HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (56) Citing Articles via Google Scholar Google Scholar Articles by Putnam, R. W. Articles by Ritucci, N. A. Search for Related Content PubMed PubMed Citation Articles by Putnam, R. W. Articles by Ritucci, N. A.

INVITED REVIEW

Cellular mechanisms involved in CO₂ and acid signaling in chemosensitive neurons

Department of Anatomy and Physiology, Wright State University School of Medicine, Dayton, Ohio 45435

	ABSTRACT

TOP ABSTRACT CO2/H+-SENSITIVE CELLS HISTORY OF STUDIES OF... REDUCED PREPARATIONS SIGNALS IN CENTRAL... OTHER FACTORS IN CENTRAL... POSSIBLE TARGETS IN CENTRAL... A REVISED MODEL FOR... OTHER ACID-SENSITIVE CELLS SUMMARY AND CONCLUSION GRANTS REFERENCES

 
An increase in CO₂/H⁺ is a major stimulus for increased ventilation and is sensed by specialized brain stem neurons called central chemosensitive neurons. These neurons appear to be spread among numerous brain stem regions, and neurons from different regions have different levels of chemosensitivity. Early studies implicated changes of pH as playing a role in chemosensitive signaling, most likely by inhibiting a K⁺ channel, depolarizing chemosensitive neurons, and thereby increasing their firing rate. Considerable progress has been made over the past decade in understanding the cellular mechanisms of chemosensitive signaling using reduced preparations. Recent evidence has pointed to an important role of changes of intracellular pH in the response of central chemosensitive neurons to increased CO₂/H⁺ levels. The signaling mechanisms for chemosensitivity may also involve changes of extracellular pH, intracellular Ca²⁺, gap junctions, oxidative stress, glial cells, bicarbonate, CO₂, and neurotransmitters. The normal target for these signals is generally believed to be a K⁺ channel, although it is likely that many K⁺ channels as well as Ca²⁺ channels are involved as targets of chemosensitive signals. The results of studies of cellular signaling in central chemosensitive neurons are compared with results in other CO₂- and/or H⁺-sensitive cells, including peripheral chemoreceptors (carotid body glomus cells), invertebrate central chemoreceptors, avian intrapulmonary chemoreceptors, acid-sensitive taste receptor cells on the tongue, and pain-sensitive nociceptors. A multiple factors model is proposed for central chemosensitive neurons in which multiple signals that affect multiple ion channel targets result in the final neuronal response to changes in CO₂/H⁺.

hypercapnia; brain stem; ventilation; peripheral chemoreceptor; glia; gap junction; glomus; channel; calcium; potassium; carbonic anhydrase; taste receptor; nociception

	CO2/H+-SENSITIVE CELLS

TOP ABSTRACT CO2/H+-SENSITIVE CELLS HISTORY OF STUDIES OF... REDUCED PREPARATIONS SIGNALS IN CENTRAL... OTHER FACTORS IN CENTRAL... POSSIBLE TARGETS IN CENTRAL... A REVISED MODEL FOR... OTHER ACID-SENSITIVE CELLS SUMMARY AND CONCLUSION GRANTS REFERENCES

 
A variety of cells in the body are able to sense altered levels of CO₂ and/or H⁺. Some such cells are nonexcitable, like renal proximal tubule cells involved in blood acid-base regulation, which respond to HA with increased HCO₃^– reabsorption (412). Others include the glomus cells of the carotid body, which serve as the peripheral chemoreceptors for the control of ventilation (227), nociceptors responsible for pain sensation (388), and taste receptor cells that sense acid as a sour taste (339). These cells are easily identified in that they respond to an increase in CO₂ or H⁺ with an increase in the firing rate of their associated afferent nerves. The identification of central chemosensitive neurons is more problematic, and no neuron has unequivocally been identified as such. To be a central respiratory chemosensor, a neuron would: 1) have to respond to changes of CO₂/H⁺, 2) have axonal projections to a respiratory control center, and 3) ultimately lead to altered ventilation. Despite the lack of certain identification of any central chemosensitive neuron, a number of traits can be used to identify candidate neurons.

The first trait of a central chemosensitive neuron is that the firing rate of the neuron should be altered by changes in CO₂/H⁺ (Fig. 1). Because of concerns that CO₂ can have a generally inhibitory effect on mammalian neurons (51), most of the studies on central chemosensitive neurons have focused on CO₂-excited neurons. Nevertheless, a chemosensitive neuron could respond to increased CO₂/H⁺ level with either increased or decreased firing rate, providing excitatory drive or removing inhibitory drive, respectively, either of which could lead to increased ventilation. In slices from the dorsal medulla, of the 50% of neurons that responded to increased CO₂/H⁺ level, about half were stimulated and about half were inhibited by hypercapnia (159). Similar findings were made in explant cultures from the ventral medulla (389), brain slices from the medullary raphe (295), and raphe neurons in cell culture (383). In contrast, in the locus coeruleus (LC), a CO₂-chemosensitive region of the pons, studies with brain slices revealed that >80% of the neurons responded to elevated CO₂/H⁺ level, and all of these cells were stimulated (116, 262, 276). Thus a variety of brain stem regions contain CO₂/H⁺-sensitive neurons.

View larger version (24K): [in this window] [in a new window]   Fig. 1. Firing rate response to hypercapnia (10% CO2) of a neuron from the retrotrapezoid nucleus, measured in a brain stem slice from a neonatal rat (postnatal day 11; P11) using a whole cell patch pipette at 35°C. The integrated firing rate (Hz, 10-s bins) is plotted at top. For the numbered regions (1, control, 5% CO2, pHo 7.45; 2, hypercapnia, 10% CO2, pHo 7.15; 3, recovery, 5% CO2), a segment of the membrane potential trace is shown at faster speed (bottom), illustrating individual action potentials as vertical upward spikes.

A final feature of a putative central respiratory chemosensitive neuron is that it should reside in a region shown to alter ventilation when locally stimulated by acidification. Experimentally, local stimulation has been produced in at least two different ways: 1) focal injections of solutions containing acetazolamide, which produce a decrementing sphere of acidification extending between 100 and 300–400 µm from the site of injection (67, 68, 245); and 2) focal microdialysis of a solution that is acidified by equilibrating with 25% CO₂ (208, 246, 247). Focal acetazolamide was injected into various brain stem regions in anesthetized animals, and the ventilatory response (as measured by an increase in phrenic nerve activity, which is a measure of the motor output to the diaphragm) was measured (240). Ventilation was increased by focal acetazolamide stimulation of the LC by 30% (compared with the ventilatory increase induced by the whole animal breathing 9% CO₂), of the NTS by 34%, of the VLM, including the retrotrapezoid nucleus (RTN), by 20%, and of the medullary raphe by 32%. Focal acidosis of the midline caudal raphe (23) and the pre-Bötzinger complex (332) also resulted in increased phrenic nerve discharge. These results indicate that no one area predominated, and it appears that several chemosensitive areas would need to be stimulated to get a full ventilatory response. Furthermore, the chemosensitivity of a given area appears to be state dependent. For instance, microdialysis of hypercapnic solution in the medullary raphe stimulated ventilation by 15% compared with the whole animal response, but only in unanesthetized animals and only when the animal was asleep (246). Regardless, there appears to be excellent agreement between regions in the brain stem shown to contribute to ventilatory control and regions that contain CO₂/H⁺-sensitive neurons. We thus refer to putative respiratory chemosensitive neurons in this review as neurons that fulfill all three criteria: 1) respond to alterations in CO₂/H⁺ level with an increase in firing rate, 2) intrinsically chemosensitive, and 3) localized to a region known to contribute to ventilatory control.

It has been suggested that an additional criterion, the degree of chemosensitivity, need be considered (383, 384). In fact, the degree of chemosensitivity of neurons from various chemosensitive areas has not been compared under similar conditions by the same laboratory. Furthermore, there is no standard way in which to express the degree of chemosensitivity of a given neuron. In general, the firing rate response to an acid stimulus is indexed to the change of extracellular pH (pH_o), and not CO₂ level, to be able to compare the responses to different acid stimuli. In Table 1, we show a compilation of the degree of chemosensitivity for neurons from five different chemosensitive regions as estimated from appropriate references. We include three different estimates of chemosensitivity. If a neuron has a firing rate in control solution (FR_C) that increases in acidic solution (FR_A), then we calculate the chemosensitivity of this neuron as the percent increase in firing rate for a 0.1 pH unit acidification {using the equation [(FR_A – FR_C)/FR_C]/(pH/0.1) x 100}. With this expression, a neuron that is not chemosensitive would have a 0% increase in firing rate. Alternatively, the absolute increase in firing rate per unit pH change could be calculated [using the equation (FR_A – FR_C)/pH]. Finally, the chemosensitive index (CI) as described by Wang et al. (383) was calculated. The CI is based on the percentage of control firing rate to which firing rate increases in response to a 0.2 pH unit acid challenge. With this expression, a neuron that is not chemosensitive would have a value of 100% for CI. In general, the two methods based on percent increase in firing rate per pH change appear to be the most robust, being similar over a wide range of initial firing rates (Table 1). In contrast, the chemosensitivity determined from the slope of absolute firing rate vs. pH are highly variable and very sensitive to differences in the initial firing rate (Table 1). Thus we propose that measures of the relative change in firing rate, rather than the absolute change, are better indicators of the relative chemosensitivity of neurons from different brain stem regions.

These variations might give clues to the role played by neurons in these various regions. For instance, the LC is a region associated with pain sensation, attention, learning, and anxiety (9, 11, 154). The primary function of CO₂/H⁺-sensitive neurons within the LC may be to produce an aversive or anxiety response to elevated CO₂ levels (179, 326). In this regard, LC neuron response would be most significant at high levels of CO₂, which is consistent with the relatively low level of chemosensitivity of neurons in this region (Table 1).

Finally, the chemosensitive response of the neurons from no one of these areas appears to be sufficient to explain the large ventilatory response to a 0.1 pH unit decrease, which is sometimes reported to be as high as an increase of 400–500% (114) (see HISTORY OF STUDIES OF CENTRAL CHEMORECEPTION). It is interesting that these high values for the ventilatory response to HA have been obtained for the most part in large animals, such as cats, dogs, goats, and humans (48, 114, 224, 253, 290, 322, 359, 373), whereas the cellular studies of chemosensitivity have been performed almost exclusively in rodents. A comparison of the ventilatory response of different animals to increased CO₂ levels reveals that the most commonly used animal model for the study of the cellular basis of chemosensitivity, the rat, appears to have a response that is two to three times lower than that of larger animals (192, 381) (Table 2). This means that most of the cellular studies performed to date have employed the least sensitive animal for studying chemosensitivity and suggests that studies of the cellular basis of chemosensitivity using neurons from more sensitive animals might be quite interesting. Nevertheless, these cellular findings are in general agreement with the focal stimulation studies discussed above and suggest that, assuming a linear relationship between chemosensitive neuron firing rate and increased ventilation, multiple central chemosensitive sites need to be stimulated to account for the full ventilatory response in the intact animal.

	HISTORY OF STUDIES OF CENTRAL CHEMORECEPTION

TOP ABSTRACT CO2/H+-SENSITIVE CELLS HISTORY OF STUDIES OF... REDUCED PREPARATIONS SIGNALS IN CENTRAL... OTHER FACTORS IN CENTRAL... POSSIBLE TARGETS IN CENTRAL... A REVISED MODEL FOR... OTHER ACID-SENSITIVE CELLS SUMMARY AND CONCLUSION GRANTS REFERENCES

Winterstein (400) proposed that a change of pH is the adequate stimulus for the chemical control of breathing. His initial theory, called the reaction theory (at that time pH was referred to as reaction), was an attempt to unify the findings that both elevated CO₂ and decreased O₂ could stimulate ventilation. Winterstein, in the first version of his reaction theory, argued that it was decreased blood pH, due to either elevated CO₂, producing H⁺ through its hydration and dissociation reactions, or the accumulation of metabolic acids during hypoxia, that resulted in activation of the chemosensory mechanisms that controlled ventilation. Thus a single common signal was proposed to explain the ventilatory response to both stimuli. Interestingly, this model was proposed more than 15 years before the identification of the peripheral chemoreceptors and their importance in the ventilatory response to O₂ (153).

The reaction theory came under question over the years on the basis of a number of findings. When HCO₃^– is added to the blood, pH alkalinizes yet ventilation is increased (69). In contrast, addition of NH₄Cl causes blood acidification but reduction in ventilation, in seeming contrast to the reaction theory (188, 399). Winterstein was able to account for these findings based on the studies of Jacobs (166), who showed that at constant pH_o, starfish eggs acidified when exposed to HCO₃^– and alkalinized when exposed to NH₄Cl because of the membrane permeability of the uncharged partners CO₂ and NH₃, respectively. Thus, in the second version of the reaction theory, Winterstein differentiated the effects of changes in blood pH ("hematogenic" effects) from "... a change of the center itself" ("centrogenic" effects) (399). It appears that by the "center itself," Winterstein may have been thinking of the intracellular pH (pH_i) of chemosensitive neurons themselves (188), thus representing the first suggestion of pH_i as the adequate stimulus, although he may have been referring to the pH_o in the vicinity of the central chemosensitive neurons.

An alternative to Winterstein’s reaction theory was the multiple factors theory of Gray (133, 134). This theory focused on the factors that contribute to the eventual increase in ventilation; O₂ and especially CO₂ (both peripherally and centrally) are proposed to contribute to respiratory drive along with pH, i.e., multiple factors contribute to increased ventilation, not a single unifying adequate stimulus. On the basis of a mathematical model, Gray (133) described the partial effects of changes in O₂, CO₂, H⁺, and muscle reflexes on the overall ventilatory increase in response to stimuli like HA, metabolic acidosis, exercise, and high-altitude anoxia. Because we now realize that control of ventilation does indeed involve input from central chemoreceptors, peripheral chemoreceptors, and reflex pathways, Gray’s conception of a multiple factors theory can hardly be challenged, and as such, the theory does not directly address the site of chemosensitivity or the pathways of activation of these sites. However, Gray (133) implied multiplicity of chemosensitive stimuli when he stated that "... the present theory ... resolves the most persistent and controversial question in the field of respiration: Should H-ion or CO₂ be considered the true respiratory stimulus? From the standpoint of the multiple factors theory, this question should be framed as follows: To what extent does each of the two agents influence ventilation?" In addition to Gray’s theory, two other lines of evidence supported the idea that CO₂ itself is a chemosensitive signal, independent of its effect on pH. The first was the finding that the effects of elevated CO₂ (HA) on ventilation often exceeded the effects of metabolic acidosis (31, 32). The other line of evidence was that elevated CO₂ increased ventilation by increasing both tidal volume and respiratory frequency, whereas metabolic acidosis resulted in increased ventilation by increasing tidal volume only (see DISCUSSION in Ref. 188).

It is clear that these early theories suffered from a lack of awareness of the complex relationships between pH in various compartments, including blood, cerebral spinal fluid (CSF), extracellular fluid pH in the brain, and intracellular fluid. For example, Gesell and Hertzman (128, 129) showed that injection of HCO₃^– into the blood system resulted in alkalinization of the blood but acidification of the CSF. Injection of NH₄Cl into the blood resulted in acidification of the blood but alkalinization of the CSF (399). It was proposed that a barrier appears to exist between the CSF and blood that allows passage of uncharged species predominantly and acts in an analogous fashion to the cell membrane, as described by Jacobs (166). Thus blood pH could be a poor reflection of CSF pH. Similarly, it was suggested that blood pH could also be a poor reflection of the pH_o in the brain (106). Furthermore, in many early studies (see, for example, Ref. 114), extracellular fluid pH (referred to in this review as pH_o but also referred to as pH_ecf or pH_e) was estimated from the pH of the CSF during ventriculocisternal perfusion. However, this method of estimation was called into question because of the unknown effects of the volume and composition of endogenously produced CSF (28). Therefore, the relationship between ventilation and pH as determined in a number of early studies must be questioned because of uncertainty regarding the appropriateness of the estimated values of pH.

These concerns were eliminated by the introduction of reliable techniques to measure pH_o with the use of either pH-sensitive microelectrodes within the medulla oblongata (75) or macro pH-sensitive electrodes on the ventral surface of the medulla (5, 106, 322, 354). With the introduction of techniques to measure pH_o directly, studies were done to determine whether ventilation has a unique relationship to pH_o. In a review of central chemosensitivity and the reaction theory, Loeschcke (211) concluded that "... extracellular pH in the brain is the main chemical signal determining ventilation." However, to reach this conclusion, he had to postulate that the acid-sensitive mechanism is accessible to CO₂ but has limited accessibility to H⁺ from metabolic acids. More recently, numerous studies demonstrated a different ventilation-pH_o relationship for HA compared with metabolic acidosis (190, 322, 353). For instance, Eldridge et al. (106) showed that the frequency of phrenic nerve firing was two to four times higher at the same pH_o with HA than with metabolic acidosis. In other words, pH_o could not be uniquely associated with a given level of ventilation. Eldridge et al. (106) offered three possible alternative explanations for this disagreement with the reaction theory. The first is the limited access theory proposed by Loeschcke (211). The second is that CO₂ has an effect separate from the effect of H⁺ or that there are separate sites for sensing H⁺ and CO₂. This proposal is consistent with the multiple factors theory. The third proposal is that pH_i, and not pH_o, is the adequate stimulus for chemoreception, a suggestion also made by Shams (322). A variation of this latter concept was also proposed by Kiwull-Schöne and Kiwull (190) and Xu et al. (403), who suggested that it is the transmembrane pH gradient (the difference between pH_i and pH_o) that serves as the chemosensitive signal. Exposure of chemosensitive neurons to CO₂ or to metabolic acids was proposed to result in a larger fall of pH_o than of pH_i due to internal H⁺ buffering by proteins. This would result in a reduction of the transmembrane pH gradient and a stimulation of chemosensitive cells. This idea is supported by the findings of Nattie (242), who showed that exposure of the ventral medullary surface to diethyl pyrocarbonate, an agent that can reduce intracellular H⁺ buffering by binding to imidazole groups on histidines, eliminates the ventilatory response to elevated CO₂, although it is possible that diethyl pyrocarbonate is directly inhibiting a pH sensor that involves a histidine residue. The importance of changes in the pH gradient across the membrane is also proposed to explain the lack of a ventilatory stimulation during hypoxia-induced acidification (the hypoxia paradox, see Ref. 239), where the internally generated lactic acid is presumed to result in a greater fall of pH_i than of pH_o (403). Evaluation of the possibility that it is pH_i or the transmembrane pH gradient that is the adequate chemosensitive signal had to await reliable measurements of pH_i in chemosensitive neurons.

Many of these earlier studies suffer from various methodological concerns. As mentioned above, most of the earlier studies are hard to interpret because of doubts about the appropriate compartment in which to determine pH. Another issue is that many of these studies were done without accounting for contributions from peripheral chemoreceptors or other feedback mechanisms (106), which complicates the interpretation of the data regarding central chemoreception. Furthermore, under some of the experimental conditions used, it is unclear whether true "isocapnic" conditions were maintained (189). Finally, the vast majority of the earlier studies aimed at determining the adequate stimulus focused on pH_o measurements at the surface of the ventral medulla as the sole site of chemoreception, whereas the experimental treatments may well have affected other chemosensitive sites (such as increasing inspired CO₂ or intravenous infusion of acid as in Ref. 106). Thus the likely presence of multiple sites for chemosensitivity throughout the brain stem makes studies of the adequate stimulus far more difficult.

Despite these limitations, certain findings are so common that they are undoubtedly true and, as such, must be explained by any model of central chemosensitivity. One such observation is that acidosis induced by increased CO₂ levels appears to be a stronger stimulus to ventilation than is metabolic acidosis. In other words, for a similar degree of extracellular acidification, it appears that the chemosensitive system is more sensitive to HA than to metabolic acidosis (see, for example, Ref. 106). Another common finding is that there is a very high degree of sensitivity of ventilation to changes of pH, at least in some animals (Table 2). The work of Fencl et al. (114) on goats is often cited as an example of the high degree of sensitivity of ventilation to changes of pH. In fact, these authors reported a five- to sixfold increase in ventilation for a 0.1 pH unit change of CSF pH. As mentioned above, the method used to calculate pH in their study has been questioned (28). Nevertheless, more recent work with cats, using reliable measures of pH_o, has still reported a considerable sensitivity with ventilation changing from twofold (5) to fourfold (322) for a 0.1 pH unit change of extracellular fluid pH. Finally, despite years of research, conflicting data seem to indicate that there may be no unifying theory for central chemosensitivity and that there may not be a single adequate stimulus.

	REDUCED PREPARATIONS

TOP ABSTRACT CO2/H+-SENSITIVE CELLS HISTORY OF STUDIES OF... REDUCED PREPARATIONS SIGNALS IN CENTRAL... OTHER FACTORS IN CENTRAL... POSSIBLE TARGETS IN CENTRAL... A REVISED MODEL FOR... OTHER ACID-SENSITIVE CELLS SUMMARY AND CONCLUSION GRANTS REFERENCES

 
Central chemosensitivity is clearly a complex process, involving numerous inputs and multiple possible signals. Thus, to study and define chemosensitive signaling at the cellular level, the use of reduced preparations is required. Several reduced preparations for the study of central chemosensitivity have been developed, including the isolated brain stem-spinal cord preparation, brain stem slices, and tissue culture.

Brain stem-spinal cord preparations. The entire brain stem, either with or without part of the spinal cord attached, can be removed and studied in vitro. Changes in "ventilatory output" in these preparations can be monitored by placing suction electrodes on the phrenic nerve root (with spinal cord attached) or on the hypoglossal rootlet (without spinal cord attached) (144), whereas chemosensitive neuron output can be measured with either extracellular or intracellular electrodes (92, 262). These preparations were originally taken from neonatal rats to minimize their size and therefore minimize problems with maintaining their oxygenation. They have been used to study a variety of questions related to central ventilatory control, including respiratory rhythm generation and control (347, 348), central chemoreception (144, 180, 262, 376), and the effects of anoxia on ventral medullary neurons (15, 386). An advantage of the brain stem-spinal cord preparation for studying central chemosensitivity is that it is possible to monitor a system output that is reflective of all the various central CO₂ chemosensitive areas without any peripheral inputs. The disadvantages of this preparation are that individual neurons, other than those near the surface, are relatively inaccessible to microelectrodes and that maintaining adequate oxygenation can be difficult, especially in the core of the preparation (for a discussion of this issue, see Ref. 87).

Two variations on the brain stem-spinal cord preparation have been described. In the first, the brain stem only was removed from an adult guinea pig (92, 234). In this preparation, ventilatory output was measured with suction electrodes on the hypoglossal roots (also reflective of respiration). Because this was a larger preparation, adequate oxygenation required both superfusion with artificial cerebral spinal fluid (aCSF) and perfusion by cannulation of the basilar artery. When this preparation was superfused alone (with aCSF equilibrated with 95% O₂), tissue PO₂ dropped from 500 Torr at the surface to essentially 0 Torr at a depth of 400 µm (these brain stems were over 6 mm thick). However, when this preparation was perfused as well as superfused, surface PO₂ was 500 Torr but fell to a stable level of 200 Torr at a depth of 1 mm into the brain stem (92). This level of 200 Torr PO₂ was maintained throughout the remainder of the thickness of the brain stem. Although this finding indicates that in this preparation core hypoxia is not an issue, Dean et al. (87) recently pointed out that brain PO₂ in vivo is <40 Torr, and so the level of oxygenation in these brain stem preparations is considerably hyperoxic (at least at the surface in the superfused brain stem and throughout the entire perfused brain stem). The effect of this hyperoxia on the neuronal properties has yet to be determined.

The other preparation is the working heart-brain stem preparation (267). In this preparation, the animal is bisected below the diaphragm, decerebrated, and the dorsal surface of the medulla exposed and superfused. The descending aorta is cannulated, and the preparation is perfused. The heart continues to beat, receives venous return from the superior vena cava, and maintains a normal beat frequency. Respiratory activity is recorded by suction electrode from the phrenic nerve. This preparation gives a normal respiratory pattern, as evidenced by the cyclic bursting pattern in the phrenic nerve. In addition, intracellular recordings can be made from dorsal medullary respiratory neurons, and peripheral reflexes appear to be intact, because activation of peripheral chemoreceptors increases phrenic activity. This preparation has been used with both mice (267) and neonatal rats (101). Although use of this preparation requires considerable surgical skill and may be limited in studying the properties of individual neurons, it benefits by maintaining a nearly fully intact regulatory pathway (including central and peripheral inputs), allowing studies of respiratory control in the absence of anesthetics, and maintaining the preparation with adequate levels of oxygen to yield normal respiratory patterns and responses. This technique was exploited further recently by the ability to study ventral respiratory neurons with whole cell patch clamp (100) and the use of fluorescence recordings to study calcium transients in medullary respiratory neurons (38). This preparation appears to be a good preparation with which to study the mechanisms of central chemoreception (268).

The results of studies on central chemoreception using the brain stem-spinal cord preparation agree in general with results of earlier studies. For instance, Denavit-Saubié et al. (92) found different responses of the adult brain stem to hypercapnia in the superfusate vs. hypercapnia in the perfusate. In the former case, when only chemosensitive neurons in the surface layers were activated, there was an increase in the burst amplitude but a decrease in burst frequency in the output of the hypoglossal root. These findings agree with previous findings in intact animals exposed to hypercapnia on the medullary surface, showing superficial chemosensitive neurons in the medulla (211). When the superfusate was made acid by decreasing solution HCO₃^– content at constant PCO₂ (metabolic acidosis), burst frequency increased, indicating that these superficial chemosensitive neurons respond differently to acidosis induced by increased CO₂ level compared with metabolic acidosis. Denavit-Saubié et al. (92) and Morin-Surun et al. (234) further showed that when the perfusate was made hypercapnic, thus activating deeper chemosensitive neurons, there was an increase in burst frequency but a decrease in amplitude. The finding that deeper chemosensitive neurons give a different response to hypercapnia than superficial neurons suggests that there are differences in the response of chemosensitive neurons from different regions to the same stimulus.

Harada et al. (144) used a brain stem-spinal cord preparation from neonatal rats to study the effects of various acid challenges on overall ventilatory output (measured as integrated phrenic nerve activity). By comparing phrenic nerve output in brain stems exposed to solutions of varying pH at constant CO₂ (made by altering HCO₃^–) with solutions of varying CO₂ at constant pH, these authors were able to differentiate the ventilatory effects of CO₂ from those of pH. They showed that pH and CO₂ had separate and additive effects on ventilatory output, suggesting that there are separate chemoreceptors for pH and CO₂, a conclusion that agrees with Eldridge et al. (106). This conclusion is further strengthened by the finding that decreased pH altered both phrenic burst magnitude and frequency, whereas increased CO₂ only increased phrenic burst magnitude. Also, the effects of altered CO₂ levels were transient, whereas those of altered pH were sustained. On the basis of these findings, Harada et al. (144) proposed that there are different sites of chemoreception for CO₂ and pH and that the response to altered CO₂ levels may be mediated by changes of pH_i as well as pH_o. Finally, their findings also suggested a small and transient effect due to altered HCO₃^–.

Brain stem slices. Slice preparations have been used extensively to study cellular properties within the central nervous system (CNS). The brain stem can be sliced either transversely (82, 86, 295, 306) or horizontally (126, 168). Slices are often made as thin as possible (100–400 µm thick) to maintain adequate oxygenation and to easily visualize individual neurons (237, 306). A thicker brain stem slice preparation (600–700 µm thick) was used to maintain a spontaneous respiratory rhythm (328) that is synchronized between the pre-Bötzinger complex and the hypoglossal roots (286, 287). The advantage of such preparations is that individual neurons can often be easily seen for patching, there is reasonably good control of the microenvironment around the neurons, and neuronal association with glia are intact, as are most of the dendritic processes and synaptic connections.

Several disadvantages are also involved with the use of brain stem slices. During slicing there can be damage to neurons and glia, especially in the surface layers. This damage can arise directly from trauma induced by slicing, from ischemia induced during brain stem dissection and preparation for slicing (297), from the loss of endogenous antioxidants and subsequent oxidative damage (292, 293), and from acidification due to CO₂ accumulation (221). All of these insults appear to lead to a cytotoxic edema that often results in neuronal death within a few hours (221, 292, 297). A number of techniques have been developed in an attempt to minimize this damage. Richerson and Messer (297) prepared brain stem slices (100 µm) in a hyperosmotic solution to reduce slice swelling, with low Ca²⁺ plus kynurenic acid to prevent neuronal uptake of Ca²⁺ and Na⁺, which have been associated with neuronal injury. This technique resulted in reduced slice swelling and increased neuronal survival. Rice and colleagues (292, 293) showed that slices lose most of their endogenous ascorbate and glutathione during incubation after slicing and that this loss is associated with slice edema. Slice edema and neuronal histology were markedly improved when the slices were made and incubated in aCSF to which 400 µM ascorbate had been added. Finally, MacGregor et al. (221) showed that the addition of the buffer N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid (HEPES) reduced slice (400 µm) edema, presumably due to reduced slice acidification. The use of a combination of these techniques will likely reduce slice damage during preparation.

The slicing procedure also undoubtedly results in some degree of dendritic pruning and the loss of synaptic input from neurons from more distant brain regions, eliminating either excitatory or inhibitory inputs. Recently, an additional concern was raised, i.e., the proper level of oxygenation to be used with slices (87). Originally, slices were bathed in solutions equilibrated with 95% O₂ to maintain oxygenation in the center of the slice. Because slicing procedures have improved and thinner slices are obtained, this oxygenation protocol has not been altered. Recent measurements of O₂ profiles within 300-µm brain stem slices superfused with aCSF equilibrated with 95% O₂ showed that the PO₂ at the surface of the slice was >400 Torr and at the center was nearly 300 Torr (237). These values are nearly an order of magnitude greater than PO₂ values seen in neural tissue in vivo (87). Given that oxidative stress has been shown to damage slices (292), an investigation of the appropriate level of O₂ with which to equilibrate the aCSF bathing slices is warranted (87).

Despite these disadvantages, brain stem slices have been used for nearly 30 years to study the cellular basis of central chemosensitivity (70, 82, 84, 116, 117, 121, 122, 125, 159, 168, 295, 304–306, 385), and many of the findings made using the brain slice preparation are reviewed below.

Tissue and cell culture. Culture techniques have been employed in studies of central chemosensitivity as well. The use of cultured chemosensitive neurons offers some unique advantages, especially in the search for the adequate stimulus of central chemosensitivity. Among these advantages are the ease of performing electrophysiological measurements on these cells, the ready ability to control the external environment of these neurons, the elimination of excitatory and inhibitory synaptic input that can often complicate the interpretation of responses in more intact systems, and the ability of neurons to recover from slice damage (383).

Several laboratories have employed organotypic culturing of brain slices. This approach starts with a brain stem slice (225–500 µm). The slice can be minced into smaller pieces (250) or placed directly into culture in a chemically defined medium on permeable plastic or nylon grids (10, 27, 311, 393, 394) or on glass slides (389, 390). Slices are obtained from either neonatal rats [postnatal days 0–6 (P0–P6)] (10, 27, 250, 393, 394) or rat fetuses (embryonic day 16) (389, 390). These various explant cultures are used in some cases after 8 days in culture (250) but mostly after 2–4 wk (10, 394). The explant slices usually flatten out, reducing their thickness to 50–200 µm (10), and have healthy-looking neurons that make synaptic contacts (27). When plated on glass slides, the slices exhibit outgrowths that are originally largely of glial origin but develop with time into a neural network (389, 390).

Primary cell culture has also been used to study central chemosensitive neurons. Cell cultures of CO₂-sensitive neurons have been derived from either fetal (119, 302, 303) or neonatal (343, 383) medullary brain stems. The technique involves microdissection of the brain stem to include the areas of interest, followed by protease digestion and then plating of the resulting cells. These cells can be plated either on a monolayer of nonneuronal cells (119, 302, 303) or on laminin-coated coverslips in glia-conditioned medium (383). Neurons grow well under culture conditions and form either complex networks (119, 303) or individual cells (383). These cultured neurons are usually studied 2–5 wk after initial culturing (anywhere from 2 to 94 days), using standard electrophysiological and histological techniques. Rigatto et al. (302) observed that 10–20% of the neurons cultured from either the nucleus ambiguus or the NTS exhibited regular spontaneous firing and responded to HA with an increase in firing rate (measured with whole cell patch techniques). The remaining neurons fired irregularly or were silent, and far fewer were stimulated by hypercapnic acidosis. Wang et al. (383) also found a proportion of their cultured medullary raphe neurons that exhibited regular spontaneous firing. Nearly 25% of the neurons studied were stimulated by HA (many of these exhibited regular firing patterns), 25% were inhibited by HA, and the remaining 50% were unaffected. Wang et al. (383) further showed that the morphologies of HA-excited and HA-inhibited neurons were distinct. Thus it is clear that the use of cell culture yields neurons that exhibit a response to chemosensitive stimuli.

Cultured cell preparations can be problematic in terms of their representation of the in vivo preparation. For any given preparation, it must be shown that the prolonged period in culture during which neuronal growth occurs does not result in neurons with altered properties, such as the expression of ion channels or transporters that are not normally expressed in vivo. For example, the pH_i of hippocampal neurons cultured for 10–12 days was significantly lower than the pH_i of freshly dissociated neurons (285). If this is true in cultured chemosensitive neurons, it could be highly problematic, because these cells are likely sensitive to changes of pH_i (116). In addition, the lack of synaptic input itself suggests that cultured neurons may well respond differently to stimuli than they would in the context of the network in vivo. These cultured cells also have nearly unrestricted access to the bathing medium, a major advantage as stated above. However, in some cases, these neurons have been studied in solutions equilibrated with 95% O₂ (383), which should expose them to nearly 700 Torr of O₂. Thus, to the extent that oxidative environments affect central chemosensitivity (238), such preparations should be highly affected. Interestingly, Rigatto et al. (303) studied their cultured neurons in growth medium that had a more normal PO₂ of 60 Torr. In fact, their cultured neurons hyperpolarized and lost their spontaneous firing rate when exposed to medium equilibrated with 100% O₂ (640 Torr) (302, 303). The question of the appropriate level of O₂ to use with cultured neurons has been addressed, and it has been shown that cortical neurons grow better in culture medium equilibrated with 9% O₂ as opposed to 20–21% O₂ (40, 176). Therefore, the effect of the use of medium equilibrated with 95% O₂ to study cultured chemosensitive neurons needs to be examined. Finally, recent findings (see Glial cells in OTHER FACTORS IN CENTRAL CHEMOSENSITIVE SIGNALING) suggest that glial cells may well play a role in central chemosensitivity, and the removal of the association between glial cells and neurons in cell culture will undoubtedly alter the response of cultured neurons compared with neurons in vivo.

Despite these concerns, cultured neurons often display many of the properties that they display in the brain slice (295, 383) and have been used to make important findings regarding the mechanisms of central chemosensitivity (298, 382).

	SIGNALS IN CENTRAL CHEMOSENSITIVE NEURONS

TOP ABSTRACT CO2/H+-SENSITIVE CELLS HISTORY OF STUDIES OF... REDUCED PREPARATIONS SIGNALS IN CENTRAL... OTHER FACTORS IN CENTRAL... POSSIBLE TARGETS IN CENTRAL... A REVISED MODEL FOR... OTHER ACID-SENSITIVE CELLS SUMMARY AND CONCLUSION GRANTS REFERENCES

 
There have been numerous proposed signals for central chemosensitivity, but the three most likely signals are changes in CO₂, pH, and, in some cases, HCO₃^– (see HISTORY OF STUDIES OF CENTRAL CHEMORECEPTION). Many other agents have been suggested to play a role as well. We review the major findings for each of these possibilities.

Carbon dioxide. The hyperventilatory response to elevated CO₂, especially when accompanied by extracellular acidosis (HA) can be substantial (see HISTORY OF STUDIES OF CENTRAL CHEMORECEPTION). In fact, the main argument for a role for molecular CO₂ as a signal in central chemosensitivity is the greater ventilatory response to HA compared with metabolic acidosis that is often observed (106, 144, 189, 322). In many of these studies, ventilation could not be plotted as a unique function of pH_o, and it was thus assumed that another factor, i.e., CO₂, must be serving as an independent signal. The strong response to elevated CO₂ is also seen in reduced preparations (e.g., slice or cell culture), where putative central chemosensitive neurons respond to HA [e.g., slice (82, 116, 117); cell culture (250, 295, 303, 382, 383)]. In fact, a response to increased CO₂ is often taken as an indication that a neuron is chemosensitive, although this definition alone is undoubtedly too inclusive (for discussion, see CO₂/H⁺-SENSITIVE CELLS above and Refs. 294 and 384). There are two main difficulties in proposing that CO₂ per se is part of the cellular signaling pathway for central chemosensitivity. The first is that all of the earlier theories that proposed CO₂ to be a unique and independent signaling factor considered only the inadequacy of pH_o to explain the respiratory responses to various chemosensitive stimuli. However, no measurements of pH_i in central chemosensitive neurons existed even though cellular pH is markedly affected by CO₂ (279, 281). Furthermore, the effects of the various chemosensitive stimuli on pH_i, and thus the possibility that changes of pH_i serve as the adequate stimulus, were only occasionally considered (e.g., Ref. 106). The second major difficulty with CO₂ being a chemosensitive signal per se is that no appropriate cellular model has ever been proposed by which CO₂ could be a chemosensitive signal separate from its effects on pH_o and pH_i. However, we cannot rule out a CO₂ receptor, because one has been proposed for the response of renal proximal tubule cells to hypercapnia (412) and because receptors for other gases, such as nitric oxide (249), carbon monoxide (410), and oxygen (374), have been reported. Recently, increased molecular CO₂, independent of its effects on pH, was proposed to activate L-type Ca²⁺ channels and raise intracellular Ca²⁺ in glomus cells (344–346) (see Carotid body cells in OTHER ACID-SENSITIVE CELLS). This effect involves elevated cAMP, and it has been suggested that molecular CO₂ could work through elevating intracellular HCO₃^– and activating soluble adenylate cyclase, which has been shown to be HCO₃^– dependent (414) (Fig. 2). It is not known whether such a mechanism is at play in central chemosensitive neurons; however, in support of this possibility, it was recently shown in LC neurons that L-type Ca²⁺ channels are only activated when CO₂ is elevated, and not by decreased pH_i alone (117). Therefore, it will be of interest to determine whether molecular CO₂ can activate central chemosensitive neurons in a fashion that is independent of its effects on pH.

View larger version (20K): [in this window] [in a new window]   Fig. 2. Model of CO2 activation of L-type Ca2+ channels in glomus cells from the carotid body. Hypercapnia elevates intracellular HCO3– level, which in turn activates soluble adenylate cyclase (AC). This in turn opens L-type Ca2+ channels by elevating cAMP and activating protein kinase A (PKA). This model is based on the findings of Summers et al. (346).

Theoretically, there are several ways by which HCO₃^– could affect neuronal activity. The exchange of HCO₃^– and Cl^– across the blood-brain barrier is involved in the response of the brain extracellular space to respiratory acidosis in the intact organism (3, 4). At the cellular level, HCO₃^–-dependent transporters, including Cl^–/HCO₃^– exchange, Na⁺-driven Cl^–/HCO₃^– exchange, and Na⁺-HCO₃^– cotransport are involved in the regulation of pH_i in neurons and glial cells (90, 279, 304). It has been shown that HCO₃^– is needed for the recovery of synaptic transmission from anoxia and that this effect is independent of the effects of HCO₃^– on pH_i or pH_o (309). It was speculated that this effect could be mediated by the HCO₃^– dependence of the glutamate uptake transporter in astrocytes (35, 171), although it now appears that glutamate uptake is accompanied by movement of H⁺, not HCO₃^– (175, 411), or by the ability of HCO₃^– to reduce free radical production by promoting the binding of iron to transferrin (177, 178, 288). This latter possibility is especially interesting, because increased production of free radicals was recently shown to selectively activate chemosensitive neurons within the NTS (238). As stated above, increased HCO₃^– can also activate soluble adenylate cyclase (414) and thereby increase cAMP and, ultimately, intracellular Ca²⁺. Finally, the GABA_A channel has been shown to be permeable to HCO₃^– (172) with a selectivity for HCO₃^–:Cl^– of 1:5 (173). Activation of the GABA_A channel can result in an efflux of HCO₃^– that leads to an intracellular acidification and an extracellular alkalinization (58, 172, 173). Thus changes in HCO₃^– in the extracellular solution may impact chemosensitive signaling by affecting neuronal pH_i, causing changes of pH_o, altering free radical production, changing cellular cAMP levels, and/or modifying the uptake of glutamate by astrocytes. A detailed examination of the role of changes of HCO₃^– in chemosensitive signaling would seem to be warranted.

Extracellular and intracellular pH. For nearly a century, H ions have been proposed to be the signal for chemosensitive cells in the control of ventilation (400). However, it has proved difficult to separate the effects of pH_o from pH_i. Much of the early work to differentiate these two possible signals employed comparisons of the effects of respiratory (hypercapnia and hypocapnia) vs. metabolic (weak acids and bases) acid-base changes on ventilation (e.g., Refs. 106, 190, 322, 353). Although these techniques often showed a difference between the ventilatory response to the stimuli, no information was available on the change of pH_i induced in putative chemosensitive neurons by respiratory vs. metabolic changes.

The study of signaling in chemosensitive cells has benefited greatly from the development of techniques to reliably study pH_i. Early techniques to measure pH_i in brain cells used the distribution of the weak acid dimethyl-2,4-oxazolidinedione (312, 313) or nuclear magnetic resonance (26, 170, 378). Both of these techniques involved more global measurements of pH_i, and thus their relevance to changes of pH_i in chemosensitive neurons is problematic. However, on the basis of these findings, Lassen (199) made the first specific proposal that it is the change of pH_i, rather than the change of pH_o, that is the adequate stimulus for central chemosensitivity. Lassen’s argument was based on the relative changes of pH_o, pH_i, and ventilation in response to hypercapnia (7% CO₂) and acetazolamide (an inhibitor of CA). It had been shown that both hypercapnia and acetazolamide result in a fall of brain pH_o of 0.12 pH unit (26, 199). However, whereas hypercapnia (7% CO₂) resulted in a significant drop in brain pH_i of 0.06 pH unit (170), acetazolamide did not significantly affect brain pH_i when brain surface CO₂ was maintained constant (26, 378). In parallel with the changes of pH_i, but not the changes of pH_o, ventilation was markedly enhanced by hypercapnia but not by acetazolamide. These data strongly implicated changes of pH_i as the adequate stimulus for ventilation, but there is a concern that these measurements of brain pH_o and pH_i may not reflect the changes of these variables within chemosensitive neurons themselves. However, over the last decade, the emphasis of the studies of signaling pathways in central chemosensitivity has been on changes of pH_i.

The earliest measurements of pH_i in single respiratory-related neurons are those of Ballanyi et al. (14) and Cowan and Martin (73). Ballanyi et al. (14) used double-barreled pH-sensitive microelectrodes (using a pH-sensitive resin for the pH electrode) to measure pH_i and membrane potential (V_m) in expiratory neurons of the ventral respiratory group. They found that when activity in the expiratory neurons was inhibited (presumably by input from the inspiratory neurons), the neurons rapidly (within 1–2 s) acidified by 0.2 pH unit from resting pH_i levels of 7.15. This acidification was attributed to HCO₃^– efflux through GABA_A channels, which are believed to be activated during inhibitory input. Interestingly, similar measurements in the axons of these neurons did not show alterations of pH_i with inhibition of neuronal output. These results suggest that pH_i changes in the axons (and presumably the dendrites) might well differ from those of the soma, and given the speed of the somatic acidification, GABA_A channels must reside close to, if not on, the soma in these expiratory neurons. Furthermore, a few measurements from associated glial cells showed that these cells alkalinize upon depolarization, and this alkalinization was attributed to the voltage-sensitive Na⁺-HCO₃^– cotransporter mediating HCO₃^– influx during glial depolarization (at resting V_m, this transporter is often outwardly directed, which would result in glial acidification). Cowan and Martin (73) also used double-barreled pH-sensitive microelectrodes (with a pH-sensitive resin) and measured V_m and pH_i simultaneously in individual neurons in slices of the dorsal respiratory group. Studying the effects of hypoxia on pH_i and V_m, these authors found that acidification of dorsal respiratory group neurons resulted in membrane depolarization, and they attributed this to effects of intracellular acidification on ion channels.

Trapp et al. (366) described a different method to simultaneously measure pH_i and V_m in dorsal vagal neurons. These neurons are from a region believed to contain chemosensitive neurons (159). To obtain simultaneous recordings of V_m and pH_i, the pH-sensitive fluorescent dye 2',7'-bis(2-carboxyethyl)-5(6)-carboxyfluorescein (BCECF) was loaded into the neuronal cell body from a whole cell patch pipette. This study (366) showed that dye-loaded and patched neurons gave normal pH_i responses to exposure to NH₄Cl (33) or the base trimethylamine (TMA), indicating that neuronal pH_i was not "clamped" by buffer diffusing from the patch pipette. This technique was used to show that these neurons acidify in response to anoxia, alkalinize in response to external buffer change from CO₂/HCO₃^– to HEPES, acidify in response to increased neuronal activity (366), and acidify in response to GABA and glycine, as a result of HCO₃^– efflux from a receptor-coupled channel (215). The response of these neurons to chemosensitive signals (e.g., HA) was not studied, but no marked change in firing rate was observed in response to anoxia-induced acidification or to alkalinization induced by the buffer change (366). This could indicate that these cells were not chemosensitive or that the electrical response was washed out (88, 117, 295, 308) by the use of whole cell recordings.

Ritucci et al. (306) introduced a technique to measure pH_i in individual neurons in brain stem slices from putative chemosensitive regions using the pH-sensitive fluorescent dye BCECF. In this technique, cells were loaded with BCECF by incubating the slice in the membrane-permeable acetoxymethyl ester (AM) form of BCECF, BCECF-AM. This form of the dye readily enters the cell, where the AM groups are cleaved by internal esterases. The resulting BCECF is charged, and thus relatively impermeable to the cell membrane, and its fluorescence is sensitive to pH_i. Slices loaded in such a fashion showed several large spheres of intense fluorescence and a diffuse background fluorescence, with the latter presumably due to dye loaded into neuronal processes, glial cells, and out-of-focus cells. The large spheres were shown to be dye-loaded neuronal cell bodies by immunocytochemical stains for neurons, by retrograde labeling of another fluorescent dye, and by direct electrical recordings. These cells were visualized and studied using fluorescence imaging microscopy.

By using this technique with brain stem slices from neonatal (P0–P12) rats, the response of neurons from two chemosensitive regions, the VLM and the NTS, were compared with the response of neurons from two nonchemosensitive regions, the inferior olive (IO) and the hypoglossal (305). Upon exposure to HA (10% CO₂, 26 mM HCO₃^–, pH_o 7.15), neurons from the chemosensitive VLM and NTS regions acidified and remained acid during the entire exposure to HA, with pH_i returning to control values upon return to normocapnic solution (5% CO₂, 26 mM HCO₃^–, pH_o 7.48). In contrast, neurons from the nonchemosensitive IO and hypoglossal regions showed pH_i recovery from the acidification induced by HA and an overshoot of pH_i, indicative of recovery during the acid exposure (281), upon return to normocapnic solution. Neurons from VLM and NTS were shown to have pH-regulating transporters by being exposed to isohydric hypercapnia (IH: 10% CO₂, 52 mM HCO₃^–, pH_o 7.48). VLM and NTS neurons exhibited a brisk pH_i recovery from these solutions, and this recovery was unaffected by the HCO₃^–-dependent transport inhibitor 4,4'-diisothiocyanostilbene-2,2'-disulfonic acid (DIDS) but completely inhibited by the Na⁺/H⁺ exchange (NHE) inhibitor amiloride. These studies were extended to show that neurons from all four regions used NHE solely to recover from intracellular acidification (304).

The question arose as to how all four regions contained a pH recovery transporter capable of responding to cellular acidification, but this recovery was not evident in neurons from the chemosensitive VLM and NTS regions during HA. This question was answered by Ritucci et al. (304), who showed that the NHE from VLM and NTS neurons was far more sensitive to inhibition by a fall in pH_o than was the NHE in neurons from the nonchemosensitive IO and hypoglossal regions. NHE from IO and hypoglossal neurons were completely inhibited at pH_o of 6.7–6.8, whereas NHE from VLM and NTS neurons were completely inhibited by pH_o of only 7.0–7.1. These findings agree with the results of studies on the effect of anoxia on pH_i in neurons from the same regions (54). In neurons from nonchemosensitive regions (IO and hypoglossal), the fall of pH_i induced by anoxia was greater in the presence of amiloride, suggesting that NHE activity blunted the anoxia-induced acidification. In contrast, amiloride did not affect anoxia-induced acidification in neurons from chemosensitive medullary regions (VLM and NTS), indicating that NHE was inhibited during anoxia (presumably by decreased pH_o) in these neurons. Together, these studies suggest that a maintained fall of pH_i is an important part of the chemosensitive signaling pathway but that a fall of pH_o also appears to be important in the maintenance of a decreased pH_i.

Similar findings of a maintained fall of pH_i in response to HA have been made in cultured medullary raphe neurons (36), in RTN neurons from brain stem slices (255), and in neurons from organotypic medullary cultures (393). The latter study involved the use of a number of acid challenges to compare their effects on pH_i (measured in BCECF-loaded neurons) and their effects on firing rate (measured separately). A procedure that acidified both pH_o and pH_i (HA) resulted in an increase in firing rate. Replacing CO₂/HCO₃^– buffer with HEPES, which involved no change of pH_o and a slow decrease of pH_i, also resulted in an increase in firing rate. Exposure to NH₄Cl, which leads to increased pH_i (at constant pH_o), resulted in decreased firing rate, whereas removal of NH₄Cl, which leads to decreased pH_i (at constant pH_o), resulted in increased firing rate (393). These findings implicated changes of pH_i in chemosensitive signaling. This conclusion was supported by Wang et al. (382), who varied external CO₂, pH_o, and HCO₃^– independently and measured the resultant increase in firing rate in cultured raphe neurons (they did not measure pH_i). Exposure to HA (9% CO₂, 26 mM HCO₃^–, pH_o 7.17), IH (9% CO₂, 40 mM HCO₃^–, pH_o 7.4), IA (5% CO₂, 15 mM HCO₃^–, pH_o 7.16), and acidified HEPES-buffered medium (pH_o 7.2) all induced an increase in firing rate, indicating that a change of CO₂, pH_o, or HCO₃^– is not required for increased firing rate and suggesting that a change of pH_i may be the primary stimulus for chemosensitivity (382).

Filosa et al. (116) developed a new technique for simultaneously measuring V_m and pH_i in CO₂-sensitive neurons from LC. This technique involved the use of perforated patch pipettes to measure V_m. The membrane-permeable form of BCECF (BCECF-AM) was placed in the patch pipette, and loading occurred through the perforated patch. The AM groups were cleaved by intracellular esterases, thus generating the pH-sensitive and membrane-impermeable form of the dye, BCECF. In this way, with the use of fluorescence imaging microscopy, the firing rate and pH_i changes induced by various solutions could be measured simultaneously. The protocols were similar to those used by Wang et al. (382). Exposure of LC neurons to HA, IH, IA, and acidified HEPES solutions all resulted in an increase in firing rate. A plot of the magnitude of the measured change of pH_i vs. the magnitude of the increased firing rate showed a good correlation (Fig. 3A). The change in firing rate was poorly correlated with the change of CO₂, the change of HCO₃^–, or the pH_i-pH_o gradient (Fig. 3B). Furthermore, the rate of change of pH_i was more closely correlated with the rate of change of firing rate than was the rate of change of pH_o (116). This was especially clear when comparing neuronal response to HA exposure vs. IA exposure. HA resulted in a fast change of pH_o, pH_i, and firing rate, whereas IA resulted in a fast change of pH_o but a slow change of pH_i and firing rate. These findings strongly implicate changes of pH_i as the primary stimulus for chemosensitivity but do not rule out a role for changes of pH_o.

View larger version (16K): [in this window] [in a new window]   Fig. 3. A: relationship between integrated firing rate (Hz, 10-s bins) and the magnitude of the change of intracellular pH (pHi) in locus coeruleus (LC) neurons in response to a variety of acid challenges. Firing rate and pHi were measured simultaneously in individual LC neurons in a brain stem slice from neonatal rats by using perforated patch pipettes loaded with the membrane-permeable form of the pH-sensitive dye BCECF-AM (116). There is a significant correlation between the degree of intracellular acidification and the increase in the spontaneous firing rate: firing rate = 4.90(pHi) – 0.14, R2 = 0.71, P < 0.05. B: relationship between integrated firing rate and the transmembrane pH gradient (pHo – pHi, where pHo is extracellular pH) under the same conditions as in A. The relationship between firing rate and the transmembrane pH gradient is not significant: firing rate = –2.10(pHi) + 1.37, R2 = 0.33, P > 0.05. The various acid challenges were as follows (all at 35°C): isohydric hypercapnia (: 15% CO2, 77 mM HCO3–, pHo 7.45), weak acid propionate (: 50 mM propionate, 5% CO2, 26 mM HCO3–, pHo 7.45), hypercapnic acidosis (: 15% CO2, 26 mM HCO3–, pHo 6.8), isocapnic acidosis (: 5% CO2, 7 mM HCO3–, pHo 6.8), and acidified HEPES-buffered solution (: nominal absence of CO2/HCO3–, pHo 6.8). Data are represented as means ± SE of between 5 and 24 values. [Adapted from Fig. 9 of Filosa et al. (116) with permission from The Physiological Society.]