INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

THE LEVEL OF CO₂/H⁺ in the blood is carefully regulated by certain neurons in several areas of the medulla oblongata. These neurons are collectively known as central chemoreceptors. The areas in which these neurons are located are known as chemosensitive areas and include the ventrolateral medulla (VLM), the nucleus of the solitary tract (NTS), and the medullary raphe (16). It is hypothesized that an increased level of CO₂/H⁺ stimulates the central chemoreceptors, which in turn, via the respiratory central pattern generator neurons (which they presumably innervate), increase ventilation (8). The stimulus to the central chemoreceptors has been the subject of much debate. It has been hypothesized that the stimulus may be an increase in molecular CO₂, a decrease in extracellular pH (pH_o), a decrease in intracellular pH (pH_i), or a combination of any of the three (1, 7, 16, 24, 25, 30). We have previously shown that both pH_i and pH_o may play a role in central chemosensitivity (22).

It would seem logical that if a change in pH is the major signaling pathway by which central chemoreceptors monitor a change in blood CO₂/H⁺, the manner in which these cells respond to acid/base disturbances should be different from that of cells that are not chemoreceptors (nonchemoreceptors). In a previous study, we found that Na⁺/H⁺ exchange is the only pH_i-regulating mechanism involved during recovery from intracellular acidification in neurons from both chemosensitive (NTS and VLM) and nonchemosensitive [hypoglossal nucleus (Hyp) and inferior olive (IO)] areas of the medulla (22). We also found that neurons from chemosensitive areas (NTS and VLM) respond with a maintained intracellular acidification during hypercapnic acidosis, but exhibit pH_i recovery during isohydric hypercapnia. This is in contrast to neurons from nonchemosensitive areas (Hyp and IO) that exhibit pH_i recovery even during hypercapnic acidosis (22). These findings suggest that the Na⁺/H⁺ exchanger is more easily inhibited by a decrease of pH_o in neurons from chemosensitive areas versus nonchemosensitive areas.

The major aim of the present study was to examine pH_i regulation in greater detail in individual neurons from chemosensitive and nonchemosensitive areas of the medulla to investigate whether other differences in pH_i regulation are present. It must be noted that these data are from neurons in known chemosensitive areas (16) but that the individual neurons themselves may or may not be chemoreceptors. Our data show the following: 1) intrinsic buffering power (_int) is the same in all neurons tested; 2) removal of extracellular chloride at steady-state pH_i results in intracellular alkalinization in all Hyp, IO, and VLM neurons but results in intracellular acidification in most NTS neurons, suggesting that Cl/HCO₃ exchange is present in all Hyp, IO, and VLM neurons but not in most NTS neurons; 3) steady-state pH_i is more dependent on pH_o in NTS and VLM neurons than in Hyp and IO neurons; 4) pH_i recovery from an intracellular acidification (mediated exclusively by Na⁺/H⁺ exchange) is inhibited at a higher pH_o in NTS and VLM neurons than in Hyp and IO neurons; and 5) pH_i recovery during intracellular alkalinization is inhibited by elevated pH_o in most NTS and VLM neurons but not in Hyp and IO neurons. These data indicate that although there are some similarities in pH_i regulation between neurons in chemosensitive and nonchemosensitive areas, there are also some major differences in pH_i regulation. These differences are consistent with a role for changes of pH_i as the signal in central chemoreception and a role for pH_o in modulating pH_i regulatory transporters. A preliminary report of these findings was previously made (21).

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Solutions and materials. Normal saline buffer (NSB) contained (in mM) 124 NaCl, 5 KCl, 2.4 CaCl₂, 1.3 MgSO₄, 1.24 KH₂PO₄, 26 NaHCO₃, and 10 glucose and was equilibrated with 5% CO₂-95% O₂ (pH 7.48 at 37°C). In the experiments on the effect of pH_o on steady-state pH_i, pH_o was manipulated in two ways: 1) CO₂ was varied at a constant NaHCO₃ concentration of 26 mM (to obtain a pH of 7.15 and 7.90; NSB was equilibrated with 10% CO₂-90% O₂ and 2% CO₂-98% O₂, respectively) and 2) NaHCO₃ was varied at 5% CO₂-95% O₂ (to obtain a pH of 7.15 and 7.90; NaHCO₃ was 13 and 65 mM, respectively). To keep pH_o constant at 7.48 (isohydric conditions) while equilibrating NSB with 2% CO₂-98% O₂ or with 10% CO₂-90% O₂, NaHCO₃ was decreased to 10.4 mM or increased to 52 mM, respectively. In all experiments where NaHCO₃ was varied, NaCl was varied reciprocally to maintain osmolarity. Normal HEPES buffer (NHB) contained the same constituents as NSB, except Na-HEPES replaced NaHCO₃ isosmotically and was equilibrated with 100% O₂. During experiments that used the NH₄Cl prepulse, NaCl was replaced isosmotically with NH₄Cl. Glucuronate replaced chloride isosmotically in the 0-chloride experiments. The concentration of calcium (as Ca-glucuronate) was increased to 12 mM to compensate for binding of calcium to glucuronate (28). The calibration solution contained (in mM) 104 KCl, 2.4 CaCl₂, 1.3 MgSO₄, 1.24 KH₂PO₄, 25 N-methyl-D-glucamine-HEPES, 25 K-HEPES, 10 glucose, and 0.016 nigericin titrated with either KOH or HCl to a pH of 7.2. Nigericin and DIDS were purchased from Sigma (St. Louis, MO) and amiloride was generously given to us by Merck (Rahway, NJ). Nigericin was added from a 16.7 mM stock solution (in DMSO), amiloride was added from a 500 mM stock solution (in DMSO), and DIDS was added directly to the solution at a final concentration of 0.5 mM. Solutions containing amiloride or DIDS were light protected. The membrane-permeable acetoxymethyl ester form of 2',7'-bis-(2-carboxyethyl)-5 (and 6)-carboxyfluorescein (BCECF-AM) was purchased from Molecular Probes (Eugene, OR) and made up in a 5 mM stock solution (in DMSO).

Preparation, BCECF-AM loading, and imaging of medullary slices. The preparation of medullary brain slices, the loading of brain slices with BCECF-AM, as well as details of imaging BCECF-AM-loaded slices have previously been described (22, 23). Briefly, transverse medullary brain slices (200-300 µm) from neonatal rats (postnatal days 0-12) were loaded with 20 µM BCECF-AM for 15 min at 37°C and washed at room temperature. Individual slices were then placed into a superfusion chamber (experiments performed at 37°C) that was then placed on the stage of an inverted Nikon Diaphot microscope. Slices were excited alternately at 500 and 440 nm, emitted fluorescence was collected at 530 nm, and fluorescence ratios (500/440) were determined. Images were collected and processed by Image-1/FL software (Universal Imaging) and stored on a Panasonic 1GB rewritable optical disk for later analysis. Fluorescence ratios were converted to pH_i by means of a calibration curve derived from the high K⁺/nigericin technique (26) as previously described (22).

Data analysis. Values are given as means ± SE. Curve fitting was done using NFIT (Island Products, Galveston, TX), with pH_o-pH_i relationships determined using a linear equation and pH_i recovery rate-pH_o relationships determined using a sigmoid equation. Student's paired t-tests and ANOVA with Bonferroni multiple-comparison t-tests with a level of significance of P < 0.05 were used for determining statistical significance.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

_int. If sensing changes in pH_i is the function of central chemoreceptors, it would be expected that they would show a maximum change in pH_i during acid/base disturbances. This would be accomplished if the central chemoreceptors possessed a low _int. We therefore measured _int using the weak acid propionate or the weak base NH₄Cl. Addition of 20 mM propionate/1 mM amiloride (amiloride was used to prevent pH_i regulation during intracellular acidification) in NHB (in the absence of CO₂/HCO₃) caused a maintained intracellular acidification in all neurons tested (Fig. 1A). _int was calculated by dividing the amount of acid (i.e., propionic acid or NH⁺₄) added to the cell by the decrease in pH_i indicated by the downward arrows in Fig. 1 (for full details, see Ref. 4). In Hyp and IO neurons (neurons from nonchemosensitive areas), _int is 40.6 ± 3.40 (n = 5) and 50.4 ± 7.49 mM/pH U (n = 13), respectively. In NTS and VLM neurons (neurons from chemosensitive areas), _int is 48.4 ± 3.13 (n = 11) and 46.9 ± 8.25 mM/pH U (n = 9), respectively. _int was also measured from the off response during an NH₄Cl prepulse (again, 1 mM amiloride was added during the off response to inhibit pH_i regulation). Similar values for _int were obtained with this technique (Fig. 1B). In Hyp and IO neurons, _int is 54.0 ± 9.15 (n = 7) and 49.0 ± 11.66 mM/pH U (n = 6), respectively. In NTS and VLM neurons, _int is 45.0 ± 7.28 (n = 14) and 37.7 ± 5.46 mM/pH U (n = 13), respectively. It was found that the _int values for each area obtained using the propionate technique are not significantly different from those obtained using the NH₄Cl technique. It was also found that _int is not significantly different between chemosensitive and nonchemosensitive areas. Therefore, the best estimate for _int for medullary neurons from neonatal rats is the overall mean for all areas, which is 46.2 ± 2.54 mM/pH U (n = 78).

View larger version (14K): [in this window] [in a new window]   Fig. 1.   Measurement of intrinsic buffering power (int). A: normal HEPES buffer (NHB) was superfused over the slice until an initial baseline intracellular pH (pHi) was obtained. This was followed by addition of 20 mM Na-propionate (prop; replacing 20 mM NaCl) + 1 mM amiloride (amil). int was estimated by the subsequent decrease of pHi (arrow). B: NHB was superfused over the slice until an initial baseline pHi was obtained. NH4Cl (15 mM; replacing 15 mM NaCl) was then added and subsequently replaced by NHB + 1 mM amiloride. int was estimated by the decrease of pHi on removal of NH4Cl (arrow).

Effect of acute removal of extracellular chloride at steady-state pH_i. We previously studied the regulation of pH_i in response to intracellular acidification in medullary neurons and found that it involves Na⁺/H⁺ exchange exclusively (22). To investigate whether these neurons also contain acidifying Cl/HCO₃ exchange (which mediates pH_i recovery from intracellular alkalinization), we followed pH_i during acute removal of extracellular chloride (replaced isosmotically with glucuronate) in the presence of CO₂/HCO₃. In other cells, if the Cl/HCO₃ exchanger is present, this procedure results in cell alkalinization due to reversal of Cl/HCO₃ exchange (e.g., Refs. 3, 6, 9, 20, 31). Removal of extracellular chloride results in a gradual intracellular alkalinization of 0.15 ± 0.01 (n = 4) and 0.09 ± 0.01 pH U (n = 5) in Hyp and IO neurons, respectively (Figs. 2, A and B, and 3), and this alkalinization is reversible on return to normal Cl-containing solution (data not shown). The 0 chloride-induced alkalinization is totally abolished by 0.5 mM DIDS (an anion exchange inhibitor) and in fact shows a slight intracellular acidification of 0.04 ± 0.008 (n = 5) and 0.005 ± 0.01 pH U (n = 4) in Hyp and IO neurons, respectively (Figs. 2, A and B, and 3). The same results were seen in VLM neurons, with an intracellular alkalinization of 0.14 ± 0.009 pH U (n = 5) on acute removal of extracellular chloride (Figs. 2C and 3). Again, this alkalinization is reversible on return to normal Cl-containing solution (data not shown). This alkalinization is also abolished by 0.5 mM DIDS and shows a slight intracellular acidification of 0.03 ± 0.008 pH U (n = 6) (Figs. 2C and 3). In ~23% of NTS neurons, acute removal of extracellular chloride also causes a gradual intracellular alkalinization of 0.14 ± 0.008 pH U (n = 3) (Figs. 4A and 5). However, a majority of NTS neurons exhibit an intracellular acidification of 0.15 ± 0.02 pH U (n = 10) in response to acute removal of extracellular chloride (Figs. 4A and 5). This intracellular acidification was most likely due to the influx of glucuronic acid (glucuronate replaced chloride isosmotically in the 0-chloride experiments). This acidification is reversible on return to normal Cl-containing solution (data not shown). In the presence of 0.5 mM DIDS, an intracellular acidification of 0.16 ± 0.02 pH U (n = 8) was seen in all NTS neurons acutely exposed to the removal of extracellular chloride (Figs. 4B and 5). Thus the Cl/HCO₃ exchanger appears to be present in Hyp, IO, and VLM neurons but absent in the majority of NTS neurons.

View larger version (15K): [in this window] [in a new window]   Fig. 2.   Effect of removal of extracellular chloride at steady-state pHi in the absence and presence of DIDS in hypoglossal (Hyp; A), inferior olive (IO; B), and ventrolateral medulla (VLM; C) neurons. Normal saline buffer (NSB) was superfused over the slice until an initial pHi baseline was obtained. Extracellular chloride was then removed (indicated by the thick line), resulting in an intracellular alkalinization. This experiment was then repeated in the presence of 0.5 mM DIDS (indicated by the thin line) and showed no intracellular alkalinization.

View larger version (18K): [in this window] [in a new window]   Fig. 3.   Change of pHi as a result of the removal of extracellular chloride at steady-state pHi in the absence and presence of DIDS in Hyp, IO, and VLM neurons. Removal of extracellular chloride caused intracellular alkalinization in all Hyp, IO, and VLM neurons (open bars). The 0 chloride-induced intracellular alkalinization was abolished in the presence of 0.5 mM DIDS (solid bars). Bars represent means ± SE. Number with each bar represents number of neurons.

View larger version (14K): [in this window] [in a new window]   Fig. 4.   Effect of removal of extracellular chloride at steady-state pHi in the absence and presence of DIDS in nucleus of the solitarty tract (NTS) neurons. A: NSB was superfused over the slice until an initial pHi baseline was obtained. Extracellular chloride was then removed (replaced isosmotically with glucuronate), resulting in an intracellular acidification in 10 of 13 NTS neurons (e.g., thin line) and intracellular alkalinization in 3 of 13 NTS neurons (e.g., thick line). B: NSB was superfused over the slice until an initial pHi baseline was obtained. Extracellular chloride was then removed (replaced isosmotically with glucuronate) in the presence of 0.5 mM DIDS and caused an intracellular acidification in all NTS neurons.

View larger version (20K): [in this window] [in a new window]   Fig. 5.   Change of pHi as a result of the removal of extracellular chloride at steady-state pHi in the absence and presence of DIDS in NTS neurons. Removal of extracellular chloride caused an intracellular alkalinization in 3 of 13 NTS neurons and caused an intracellular acidification in 10 of 13 NTS neurons (open bars). Removal of extracellular chloride in the presence of 0.5 mM DIDS produced an intracellular acidification in all NTS neurons (solid bar). Bars represent means ± SE. Number with each bar represents number of neurons.

Effect of amiloride and DIDS on steady-state pH_i. Slices were exposed to either 1 mM amiloride or 0.5 mM DIDS at steady-state pH_i for 5 to 10 min. In Hyp and IO neurons, steady-state pH_i was 7.39 ± 0.05 (n = 10) and 7.40 ± 0.03 (n = 6), respectively. On addition of 1 mM amiloride, pH_i was 7.41 ± 0.05 (n = 10) and 7.41 ± 0.04 (n = 6), respectively. In NTS and VLM neurons, steady-state pH_i was 7.48 ± 0.03 (n = 6) and 7.30 ± 0.05 (n = 9), respectively. On addition of 1 mM amiloride, pH_i was 7.46 ± 0.03 (n = 6) and 7.34 ± 0.05 (n = 9), respectively. In another set of experiments, Hyp and IO neurons had steady-state pH_i values of 7.16 ± 0.02 (n = 5) and 7.22 ± 0.03 (n = 4), respectively. On addition of 0.5 mM DIDS, pH_i was 7.16 ± 0.01 (n = 5) and 7.20 ± 0.02, respectively. In NTS and VLM neurons, steady-state pH_i was 7.50 ± 0.02 (n = 10) and 7.27 ± 0.04 (n = 6), respectively. On addition of 0.5 mM DIDS, pH_i was 7.51 ± 0.02 (n = 10) and 7.26 ± 0.03, respectively. In each case above, it was found that neither amiloride nor DIDS had any significant effect on steady-state pH_i, suggesting that the activities of both the Na⁺/H⁺ and Cl/HCO₃ exchangers are low at steady-state pH_i.

pH_o effect on steady-state pH_i. We have previously shown that neurons from chemosensitive areas of the medulla (NTS and VLM) do not regulate pH_i during hypercapnic acidosis, whereas neurons from nonchemosensitive areas of the medulla (Hyp and IO) do regulate pH_i (22). These findings suggest that pH_i may be more dependent on pH_o in neurons from chemosensitive areas compared with nonchemosensitive areas. We therefore studied the effect of pH_o on steady-state pH_i in two different ways: 1) altering pH_o by changing CO₂ at constant extracellular HCO₃ concentration ([HCO₃]_o) (Figs. 6 and 7, A and B) and 2) altering pH_o by changing [HCO₃]_o at constant CO₂ (Figs. 6 and 7, C and D). pH_o was altered over a range of 7.15-7.90 in both protocols, and at each value of pH_o, pH_i was measured after it had assumed a new steady-state value. In the Hyp and IO neurons, the slope of the relationship between pH_i and pH_o while changing CO₂ at constant [HCO₃]_o was linear and had values of 0.16 ± 0.008 (n = 38) and 0.25 ± 0.003 (n = 94), respectively (Fig. 7A). In NTS and VLM neurons, the slope was also linear but had higher values of 0.68 ± 0.004 (n = 110) and 0.56 ± 0.003 (n = 144), respectively (Fig. 7B). In the Hyp and IO neurons, the slope of the relationship between pH_i and pH_o while changing [HCO₃]_o at constant CO₂ was again linear and had values of 0.26 ± 0.014 (n = 24) and 0.35 ± 0.014 (n = 30), respectively (Fig. 7C). In NTS and VLM neurons, the slope was also linear and again had higher values of 0.84 ± 0.014 (n = 32) and 0.72 ± 0.008 (n = 30), respectively (Fig. 7D). Therefore, in both protocols, the neurons from the chemosensitive areas (NTS and VLM) had a significantly steeper pH_i versus pH_o slope than the neurons from the nonchemosensitive areas (Hyp and IO), as predicted.

View larger version (24K): [in this window] [in a new window]   Fig. 6.   Extracellular pH (pHo) effect on steady-state pHi. A: NSB equilibrated with 5% CO2-26 mM NaHCO3 (pHo 7.48) was superfused over the slice until an initial pHi baseline was obtained. This was followed by superfusion of 5% CO2-65 mM NaHCO3 (pHo 7.9). pHi increased followed by recovery to a new steady-state pHi. Similar results were seen in Hyp neurons under these conditions. Also, similar results were seen in IO and Hyp neurons when 2% CO2-26 mM NaHCO3 (pHo 7.9) was used. B: NSB equilibrated with 5% CO2-26 mM NaHCO3 (pHo 7.48) was superfused over the slice until an initial pHi baseline was obtained. This was followed by superfusion of 5% CO2-65 mM NaHCO3 (pHo 7.9). pHi increased to a new steady-state pHi with no recovery. Similar results were seen in NTS neurons under these conditions. Also, similar results were seen in VLM and NTS neurons when 2% CO2-26 mM NaHCO3 (pHo 7.9) was used. C: NSB equilibrated with 5% CO2-26 mM NaHCO3 (pHo 7.48) was superfused over the slice until an initial pHi baseline was obtained. This was followed by superfusion of 5% CO2-13 mM NaHCO3 (pHo 7.15). pHi decreased, followed by recovery to a new steady-state pHi. Similar results were seen in IO neurons under these conditions. Also, similar results were seen in Hyp and IO neurons when 10% CO2-26 mM NaHCO3 (pHo 7.15) was used. D: NSB equilibrated with 5% CO2-26 mM NaHCO3 (pHo 7.48) was superfused over the slice until an initial pHi baseline was obtained. This was followed by superfusion of 5% CO2-13 mM NaHCO3 (pHo 7.15). pHi decreased to a new steady-state pHi with no recovery. Similar results were seen in VLM neurons under these conditions. Also, similar results were seen in NTS and VLM neurons when 10% CO2-26 mM NaHCO3 (pHo 7.15) was used.

View larger version (26K): [in this window] [in a new window]   Fig. 7.   Relationship between pHo and steady-state pHi. pHo was altered in 2 ways: changing CO2 at constant HCO3 (A and B) and changing HCO3 at constant CO2 (C and D). In NTS and VLM neurons (B and D), the slope of the relationship between pHi and pHo was between 0.6 and 0.8 in both protocols. In Hyp and IO neurons (A and C), the slope of the relationship between pHi and pHo was between 0.2 and 0.3. Therefore, the neurons from the chemosensitive areas (NTS and VLM) had a steeper pHi versus pHo slope than the neurons from the nonchemosensitive areas (Hyp and IO). Each point represents mean ± SE of 5-31 neurons.

pH_o effect on pH_i recovery during intracellular acidification. The steeper pH_i-pH_o relationship in neurons from chemosensitive regions suggests that these neurons respond differently to acidification and to alkalinization than neurons from nonchemosensitive regions. As discussed above, we found that pH_i recovery from intracellular acidification in all neurons tested is mediated solely by Na⁺/H⁺ exchange. Furthermore, we found that pH_i regulation in neurons from chemosensitive areas (NTS and VLM) is inhibited during hypercapnic acidosis, whereas pH_i regulation in neurons from nonchemosensitive areas (Hyp and IO) is not inhibited (22). On the basis of these findings, we propose that the Na⁺/H⁺ exchanger from NTS and VLM neurons is more sensitive to inhibition by decreased pH_o than the exchanger from Hyp and IO neurons. To test this, we compared the effect of pH_o on pH_i recovery rate in neurons from chemosensitive areas and nonchemosensitive areas. This was done by acidifying neurons with an NH₄Cl prepulse in NHB, titrated to different values of pH_o (pH_o was titrated to the following values: 6.5, 7.0, 7.15, 7.30, 7.45, and 7.90), and calculating the initial rate of pH_i recovery. The longer the exposure to NH₄Cl, the greater the intracellular acidification on its removal. Therefore, NH₄Cl exposures were varied between 3 and 7 min, with the shorter exposures at lower pH_o values and longer exposures at higher pH_o values to produce the same intracellular acidification of between 6.7 and 6.8. In all neurons tested, the rate of recovery from acidification increased with increased pH_o and this relationship was sigmoidal (Fig. 8). However, at a pH_o of 7.15, pH_i recovery was nearly completely inhibited in neurons from chemosensitive areas (NTS and VLM) (Fig. 8, C and D), whereas it was still relatively high in neurons from nonchemosensitive areas (Hyp and VLM) (Fig. 8, A and B). The half-maximal inhibition of the Na⁺/H⁺ exchanger in Hyp and IO neurons occurs at a pH_o of 7.12 ± 0.001 (n = 20) and 7.10 ± 0.002 (n = 22), respectively, whereas the half-maximal inhibition of the Na⁺/H⁺ exchanger in NTS and VLM neurons occurs at a pH_o of 7.32 ± 0.004 (n = 30) and 7.34 ± 0.001 (n = 20), respectively. Thus, as predicted, the Na⁺/H⁺ exchanger in neurons from chemosensitive areas is significantly more sensitive to inhibition by decreased pH_o, because it is inhibited at higher values of pH_o than the exchanger in neurons from nonchemosensitive areas.

View larger version (33K): [in this window] [in a new window]   Fig. 8.   pHo effect on pHi recovery during intracellular acidification in Hyp (A), IO (B), NTS (C), and VLM (D) neurons. Neurons were acidified with an NH4Cl prepulse in NHB and titrated to different values of pHo, ranging from 6.5 to 7.9. Exposure to NH4Cl was varied at each pHo value to produce the same intracellular acidification of between 6.7 and 6.8 (shorter NH4Cl exposures at lower pHo values and longer NH4Cl exposures at higher pHo values). pHi recovery was estimated as the initial slope of the pHi vs. time trace as pHi recovered back toward initial steady-state pHi. This recovery was previously showed to be entirely due to Na+/H+ exchange (see Ref. 22). Each point represents mean ± SE of 4-10 neurons.

pH_o effect on pH_i recovery during intracellular alkalinization. We showed that the slope of the relationship between pH_i and pH_o is much more steep in NTS and VLM neurons compared with Hyp and IO neurons. This is due in part to the sensitivity of Na⁺/H⁺ exchange to inhibition by decreased pH_o in NTS and VLM neurons, as discussed above. It would seem logical to assume that the steepness of the relationship in NTS and VLM neurons is also due to the sensitivity of pH_i regulation in response to an increase in pH_o. We showed that Hyp, IO, and VLM neurons contain the acidifying Cl/HCO₃ exchanger, whereas most NTS neurons do not. The absence of Cl/HCO₃ exchange in NTS neurons would then undoubtedly contribute to the steepness of the pH_i versus pH_o relationship found in these neurons. However, the absence of Cl/HCO₃ exchange could not account for the steep relationship found in VLM neurons, because these neurons contain this exchanger. Therefore, the Cl/HCO₃ exchanger in VLM neurons must be more sensitive to an increase in pH_o than the Cl/HCO₃ exchanger in Hyp and IO neurons.

To test this possibility, neurons were exposed to isohydric hypercapnia (increased CO₂ at constant pH_o) and were allowed to recover from this acid load for a few minutes. Intracellular pH exhibits an alkaline overshoot due to pH_i recovery when hypercapnic conditions are removed (22). To accentuate this increase in pH_i and to elevate pH_o, neurons were exposed to a solution of pH_o 7.9 (hypocapnic alkalosis, 2% CO₂-26 mM HCO₃). Neurons were alkalinized by 0.2-0.4 pH U under such conditions (Figs. 9, A and B, and 10, A and B). In Hyp (Fig. 9A) and IO (Fig. 9B) neurons, acidifying pH_i recovery from this alkalinization was evident. The initial rate of this recovery was 0.017 ± 0.02 (n = 9) and 0.025 ± 0.05 (n = 5) pH U/min in Hyp and IO neurons, respectively. This recovery was completely inhibited by 0.5 mM DIDS (Fig. 9, A-C), indicating that it is mediated by Cl/HCO₃ exchange.

View larger version (24K): [in this window] [in a new window]   Fig. 9.   Effect of hypocapnic alkalosis in the absence and presence of DIDS in Hyp (A) and IO (B) neurons. Arrows indicate when DIDS was applied. NSB was superfused over the slice until an initial pHi baseline was achieved. This was followed by isohydric hypercapnia, which caused an intracellular acidification with subsequent pHi recovery toward normal. This was then followed by hypocapnic alkalosis in the absence (thin line) and presence (thick line) of 0.5 mM DIDS. In the absence of DIDS, pHi recovered toward normal during hypocapnic alkalosis. In the presence of 0.5 mM DIDS, pHi recovery was inhibited during hypocapnic alkalosis. C: initial pHi recovery rates (pH U/min ± SE) in the absence (open bars) and presence (solid bars) of 0.5 mM DIDS. [HCO3]o, extracellular HCO3 concentration. Bars represent means ± SE. Number with each bar represents number of neurons.

View larger version (20K): [in this window] [in a new window]   Fig. 10.   Effect of hypocapnic alkalosis and isohydric hypocapnia in NTS and VLM neurons. A and B: NSB was superfused over the slice until an initial pHi baseline was achieved. This was followed by isohydric hypercapnia, which caused an intracellular acidification with subsequent pHi recovery toward normal. This was then followed by hypocapnic alkalosis, which caused a maintained intracellular alkalinization in most NTS and VLM neurons. C and D: NSB was superfused over the slice until an initial pHi baseline was achieved. This was followed by isohydric hypercapnia, which caused an intracellular acidification with subsequent pHi recovery toward normal. This was then followed by isohydric hypocapnia, which caused a maintained intracellular alkalinization in most NTS neurons. VLM neurons, however, exhibited pHi recovery toward normal during the isohydric hypocapnic exposure.

In contrast to the pH_i recovery seen in Hyp and IO neurons, under similar conditions, most NTS (Fig. 10A) and VLM (Fig. 10B) neurons exhibited no pH_i recovery from intracellular alkalinization when pH_o was 7.9. In fact, NTS and VLM neurons exhibited a slight alkaline drift under these conditions, amounting to 0.013 ± 0.006 (n = 10) and 0.003 ± 0.001 (n = 16) pH U/min, respectively. In each region, a few neurons did exhibit acidifying recovery, with a rate of 0.009 ± 0.003 (n = 2) pH U/min in NTS neurons and 0.006 ± 0.001 (n = 3) pH U/min in VLM neurons.

The lack of pH_i recovery in the majority of NTS neurons is most likely due to the fact that the majority of these neurons do not have the acidifying Cl/HCO₃ exchanger. We propose that the lack of pH_i recovery in the majority of VLM neurons at pH_o 7.9 is due to inhibition of the Cl/HCO₃ exchanger by elevated pH_o. To test this hypothesis, NTS and VLM neurons were again exposed to isohydric hypercapnia and pH_i was allowed to recover from this acid load for a few minutes. Neurons were then exposed to isohydric hypocapnia (2% CO₂-10.4 mM HCO₃, with constant pH_o of 7.48), which resulted in an intracellular alkalinization of 0.2-0.4 pH U with no change in pH_o (Figs. 10, C and D, and 11). The majority of NTS neurons again did not exhibit acidifying pH_i recovery (Figs. 10C and 11) due to the lack of Cl/HCO₃ exchange. In those neurons that did not show an acidifying pH_i recovery, a slight alkaline drift of 0.009 ± 0.002 (n = 10) pH U/min was seen. In those neurons that did recover, the rate of recovery was 0.020 ± 0.007 (n = 2) pH U/min. In VLM neurons, however, acidifying pH_i recovery was seen in all neurons tested (Figs. 10D and 11). The rate of recovery was 0.011 ± 0.001 (n = 10) pH U/min. This recovery was completely inhibited by 0.5 mM DIDS, amounting to only 0.005 ± 0.003 pH U/min (n = 8) (data not shown). These data are consistent with the hypothesis that elevated pH_o inhibits the Cl/HCO₃ exchanger in VLM neurons.

View larger version (20K): [in this window] [in a new window]   Fig. 11.   Initial pHi recovery rates ± SE during hypocapnic alkalosis (open bars) and isohydric hypocapnia (solid bars). Number with each bar represents number of neurons.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

We have hypothesized that a change of pH_i in central chemoreceptors is the proximate signal required for transducing a change in blood CO₂/H⁺ into a change in ventilation (22). If this is the case, one would expect that pH_i regulation is different in chemoreceptor compared with nonchemoreceptor cells. For instance, central chemoreceptors, in response to an appropriate stimulus (i.e., during an acid/base disturbance), should produce a large signal (i.e., a large change in pH_i), and this signal should be maintained for the entire duration of the stimulus. This is in contrast to nonchemoreceptors, which typically maintain a more constant pH_i in the face of acid/base disturbances due to membrane-bound pH-regulating transporters (19). The following characteristics would allow central chemoreceptors to function in this manner: 1) a low _int to maximize the change in pH_i during acid/base disturbances, 2) the ability of pH_i to closely track pH_o, and 3) the lack of and/or inhibition of pH_i-regulating mechanisms during acid/base disturbances.

For any given acid/base disturbance, the change in pH_i will be larger the smaller the _int. We therefore hypothesized that the neurons from chemosensitive areas (NTS and VLM) would possess a low _int to maximize changes in pH_i during acid/base disturbances. This is not the case, however. We found that in all neurons tested, _int is not significantly different in chemosensitive versus nonchemosensitive areas and has a value of ~46 mM/pH U. The _int values previously measured in neurons vary widely from ~5 to 60 mM/pH U (14), which makes our measured value on the high end. This may be due to the fact that our measurements were made in neonatal rats. It is well established that the newborn brain is much more resistant than the adult brain to anoxic and hypoxic damage (10, 13). During anoxia and hypoxia, neurons are faced with an increase in intracellular acid production, and large decreases in pH_i (below pH 6.6) have been found to contribute to cell death (12, 17). To minimize a change in pH_i during anoxia or hypoxia, neurons would need a high _int. Thus neurons from neonates may have large values of _int to protect the brain from periods of anoxia and hypoxia.

One would still expect that NTS and VLM neurons (neurons from chemosensitive areas) should have a smaller _int than the Hyp and IO neurons (neurons from nonchemosensitive areas). However, the likelihood of acid-induced neuronal death is also directly related to the duration of the acid exposure, such that the longer the exposure the greater the incidence of cell death (12, 17). Because an ideal chemoreceptor would maintain an intracellular acidification during, for example, metabolic acidosis and the acidification may take place for extended periods of time, the neuron may not be able to survive if the maintained acidification is too great. Therefore, chemoreceptors probably would not be able to survive if _int were too low. We thus propose that medullary chemosensitive neurons possess a large _int to decrease the chances of cell death during a prolonged acid disturbance and that an alteration of _int is not a part of the mechanism that allows central chemoreceptors to monitor blood CO₂/H⁺ levels.

The regulation of pH_i is mediated by membrane transport systems that move acid/base equivalents across the cell membrane (19). We have previously shown that in response to acidification medullary neurons from all four areas studied exhibit recovery that is mediated solely by Na⁺/H⁺ exchange, with no contribution from HCO₃-dependent transport (22). In most cells (19), including neurons (3, 9, 18, 20), there is Na⁺-independent Cl/HCO₃ exchange, which is believed to acidify the cell in response to an alkaline load. In agreement with these studies, we found evidence for Cl/HCO₃ exchange in neurons from three of the four areas that we studied (VLM, IO, and Hyp) (Figs. 2 and 3). In contrast, nearly 80% of the neurons from the NTS showed no Cl/HCO₃ exchange (Figs. 4 and 5). Although the absence of Cl/HCO₃ exchange does not appear to be common in neurons, it is not surprising that a group of neurons that does not appear to regulate pH_i in the face of a base disturbance (i.e., NTS neurons) does not possess Cl/HCO₃ exchange.

One interesting aspect of the 0-chloride experiments is the acidification seen in the majority of NTS neurons. Because chloride was replaced by glucuronic acid, we propose that an influx pathway for glucuronic acid is present in NTS neurons and that this pathway causes the acidification. If the influx pathway for glucuronic acid is present in Hyp, IO, and VLM neurons also, we would expect that in 0-chloride solutions (in the presence of DIDS), neurons from these regions would acidify to the same extent as NTS neurons (~0.16 pH U). However, neurons from these regions only acidify by at most 0.04 pH U under these conditions. Thus the influx pathway for glucuronic acid appears to be present only in NTS neurons. Because glucuronic acid is not normally present in the brain, we assume this putative influx pathway normally transports other organic compounds. Why it would be localized to NTS neurons alone is unclear.

These data give a clearer picture of pH_i regulation in medullary neurons. In most of these neurons, pH_i is regulated back toward normal from an acid load by Na⁺/H⁺ exchange and from an alkaline load by Cl/HCO₃ exchange. This is similar to the pattern of pH_i regulation found in rat cortical neurons (18). In such cells, steady-state pH_i is maintained at the point at which acid extrusion (due to Na⁺/H⁺ exchange) is equal to acid loading (due to Cl/HCO₃ exchange, H⁺ influx, and metabolic acid production) (19). In medullary neurons, our data suggest that the activity of each exchanger at this "steady-state" pH_i is probably small because neither the inhibition of the Na⁺/H⁺ exchanger with amiloride nor the inhibition of the Cl/HCO₃ exchanger with DIDS has a significant effect on pH_i. These findings also suggest that the background acid loading of these cells (due to H⁺ influx and metabolic acid production) is also quite small.

The pattern of pH_i regulation appears to vary in neurons from different chemosensitive areas. VLM neurons contain the Cl/HCO₃ exchanger and fit the pattern of pH_i regulation described above. In contrast, the majority of NTS neurons do not appear to contain the Cl/HCO₃ exchanger. The significance of this difference in the neurons from these two different chemosensitive areas is not clear. However, the lack of the anion exchanger in NTS neurons indicates that the Na⁺/H⁺ exchanger plays a predominant role in determining steady-state pH_i in these cells. It is interesting to note that NTS neurons have a higher steady-state pH_i than the neurons from the other areas studied (22). This might be due either to the lack of the acidifying Cl/HCO₃ exchanger or possibly to a higher set point for the Na⁺/H⁺ exchanger.

The second predicted characteristic of an ideal central chemoreceptor is that its pH_i should closely track pH_o, as opposed to most cells, which maintain a fairly constant pH_i in the face of changes in pH_o. We indeed found that the relationship between pH_o and steady-state pH_i is very steep (slope of ~0.6-0.8) in neurons from chemosensitive areas (NTS and VLM) (Fig. 7). This steep relationship is similar to that found in other chemoreceptive cells, including glomus cells (peripheral chemoreceptors) of the carotid body, where a unit change in pH_o also leads to a 0.6-0.8 unit change in steady-state pH_i (5, 11, 31) and in acid-sensing taste receptor cells, where a unit change in pH_o leads to a 0.7-1.0 unit change in steady-state pH_i (15). This steep relationship is greater than in most other cells, where a unit change in pH_o leads to only an ~0.3 U change in steady-state pH_i (2, 27, 29), which is similar to what we found in neurons from nonchemosensitive areas (0.2-0.3 in Hyp and IO).

Our data suggest the basis for why the slope of the pH_i-pH_o relationship is higher in neurons from chemosensitive versus nonchemosensitive areas. We previously showed that pH_i recovery is inhibited in response to hypercapnic acidosis in NTS and VLM neurons but not in Hyp or IO neurons (22). It is known that this pH_i recovery is mediated entirely by Na⁺/H⁺ exchange (22), and a decrease in pH_o can inhibit this exchanger (19). Taken together, these findings suggest that the Na⁺/H⁺ exchanger is more sensitive to inhibition by reduced pH_o in neurons from chemosensitive areas than in neurons from nonchemosensitive areas. We verified this by directly measuring the pH_o sensitivity of the exchanger in neurons from the different medullary areas (Fig. 8). The Na⁺/H⁺ exchangers in Hyp and IO neurons have a half-maximal inhibition at about pH 7.15 and are fully inhibited only when pH_o is <6.7. In contrast, the Na⁺/H⁺ exchangers from NTS and VLM neurons have a half-maximal inhibition at ~7.35 and are fully inhibited at pH_o of 7.0. The basis for this altered pH_o sensitivity of Na⁺/H⁺ exchange in neurons from chemosensitive regions is unknown. It is unlikely that chemosensitive neurons have a completely different isoform of the exchanger. Currently four isoforms of the Na⁺/H⁺ exchanger are generally recognized, with NHE-1 being the most common (19). This is the "housekeeping" isoform, which is nearly ubiquitous and responsible for pH_i regulation. The other three isoforms are more specialized and in general have a reduced sensitivity to inhibition by amiloride and its analogs. Therefore, we think it is most likely that all of the neurons we studied contain the NHE-1 isoform but that the exchange protein is altered in chemosensitive neurons, rendering it more susceptible to inhibition by pH_o.

The greater sensitivity of Na⁺/H⁺ exchange to inhibition by decreased pH_o is in part responsible for the greater slope of the pH_i-pH_o relationship in neurons from chemosensitive areas (i.e., NTS and VLM). In these neurons, no pH_i recovery occurs from intracellular acidification caused by extracellular acidosis and thus the maintained fall in pH_i will more closely match the fall in pH_o. In contrast, neurons from nonchemosensitive areas exhibit pH_i recovery from intracellular acidification, and thus steady-state pH_i will be much less affected by a decrease in pH_o.

A reduced pH_i recovery from alkalinization when pH_o is elevated would also contribute to a steeper pH_i-pH_o relationship in neurons from chemosensitive regions. Our data clearly show that even at pH_o 7.9, neurons from the nonchemosensitive Hyp and IO regions exhibit pH_i recovery from alkalinization (Fig. 9). This recovery is completely abolished by DIDS, indicating that it is mediated by Cl/HCO₃ exchange. This recovery would tend to minimize the change in pH_i when pH_o is elevated. In contrast, most neurons from the chemosensitive NTS and VLM regions do not exhibit pH_i recovery from alkalinization when pH_o is 7.9 (Fig. 10, A and B). This lack of pH_i recovery is expected for NTS neurons, because most of them lack Cl/HCO₃ exchange (Figs. 4 and 5) and do not exhibit pH_i recovery from alkalinization even when pH_o is normal (Figs. 10C and 11). However, this lack of pH_i recovery is surprising in VLM neurons, where the Cl/HCO₃ exchanger is present and active at normal pH_o (Figs. 2C, 3, 10D, and 11). We observed that the Cl/HCO₃ exchanger from VLM neurons appears to be more sensitive to inhibition by elevated pH_o than the exchanger from Hyp and IO neurons. To our knowledge, inhibition of Cl/HCO₃ exchange by elevated pH_o has not previously been shown.

It is possible that the pH_i recovery seen in these neurons is due to an HCO₃ channel that would mediate acidifying HCO₃ efflux when pH_i is alkaline. Such a channel would have to be inhibited by DIDS, absent from most NTS neurons, and inhibited by high pH_o in VLM neurons to be consistent with our data. The efflux of HCO₃ through such a channel would be largely dependent on membrane potential (V_m), and thus definitive evidence for such a channel must await our simultaneous measurements of pH_i and V_m in individual medullary neurons. Because we already have evidence for the presence of Cl/HCO₃ exchange in Hyp, IO, and VLM neurons (Figs. 2 and 3), we believe that it is most reasonable to assume that the acidifying recovery we see in alkaline neurons is mediated by Cl/HCO₃ exchange.

The lack of pH_i recovery in response to intracellular alkalinization in neurons from chemosensitive regions when pH_o is elevated would contribute to the steeper pH_i-pH_o relationship in these cells. In fact, this relationship is steepest in NTS neurons, most of which lack the Cl/HCO₃ exchanger. It is interesting that the basis for this lack of recovery from alkalinization differs between NTS neurons (which lack the exchanger) and VLM neurons (where the exchanger is inhibited by elevated pH_o). It is not presently clear why there are differences in the pH_i regulation between the neurons in these two regions.

With the exception of _int, our data show that neurons from chemosensitive areas do indeed have altered pH_i regulation. Both the Na⁺/H⁺ exchanger and Cl/HCO₃ exchanger (if present) appear to be more sensitive to inhibition by changes in external pH in neurons from chemosensitive versus nonchemosensitive areas. As a consequence, neurons from chemosensitive areas exhibit a greater dependence of pH_i on pH_o than neurons from nonchemosensitive areas. These alterations in pH_i regulation are likely a part of the signaling pathway that enables chemosensitive neurons to respond to changes in extracellular CO₂/H⁺ and thus function as central chemoreceptors.

Perspectives

Future studies will focus on several issues that arise from this work. As neurons from various regions of the brain are studied, it is hoped that a pattern for the distribution of the different pH_i-regulating transporters will become evident. We are currently developing techniques to measure the electrical and pH_i responses of individual neurons to hypercapnia simultaneously to correlate pH changes with alterations in chemoreceptor excitability. Using this technique, we also hope to determine the cellular distribution of the Na⁺/H⁺ and the Cl/HCO₃ exchangers between the cell soma and the dendritic processes. It will also be of significance to determine the molecular alterations that render the exchangers more sensitive to pH_o in neurons from chemosensitive regions. Finally, determining the intracellular targets of altered pH_i should further elucidate the signaling pathway involved in chemoreception.