Distinct target cell-dependent forms of short-term plasticity of the central visceral afferent synapses of the rat
© Yamamoto et al. 2010
Received: 31 May 2010
Accepted: 20 October 2010
Published: 20 October 2010
Skip to main content
© Yamamoto et al. 2010
Received: 31 May 2010
Accepted: 20 October 2010
Published: 20 October 2010
The visceral afferents from various cervico-abdominal sensory receptors project to the dorsal vagal complex (DVC), which is composed of the nucleus of the solitary tract (NTS), the area postrema and the dorsal motor nucleus of the vagus nerve (DMX), via the vagus and glossopharyngeal nerves and then the solitary tract (TS) in the brainstem. While the excitatory transmission at the TS-NTS synapses shows strong frequency-dependent suppression in response to repeated stimulation of the afferents, the frequency dependence and short-term plasticity at the TS-DMX synapses, which also transmit monosynaptic information from the visceral afferents to the DVC neurons, remain largely unknown.
Recording of the EPSCs activated by paired or repeated TS stimulation in the brainstem slices of rats revealed that, unlike NTS neurons whose paired-pulse ratio (PPR) is consistently below 0.6, the distribution of the PPR of DMX neurons shows bimodal peaks that are composed of type I (PPR, 0.6-1.5; 53% of 120 neurons recorded) and type II (PPR, < 0.6; 47%) neurons. Some of the type I DMX neurons showed paired-pulse potentiation. The distinction of these two types depended on the presynaptic release probability and the projection target of the postsynaptic cells; the distinction was not dependent on the location or soma size of the cell, intensity or site of the stimulation, the latency, standard deviation of latency or the quantal size. Repeated stimulation at 20 Hz resulted in gradual and potent decreases in EPSC amplitude in the NTS and type II DMX neurons, whereas type I DMX neurons displayed only slight decreases, which indicates that the DMX neurons of this type could be continuously activated by repeated firing of primary afferent fibers at a high (~10 Hz) frequency.
These two general types of short-term plasticity might contribute to the differential activation of distinct vago-vagal reflex circuits, depending on the firing frequency and type of visceral afferents.
The visceral afferents arising from various kinds of receptors carry a wide range of information about the status of the cervico-abdominal organs including gastric load , esophageal tension [2, 3], lung volume [4–6], arterial blood pressure , chemosensory inputs from the carotid bodies  and intragastric concentrations of bioactive substances [1, 9]. Compared to somatosensory sensation that reports rapid touch pressure changes and acute nociceptive information, these sets of visceral information are encoded in a long-lasting and slowly changing frequency-modulated series of action potentials and transmitted to the brain via primary afferent fibers that pass through the vagus and glossopharyngeal nerves. These afferent axons then form their first intracerebral synapses in the dorsal vagal complex (DVC) composed of the nucleus of the solitary tract (NTS), area postrema and the dorsal motor nucleus of the vagus nerve (DMX). The DVC is located on or close to the dorsal aspect of the medulla beneath the fourth ventricle [10–12]. In this regard, frequency-dependent transfer properties, such as short-term plasticity and frequency-dependent suppression [13, 14], at these first synapses in the DVC should play the primary role in determining how the central neurons respond to frequency modulation-encoded visceral information .
Synaptic transmission between the baroreceptor afferents and the second-order neurons in the NTS has been well studied and shown to be strongly suppressed at elevated input frequencies in anesthetized rats  and in brainstem slice preparations [13, 16–20]; i.e., neurons cannot respond to inputs at an elevated frequency or at short inter-spike intervals. One of the advantages of such "low-pass filter" characteristics of synaptic transmission is that it can attenuate excessive rapid fluctuations in central reflex responses of the autonomic output . This is a property that is suitable for dealing with "phasic" inputs, whereas an apparent disadvantage is that neurons cannot faithfully respond to continued high-frequency "tonic" inputs. This is however apparently contradictory because a subset of vagal afferents shows continuous discharge at approximately 10 Hz in response to elevated gastric load  and esophageal tension [2, 3].
Here, we analyzed the short-term plasticity of the synapses between the primary afferents and various types of DVC neurons in acute slice preparations. In particular, we compared the frequency dependence between the well-studied NTS neurons [7, 13, 16–20] and the much less studied DMX neurons [10, 11, 16, 21–24] that also form a part of vago-vagal reflex pathway . Our findings show that distinct classes of postsynaptic neurons show distinct types of short-term plasticity and frequency dependence resulting from distinct presynaptic mechanisms, depending on the type and function of each neuron.
In contrast, DMX neurons showed much larger variation in terms of short-term plasticity. For example, two representative DMX neurons, which were located immediately adjacent to each other in the same slice (cell 1 and cell 2 in Figure 1A-DMX neuron), showed distinct paired-pulse plasticity (Figure 1B); one with almost no paired-pulse depression (cell 1, Figure 1B-cell 1; PPR = 0.90) and another with strong paired-pulse depression (cell 2; PPR = 0.47). The histogram with blue bars in Figure 1C, constructed on the basis of recordings from 115 DMX neurons, further confirms that the distribution of PPR in DMX shows greater variation (ranging from 0.2 to 1.5) than that of dm-cNTS neurons. This distribution of PPR among DMX neurons was well fitted with a composite of two Gaussian distributions (blue and green lines in Figure 1C). We classified DMX neurons in two groups because: i) these two Gaussian distributions showed the smallest overlap at a PPR value around 0.6; ii) there was a trough at around 0.6 for the distribution of PPR among DMX neurons; iii) the PPR of dm-cNTS neurons never exceeded 0.6. The groups were as follows: those exhibiting PPRs larger than 0.6 ("type I" DMX neurons; 47.5% of 120 DMX neurons) and those with PPRs smaller than 0.6 ("type II" DMX neurons; 52.5%) (Figure 1C). According to this classification, cell 1 in Figure 1B was classified as a type I DMX neuron and cell 2 as a type II DMX neuron. The mean PPR values of type I and type II DMX neurons were 0.92 ± 0.02 (n = 63) and 0.46 ± 0.10 (n = 57), respectively. Visualization of cell forms with intracellular injection Alexa Fluor or Neurobiotin was made in 11 DMX cells, of which 5 type I and 6 type II neurons, and there was no apparent difference in the soma shape, the direction and the number of thick dendrites.
To our knowledge, this distinction with regard to short-term plasticity in DMX neurons has not been described before. These results were unexpected because it has been believed that DMX neurons, the majority of which are parasympathetic preganglionic neurons, represent a relatively homogeneous population . The following four lines of evidence strongly support the notion that these differences are intrinsic to the properties of synaptic transmission between TS and each distinct neuron type and do not reflect experimental artifacts.
Second, this distinction between type I and type II DMX neurons was robustly observed at a wide range of stimulation intensities. To examine the hypothesis that a specific type of DMX neuron predominantly receives afferent fibers with a similar class of short-term plasticity (i.e., the type of postsynaptic neuron determines the type of afferents that converge), we analyzed the effect of changing the stimulation intensity from sub-threshold to supra-maximal levels on EPSC amplitude and PPR (Figure 2C) because this maneuver should gradually expand the population of activated afferents. As expected, augmentation of stimulation intensity increased the EPSC amplitude in all three groups, presumably by recruiting higher threshold primary afferent fibers. Compared to dm-cNTS neurons that showed abrupt increases in the EPSC amplitude upon increases in stimulation intensity , DMX neurons showed relatively gradual increases in amplitude (Figure 2D). The rate of rise of the intensity-amplitude curve estimated by sigmoidal curve fitting (Figure 2D) was significantly (Mann-Whitney U-test, p < 0.05) and markedly slower for DMX neurons than that for dm-cNTS neurons (dm-cNTS, type I and type II DMX neurons; 447 ± 282 mA-1, 5.2 ± 0.7 mA-1 and 8.5 ± 1.5 mA-1, respectively); this suggests that the afferent fibers that converge onto the DMX neurons possess a larger variation in threshold than those that converge onto the NTS neurons. This limited convergence of afferents to the NTS neurons is consistent with the previous findings [26, 28]. However, the intensity-PPR plot indicates that the PPR-dependent classification between NTS, type I and type II DMX neurons was relatively resistant to the changes in stimulation intensity (Figure 2D). This result argues against the possibility that the classification of type I and II DMX neurons is simply attributed to the properties of a set of fibers that are stimulated at a specific condition but supports the possibility that this classification is attributed to the type of the neuron of concern.
The above results clearly indicate that aspects of short-term plasticity vary among these three types of synapses of dm-cNTS and DMX neurons. We sought to identify distinct cellular mechanisms that underlie these differences in short-term plasticity.
The distribution of PPRsuccess for the DMX was divided into two populations (< 0.6 and >0.6) in a similar manner to PPRampli. There was no significant difference between the distributions of PPRsuccess (gray bars) and PPRampli (red cityscape plots) in the DMX and the dm-cNTS (Figure 5E; Kolmogorov-Smirnov test; p = 0.95 and p = 0.40 for the DMX and the dm-cNTS, respectively). There was no significant difference between the proportion of type I neurons, as classified according to PPRampli and PPRsuccess (according to PPRampli, 63 neurons among 120 cells (52.5%); according to PPRsuccess, 6 neurons among 11 cells (54.5%); p = 0.9, Chi-square test). These data also strongly support that the distinction between type I and II DMX neurons does not depend on the intensity of the stimulation. These results support the hypothesis that the paired-pulse plasticity observed in the amplitude of EPSC that is evoked by submaximal stimulation described above results mostly from short-term plasticity in the release probability and not from short-term plasticity in the postsynaptic AMPA receptor responses. These data indicate that the specific properties of short-term plasticity of the EPSC amplitude in different postsynaptic neurons result primarily from the properties of transmitter release from the presynaptic terminals.
Cell and synaptic characteristics of dm-cNTS and DMX neurons
Longitudinal soma size (μm)
13.0 ± 0.5 (31)
26.1 ± 0.5 (110)
26.0 ± 0.6 (58)
26.2 ± 0.7 (52)
Resting membrane potential (mV)
-64.7 ± 2.2 (33)
-63.5 ± 0.9 (115)
-65.5 ± 1.2 (59)
-61.3 ± 1.1 (56)
Cell capacitance (pF)
12.1 ± 1.1 (34)
31.2 ± 1.0 (117)
31.0 ± 1.5 (61)
31.2 ± 1.5 (56)
Input resistance (MΩ)
755.6 ± 70.5 (34)
270.1 ± 11.6 (113)
277.9 ± 17.1 (61)
261.0 ± 15.5 (52)
EPSC amplitude (pA) (submaximal stimulation)
243.5 ± 22.0 (35)
266.5 ± 14.1 (120)
211.3 ± 14.5 (63)
323.2 ± 22.7 (57)
Decay time-constant (ms)
6.0 ± 0.3 (35)
8.1 ± 0.2 (120)
8.6 ± 0.3 (63)
7.5 ± 0.3 (57)
9.3 ± 0.3 (35)
9.4 ± 0.1 (120)
9.6 ± 0.2 (63)
9.2 ± 0.2 (57)
Cell and synaptic characteristics of DiI(+) type I and type II DMX neurons
DiI(+) DMX neuron
Longitudinal soma size (μm)
29.3 ± 1.3 (21)
29.2 ± 1.4 (17)
30.0 ± 4.6 (4)
Resting membrane potential (mV)
-60.0 ± 1.1 (20)
-60.6 ± 1.3 (16)
-57.75 ± 1.5 (4)
Cell capacitance (pF)
26.8 ± 1.8 (21)
27.3 ± 2.2 (17)
24.9 ± 3.0 (4)
Input resistance (MΩ)
515.8 ± 59.0 (21)
559.9 ± 63.4 (17)
328.0 ± 143.2 (4)
EPSC amplitude (pA) (submaximal stimulation)
114.0 ± 17.4 (21)
110.4 ± 18.8 (17)
129.2 ± 62.4 (4)
Decay time-constant (ms)
8.5 ± 0.9 (21)
7.7 ± 0.5 (17)
12.2 ± 1.0 (4)
9.9 ± 0.6 (21)
9.3 ± 0.4 (17)
12.6 ± 0.2 (4)
This study has described in detail, for the first time to our knowledge, distinct forms of short-term plasticity of the first intracerebral synapses formed between the visceral afferents and the second-order neurons in the DMX. Our data confirm neurophysiologically the previously published anatomical observations showing that the TS terminals make monosynaptic asymmetric contact with the dendrites of viscerally projecting DMX neurons in the cNTS regions [11, 21, 25]. Such a monosynaptic vago-vagal connection has also been functionally suggested by the authors of studies using slice preparations of rats and guinea pigs [16, 22, 24]. The novel finding of the present study is that, unlike the TS-NTS synapses that consistently exhibit strong short-term depression [13, 26, 36, 37] and this study, the TS-DMX synapses show distinct types of short-term plasticity ranging from paired-pulse depression to facilitation, which results primarily from the distinct release properties of each afferent fiber.
In accordance with previously published studies [26, 37], the TS-dm-cNTS transmission consistently exhibited a small PPR for EPSC amplitude, which is likely to result from paired-pulse depression of the release probability from single terminals, indicating the number of glutamate-containing vesicles recruited for exocytosis is reduced during repeated stimulation. In a similar manner to these dm-cNTS neurons [37–40], about half of DMX neurons were classified to "type II" DMX neurons that show short-term depression (PPR < 0.6) and another half were classified as "type I" DMX neurons exhibiting markedly higher PPR values (PPR > 0.6). To our knowledge, such short-term facilitation has never been reported with TS-DVC synapses.
The results of this study suggest that differences in short-term plasticity between type I and II DNX neurons result from the properties of vesicular release from the TS terminal, as has been demonstrated with TS-NTS synapses [26, 41] as well as in this study. While the properties of the type II DMX synapses described in this study favor short-term depression primarily due to large P r, more frequent multi-vesicular release, and a smaller pool of docked vesicles [40, 29, 30], the properties of TS-type I DMX neuron synapses with much less short-term depression or rather facilitation in response to both submaximal and minimal stimulation are likely to result from small P r and a sufficient number of vesicles that are ready for exocytosis even after a short period of time. It will therefore be necessary for future studies to identify the molecular determinants of such distinct short-term plasticity in these DMX synapses .
To our knowledge, this study provides the first detailed description of the short-term plasticity of the TS-DMX synapses. Because the synaptic properties, especially those for TS-type I DMX transmissions, are distinctly different from those that have been demonstrated for TS-NTS synapses, it is essential to confirm that these transmissions are monosynaptic. For the monosynaptic transmission between arterial baroreceptor afferents and second-order cNTS neurons that receive these afferents at their perisomatic synapses, a pioneering study by Doyle & Andresen  has demonstrated that this transmission consistently shows smaller SD of latency over stimulation trials (the mathematical definition of a frequently used jargon, "jitter") in the horizontal slices at 34°C - 37°C. However, for the synapses between TS afferents and DMX neurons are formed on distant dendrites, and not on the soma [11, 21]. Because, to our knowledge, there is no information on the SD of latency for synaptic transmission of the non-baroreceptor TS afferents that form synapses on the dendrites of the second-order DVC neurons, we could not use the smaller SD of latency as an a priori criterion for monosynaptic transmission. Whatever the case, the SD of latency in this study for the DMX and dm-cNTS neurons was consistently smaller than 500 μs (Figure 4D), which is a value used by Chen et al.  for the criteria for monosynaptic transmission in their study using a similar type of coronal slices and room temperature recording.
Instead, the following two pieces of evidence support the notion that the EPSCs in both type I and II DMX neurons are monosynaptically generated and not caused subsequently by the firing of second-order interneurons. First, there was no significant difference between groups in the latency of sfEPSC and EPSC, which argues against the possibility that one cell type involves the activation of more synapses than the other. If this was the case, it would be expected that polysynaptically evoked EPSCs would exhibit longer latency due to the time required for EPSP accumulation and the generation and conduction of action potentials in second-order interneurons. Such a pattern was not observed. Second, there was no significant difference between the required intensity of minimal stimulation for the all-or-none responses (i.e., sfEPSC); a result that argues against the possibility that minimal stimulation directly elicited interneuron firing by directly stimulating the somatodendritic regions of such interneurons in the vicinity of the stimulation electrode. If this possibility is true for any of these neuron groups, higher stimulation intensity should have been required to generate action potentials, due to their lower expression density of voltage-dependent Na+ channels and lower specific membrane input resistance than in the axons. In addition, if firings of second-order NTS neurons underlie the EPSCs in the DMX cells, then the required minimal stimulation intensity should be higher so that the EPSP in the second-order neurons becomes large enough to generate action potentials by spatial summation in response to single-fiber stimulation. On the basis of these arguments, we conclude that both responses in type I and II DMX neurons occur through monosynaptic activation of neurons that are postsynaptic to the TS afferents.
In a physiological sense, two pulses separated by a few hundred ms, as used in the PPR measurements, do not have any functional meaning. Rather, such short-term plasticity plays an essential role in determining the cut-off frequency of a synapse, thus limiting the content of information flow at a specific pathway [32, 30, 42–45]. The most important consequences of the higher PPR of type I DMX neurons are that these neurons are capable of being activated by repeated high-frequency afferent inputs up to a few tens of Hz in a manner that contrasts with that of the dm-cNTS and type II DMX neurons that filter out high-frequency components and can be sufficiently excited only by low (~0.1 Hz) frequency inputs. In contrast to paired pulse stimulation, continuously repeated action potentials might trigger other mechanisms that additionally affect synaptic depression, such as the activation of presynaptic metabotropic glutamate receptors  or adenosine receptors . These possibilities are not explored in detail in the present study and are the subjects of future studies.
As a consequence, the findings of this study could be used to elaborate a "synaptic filter" model of these TS-DMX synapses. 1) At TS-type I DMX neuron synapses, the cut-off frequency  is elevated through a decreased release probability. 2) At TS-type II DMX neuron and TS-dm-cNTS neuron synapses, the cut-off frequency is lowered through increased release probability and short-term depression.
One of the possible functional implications of different forms of short-term plasticity would be the ability to filter the information represented by a series of repeated firings from the afferents arising from visceral organs depending on the discharge frequency. Such a mechanism could facilitate the frequency-dependent decoding of the afferent information to optimize the direct vago-vagal loop regulation of the visceral organs. Type I DMX neurons might receive afferents from so-called "dynamic range" receptors that start to generate a tonic series of action potentials at a low level of distention of digestive walls or at a low concentration of intragastric substances, such as cholecystokinin . In contrast, type II DMX neurons might receive afferents arising from low-frequency-discharge receptors and/or selectively transmit information related to the "onset" of changes in the visceral environment. In these senses, type I DMX neurons can be denoted "tonic" receivers and type II neurons "phasic" receivers. Such distinct frequency filter properties of the first-step synapses of visceral afferents might contribute to the formation of appropriate discharge patterns of the neurons in the central circuit in response to various forms of afferent inputs, including those activated by duodenal acidification, intestinal distension, antral distension, cytokine generation during sepsis, and poisoning with emetics .
To our knowledge, the frequency dependence and short-term plasticity of the TS-DMX synapses have not previously been reported; therefore, it has been considered, based on the data on the TS-dm-cNTS synapses, that the synaptic transmission between the TS and DVC neurons does not transfer high-frequency components and is therefore high-cut filtered. Nevertheless, this is not always the case because our first description of the frequency dependence of TS-DMX transmission presented here clearly indicates that there are distinct forms of short-term plasticity and filtering properties in these synapses. These results provide another example of pre- and postsynaptic functional association in which the role and type of a postsynaptic neuron determines the mode and characteristics of transmitter release from the presynaptic terminal that forms synapses onto it.
Animal handling conformed to the Guiding Principles for the Care and Use of Animals in the Field of Physiological Sciences of the Physiological Society of Japan (1988) and was approved by the Animal Care Committee of The Jikei University. Transverse brainstem slices from Wistar rats (2-5 weeks) were prepared according to a method described previously [48, 49]. Briefly, following the decapitation under sufficient anesthesia with overdose ketamine (100-150 mg kg-1, i.p.; in 133 rats) or isoflurane (5% in 100% O2; in 91 rats), the lower brainstem was dissected out and secured on the cutting stage of a vibrating blade slicer (DSK-1000, Dosaka EM, Kyoto, Japan) with the caudal end upward. As there was no apparent difference between the use of ketamine compared to isoflurane for the initial anesthesia, the data from both groups are pooled. Two to three transverse slices (400-μm thick) containing the DVC were cut in ice-cold "cutting" artificial cerebrospinal fluid (ACSF) composed of (in mM) NaCl 125, KCl 3, CaCl2 0.1, MgCl2 5, NaH2PO4 1.25, D-glucose 10, L-ascorbic acid 0.4 and NaHCO3 25 (pH 7.4 bubbled with 95% O2 + 5% CO2; osmolarity, approximately 310 mOsm kg-1). The use of coronal transverse slices was necessary because of the following: 1) the DMX neurons could be clearly distinguished by microscopic observation from those in the adjacent NTS and the hypoglossal nucleus ; 2) TS fibers (which are dorsally located in the DVC) and DMX neuron soma (which are ventrally located) are well conserved in a single slice; and 3) recordings from the dm-cNTS and the DMX neurons without repositioning of the stimulation electrode in the same slice are possible, which allows for direct comparison of the synaptic characters between them. All of these are impossible or extremely difficult in horizontal slices . We used the caudal part of the NTS (cNTS) that is mostly involved in circulatory, respiratory and digestive functions . The slices were first incubated in a holding chamber with "standard" ACSF (the concentrations of CaCl2 and MgCl2 in the cutting ACSF were changed to 2 mM and 1.3 mM, respectively) at 37°C for 30-45 min and then kept at room temperature (25°C) in the same chamber for 0.5-8 hr until recording. One slice was transferred to a recording chamber (approximately 0.4 ml volume), submerged and continuously superfused with standard ACSF at a rate of 1-2 ml/min.
The whole-cell transmembrane current was recorded from small second-order dm-cNTS neurons and DMX neurons that were visually identified with an upright microscope (BX-50, Olympus) with infrared differential interference contrast optics (IR-DIC). The reason for the selective recording from the dorsomedial part of the cNTS was threefold. First, this part contains (but not exclusively) a large number of neurons receiving afferents from gastrointestinal regions , which might be suitable to compare with DMX that also receives mostly (but not exclusively) the gastrointestinal afferents. Second, this part roughly corresponds to "area 1" defined by Okada et al.  in which small neurons expressing VGLUT2 and GAD67 are localized, allowing comparison with a limited and defined group of cNTS neurons. Third, this part is similarly distant from the TS, at which the stimulation electrode was positioned, to that for the DMX, allowing the direct influence of stimulation current flow to the neuron being recorded to be similar and minimum. Neurons showing DiI fluorescence were identified prior to electrophysiological recordings using fluorescence microscopy . Patch-clamp electrodes were made of borosilicate glass pipettes (1B120F-4, World Precision Instruments). The composition of the internal solution was (in mM) 120 potassium gluconate, 6 NaCl, 1 CaCl2, 2 MgCl2, 2 ATP magnesium, 0.5 GTP sodium, 5 EGTA, 10 HEPES hemisodium, 12 phosphocreatine disodium (pH was adjusted to 7.2 with KOH; osmolarity, ~303 mOsm kg-1). In one subset of experiments, Alexa Fluor 488 or 568 (Invitrogen) was added to the internal solution and fluorescence was visualized with a confocal scanning microscope (FV300, Olympus, Tokyo, Japan). In another series, neurobiotin (5 μg/μl; Vector Laboratories) added to the internal solution and the fluorescence was visualized following reaction with streptavidin-conjugated AlexaFluor 488 (Invitrogen) after fixation with 4% paraformaldehyde in 0.02 M PBS overnight at 4°C. The tip resistance of the electrode was 4-8 MΩ with the intracellular solution above. Neurons with either break-in membrane potential above -50 mV or non-overshooting action potential were discarded, but such neurons were quite rare in our slice preparations and IR-DIC-based pre-selection of healthy-looking cells. The longitudinal soma size (Table 1) was measured on the basis of the IR-DIC images focused at the circumference of the soma (e.g., Figure 6A, middle), captured at 0.336-μm/pixel resolution (640 pixel by 480 pixel), and stored on a computer, therefore providing a rough estimate of the soma diameter projected on the coronal plane. All DMX neurons showed smaller-or-larger delayed excitation upon depolarization from a hyperpolarized level . The membrane current was recorded with an Axopatch 200B amplifier (Axon Instruments), low-pass filtered at 2 kHz and sampled at 4-20 kHz with a PowerLab interface (ADInstruments) together with the holding potential, stimulation timing pulse and TTL pulse from the electromagnetic valve controller used for drug application (VC-6, Warner Instruments). The membrane potential was held at -70 mV (liquid junction potential compensated) during the voltage-clamp recordings. All recordings were made at room temperature (20-25°C).
The tip of a bipolar concentric electrode (interpolar distance, 100 μm) was placed on the TS (Figure 1A). The distances between the soma of the recorded neurons and the tip of the stimulation electrode were 546.9 ± 22.1 μm and 414.1 ± 7.7 μm for dm-cNTS and DMX neurons, respectively (mean ± SEM; dm-cNTS, n = 33, range, 216.9-571.6 μm; DMX, n = 105, range 277.7-596.7 μm; as measured based on the digitally captured images of the slice with a spatial resolution of 0.336 μm/pixel). The short-term plasticity of all neurons recorded was evaluated using the paired-pulse ratio (PPR), the ratio of the EPSC amplitude evoked by the second stimulus (designated as EPSC2) to that evoked by the first stimulus (EPSC1), in response to two stimuli delivered to the TS with a 100-ms inter-stimulus interval (ISI). This paired-pulse stimulation was delivered at an interval of 10 s according to the recovery time course of the paired-pulse depression (See Figure 3). In a series of experiments, ISI was varied from 20 ms to 10 s (e.g., Figure 3). The intensity of the stimulation was set at the minimum that yielded a submaximal EPSC amplitude (0.01-2.5 mA, 100 μsec; called as "submaximal stimulation"; see Figure 2D). Because this pulse width of 100 μs is smaller than or equal to the sampling interval of the data acquisition, the stimulus artifacts are not faithfully represented by the digitally converted traces and therefore the artifacts in the traces are digitally masked in the figures. To evaluate the "frequency-dependency index" (fdi) of the synapse between the TS and the recorded neuron, the TS was stimulated at 0.1, 0.5, 1, 5, 10 and 20 Hz for 15 pulses at 1-mA intensity. The same stimulation protocol was carried out 3-4 times. The fdi was defined as follows: first, the mean amplitudes of five consecutive responses (EPSC5 to EPSC9) were normalized to EPSC1 in response to TS stimuli repeated at 20 Hz and 0.1 Hz. Then, their ratio (mean relative EPSC5-9 at 20 Hz divided by that at 0.1 Hz) was defined as fdi. This value represents the reduction in the EPSC amplitude after the 5th stimulus when stimulated at 20 Hz compared to stimulation at 0.1 Hz. In order to record postsynaptic responses evoked by an action potential generated in a single fiber located precisely at the place of stimulation, we performed "minimal" bipolar stimulation . The stimulation pipettes were fabricated from theta capillaries (WPI) with similar pulling procedures to those used to fabricate patch electrodes (tip diameter, ~1-2 μm), which were filled with ACSF and placed on the TS (see e.g., Figure 5). The stimulus intensity was gradually increased to a fixed value at which EPSC appeared in an all-or-none manner with consistent amplitude and waveform (usually 20 -120 μA).
Twenty-two Wistar rats (10-15 days-old) were lightly anaesthetized in an anesthesia box filled with with diethyl ether vapor for initial sedation for a few tens seconds and mounted in a prone position. Rats underwent subdiaphragm laparotomy under continuous isoflurane (1.5-2% in 100% O2) inhalation through a facial mask. Rats breathed spontaneously. Only when a spontaneous movement was detected, the concentration of isoflurane was eventually increased to 3%. The anterior gastric branches of the vagus nerve were isolated and a part (1-2 mm-length) of the nerve was covered with a cylinder-shaped silicone tube (inner diameter of 0.4 mm), inside of which was filled with1,1'-dioctadecyl-3,3,3',3'-tetramethylindocarbocyanine perchlorate (DiI) crystals mixed with DiI gel (Invitrogen, NeuroTrace DiI tissue-labeling paste), and the both ends of this tube were closed also with dental silicone (GC Corporation, Tokyo, Japan; Exafine injection) and the wound was sutured with 5-0 silk threads. The DVC slices were made 10-15 days after the operation. The DiI-positive cells were identified with epifluorescence optics for targeted patch-clamp recordings (e.g., Figure 6).
The variance (V) and the mean (M) of the amplitude of TS stimulation-evoked EPSCs over 49-96 trials (stimulation at every 10 s) recorded at different [Ca2+]o (0.25, 0.5, 1.0, 2.0 and 5.0 mM) were calculated. Assuming that the number of vesicles released from N release sites is governed by a binominal distribution, V is expressed as a quadratic function of M as follows: V = qM-M2/N, where q is the quantal postsynaptic response evoked by a single vesicular release. With theoretical constraints that V = 0 at M = 0 (zero variance and zero mean amplitude when no vesicle is released) and M = Nq (each whole N release site releases one vesicle, each causing q), the N and q are estimated by fitting this function on actual values obtained at different release probabilities (i.e., at different [Ca2+]o . The sum of [Ca2+]o and [Mg2+]o was kept constant when manipulating [Ca2+]o. There is no existing mathematical method to examine the applicability of this method to actual synapses in the central nervous system that contain multiple release sites with heterogeneous properties. However, we consider that the synaptic transmission is well represented by this model because, in any of the neurons examined, the values obtained from experiments were well fitted by the theoretical quadratic function above with little deviation (see Figure 6Ab-d).
Picrotoxin (100 μM), strychnine HCl (1 μM) and MK-801 (20 μM) were added to the ACSF in all experiments to block GABAA/C, glycine and NMDA receptors, respectively. In a subset of the experiments, tetrodotoxin citrate (TTX, 1 μM) was also added to the ACSF after the recording of evoked postsynaptic currents to calculate the artifact waveform, which was subtracted from the averaged postsynaptic responses recorded in the absence of TTX. TTX, picrotoxin, strychnine hydrochloride, adenosine, and (+)MK-801 were purchased from Sigma. Other compounds were purchased from Sigma and Wako Pure Chemical Industries (Osaka, Japan).
The recorded membrane current was analyzed off-line with the procedures written by F.K. running on Igor Pro 5 (WaveMetrics, Oregon, USA). The values are expressed as mean values ± standard error of the mean (S. E. M.). Differences in the values were compared with the Mann-Whitney U-test, the Kruskal-Wallis test, the ANOVA test, t-test or the Kolmogorov-Smirnov test (for distributions comparison). Differences with a probability (p) less than 0.05 were considered significant.
artificial cerebrospinal fluid
caudal part of the nucleus of the solitary tract
dorsomedial part of the caudal nucleus of the solitary tract
dorsal motor nucleus of the vagus nerve
dorsal vagal complex
infrared differential interference contrast
the fourth ventricle
number of release sites
paired-pulse ratio of the success rate
paired-pulse ratio of the EPSC amplitude
the solitary tract
hypoglossal motor nucleus
This work was supported by Grant-in-Aid for Scientific Research on Priority Areas - Elucidation of neural network function in the brain - from the Ministry of Education, Culture, Sports, Science and Technology of Japan (17023042, 18021036 and 20021026), Grants for the Research on Health Sciences focusing on Drug Innovation from The Japan Health Sciences Foundation and the Grants from The Danone Institute Japan Foundation to F. K. The expert assistance in the experiments and during preparation of the manuscript by T. Matsuo-Tarumi is acknowledged.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.