurones is a monophasic membrane depolarization even when inspired PO2 levels were reduced to only 12 vol.%. This may reflect both a direct action of hypoxia on the neurone together with an action of input, related to the activation of the arterial chemoreceptors. However, the absence of an increase in sympathetic noise during hypoxia makes the first suggestion the most plausible. The degree of hypoxic depolarization following a more severe decrease of PO2 to 6 vol.% seemed, nevertheless, smaller than that observed in vitro, especially if one considers that neonatal neurones depolarize less than adult neurones (Haddad & Donnelly, 1990; Donnelly et al. 1992). However, the smaller magnitude of in vivo depolarizations was probably due to the less severe level of hypoxia when compared with the anoxic tests performed in vitro. This conclusion is supported by comparison of hypoxic delay times. The hypoxic responses reported for in vitro slice preparations of rats started after 4–5 min (Haddad & Donnelly, 1990; Donnelly et al. 1992; Doutheil et al. 1992), whilst the hypoxic responses of in vivo hypoglossal motoneurones were much faster. It became evident as a surprisingly rapid change in membrane potential whenever inspired oxygen levels were changed. The factors determining the fast kinetics of in vivo responses are as yet unclear. Factors that require consideration are the differences in temperature and hence metabolic rate, and regional blood supply of in vivo brainstem tissue. Compared with direct measurements in the vicinity of respiratory neurones (Acker & Richter, 1985; Richter & Acker, 1989; Morawietz, Ballanyi, Kuwana & Richter, 1995), hypoglossal motoneurones obviously start to depolarize at a time when tissue PO2 levels have fallen by only 50%. Biophysical properties of respiratory neurones are still unchanged when tested at such tissue PO2 levels (Richter et al. 1991) and even later they display long periods of voltage stabilization.

Another striking in vivo observation was the obvious instability of cell recordings. Although we used fine-tipped electrodes filled with the Ca2+ chelator BAPTA, our recordings tended to abort relatively early during hypoxia. This effect was not attributable to changes in arterial blood pressure (which altered to only a small degree in the first minute) in contrast to our experiences with other brainstem neurones where hypoxic periods lasted considerably longer inducing much stronger cardiovascular responses in those cases (Richter et al. 1991; Richter & Ballanyi, 1996). We therefore consider the possibility that apparent instabilities of recordings originated, at least in part, from prominent inward cation currents within the dendrites (Viana, Bayliss & Berger, 1993).

Synaptic transmission during hypoxia

Our studies also revealed a marked influence of hypoxia on synaptic inputs to hypoglossal motoneurones. SLN stimulation-evoked postsynaptic currents and potentials were quickly attenuated during hypoxia. Sometimes the effects were so marked that synaptic transmission was totally blocked during early periods of hypoxia at times when central respiratory activity was still enhanced. This implies a hypoxic depression acting at pre- and post synaptic sites that occurs earlier than in other brainstem neurones, e.g. respiratory neurones. In respiratory neurones, a comparable effect was not seen before the beginning of the apnoeic response to hypoxia, which occurs several minutes later than the changes seen in hypoglossal motoneurones as described in the present report (Richter et al. 1991). Excitatory synaptic inputs in hypoglossal motoneurones were found to be more sensitive to hypoxia than inhibitory inputs. This observation is consistent with the assumption that early suppression of inhibitory synaptic transmission (Fujiwara et al. 1987; Krnjevic, Xu & Zhang, 1991; Luhmann & Heinemann, 1992; Luhmann et al. 1993; Rosen & Morris, 1993; Zhu & Krnjevic, 1993) originates from a more sensitive suppression of excitatory synaptic transmission by inhibitory interneurones (Khazipov, Bregestovsky & Ben-Ari, 1993; Congar, Khazipov & Ben-Ari, 1995).

The role of adenosine and ATP-regulated channels

One factor to be considered for adjustment of neuronal excitability to metabolic demand is adenosine that derives from cellular metabolism and quickly equilibrates with the extracellular compartment (Greene & Haas, 1991) and appears in increasing amounts during hypoxia (Schmidt-Garcon, Nagel & Richter, 1992). Through activation of A1 receptors, adenosine down-regulates cAMP-dependent protein kinase A and thus primarily blocks Ca2+ influx in presynaptic terminals to reduce exocytosis of transmitter and may act directly on both Ca2+ and K+ currents which may also directly modulate transmitter release (Dunwiddie & Haas, 1985; Scanziani, Capogna, Gahwiler & Thompson, 1992; Mogul, Adams & Fox, 1993; Umemiya & Berger, 1994; Wu & Saggau, 1994; Ulrich & Huguenard, 1995). Adenosine also down-regulates ligand-controlled receptor activity on postsynaptic sites (Siggins & Schubert, 1981; Zhu & Krnjevic, 1993; Schmidt, Bellingham & Bichter, 1995). The effects of extracellular adenosine application on synaptic transmission seen in the present study are consistent with these data and confirm previous findings on rat brain slices, where adenosine blocks excitatory synaptic transmission in hypoglossal motoneurones through its presynaptic action (Bellingham & Berger, 1994; Umemiya & Berger, 1994). The more effective adenosinergic depression of IPSUP is consistent with similar findings in respiratory neurones (Schmidt, Bellingham & Bichter, 1995) and indicates additional effects on postsynaptic receptors for inhibitory transmitters as a consequence of the activation of A2 receptors (Ben-Ari, Krnjevic & Crepel, 1990; Greene & Haas, 1991; Marks, Donnelly & Haddad, 1993; Mogul, Adams & Fox, 1993). Depression of synaptic transmission, however, cannot explain hypoxic membrane depolarization as there was no evidence in our voltage clamp measurements (Fig. 3) that there is continuous bombardment of in vivo hypoglossal motoneurones with inhibitory synaptic inputs that are blocked during hypoxic episodes.

Our finding of hypoxic activation of a persistent net inward current (Fig. 3) shows that protective K+ currents are relatively ineffective. We therefore studied the functional role of KATP channels and found that KATP channels are persistently activated under normoxia and these down-regulate EPSCs. This effect seems to occur largely at a presynaptic release site, as it occurs without major changes of postsynaptic membrane conductances. This leads to the conclusion that the contribution of KATP currents to resting membrane potential is small. This becomes obvious when the effect of KATP channel blockade is compared with that of respiratory neurones (Pierrefiche et al. 1996). Further activation of KATP currents also seems to be ineffective and cannot protect hypoglossal motoneurones against hypoxic membrane depolarization (Kerry, Murphy & Greenfield, 1992; Jiang, Xia & Haddad, 1992; Haddad & Jiang, 1993a). All this supeaks for a low KATP channel expression in hypoglossal motoneurones.

The physiological role of hypoglossal motoneurones in hypoxia

Taken together, our observations indicate that, under in vivo conditions, the hypoglossal nucleus is not well supplied by arterial blood flow, so that hypoglossal motoneurones are under permanent metabolic stress even during normoxia. In addition, hypoglossal motoneurones seem to have only low compensatory capacities when energy supply is reduced. Therefore, hypoglossal motoneurones depolarize rapidly during hypoxia leading to a transient increase of action potential discharge frequency before Na+ action potentials inactivate. This only transiently increases airway patency and lowers the resistance to airflow to minimize work load during enhanced tidal ventilation. Shortly afterwards, synaptic transmission is blocked, which results in failure of respiratory adjustment of tongue muscle tone although central respiratory activity is still augmented. This must produce relaxation of tongue muscles and airway obstruction. Such mechanisms might contribute to obstructive sleep apnoea (Henderson-Smart & Bead, 1980; Kavey, Whyte, Blitzer & Gidro-Frank, 1989; Säriås et al. 1989; Lowe, 1990; Bichard & Harper, 1991; Henke, Dempsey, Badr, Kowitz & Skatrud, 1991).


This work was supported by the SFB 406. K. M. Supyer is supported by a Wellcome Trust Programme Grant.