- Top of page
The role of AMPA receptors (AMPARs) in generation and propagation of respiratory rhythm is well documented both in vivo and in vitro, whereas the functional significance of NMDA receptors (NMDARs) in preBötzinger complex (preBötC) neurons has not been explored. Here we examined the interactions between AMPARs and NMDARs during spontaneous respiratory rhythm generation in slices from neonatal rats in vitro. We tested the hypothesis that activation of NMDARs can drive respiratory rhythm in the absence of other excitatory drives. Blockade of NMDARs with dizocilpine hydrogen maleate (MK-801, 20 μm) had a negligible effect on respiratory rhythm and pattern under standard conditions in vitro, whereas blockade of AMPARs with NBQX (0.5 μm) completely abolished respiratory activity. Removal of extracellular Mg2+ to relieve the voltage-dependent block of NMDARs maintained respiratory rhythm without a significant effect on period, even in the presence of high NBQX concentrations (≤ 100 μm). Removal of Mg2+ increased inspiratory-modulated inward current peak (II) and charge (QI) in preBötC neurons voltage-clamped at −60 mV by 245% and 309%, respectively, with respect to basal values. We conclude that the normal AMPAR-mediated postsynaptic current underlying respiratory drive can be replaced by NMDAR-mediated postsynaptic current when the voltage-dependent Mg2+ block is removed. Under this condition, respiratory-related frequency is unaffected by changes in II, suggesting that the two can be independently regulated.
Glutamate is the major fast excitatory neurotransmitter underlying respiratory rhythm generation. AMPAR and NMDAR antagonist microinjections in vivo in adult mammals suggest a synergistic role of these receptors in the transmission of inspiratory drive to motoneurons. Microinjection of either the AMPAR antagonist 2,3-dihydroxy-6-nitro-7-sulfamoyl-benzo(f)quinoxalline (NBQX) or the NMDAR antagonist d(–)-2-amino-7-phosphonoheptanoic acid (AP-7) into the phrenic motor nucleus decreases the amplitude of phrenic nerve bursts in rats. However, simultaneous blockade of both receptors decreases the amplitude in a synergistic way (Chitravanshi & Sapru, 1996). Although NMDARs and AMPARs coexist in respiratory rhythm generation-related areas, only AMPAR-mediated transmission is essential for rhythm generation and propagation both in vivo in adult rat (Connelly et al. 1992; Chitravanshi & Sapru, 1996) and cat (Anderson & Speck, 1999), and in vitro in neonatal rat (Greer et al. 1991; Funk et al. 1993). Moreover, in vitro preparations from neonatal mutant mice lacking the NMDAR1 subunit generate a rhythm that is indistinguishable from that obtained from neonatal wild-type mice, demonstrating that NMDARs are not essential for respiratory rhythm generation or drive transmission in the neonate (Funk et al. 1997). However, this does not mean that NMDARs are superfluous. In in vitro preparations generating respiratory rhythm, while the NMDAR antagonist dizocilpine hydrogen maleate (MK-801) does not perturb spontaneous respiratory burst frequency, bath application of NMDA produces a dose-dependent increase in respiratory frequency (Greer et al. 1991; Funk et al. 1993). Furthermore, in vivo, anaesthetized cats breathe normally after systemic administration of MK-801, but subsequent vagotomy produces apneusis (Foutz et al. 1988, 1989; Feldman et al. 1992).
Complicating our understanding of the contribution of NMDARs to rhythm generation is its voltage dependence: at resting membrane potentials (≤−60 mV), currents through activated NMDARs are substantially attenuated by Mg2+ in physiological concentrations (0.8–1.2 mm), but as the membrane depolarizes, the Mg2+ blockade is relieved. Here we examined the effects of removing NMDAR blockade during perturbations of AMPAR-mediated transmission on rhythmic activity of preBötC neurons and integrated hypoglossal nerve (∫XIIn) activity in a neonatal rat medullary slice preparation. We found that under conditions where the voltage-dependent Mg2+ block of NMDARs is relieved, substantial currents can pass through NMDARs sufficient to drive the rhythm when AMPARs are blocked. Moreover, even though in the absence of Mg2+ there is a 4-fold increase on preBötC neuron inspiratory-modulated inward current peak (II), respiratory frequency remained unaffected. We show that the NMDAR can substitute, after removing its voltage-dependent block due to Mg2+, for the AMPAR glutamatergic transmission normally underlying respiratory pattern generation in the in vitro slice preparation.
- Top of page
Under control conditions in respiratory rhythmic slices from neonatal rats, NMDARs in preBötC neurons appear activated during inspiration but do not pass significant amounts of current due to their voltage-dependent block. However, NMDARs can significantly contribute to or even solely drive the respiratory rhythm in these slices in situations where the voltage-dependent block is substantially attenuated.
NMDARs are not required for prenatal development of respiratory networks (Funk et al. 1997). In medullary slice or en bloc brainstem–spinal cord preparations from neonatal rodents under baseline conditions, NMDARs are not required for generation of respiratory rhythm or motor output, yet exogenous application of NMDA produces a robust response (Greer et al. 1991; Funk et al. 1993, 1997). Furthermore, removal of extracellular Mg2+ enhances inspiratory currents in preBötC neurons, suggesting that endogenous NMDAR activation can enhance the discharge normally due to currents through AMPARs (Pierrefiche et al. 1991).
We suggest that during each inspiratory cycle only a small fraction of preBötC neuron NMDARs are sufficiently depolarized to remove the Mg2+ block. How big a depolarization is needed to remove the Mg2+ block? That depends on NMDAR subunit composition. NR1/2A (composed of NR1 and NR2A subunits) and NR1/2B receptors are more strongly inhibited at hyperpolarized potentials by Mg2+ than NR1/2C or NR1/2D receptors (Kuner & Schoepfer, 1996). In recombinant systems, inclusion of NR3A with NR1 and NR2 in heteromultimeric channels reduces the sensitivity to Mg2+ block and results in a smaller unitary conductance than NR1/NR2 channels. Consistent with that, in NR3A-deficient mice, NMDA-evoked currents of cortical neurons are larger than in wild-type littermates (Das et al. 1998; Sasaki et al. 2002).
Single-cell RT-PCR analysis reveals that NR2A, NR2B and NR2D are expressed in similar amounts in preBötC neurons, XII motoneurons and neurons from the nucleus of the solitary tract (NTS) in young rats, whereas NR3A is expressed in all preBötC neurons but only in one-third of XII motoneurons and NTS neurons (Paarmann et al. 2005). We suggest that under control conditions, the depolarization achieved during the initial phase of each inspiratory burst relieves the Mg2+ block of NR3A-containing receptors, which are less sensitive to the voltage-dependent blockade. Since these receptors have a small unitary conductance, their contribution to QI is not significant. Depolarizing a neuron or removing the extracellular Mg2+ unmasks the fraction of NMDARs that under control conditions are presumably active, i.e. bound by glutamate, but not passing current. This fraction must contain NR1/NR2A and NR1/NR2B receptors that are strongly blocked near resting membrane potentials by Mg2+ at physiological concentrations.
An interesting observation is that inhibition of glutamate uptake enhances both the AMPAR- and the NMDAR-mediated components of IGlu in preBötC neurons voltage-clamped at −60 mV. We suggest that increased glutamate accumulation in the synapse increases the AMPAR-mediated depolarization of preBötC neurons sufficiently to remove the voltage-dependent block of NMDARs (MacDonald & Nowak, 1990) allowing them to pass current.
In the absence of Mg2+, NBQX concentrations higher than 1 μm decreased ∫XIIn amplitude while II peak and period remain virtually unchanged. We suggest that the AMPAR/NMDAR ratio is higher in XIIn motoneurons than in preBötC neurons, making them more sensitive to AMPA blockade.
Though in standard control conditions NMDARs are not the major charge carrier for II, the Ca2+ influx they provide could contribute to activation of much larger inward currents such as the Ca2+-activated mixed cationic current (ICAN) that is present in all preBötC neurons and are hypothesized to be an important intrinsic burst-generating current (Rekling & Feldman, 1998; Pena et al. 2004; Del Negro et al. 2005).
Implications for respiratory rhythm generation
The group-pacemaker hypothesis posits that preBötC inspiratory neurons mutually interconnected by glutamatergic synapses initiate inspiration by generating a population burst of activity arising from a recurrent network with positive feedback (Rekling & Feldman, 1998). Here we show that, in the absence of AMPAR-mediated synaptic transmission, NMDARs can provide the excitatory drive necessary to initiate and propagate the inspiratory burst. Under our experimental conditions, NMDARs in preBötC neurons can replace AMPARs in mediating excitatory interactions since: (i) Both receptors are coexpressed in preBötC neurons (at present we do not know if they have similar somatodendritic or synaptic distribution; Paarmann et al. 2000); (ii) NMDAR current slope in the absence of Mg2+ is similar to that of AMPARs; and (iii) NMDAR activation can recruit a set of intrinsic conductances resembling those activated by AMPARs, and also induce pacemaker-like membrane potential oscillations (Grillner & Wallen, 1985). These currents are not identical, however. A striking difference is that NMDARs are permeable to Ca2+ whereas in preBötC neurons, AMPARs almost exclusively contain R-edited GluR2 (Paarmann et al. 2000) and are mainly permeable to Na+ (authors' unpublished observations) but not Ca2+. This would suggest that Ca2+ entry through synaptic receptors is not playing a critical role in rhythmogenesis, which is of considerable interest because the Ca2+ buffering of these neurons appears to be limited (Alheid et al. 2002). We suggest that the major source of Ca2+ entry in preBötC neurons is through voltage-gated Ca2+ channels activated during action potentials (C. Morgado-Valle and J. L. Feldman, unpublished data).
Removal of extracellular Mg2+ increased the II and the ∫XIIn amplitude but slightly decreased the period. In contrast, lowering the ACSF [K+] to 3 mm in the absence of Mg2+ increased the II and ∫XIIn amplitude but almost tripled the period. Thus, an increase in II does not necessarily result in an increase in frequency. This would suggest that II affects the amplitude of the motor output whereas frequency depends on the level of excitability of preBötC neurons.
Our findings may have clinical relevance. Patients with hyperventilation syndrome (HVS), a breathing pattern disorder characterized by bouts of inappropriately high ventilation associate with elevated frequency, often have significant hypomagnesaemia (Fehlinger & Seidel, 1985; Durlach et al. 1997). Patients with Rett's syndrome also have frequent episodes of hyperventilation; supplemental dietary Mg2+ ameliorates these episodes (Egger et al. 1992). Our work suggests a potential causal link since decreased levels of Mg2+ could increase respiratory output (inappropriately) that is driven by the preBötC by mechanisms described above.
In summary, NMDARs are not necessary for respiratory rhythmogenesis under standard in vitro conditions. However, they can, after removing their voltage-dependent block due to Mg2+, substitute for AMPAR-mediated glutamatergic transmission normally underlying respiratory pattern generation (at least in our experimental conditions). Moreover since II can be modulated independent of frequency and recurrent excitation is necessary for rhythm generation, network connectivity is an essential element underlying respiratory rhythmogenesis. Since Mg2+ levels can affect neuronal plasticity (Slutsky et al. 2004), an additional role of Mg2+ in breathing could be to modulate respiratory plasticity (Feldman et al. 2003), essential for adaptation of breathing to changing demands.