7.2.1. Central Mapping of Oral Receptive Fields

Soon after the rNST was functionally defined by extracellular recordings of neuron responses to chemicals applied to the tongue [1], the extent of the brainstem central gustatory projection was mapped [2]. This mapping study published in 1963 was remarkable for its time and consisted of sequential electrode penetrations made lateral to the midline at two planes, 2.9 and 1.5 mm rostral to the obex, and at different depths. Responses were recorded to taste, tactile, and thermal stimulations applied not only to intraoral receptive fields but also to the lateral face. The results were presented as a foldout map of the brainstem and provide the location in three coordinates of the chemosensory area of the rat brainstem. These pioneering findings were later expanded to determine if the gustatory relay had a chemotopic map in which the different taste qualities were localized in different dorso-ventral depths or areas of the rNST [3,4]. Although there were suggestions of variations in response magnitude with depth to different taste qualities, the existence of a chemotopic map similar to the tonotopic organization of the auditory brainstem nucleus was not found. However, later investigations described some level of a chemotopic organization [5].

In investigations of the rostral-caudal response pattern to tongue stimulation, it is not surprising, given the anatomically defined projection pattern, to find that the electrophysiological responses to anterior tongue stimulation are maximum in the most rostral part of the rNST, whereas the posterior tongue chemosensitivity is recorded more caudally [6–8]. Caudal and medial to the rostral tongue area is an area that responds to both anterior and posterior tongue stimulation, suggesting interactions between anterior and posterior tongue receptive fields at the level of rNST.

Although early investigators restricted their studies to the projection pattern of the anterior and posterior tongue, the gustatory system in the rat is more widespread and, in addition to the anterior and posterior tongue, includes receptive fields on the anterior hard palate including the nasoincisor ducts, the soft palate (Geshmacksstreifen), buccal wall, sublingual organ, and epiglottis. In addition, the posterior tongue field consists of taste buds in both circumvallate and foliate papillae [9,10]. Taste buds in these areas are innervated by various branches of the VIIth, IXth, and Xth cranial nerves. Despite this widespread input to the rNST, most investigators have restricted analysis to the anterior tongue field principally because of expediency. Thus, much of the information on how the rNST processes taste information is based on recordings from only a small proportion of the peripheral gustatory receptors.

However, there are a few studies in which the response of rNST neurons to stimulation of the other gustatory receptive fields has been investigated [11–14]. It was discovered that rNST neurons have convergent input from the anterior tongue and nasoincisor duct taste buds, but the responses to taste stimuli applied to these two receptive fields differ. Apparently, the nasoincisor and soft palate taste buds are especially responsive to sucrose, whereas the anterior tongue taste buds are more responsive to NaCl. Some rNST neurons respond to whole-mouth stimulation but the majority respond to circumscribed stimulation of taste buds in one of the other receptive fields [14]. Furthermore, not all rNST neurons respond to gustatory stimulation; a significant proportion respond exclusively to tactile stimulation of various receptive fields in the oral cavity, suggesting that the designation of the rNST as a gustatory nucleus is somewhat of a misnomer. Even the gustatory responsive neurons are multimodal, responding as well to thermal and mechanical stimulation. In addition to differences in the response characteristics of the gustatory and mechanoresponsive neurons, they are located in different parts of the rNST: Gustatory responsive neurons are situated medial to the solitary tract, whereas the mechanoresponsive neurons are located lateral to the solitary tract. Thus, the rNST gustatory and mechanoreceptive neurons form parallel populations with nearly identical orotopic organizations, suggesting that in addition to being involved in sensory processes, these neurons provide information on location of ingested substances.

7.2.2. Some Basic Assumptions Derived from the Extracellular Recordings

Most investigators have used extracellular recordings of rNST neurons to investigate how taste information is encoded (Chapter 5). One of the underlying rationales for the early studies was to investigate if, at the level of the rNST, the relatively quality nonspecific afferent discharge patterns were somehow sorted out to become more specific. The hypothesis was that rNST neurons would be divided into sets of neurons more specifically responsive to the basic taste qualities. In one of the first investigations [15] of the transfer of information between afferent taste fibers and rNST neurons, the discharge characteristics to a single concentration of 14 taste stimuli recorded in chorda tympani fibers was compared to the response of rNST neurons to the same stimulus set. The results demonstrated that the characteristics of chorda tympani fibers were very similar to those of the rNST neurons with respect to response specificity. Changes did occur in response frequency, probably reflecting convergence between the afferent fibers and the second-order neurons.

Many later experiments used essentially the same animal preparation but employed different analyses (e.g., see References [16] and [17]) to examine different characteristics of the neuronal discharge pattern. Usually, these investigators used a limited set of chemical stimuli, at single concentrations and at an undefined location of the rostral caudal extent of the rNST, although there are notable exceptions [18]. Mechanical and thermal sensitivity of the neurons was not usually determined. Also, because it is possible to record relatively few neurons per animal, in these experiments a large number of animals have to be used to obtain a suitable data set for analysis. Thus, the analysis relies on a limited sampling from each animal, but is analyzed as if the pooled set of neurons represents what would be recorded if all the neurons were sampled in a single animal. This becomes pertinent if the data are analyzed to look for cross-neuron discharge patterns as a way in which taste information is encoded.

At the time of the initial electrophysiological investigations of the rNST, little was known about the basic cell biology of the nucleus; electrodes were advanced into the rNST to isolate a neuron that responded to tongue stimulation with chemicals. Beginning in the mid-1980s, investigators began to describe the neuronal composition of the rNST [19–21]. As detailed in Chapter 2, neurons of the rNST have been divided into three morphological groups, one of which is comprised of interneurons. Beyond a general marker of the recording site (electrolytic lesion or dye mark), it is not possible using extracellular techniques to actually determine with any certainty the identity of the neuron that is being recorded from; thus, many of the electrophysiological recordings could have been derived from interneurons or other neurons with an undetermined projection pattern. However, the only definition of an rNST neuron is based on whether it responds to chemicals flowed over the tongue. These neurons are loosely called taste neurons, implying that they are involved in the perceptual properties of the sense of taste.

In a few later experiments, the projection pattern of the rNST neurons was determined by simultaneously electrically stimulating the pontine taste relay nucleus (PBN ) [22–24]. Thus, in a limited number of experiments, the recordings were obtained from neurons in the rNST that projected to the PBN, which presumably do participate in some aspect of the perceptual aspects of taste sensation, but does not rule out the possibility that these neurons also project or connect to neurons with other functions. In these studies, very few of the total population of rNST neurons were reported to project to the PBN (21–45% [22,23]). Anatomical studies, on the other hand, suggest a much higher number, although the counting was done at only two levels in the rNST and, therefore, did not represent a percentage of total rNST neurons [25]. In a more recent study of 101 rNST neurons that responded to tongue stimulation, 81 (82%) could be antidromically driven from the PBN [24]. However, the 101 neurons were obtained by recording from 73 animals, so that only a few rNST neurons were obtained per animal. The assumption is made that the total represents the results that would occur from recording from 101 neurons in a single animal. Differences in the results of these three studies relate to sampling problems inherent in the extracellular recordings. In the ideal experimental protocol, a large number of neurons would be sampled at both lateral-medial and rostral-caudal extents of the rNST, preferably in a single animal, to determine how the rNST processes information. The stimuli used should encompass several concentrations and include several members of the taste qualities as well as thermal and mechanical stimuli. It is also important to determine the identity and connections of the neurons being characterized. Current limitations of technology prevent such an experimental approach, although recent advances in the use of microwire and micromachined arrays of recording electrodes [26–28] may provide a promising new approach.

7.3.1. IN VITRO RECORDINGS

Neurons of the rNST isolated from their input are not spontaneously active. Using a series of positive and negative current injections to determine basic membrane properties, rNST neurons were found to have a mean resting membrane potential of about –50 mV and an input resistance of 500 MΩ [31]. There is nothing unusual about these numbers, but it was possible to divide the rNST neurons into groups based on their repetitive discharge characteristics. Different neurons responded differently to a current injection protocol consisting of membrane hyperpolarization followed by a long depolarizing current injection. For some neurons, the initial hyperpolarization had no effect on the subsequent discharge pattern resulting from membrane depolarization. In contrast, other neurons responded to this current protocol by a delay in the initiation of action potentials (termed delayed excitation), and the length of the delay was related to either the length or magnitude of the hyperpolarizing prepulse (Figure 7.2) [32].

FIGURE 7.2

Intracellular recording from an rNST neuron. This neuron was briefly hyper-polarized and then depolarized by current injection (lower trace). The hyperpolarizing prepulse results in a long delay before action potentials are initiated by the depolarization. (more...)

The regular repetitive discharge pattern of a further group of neurons was changed to a highly irregular pattern by the hyperpolarizing prepulse, and the final group produced only a short burst of action potentials when depolarized. There was no correlation between the discharge pattern of the neurons and neuron morphology [33]. The ion channels underlying these different repetitive discharge patterns have also been investigated [34]. The variety of temporal firing patterns are important determinants of the way neurons transform synaptic input into spike output and, therefore, control information processing by the rNST. Importantly, afferent gustatory information has been shown to both excite and inhibit rNST neurons [15,35]. Excitatory input to rNST neurons would depolarize the membrane and initiate action potentials, whereas inhibitory input would result in membrane hyperpolarization. Thus, the neuron group unaffected by membrane hyperpolarization would pass on the afferent action potential train relatively unchanged. On the other hand, neurons with a delayed excitation would, depending on the combination of inputs, not pass on discharge pattern arriving from the primary afferent fiber input but change the pattern of neural discharges evoked in these neurons. These results demonstrated for the first time that rNST neuron groups could potentially process gustatory information differently.

Synaptic properties of rNST neurons have been detailed in Chapter 3. Besides identifying glutamate as the transmitter between the afferent input and all morphological types of rNST neurons, the role of inhibitory synaptic activity mediated by GABA is especially interesting. For example, the rNST neurons are chronically inhibited [36], repetitive stimulation results in long-term potentiation of inhibitory activity [37,38], and all neuron morphological types respond to GABA [39]. The way that rNST neurons respond to trains of action potentials is a significant factor in how taste information is processed by the rNST and has been shown to influence the response properties of rNST neurons that respond to chemical stimulation of the tongue [40]. The role of the other neurotransmitters identified in the rNST (Chapter 3) in taste processing remains to be determined.

7.4.1. rNST Circuits Based on Current Knowledge

In early experiments, investigators had a limited knowledge of the basic rNST circuit that processed chemosensory information. It consisted of an afferent input that synapsed with NST neurons that then project to more rostral brain areas [1]. Because response frequencies of rNST neurons are higher than afferent fibers in response to the same stimulus concentration applied to the tongue, it was assumed that convergence occurred at the first central synapse in the taste pathway (Figure 7.3). This conceptual model was the circuit underlying all the early experiments. Extracellular recordings were made from an rNST neuron in response to stimulation of taste buds with taste stimuli. With the discovery of projection patterns and different neuron morphologies, the basic circuit becomes more complex especially if input is derived from both the VIIth and IXth nerves (Figure 7.4). Two neuron types are illustrated in Figure 7.4. Projection neurons either connect to the PBN or synapse with other brainstem motor systems. The connections of the local circuit neurons have never been directly investigated, although anatomical and brain slice investigations have provided some details. They are known to be GABAergic [20,41] and are, therefore, presumed to synapse on the projection neurons to maintain tonic inhibitory activity.

FIGURE 7.3

Basic circuit of the rNST that formed the basis of the early extracellular recordings. Convergent afferent input synapses on rNST neurons, which then project their output to rostral brain areas.

FIGURE 7.4

Circuit of the rNST that incorporates convergent input from both the VIIth and IXth nerves onto both projection and local circuit neurons. The projection neuron connects to either or both rostral brain areas as well as brainstem reflex connections. Filled (more...)

Ultrastructural studies using GABA immunostaining and anterograde labeling of afferent terminals indicate that inhibitory synapses occur on the terminal endings, possibly derived from the inhibitory interneurons [42,43]. Stimulation of the solitary tract in brain slices initiates mixed postsynaptic potentials in rNST neurons. By blocking either the excitatory (glutamate) or inhibitory (GABA) component, it is possible to measure the latency between the stimulation pulse and the rise time of the synaptic potential. Excitatory postsynaptic potentials were found to have a mean latency of 4.8 ms and the inhibitory postsynaptic potential of 8.8 ms [31]. This difference in latency suggests that the excitatory postsynaptic potentials are mediated monosynaptically, whereas at least two synapses are involved in the inhibitory postsynaptic potential. Hence, the interneurons inhibit the afferent input by feedback inhibition (Figure 7.5).

FIGURE 7.5

Circuit of the rNST that includes both excitatory and inhibitory synapses. Excitatory input to the local circuit neurons is diagrammed to inhibit the projection neuron, which is therefore chronically inhibited. As suggested by ultrastructural investigations, (more...)

7.4.2. Model rNST Circuits

Neurons in the rNST receive input from afferent fibers innervating receptors in the oral cavity. The response of these rNST neurons is, therefore, determined by the response characteristics of the afferent fibers that synapse with the neuron. Despite recent advances that have suggested that taste quality may be coded using labeled lines [44–46], most of the evidence based on recordings from single afferent gustatory fibers supports a population code [47]. Thus, peripheral gustatory fibers can be separated into groups according to greatest response to one of the taste qualities [48]. Recordings from rNST neurons reveal a similar grouping of responses (see Chapter 5) [16]. During stimulation, not all the afferent fibers synapsing with a neuron will be activated equally because different fibers respond differently to particular taste qualities, so the response of an rNST neuron will reflect the overall level of activity in the afferent fibers.

At present, there is no information on the number of synapses between afferent chemosensory fibers and an rNST neuron. However, the receptive field size of rNST neurons is significantly larger than those of afferent taste fibers, demonstrating that afferent fibers innervating adjacent tongue-receptive fields converge on an rNST neuron [49]. Also, we do not know how many rNST neurons are synaptically connected to a single afferent fiber, although neural tracing studies show the terminal fields to be a complex meshwork of fibers. All of these factors will determine the response characteristics of an rNST neuron, as suggested in Figure 7.6. Moreover, little is known about the stability of these synaptic connections over time. The response characteristics of the afferent fibers may change as the taste receptor cells turn over [50,51].

FIGURE 7.6

Circuit diagram that illustrates sorting of afferent fibers with similar response characteristics onto rNST neurons. rNST neurons will, therefore, have similar response characteristics to the afferent fibers. S = sucrose best fiber and rNST neuron, N (more...)

Because of this pattern of connections, the only way to alter the response profile of the rNST neurons is to change the afferent connections. This applies to all the neurons at each relay in the central taste pathway. It is not possible, for example, to modify a neuron at the PbN taste relay to change its response profile, because the connection pattern determines how the neuron responds to tongue stimulation.

Once the gustatory information has passed via the first central synapse to the second-order neuron in the taste pathway, it is uncertain what happens next, before the information is passed beyond the rNST. There is even the possibility that the rNST may be bypassed. For example, there is evidence that stimulation of the solitary tract can initiate postsynaptic potentials in some salivatory nucleus neurons with a latency of less than 4 ms, suggesting that this is a monosynaptic connection, indicating a direct reflex link between afferent input and the efferent output. The fact that the rNST contains a sizeable population of interneurons suggests processing, and as described above, inhibitory activity plays a significant role in rNST processing.

7.4.3. Separation of Function

Anatomical and electrophysiological studies reviewed above demonstrate that only a certain proportion of rNST neurons project rostrally to the PbN. This suggests that rNST neurons are separated into groups, some projecting rostrally, whereas others synapse with neurons involved in brainstem reflexes (Figure 7.7). Electrophysiological recordings from neurons that do not relay to the PbN revealed that they respond similarly to the relay neurons, albeit with some subtle differences [23,24]. Notably, rNST neurons responding best to sucrose preferentially project to the PbN.

FIGURE 7.7

Diagram illustrating separation of input onto rNST neurons grouped according to projection pattern and function.

Figure 7.7 is perhaps what many investigators assume happens to information arriving at the rNST. Converging afferent information arrives at an rNST neuron that is then distributed to rostral and brainstem sites. Conceptually, this circuit suggests that the rNST plays little or no role in sensory processing. Most of the processing takes place at the first synapse, where the rNST neuron averages all the sensory input and then sends it to various locations. Perhaps that is all that is necessary at this relay in the central taste pathway. However, because different taste qualities can result in different types and volumes of saliva, and different facial expressions, it would seem that some organization of the sensory input takes place in the rNST.