Corresponding author S. L. Mironov: II Department of Physiology, University of Göttingen, Humboldtallee 23, 37075 Göttingen, Germany. Email: email@example.com WWW: www.neuro-physiol.med.uni-goettingen.de
1Whole-cell (ICa) and single Ca2+ channel currents were measured in inspiratory neurones of neonatal mice (4–12 days old). During whole-cell recordings, ICa slowly declined and disappeared within 10–20 min. The run-down was delayed during hypoxia, indicating ICa potentiation.
2Ca2+ channels were recorded in cell-attached patches using pipettes which contained 110 mm Ba2+. L-type Ca2+ channels exhibited a non-ohmic I–V relationship. The slope conductance was 24 pS below and 50 pS above their null potential. The open probability of the channels increased during oxygen depletion, reaching a maximum 2 min after the onset of hypoxia. Restoration of the oxygen supply brought the channel activity back to initial levels.
3The channel activity was enhanced by 3–30 μmS(–)Bay K 8644, an agonist of L-type Ca2+ channels. The open probability was increased about 3-fold and the activation curve was shifted by 20 mV in the hyperpolarizing direction. In the presence of the agonist, channel open time increased and long openings appeared. Agonist-modulated channels were also potentiated during oxygen depletion. The effect was due to an increase in open time and a decrease in closed time. The channels were inhibited by bath application of nifedipine (10 μm) and nitrendipine (20 μm).
4Weak bases such as NH4Cl and TMA increased and weak acids such as sodium acetate and propionate decreased activity of the channels, indicating that they are modulated by intracellular pH. Bath application of 1 μm forskolin enhanced the channel activity, whereas 500 μm NaF suppressed it.
5L-type Ca2+ channels were modulated by an agonist for mGluR1/5 receptors, (S)-3,5-dihydrophenylglycine (DHPG, 5 μm). In its presence, the hypoxic facilitation of channels was abolished.
6After blockade of L-type Ca2+ channels, the respiratory response to hypoxia was modified. The transient enhancement of the respiratory rhythm (augmentation) was no longer evident and the secondary depression occurred earlier.
7We suggest that L-type Ca2+ channels contribute to the early hypoxic response of the respiratory centre. Glutamate release during hypoxia stimulates postsynaptic metabotropic glutamate receptors, which activate the Ca2+ channels.
The present study was designed to identify Ca2+ channels in respiratory neurones and to determine whether their activity is modulated by hypoxia. Whole-cell measurements of high-voltage-activated Ca2+ currents (ICa) suggested that they are potentiated during hypoxia. Due to the small amplitude and relatively fast run-down of ICa in inspiratory neurones, the effects on whole-cell currents could not be investigated in much detail. Therefore, we analysed single Ca2+ channels in cell-attached patches and identified them by their conductance and gating pattern. According to the conductance, kinetics, ion permeation properties and pharmacology, the channels were classified as dihydropyridine-sensitive, voltage-gated Ca2+ channels or L-type Ca2+ channels. Similar to other cells, L-type Ca2+ channels in inspiratory neurones were found to be modulated by intracellular pH and cAMP. The novel finding is that single Ca2+ channels were activated during hypoxia. They were also potentiated by agonists of mGluR1/5 receptors, after which hypoxic facilitation of the channels could no longer be observed. We suggest that L-type Ca2+ channels participate in the transient enhancement of the respiratory rhythm during early hypoxia. Their activation is caused by glutamate release, leading to activation of postsynaptic metabotropic receptors.
Experiments were performed on medullary slice preparations from neonatal mice (P4–12) that contain the functional respiratory network and generate slow oscillatory activity. The slices were prepared as documented in detail by Smith et al. (1991) with modifications described previously (Mironov et al. 1998). All animals were housed, cared for and killed in accordance with the recommendations of the European Commission (No. L 358, ISSN 0378-6978) and the protocols were approved by the Committee on Animal Research, Göttingen University. The animals were put in an anaesthetic chamber with cotton wool soaked in ether. When respiration ceased, the animals were removed from the chamber and rapidly decapitated. The brainstem-spinal cord was isolated in ice-cold artificial cerebrospinal fluid (ACSF, for composition see below) that was saturated with carbogen (95 % O2 and 5 % CO2). Using a vibratome (Campden Instruments, UK), the brainstem was sectioned serially from rostral to caudal in the transverse plane at an angle of 135 deg between vibratome blade and neuraxis. A single transverse 700 μm thick slice containing the pre-Bötzinger complex (Smith et al. 1991) was cut from the brainstem, transferred to a recording chamber and continuously superfused at 28°C with carbogen-saturated ACSF. To prevent loss of dissolved gases, the perfusing solution was delivered to the experimental chamber via stainless steel tubes. The respiratory rhythm was stabilized at elevated extracellular K+ concentrations ranging from 8 to 10 mm.
Oxygen tissue measurements
In order to induce hypoxic conditions, oxygen was replaced by nitrogen in the bubbling gas mixture. The oxygen level was measured using oxygen-sensitive electrodes (tip diameter 10–20 μm) connected to a chemical microsensor (Diamond Electro-Tech Inc., Ann Arbor, MI, USA). Before measurements, the PO2-sensitive electrode was polarized for 1 h by passing −750 nA and calibrated in the perfusion chamber using two ACSF solutions that were saturated with carbogen or with a gas mixture containing 95 % N2 and 5 % CO2, and 1 mm Na2S2O3 to determine a zero oxygen level. The oxygen-sensitive electrode was gently driven into the tissue using a piezomanipulator (SPI, Oppenheim, Germany). The electrode was placed about 100 μm below the slice surface, in the layer where most inspiratory neurones were recorded. Fifteen to twenty seconds after exchanging oxygen for nitrogen in the perfusing solution, extracellular PO2 dropped from 200 ± 50 to 5 ± 3 mmHg and then remained constant. The responses to hypoxia were readily reproducible and could be repeated 20–30 times for a given slice with complete restoration of rhythmic activity, even when O2 had been removed for 20–30 min.
The respiratory rhythm in the slices was monitored from the cut end of the hypoglossal (XII) nerve roots with a suction electrode. Recording electrodes were made from broken patch pipettes with fire-polished tips (5–10 μm openings). Extracellularly recorded signals were amplified 20 000 times, bandpass filtered (0.1–3 kHz), rectified and integrated (50–100 ms time constant).
Patch electrodes manufactured from borosilicate glass capillaries (Clark Instruments) had tip openings of 1.5–2 μm and resistances of 2–4 MΩ. Intracellular signals were recorded with a patch-clamp amplifier (EPC-7, ESF, Friedland, Germany). Membrane currents were filtered at 3 kHz (−3 dB), digitized at 5 kHz, and stored for off-line analysis with an IBM-compatible PC. Data analysis was performed using home-written programs on Turbo-Pascal 7.0. Data are presented as means ±s.e.m. Statistical significance was determined using Student's t test and accepted if P was less than 0.05. All single-channel data are presented according to conventions adapted for intracellular recordings, i.e. the current and patch command potential were taken as the value inside the cell minus the outside value. The open probability, Popen, was obtained by dividing the mean current by the unitary current and the number of active channels, which was determined as the maximal number of simultaneous openings ever seen during a given experiment. This was confirmed by binomial analysis.
In whole-cell measurements made with potassium gluconate solutions, the resting potentials of inspiratory neurones ranged from −53 to −67 mV (mean −60 ± 3 mV, n= 17). In cell-attached recording mode, we therefore assumed that the intracellular voltage is −60 mV. As Ca2+ channels are voltage dependent, their activity was monitored by stepping the patch command potential to positive voltages. In terms of membrane potential, the voltage steps from 0 to +60 mV in cell-attached mode would correspond to voltage shifts from −60 to 0 mV where activation of Ca2+ channels reaches a maximum (Figs 1 and 6). All single-channel measurements were made in cell-attached mode. Excised patches were stable, but channel activity quickly faded, which is typical for high-voltage-activated Ca2+ channels (cf. Chad & Eckert, 1984; Levitan, 1988).
Inspiratory neurones were identified by their on-going activity that correlated with the hypoglossal rhythm (see Fig. 1A in Mironov et al. 1998). In comparison with other cells, e.g. expiratory and tonic neurones, spikes on current recordings obtained in cell-attached mode occurred only during the inspiratory phase and therefore did not contaminate channel activity measured during the interburst intervals. Therefore, all single-channel data were measured and analysed within windows between inspiratory bursts. In whole-cell mode, the inspiratory neurones demonstrated synaptic drive currents (Smith et al. 1991; see Fig. 5).
Solutions and drugs
ACSF contained (mm): 128 NaCl, 3 KCl, 1.5 CaCl2, 1.0 MgSO4, 21 NaHCO3, 0.5 NaH2PO4 and 30 D-glucose. The pH was adjusted to 7.4 with NaOH. Solutions with elevated K+ levels (8–10 mm) were obtained by replacing NaCl with KCl. The pipette solution for single-channel recordings contained (mm): 110 BaCl2 and 10 Hepes. The pH (7.4) was adjusted with NaOH. The osmolarity was 285–290 mosmol l−1. The intracellular solution used for whole-cell recording contained (mm): 80 caesium gluconate, 40 tetraethylammonium (TEA+) gluconate, 15 NaCl, 2 MgCl2, 10 Hepes, 0.5 Na2ATP, 1 CaCl2 and 3 BAPTA. This Ca2+-buffering cocktail corresponds to an estimated free Ca2+ concentration of about 0.1 μm (Kay, 1992). Solutions containing potassium gluconate instead of (Cs++ TEA+) gluconate were used to measure the resting potential in inspiratory neurones. All salts used for extra- and intracellular solutions were obtained from Sigma. The Ca2+ channel agonist S (-)Bay K 8644, and the antagonists nifedipine and nitrendipine were purchased from Calbiochem. Drugs selective for metabotropic glutamate receptors ((1S, 3R)-1-aminocyclopentane-1,3-dicarboxylic acid (t-ACPD); (S)-3,5-dihydrophenylglycine (DHPG); (RS)-1-aminoindan-1,5-dicarboxylic acid (AIDA); N-acetyl-L-aspartyl-L-glutamic acid (NAAG); (2S, 1′S, 2′S)-2-(carboxycyclopropyl)-glycine (L-CCG-I); and L-2-amino-4-phosphonobutyric acid (L-AP4)) were obtained from Tocris Cookson.
Inward Ca2+ current in inspiratory neurones and the effects of hypoxia
Whole-cell currents were recorded in inspiratory neurones using intracellular solution containing 40 mm TEA+ and 80 mm caesium gluconate. Voltage steps applied from a holding potential of −60 mV activated a long-lasting inward current that remained nearly unchanged after 3 min hypoxia, but was markedly reduced 4 min after restoration of oxygen content in the bath (Fig. 1). Similar changes in current were measured in six other inspiratory neurones. The normalized mean amplitudes of the inward current measured after 3 min hypoxia and after 4 min reoxygenation were 0.89 ± 0.11 and 0.44 ± 0.09 of the initial value, respectively. During hypoxia, the activation curve was shifted in the hyperpolarizing direction. During reoxygenation its position along the voltage axis was restored (Fig. 1B). These effects can be expressed in terms of potentials (V½), for which peak current is half-maximal (cf. Kostyuk et al. 1982). For seven cells, mean V½ values were −20 ± 2 mV (control), −29 ± 3 mV (3 min hypoxia) and −19 ± 2 mV (5 min recovery).
In control experiments, when the inward current was monitored immediately after breaking into the cell, the current amplitude steadily declined at the average rate of 9 ± 2 % min−1 (n= 9, Fig. 1C) and during hypoxia the run-down rate decreased to 1 ± 1 % min−1 (n= 7). Thus without run-down, the inward current would have been increased by about 20 % after 3 min hypoxia. The voltage dependence, kinetics and the lability of the inward current match the properties of L-type Ca2+ channels (cf. Kostyuk & Krishtal, 1977; Chad & Eckert, 1984; Mironov & Lux, 1991b). We were not able to determine the composition of ICa pharmacologically using whole-cell recordings, since the action of dihydropyridines in slices was slower than run-down of ICa. Nevertheless, the recording conditions (Vh=−60 mV) and the slow inactivation of the current allowed us to assume that a large portion of this current passed through L-type Ca2+ channels. Other Ca2+ channel types, which could have contributed to the inward current at this Vh, show different kinetics and are largely inactivated. Another candidate which has to be considered is the persistent sodium current, which was reported to increase during hypoxia in myocytes (Ju et al. 1996). However, it is robust and does not demonstrate run-down. Therefore we focused our efforts to identify single L-type Ca2+ channels, aiming to answer the question of whether they could be activated by hypoxia.
Dihydropiridine-sensitive Ca2+ channels in cell-attached patches are up-regulated by hypoxia
Single Ca2+ channels were measured in cell-attached patches using pipettes filled with 110 mm Ba2+. When oxygen was removed from the bath, the activity of Ca2+ channels increased (Fig. 2A) and decreased again after restoration of the oxygen content (Fig. 2B). After 2 min of hypoxia, the open probability increased from 0.09 ± 0.02 to 0.18 ± 0.03 (n= 7). Addition of the L-type Ca2+ channels agonist Bay K 8644 (10 μm) to the bath, 5 min after recovery from hypoxia, produced a slow increase in the channel activity (Fig. 2C).
In order to exclude possible indirect effects of bath-applied Bay K 8644, the drug was added to the pipette solution. In about half of all neurones examined (21/47), there was initially no channel activity, but then it appeared slowly (Fig. 3A). In other patches (n= 26), the channel activity was observed from the very beginning and slowly increased over a similar time course, indicating that channels were activated by the agonist that diffused out of the pipette. Ten minutes after seal formation the effect reached a steady state, and, on average, the open probability increased from 0.15 ± 0.02 to 0.43 ± 0.11 (n= 26). Agonist-modulated channels were also potentiated by oxygen withdrawal (Fig. 3B). The effect started to develop within the first minute of hypoxia and reached a steady state after 2 min. The mean increase in Popen ranged from 1.5- to 3-fold with a mean of 2.2 ± 0.7 (n= 9). Spontaneously active as well as agonist-modulated channels were blocked by 10 μm nifedipine (n= 8, data not shown).
Hypoxia did not change the single-channel conductance in either spontaneously active or agonist-modulated channels. The activation curve was shifted in the hyperpolarizing direction by 8 ± 5 mV (n= 7, P= 0.1, data not shown). To exclude possible variations in the resting potential, changes in channel gating were analysed for maximal values of Popen at a patch command potential of +160 mV, at which the channels carry outward currents. For spontaneously active channels, the distribution of open times was best fitted by a single exponential function (Fig. 4). In the presence of Bay K 8644, long-lasting channel closures disappeared and long openings became more frequent as evidenced by the long tail in the histogram of the open time distribution (Fig. 4). We did not estimate the slow time constant, because a much larger number of long openings would have been necessary, but they were often masked by on-going spiking activity.
The mean values of the fast open time constant obtained for six different patches at patch command potential +160 mV were 1.5 ± 0.2 ms (control), 2.4 ± 0.3 ms (after agonist activation) and 3.0 ± 0.3 ms (after 2 min of hypoxia). The mean values of closed time constants were 4.8 ± 0.4 ms, 2.0 ± 0.3 ms and 1.4 ± 0.3 ms, respectively (P < 0.01). For the simplest kinetic scheme of channel gating containing only one closed and one open state, Popen= 1/(1 +τcl/τop). Three pairs of values for τop and τcl gave mean open probabilities of 0.23, 0.55 and 0.68, respectively, in agreement with measured values of Popen. The small differences may reflect the influence of long closures for spontaneously active channels and long openings recorded in the presence of agonist. These two effects would lower control Popen values, but would increase them for agonist- and hypoxia-modulated channels.
Involvement of L-type Ca2+ channels in the respiratory hypoxic response
The hypoxic potentiation of L-type Ca2+ channels appeared to be relevant for the response of respiratory network to oxygen depletion. Accordingly, we recorded changes in synaptic currents during hypoxia in the absence and in the presence of Ca2+ channel blockers. In comparison with Ca2+ channel currents, synaptic currents were more stable and deteriorated much more slowly (> 30 min of intracellular recording).
Figure 5 presents a typical hypoxic response of an inspiratory neurone measured at a holding potential of −50 mV. Such cells exhibit both spontaneous inhibitory and excitatory synaptic currents. The latter summate to produce synaptic drive currents (long-lasting inward deflections) that coincide with the inspiratory burst discharges recorded from hypoglossal rootlets. About 1 min after onset of hypoxia, both the amplitude and frequency of synaptic drive currents increased in association with the initial hypoxic augmentation of the respiratory rhythm. The transient enhancement was followed by respiratory depression.
In the presence of 20 μm nitrendipine hypoxic augmentation of respiratory activity was greatly reduced and depression occurred earlier (Fig. 5). Similar effects were observed in four other inspiratory neurones. The mean interval between inspiratory bursts was 6.1 ± 0.5 s during control, 4.2 ± 0.3 s at the peak of augmentation and 9.5 ± 0.7 s at the end of augmentation, respectively. In the presence of 20 μm nitrendipine, corresponding values were 3.8 ± 0.3, 4.5 ± 0.4 and 5.8 ± 0.5 s, respectively (n= 5).
Properties of single Ca2+ channels in cell-attached patches
Ca2+ channels are also able to carry monovalent ions (Kostyuk & Krishtal, 1977; Kostyuk et al. 1983). In the presence of extracellular divalent cations, this outward current can be observed only at high positive voltages, whose electromotive force prevents Ca2+ from entering the channel. The I–V relationship of the L-type Ca2+ channel currents demonstrated non-ohmic behaviour, resembling that measured for whole-cell currents in heart cells (McDonald et al. 1986). This relationship is also consistent with theoretical predictions of Ca2+ channel permeability (Mironov, 1992). The null potential (Enull) corresponded to a command voltage of +120 mV. Considering the resting membrane potential of inspiratory neurones of about −60 mV, this gives an apparent reversal potential of +60 mV. The conductance of non-selective outward currents carried through the channel predominantly by intracellular K+ ions was difficult to define unambiguously due to the non-linear behaviour of the single-channel current at potentials positive to Enull. The linear regression of data points above Enull (Fig. 6A) corresponded to a conductance of 50 pS, close to values (70–90 pS) obtained for the inward Ca2+ channel currents carried by monovalent cations in the complete absence of divalent cations (Matsuda, 1986; Kuo & Hess, 1993).
The channel gating was voltage dependent. In the presence of Bay K 8644, the open probability increased and the activation curve was shifted in the hyperpolarizing direction (Fig. 6B). The maximal Popen values obtained for command potentials larger than +90 mV were 0.16 ± 0.07 in control and 0.47 ± 0.13 in the presence of agonist (n= 11). The activation curve recorded for single channels was also consistent with that determined for whole-cell currents (Fig. 1), assuming the cell resting potential of about −60 mV. The unitary current flowing through open channels represents the instantaneous I–V relationship for the whole-cell current. Multiplying the unitary current by Popen results in a bell-shaped curve for the mean current, which is characteristic for the whole-cell Ca2+ current (cf. Kostyuk & Krishtal, 1977; Chad & Eckert, 1986). Prolonged depolarizations induced channel inactivation. The openings became less frequent at the end of depolarizing pulses, producing a slowly declining ensemble current (Fig. 6C). Patch depolarization by 30–50 mV above the resting potential for the duration of 10–20 s completely inactivated the channels (n= 5, data not shown).
L-type Ca2+ channels and intracellular metabolism
A distinctive feature of L-type Ca2+ channels is their modulation by various metabolic factors (Chad & Eckert, 1984; Levitan, 1988). In inspiratory neurones, the channels behaved similarly to the L-type channels in other preparations. For example, channel activity in inside-out patches irreversibly declined within a minute after the patch had been excised (mean channel lifetime was 1.5 ± 0.6 min, n= 10).
In cell-attached patches, changes in intracellular cAMP modulated L-type Ca2+ channels. The channels were potentiated by 1 μm forskolin, an activator of adenylyl cyclase (Fig. 7A). In contrast, NaF, which inhibits the adenylyl cyclase at submillimolar concentration (Blackmore et al. 1985), suppressed Ca2+ channel activity (Fig. 7B). Both NaF and forskolin acted slowly and the maximal effects were obtained 5–10 min after the drugs were applied. The effects did not involve changes in channel conductance or voltage dependence of channel gating. Nevertheless, in order to exclude possible variations in cell resting potentials, the effects of the two agents were estimated at patch command voltages above +150 mV where the open probability reached maximal values (Fig. 6B). Accordingly, Popen values were calculated for non-specific outward currents through the Ca2+ channels. For agonist-modulated channels in the presence of NaF, maximal Popen values decreased from 0.51 ± 0.09 to 0.19 ± 0.03 (n= 8). Ten minutes after addition of 1 μm forskolin, the maximal Popen increased from 0.43 ± 0.09 to 0.62 ± 0.12 (n= 9).
Among Ca2+ channels, only L-type channels are known to be modulated by intracellular pH (Mironov & Lux, 1991a). A similar modulatory mechanism seems to be operative in inspiratory neurones. The channel activity was increased by NH4Cl or trimethylammonium (TMA) and decreased in the presence of acetate or propionate (Fig. 8). These weak bases and acids readily permeate the cell membrane in neutral form, making the cytoplasm alkaline or acidic, respectively. Mean increases in maximal Popen values for agonist-modulated channels were 1.98 ± 0.23 (n= 7) for NH4Cl and 1.87 ± 0.21 (n= 6) for TMA, respectively. Sodium acetate and sodium propionate decreased maximal Popen values to 0.48 ± 0.07 (n= 7) and 0.42 ± 0.06 (n= 6) from that of control. In dorsal root ganglion (DRG) neurones (Mironov & Lux, 1991a) and hippocampal pyramidal cells (Mironov, 1995), NH4Cl and sodium acetate changed pHi from 7.2 to 7.9 and 6.8, respectively. A similar range of pHi changes may also have been presumed in inspiratory neurones.
Activation of L-type Ca2+ channels by group I metabotropic glutamate receptors
Since glutamate levels are increased during hypoxia and Ca2+ channels can be modulated by metabotropic glutamate receptors (mGluR), tests were made to determine whether such effects could underlie the facilitation of Ca2+ channels during hypoxia. Of the currently available specific activators of mGluRs (Conn & Pin, 1997), only group I (mGluR1/5) agonists potentiated the activity of single L-type Ca2+ channels (Fig. 9A). Two minutes after addition of 5 μm DHPG to the bath, maximal Popen values increased from 0.44 ± 0.07 to 0.62 ± 0.06 (n= 6). The effects of DHPG were prevented by AIDA (10 μm, n= 4, data not shown), a specific antagonist of the mGluR1/5 receptors. In the presence of DHPG, the channels could not be further modulated by hypoxia (Fig. 9B). On the contrary, they were slightly inhibited and the maximal Popen values were 0.64 ± 0.07 in control and 0.54 ± 0.08 after 2 min of hypoxia (n= 6, P < 0.05). Other types of mGluR agonists (Conn & Pin, 1997), such as NAAG (5–20 μm), L-CCG-I (4–18 μm), L-AP4 (10–50 μm) were not effective. For Bay K-modulated channels, the maximal Popen values before and 5 min after application of mGluR agonists were 0.48 ± 0.09 and 0.42 ± 0.07 (n= 4, 20 μm NAAG), 0.52 ± 0.10 and 0.44 ± 0.08 (n= 3, 18 μm L-CCG-I), 0.49 ± 0.11 and 0.46 ± 0.08 (n= 4, 40 μm L-AP4), respectively.
The changes in the respiratory activity and its hypoxic modulation produced by 10 μm AIDA (n= 4, data not shown), were similar to those observed for nitrendipine (Fig. 5). Interestingly 5 μm DHPG also abolished the hypoxic augmentation (n= 5, data not shown), which was probably caused by a maximal potentiation of Ca2+ channels induced by DHPG.
Ca2+ channel currents are important in the process of respiratory rhythm generation, since changes in intracellular Ca2+ during each respiratory cycle induce phasic alterations of Ca2+-activated conductances (Richter et al. 1993; Pierrefiche et al. 1995). In contrast to other high-voltage-activated channels, which are pre-synaptic and participate in synaptic transmission, L-type Ca2+ channels seem to be localized predominantly postsynaptically (Elliot et al. 1995) and may serve to shape the discharge patterns of respiratory neurones. L-type channel blockers alter firing properties of in vitro neonatal rat respiratory neurones (Onimaru et al. 1996) and activity of bursting neurones in cultured organotypic medullary tissue (Bingmann et al. 1995). In the present investigation we have identified single L-type Ca2+ channels in inspiratory neurones and showed the modulation of their activity by mechanisms which may be relevant to various physiological and pathophysiological phenomena.
Properties of L-type Ca2+ channels in inspiratory neurones
It is well established that in comparison to other components of whole-cell Ca2+ currents, L-type channels are labile in patch-clamp recordings (cf. Kostyuk & Krishtal, 1977; Chad & Eckert, 1986; Mironov & Lux, 1991b). Accordingly, whole-cell high-voltage-activated Ca2+ currents recorded from respiratory neurones in the present study had an average lifetime of 10 min, which hampered quantitative studies of metabolic ICa modulation, including the changes induced by hypoxia. Nonetheless, a potentiation of ICa by hypoxia can be estimated, when run-down of the current is taken into consideration (Fig. 1). In mouse sensory neurones, the whole-cell high-voltage-activated Ca2+ current declined during metabolic poisoning with CN− and decreased irreversibly further with its wash-out (Duchen, 1990). Correction for this run-down revealed an increase of ICa. In the presence of CN− the activation curve was shifted in the hyperpolarizing direction by about 10 mV. In the present study a comparable shift of 9 mV was seen. This might indicate that Ca2+ channels were indeed potentiated by CN− and hypoxia, since most channel openers act in this way, similarly to that observed for Bay K 8644 in the present study (Fig. 6B). We recorded single Ca2+ channels that were identified by their conductance (24 pS), ion selectivity, voltage dependence and sensitivity to dihydropyridines. The effects of Bay K 8644 were similar to those observed in other cells (Ganitkevich & Isenberg, 1990; Slesinger & Lansman, 1991).
Intracellular modulation of L-type Ca2+ channels
In inspiratory neurones, the operation of Ca2+ channels was found to be closely linked to intracellular signalling pathways. The channel activity followed changes in intracellular pH in a reciprocal fashion. Weak bases increased the channel activity and weak acids decreased it as documented for rat DRG cells (Mironov & Lux, 1991a). cAMP-mediated faciliation is well documented for whole-cell Ca2+ currents (cf. Chad & Eckert, 1984; Levitan, 1988; Mironov & Lux, 1991b). Accordingly, NaF inhibited the channel activity, whereas forskolin potentiated it. The inhibitory effect of submillimolar NaF was probably due to activation of Gi proteins which suppress the adenylyl cyclase (Blackmore et al. 1985).
L-type Ca2+ channels and respiratory hypoxic response
Even brief hypoxic episodes change cell metabolism and alter the operation of ion channels and receptors. For example, hypoxia activates the persistent sodium current (Ju et al. 1996), Ca2+ currents (Sun & Reis, 1994), and Ca2+- and ATP-sensitive K+ channels (Jiang et al. 1994; Leyssens et al. 1996; Mironov et al. 1998). Activation of such conductances may contribute to the biphasic hypoxic response of the respiratory network. The results presented here point out the involvement of the L-type Ca2+ channel in the augmentation of respiratory discharge during early hypoxia.
The hypoxic increase in Ca2+ channel activity developed faster than the potentiation of KATP channel activity that is responsible for the secondary respiratory depression (Mironov et al. 1998). In the presence of the L-type channel blocker nitrendipine, the hypoxic augmentation could not be observed and depression occurred earlier. Thus, ICa potentiation could underlie the decrease in extracellular Ca2+ as measured during early hypoxia (Trippenbach et al. 1990; Martin et al. 1995).
In a recent abstract, Elsen & Ramirez (1997) reported a hypoxic depression of whole-cell Ca2+ currents in a brainstem preparation of mice. The reason for the discrepancy between their results and ours is as yet unclear. One possibility is that the authors measured a mixture of Ca2+ currents, which were activated from a holding potential of −100 mV, which might respond differently to oxygen withdrawal. Secondly, as was shown in the present investigation, run-down of L-type Ca2+ currents complicates the analysis of hypoxic effects by using the whole-cell recordings only.
Hypoxic potentiation of L-type Ca2+ channels and postsynaptic metabotropic glutamate receptors
Our findings seem to be closely related to facilitation of L-type Ca2+ channels via mGluR1/5 receptors observed in cultured cerebellar granule cells (Chavis et al. 1995a, b). The effects recorded were very fast, resembling hypoxic activation of Ca2+ channels. In this study, a specific agonist of mGluR1/5 receptors (DHPG) increased the activity of L-type Ca2+ channels and occluded their facilitation by hypoxia. Interestingly both mGluR1/5 receptors (Ottersen & Landsend, 1997) and L-type Ca2+ channels (Elliot et al. 1995) show predominantly postsynaptic localization. Therefore such mechanisms as glutamate release, activation of mGluR1/5 receptors and facilitation of L-type Ca2+ channels would explain the hypoxic augmentation. The intermediate steps, involving specific biochemical pathways, should be elucidated, however.
The study was supported by SFB 406. The authors thank Hella Timmermann for technical assistance, Kersten Langohr for help with the preparations and Peter Lalley for critical reading of the manuscript.