INTRODUCTION

Top Abstract Introduction Methods Results Discussion References

GLIAL CELLS HAVE BEEN IMPLICATED in brain extracellular fluid (ECF) ion and acid-base homeostasis (9, 15, 16, 22, 25, 26). The brain extracellular space is small, and increases in neuronal activity result in rapid changes in ECF K⁺ and pH (1, 14, 22, 25). These physiological changes are regulated, in part, by glia (16, 22, 25), as are changes that occur in disorders of these variables. In the brain stem, changes in ECF pH that occur with respiratory and metabolic acidosis increase ventilation. This response originates from central chemoreceptor cells located at many sites within the brain stem (4, 6, 8, 17-19, 27). The unique stimulus to these cells is not known, but changes in ECF pH are correlated with ventilatory output in many studies (9, 10, 18, 28). Therefore, factors that influence pH in or around sites of chemosensitivity may affect ventilation. The role of glia in brain stem ECF pH regulation and in the control of breathing is unknown.

Fluorocitrate (FC) has recently been described as a specific metabolic inhibitor within glial cells (3, 5, 12, 16, 23). FC selectively blocks the "synthetic" tricarboxylic acid cycle within the glial glutamine metabolic pathway by inhibiting the enzyme aconitase (3, 5, 12, 16, 23). FC treatment disrupts the glia functionally and structurally (16, 23). Glutamine levels decrease markedly and glia swell. At low doses (1 mM or less) and short durations of application (<2-3 h), these effects are reversible (3, 12, 16, 23). Largo et al. (16) used microdialysis of FC (1 mM) for 4-8 h to study the role of glia in hippocampal ECF K⁺ and pH regulation in vivo. In the first 2 h, ECF K⁺ increased slightly, and ECF pH decreased (16). At 4 h, glia were swollen. With longer FC application, more severe changes were noted in ECF K⁺ and pH, in synaptic function, and in glial morphology.

In this study, we report the effects of FC administration by diffusion pipette (1 mM) for 45-60 min in a brain stem region known to contain central chemoreceptors and to be important in the control of breathing, the retrotrapezoid nucleus (RTN). The RTN is located within a few hundred micrometers of the ventrolateral medullary surface and extends from the rostral border of the nucleus ambiguus to the rostral border of the facial nucleus bounded laterally by the spinotrigeminal tract and medially by the pyramidal tract (20, 21, 24, 29). RTN neurons discharge tonically or phasically with respiratory rhythm and increase their firing rate with increased systemic CO₂ (7). Focal acidosis in the region of the RTN increases phrenic nerve output (17), indicating the presence of CO₂ chemoreceptors within this region. Neuronal ablation at this site decreases ventilation and CO₂ sensitivity (2, 20, 21). We hypothesize that glia play an important role in the regulation of ECF pH in the RTN and that selective, reversible impairment of glial function by FC will result in an acidification of ECF pH and an increase in ventilatory output.

	METHODS

Top Abstract Introduction Methods Results Discussion References

There were four experimental groups. Group 1, the FC-treatment group, consisted of 10 rats, all of which received 1 mM FC focally administered in the RTN by using a diffusion pipette (see below). All 10 had integrated phrenic nerve amplitude (PNA) and frequency measurements before, during, and after FC administration. The PNA responses to systemic CO₂ administration were evaluated at the beginning and the end of the experiment. In two of these 10 animals (and in 1 other), ECF pH was measured at the tip of the diffusion pipette with the use of a glass pH electrode. In 6 of these 10 animals, the membrane permeability of cells affected by FC was evaluated anatomically by using ethidium homodimer-1 (DEAD Red) (13), which was injected just before FC diffusion. The presence of glial cells was evaluated by immunohistochemistry for glial fibrillary acidic protein (GFAP). Group 2 was a control group of two rats that were treated with the FC vehicle in the diffusion pipette, had measurements of PNA and frequency, and received DEAD Red. Group 3 was a control group of four rats that received DEAD Red only and were killed at 30-60 min (n = 2) or 120 min (n = 2) to monitor the spread of DEAD Red in the brain stem extracellular space. Group 4 consisted of two rats that received RTN injection of the neurotoxin kainic acid (50 nl; 5 mM), with measurement of pH at the injection site as a control for the possible effects of cell destruction on RTN pH.

The experimental protocol was approved by the Dartmouth College Institutional Animal Care and Use Committee. Adult male Sprague-Dawley rats (290-410 g) were initially anesthetized with 2.0% halothane in oxygen. Catheters were placed in the trachea, femoral artery, and vein. While monitoring blood pressure, a mixture of urethan (550 mg/kg) and -chloralose (60 mg/kg) was injected intravenously over several minutes while the inspired halothane concentration was slowly decreased. Gallamine triethiodide (3 mg/kg) was used to paralyze the rats with supplemental doses administered as needed. The rats were artificially ventilated (Harvard Instruments, rodent ventilator model 683) with 100% O₂. Rectal temperature was maintained at 37°C by using a feedback system, and end-tidal PCO₂ was monitored with a CO₂ analyzer (Biochem, ETCO₂ Monitor, Lifespan 100) and maintained at any set value by manually controlling the ventilator rate and/or tidal volume. Bilateral thoracotomies were performed, and a positive end-expiratory pressure of 3-5 cmH₂O was maintained during the experiment.

Both vagi were cut, and the ventral medullary surface was exposed. The phrenic nerve was isolated, cut, and the central end placed on bipolar electrodes and covered with Wacker Sil-gel (an insulating medium). Phrenic nerve activity was amplified (BMA 831 amplifiers, Charles Ward), integrated (Paynter filter in an MA-821 moving averager, Charles Ward), and displayed on a storage oscilloscope. Integrated PNA, arterial blood pressure, and end-tidal PCO₂ were recorded on a strip-chart recorder (MFE 1400). An on-line program using a 12-bit processor and a 150- to 200-Hz sampling rate calculated the amplitude of the integrated phrenic activity, phrenic burst frequency, blood pressure, and end-tidal PCO₂.

Blood pressure and the frequency and integrated amplitude of phrenic discharge were continuously monitored during the experiment to evaluate the depth of anesthesia. An increase in respiratory frequency or blood pressure that could not be attributed to an experimental perturbation or that occurred in response to a noxious stimulus (e.g., a pinch to the hind paw) were viewed as signs that the animal needed additional anesthesia. This was given in the form of one-fourth to one-third the initial dose.

The construction of the diffusion pipette used to deliver FC to the RTN has been described in detail elsewhere (17). Briefly, a double-barreled capillary tubing (1.5 mm OD, Frederick Haer) was pulled with a vertical pipette puller (Narashige), and the tip was broken back to a diameter of ~100 µm. One barrel was used for a pH electrode or microinjection of DEAD Red, and the other for the diffusion pipette. Polyimide capillary tubing [World Precision Instruments (WPI)] was inserted through the perfusion port of a pipette holder (model A003-2, E. W. Wright) and positioned into the diffusion pipette barrel. A second port on the pipette holder permitted the removal of solutions circulating within the diffusion pipette. The pH barrel was silanized in an oven with 5% dimethylchlorosilane (40136, Fluka) in xylene (80-100°C) overnight. The pipette tip was dipped in Hydrogen Ionophore II-Cocktail A (95297, Fluka) until hydrogen-ion exchanger was drawn up into the barrel a distance of 500-1,000 µm. Then the barrel was backfilled with 100 mM NaCl (pH 6.1). The pH barrel was coupled to a potentiometer (EA 920, Orion Research) by using a platinum-iridium wire and referenced to a KCl electrode located in the cerebral spinal fluid (Dri-Ref, WPI). All electrodes were calibrated before and after the experiment by using buffered solutions with pH values of 4, 6, 7, and 10 at 24°C. Acceptable electrodes showed a 58 mV/pH unit response, a linear calibration, and a response time of 1-5 s. In addition, the electrode responses to the initial and the final whole animal CO₂-response tests were compared as an in vivo calibration. To allow comparison of tissue pH data among animals and groups, the responses in each animal were normalized to the response of that electrode in vivo to changes in end-tidal CO₂- from 4 to 9%. The change in tissue pH was estimated by assuming a tissue value for bicarbonate of 22 mM, a tissue CO₂ solubility of 0.03 mM · l¹ · mmHg¹, and the negative log of the apparent dissociation constant, 6.1.

FC was prepared as described previously by Paulsen et al. (23). Briefly, the barium salt of FC was first dissolved in 0.1 M HCl, and the barium was precipitated from the solution with the addition of 0.1 M Na₂SO₄. This solution was then buffered (0.1 M NaPO₄) and centrifuged at 800 g for 10 min. The supernatant containing FC was removed and added to artificial cerebrospinal fluid (aCSF) to a final concentration of 1 mM. In separate control experiments, the solution was prepared in the same manner, except that FC was omitted. The aCSF contained (in mM): 126 NaCl, 3.0 KCl, 2.1 MgCl₂, 2.2 CaCl₂, 26.2 NaHCO₃, and 5 mM glucose. The perfusion rate through the pipette of these solutions was ~100 µl/min.

To determine the pattern of cells with altered membrane permeability elicited by the diffusion of FC into the RTN, the fluorescent, nucleic acid stain DEAD Red (Molecular Probes) was microinjected into the RTN. DEAD Red is a cell-impermeant dye that can only diffuse across leaky or otherwise compromised cell membranes (13). Approximately 1 µl of aCSF containing DEAD Red (1:50 dilution) was slowly microinjected into the RTN over a period of 10 min. The large volume of DEAD Red was used to ensure an adequate concentration of the dye in the extracellular space in the region of the RTN exposed to the diffusion pipette solutions. In preliminary experiments, the microinjection of DEAD Red alone into the RTN had no noticeable effect on baseline PNA. We injected DEAD Red into the RTN just before diffusion of FC to have dye present during the period of FC-induced glial dysfunction. At the end of the study period, the brain was fixed in situ by serially perfusing through the left ventricle of the heart 0.1 M PBS, pH 7.3, followed by perfusion with PBS containing 4% paraformaldehyde. After fixation, the brain was removed, frozen in dry ice, and sectioned at a 50-µm thickness with a cryostat (Cryocut 1800) maintained at 20°C. Tissue sections were mounted on glass slides, cleared in 100% DMSO, and viewed under epifluorescence by using a long-pass, FITC filter set.

To compare the staining pattern of DEAD Red with the morphology and distribution of glial cells, tissue sections were stained for GFAP. Slices were fixed as previously described and washed three times with PBS + 0.5% Triton X-100 (pH 7.4) for 30 min at room temperature. We diluted 150 µl of normal goat serum (NGS; Jackson Laboratories) into 10 ml of PBS/Triton and incubated the sections in this solution for 1 h. A mouse monoclonal anti-GFAP anti-body (Sigma Chemical) was added to the NGS/PBS/Triton solution at a dilution of 1:60, and the slices were incubated at 4°C overnight. Sections were then washed three times with PBS/Triton for 1 h. Secondary labeling with 7-amino-4-methyl coumarin-3 acetic acid-conjugated anti-mouse antibody (1:50 dilution NGS/PBS/Triton) was then performed. performed. Sections were incubated for 2 h in the dark with agitation, washed with PBS/Triton for 1 h in the dark, and cleared with 100% DMSO. Processed tissue sections were mounted on glass slides and viewed with a Nikon Optiphot 2-UD microscope. The volume of tissue containing DEAD Red-positive cells and cell diameter were measured from the digitized images by using a micrometer.

The diffusion pipette and, in some cases, the pH electrode were inserted directly into the ventrolateral medulla with the use of landmarks; 2-3 mm lateral to the midline and 1-3 mm caudal to the point of exit of the sixth cranial nerve and the caudal aspect of the trapezoid body. The tips were placed just beneath the ventral surface within 200-600 µm of the surface. All animals were first tested for their responsiveness to inspired CO₂. Baseline end-tidal PCO₂ was set at 28 Torr, slightly above the apneic threshold, whereas the animal was ventilated with 100% O₂. The end-tidal CO₂ was increased in 7-Torr steps by adding controlled amounts of 100% CO₂ into the inspired air while maintaining ventilatory frequency and tidal volume constant. Responses were measured when phrenic amplitude had stabilized (~3-5 min at any CO₂ level). When the animal had completely recovered from the CO₂ challenge and showed stable baseline phrenic activity, focal FC exposure began for 45 or 60 min, followed by 1 h of recovery; then the animal was exposed to a second hypercapnic challenge. In experiments in which DEAD Red was used, it was microinjected into the RTN before FC exposure proceeded. PNA was monitored throughout the injection period to ensure that baseline values remained stable during the period of injection.

	RESULTS

Top Abstract Introduction Methods Results Discussion References

Diffusion of FC into the RTN resulted in a rapid and progressive increase in PNA in all animals tested (n = 10; one-way ANOVA of PNA vs. time, P < 0.003; Fig. 1A). Within 5 min from the onset of FC exposure, PNA had increased 16 ± 4% (all values are means ± SE) above baseline values. PNA continued to rise for the first 30 min of FC treatment, whereupon the response began to plateau and remained elevated for the duration of FC exposure. At the end of 45-60 min of exposure to FC, PNA had increased 43 ± 5% above baseline. Cessation of FC perfusion resulted in a prompt reduction in PNA back toward control values. PNA measured 120 min after the onset of FC exposure was 11 ± 9% above baseline, not significantly different from baseline. In contrast to our findings with FC, the diffusion of vehicle alone had no significant effect on PNA (n = 2; data not shown). Respiratory frequency showed a tendency to increase in both the FC- and vehicle-treated groups during the 120 min of the experiment. In the FC group, the initial and final mean frequencies were 41 ± 3 and 52 ± 3 breaths/min, respectively. In the vehicle group, these values were 40 ± 0 and 51 ± 3 breaths/min, respectively. This frequency change is regarded as nonspecific and unrelated to FC treatment.

View larger version (10K): [in this window] [in a new window]   Fig. 1.   Integrated phrenic nerve amplitude (PNA: A) expressed as %baseline (values are means ± SE; n = 10 animals) is shown just before (t = 0), during, and after administration of 1 mM fluorocitrate (FC) by diffusion pipette for 45-60 min. B: change in tissue pH (n = 3 animals) measured at diffusion pipette tip and normalized to change in tissue pH observed with a change in end-tidal CO2 from 4 to 9% (see text for details).

The average (±SE) mean arterial blood pressure at the beginning of the experiments was 140 ± 4 mmHg, and at 120 min it was 127 ± 8 mmHg. In four animals, mean arterial blood pressure increased during FC administration, with a peak increase of 32 ± 5 mmHg that occurred at 34 ± 6 min.

The tissue pH change observed with FC treatment, normalized to the change observed during systemic stimulation with an increase in end-tidal CO₂ from 4 to 9% CO₂, showed a significant acidification (n = 3, one-way ANOVA of the change in pH vs. time, P < 0.001; Fig. 1B). Tissue acidification was most rapid during the first 15 min of exposure (Fig. 1B). At 5 and 15 min, pH had decreased by 39 ± 20 and 67 ± 49% of the maximum. We can estimate that these pH values would be approximately the same as those observed with tissue PCO₂ in the 50- to 65-Torr range. The tissue acidosis began to recover after FC exposure, and at 120 min the change in pH was 23 ± 28% of the maximum, not significantly different from baseline. The time course of the PNA and ECF pH changes was similar during both the period of FC perfusion and recovery.

Responses of PNA and frequency to elevated end-tidal CO₂ were obtained at the beginning and end of each experiment. As in our prior studies, the frequency response to increased CO₂ in this preparation was nil. For the comparison of CO₂ responses among animals, each PNA response was normalized to the maximum response observed with initial exposure to 9% end-tidal CO₂. At the baseline end-tidal PCO₂ value of 28 Torr, the control group with diffusion of FC vehicle had a normalized PNA of 62 ± 2% of maximum and the FC-treatment group had 57 ± 2% of maximum. With CO₂ stimulation, at the end of the experiment, the control group PNA increased to 83 ± 1% of maximum, and the FC treatment group increased to 73 ± 6% of maximum.

As a control for the effects of cell destruction with spillage of cell contents into the ECF, we measured pH at the tip of a double-barrelled pipette before, during the injection over 5 s of kainic acid (50 nl; 5 mM), and for the ensuing 120 min. In both cases, PNA decreased steadily, as previously reported (21), such that at 80 min it was decreased by 65 ± 14%. At 120 min, the PNA response to increased systemic CO₂ was absent. ECF pH at the pipette tip did not decrease; initially, it increased by 10 and 20% in the two cases. Over the 120-min time course, pH decreased by a small amount, such that at 120 min it was 10 and 20% below the initial value.

The locations within the RTN of the center of the FC diffusions as determined in four rats by DEAD Red fluorescence are shown in Fig. 2. The other FC- and vehicle-treated rats had diffusion pipette tips within the RTN, as judged by the pipette tracks in stained brain stem sections (not shown).

View larger version (23K): [in this window] [in a new window]   Fig. 2.   Location and size of center of 4 DEAD Red injections is shown as representative of the location of all FC administrations, made by using a cresyl violet-stained image from 1 brain shown as the small darkly stained cells. Superimposed is DEAD Red fluorescence from that and 3 other animals (Image Pro) shown as the dark, large areas at ventral surface beneath the facial nucleus. Sky Blue is added to each perfusion, and its absence in tissue sections shows that no volume has been injected at the pipette tip (see Ref. 17). VII, facial nucleus; P, pyramidal tract. Bar = 500 µm.

In group 3, we injected 1 µl of DEAD Red alone and killed the animals at 30-60 min (n = 2) and at 120 min (n = 2) to describe the observed volume of extracellular DEAD Red distribution during the time course of our protocol. We measured the mean diameter of the DEAD Red fluorescence observed in each section and the number of sections (length) that contained the fluorescence. After 30-60 min, the mean diameter was 469 µm, the length 1,200 µm, and the calculated volume was 211 nl (using a cylinder as a geometric model). After 120 min, the mean diameter was 245 µm, the length 850 µl, and the calculated volume was 55 nl.

The tissue volume that contained cells with DEAD Red fluorescence was estimated, by using a cylinder as a geometric model, in three rats that received both DEAD Red and FC. The mean diameter of the tissue volume containing DEAD Red-positive cells was 455 µm, its length 1,167 µm, and the volume 271 nl. Measurements made in a total of 283 clearly identifiable DEAD Red-labeled cells in these three brains showed an average diameter of 5.1 ± 3.9 µm (see Fig. 3).

View larger version (155K): [in this window] [in a new window]   Fig. 3.   A computer-modified image at ×40 of retrotrapezoid nucleus of a rat treated with FC and injected with DEAD Red. White areas represent DEAD Red-positive cells, presumably glia. Darker staining cells are superimposed computer-modified images of glial fibrillary acidic protein (GFAP) staining pattern of same section, showing glia with small soma and fibrillary processes. There are fewer GFAP-containing cells in the region of DEAD Red-positive cells, and GFAP staining in this region is fragmented. GFAP staining cells in the perimeter outside DEAD Red-positive cells appear normal. Bar = 100 µm.

Focal exposure of the RTN to FC had effects on the GFAP staining. Glia labeled with GFAP outside the region of FC exposure, in the contralateral RTN or the RTN exposed to control solution (no FC present), had a normal morphology in that there was punctate staining of the cell body and well-defined labeling of the projections (see Fig. 3). However, the GFAP staining pattern at the site of DEAD Red-labeled cells was distinctly different. The glial cell bodies still were well labeled, but the glial projections were no longer distinct and had a diffuse appearance (Fig. 3). These GFAP-positive cells colocalized in regions that contained many DEAD Red-labeled cells (Fig. 3).

	DISCUSSION

Top Abstract Introduction Methods Results Discussion References

The reversible impairment of glial function by FC administration (1 mM; 45-60 min) in the RTN region of the brain stem disrupts local tissue pH regulation, produces a focal tissue acidosis, and increases ventilatory output. Although a similar finding, in respect to the tissue acidosis, has been reported in the hippocampus after FC administration (16), this study provides the first evidence in vivo of the important role of glia in brain stem ECF pH regulation. The cause of the acidosis is uncertain, although glia are thought to be important in ECF pH regulation by the presence within of carbonic anhydrase and the presence at the cell membrane of ion-translocating proteins (16). An alternative cause of the ECF pH change might be the release of acid equivalents by disrupted glia. Our interpretation, i.e., that the effects of FC on PNA and ECF pH can be attributed to reversible impaired glial function, is based on three lines of evidence.

First, our anatomic evidence suggests that cells with membrane permeability altered by FC and labeled with DEAD Red also contain the glial marker GFAP and that cells labeled with DEAD Red after FC exposure were, on average, much smaller than such labeled cells after the microinjection of the neurotoxin ibotenic acid (~5 vs. 17 µm; unpublished observations). Although we cannot discount the possibility that FC selectively damages a subpopulation of small neurons, there is no support for this in the literature. The effects of FC on cells studied both in vitro and in vivo suggest that neuronal damage is related to the concentration of FC and duration of exposure (16). The concentration of FC we have used in the present study is the same used previously by Largo et al. (16) in the rat hippocampus in vivo. In their study, microdialysis of 1 mM FC for 4 h resulted in anatomic changes in glial morphology. Some limited neuronal damage was described, but only after 8 h of FC perfusion. Because our protocol restricted the perfusion of 1 mM FC for 1 h or less, neuronal involvement seems unlikely.

Second, the nature of the physiological effects of FC on PNA supports the interpretation that they are glial in origin. Diffusion of FC increased steady-state PNA 43% above baseline values. Were the effects attributable to neuronal rather than glia impairment, we would expect neurotoxin administration to mimic the results. Previous studies using the neuronal excitotoxins kainic acid and ibotenic acid (2, 20, 21) indicate that the opposite is true. Neuronal destruction in the RTN with these agents diminishes baseline PNA and decreases CO₂ sensitivity in vivo. Although glia have many of the same receptors found on neurons, including glutamate (kainate-selective) receptors (11), there is no evidence suggesting that glia are killed by these excitotoxins. Moreover, gliosis is often observed after neurotoxin injections in chronically maintained animals, suggesting that glia are spared from the effects of these agents (2). Thus, it is unlikely that the effect of FC on PNA that we report was a result of neuronal destruction.

Third, we do not think that the decrease in brain stem ECF pH is the result of leakage of acidic contents of glial cells destroyed by the FC. This interpretation is based on a number of points. Both PNA and ECF pH returned essentially to baseline, beginning immediately after the FC diffusion was stopped. The response to increased systemic CO₂ measured at the end of the experiment, 1 h after stopping the FC diffusion, was not different from the response in the control group. These observations indicate that the FC-induced effects were reversible. Data in the literature support the contention that the use of this FC dose and this period of administration produces reversible effects on glial function (3, 12, 16, 23). Furthermore, injection into the RTN of kainic acid, a procedure known to dramatically decrease respiratory output, kills neurons at the injection site (20, 21). If loss of cell contents acidifies ECF pH, then we should observe such a pH change at the pipette tip. In two rats, we did not see such an acidification at the pipette tip. In fact, there initially was an alkaline pH shift, an effect that must be specifically attributable to the kainic acid injection, since in prior experiments with injection of an aCSF control no such alkalinization was observed (19). After 2 h, there was a slight acidification, a drop of 10-20% of the maximum, much less than observed with FC treatment.

We used DEAD Red to monitor for the presence of cells within the region of FC administration with membranes sufficiently dysfunctional to allow the fluoroprobe to enter and bind to nucleic acids (13). The DEAD Red was injected in a single volume of 1.0 µl just before the period of FC diffusion in order that it be present during the period of glial disruption. Surprisingly, our control experiments suggest that DEAD Red is cleared rapidly from the ECF, its volume of distribution being 211 nl at 30-60 min and 55 nl at 120 min. These volumes are much less than expected for a 1-µl injection, and there are two likely reasons for this discrepancy. First, some dye might move up the outer surface of the pipette onto the surface of the medulla, rather than being injected into the tissue. Second, some of the DEAD Red was cleared from the extracellular space by bulk flow of ECF. In practice, the dye appears to be present initially in the extracellular space at sites more distant than this estimated volume of distribution, since we observed DEAD Red-labeled cells outside of this volume. It seems reasonable to assume that the DEAD Red volume of distribution in the initial 30 min would be larger and that FC-induced changes in glial cell function would begin within that time period. That our measured volume of DEAD Red-positive cells was 271 nl supports this interpretation. Given the apparent rapid clearance of DEAD Red, however, it seems possible that we are underestimating the tissue region that contains FC-affected glia. For these DEAD Red-positive cells, our interpretation is that the FC effect was sufficient to alter membrane permeability and allow us to visualize the cells and region affected. Overall, the functional effect of FC administered at this dose and time period was reversible. Thus either these DEAD Red-positive cells shown at the center of the FC administration site (Fig. 3) were reversibly damaged or, if their damage was irreversible, they represent but a small portion of the total FC-induced disfunction.

We attribute the initial increase in PNA to the decrease in tissue pH that occurred with FC perfusion. Focal acidification of the RTN by CO₂ diffusion pipette (17) results in an increase in baseline PNA and a decrease in ECF pH. For comparison, the estimated tissue PCO₂ that is present at the 100% normalized value is in the 70- to 72-Torr range. With ECF pH decreased 51 or 107% of maximum, PNA increased by 6 or 13% of baseline. These exposures were brief, 1-2 min. Here, with FC administration, ECF pH decreased 67% of the maximum at 15 min, and PNA increased 16% of baseline, values quite comparable to those observed with focal CO₂ diffusion. However, after 15 min, pH remained at about the same value, whereas PNA increased further to values of 43% of baseline. It is possible that this secondary response of PNA may be related to non-pH effects of FC-induced glial disruption, e.g., elevated extracellular concentrations of K⁺ or amino acids (e.g., glutamate) (16). However, the accumulation of these substances in the extracellular space after FC appears to occur over a time course longer than the duration of the present experiments (16). An alternative explanation for this secondary increase in PNA may be that a greater tissue volume is affected with FC administration than with CO₂ diffusion. With CO₂ diffusion, the volume of the tissue region with pH decreased to 67% of maximum was 33 nl. In the FC experiment, the region containing DEAD Red-labeled small cells (which presumably would also be the region with decreased pH) is estimated as 271 nl, a tissue region that is considerably larger but still only ~23% of the total volume of one RTN.

We interpret our results to indicate that reversible impairment of glial function in the RTN with FC decreases ECF pH, resulting in an increase in phrenic nerve activity via activation of RTN chemoreceptors. This finding suggests that the microenvironment surrounding chemoreceptor cells is determined, in part, by glial function. Although the effects of acidic stimuli on ventilation and putative chemoreceptor activity are well documented, this is the first study to document an important role for glia in central chemoreception. The role of glia in the sensitivity of central chemoreceptors requires further study, preferably in an unanesthetized animal model in which the depressant effects of anesthesia are absent.