Corresponding author J. L. Feldman: Department of Neurobiology, Box 951763, University of California, Los Angeles, CA 90095-1763, USA. Email: email@example.com
1We investigated the role of α-amino-3-hydroxy-5-methylisoxazole-4-propionate (AMPA) receptors and their regulation in affecting respiratory-related neurones in a neonatal rat medullary slice that spontaneously generates respiratory-related rhythm and motor output in the hypoglossal (XII) nerve.
2Bath application of the AMPA receptor antagonist 1-(4-aminophenyl)-4-methyl-7,8-methylenedioxy-5H-2,3-benzodiazepine (GYKI) completely blocked XII nerve activity, as well as respiratory-related synaptic drives in neurones within the preBötzinger Complex (preBötC), site of rhythm generation in the slice.
3Local application of GYKI to the preBötC blocked respiratory rhythm. Local application of AMPA to the preBötC increased rhythm frequency and depolarized respiratory-related neurones.
4In the presence of tetrodotoxin (TTX), GYKI completely blocked the inward current induced by local application of AMPA, but not that induced by kainate.
5Local application of okadaic acid, a membrane-permeable inhibitor of phosphatase 1 and 2A, to the preBötC increased the frequency of respiratory motor discharge.
6Intracellular application of microcystin, a membrane-impermeable inhibitor of phosphatase 1 and 2A, enhanced endogenous inspiratory drive and exogenous AMPA-induced current (in the presence of TTX) in preBötC inspiratory neurones. Both the enhanced inspiratory drive and the increased AMPA-induced current were completely blocked by GYKI.
7We suggest that AMPA receptor activation and AMPA receptor modulation by phosphorylation are crucial for the rhythm generation within the preBötC.
Experiments were performed on a medullary slice preparation spontaneously generating respiratory pattern (Smith et al. 1991). Briefly, Sprague-Dawley neonatal rats (age, 0-3 days) were anaesthetized with ether, decerebrated, and the neuraxis was isolated. The en bloc brainstem-spinal cord was pinned down with the ventral surface facing upwards and mounted in the specimen vice of a Vibratome (VT 1000, Technical Products International, St Louis, MO, USA) oriented vertically (rostral end upwards). It was then sectioned serially in the transverse plane until the landmarks at the rostral boundary of the preBötC, i.e. nucleus ambiguus and inferior olive, were visible. One transverse slice (600-750 μm thick) containing the preBötC was cut. The slice was transferred to a recording chamber (8-10 ml volume), and pinned down on a Sylgard elastomer. In experiments where rapid bath application of drug was desirable (e.g. fast blocking of synaptic transmission by TTX), a 0.9 ml recording chamber was used instead. The standard bath solution for dissection and slicing contained (mm): 128 NaCl, 3.0 KCl, 1.5 CaCl2, 1.0 MgSO4, 23.5 NaHCO3, 0.5 NaH2PO4, 30 glucose and 1 ascorbic acid, bubbled with 95 % O2-5 % CO2 at room temperature (22-24°C). During electrophysiological recording, the slice was continuously superfused (10-20 ml min−1) with the standard solution with increased KCl (9 mm), and was recycled into a 200 ml reservoir equilibrated with 95 % O2-5 % CO2 at 27 ± 0.5°C. Nerve activity was recorded from the cut end of the XII nerve with a suction electrode.
Whole-cell recordings from preBötC respiratory neurones
Blind whole-cell patch-clamp recordings were made from respiratory neurones in the preBötC (Smith et al. 1991). Patch electrodes were pulled from borosilicate glass (Model P-87, Sutter Instruments) with a tip size of 1-1.5 μm (4-6.5 MΩ). Electrodes were filled with a solution containing (mm): 120 potassium gluconate, 5.0 NaCl, 1.0 CaCl2, 10 BAPTA (tetra-K+ salt) and 2.0 ATP (Mg2+ salt); pH adjusted to 7.3 using KOH. Intracellular signals were amplified (Axopatch-1D, Axon Instruments). Electrodes were mounted on a microdrive and positioned over the preBötC. Positive pressure (150-180 mmHg) was applied to the electrodes as they were advanced 100 μm below the surface of the slice. The positive pressure was then released and the electrode advanced in 1 μm steps until a rapid increase in electrode resistance was observed. A slight negative pressure was applied to form a gigaohm (> 1 GΩ) seal. Cells were then ruptured using short-duration negative pressure pulses. Whole-cell capacitance was compensated and series resistance was compensated 60-90 %. Cells with large, unstable access resistance (> 40 MΩ), indicating inadequate membrane rupture, were not used. The junction potential between bath solution and pipette filling solution was corrected in the values reported in this study.
In experiments where GYKI was bath applied, one neurone was studied per slice because incomplete washout of GYKI prevented re-establishment of control conditions. In local application experiments, single- or multi-barrel pressure ejection pipettes (tip diameter, 6-8 μm per barrel: pressure, 50-70 kPa) were positioned over or beneath the rostral surface of the preBötC. GYKI or okadaic acid was microinjected into the tissue, while AMPA or kainate was perfused over the slice to avoid perturbing whole-cell recording in the same area. Injection volumes were determined by measuring the displacement of the pipette fluid meniscus with a microscope equipped with a calibrated eyepiece reticule. In perfusion experiments, the tip of the application pipette was first painted with ink to make it clearly visible when advancing towards the slice. After gently touching the slice surface under visual inspection, the pipette was quickly withdrawn by about 50 μm so that the ejection would be close to the surface while minimizing pressure artefacts. Slices were oriented so that the direction of superfusate flowed from dorsal to ventral to minimize spread of drugs to the XII motor nucleus or other neuronal populations dorsal to the preBötC. The buffer solution for each drug was applied as control. In experiments where the change in AMPA-induced current was monitored, only neurones within 160 μm of the slice surface were used, in order to ensure effective and consistent perfusion. All agonist-induced currents were allowed to return to the steady state for at least 15 s before another local application.
Microcystin was applied intracellularly via dialysis from the patch-clamp recording pipette solution. Microcystin stock solution was dissolved in DMSO or ethanol. Okadaic acid stock solution was prepared using minimal DMSO. The concentration of DMSO (v/v) in okadaic acid (OA) solutions was: 100 μm OA, 1.6 %; 30 μm OA, 0.48 %; 10 μm OA, 0.16 %; 250 nm OA, 0.004 %. GYKI (GYKI 52466), microcystin-LR, AMPA-HBr and kainate were obtained from RBI. Okadaic acid was obtained from Calbiochem. All other drugs, including TTX and components of the recording pipette solution, bath and buffer solution were obtained from Sigma.
Signals were recorded on tape. Selected segments of whole-cell recording data were filtered and digitized at 1-2 kHz per channel and analysed using Axoscope software (Axon Instruments). To quantify the change in inspiratory drive or AMPA-induced current due to phosphatase inhibition, currents were integrated using pCLAMP software (Axon Instruments). Zero baseline for integration was set at the steady-state membrane current. For each neurone, the duration of integration was set to enclose the longest inspiratory drive or AMPA-induced current. This duration was then fixed for integration of all the inspiratory drives or AMPA-induced currents for this neurone. Periods of the respiratory rhythm at the XII nerve were obtained using AxoData and AxoGraph software (Axon Instruments).
The concentration-dependent effect of okadaic acid on respiratory rhythm period was analysed using non-linear regression. The average normalized ratio of maximal enhancement in inspiratory drive charge transfer and AMPA-induced charge transfer for inspiratory neurones with microcystin were analysed using Student's unpaired t test.
GYKI blocks endogenous inspiratory drive and exogenous AMPA-induced current
To determine the effect of endogenous AMPA receptor activation on respiratory motor pattern, GYKI was bath applied, where it could affect glutamate transmission among respiratory rhythm-generating neurones, premotoneurones and motoneurones. GYKI (≥ 50 μm; n= 7) completely blocked XII nerve discharge, as well as the rhythmic activity of both inspiratory (I) and expiratory (E) neurones (Fig. 1A and B). Spontaneous excitatory postsynaptic potentials (sEPSPs) in these neurones were also blocked. Complete blockade took ∼12-15 min with 50 μm GYKI (n= 4). Higher concentrations of GYKI (e.g. 100 μm; n= 3) shortened the blocking time to ∼10 min. After 35-40 min washout time, the rhythm in both XII nerve and respiratory neurones partially recovered (n= 3; data not shown). In some slices, GYKI decreased respiratory frequency before it decreased the amplitude of the integrated XII nerve discharge; in others, the effect was reversed.
GYKI (2.0 mm, n= 3; or 3.8 mm, n= 2), when microinjected within the preBötC, either ipsilateral or contralateral to the recorded XII nerve, reversibly blocked XII nerve discharge (Fig. 2). Control injections with GYKI buffer solution had no blocking effect on the motor output (n= 3; data not shown).
When we locally perfused AMPA (100 μm, 1 or 2 s) over the rostral surface of the preBötC, both inspiratory (Fig. 3A, top) and expiratory (Fig. 3A, middle) neurones (n= 11) depolarized (12-25 mV), and the respiratory frequency in the XII nerve increased by 44-217 % (Fig. 3A, bottom; n= 11). Local application of kainate to the preBötC also depolarized respiratory neurones (data not shown). In order to identify the effects of GYKI on AMPA- or kainate-induced currents, TTX (0.5 μm) was bath applied. Subsequently, AMPA (100 μm) or kainate (1.0 mm) was locally applied to the preBötC, before and after bath application of GYKI (100 μm, 12-15 min). Both AMPA and kainate caused inward currents in respiratory neurones (Fig. 3B;n= 25 for AMPA, n= 4 for kainate). GYKI completely blocked the AMPA-induced currents (Fig. 3B, top; n= 5); while it only partially reduced the kainate-induced currents (Fig. 3B, bottom; n= 3).
Phosphatase inhibition within the preBötC increased the frequency of respiratory rhythm
To determine the effect of phosphatase inhibition on the rhythm generation network, we locally applied okadaic acid to the preBötC. In some experiments (n= 3), (+)-5-methyl-10,11-dihydro-5H-dibenzo[a,d]cyclohepten-5,10-imin maleate (MK-801; 100 μm) was added to the bath before application of okadaic acid to control for the possible effect of activation of NMDA receptors; since the effect of okadaic acid was comparable with or without MK-801, we pooled the data (Fig. 4). Okadaic acid induced a concentration-dependent increase in frequency of the XII nerve activity (Fig. 4E).
Phosphatase inhibition enhanced endogenous inspiratory drive and exogenous AMPA-induced current in inspiratory neurones
To determine the effect of phosphatase inhibition on respiratory neurones, microcystin was applied intracellularly. Microcystin (50 μm) increased the duration, amplitude, or both, of inspiratory drive in inspiratory neurones (n= 10); the amplitude of sEPSPs/sEPSCs also increased (Fig. 5). Lower concentrations of microcystin were less effective (data not shown). The microcystin-enhanced inspiratory drive potential or current, and sEPSPs were completely blocked by bath application of GYKI (50 μm, Fig. 5). Because the time course of inspiratory drive current varied amongst inspiratory neurones, and the extent of change in amplitude or duration due to microcystin varied, the integral (instead of peak or decay time constant) of the drive current, i.e. charge transfer, was used as an index for this enhancement. Microcystin enhanced the endogenous inspiratory drive (P < 0.05, Student's t test). The ratio of maximal enhancement in charge transfer (with respect to the state at the beginning of each recording), when averaged amongst inspiratory neurones dialysed with microcystin (Vh= -60 mV; n= 5; 175 ± 14 %, mean ±s.e.m.), was significant in comparison with the control inspiratory neurones (Vh= -60 mV; n= 4; 69 ± 25 %) (Fig. 5E). The enhancement in inspiratory drive due to microcystin was transient, with a maximum increase that occurred from 8 to 30 min after whole-cell break-in. Recordings from different parts of individual neurones (e.g. soma vs. proximal dendrite) may have contributed to this variability. In Fig. 5D, we give an example comparison of the time-dependent change in charge transfer for the inspiratory drive current of a representative microcystin-dialysed inspiratory neurone to that of a representative control inspiratory neurone, i.e. recorded with exactly the same pipette solution except for microcystin. The decline in the relative charge transfer under control conditions may be due to the effects of intracellular dialysis by the control pipette solution.
Slight changes in steady-state membrane potential/current were observed in two out of the ten microcystin-dialysed inspiratory neurones. One neurone hyperpolarized shortly (3 min) after whole-cell break-in, but the steady-state membrane potential stabilized 1-2 min later; the other inspiratory neurone slightly depolarized during the enhancement of inspiratory drive.
To establish the effect of phosphorylation on exogenous AMPA-induced currents, TTX (1 μm) was bath applied immediately upon obtaining successful patch recording of an inspiratory neurone (Vh= -60 mV; n= 7). After TTX had completely abolished synaptic activity, AMPA (100 μm) was locally applied to the preBötC. The ratio of maximal enhancement of AMPA-induced charge transfer (with respect to the state at the beginning of each recording) was averaged amongst inspiratory neurones dialysed with microcystin (n= 4; 187 ± 27 %, mean ±s.e.m.), and amongst the control inspiratory neurones (n= 3; 68 ± 12 %) (Fig. 6C). Microcystin enhanced the exogenous AMPA-induced current (P < 0.05, Student's t test). Figure 6A shows the current induced by exogenous AMPA (1 s) at various times after membrane rupture in a typical microcystin-dialysed inspiratory neurone; the enhanced AMPA-induced current was subsequently blocked by bath application of GYKI. The time-dependent change in AMPA-induced charge transfer/peak current for this inspiratory neurone was compared with that of a representative control inspiratory neurone (Fig. 6B). Similar to the endogenous response, the effect of microcystin was transient, and the time to maximum increase in AMPA-induced current varied amongst the inspiratory neurones.
Among the microcystin-dialysed inspiratory neurones studied in the presence of TTX, one exhibited a small inward current during the period when AMPA-induced current increased; another inspiratory neurone transiently exhibited an inward current, which stabilized at the original level (-70 pA). Enhancement of AMPA-induced current in this neurone was observed after steady-state membrane current was stabilized.
In order to compare directly the time course of change in charge transfers of endogenous inspiratory drive and exogenous AMPA-induced current, both currents were monitored simultaneously in several microcystin-dialysed inspiratory neurones (Fig. 6D;n= 4) without TTX. Phosphatase inhibition by microcystin caused nearly concurrent enhancement in these currents (Fig. 6D).
In inspiratory neurones dialysed with microcystin, input conductance increased (n= 7) along with the enhanced AMPA-induced currents; it then decreased when the AMPA-induced charge transfer passed its maximum. Since no significant simultaneous change in steady-state membrane current/potential was observed, this conductance change might reflect enhancement of opening states or numbers of AMPA receptors (or other receptors/channels) affecting steady-state excitability (see Discussion).
Effect of phosphatase inhibition on expiratory neurones
Intracellular microcystin (50 μm) enhanced the amplitude of sEPSPs/sEPSCs in expiratory neurones (n= 9). Figure 7A compares the sEPSCs of a microcystin-dialysed expiratory neurone at 6 and 11.5 min after whole-cell break-in. In the presence of TTX, intracellular microcystin also increased the local AMPA-induced current in expiratory neurones (n= 7; Fig. 7B). Similar to inspiratory neurones, the enhancement in sEPSCs and AMPA-induced current (without TTX) in microcystin-dialysed expiratory neurones occurred nearly concurrently (n= 3; Fig. 7C). The above observations did not occur in the control expiratory neurones (n= 5; data not shown).
Activated AMPA receptors and their regulation play an important role in neonatal respiratory rhythm generation and modulation in vitro. Blocking AMPA receptors with GYKI in the preBötC, the hypothesized site of respiratory rhythm generation (Smith et al. 1991), completely abolished respiratory motor discharge and rhythmic drives in respiratory neurones, while local application of AMPA to the preBötC increased respiratory frequency and depolarized respiratory neurones. GYKI also completely blocked the inward current induced by AMPA but not by kainate in preBötC respiratory neurones. Moreover, phosphatase inhibition within the preBötC modulated the rhythm-generating network by increasing the frequency of respiratory-related discharge. Intracellular inhibition of the same phosphatases enhanced inspiratory synaptic drive and exogenous AMPA-induced current in preBötC inspiratory neurones; both currents were completely blocked by GYKI.
AMPA receptors primarily mediate inspiratory drives in inspiratory neurones
Unlike 6-cyano-7-dinitroquinoxaline-2,3-dione (CNQX) or 6,7-dinitroquinoxaline-2,3-dione (DNQX), the blocking effect of GYKI is not influenced by phasic changes in glutamate concentration, nor is its effect reduced by increasing agonist, e.g. AMPA, concentrations (Donevan & Rogawski, 1993; Zorumski et al. 1993; Paternain et al. 1995; Wilding & Huettner, 1995). The non-competitive nature of GYKI has made it useful (Paternain et al. 1995) for differentiating currents mediated by AMPA or kainate receptors. In addition, because GYKI completely blocks AMPA receptor-mediated currents, including the current induced by kainate, but minimally affects kainate-induced current through kainate receptors, it can uncover the functional role played by kainate receptors (Paternain et al. 1995).
We compared endogenous respiratory drive in preBötC respiratory neurones and the current induced by exogenous AMPA or kainate, before and after GYKI application. GYKI completely blocked the endogenous respiratory drive, sEPSPs and exogenous AMPA-induced current, but not the exogenous kainate-induced current in preBötC respiratory neurones. Therefore, kainate receptors, while present in the respiratory neurones, are unlikely to be the principal mediator of excitatory synaptic transmission in respiratory rhythmogenesis, since both the respiratory-modulated synaptic currents and rhythmic XII nerve motor discharge were completely blocked by bath or local application of GYKI. Thus, in the neural network that generates respiratory rhythm, AMPA, but not kainate, receptors appear to mediate primarily inspiratory drive and sEPSPs, and are essential for the rhythm and pattern generation. It is not surprising that glutamate, the ubiquitous fast excitatory neurotransmitter in mammalian brain, mediates respiratory drive (Greer et al. 1991), but why should AMPA receptors be the principal transducer? For reliable neurotransmission under a wide range of ventilatory demands and neuronal excitability essential for breathing, AMPA receptors are suitable. Moreover, we hypothesize that the possibility that AMPA receptors in these neurones can be modulated by other inputs affecting their phosphorylation state provides them with properties appropriate for their vital homeostatic function. Further illumination of the specific properties of AMPA receptors in respiratory neurones and their particular roles in the central networks controlling breathing will be necessary to test this hypothesis.
The burst frequency of the XII nerve respiratory discharge increased in a concentration-dependent manner when okadaic acid was applied to the preBötC. This network effect reflects the altered states of, and thus interactions among, its constituent neurones, including inspiratory and expiratory neurones. Enhanced phosphorylation may also have induced network reconfiguration.
In inspiratory and expiratory neurones of the preBötC, intracellular microcystin increased excitability by enhancing sEPSPs and the current induced by exogenous AMPA. Of most interest with respect to generation of respiratory pattern is that intracellular microcystin enhanced the endogenous inspiratory drive currents in inspiratory neurones, which were presumably mediated by AMPA receptors (see above). We propose that these effects were due to alteration of the phosphorylation state of one or more proteins that specifically affect AMPA receptor function in preBötC respiratory neurones; the target proteins could be subunits of the AMPA receptor itself. We did not pursue the mechanism underlying enhanced respiratory drive or the AMPA-induced currents due to phosphatase inhibition. Possible mechanisms include: prolonged channel opening time (Greengard et al. 1991; Lieberman & Mody, 1994); increased probability of opening (Greengard et al. 1991; Wang, Orser, Brautigan & MacDonald, 1994b); or additional recruitment of functional postsynaptic receptors (Isaac, Nicoll & Malenka, 1995; Liao, Hessler & Malinow, 1995; Durand, Kovalchuk & Konnerth, 1996).
The phosphorylation state of other receptors and/or voltage-gated channels might also be affected by okadaic acid or microcystin. However, while there was significant enhancement of endogenous inspiratory drives or exogenous AMPA-induced currents in microcystin-dialysed inspiratory neurones, most neurones exhibited an increase in conductance without an associated change in steady-state membrane current. The increase in conductance may be due to an enhanced opening state of tonically activated AMPA receptors, or other receptors/channels.
We used an elevated K+ concentration in the bath solution to maintain a robust respiratory-related rhythm in the slice. We recognize that this situation might have caused an increase in inward Ca2+ currents due to the depolarization, and thus affected the activities of Ca2+-CaMKII, PKA or PKC. However, since both control and test observations were made under similar conditions, our results indicate that phosphatase inhibition can still significantly enhance respiratory neuronal excitability.
Since the slice preparation we used for these experiments generates an integrative nervous system behaviour, i.e. respiratory-related motor outflow, interpretation of results in the context of function is possible. The functional relevance of the present results is considerable, suggesting that respiratory-modulated excitatory postsynaptic currents, mediated primarily by AMPA receptors, can be altered by phosphorylation and dephosphorylation of these receptors or associated proteins.
We hypothesize that AMPA receptor function is modulated in respiratory neurones to regulate the net inward inspiratory or expiratory current that is produced for a given amount of presynaptic inspiratory or expiratory activity, ultimately to affect the strength and timing of respiratory muscle contractions that produce breathing movements appropriate for blood gas homeostasis. This modulation could play a role in adjusting ventilation to changes in respiratory demand, such as during exercise, hypoxia (due to disease or altitude), hypercapnia, hyperthermia and sleep. We further hypothesize that this modulation is achieved, at least in part, by phosphorylation and dephosphorylation of AMPA receptors.
In conclusion, we have shown that AMPA receptor activation is crucial for neonatal respiratory rhythm generation. We have also provided evidence of the modulation of synaptic efficacy by phosphatase inhibition in the neural network generating breathing rhythm. Control of the phosphorylation state of AMPA receptor subunits or associated proteins may be an essential component of the mechanisms regulating respiration, and a target site of various neuromodulators, such as serotonin, noradrenaline and various peptides that affect breathing.
This work was supported by a grant (HL40959) from the National Institutes of Health. The authors particularly thank Dr Xuesi M. Shao for his support, technical help, and constructive discussions throughout the project, and our colleagues at the Systems Neurobiology Laboratory for critical comments on the manuscript. Q. G. thanks Dr Steven M. Johnson for teaching the slice preparation technique.