Neuroscience 122 (2003) 831–841

A COMBINED BLOCKADE OF AND CALCIUM-DEPENDENT POTASSIUM CHANNELS ABOLISHES THE RESPIRATORY RHYTHM

D. BU¨ SSELBERG,* A. M. BISCHOFF AND of the glycine blocker into the pre- D. W. RICHTER Bo¨tzinger complex of anaesthetized mature cats resulted Georg-August-Universita¨t Go¨ttingen, Abteilung fu¨r Neuro- und Sin- in an absolute cessation of rhythmic respiratory discharges nesphysiologie, Humboldtallee 23, D-37073 Go¨ttingen, Germany (Pierrefiche et al., 1998). A recent study, however, showed that the respiratory rhythm persists in the non-anaesthe- Abstract—In order to test whether glycinergic inhibition is tized brainstem preparation in wild type mice after failure of essential for the in vivo respiratory rhythm, we analysed the glycinergic inhibition (Bu¨sselberg et al., 2001a) as well as discharge properties of neurones in the medullary respira- in mutant oscillator mice (Bu¨sselberg et al., 2001b). This tory network after blockade of glycine receptors in the in situ observation was unexpected, since these oscillator mice perfused brainstem preparation of mature wild type and os- were shown to have a complete loss of the mature ␣1, ␤ cillator mice with a deficient . type glycine receptors (Kling et al., 1997), whilst the em- In wild type mice, selective blockade of glycine receptors bryonic ␣2 and the ␣3, ␤ receptor subtypes were supposed with low concentrations of strychnine (0.03–0.3 ␮M) pro- voked considerable changes in neuronal discharge charac- to be absent at the age these animals were used at post- teristics: The cycle phase relationship of inspiratory, post- natal age of P18 to P22. Persistence of the respiratory inspiratory and expiratory specific patterns of membrane po- rhythm after failure of glycinergic inhibition raised the ques- tential changes was altered profoundly. Inspiratory, post- tion whether membrane properties of neurones may play a inspiratory and expiratory neurones developed a propensity key role in the in vivo respiratory rhythm generation of for fast voltage oscillations that were accompanied by multi- non-anaesthetized mammals (Richter, 1996). ple burst discharges. These burst discharges were followed To further investigate the significance of glycinergic by “after-burst” hyperpolarisations that were capable of trig- gering secondary burst discharges. Blockade of glycine re- inhibition for respiratory rhythm generation in an in vivo like -ceptors and the “big” Ca2؉-dependent K؉-conductance by condition, we have analysed cellular and network re charybdotoxin (3.3 nM) resulted in loss of the respiratory sponses after selective blockade of glycine receptors in rhythm, whilst only tonic discharge activity remained. In con- wild type mice (C57BL/6J or NMRI) as well as in homocy- trast, rhythmic activity was only weakened, but preserved gote oscillator mice (spdot/spdot)inthein situ perfused 2؉ ؉ after the “small” Ca -dependent activated K conductance brain stem preparation (Paton, 1996). This preparation was blocked with apamin (8 nM). Also low concentrations of allows analysis of specific synaptic processes and testing sodium (6 mg/kg) abolished rhythmic respira- tory activity after blockade of glycine receptors in the wild for the functional role of membrane properties in an intact type mice and in glycine receptor deficient oscillator mice. respiratory network under non-anaesthetized conditions. The data imply that failure of glycine receptors provokes Our studies show that respiratory neurones develop spon- enhanced bursting behaviour of respiratory neurones, whilst taneous voltage oscillations and secondary bursts when the additional blockade of BKCa channels by charybdotoxin glycinergic inhibition fails and suggest that a “big” Ca2ϩ- or with pentobarbital abolishes the respiratory rhythm. dependent Kϩ-conductance (BK ) conductance is essen- © 2003 IBRO. Published by Elsevier Ltd. All rights reserved. Ca tial for the maintenance of respiratory rhythm. A clinically Key words: rhythm generation, synaptic interactions, net- relevant finding was that low doses of pentobarbital abolish work functions, breathing behaviour, mice, oscillator mice. rhythmic respiratory discharges after failure of glycine re- ceptors as it may occur in patients with hyperekplexia. A preliminary report has been published in abstract Under in vivo conditions, periodic synaptic inhibition form (Bu¨sselberg et al., 2000). through glycinergic and GABAergic synapses is believed to exert a principal function in the respiratory network (Richter, 1996; Champagnat et al., 1982; Haji et al., 1992; EXPERIMENTAL PROCEDURES Schmid et al., 1996; Shao and Feldman, 1997). The func- Surgical procedures and preparation tional significance of glycinergic inhibition for respiratory rhythm generation became obvious when microinjections Analyses were performed on the in situ perfused brainstem prep- aration (Paton, 1996) of wild-type (C57BL/6J or NMRI; postnatal *Correspondence to: D. Bu¨sselberg, Universita¨tsklinikum Essen, Insti- age P20–P25) and mutant oscillator (spdot/spdot; postnatal age tut fu¨r Physiologie, Hufelandstrasse 55, 45122 Essen, Germany. Tel: P18–23) mice, which were bred and housed in the University of ϩ49-0-201-723-4627; fax: ϩ49-0-201-723-4627. Go¨ttingen (Go¨ttingen, Germany) animal facility. Oscillator mice E-mail address: [email protected] (D. Bu¨sselberg). 2ϩ were identified by their phenotype and by PCR analysis of tail Abbreviations: ACSF, artificial cerebrospinal fluid; BKCa, “big” Ca - dependent Kϩ-conductance; CTX, charybdotoxin; PN, phrenicus tissue. 2ϩ ϩ ot ot nerve; SKCa, “small” Ca -dependent K -conductance; spd /spd , All experimental procedures conformed to the recommenda- homozygous mutant oscillator mice. tions of the European Commission (No L 358, ISSN 0378-6978), 0306-4522/03$30.00ϩ0.00 © 2003 IBRO. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.neuroscience.2003.07.014

831 832 D. Bu¨sselberg et al. / Neuroscience 122 (2003) 831–841 and protocols were approved by the Committee on Animal Re- out the drugs did not have any effects. Pentobarbital sodium search of the University of Go¨ttingen. Every attempt was made to (Nembutal; Sanofi, CEVA; Germany) was administered in sub- minimize the number of animals used in this study. Our proce- anaesthetic doses of 6 mg/kg (Kissin et al., 1987). dures minimized pain and discomfort of animals used. Further- more mice were deeply anaesthetized with as verified Data analysis by the lack of nociceptive reactions to pinches applied to the paws and the tail. During maintained anesthesia, brains were transected All data, including arterial perfusion pressure, were stored on at the pre-collicular level and the rostral part to the transection as magnetic tape for off-line analysis. Phrenicus nerve (PN) dis- well as the cerebellum was removed by suction. Following a charges were measured over periods of 1 min before and at second transection through the thorax below the level of the several different time points after the application of the drugs. Ϯ diaphragm, the thorax–brainstem preparation was isolated and Statistical analysis (mean S.D.) was done using the Windows placed in an ice-chilled bath of artificial cerebrospinal fluid (ACSF) Excel programme. To calculate the statistical significance of changes before and after drug application the paired t-test was that contained: 125 mM NaCl, 24 mM NaHCO3, 5 mM KCl, used and results are presented as P-values. 2.5 mM CaCl2, 1.25 mM MgSO4, 1.25 mM KH2PO4, 1.1 mM glucose. The ACSF had an osmolarity of 290 mosmol and was gassed with 95% O2/5% CO2 (pH was adjusted to 7.4 with HCl or RESULTS NaOH). The preparation was transferred to a recording chamber where the descending abdominal aorta was quickly (within 7–10 Results were obtained from 69 adult wild-type mice and 13 min) cannulated for retrograde perfusion with ACSF containing mutant oscillator mice of an age of P18–P22, when they 2.1 g dextran/100 ml (total osmolarity of 300–320 mosmol), have lost the mature ␣1, ␤ type glycine receptors, whilst gassed with either 8% (ACSF pH 7.3) or 12% CO2 (ACSF pH 7.2) the embryonic ␣2 and the ␣3, ␤ receptor subtypes are no in oxygen and maintained at a temperature of 31 °C. The perfus- ate was driven by a peristaltic pump and re-circulated through longer expressed (Kling et al., 1997). filters and bubble traps. The perfusion pressure was measured Effects of the selective glycine receptor blocker strych- continuously by a transducer at the level of the descending ab- nine were analysed in 53 neurones of wild type mice and dominal aorta and maintained at 50–60 mm Hg by adjusting the nine neurones of oscillator mice. flow rate. Vecuronium bromide 0.04 ␮gmlϪ1 (Norcuron, Organon Teknika, Germany) was added to the perfusate to maintain mus- Blockade of glycinergic inhibition induced multiple cle paralysis. burst discharges in PN output Phrenic nerves (PN) were prepared for recordings by a dorsal approach. In wild type mice, inspiratory PN burst discharges had a typical augmenting pattern with peak activity during late Electrophysiological analysis periods of inspiration (control in Fig. 1A, B). Strychnine Neural discharges from the PN were recorded with glass suction application provoked several changes: electrodes to monitor central respiratory activities. The neural 1) PN burst discharges occurred at shorter and more activities were amplified (2000–10,000ϫ), band-pass filtered irregular intervals. Mean burst frequency changed from (100–5000 Hz), monitored on an oscilloscope (Tektronix, Beaver- 14.5Ϯ4.6 PN bursts/min to 24.5Ϯ7.5 discharges/min ton, OR, USA) and registered as raw discharge activity or in (Nϭ11; P-valueϽ0.01). integrated form (␶ϭ0.1 s) on a strip chart recorder (Yokogawa, Japan). 2) Individual burst discharges augmented more rapidly Intracellular recordings from medullary respiratory neurones reaching an early maximum of activity (Figs. 1A, B; 2A; were obtained with an npi amplifier (Tamm, Germany) using fine- 3A). tipped glass micropipettes (tip size approximately 0.5 ␮m) filled 3) Secondary “doublet burst” discharges following a with 2 M potassium methylsulfate (ICN Biomedicals GmbH, Ger- preceding burst at short intervals (within 0.2–2 s) were Ј Ј many) and 5 mM 1,2-bis[2-aminophenoxy]ethane-N,N,N ,N -tet- observed in 28.8% (17 of 59) of the preparations (Figs. 1A; raacetic acid (Sigma-Aldrich Chemie GmbH, Germany). Series 3). In these cases, the frequency of phrenic burst dis- resistances as measured in extracellular fluid ranged between 40 Ϯ and 80 M⍀. Respiratory neurones were recorded in the ventral charges was almost doubled (29 8 discharges/min). respiratory group of the medulla oblongata and functionally iden- 4) In other 22% (13 of 59) of the preparations, PN tified by the periodic fluctuations of their membrane potential and discharges were fractionated into multiple bursts, each of their discharge in synchrony with PN activity. According to the short duration (Fig. 1B, C). fairly ventral position of recording sites, we have presumably 5) In 10.2% (6 of 59) of the preparations, a prolonged excluded cranial motoneurones. The membrane potential was PN discharge was followed by multiple brief bursts at high monitored on-line on an oscilloscope and on a strip chart recorder. frequencies of 4–6 Hz (Figs. 1C, 4A). Drug solutions These changes of the output activity of the respiratory network indicate that glycine receptor blockade provokes Drugs were applied to the perfusate solution. Strychnine hydro- marked disturbances of burst generation and patterning chloride (Sigma, Taufuirchen, Germany), a selective glycine re- not only during the inspiratory phase, but throughout the ceptor antagonist, was applied in 0.03–0.3 ␮M concentrations to ensure receptor specificity (Jonas et al., 1998). The “small-con- respiratory cycle. ductance” SKCa apamin (Sigma) was applied in a concentration of 8 nM (Pennefather et al., 1985; Hill et al., 1992) Blockade of glycinergic inhibition induced multiple and the “big-conductance” BKCa channel blocker charybdotoxin bursting and voltage oscillations in medullary (CTX, Alomone, Jerusalem, Israel) was applied in a concentration respiratory neurones of 3.3 nM (Egan et al., 1993). Both drugs were dissolved in ACSF with bovine serum albumin (Sigma), 0.03 g/100 ml instead of Intracellular recordings revealed severe disturbances in dextran. Control solutions containing bovine serum albumin with- bursting behaviour of various types of medullary neurones, D. Bu¨sselberg et al. / Neuroscience 122 (2003) 831–841 833

Fig. 1. Characteristic changes of burst discharges in PN of wild type mice under control conditions and after blockade of glycine receptors with strychnine: A) Prolongation of inspiratory bursts and secondary bursts after application of strychnine. B) Double or triple bursts occur when the glycine receptor is blocked. Traces on the right panel demonstrate the integrated PN activity. C) Fast voltage oscillations during the post-inspiratory period after application of strychnine. i.e. inspiratory (Nϭ39), post-inspiratory (Nϭ10) and expi- Changes in inspiratory neurones ratory (Nϭ14) neurones after glycine receptor blockade, Under control conditions, inspiratory neurones exhibited whilst respiratory rhythm generation was preserved. an augmenting pattern of membrane depolarisation and

Fig. 2. PN discharges in relation to changes of membrane voltages in an inspiratory (A) and an expiratory (B) neuron. A) Double and triple bursts occur after the application of strychnine. B) Burst discharges in an augmenting expiratory neuron after glycine receptor blockade. Control, showing maximal synaptic inhibition during early inspiration and a slowly augmenting discharge of action potentials with small after-hyperpolarisations during the expiratory interval. During glycine receptor blockade with strychnine (0.3 ␮M), PN was silent for a short period of time during inspiration (arrow). ECG, electrocardiographic activity; PN, PN discharge. PN and ECG are recorded simultaneously. 834 D. Bu¨sselberg et al. / Neuroscience 122 (2003) 831–841

Fig. 3. Blockade of glycinergic inhibition revealed double bursting. A/B) Potentiation of burst-related membrane depolarisations and prominent after-burst hyperpolarisation that were followed by extra-burst (arrows) depolarisations in two different inspiratory neurones. Upper traces: membrane potentials; ECG, electrocardiographic activity; PN: PN activity. action potential discharges which lasted throughout In the hyperpolarisation normally occurring during post-inspira- presence of strychnine, several notable changes in the tion disappeared. (iv) Clustered waves of depolarizing po- spontaneous variations of the membrane potential and tentials occurred at 4–6 Hz (Fig. 4A) in association with action potential discharge occurred consistently. (i) In- doublet bursts in PN output activity (Figs. 2A; 3A, B; 4A) spiratory membrane depolarisation became significantly and sometimes (Nϭ4) even in the absence of PN dis- steeper, whilst the duration from the resting membrane charges (Fig. 4B).Strychnine blockade of glycine receptors potential to the peak of depolarisation decreased from enhanced inspiratory depolarisations in most inspiratory 0.5Ϯ0.12sto0.27Ϯ0.04 s (P-valueϽ0.01), the amplitude neurones (Nϭ30 of 39; Fig. 3A, B). The strongly aug- of the depolarisation increased from 17.1Ϯ1.2 mV to mented inspiratory depolarisation was followed by a pro- 20.1Ϯ1.4 mV (P-value PϽ0.01). The action potential dis- nounced after-burst hyperpolarisation, their amplitudes charges became more intense and changed from changing from 1.6Ϯ0.1 mV under control conditions to 10.7Ϯ2.9 action potentials per bursts to 26.3Ϯ4.3 action 16.5Ϯ5.4 mV after the application of strychnine (P-val- potentials per burst (P-valueϽ0.01) whilst the frequency ueϽ0.01), whilst the after-burst hyperpolarisation after ex- increased from 32.2Ϯ7.8–58.4Ϯ8.1 action potentials per tra-bursts was 21.5Ϯ6.3 mV (P-value compared with the second (P-valueϽ0.01). (ii) The duration of the membrane strychnine data: 0.08; Nϭ8). These enhanced after-burst depolarisation was reduced to roughly half compared with hyperpolarisations brought the membrane potential below control (from 0.92Ϯ0.4 s to 0.54Ϯ0.2 s; Nϭ5; P-val- pre-inspiratory, i.e. even below inhibitory synaptic levels ueϭ0.06). (iii) The steepness of membrane repolarisation (Fig. 3A) and were frequently followed by an interposed following inspiratory bursts was slowed and the voltage “doublet-burst” within the same cycle phase, i.e. without D. Bu¨sselberg et al. / Neuroscience 122 (2003) 831–841 835

Fig. 4. Fast oscillations occur in membrane voltage as well as in PN activity in different types of respiratory neurones. A) Blockade of post-inspiratory membrane potential hyperpolarisation and fast voltage oscillations during the post-inspiratory period in an inspiratory neuron. B) An inspiratory neuron with repetitive depolarization waves during the post-inspiratory period that were accompanied by brief discharges of PN activity. (Note different time scales!) C) During strychnine application a post-inspiratory neuron started to depolarise during inspiration and rapid voltage oscillations in synchrony with PN activity occurred during the inter-burst interval. preceding expiratory interval (Fig. 3B, arrow).Under con- expiratory intervals. This was accompanied by 4–6 Hz trol conditions, inspiratory neurones (Nϭ9 of 39) dis- oscillations in PN discharges (Fig. 4A). charged only a short series of fast action potentials at the very onset of inspiratory depolarisation (Fig. 3B). Changes in expiratory neurones After strychnine application, the neurons changed this “adaptive” firing pattern and discharged action potentials During control, augmenting expiratory neurones were hy- even throughout part of the expiratory interval (Fig. 3B, perpolarised to maximal voltages during early inspiration, right side) until they were finally stopped shortly before and significant synaptic inhibition outlasted inspiration into onset of the next inspiratory phase.In several inspiratory post-inspiration. Therefore, expiratory neurones started to neurones (Nϭ5 from 39), fast sub-threshold oscillations discharge only gradually after end of postinspiration. Ter- of membrane depolarisations appeared repeatedly after mination of membrane depolarisation and expiratory dis- the end of inspiratory bursts and continued into the charge occurred rapidly during late-expiration and onset of 836 D. Bu¨sselberg et al. / Neuroscience 122 (2003) 831–841

Fig. 5. Pentobarbital induced central apnoea when glycinergic inhibition is absent. A) Action of pentobarbital and strychnine in wild type mice. Upper trace: PN discharges before treatment. Middle trace: PN discharges were reduced after administration of pentobarbital sodium (Nembutal; 6 mg/kg); Lower trace: PN discharges were blocked by additional application of strychnine, 0.3 ␮M. B) Pentobarbital applied in oscillator mice induces apnoea. PN discharges (Fig. 2B). After blockade of glycine recep- charges were greatly altered. The initial post-inspiratory tors, the voltage fluctuations attained a step-like pattern, burst discharges were intensified and followed by a sec- because post-inspiratory inhibition was absent, whilst syn- ondary burst-depolarisation. At the same time, PN activity aptic hyperpolarisation persisted during the inspiratory was depressed (not shown). phase (Fig. 2B). Previously augmenting expiratory neu- Another characteristic feature that appeared after glycine rones therefore depolarised rapidly from strongly hyperpo- receptor blockade in post-inspiratory neurones (Nϭ8of10) larised inspiratory levels to maximal expiratory depolarisa- were rapid voltage oscillations combined with action potential tion. The duration of expiratory depolarisation did not discharges in synchrony with high-frequency PN discharges change (2.1Ϯ0.8 s versus 2.1Ϯ1.1 s; Nϭ6) after applica- (Fig. 4C). In such cases, the silent period characteristic of the tion of strychnine. An important observation made in sev- expiratory interval was missing in PN activity. eral cases (Nϭ3) was that inspiratory inhibition did not reveal clear fragmentations, although strychnine blockade Glycine receptor blockade abolished the respiratory of glycinergic inhibition had induced doublet bursting in PN rhythm in pentobarbital-anaesthetized and mutant discharge (Fig. 2B). oscillator mice The data described above are in apparent conflict with a Changes in post-inspiratory neurones previous study, which demonstrated that in the pentobar- Blockade of glycine receptors also dramatically changed bital-anaesthetized cat, strychnine injected into the pre- the discharge of post-inspiratory neurones. Under control Bo¨tzinger complex, blocked rhythmic respiratory activity conditions, post-inspiratory neurones were hyperpolarised (Pierrefiche et al., 1998). Since a major difference in the to maximal voltage levels at the beginning of the inspira- experimental approach of the two studies was anesthe- tory phase (Fig. 4C). After blockade of glycine receptors, (i) sia, we tested the effect of pentobarbital after glycine membrane depolarisation and onset of action potential receptors were blocked with strychnine (Nϭ15). Under discharges were shifted into the inspiratory phase (see control conditions small doses of pentobarbital sodium Bu¨sselberg et al., 2001a). (ii) The periodic discharge was (6 mg/kg) produced only a modest slowing of phrenic intensified from 35Ϯ12 action potentials per depolarisation burst frequency from 14.5Ϯ4.3 to 13.6Ϯ4.1 per minute to 87Ϯ14 action potentials per depolarisation (P-val- (Nϭ5). The striking finding, however, was that after ueϽ0.01) and the membrane depolarisation was pro- additional administration of strychnine (0.3 ␮M) all rhyth- longed into the inspiratory phase. (iii) In four (out of 10) mic PN activity disappeared within 2–5 min (Fig. 5). The neurones the slowly decaying “post-inspiratory” burst dis- same neural apneic effect was seen when the sequence D. Bu¨sselberg et al. / Neuroscience 122 (2003) 831–841 837

Fig. 6. Blockade of the SKCa by apamin (8 nM) does not change the respiratory rhythm after blockade of glycine receptors in wild type mice. Blockade of glycine receptors induced irregular and extra-bursting of the expiratory neuron. PN recordings revealed the described irregularities of the rhythmic burst discharges. The discharge characteristics remained unchanged and the respiratory rhythm was not disturbed further after additional blockade of the SKCa with apamin. of drug application was reversed (Nϭ3), strychnine be- between the two studies. Administration of the same ing applied before pentobarbital. Such findings revealed small dose of pentobarbital to oscillator mice (spdot/ a depressant action of pentobarbital on the bursting spdot; Nϭ3) also resulted in a loss of rhythmic PN activity mechanisms of neurones and resolved the discrepancy within short times (Fig. 5B). 838 D. Bu¨sselberg et al. / Neuroscience 122 (2003) 831–841

Fig. 7. Blockade of the BKCa by CTX (3.3 nM) abolished the respiratory rhythm when glycinergic inhibition is blocked in wild type mice (A–F) and oscillator mice (G–I). Elimination of PN burst-discharges and rhythmic activity after blocking calcium-activated potassium channels in wild type mice during glycine receptor blockade (A–E). Note the reduction in burst amplitude and the appearance of tonic PN activity between bursts after application of CTX (C1,D1). The post-inspiratory decay of PN activity (F) was relatively short under control conditions (upper left panel), but became progressively longer when strychnine was applied (upper right panel). Tonic expiratory discharges occupied the whole expiratory interval after application of CTX until phrenic bursts finally disappeared whilst tonic discharges persisted. In oscillator mice with defective glycine receptors (G), apamin prolonged PN burst discharges, but did not abolish the rhythm (H). Blockade of the rhythm occurred only after the additional blockade of big-conductance KCa (I).

Disruption of the respiratory rhythm after blockade charges that normally declined quite readily after strych- of calcium-activated potassium channels nine (from 0.18 sϮ0.02 s to 0.29 sϮ0.05 s) was slowed when BK channels were blocked with CTX (Fig. 7F) To test the possibility that after-burst hyperpolarisations Ca (to 0.58 sϮ0.11 s after 20 min CTX application and to preceding doublet bursting after glycine receptor blockade Ϯ 2ϩ ϩ 0.74 s 0.18 after 80 min of CTX application). The were linked to activation of Ca -dependent K currents, changes progressed with time until all rhythmic burst we examined the effects of blocking the small-conduc- 2ϩ ϩ discharges disappeared, whilst enhanced tonic dis- tance of Ca -dependent K channels (SKCa) with apamin charges persisted (Fig. 7E, I). and the BKCa with CTX. CTX given without blockade of glycine receptors only ϭ Apamin administration (N 7) after strychnine block- led to a minor increase in phrenic burst frequency (Nϭ5; ade of glycine receptors in wild type mice did not abolish data not shown). respiratory activity (Fig. 6). The intensity of both PN burst In oscillator mice at P 21 (Fig. 7G), respiratory burst discharges and cellular discharges gradually declined and amplitudes were also significantly reduced from became obvious after 15–45 min, but there was no obvi- 24.1Ϯ1.8 to 14.4Ϯ1.2. Simultaneously burst frequency ous change in burst characteristics (Fig. 6). The shape of increased from 21Ϯ5.9 burst/min to 29.8Ϯ8.3 bursts/ single action potentials and after-hyperpolarisations also min 15 min after blocking SKCa channels with apamin remained unchanged (not illustrated). (Nϭ6; Fig. 7H). However, rhythmic PN activity was com- Strikingly different was the response to blockade of pletely abolished when BK channels (Fig. 7I) were ϭ Ca BKCa with CTX (N 6; Fig. 7A–F, I). CTX developed its blocked by CTX (Nϭ9). maximum effect after a fairly long delay (more than 45 min), apparently because it only slowly passed through the blood–brain barrier possibly do to its high molecular DISCUSSION weight. However, there were gradual changes in post- The “Conditional Network-Burster Theory” (Richter et inspiratory discharges and subsequently all rhythmic PN al., 1986, 2000; Richter, 1996; Richter and Spyer, 2001) discharges were abolished in the strychnine treated wild postulates that synaptic inhibition via glycinergic and/or type animals. The typical changes preceding elimination GABAergic synapses (Champagnat et al., 1982; Haji et of rhythmic discharges were a decline in the amplitude al., 1992, 2000; Schmid et al., 1996) is important for and a prolongation of individual PN bursts (Fig. 7C, D). normal respiratory rhythm generation in the in vivo ma- The post-inspiratory decay of inspiratory burst dis- ture mammal (Pierrefiche et al., 1998). Since it is known, D. Bu¨sselberg et al. / Neuroscience 122 (2003) 831–841 839 that suppress respiratory rhythm (Robson glycine receptors (Kling et al., 1997) and in the in vitro et al., 1963), we used the non-anaesthetized in situ neonatal brainstem after blockade of glycine receptors perfused brainstem preparations (Paton, 1996) to spe- (Onimaru et al., 1987, 1989; Feldman et al., 1990). There cifically test the contribution of glycinergic inhibition by are two possible explanations for these differences: (i) using low concentrations of strychnine (in the range of respiratory neurones have the potential for endogenous 0.03–0.3 ␮M) in order not to affect at the same time bursting, which is normally overwritten by inhibitory synap- GABAA receptors (Jonas et al., 1998) and provoke ep- tic control and concomitant conductance changes (Richter ileptiform activities as verified by the absence of drug- and Spyer, 2001); (ii) endogenous bursting depends on induced discharges in non-respiratory nerves of the bra- specific membrane conductances, which are suppressed chial plexus (Bu¨sselberg et al., 2001a). The study re- by barbiturates. Such sensitivity was indeed vealed that failure of glycinergic inhibition induces verified for fast Naf channels (Rehberg and Duch, 1999), multiple forms of rhythmic discharges and voltage oscil- rectifying Kϩ channels (Gibbons et al., 1996), pre- and lations in all three types of medullary respiratory neu- postsynaptic high voltage-activated Ca2ϩ channels (Werz rones. Whilst these changes of firing behaviour could and Macdonald, 1985), the BKCa channels (Ishida et al., also be explained by changes in the efficacy of synaptic 1999) as well as GABAA receptor currents (Saunders and interactions, our interpretation is that medullary respira- Ho, 1990). We assume that this explains why endogenous tory neurones developed an intrinsic tendency to cellular respiratory bursting is attenuated in anaesthetized bursting when they are released from inhibitory synaptic preparations. control through glycine receptors. We speculated there- fore that Ca2ϩ-dependent Kϩ currents, activated by pre- Mechanisms of double bursts and fast oscillations 2ϩ ceding Ca currents (Mironov and Richter, 1998; Elsen Several changes in inspiratory activity developed when and Ramirez, 1998; Pierrefiche et al., 1999), are involved glycinergic synapses were blocked: (i) respiratory move- in termination of rhythmic bursting. A consequence there- ments producing long-lasting bursts were doubled; (ii) fore should be loss of rhythmic respiratory bursting, when membrane hyperpolarisation of inspiratory neurones start- “big” conductance BKCa channels were blocked with CTX. ing with onset of the post-inspiratory phase was slowed; This was indeed the case; rhythmic respiratory activity and (iii) all types of neurones developed a preference for disappeared prevailing irregular tonic discharges. There- voltage oscillations at a frequency of 4–6 Hz. All these fore we suggest that the mature in vivo respiratory network changes are consistent with the view that glycinergic inhi- is capable of changing between two operational modes: bition is required for effective late- and post-inspiratory Synaptic inhibitory interaction between respiratory neu- “off-switching” of respiratory phase activities (Richter et al., rones seems to be important for respiratory rhythm gener- 1979; Richter, 1982; Schmid et al., 1996). ation and stabilisation under normal conditions (Richter et The various types of respiratory neurones tested devel- al., 1986; Richter and Spyer, 2001). However, whenever oped a tendency to generate multiple burst discharges, which glycinergic inhibition is diminished, mature respiratory neu- often were accompanied by corresponding changes in PN rones may enter a different operational state in which output activity. The responses of individual neurones resem- intrinsic bursting becomes increasingly important, as is bled the general pattern described for endogenous bursting described for neonatal neurones (Smith et al., 1991; of respiratory neurones in the in vitro slice preparation (John- Ramirez and Richter, 1996). son et al., 1994). In our study, “additional” bursts were initi- ated by a pronounced membrane hyperpolarisation following Persistence of respiratory rhythm generation after enhanced preceding bursts. We assume that such post-burst glycine receptor blockade membrane hyperpolarisation originated from KCa currents A series of previous studies (Pierrefiche et al., 1998; Bu¨s- (Mifflin et al., 1985; Richter et al., 1993; Pierrefiche et al., selberg et al., 2001a) revealed that inspiratory, post-in- 1995), which obviously are competent to re-activate inacti- spiratory and stage two expiratory neurones receive gly- vating voltage-regulated membrane conductances to trigger cinergic inhibition during the inspiratory phase. In addition, a secondary burst. The obvious test therefore was to specif- there is glycinergic inhibition during the post-inspiratory ically block BKCa channels. Such blockade abolished all phase in inspiratory and stage two expiratory neurones rhythmic discharges, inspiratory bursts becoming longer and (Champagnat et al., 1982; Schmid et al., 1996; Haji et al., smaller, whilst tonic discharge activity persisted. Therefore, 1992). The importance of glycinergic inhibition became activation of BKCa channels seems to be essential for main- evident when microinjections of strychnine into the pre- tenance of repetitive firing during bursts and burst termination Bo¨tzinger complex of anaesthetized cats eliminated rhyth- in glycine receptor-deficient mice (Mironov, 1983; Bennett et mic respiratory activity (Pierrefiche et al., 1998). The most al., 2000; Shao et al., 1999; Golding et al., 1999; Pedarzani et reasonable explanation was that such neural apnoea re- al., 2000). sulted from removal of glycinergic synaptic interactions Therapeutic implications between the different types of respiratory neurones in the pre-Bo¨tzinger complex (Connelly et al., 1992; Schwar- Under mature in vivo conditions, periodic breathing re- zacher et al., 1995). Conversely however, respiratory quires not only a persistent excitatory synaptic drive, but rhythm persisted in non-anaesthetized mature mutant os- also mutual inhibitory control by glycinergic and GABAer- cillator mice (Bu¨sselberg et al., 2001b) that lack functional gic synaptic inhibition. Our findings suggest that endoge- 840 D. Bu¨sselberg et al. / Neuroscience 122 (2003) 831–841 nous bursting behaviour of respiratory neurones are sup- Kling C, Koch M, Saul B, Becker CM (1997) The frameshift mutation pressed by synaptic inhibition, but might be released, oscillator (Glra1(spd-ot)) produces a complete loss of glycine re- whenever inhibitory synaptic control declines or is sup- ceptor alpha1-polypeptide in mouse central nervous system. Neu- pressed by pathologic processes. Patients suffering from roscience 78:411–417. Mifflin SW, Ballantyne D, Backmann SB, Richter DW (1985) Evidence hyperekplexia lack functional glycine receptors. In these for a calcium activated potassium conductance in medullary respi- patients, respiratory rhythm might depend on a certain ratory neurones. In: Neurogenesis of central respiratory rhythm endogenous bursting capacity of respiratory neurones. (Bianchi AL, Denavit-Saubie M, eds), pp 179–182. Lancaster: MTP The high sensitivity of these bursting mechanisms to pen- Press. tobarbital calls for careful monitoring and cautious admin- Mironov SL (1983) Theoretical model of slow-wave membrane poten- istration of barbiturates when they are necessary in the tial oscillations in molluscan neurons. Neuroscience 10:899–905. 2ϩ treatment of hyperekplexic patients. Mironov SL, Richter DW (1998) L-type Ca channels in inspiratory neurones of mice and their modulation by hypoxia. J Physiol (Lond) 512:75–87. Acknowledgements—Supported by DFG. Onimaru H, Arata A, Homma I (1989) Firing properties of respiratory rhythm generating neurons in the absence of synaptic transmission REFERENCES in rat medulla in vitro. Exp Brain Res 76:530–536. Paton JF (1996) A working heart-brainstem preparation of the mouse. Bu¨sselberg D, Bischoff AM, Richter DW (2000) Blockade of KCa abol- ishes respiratory rhythm after loss of glycinergic inhibition. Soc J Neurosci Methods 65:63–68. Neurosci Abstr 26:p926. Pedarzani P, Kulik A, Mu¨ller M, Ballanyi K, Stocker M (2000) Molecular ϩ ϩ Bu¨sselberg D, Bischoff AM, Paton JF, Richter DW (2001a) Reorgani- determinants of Ca2 -dependent K channel function in rat dorsal sation of respiratory network activity after loss of glycinergic inhibi- vagal neurones. J Physiol 527(2):283–290. tion. Pflu¨gers Arch 441:444–449. Pennefather P, Lancaster B, Adams PR, Nicoll RA (1985) Two distinct ϩ ϩ Bu¨sselberg D, Bischoff AM, Becker K, Becker CM, Richter DW Ca2 -dependent K currents in bullfrog sympathetic ganglion cells. (2001b) The respiratory rhythm in mutant oscillator mice. Neurosci Proc Nat Acad Sci USA 82:3040–3044. Lett 316:99–102. Pierrefiche O, Champagnat J, Richter DW (1995) Calcium-dependent Champagnat J, Denavit-Saubie M, Moyanova S, Rondouin G (1982) conductances control neurones involved in termination of inspira- Involvement of amino acids in periodic inhibitions of bulbar respi- tion in cats. Neurosci Lett 184:101–104. ratory neurones. Brain Res 237:351–365. Pierrefiche O, Schwarzacher SW, Bischoff AM, Richter DW (1998) Connelly CA, Dobbins EG, Feldman JL (1992) Pre-Bo¨tzinger complex Blockade of synaptic inhibition within the pre-Bo¨tzinger complex in in cats: respiratory neuronal discharge patterns. Brain Res 590: the cat suppresses respiratory rhythm generation in vivo. J Physiol 337–340. (Lond) 509:245–254. Egan TM, Dagan D, Levitan IB (1993) Properties and modulation of a Pierrefiche O, Haji A, Bischoff A, Richter DW (1999) Calcium currents calcium-activated potassium channel in rat olfactory bulb neurons. in respiratory neurons of the cat in vivo. Pflu¨gers Arch 438:817– J Neurophysiol 69:1433–1442. 826. Elsen FP, Ramirez JM (1998) Calcium currents of rhythmic neurones Ramirez JM, Richter DW (1996) The neuronal mechanisms of respi- recorded in the isolated respiratory network of neonatal mice. ratory rhythm generation. Curr Opin Neurobiol 6:817–825. J Neurosci 18:10652–10662. Rehberg B, Duch DS (1999) Suppression of central nervous system Feldman JL, Smith JC, Ellenberger HH, Connelly CA, Liu GS, Greer sodium channels by . Anesthesiology 91:512–520. JJ, Lindsay AD, Otto MR (1990) Neurogenesis of respiratory rhythm Richter DW (1982) Generation and maintenance of the respiratory and pattern: emerging concepts. Am J Physiol 259:879–886. rhythm. J Exp Biol 100:193–107. Gibbons SJ, Nunez-Hernandez R, Maze G, Harrison NL (1996) Inhi- Richter DW (1996) Neural regulation of respiration: rhythmogenesis bition of a fast inwardly rectifying potassium conductance by bar- and afferent control. In: Comprehensive human physiology: from biturates. Anesth Analg 82:1242–1246. cellular mechanisms to integration, Vol. 2 (Greger R, Windhorst U, Golding NL, Jung HY, Mikus T, Spruston N (1999) Dendritic calcium eds), pp 2079–2095. Berlin: Springer Verlag. spike initiation and repolarization are controlled by distinct potas- Richter DW, Ballantyne D, Remmers JE (1986a) How is the respiratory sium channel subtypes in CA1 pyramidal neurons. J Neurosci rhythm generated? a model. News Physiol Sci 1:109–112. 19:8789–8798. Richter DW, Camerer H, Meesmann M, Ro¨hrig N (1979) Studies on Haji A, Takeda R, Remmers JE (1992) Evidence that glycine and the synaptic interconnection between bulbar respiratory neurones GABA mediate postsynaptic inhibition of bulbar respiratory neurons of cats. Pflu¨gers Arch 380:245–257. in the cat. J Appl Physiol 73:2333–2342. Haji A, Takeda R, Okazaki M (2000) Neuropharmacology of control of Richter DW, Champagnat J, Jacquin T, Benacka R (1993) Calcium respiratory rhythm and pattern in mature mammals. Pharmacol and calcium-dependent potassium currents in medullary respiratory Ther 86:277–304. neurones. J Physiol (Lond.) 470:23–33. Hill R, Matsushima T, Schotland J, Grillner S (1992) Apamin blocks the Richter DW, Mironov SL, Bu¨sselberg D, Lalley P, Bischoff AM, Wilken slow AHP in lamprey and delays termination of locomotor bursts. B (2000) Respiratory rhythm generation: plasticity of a neuronal Neuroreport 3:943–945. network. Neuroscientist 6:188–205. Ishida K, Kinoshita H, Kobayashi S, Sakabe T (1999) Thiopentone Richter DW, Spyer KM (2001) Studying rhythmogenesis of breathing: inhibits endothelium-dependent relaxations of rat aortas regulated comparison of in vivo and in vitro models. Trends Neurosci 24:464– by endothelial Ca2ϩ-dependent Kϩ channels. Eur J Pharmacol 472. 371:179–185. Robson JG, Housley MA, Solis-Quiroga OH (1963) The mechanism of Johnson SM, Smith JC, Funk GD, Feldman JL (1994) Pacemaker respiratory arrest with sodium pentobarbital and . behaviour of respiratory neurons in medullary slices from neonatal Ann NY Acad Sci 109:494. rat. J Neurophysiol 72:2598–2608. Saunders PA, Ho IK (1990) Barbiturates and the GABAA receptor Jonas P, Bischofberger J, Sandku¨hler J (1998) Corelease of two fast complex. Prog Drug Res 34:261–286. neurotransmitters at a central synapse. Science 281:419–424. Schmid K, Foutz AS, Denavit-Saubie M (1996) Inhibitions mediated by Kissin I, Mason JO, Vinik HR, McDanal J, Bradley EL (1987) Barbitu- glycine and GABAA receptors shape the discharge pattern of bul- rates inhibit stress-induced analgesia. Can J Anaesth 34:146–151. bar respiratory neurons. Brain Res 710:150–160. D. Bu¨sselberg et al. / Neuroscience 122 (2003) 831–841 841

Schwarzacher SW, Smith JC, Richter DW (1995) Pre-Bo¨tzinger com- differential roles of glycinergic and GABAergic neural transmission. plex in the cat. J Neurophysiol 73:1452–1459. J Neurophysiol 77:1853–1860. Shao LR, Halvorsrud R, Borg-Graham L, Storm JF (1999) The role of Smith JC, Ellenberger HH, Ballanyi K, Richter DW, Feldman JL (1991) BK-type Ca2ϩ-dependent Kϩ channels in spike broadening during Pre-Bo¨tzinger complex: a brainstem region that may generate re- repetitive firing in rat hippocampal pyramidal cells. J Physiol 521 (Pt spiratory rhythm in mammals. Science 254:726–729. 1):134–146. Werz MA, Macdonald RL (1985) Barbiturates decrease voltage-de- Shao XM, Feldman JL (1997) Respiratory rhythm generation and pendent calcium conductance of mouse neurons in dissociated cell synaptic inhibition of expiratory neurons in pre-Bo¨tzinger complex: culture. Mol Pharmacol 28:269–277.

(Accepted 31 July 2003)