Discussion
Water delivered to the oropharyngeal cavity evoked activity in a diverse group of neurons in the NTS and PbN, the first and second central gustatory relays. About a third of cells in each structure (30 of 91, 33% in the NTS; 17 of 44, 39% in the PbN) responded to water either preceding or following a taste stimulus. Three water responsive cell types were observed in both the NTS and PbN. These were excitatory, including water specialists, inhibitory, and conditional. Both excitatory and inhibitory responses to water when presented alone and/or following a taste stimulus were seen. The majority of cells that showed excitatory responses to water, found almost exclusively in the PbN, actually responded more to water than to any taste stimulus. Four cells were water specialists, responding exclusively to water. In conditional water cells, water responses were significantly higher after delivery of a subset of taste stimuli. A separate group of cells were classified as “somatosensory” because they responded equivalently to all water and taste stimuli. These findings, along with data from the existing literature, provide evidence for the idea that water is encoded by a separate information “channel” that begins in the taste receptor and is transmitted through the gustatory neuraxis along with information about other taste stimuli. The view that water is an independent taste modality is consistent with the idea that the function of the gustatory system is to detect and identify chemical stimuli that are essential for survival.

Water as an independent taste modality
As support for the classification of water as an independent taste quality, we argue three lines of evidence. First, the existence of a discrete and dedicated transduction mechanism for water in the oropharyngeal cavity provides a basis for peripheral sensitivity underlying central neural responses. Second, electrophysiological responses in dedicated (specialist, i.e., exclusively responsive to a single taste quality) taste-related peripheral nerves argue that the sensation of water is transmitted to the central nervous system over a separate information channel. Third, data showing responses in water specialist cells in central gustatory-related structures provide strong evidence that the central representation of water is distinct from that associated with other taste qualities.
Recently, a dedicated transduction mechanism for water (or hypo-osmolarity), has been identified in the mammalian oropharyngeal cavity. Water enters directly into the taste receptor cell through a channel called an aquaporin (Watson et al., 2007). Several types of aquaporins are expressed by taste receptor cells including AQP5, which is expressed on the apical membrane, and AQP1 and AQP2 which are expressed on the basolateral membrane (Gilbertson et al., 2006). When water enters taste receptor cells they swell. This activates volume-regulated anion channels and results in cell depolarization (Gilbertson, 2002; Gilbertson et al., 2006). Depolarization leads to activation of voltage-sensitive calcium channels that facilitate neurotransmitter release (Gilbertson et al., 2006). Thus a mechanism exists in mammals for transduction of water, apart from other taste stimuli.
Water is also an effective stimulus in peripheral nerves that respond to more traditional taste qualities. Specifically, water responses have been observed in the chorda tympani nerve (a branch of the facial nerve innervating taste buds on the rostral 2/3 of the tongue) of the rat, cat, and dog (Pfaffmann and Bare, 1950; Liljestrand and Zotterman, 1954; Zotterman, 1956), superior laryngeal nerve (a branch of the vagus nerve innervating taste buds on the palate) of the rat (Shinghai, 1980; Hanamori, 2001) and glossopharyngeal nerve (innervating taste buds on the caudal 1/3 of the tongue) of the frog, hamster, and rat (Zotterman, 1949; Hanamori et al., 1988; Frank, 1991). Water specialist fibers have been observed in the superior laryngeal nerve (Shinghai, 1980), a nerve that plays an important role in swallowing (Kitagawa et al., 2009). Interestingly, water applied to the posterior tongue/larynx in humans was shown to be particularly effective at evoking a swallowing reflex as compared to other taste stimuli (Shinghai et al., 1989). In addition, many neurons in the intermediate NTS that mediate swallowing have been shown to receive convergent input from the superior laryngeal and glossopharyngeal nerves (Ootani et al., 1995). In sum, water-responsive fibers have been described in several gustatory nerves and may play a role in the swallowing reflex.
Water-specific responses have also been observed in many gustatory processing regions of the brain including the NTS (Nakamura and Norgren, 1991), PbN (Nishijo and Norgren, 1990), thalamus (Verhagen et al., 2003), and gustatory cortex (de Araujo et al., 2003). These responses, though widely observed, are seldom described in detail. In the current investigation, we extend these observations by describing responses to water elicited in cells of the NTS and PbN. Though some of these responses may be due to tactile or thermal stimulation, we detail some responses for which these explanations are inadequate. Based on this evidence, we argue that the neural representation of water parallels that of more traditional taste qualities.
In addition to the role of water in evoking a swallowing reflex, the perceptual consequences of water taste may play a critical role in regulating fluid intake (thirst) and the maintenance of hydration. The responses described here were recorded with passive stimulus delivery in a non-deprived state and in the absence of post-ingestional effects. It is possible, however, that homeostatic variables such as thirst may modulate the responses to water. For example, in an imaging study of thirst and water taste processing in the human gustatory cortex, the primary gustatory cortex (anterior insula and frontal operculum) was activated by water regardless of thirst but the secondary gustatory cortex (caudal orbitofrontal cortex) only showed water-evoked activation in a water-deprived state (de Araujo et al., 2003). The most likely source of this water-selective cortical activation is taste-related structures that receive input from the oropharyngeal area. The ability of cortical neurons to distinguish between water and other tastes supports the idea that water may be processed as an independent taste quality.
In addition to oral sensitivity, water may also be detected by chemoreceptors located in the gut (Rozengurt and Sternini, 2007). Chemical transduction in the enteric nervous system plays an important role in gut motility, the regulation of nutrient absorption, gastric emptying and acid secretion (Baggio and Drucker, 2007). The vago-vagal reflex, essential for digestives processes such as gastric emptying, involves vagal projections to NTS and from there to the dorsal motor nucleus of the vagus which extends efferents that mediate gastric function (McCann and Rogers, 1992; Konturek et al., 2004). The gastric and intestinal mucosa is sensitive to both mechanical and chemical stimulation (Cottrell and Iggo, 1984). Studies have characterized modality specific enteroreceptors for sugars, acids, fats, amino acids as well as water (Iggo, 1957; Mei, 1978; Jeanningros, 1982; Mei and Garnier, 1986). Endocrine cells of the luminal tissue transduce mechanical and chemical stimuli via paracrine activation of vagal afferents which synapse in the caudal NTS (Zhu et al., 2001; Young et al., 2008) and spinal afferents which synapse in the dorsal horn of the spinal cord (Konturek et al., 2004). Stimulation of the small intestine of the cat with various chemical solutions has shown that water evokes large responses in the nodose ganglion containing cells of the vagus nerve (Mei and Garnier, 1986).

Water-responsive and somatosensory cell types
The idea that the taste of water is independent from that of other taste qualities represents a substantial departure from the conventional supposition that water responses are entirely somatosensory. Present data support a clear distinction between water-responsive and somatosensory cell types. Somatosensory cells appeared to respond only to the mechanical and/or thermal components of a fluidic stimulus. That is, their responses to water and tastants were indistinguishable. As can be seen in Figure 1, some of these responses were phasic, suggesting rapid adaptation, and others were more tonic, suggesting slow adaptation. The presence of somatosensory cells is not surprising given the fact that most taste-responsive cells in both NTS and PbN receive convergent mechanosensory input (Ogawa and Kaisaku, 1982; Ogawa et al., 1982, 1984; Ogawa and Hayama, 1984). In contrast, cells identified as water-responsive in the present study showed responses to water that could not be accounted for by the mechanosensory aspects of the stimulus. For example, water best cells responded more vigorously to water than to any of the four other taste stimuli even though all stimuli would be expected to evoke equivalent tactile and thermal sensations. It is possible, perhaps likely, that responses in these cells reflect the osmolarity of the solution, with water being the most hypo-osmotic. In effect, these cells might respond to taste stimuli as a mixture of water and taste stimulus, with water being the effective stimulus that drives neural activity. Water-inhibitory and conditional water cells showed water-specific inhibitory responses that could be predicted by the taste stimulus that preceded water presentation. In this case, an interaction of tastant and water, likely peripheral in nature (Bartoshuk, 1977), might be at play in these cells. In fact, it is well known that water presented after some taste stimuli, usually acid, induce tastes of their own (e.g., Oakley, 1985).

Comparison of NTS and PbN water responsivity
In general, differences between the NTS and PbN suggest that water sensibility may serve different functions in each structure. In the NTS, the relatively large proportion of somatosensory responses to water may be part of neural circuits extending from the caudal NTS that produce ingestive reflexes such as swallowing (Lang, 2009). In contrast, the significantly higher water-evoked firing rates in the PbN suggest that the PbN may be part of the chemosensory pathway for the perception of water along with other taste stimuli.
The sensitivity of gustatory NTS cells to water may play an essential role in the initiation of swallowing. The superior laryngeal nerve of the rat includes water-selective fibers (Shinghai, 1980) and has been shown to converge with the glossopharyngeal nerve onto neurons that mediate swallowing in the intermediate NTS (Ootani et al., 1995). Neurons in the rostral and ventral NTS project caudally to the intermediate NTS (Streefland and Jansen, 1999) which contains premotor neurons that initiate swallowing when water is applied to the pharynx (Lang, 2009). Physiological and behavioral evidence suggests that the chemosensory properties of water may be important in eliciting the swallowing reflex. For example, water has been shown to be an exceptionally effective stimulus for eliciting a swallowing reflex when applied to the pharynx and larynx (Storey, 1968; Shinghai et al., 1989). The extensive connectivity of the rostral NTS with the intermediate NTS, reticular formation and various oromotor nuclei may facilitate swallowing and other taste and somatosensory-mediated ingestive reflexes. Only about a third of NTS cells project directly to the PbN in rat (Ogawa and Kaisaku, 1982; Ogawa and Hayama, 1984; Monroe and Di Lorenzo, 1995), therefore at least some PbN activity may reflect taste processing that is independent of NTS input (see Di Lorenzo et al., 2009). Gustatory and hepatic-vagal afferents have been shown to converge to a greater degree in the postero-medial PbN as compared to the NTS (Hermann et al., 1983). Vagal fibers innervating luminal tissue respond robustly to water (Mei and Garnier, 1986) and may therefore contribute to the enhanced water responsivity observed in the PbN.

Water taste and “after-tastes”
Conditional water responses were more prevalent in the NTS than the PbN. This observation suggests that the NTS responses may be more closely tied to peripheral nerve input than the PbN. Psychophysical studies over the last 40 years have emphasized the interdependence of water responses and sensitivity to other taste stimuli. In 1974, Bartoshuk showed that water-evoked a taste but that it was highly dependent on adaptation to a preceding taste stimulus even inclusive of saliva. In addition, recordings from afferent taste nerves have shown that some water responses were only observed after pre-exposure to a particular taste stimulus (Bartoshuk and Pfaffmann, 1965; Bartoshuk et al., 1971). These responses were similar to those of the conditional cells reported in the current study, suggesting that responses in these cells directly reflect peripheral processing. The fact that water that follows certain tastes produces a taste sensation of its own implies that these responses to water access the information channels of the taste sensations they evoke as the signal ascends through the central nervous system.

Conclusions and caveats
The present results are not without their limitations. For example, some responses to water may have been affected by the anesthetic under which they were recorded. Urethane anesthesia delivered intraperitoneally (IP), as in the present study, has been shown to produce hyperglycemia, increased hematocrit and decreased blood plasma protein levels (Van Der Meer et al., 1975). Since polydipsia (thirst) is a symptom of hyperglycemia, it is possible that urethane anesthesia could modify responses to water and other taste stimuli.
Sampling errors might also have affected present results. That is, in the recordings of taste responses in both NTS and PbN, most water responses were observed only incidentally. That is, water responses were only noted “after the fact”, when analyses of responses to other tastants were analyzed. It is therefore possible that we missed some cells that might have been responsive to water. Even so, our evidence, and those of others who have described water responses, clearly suggests that the neural representation of water in the chemosensory pathway is weaker than that of other taste qualities. On the other hand, the consistent presence of water responses in almost every electrophysiological study of taste, including the present one, implies that sensibility to water is clearly present and may serve an important purpose.
In conclusion, we have described responsivity to water in two brainstem nuclei, the NTS and PbN, in the anesthetized rat. A novel finding of the current study is the observation of water best neurons and water specialist cells in both NTS and PbN. Importantly, these data underscore the literature showing that water is processed by cells in gustatory nuclei. This implies that water evokes a gustatory sensation that is supported by an independent neural representation, different from that of other taste qualities.