Volume 593, Issue 1 p. 111-125
Research Paper
Open Access

Frequency-dependent facilitation of synaptic throughput via postsynaptic NMDA receptors in the nucleus of the solitary tract

Huan Zhao

Huan Zhao

Program in Neuroscience, Department of Integrative Physiology and Neuroscience, Washington State University, Pullman, WA, 99164 USA

Search for more papers by this author
James H. Peters

James H. Peters

Program in Neuroscience, Department of Integrative Physiology and Neuroscience, Washington State University, Pullman, WA, 99164 USA

Search for more papers by this author
Mingyan Zhu

Mingyan Zhu

Program in Neuroscience, Department of Integrative Physiology and Neuroscience, Washington State University, Pullman, WA, 99164 USA

Search for more papers by this author
Stephen J. Page

Stephen J. Page

Program in Neuroscience, Department of Integrative Physiology and Neuroscience, Washington State University, Pullman, WA, 99164 USA

Search for more papers by this author
Robert C. Ritter

Robert C. Ritter

Program in Neuroscience, Department of Integrative Physiology and Neuroscience, Washington State University, Pullman, WA, 99164 USA

Search for more papers by this author
Suzanne M. Appleyard

Corresponding Author

Suzanne M. Appleyard

Program in Neuroscience, Department of Integrative Physiology and Neuroscience, Washington State University, Pullman, WA, 99164 USA

Corresponding author S. M. Appleyard: Department of IPN, Program in Neuroscience, Washington State University, 100 Dairy Road, Pullman, WA 99164, USA.Email: [email protected]Search for more papers by this author
First published: 29 September 2014
Citations: 19

Key points

  • Hindbrain NMDA receptors play important roles in reflexive and behavioural responses to vagal activation.

  • NMDA receptors have also been shown to contribute to the synaptic responses of neurons in the nucleus of the solitary tract (NTS), but their exact role remains unclear.

  • In this study we used whole cell patch-clamping techniques in rat horizontal brain slice to investigate the role of NMDA receptors in the fidelity of transmission across solitary tract afferent–NTS neuron synapses.

  • Results show that NMDA receptors contribute up to 70% of the charge transferred across the synapse at high (>5 Hz) firing rates, but have little contribution at lower firing frequencies. Results also show that NMDA receptors critically contribute to the fidelity of transmission across these synapses during high frequency (>5 Hz) afferent discharge rates.

  • This novel role of NMDA receptors may explain in part how primary visceral afferents, including vagal afferents, can maintain fidelity of transmission across a broad range of firing frequencies.

Neurons within the nucleus of the solitary tract (NTS) receive vagal afferent innervations that initiate gastrointestinal and cardiovascular reflexes. Glutamate is the fast excitatory neurotransmitter released in the NTS by vagal afferents, which arrive there via the solitary tract (ST). ST stimulation elicits excitatory postsynaptic currents (EPSCs) in NTS neurons mediated by both AMPA- and NMDA-type glutamate receptors (-Rs). Vagal afferents exhibit a high probability of vesicle release and exhibit robust frequency-dependent depression due to presynaptic vesicle depletion. Nonetheless, synaptic throughput is maintained even at high frequencies of afferent activation. Here we test the hypothesis that postsynaptic NMDA-Rs are essential in maintaining throughput across ST–NTS synapses. Using patch clamp electrophysiology in horizontal brainstem slices, we found that NMDA-Rs, including NR2B subtypes, carry up to 70% of the charge transferred across the synapse during high frequency stimulations (>5 Hz). In contrast, their relative contribution to the ST-EPSC is much less during low (<2 Hz) frequency stimulations. Afferent-driven activation of NMDA-Rs produces a sustained depolarization during high, but not low, frequencies of stimulation as a result of relatively slow decay kinetics. Hence, NMDA-Rs are critical for maintaining action potential generation at high firing rates. These results demonstrate a novel role for NMDA-Rs enabling a high probability of release synapse to maintain the fidelity of synaptic transmission during high frequency firing when glutamate release and AMPA-R responses are reduced. They also suggest why NMDA-Rs are critical for responses that may depend on high rates of afferent discharge.

Abbreviations

  • AP
  • action potential
  • APV
  • (2R)-amino-5-phosphonovaleric acid
  • CC
  • current clamp
  • DCP
  • d-CPP-ene
  • EPSC
  • excitatory postsynaptic current
  • NBQX
  • 2,3-dihydroxy-6-nitro-7-sulfamoyl-benzo[f]quinoxaline-2,3-dione
  • NTS
  • nucleus of the solitary tract
  • -R
  • receptor
  • TBOA
  • DL-threo-β-benzyloxyaspartic acid
  • ST
  • solitary tract
  • ST-EPSC
  • ST-evoked EPSC
  • Introduction

    Vagal afferent neurons make excitatory glutamatergic synapses with second-order neurons in the nucleus of the solitary tract (NTS) and relay information from the viscera to the brain (Andresen & Kunze, 1994; Moran et al. 2001; Berthoud, 2008; Browning & Travagli, 2011). Depolarization of vagal afferent terminals releases glutamate with an intrinsically high probability of release (Pr) (Talman et al. 1980; Bailey et al. 2006; Peters et al. 2008; Baude et al. 2009). Glutamate acts at both 2-amino-3-(5-methyl-3-oxo-1,2-oxazol-4-yl)propanoic acid (AMPA)- and N-methyl-d-aspartate (NMDA)-type ionotropic glutamate receptors (Aicher et al. 1999; Lachamp et al. 2003; Antunes et al. 2004; Almado & Machado, 2005; Mutolo et al. 2007; Sartor & Verberne, 2007; Balland et al. 2008; Liu et al. 2012; Marques-Lopes et al. 2012). Electrophysiological experiments have clearly established that AMPA-R activation following vagal afferent stimulation triggers fast excitatory postsynaptic currents (EPSCs) in NTS neurons (Andresen et al. 2012). Moreover, there is abundant evidence that AMPA-R EPSCs critically contribute to synaptic transmission and successful generation of action potentials (APs) in NTS neurons (Bailey et al. 2006; Appleyard et al. 2007; Peters et al. 2010).

    In contrast to our understanding of the role of AMPA-Rs in vagal afferent synaptic transmission, the contribution of NMDA-Rs to the fidelity of transmission across this synapse is not well understood. NMDA-Rs in the NTS appear to play an important role in some reflexes and behavioural responses in vivo (Treece et al. 1998; Hung et al. 2006; Porres et al. 2011; Wright et al. 2011; Budisantoso et al. 2012; Campos et al. 2012) but just how NMDA-Rs contribute to these responses to vagal activation is not clear. Application of NMDA activates a depolarizing current in a subpopulation of NTS neurons in both slice (de Paula et al. 2007) and dissociated preparations (Bonham & Chen, 2002). In addition, activation of NMDA-Rs increases glutamate release from NTS neurons projecting to the dorsal motor nucleus of the vagus (Bach & Smith, 2012). NMDA-Rs have also been reported to contribute to some ST-evoked EPSCs (ST-EPSCs) (Aylwin et al. 1997; Yen et al. 1999; Jin et al. 2003; Baptista et al. 2005; Baptista & Varanda, 2005; Balland et al. 2006, 2008). What is not known is how the NMDA-R component of the ST-EPSC contributes to the successful generation of APs in NTS neurons.

    Synaptic efficacy is maintained in the NTS over a broad range of vagal afferent discharge frequencies in spite of rapid diminishment of synaptic glutamate release due to vesicle depletion resulting in strong frequency-dependent depression (Chen et al. 1999; Doyle & Andresen, 2001). We hypothesize that postsynaptic NMDA-Rs, with their slow decay kinetics and relatively high affinity for glutamate (Patneau & Mayer, 1990; Lester & Jahr, 1992; McBain & Mayer, 1994), provide a mechanism to maintain charge transfer across ST–NTS synapses and compensate for use-dependent vesicle depletion. Such a mechanism could maintain synaptic fidelity from very low to high frequency firing and thereby support sustained vagally mediated responses. In the present study we test this hypothesis using patch-clamping techniques in a rat horizontal brain stem slice preparation.

    Methods

    Animal welfare assurances

    All animal procedures were conducted with the approval of the Institutional Animal Care and Use Committees at WSU and in accordance with the US Public Health Service Policy on Humane Care and Use of Laboratory Animals (PHS Policy) and the National Institutes of Health Guide for the Care and Use of Laboratory Animals (NIH Guide).

    NTS slice preparation

    Hindbrain slices were prepared from adult male Sprague–Dawley rats (7–10 weeks, 240–320 g). Rats were anaesthetized with isoflurane and then killed by thoracic compression as previously described (Doyle & Andresen, 2001; Peters et al. 2010). The hindbrain was removed and placed for 1 min in cold (0–4ºC) artificial cerebrospinal fluid (aCSF) composed of (in mm): 125 NaCl, 3 KCl, 1.2 KH2PO4, 1.2 MgSO4, 25 NaHCO3, 10 dextrose, 2 CaCl2, and bubbled with 95% O2–5% CO2. The final osmolarity was adjusted to 300–306 mosmol l–1 using dextrose. For Mg2+-free experiments MgSO4 was omitted from the aCSF (osmolarity was kept in the same range with extra dextrose). The medulla was trimmed and a wedge of tissue was removed from the ventral surface to align the solitary tract (ST) with the NTS in the same cutting plane when mounted in a vibrating microtome (Leica VT-1000S). Hindbrain slices (250 μm) were cut with a sapphire knife (Delaware Diamond Knives, Wilmington, DE, USA) and contained a long section of the ST. Slices were submerged in a perfusion chamber and all recordings were performed at a constant temperature (∼33 °C) and pH 7.4.

    Electrophysiology

    Recording electrodes, 3–5 MΩ, were filled with an internal solution composed of (in mm): 10 NaCl, 130 potassium gluconate, 11 EGTA, 1 CaCl2, 2 MgCl2, 10 Hepes, 2 NaATP, 0.2 NaGTP; pH 7.3; 297–301 mosmol l–1. For the depolarized ‘Flip’ protocol, where neurons were held at +40 mV in 5 μM gabazine, NaCl was replaced with CsCl, potassium gluconate was replaced with caesium methanesulfonate, and the internal solution was adjusted to pH 7.4 with CsOH. Neurons were recorded from NTS within 250 μm rostral or caudal from obex and medial to the ST. Whole cell patch clamp recordings were made with an Axopatch 700 B amplifier, Digidata 1440 A digitizer and pCLAMP 10 software (all from Molecular Devices, Sunnyvale, CA, USA). Solutions were preheated using a HPRE2 pre-heater (Cell Micro Controls, Norfolk, VA, USA) and bath temperature was monitored through a probe positioned next to the slice. Local applications of NMDA were made using a picospritzer (Parker Hannifin Corporation, Cleveland, OH, USA) via a delivery pipette (application pressure 1 p.s.i. with 5 s duration) similar to those used for patch clamp recordings. The tip of the delivery pipette was positioned within 25 μm from the soma of the recorded neuron. Only neurons with holding currents not exceeding ±50 pA (or 100 pA with Cs internals) at Vh = –60 mV for the 10 min control period (input resistance >150 MΩ) were studied further. Series resistance was monitored throughout the recordings and neurons were not included in further analysis if it exceeded 20 MΩ, or drifted >25%. Synaptic responses were evoked with an ultrafine concentric bipolar stimulating electrode (50 μm i.d., FHC Inc., Bowdoin, ME, USA) placed on the ST 1–3 mm from the recording electrode. Electrical stimuli were delivered from an isolated programmable stimulator (Isoflex stimulator with Master-8, AMPI, Jerusalem, Israel) triggered to deliver a burst of stimuli (2–50 Hz). In the ‘on-threshold’ protocol for the throughput experiments, a minimum stimulation intensity was chosen such that only one ST afferent was recruited and contributed to the EPSC; while in the ‘supra-threshold’ protocol maximum stimulation was applied to evoke a compound EPSC composed of 2–5 individual afferent inputs (McDougall et al. 2009; Peters et al. 2011). The average size of the sustained current was measured as the baseline immediately after the 5th shock where it normally reached a plateau, unless specifically stated. All drugs were obtained from Tocris Cookson (Ballwin, MO, USA) or Sigma (St Louis, MO, USA). All membrane potentials reported are not corrected for junction potentials.

    Statistics

    All data are presented as means ± SEM. Percentage inhibition was first calculated for each individual neuron then averaged across the same group. Significance of a drug effect was determined using one-way ANOVA. Statistical comparisons of drug effects between groups (e.g. aCSF and NMDA antagonist) were made using Student's t test or two-way ANOVA with Tukey's or Bonferroni post hoc analysis and Fisher's exact test where appropriate. P < 0.05 was considered significantly different.

    Results

    The majority of second-order NTS neurons express an NMDA-R-mediated current

    All of our recordings with ST-EPSCs were made in neurons that were identified as second-order NTS neurons by their jitter (or variability in the latency of ST-shock to ST-EPSCs) being less than 200 μs as defined previously (Doyle & Andresen, 2001). To determine the contribution of NMDA-Rs in solitary tract-evoked postsynaptic responses in these second-order NTS neurons, we performed voltage clamp recordings under conditions that allow NMDA-R currents to be measured. First, we used a depolarizing ‘flip’ protocol that has been used extensively to study the relative contributions of AMPA-Rs vs. NMDA-Rs (Baptista & Varanda, 2005; Balland et al. 2006, 2008). ST-evoked EPSCs (ST-EPSCs) were initially recorded at a holding potential of −60 mV (a potential at which NMDA-Rs are blocked by extracellular Mg2+), then again at +40 mV (a potential at which the Mg2+ block of NMDA-Rs is removed) in 5 μM gabazine. We observed two distinctive responses in subpopulations of NTS neurons (Fig. 1). At +40 mV, 9 of 14 neurons exhibited evoked currents comprising both a fast peak and a prolonged slow component. In those neurons, the slow component (measured as the size of the sustained current 100 ms from the start of the stimulation) was largely blocked by the NMDA-R antagonist (2R)-amino-5-phosphonovaleric acid (APV; 25 μm) (110.9 ± 35 pA reduced to 22.6 ± 7 pA; or an average inhibition of 69.7 ± 4.9%, P < 0.01, n = 9, paired Student's t test). The peak or the fast component was also moderately attenuated by APV (190.6 ± 27 pA reduced to 119.6 ± 27 pA; or an average inhibition of 35.8 ± 6.5%, P < 0.05, n = 9, paired Student's t test), but an APV-insensitive fast component remained that was blocked by the AMPA-R antagonist 2,3-dihydroxy-6-nitro-7-sulfamoyl-benzo[f]quinoxaline-2,3-dione (NBQX; 20 μm) (119.6 ± 27 pA vs. 16.1 ± 2.2 pA with NBQX; P < 0.01, n = 5, paired Student's t test, Fig. 1A). NMDA-R-mediated currents accounted for 74.4 ± 3.8% of the total charge transfer in this population (Fig. 1C). A second population of NTS neurons (5 of 14) lacked the slow component at +40 mV, and the ST-EPSCs were not significantly altered with the application of APV (160.2 ± 42.2 pA vs. 153.3 ± 38.3 pA, n = 5, not significant, P = 0.25, paired Student's t test), but were completely blocked by NBQX (10.1 ± 7.3 pA with NBQX; n = 5, P < 0.01, paired Student's t test, Fig. 1B and C). In cells examined from both populations the ST-EPSCs at –60 mV were almost entirely mediated by AMPA-Rs as they were blocked by the AMPA-R antagonist NBQX (20 μm) (a 90.0 ± 3.1% reduction; n = 6; P < 0.05, paired Student's t test), but insensitive to APV (a 3.1 ± 5.6% reduction; n = 6; not significant, P = 0.13; paired Student's t test, Fig. 1A).

    Details are in the caption following the image
    Figure 1. The majority of NTS neurons express a slow component of the ST-EPSC mediated by NMDA-Rs

    ST-EPSCs are recorded at both +40 mV and −60 mV. Representative traces are shown in A and B (baseline has been adjusted). In 14 neurons examined, 9 neurons exhibit a slow component in EPSC when flipped to +40 mV (NMDAR+; shown in A). This slow component is blocked with the application of APV (25 μm; dark grey), whereas the fast peak is blocked only with further application of NBQX (20 μm; light grey). The other 5 neurons lack the slow component (NMDAR–; shown in B); the EPSCs are not altered by APV, but are completely blocked by NBQX. Quantifications of the relative contribution of NBQX-sensitive (NBQX-Sens) and APV-sensitive (APV-Sens) components are shown in C. Peak amplitude (Peak) is measured within 20 ms from the shock. Sustained current (Sus) is measured at 100 ms from the shock. Total charge transfer (Area) is measured as the area under the trace 200 ms from the shock. For NMDAR-negative neurons, sustained current is below the detectable level and therefore not quantified.

    Direct activation of postsynaptic NMDA-R with local applications of NMDA

    To further demonstrate the presence of functional NMDA-Rs on NTS neurons, we locally applied NMDA (100 μm) using a picospritzer under Mg2+-free conditions and monitored the postsynaptic responses in the presence and absence of bath APV (50 μm). We found that the brief application of NMDA induced an inward current in all neurons tested (Fig. 2A, n = 11), which was almost completely abolished following bath application of APV (Fig. 2AC; n = 9; P < 0.001, one-way ANOVA). In contrast a second application of NMDA 20 min later in the absence of APV evoked a current similar in size to the first application (125.1 ± 63.2 pA vs. 124.2 ± 62.1 pA; n = 6; data not shown). Under our experimental conditions NMDA did not cause a significant change in the frequency of spontaneous EPSCs (sEPSCs; Fig. 2D, n = 6; P > 0.05, one-way ANOVA). The sEPSCs were blocked by application of the glutamate receptor antagonists NBQX and APV (data not shown).

    Details are in the caption following the image
    Figure 2. Local application of NMDA directly activates postsynaptic currents

    Picospritzing NMDA (100 μm; at the arrow) induces an inward current (A) that is completely blocked by bath application of 50 μm APV (B). Quantification of the currents is shown in C (n = 9). The frequency of the spontaneous EPSCs is not altered by picospritzing NMDA or bath application of APV (D; n = 6). *P < 0.05, one-way ANOVA.

    The relative contribution of NMDA-Rs to the total charge-transferred increases during a train of stimulations

    We next examined the contribution of the NMDA-R-mediated current to the ST-EPSCs at –60 mV in more detail. As the neurons were voltage clamped we recorded ST-EPSCs in Mg2+-free aCSF (to allow measurement of the NMDA-R contribution as this removes the Mg2+ block from the NMDA-R; Mayer et al. 1984; Qian et al. 2002). We found 10/14 NTS neurons exhibited a sustained current during a train of 10 stimulations at 50 Hz. This sustained current was not sensitive to NBQX but was blocked by the NMDA-R antagonist d-CPP-ene (DCP; 20 μm) (Fig. 3A). To evoke the maximal NMDA-mediated current it was necessary to incubate the slices in magnesium-free conditions for at least 1 h, consistent with the very high affinity of magnesium for the NMDA receptor (Mayer et al. 1984). The peak of the ST-EPSCs and area (charge transfer) were both analysed across 10 stimulations at 50 Hz (Fig. 3B and C, respectively, only including neurons exhibiting a sustained component). The proportion of the ST-evoked current mediated by NMDA-Rs (DCP-sensitive current) increased across the train of 10 stimulations as measured by both peak amplitude of the EPSC (Fig. 3B) and area (charge transfer, Fig. 3C). By the 10th stimulation in the train the averaged sustained current was 126.2 ± 37.8 pA, while the amplitude of the AMPA-R-mediated EPSC was depressed to an average of 28 ± 4.5 pA (n = 7). Thus NMDA-Rs carried the majority (∼70%) of the total charge transferred later in the stimulus train (Fig. 3C). In some neurons a very small proportion of this sustained current (5–10%) was not blocked by a combination of both NMDA-R and AMPA-R antagonists and probably reflects either the contribution of another unidentified current or small changes in the size of the sustained current over time (30–40 min of recording; this is also why the combined percentage contribution of AMPA-Rs and NMDA-Rs do not always sum to 100%). Input resistance was monitored throughout the experiment, and no significant change was observed with NMDA-R blockade (459 ± 47 MΩ vs. 452 ± 39 MΩ, n = 11, including all neurons, P = 0.83, paired Student's t test). Similarly, in a separate set of experiments, we observed the sustained current in 5 of 7 neurons, and APV (25 μm) attenuated the sustained current in those 5 neurons (reduced by an average of 57.0 ± 7.0%, n = 5, P < 0.05, paired Student's t test, data not shown).

    Details are in the caption following the image
    Figure 3. The relative contribution of NMDA-Rs increases and AMPA-Rs decreases

    An example of ST-EPSC under Mg2+-free conditions is shown in A. Note the sustained current revealed by application of NBQX (20 μm; dark grey) and blocked by application of DCP (20 μm; light grey). The peak amplitude and the charge transfer (measured as the area under trace) are quantified across 10 stimulations (50 Hz), shown in B and C, respectively. The proportion that is sensitive to NBQX is shown in dark grey while the proportion that is sensitive to DCP is shown in light grey (n = 7).

    From this point on, for all the experiments performed under Mg2+-free conditions, only neurons that exhibited NMDAR-like characteristics were pursued, unless otherwise stated.

    Effect of blocking glutamate uptake

    Our results show that NMDA evoked an inward current in all neurons tested, while NMDA receptors only contributed to the ST-evoked EPSCs in 70% of neurons. To determine whether some NMDA receptors may be extra-synaptic and therefore not activated by ST-evoked synaptic release of glutamate, we examined whether the glutamate uptake inhibitor dl-threo-β-benzyloxyaspartic acid (TBOA) had any effect in neurons that lacked, or expressed a low level of, the sustained component. Bath application of TBOA (100 μm) induced an inward shift in the holding current in all neurons tested, and the shift was completely reversed by 20 μm DCP even in the continuous presence of TBOA (∆ = –19.2 ± 4.2 pA with TBOA compared to ∆ = 1.7 ± 4.6 pA with TBOA and DCP; n = 6; P < 0.05; Fig. 4A and C). In a separate set of experiments, the size of the sustained current during ST stimulation was significantly increased in 5 of 7 neurons, to an average of 126 ± 10% of control (n = 7, P < 0.05; paired Student's t test; Fig. 4B), presumably reflecting the higher concentration and more prolonged presence of glutamate binding the NMDA-Rs. The peak amplitude of the ST-EPSCs was reduced to 78.7 ± 4.4% of control (n = 7, P < 0.01; Fig. 4B).

    Details are in the caption following the image
    Figure 4. Effects of glutamate uptake blocker and NR2B subunit-specific antagonism on NMDA-R-mediated current

    An example of a ST-EPSC with minimal sustained component in the absence (black) or presence (grey) of TBOA (100 μm), as well as the reversal with DCP (light grey; 20 μm) is shown in A. Normalized group data (n = 7) of changes in the size of the fast component (peak; squares) and the sustained current (sustained, circles, measured at 100 ms after the 1st shock) is shown in B, and the change in the holding current and its reversal by DCP is shown in C (n = 6). Application of drugs is indicated by the horizontal bar above the plots. An example of the attenuation of NMDA-R-mediated sustained current by NR2B-specific antagonist ifenprodil (10 μm; grey) is shown in D. Inset: zoomed-in plot of the first ST-EPSC. Summarized data of the effect of ifenprodil on the peak amplitude and the size of the sustained component is shown in E and F, respectively (n = 7). Black arrows indicate the beginning of the shock. *P < 0.05 one-way ANOVA in B and C, Student's t test in F.

    NMDA-R2B receptors mediate some of the NMDA-R current

    We next examined the effect of a NR2B-selective inhibitor, ifenprodil (10 μm), on the sustained current evoked at 50 Hz (Fig. 4D). Ifenprodil inhibited the sustained current by ∼50% (140.6 ± 38.4 pA (Mg2+-free ACSF) vs. 69.5 ± 17.9 pA (Mg2+-free ACSF + ifenprodil); n = 7; P < 0.05; Fig. 4D and F). DCP (20 μm) almost completely blocked the remainder of the current (Fig. 4D). Ifenprodil also inhibited the peak amplitude of the 1st eEPSC by ∼10%, an effect not significantly different from the effect of DCP (Fig. 4E).

    The relative contribution of the NMDA-R-mediated current is greater at higher frequencies of ST stimulation

    We next examined the size of the NMDA-R-mediated sustained currents and the relative contribution of AMPA-Rs and NMDA-Rs to the total ST-EPSC amplitude at different stimulating frequencies. Stimulating the ST at 10 Hz resulted in a clear sustained current in 15 of 20 neurons tested. In six neurons, 20 μm DCP was applied, which reduced the sustained current from 37.3 ± 3.8 pA to 3.2 ± 0.5 pA (Fig. 5A; P = 0.001). In contrast, under the same conditions stimulating at 2 Hz produced none, or only a small, sustained current (8.7 ± 1.4 pA; n = 7; Fig. 5B). As shown previously (Doyle & Andresen, 2001) AMPA-R-mediated ST-EPSCs reveal the significant frequency-dependent depression at high rates of stimulation due to their rapid decay (Fig. 5A and B). On average, the amplitude of the sustained current was larger at higher frequencies of stimulation (Fig. 5C). Taken together with the greater frequency-dependent depression of the AMPA-R response (Fig. 5D), this means that the contribution of the NMDA-Rs to the overall charge transfer is greater at higher stimulation frequencies.

    Details are in the caption following the image
    Figure 5. Summation of NMDA-R-mediated sustained current is greater at higher stimulation frequencies

    Examples of sustained currents at high (10 Hz) and low (2 Hz) stimulation frequencies are shown in A and B, respectively. ST-EPSCs are recorded under Mg2+-free conditions. Note that in both cases the slow component is blocked by DCP (20 μm; on the right). The development of the sustained current (measured immediately before each ST-EPSC) is plotted at four different frequencies (C; n = 6). The frequency-dependent depression of the fast AMPA-R-mediated peak (recorded in normal Mg2+) is also plotted at different frequencies, expressed as the ratio to EPSC1 (D; 50 Hz: n = 8; 10 Hz: n = 9; 5 Hz: n = 8; 2 Hz: n = 7).

    Activation of NMDA-Rs depolarizes NTS neurons

    To examine the contribution of NMDA-Rs to afferent-induced depolarization and AP generation in NTS neurons, we performed current clamp (CC) experiments using ST shocks to elicit excitatory postsynaptic potentials and APs in the postsynaptic neuron. We observed a significant depolarization and summing of NMDA-R-mediated responses at a stimulation frequency of 10 Hz, mirroring our voltage clamp results. We therefore used 10 Hz as the standard stimulating frequency for our CC protocols unless otherwise stated. We recorded AP generation in the postsynaptic neuron while stimulating the ST with a train of 100 shocks under Mg2+-free conditions. We observed the development of a standing depolarization within 10 stimulations, which was almost completely blocked with the application of 20 μm DCP (reduced by 81.7 ± 0.6%, n = 7, P < 0.001, two-way ANOVA), consistent with our voltage clamp findings (Fig. 6A, and compareFig. 6B and Fig. 5D). The standing depolarization persisted as long as the stimulation continued (Fig. 6C). In the presence of extracellular Mg2+ the standing depolarization developed more slowly and was not observed until 10 simulations were delivered, presumably because this amount of stimulation was necessary to remove the Mg2+ block of the NMDA channel. Therefore, we used 100 stimulations as our standard protocol for the remaining experiments performed in aCSF with normal [Mg2+].

    Details are in the caption following the image
    Figure 6. NMDA-Rs mediate a ST-evoked depolarization in NTS neurons at high frequency stimulation

    Under Mg2+-free conditions, stimulation of the ST results in AP firing and development of a lasting depolarization (A; aCSF), which is largely blocked with 20 μm DCP (A; DCP). Horizontal bars under the trace indicate stimulation period of a train of 100 shocks at 10 Hz. The size of the depolarization (Depo.) is plotted across the first 10 stimulations (B) and over all 100 stimulations (C) before and after DCP (n = 7). *P < 0.001, significantly different from control, two-way ANOVA.

    Activation of NMDA-Rs increases the throughput of synaptic transmission

    To further investigate the role of NMDA-Rs in synaptic transmission, we examined the effect of NMDA-R antagonists on synaptic throughput with regular external [Mg2+]. Throughput was quantified in all CC experiments as the ratio of postsynaptic APs generated by 100 afferent stimulations unless otherwise stated. With the on-threshold protocol (see Methods) blocking NMDA-Rs with DCP (20 μm) greatly decreased the throughput by 69 ± 8% in 7 of 10 neurons (representative trace shown in Fig. 7A; summarized data in Fig. 7B). In these CC experiments performed with 2 mm Mg2+ it was not possible to determine whether a neuron had an NMDA component which contributes to the ST-EPSCs without applying NMDA antagonists, therefore cells were randomly picked without any bias. It is interesting to note that the percentage of neurons that were sensitive to NMDA-R blockade in these CC experiments is the same as the percentage found to exhibit an NMDA-R contribution to the ST-EPSCs in our earlier experiments.

    Details are in the caption following the image
    Figure 7. NMDA-R antagonism reduces throughput at high, but not low, frequencies

    A, representative current clamp traces of a train of 100 shocks at 10 Hz (indicated by the horizontal bars under the trace) in aCSF and DCP in normal external [Mg2+]. Note DCP reduced the number of APs generated. B, time course of the change in throughput. Throughput at each time point is normalized to the throughput at the start of the application of DCP (minute 6). Horizontal bar indicates bath application of drug. Throughput is reduced within 3–5 min of application of DCP (light grey; n = 7) but not aCSF (black; n = 8). C, APV (50 μm) significantly reduced throughput in 7 of 11 neurons when the ST is stimulated at 10 Hz, 3 of 11 neurons at 5 Hz, and 0 of 7 neurons at 2 Hz. Bars represent summarized data from responders only for 10 Hz and 5 Hz, and from all neurons for 2 Hz. Grey dots indicate each individual neuron (includes both responders and non-responders). *P < 0.001, significantly different from baseline, one-way ANOVA.

    When we divided the 100 stimulations into 10 segments of 10 shocks each, the effect of DCP to reduce throughput was statistically significant only later in the train (starting from shock Nos 50–59; two-way ANOVA, data not shown). APV (50 μm) had a similar effect, decreasing the throughput by 63 ± 9% in 7/11 neurons (Fig. 7C). However, when repeated with lower stimulating frequencies (5 Hz and 2 Hz), blockade of NMDA-Rs with APV had a much smaller, or no effect, on throughput (Fig. 7C; ∼25% reduction in 3/11 randomly selected cells at 5 Hz; no significant change in all 7 neurons tested at 2 Hz, P = 0.16). We also examined the effect of DCP on throughput using a supra-threshold protocol (see Methods). In those cases blockade of NMDA-Rs had a much smaller effect, only decreasing the throughput in 3 of 7 neurons by an average of 30 ± 20% (data not shown).

    In contrast, when the AMPA-R contribution was blocked with 20 μm NBQX in regular extracellular Mg2+, the throughput at 10 Hz was consistently reduced in all eight neurons tested (Fig. 8B; on average a reduction of 78 ± 13%; P < 0.001; n = 8). However, while AP generation was greatly reduced, the standing depolarization was less sensitive to AMPA-R block as on average NBQX had no significant effect (control: 11.0 ± 2.3 mV; NBQX: 6.0 ± 1.4 mV; summarized in Fig. 8C; n = 8; P = 0.12).

    Details are in the caption following the image
    Figure 8. AMPA-R antagonism reduces throughput but has less impact on the standing depolarization

    A, example of AP firing evoked by a train of 100 shocks at 10 Hz under control conditions (top panel), with 20 μm NBQX (middle panel), and with both NBQX and 20 μm DCP (bottom panel). Summarized data (n = 8) of the effects of NBQX on throughput and the size of the standing depolarization (Depo, measured as the average ∆ (in mV) from baseline immediately before the stimulation) are shown in B and C, respectively. *P < 0.001, Student's t test.

    NMDA-R antagonism did not alter the intrinsic ability of NTS neurons to generate APs in response to current injections

    To test whether NMDA-R antagonists decreased the throughput by affecting the intrinsic sensitivity of the postsynaptic neuron to depolarizing current, we monitored AP generation (Fig. 9B) and depolarization (Fig. 9C) in response to current injections of various magnitudes (20, 40, 60, 80, 100 and 150 pA) in regular (2 mm) extracellular Mg2+. DCP had no effect on the ability of current injections to depolarize NTS neurons (that were randomly selected, with no bias for the expression of NMDA currents) (Fig. 9A and B; n = 6; P = 0.28 and 0.47, respectively). Across all neurons examined, the membrane resistance was 516 ± 93 MΩ and resting membrane potentials were highly consistent between cells, allowing for comparison of injected current to AP generation relationships.

    Details are in the caption following the image
    Figure 9. NMDA-R antagonism does not alter the intrinsic ability of NTS neurons to fire

    A, example of a NTS neuron firing APs in response to current injection (60 pA), with or without the NMDA-R antagonist DCP. Current injection protocol is shown under the trace. The number of APs generated and the size of the depolarization (Depo) are plotted against the size of the current injection, shown in B and C, respectively (n = 6). Not significantly different determined by two-way ANOVA.

    NMDA-R-mediated synaptic currents are found in both NTS neurons receiving asynchronous inputs and those not receiving asynchronous inputs

    Asynchronous release has been shown to be another important mechanism by which ST-afferent synapses can increase the total charge transferred across a synapse during high frequency release (Peters et al. 2010). A neuron's asynchronous input profile has also been reported to be a good indicator of A- vs. C-fibre afferent phenotype in rats (Peters et al. 2010). To determine whether NTS neurons can employ both mechanisms we tested the association of NMDA-R afferent synaptic currents with their asynchronous release profile as described in Peters et al. (2010). We observed NMDA-R currents in 12 of 18 neurons tested; 9 of the 12 received asynchronous inputs (9/12; C fibre profile) and 3 of the 12 did not (3/12; A fibre profile). Finally, the 6 neurons that did not express NMDA-R current all received asynchronous inputs (data not shown).

    Discussion

    NMDA-Rs in the NTS have been shown to participate in a number of vagal reflex pathways (Antunes et al. 2004; Almado & Machado, 2005; Mutolo et al. 2007; Wright et al. 2011; Liu et al. 2012; Marques-Lopes et al. 2012). However, the specific mechanism(s) by which they contribute to the fidelity of transmission across afferent–NTS neuron synapses are not well understood. The central finding of the current study is that activation of postsynaptic NMDA-Rs maintains synaptic throughput at ST afferent–NTS neuron synapses during high frequency afferent firing. Unlike AMPA receptor-mediated currents, the relatively slow gating kinetics of NMDA-Rs permit the summation of phasic currents during conditions when vesicular glutamate release is low, such as during prolonged afferent volleys at frequencies >5 Hz. As a consequence the relative contribution of AMPA vs. NMDA receptors to the total synaptic charge transfer changes dynamically, in a frequency-dependent fashion, throughout the train of stimulations. This novel role of NMDA-Rs may partly explain how primary visceral afferents can maintain signalling fidelity across a broad range of firing frequencies.

    A majority of NTS neurons express an NMDA-R current

    Using three different electrophysiological approaches we found that the majority of NTS neurons exhibit NMDA-R-mediated currents. Specifically, we found that in 70% of NTS neurons NMDA-Rs contributed to 70% of the total charge transfer when recordings were made at +40 mV. Our observation is consistent with a previous report in young animals using this protocol (Balland et al. 2008); these results suggest that NMDA-Rs play a functional role in the NTS well beyond early developmental stages (Vincent et al. 1996). Consistent with previous reports we found that, when slices were incubated in Mg2+-free aCSF, the majority of NTS neurons exhibited an NMDA-R component as part of their ST-evoked glutamatergic EPSCs (Aylwin et al. 1997; Smith et al. 1998; Yen et al. 1999; Jin et al. 2003).

    In addition to the charge transfer during synchronous release of glutamate, asynchronous glutamate release from vagal afferent endings also increases charge transfer across the ST–NTS neuron synapse during bursts of high frequency firing (Peters et al. 2010). However, asynchronous release is associated only with activation of vagal C-type, but not A-type, afferent fibre synapses. In contrast, we found that the NMDA-R component of the ST-EPSC was present in both neurons that received asynchronous inputs (C fibres) and those that did not (A-type), suggesting that NMDA-Rs increase charge transfer at both A and C fibre afferent synapses during high frequency stimulation. The NMDA-R current develops fairly early on in the train of stimulations. In contrast, the asynchronous release profile is delayed and occurs after the synchronous release (Peters et al. 2010). Therefore, in neurons expressing NMDA-Rs that receive input from A-type afferents the standing NMDAR current is probably a result of the slow kinetics of the NMDA receptors activated during synchronous release and not a result of increasing charge transfer during asynchronous glutamate release. However, the majority of NTS neurons receive C-type afferent inputs and are therefore subject to the effects of asynchronous glutamate release. Hence, standing NMDA-R currents from synchronous release and currents activated by asynchronous glutamate release may well interact in this population.

    Finally, we found that all neurons responded to a direct puff of NMDA (also in Mg2+-free conditions), similar to previous reports for bath application of NMDA (Tell & Jean, 1991). A potential explanation for the higher proportion of neurons responding to exogenous NMDA is that NMDA application also activates NMDA-Rs at non-afferent glutamate synapses or extra-synaptic sites (Le Meur et al. 2007) that can be activated by ambient glutamate or glutamate released from glial cells (Araque et al. 1999). This hypothesis remains to be fully tested, but is supported by our data using TBOA, which caused a shift in the holding current in all neurons tested, an effect fully reversed by DCP, consistent with previous reports (Fleming et al. 2012; Wilson-Poe et al. 2012). Fleming and colleagues (Fleming et al. 2012) argued that this inward current is a result of the activation of extra-synaptic NMDA-Rs, although we cannot rule out NMDA-Rs on non-ST afferent synapses being activated by the increased diffusion of glutamate in the presence of the uptake blocker. TBOA did not increase the proportion of neurons expressing the prolonged sustained current during ST stimulation, however, supporting the conclusion that 60–70% of synapses between ST afferents and NTS neurons express NMDA-Rs.

    The fact that the obligatory NR1 subunit of NMDA-Rs is broadly expressed in the NTS, including in dendrites and cell bodies (Ambalavanar et al. 1998; Aicher et al. 1999; Lin & Talman, 2000; Huang & Pickel, 2002; Lachamp et al. 2003; Glass et al. 2004; Balland et al. 2006), is consistent with the high prevalence of NMDA-R-mediated responses we detected in NTS neurons. In contrast, while NMDA-Rs are also located in both nodose ganglia neurons (the cell bodies of vagal afferents) and on terminals in the NTS (Aicher et al. 1999; Czaja et al. 2006), we did not see evidence of presynaptic effects of NMDA-Rs in the present study. However, all of our recordings were made under conditions where the probability of glutamate release is extremely high (Bailey et al. 2006; Peters et al. 2008), which could explain why we did not observe a further increase in spontaneous release by the activation of NMDA-Rs. Alternatively, presynaptic NMDA-Rs may mediate longer-term changes in synaptic function, for example via phosphorylation of synapsin (Campos et al. 2012).

    NTS neurons have been found to express all four NR2 subunits (Guthmann & Herbert, 1999; Baude et al. 2009). In the present study we observed a prominent effect on throughput with the NMDA-R antagonist DCP, which has a higher affinity for NR2A/B subunits vs. NR2C/D subunits (Hrabetova et al. 2000). The NR2B-specific inhibitor ifenprodil also reduced the amplitude of the ST-evoked EPSCs, suggesting that some postsynaptic NMDA channels on NTS neurons located at afferent synapses have NR2B subunits.

    NMDA-R contribution is frequency dependent

    We observed substantial frequency-dependent depression in the amplitude of AMPA-R-mediated EPSCs at higher frequency stimulations, as has been reported previously (Doyle & Andresen, 2001). At physiological frequencies of afferent firing this depression is not due to AMPA-R desensitization (Chen et al. 1999), but is thought to be due to depletion of the readily releasable pool of glutamate owing to the high probability of glutamate release from ST terminals (Bailey et al. 2006). Interestingly, glutamate has a considerably higher affinity for NMDA-Rs than AMPA-Rs (Patneau & Mayer, 1990). Consequently, NMDA-Rs are likely to be preferentially activated at higher frequencies of afferent stimulation, when less glutamate is released. Furthermore, the slow decay kinetics of NMDA-Rs allows these currents to summate during higher frequency stimulations (Lester & Jahr, 1992; McBain & Mayer, 1994). In contrast to NMDA-R currents, AMPA-R currents did not readily summate even at the fastest frequencies tested. These use-dependent differences in glutamate receptor gating and glutamate affinity, coupled with frequency-dependent changes in presynaptic vesicle release, would predict a dynamic switch from AMPA-Rs to NMDA-Rs to provide the majority of charge carried across the ST–NTS synapse. Our data support this as at frequencies greater than 5 Hz most of the integrated charge of the ST-EPSC is carried by NMDA-Rs (60–90%), not AMPA-Rs. NMDA-Rs mediate a prolonged depolarization, which is sustained throughout a train of high frequency stimulations and critical for maintaining throughput of AP generation at higher frequency stimulations (>5 Hz). In contrast, no prolonged depolarization occurs at low frequency stimulation as the stimulations are too far apart for summation of individual NMDA-R responses. There is also little depression of the AMPA-R response at low frequencies, so AMPA-R activation remains sufficient for AP generation throughout the low frequency train and NMDA-R activation is less critical. Interestingly, if multiple (2–4) afferent inputs are recruited at the same time, larger amplitude compound ST-EPSCs are generated and NMDA-R antagonists do not have a major effect on throughput; presumably because the summated AMPA-R response is sufficient to reach threshold even with frequency-dependent depression. NMDA-Rs therefore appear to be particularly important for decreasing the threshold of activation of NTS neurons to single or relatively small convergent inputs.

    We found that the NMDA-R-mediated depolarization develops much more slowly in normal extracellular Mg2+ concentrations than in 0 Mg2+ and consequently the NMDA-R contribution to throughput was greater later in the train in physiological magnesium, consistent with the idea that NMDA-Rs became available as Mg2+ block is released by continuous afferent stimulation. These observations imply that in vivo NMDA-Rs will contribute more to reflexes involving prolonged bursts of afferent firing. Interestingly, the Mg2+ block is not removed when AMPA-R contribution is highest (early in the train), but rather when the contribution of AMPA-Rs is low during prolonged high frequency bursts, hinting that additional depolarizing mechanisms other than the fast-activation of AMPA-Rs may contribute to the removal of Mg2+ block from NMDA-Rs.

    NMDA-R facilitates synaptic transmission in other high probability of release synapses

    NMDA-Rs are traditionally thought to mediate longer-term changes in synaptic formation and plasticity (Bliss et al. 2003; Citri & Malenka, 2008; Rebola et al. 2010). However, consistent with our current findings there is accumulating data showing that NMDA-Rs are important for synaptic transmission and throughput at other central synapses such as the dorsal nucleus of the lateral lemniscus (Porres et al. 2011) and dorsal lateral geniculate nucleus, another high probability of release synapse that shows significant vesicle depletion and non-summating AMPA-R responses (Budisantoso et al. 2012). NR2B receptors in particular have recently been shown to contribute to the summation of glutamate synaptic responses in developing auditory synapses in the nucleus laminaris in the brainstem (Sanchez et al. 2012). Taken together with our findings we predict that this may be a common mechanism that allows the high probability of release synapses to maintain the fidelity of synaptic transmission, especially under conditions of prolonged high frequency activation.

    Physiological implications

    Our findings are consistent with NMDA-Rs being especially important in reflexes that are mediated by prolonged, higher frequency, discharge rates. Gastric afferents fire at sufficiently high frequencies (10–20 Hz) (Schwartz et al. 1991) to recruit the participation of NMDA-Rs following gastric distention and cholecystokinin (CCK) administration (conditions mimicking ingestion of a meal). Further, NTS neurons expressing NMDA-Rs are activated by gastric distention in vivo (Berthoud et al. 2001). In this regard it is interesting that NMDA-R activation is required for both meal-induced and CCK-induced inhibition of food intake (Treece et al. 1998; Hung et al. 2006; Wright et al. 2011), as well as CCK-induced increase of c-fos and p-Erk in NTS neurons (Wright et al. 2011; Campos et al. 2012).

    NMDA-Rs also contribute to vagally mediated cardiovascular reflexes (Kubo & Kihara, 1988a,b; Vardhan et al. 1993a,b; Haibara et al. 1995; Marques-Lopes et al. 2012). Consistent with our model, NMDA-R antagonists only attenuate aortic baroreceptor reflexes initiated by high frequency afferent discharge, while AMPA-Rs contribute to responses mediated by both low and high afferent discharge (Gordon & Leone, 1991). Decreased NMDA-R activation in the NTS also attenuates other reflexes involving high frequency vagal afferent discharge, such as cough (Mutolo et al. 2007) and pain induced c-fos signals in NTS neurons (Marques-Lopes et al. 2012). In addition, participation of NR2B subunits in NTS synaptic function is intriguing as these subunits are associated with greater calcium permeability and their expression is highly plastic and altered by many protocols that trigger long-term synaptic plasticity (Kerchner & Nicoll, 2008; Huang et al. 2009).

    In summary, our data demonstrate that postsynaptic NMDA-Rs contribute to ST-EPSCs in the majority of second-order NTS neurons. At high frequencies of stimulation NMDA-Rs carry the majority of charge transferred across the synapse and appear critical for maintaining the fidelity of AP generation across the ST–NTS neuron synapse. This use-dependent facilitation is a result of the slower decay kinetics of NMDA-Rs and perhaps a higher affinity for glutamate. Our results describe a potential mechanistic explanation for the requirement of NMDA-Rs for gastrointestinal and cardiovascular reflexes and for their participation in the control of behaviours such as food intake, which are associated with relatively high frequencies of vagal afferent discharge and prolonged afferent volleys.

    Additional information

    Competing interests

    No conflicts of interest, financial or otherwise, are declared by the authors.

    Author contributions

    H.Z. helped design, perform, analyse and interpret the experiments, contributed to the conception and design of the study, and co-wrote the manuscript. J.H.P. helped design, perform and interpret the experiments, contributed to the conception and design of the study, and co-wrote the manuscript. M.Z. and S.J.P. performed, analysed and interpreted the experiments and helped revise and critically edit the final manuscript. R.C.R. and S.M.A. contributed to the conception and design of the study, interpretation of the experiments and co-wrote the manuscript. All authors approved the final version of the manuscript.

    Funding

    This work was supported by grants from the National Institutes of Health (DK083452 to S.M.A. and DK-52849 to R.C.R.). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of Diabetes and Digestion and Kidney disease or the National Institutes for Health.