CO₂ chemoreception in cardiorespiratory control

Abstract

Considerable progress has been made elucidating the cellular signals and ion channel targets involved in the response to increased CO₂/H⁺ of brain stem neurons from chemosensitive regions. Intracellular pH (pH_i) does not exhibit recovery from an acid load when extracellular pH (pH_o) is also acid. This lack of pH_i recovery is an essential but not unique feature of all chemosensitive neurons. These neurons have pH-regulating transporters, especially Na⁺/H⁺ exchangers, but some may also contain HCO₃⁻-dependent transporters as well. Studies in locus ceruleus (LC) neurons have shown that firing rate will increase in response to decreased pH_i or pH_o but not in response to increased CO₂ alone. A number of K⁺ channels, as well as other channels, have been suggested to be targets of these pH changes with a fall of pH inhibiting these channels. In neurons from some regions it appears that multiple signals and multiple channels are involved in their chemosensitive response while in neurons from other regions a single signal and/or channel may be involved. Despite the progress, a number of key issues remain to be studied. A detailed study of chemosensitive signaling needs to be done in neurons from more brain stem regions. Fully describing the chemosensitive signaling pathways in brain stem neurons will offer new targets for therapies to alter the strength of central chemosensitivity and will yield new insights into the reason why there are multiple central chemoreceptive sites.

a major tenet of central cardiorespiratory control for nearly a century has been that central chemosensitive neurons respond to respiratory stimuli through their responses to changes of pH (19, 34, 47, 67). Over the past 15 years, fluorescence imaging techniques to measure intracellular pH (pH_i) in neurons from chemosensitive regions of the brain stem (5, 24, 53, 56, 64, 65) have enabled great strides to be made in understanding the role of pH changes in chemosensitive signaling. This minireview will provide a brief overview of that progress, highlighting our current understanding of the role played by changes of pH in central chemosensitive signaling and the implications of that understanding. The challenges that lie ahead will also be emphasized.

pH-REGULATING TRANSPORTERS IN CENTRAL CHEMOSENSITIVE NEURONS

By necessity, the vast majority of the studies of pH regulation in neurons from chemosensitive regions has been done in reduced preparations, usually brain slices (43, 53, 55), isolated ganglia (24), organotypic cultures (63, 64, 65) or in cultured neurons (5), superfused with artificial cerebral spinal fluid (aCSF). One of the earliest findings regarding the regulation of pH in these neurons is that they exhibit a maintained acidification in response to hypercapnic acidosis (HA: aCSF equilibrated with elevated CO₂ such that fluid pH, pH_o, falls) with no apparent tendency for pH_i to recover back toward its initial value (43, 53, 64, 65). This was shown not to be due to the lack of pH recovery transporters on the membrane surface since these neurons exhibited brisk pH_i recovery from acidification when pH_o was held constant using isohydric hypercapnic (IH) solutions (aCSF containing elevated HCO₃⁻ and equilibrated with elevated CO₂ such that pH_o remained constant) (24, 53). The explanation for these findings is that neurons from chemosensitive regions clearly have pH-regulating transporters in their surface membranes but the activity of these transporters is suppressed during periods of acid challenge involving a fall of pH_o (53).

Early work showed that the amiloride-sensitive Na⁺/H⁺ exchanger (NHE) played the major role in recovery from an acid load in these neurons. Whether neurons from chemosensitive regions were acidified using IH solutions or using an NH₄Cl prepulse (4), the recovery from that acidification was largely inhibited by amiloride (an inhibitor of NHE) and largely unaffected by DIDS (an inhibitor of HCO₃⁻-dependent membrane transporters). This was shown for rat neurons from the nucleus tractus solitarius (NTS) (53), the ventrolateral medulla (VLM) (53, 64, 65), and the retrotrapezoid nucleus (RTN) (43), and from snail neurons from the subesophageal ganglion (24).

There are at least nine isoforms of NHE, with only NHE1–5 thought to be present in the brain of rats (1, 9, 35). NHE1 is believed to be ubiquitous and is sensitive to inhibition by amiloride (46). Low levels of expression of mRNA for NHE2 have been reported for the brain stem (35), but so far no functional NHE2 has been described in neurons from this region. Very low levels of expression for NHE3 in brain stem have also been reported (35), but physiological evidence suggests that this isoform may play an important role in central chemoreception (see below). NHE4 activity has only been reported for hippocampal neurons (3), and its expression level is very low in the brain stem (35). Finally, NHE5 has been shown to be highly expressed in rat brain (1) although studies of its significance have been hindered by the lack of specific inhibitors for this isoform.

The nature of the NHE transporters present in neurons from chemosensitive regions has been studied with pharmacological agents (31, 63, 64). In a recent paper, Kersh et al. (31) used isoform-specific inhibitors to study the presence of NHE isoforms in neurons from three different chemosensitive regions. NHE1 was the predominant isoform present in RTN neurons, accounting for about half of the pH_i recovery from acidification, while NHE3 appeared to play a small role (5–10% of recovery). The remainder of the recovery may be due to NHE5 (31). A similar pattern was seen in NTS neurons, where ∼66% of recovery from acidification is mediated by NHE1 and ∼25% is mediated by NHE3 (31). Surprisingly, studies of pH recovery in neurons from another chemosensitive region, the locus ceruleus (LC), revealed no amiloride sensitivity to the pH recovery from acidification, suggesting that NHE isoforms play at most a small role in pH_i recovery from acidification in LC neurons (31).

Our findings of at most a small contribution to acid extrusion by NHE3 in brain stem neurons is in agreement with the findings of Ma and Haddad (35) of a low level of expression of this isoform in brain stem. However, they are at odds with physiological data suggesting a role for NHE3 in CO₂ sensing by the brain stem. In organotypic cultures of VLM neurons, an inhibitor of NHE3 resulted in neuronal acidification and an increase in firing rate (64), and this increase in firing rate was higher if both mild hypercapnia and the inhibitor were used together (63). These studies have been extended to show that exposure of rabbits to a blood-brain barrier-permeable form of an NHE3 inhibitor results in increased ventilation (32), and infusion of an NHE3 inhibitor increases respiratory frequency in anesthetized rats (50). Although these findings appear to conflict with our data showing little NHE3 activity in brain stem neurons, we believe that the two sets of data are consistent. The NHE3 inhibitors used in the animal studies may well have effects on other components of the respiratory network such as the pre-Bötzinger complex (50), which helps to set respiratory rhythm (21) and astrocytes, which affect central chemosensitivity (20, 27). Further, the NHE3 inhibitors appear to increase ventilation less than inspired hypercapnia (32). These findings suggest that NHE3 plays a role in increasing firing rate in only a limited number of central chemosensitive neurons while neurons from other central chemosensitive regions do not contain NHE3. This interpretation is consistent with the distributed nature of the central chemosensitive network (40).

The other major group of pH-regulating transporters is the HCO₃⁻-dependent transporter family (9, 46, 57). There are many members of this class of transporters, including alkalinizing Na⁺-HCO₃⁻ cotransport (NBC: both electrogenic, NBCe, and electroneutral, NBCn) and Na-driven Cl⁻/HCO₃⁻ exchange (NDCBE) as well as acidifying Na-independent Cl⁻/HCO₃⁻ exchange (AE, anion exchange). While NHE appears to play a major role in pH_i recovery in acid loads from brain stem neurons, there is evidence for the presence of functional HCO₃⁻-dependent transporters as well. The best evidence is for the presence of AE in brain stem neurons (52). In response to an alkaline load at constant pH_o, neurons from the VLM, but interestingly not the NTS, exhibit DIDS-inhibitable acidification. Further, removal of external Cl⁻ results in DIDS-inhibitable alkalinization (most likely due to reversal of AE in the absence of external Cl⁻) in VLM but not NTS neurons (52). These data indicate that in neonatal rats, an acidifying AE is present in VLM but not NTS neurons. This is consistent with the fact that NTS neurons had a more alkaline resting pH_i than VLM neurons (53), suggesting that the lack of acidifying AE in NTS neurons results in an elevated resting pH_i. We have recently found that in adult rats the resting pH_i of NTS neurons is lower than in neonates (41) and that these neurons appear to have AE activity (Martino PF and Putnam, unpublished observations). Interestingly, the pH_i response of chemosensitive neurons to alkalinization (such as seen with hypocapnia) has not received much attention.

The presence of alkalinizing HCO₃⁻-dependent transporters in response to acidification in neurons from chemosensitive regions is less clear. As mentioned above, LC neurons did not exhibit significant amiloride-inhibitable pH_i recovery from acidification. However, the recovery from acidification was enhanced in the presence of HCO₃⁻, compared with its absence, suggesting the presence of an alkalinizing HCO₃⁻-dependent transporter in LC neurons (31). Since DIDS did not reduce recovery (31) it appears that LC neurons contain an alkalinizing DIDS-insensitive HCO₃⁻-dependent transporter. Although this transporter has not been identified in LC neurons, one such transporter is an electroneutral variant of NBC, NBCn1, which has been shown to be largely unresponsive to inhibition by DIDS (10). Given the supposed importance of NDCBE for neuron pH regulation (8) it is likely that at least some neurons from some other chemosensitive regions will ultimately be shown to contain functional NDCBE and perhaps other HCO₃⁻-dependent transporters.

It was hoped that a lack of pH_i recovery from hypercapnia-induced acidification would be characteristic of chemosensitive neurons, but several lines of evidence suggest that this is not the case. In regions like the VLM or NTS, where only about 30–50% of the neurons are CO₂ sensitive (i.e., increase their firing rate in response to hypercapnia) (12, 14, 17, 41, 62), virtually all neurons from these regions exhibit a lack of pH_i recovery from hypercapnic acidosis (53). Further, although neurons from nonchemosensitive regions, such as the inferior olive and the hypoglossal, of young neonates exhibit pH_i recovery in response to hypercapnia-induced acidification, neurons from these regions in older neonates [older than postnatal day 11 (>P11)] no longer show recovery from hypercapnia-induced acidification (43). In other words, the pH_i response to hypercapnia was similar in neurons from both chemosensitive and nonchemosensitive regions. Finally, in a direct comparison of cultured neurons from the chemosensitive raphé region vs. neurons from the nonchemosensitive hippocampal region, it was found that all neurons from all regions (both chemosensitive and nonchemosensitive) did not exhibit pH_i recovery from hypercapnic acidosis (5). Thus it is clear that a lack of pH_i recovery in response to hypercapnia-induced acidification is not a unique characteristic of chemosensitive neurons.

ROLE OF pH CHANGES IN NEURONAL CHEMOSENSITIVE RESPONSES

Hypercapnic acidosis could stimulate chemosensitive neurons through several potential signals, including increased CO₂ per se, increased intracellular HCO₃⁻, or decreased pH_o or pH_i (47). Most of the focus of research in chemosensitive signaling has been on the changes of pH. The main thrust of these studies is whether it is pH_o or pH_i that is the major factor in chemosensitive signaling. A typical approach is to study the firing rate response of chemosensitive neurons to various acid solutions, some with elevated CO₂ and others not, and some with decreased pH_o and others not. Such studies in LC (22) and VLM (63) neurons found that the factor that best correlates with the increased firing rate is the decrease of pH_i and not pH_o. A similar conclusion was made for raphé neurons (60) although pH_i was not measured in that study.

Despite the strong correlation between increased firing rate and decreased pH_i, a simple relationship between these two parameters is problematic. One issue is that the increased firing rate appears to saturate, reaching a maximum in LC neurons with stimulation by solutions equilibrated with 15% CO₂ (54), despite the fact that pH_i continues to become more acid when solutions are equilibrated with values of CO₂ higher than 15% (16) (Fig. 1). This saturation of the CO₂ effect is also seen with increased phrenic nerve activity in response to increased inspired CO₂ (11). It might be expected that the chemosensors would saturate at such pathologically high levels of CO₂, but the mechanistic basis of this saturation of the chemosensitive response to very high levels of CO₂ (and thus lower levels of pH_o and presumably pH_i) has not been addressed. A lack of a straightforward relationship between the changes of pH_i and firing rate is also suggested, in LC neurons, by the fact that the decrease in pH_i is larger in response to hypercapnic acidotic solutions compared with isohydric hypercapnic solutions, but the increase in firing rate is similar in response to these two acidic stimuli (22). The same result was reported for RTN neurons (56).

Fig. 1.

A graph showing the effect of increased CO₂ [equilibrated with artificial cerebral spinal fluid (CSF) on intracellular pH (pH_i) (○; left axis)] and on the increased firing rate of locus ceruleus (LC) neurons (●; right axis). The values for pH_i were taken from Dean et al. (16) and represent the decrease in pH_i from the initial value of pH_i (at 5% CO₂) of ∼7.25. The %increase in firing rate values were taken from Ritucci et al. (54) and represent the %increase in firing rate compared with control firing rate of ∼1 Hz (at 5% CO₂). Note that the firing rate response reaches a plateau at 15–20% CO₂ while pH_i continues to decline up to 50% CO₂.

The problem with using acid solutions of various compositions to study chemosensitive signaling is that while acid challenges can be studied at constant CO₂ or constant pH_o, all acid challenges involve a decrease of pH_i. To test whether a decrease of pH_i is necessary for chemosensitive signaling, we developed a technique to clamp pH_i at normal control values (measured in 5% CO₂) during exposure to HA solutions (25). When LC neurons were exposed to HA solutions with pH_i clamped to control values, firing rate still increased, clearly indicating that a change of pH_i is not required for chemosensitive signaling in LC neurons. Interestingly, firing rate does not increase in response to IH solutions (which have the same pH_o as control pH_o) when pH_i is also clamped to control values (Hartzler LK, Dean JB, and Putnam RW, unpublished observation). These findings clearly indicate that a decrease of pH_i, a change of pH_o, or both are sufficient to increase LC firing rate, but that an increase of CO₂ alone is not sufficient to increase firing rate in LC neurons. These findings have led us to propose a multiple-factors model for chemosensitive signaling in LC neurons (47). It is not known at this point whether a similar signaling pattern occurs in neurons from other chemosensitive regions or whether the components of chemosensitive signaling are dependent on the neurons being studied.

Another issue in cellular chemosensitive signaling is where on the neuron acid stimuli are sensed. In neurons, the critical changes of pH and the ion channel targets of those changes could reside on the soma, on the dendrites, or throughout the cell. As we study more fully the cellular basis for chemosensitivity, this question takes on greater importance since the pH_i responses of dendrites to acidic stimuli can differ in magnitude and kinetics from the responses to the same stimuli in the soma (58, 66) and the distribution of channels may differ between soma and dendrites (33, 49). Based on neuronal morphology, the argument has been made that chemosensitive signaling occurs mainly on specialized regions of the dendrites. Neurons from the VLM seem to have at least one dendritic process that projects toward the ventral medullary surface (30, 44), suggesting that this process may be “sampling” fluid chemistry at the medullary surface. A similar morphology has also been observed in RTN neurons, which send dendritic projections to the ventral medullar surface that run considerable distances along the surface limiting layer, thus exposing a large dendritic surface area to the surface solution (37). A variation on this theme is the observation that chemosensitive serotonergic neurons send dendritic projections that make close contact with large medullary arteries, suggesting that these projections are sensing the chemistry of the arterial blood (6). Based on these morphological studies, it appears that the dendrites of at least some chemosensitive neurons play a critical role in chemosensitive sensing.

The role of the soma vs. the dendrites has only been directly addressed experimentally in LC neurons (54). In these studies, an LC neuron within a brain stem slice was patched and loaded with fluorescent dye so that both the soma and dendrites could be visualized. A second, larger pipette containing aCSF equilibrated with 100% CO₂ and containing a fluorescent dye for visualization was brought close to the surface of the patched neuron. Acidic solution was continuously puffed on only a large dendrite (as determined by the area of spread of the fluorescent dye in the pipette) or only on the soma (and the initial segments of the primary dendrites), and the firing rate response was measured. Surprisingly, only when acidic solution was placed on the soma did the LC neuron firing rate increase (54). There was no increase in firing rate if normal aCSF (equilibrated with 5% CO₂) was puffed on the neuron. Further, the increased firing rate of LC neurons in response to a regional acidification only on the soma was as large as the maximal firing rate response to a generalized acidification due to exposing the whole slice to aCSF equilibrated with 15% CO₂ (54). These findings clearly indicate that the chemosensitive machinery in LC neurons resides within or very close to the cell soma. It may well be that LC neurons are sampling the local extracellular space and their cellular localization of the chemosensory machinery differs from that in neurons from other chemosensitive regions. Whether LC neurons are unique in having somatic chemosensitive signaling or whether this is a generalized finding in all chemosensitive neurons must await further experiments, of the type described here, in neurons from other chemosensitive regions such as the NTS, RTN, and VLM.

ION CHANNEL TARGETS OF CHEMOSENSITIVE SIGNALING

It is widely assumed that the targets for hypercapnia-induced acidification are pH-sensitive ion channels, especially K⁺ channels (29, 47). The assumption is that acid pH, either pH_o or pH_i, results in inhibition of K⁺ channels, neuronal depolarization, and increased firing rate. There are many candidate pH-sensitive channels, most of which are K⁺ channels (29, 47). Possible K⁺ channels involved in chemosensitive signaling include inwardly rectifying K⁺ channels (Kir) (45), especially the heteromeric Kir4.1–5.1 channel with a pK of ∼7.4 (13, 69), TASK channels (2), the rapidly activating and inactivating A current (15, 18, 48), and a delayed-rectifying K⁺ channel (18, 48).

Three main approaches have been adopted to study possible channels involved in the response of chemosensitive neurons: the use of channel inhibitors; direct measurements of the effects of pH on channel function using voltage clamp; and studies of the response to CO₂/H⁺ in neurons from knockout mice. Based on the evidence to date, it appears that the number and types of channels involved in chemosensitive signaling differ in neurons from different chemosensitive regions (Fig. 2).

Fig. 2.

A summary figure of the ion channels and cellular signals involved in the chemosensitive response of neurons from 4 different brain stem regions or from the snail Helix aspersa. − indicates that the channel is inhibited while + indicates that it is activated by the stated signal. ? indicates the identity of the channel or of the signal is uncertain. Note that some neurons, like those from the LC or from snail, seem to involve multiple signals and multiple ion channels in chemosensitive signaling, while neurons from other regions such as the nucleus tractus solitarius (NTS) and retrotrapezoid nucleus (RTN), appear to involve one ion channel and one chemosensitive signal. In neurons from the raphé region, the ion channel responsible for chemosensitive signaling appears to be developmentally dependent, with TASK channels predominating in rodents younger than postnatal day 12 (<P12) while a calcium activated nonselective cation current (CAN current) mediates signaling in animals older than P12 (>P12). Kir, inwardly rectifying K⁺ channel; K_Ca, Ca²⁺-activated K⁺ channel; Kdr, delayed rectifying K⁺ channel; sAC, soluble adenylate cyclase.

In LC neurons, evidence suggests that many different channels are involved in the firing rate response to acidic stimuli. Inhibitor studies have revealed that a tetraethylammonium (TEA)-sensitive channel is active at resting pH and is inhibited by pH, and that 4-aminopyridine (4-AP) and nifedipine reduce but do not eliminate CO₂/H⁺-induced increased firing rate (23, 36, 48). These findings suggest that the chemosensitive response in LC neurons involves L-type Ca²⁺ channels and TEA- and 4-AP-sensitive K⁺ channels. Voltage-clamp studies have confirmed that LC neurons contain CO₂/H⁺-inhibitable transient (4-AP-sensitive current, which is likely an A current) and sustained (TEA-sensitive current, which may be due to a delayed-rectifying K⁺ channel) currents (48). The A current in LC neurons is largely inhibited by decreased pH_o (Li K-Y and Putnam, unpublished observation), and we speculate that the TEA-sensitive sustained current is inhibited by decreased pH_i. In addition, LC neurons have been shown to have CO₂/H⁺-inhibitable Kir channels (45). These channels are inhibited by a decrease in pH_i, which may change the charge state on intracellular polyamines thereby decreasing the activity of Kir channels (45). In LC neurons, pH_o-sensitive TASK channels are strongly expressed and may play a role in chemosensitive signaling as well. Interestingly, it appears that L-type Ca²⁺ channels are activated by increased CO₂/H⁺ in LC neurons (23), although the basis for this increase is not known. As shown in peripheral chemoreceptive cells (59), it may be that increased CO₂ results in an increase in intracellular HCO₃⁻, which activates soluble adenylate cyclase, cAMP production, and protein kinase A activity. This leads to phosphorylation and activation of L-type Ca²⁺ channels. Thus it appears likely that the response to CO₂/H⁺ in LC neurons involves multiple ion channels, perhaps as many as five different types.

A similar finding has previously been made in chemosensitive neurons from the snail Helix aspersa (18). These snails were found to contain three pH-sensitive K⁺ currents: a transient current (likely an A current), a delayed rectifying K⁺ current, and a calcium-activated K⁺ current. Thus, like LC neurons, chemosensitive neurons from the snail appear to involve increased CO₂/H⁺ modulating multiple ion channels.

A different picture is indicated for chemosensitive NTS neurons. In these neurons, the increased firing rate response is entirely eliminated by 4-AP (15, 36). Although voltage clamp has not been performed on these chemosensitive neurons, it appears, based on the inhibitor studies, that an A current may be solely responsible for the firing rate response of chemosensitive NTS neurons to CO₂/H⁺. It is not known whether the A current in NTS neurons is largely inhibited by pH_o or pH_i. Thus, unlike chemosensitive LC and snail neurons, NTS neurons may involve a single chemosensitive signal acting on a single ion channel.

Yet a different pattern for hypercapnia-induced increased firing rate is indicated for chemosensitive neurons from the medullary raphé. In young neonatal mice [younger than postnatal day 12 (P12)] that have had TASK channel isoforms 1, 3, or both 1 and 3 knocked out, the increased firing rate in response to elevated CO₂/H⁺ in serotonergic raphé neurons is eliminated (38). This suggests that chemosensitivity arises from TASK-1 and TASK-3 in raphé neurons from mice younger than P12. It may be that these two isoforms make a heteromeric channel that is inhibited by decreases in pH_o (38). The magnitude of the chemosensitive response in raphé neurons from these young mice is very small (38), and in fact Hodges et al. (26) recently showed that a large chemosensitive response to hypercapnia is only seen in raphé neurons from mice older than P12. Further, the large hypercapnia-induced increase in firing rates of raphé neurons from older neonates appears to be due to an acid-sensitive calcium-activated nonselective cation (CAN) current (68). It has not been determined whether this channel is sensitive to changes of pH_o or pH_i. The chemosensitive response of raphé neurons, therefore, appears to be developmentally dependent, with a small chemosensitive response, perhaps dependent on TASK channels in young neonates, that changes to a larger response dependent on a CAN current after age P12 (Fig. 2).

Interestingly, in these same TASK-1 and -3 knockout mice, the chemosensitive response to increased CO₂/H⁺ is unaffected in RTN neurons (38). This indicates that TASK-1 and -3 are not involved in the chemosensitive response of RTN neurons. Nevertheless, the acid-sensitive channel that is responsible for the chemosensitive response of RTN neurons has a current-voltage plot that indicates that it is a background K⁺ channel (38), either one that is not a TASK channel or an isoform of TASK channels that is neither TASK-1 nor TASK-3. It is unclear whether the chemosensitive channel(s) in RTN neurons is/are responsive to changes of pH_o or pH_i.

In summary, chemosensitive signaling in neurons from some regions (LC, snails) involves multiple channels and presumably multiple signals, while in neurons from other regions (NTS and RTN) a single channel may be responsible for chemosensitive signaling. Finally, in raphé neurons the channel responsible for chemosensitive signaling appears to change with development.

Future Studies

This brief review points to several significant unanswered questions that could serve as the focus for future studies. While we have learned a great deal about pH_i regulation in chemosensitive neurons it will ultimately be important to know which isoforms of which transporters are present in the neurons from different chemosensitive regions. This is especially true given the evidence that NHE3 may play a role in ventilatory control. The exact localization of this NHE isoform, in neurons or glia, in which neurons, and somatic vs. dendritic distribution, should be a high priority since this information could yield important insights into the role played by neurons from chemosensitive regions that contain NHE3 in ventilatory control. It is also likely that HCO₃⁻-dependent transporters play a greater role than currently appreciated in the response of chemosensitive neurons to acidic stimuli. Thus studies of the expression and activity of various HCO₃⁻-dependent transporters in neurons from different chemosensitive regions should be a focus of future research. There is already evidence that the expression of pH-regulating transporters in brain stem neurons varies with development (43; Martino and Putnam, unpublished observation), and it will be important in the future to determine developmental changes of the expression of pH-regulating transporters in neurons from the various chemosensitive regions. Finally, detailed studies such as those performed in LC neurons need to be done in neurons from other chemosensitive regions. The most important experiments will be those done with pH_i clamped during exposure to HA in an attempt to elucidate whether the neurons from various chemosensitive regions are responding to changes of pH_i, changes of pH_o, changes of CO₂, or to a combination of these stimuli.

There is also considerable work left to do identifying which ion channels are involved in the response to acid stimuli of neurons from different chemosensitive regions (Fig. 2). A major part of this work will be to further identify the cellular localization of these putative chemosensitive channels, whether on the soma, dendrites, or throughout the neuron. Especially if we find that many chemosensitive neurons are responding to acid challenges based on processes that occur on their dendrites, studies of pH_i regulation and of the proximate stimulus for these channels will need to be designed to examine pH changes and pH_i clamping on the dendritic processes, a daunting task but most likely amenable using modern technological advances in imaging.

Whatever model emerges from these studies of the signals and ion channel targets in chemosensitive neurons will have to be able to explain the apparent saturation of the firing rate response of these neurons to stimuli of increasing strength (Fig. 1). This saturation phenomenon may arise from a diminishing ability of decreasing pH to inhibit K⁺ channels further or it may be due to activation of some process that limits further increases in firing rate with increasing acidic stimulation. Such a process may involve the Ca²⁺-activated K⁺ channel (K_Ca) since it has been shown in LC neurons that hypercapnic solutions activate L-type Ca²⁺ channels (23). At some point this process could result in sufficient intracellular Ca²⁺ accumulation to activate K_Ca, limiting further membrane depolarization and increased firing rate. The mechanism by which increased acidic stimuli saturate the firing rate or the ventilatory response is an important process to elucidate.

Finally, the focus of this review has been solely on neurons that are activated by hypercapnia, but in several brain stem regions neurons have been described that are inhibited by hypercapnia (12, 17, 28, 41, 51, 61, 62). There has yet to be a cellular mechanism proposed by which acid stimuli result in a reversible decrease in neuronal firing rate. However, it has long been postulated that the normal response of neurons to acid environments is an inhibitory response (7), and the mechanism in neurons from chemosensitive regions may involve K⁺ channels that are activated by acidic stimuli. Neurons inhibited by hypercapnia are often present in low proportions, but recently a model system has been described in which acid-inhibited neurons constitute as much as 30% of NTS neurons (42). Such a system has a sufficient percentage of inhibited neurons to serve as an excellent model system for the study of the cellular mechanism(s) involved in the inhibition of neurons from chemosensitive regions by hypercapnia.

CONCLUSIONS

A detailed knowledge of the signals and targets that mediate the activation of chemosensitive neurons in response to hypercapnic acidosis would be extremely valuable. The signals being “sensed” will reveal a great deal about what is being regulated (e.g., ventilation or brain pH), and the ion channels that are affected by the signals will make attractive targets for therapies aimed at altering the strength of central chemosensitivity. Further, since central chemosensitivity appears to be due to a distributed network of chemosensitive neurons (40), knowledge of how changes of CO₂/H⁺ are sensed in neurons from different regions may yield insights into the role played by each of those regions in the overall control of ventilation in the organism. Finally, since it appears likely that chemosensitive signaling differs in neurons from different brain stem regions (Fig. 2), the findings to date are consistent with a distributed central pattern of chemosensitivity that arose in a hierarchical fashion, with new chemosensory components added throughout evolution as organisms passed through key transitions such as those from water to air breathing, the maintenance of a high and constant body temperature, and the development of sleep (39).

GRANTS

This work was supported by National Heart, Lung, and Blood Institute Grant HL-56683.

DISCLOSURES

No conflicts of interest are declared by the author.