Role of intracellular and extracellular pH in the chemosensitive response of rat locus coeruleus neurones

Abstract

The chemosensitive response of locus coeruleus (LC) neurones to changes in intracellular pH (pH_i), extracellular pH (pH_o) and molecular CO₂ were investigated using neonatal rat brainstem slices. A new technique was developed that involves the use of perforated patch recordings in combination with fluorescence imaging microscopy to simultaneously measure pH_i and membrane potential (V_m). Hypercapnic acidosis (15 % CO₂, pH_o 6.8) resulted in a maintained fall in pH_i of 0.31 pH units and a 93 % increase in the firing rate of LC neurones. On the other hand, isohydric hypercapnia (15 % CO₂, 77 mm HCO₃⁻, pH_o 7.45) resulted in a smaller and transient fall in pH_i of about 0.17 pH units and an increase in firing rate of 76 %. Acidified Hepes (N-2-hydroxyethylpiperazine-N′-2- ethanesulfonic acid)-buffered medium (pH_o 6.8) resulted in a progressive fall in pH_i of over 0.43 pH units and an increase in firing rate of 126 %. Isosmotic addition of 50 mm propionate to the standard HCO₃⁻-buffered medium (5 % CO₂, 26 mm HCO₃⁻, pH_o 7.45) resulted in a transient fall in pH_i of 0.18 pH units but little increase in firing rate. Isocapnic acidosis (5 % CO₂, 7 mm HCO₃⁻, pH_o 6.8) resulted in a slow intracellular acidification to a maximum fall of about 0.26 pH units and a 72 % increase in firing rate. For all treatments, the changes in pH_i preceded or occurred simultaneously with the changes in firing rate and were considerably slower than the changes in pH_o. In conclusion, an increased firing rate of LC neurones in response to acid challenges was best correlated with the magnitude and the rate of fall in pH_i, indicating that a decrease in pH_i is a major part of the intracellular signalling pathway that transduces an acid challenge into an increased firing rate in LC neurones.

Alterations in the partial pressure of carbon dioxide (P_CO2) and pH are known to change the activity of specialized central neurones referred to as chemosensitive neurones. We define a chemosensitive neurone as one whose firing rate is increased by an increase in CO₂/H⁺. Such neurones have been localized to a number of brainstem centres involved in a variety of physiological functions including the regulation of cardiovascular and respiratory systems. These centres include the ventral medullary surface, the ventrolateral medulla (VLM), the solitary tract (NTS), the medullary raphe and the locus coeruleus (LC) (Nattie, 1995, 1998a,b, 1999; Richerson, 1995; Ritucci et al. 1998). The last of these regions, the LC, is located in the dorsal pons of the brainstem. It is the nucleus in the central nervous system with the highest number of noradrenergic neurones and has been associated with a number of physiological and behavioural functions including the sleep-wake cycle, feeding, cardiovascular and respiratory control, nociception, attention and learning (Hobson et al. 1975; Aston-Jones, 1985; Oyamada et al. 1998). Among the different central nuclei expressing CO₂/H⁺-sensitive neurones, the LC has the largest percentage (> 80 %) of neurones excited by high CO₂/H⁺ (Pineda & Aghajanian, 1997; Oyamada et al. 1998). The high degree of chemosensitivity makes this nucleus an ideal one for the study of the cellular pathways by which an increase in CO₂/H⁺ leads to an increase in neuronal discharge.

Increased neuronal firing rate in response to high CO₂/H⁺ could be brought about by changes in three possible signals: extracellular pH (pH_o), intracellular pH (pH_i) or molecular CO₂. There is evidence indicating that a change in pH_i (intracellular acidification), rather than in pH_o or molecular CO₂, is the major signal responsible for the increase in firing rate seen in chemosensitive neurones (Lassen, 1990; Putnam, 2001; Wiemann & Bingmann, 2001). Most cells respond to intracellular acidification with pH_i recovery due to activation of pH-regulating membrane transporters. However, several studies have shown that neurones from various chemosensitive regions do not exhibit pH_i recovery from intracellular acidification when pH_o is also reduced (Ritucci et al. 1997; Putnam, 2001; Wiemann & Bingmann, 2001), e.g. in response to hypercapnic acidosis (HA). For instance, neurones from the NTS and the VLM show sustained intracellular acidification without pH_i recovery in response to hypercapnic acidosis (Ritucci et al. 1997). This lack of pH_i recovery is due to inhibition, by extracellular acidification, of the only pH regulatory transporter, the Na⁺-H⁺ exchanger (NHE), that mediates recovery from cellular acidification in NTS and VLM neurones (Ritucci et al. 1997). On the other hand, pH_i recovery does occur in these neurones in response to intracellular acidification when pH_o is normal, e.g. isohydric hypercapnia (IH) (Ritucci et al. 1997). Despite these various studies, the direct role played by changes in various potential signals in increasing neuronal firing rate is still to be addressed. One of the major goals of the current study was to address the role of changes in pH_i, pH_o and molecular CO₂ in increasing neuronal firing rate.

Simultaneous pH_i and membrane potential (V_m) measurements have provided a new means for the understanding of the cellular mechanisms underlying chemosensitivity (Putnam, 2001). One technique for making such simultaneous measurements involves impaling chemosensitive neurones using sharp-tipped electrodes and injecting pH-sensitive fluorescent dyes into the neuronal cell body (Ritucci et al. 1999). However, such measurements are difficult and thus have a low rate of success. An easier technique for simultaneously measuring V_m and pH_i involves the use of whole-cell patch recordings. When the pipette contains a fluorescent dye, the neuronal cell rapidly loads with the dye. Using whole-cell recordings to simultaneously measure V_m and pH_i in LC neurones, Ritucci et al. (1999) showed pH_i changes in response to hypercapnia that were identical to those seen with sharp-tipped electrodes, but the V_m response was lost. A similar loss of the V_m response to hypercapnia in chemosensitive neurones from other brainstem regions when using whole-cell recordings (Dean & Reddy, 1995; Richerson, 1995) suggests that a key player in the chemotransduction pathway is lost by diffusional exchange with the pipette solution. This phenomenon could explain the loss of the increased firing rate observed during hypercapnia when using whole-cell recordings (Ritucci et al. 1999). For these reasons we developed a new technique that combines perforated patch recordings with fluorescence imaging microscopy to simultaneously measure V_m and pH_i changes, respectively, in LC neurones. We used this technique to study the role of changes in pH_i, pH_o and molecular CO₂ in the response of LC neurones to different conditions of acid challenge. Our results show that pH_i changes are best correlated with the increased firing rate seen in LC neurones during acid challenges.

Preliminary reports of this work have been published previously (Filosa et al. 2000, 2001).

METHODS

Slice preparation

Pontine LC slices were prepared from neonatal Sprague-Dawley rats (postnatal (P) age P1-9). We have previously shown that the electrophysiological response of these slices to hypercapnia is identical to the adult response (Stunden et al. 2001). All procedures were in agreement with the Wright State University Institutional Animal Care and Use Committee guidelines and were approved by the committee. Wright State University is accredited by AAALAC and is covered by NIH Assurance (no. A3632-01). Rats were anaesthetized by hypothermia (to ∼15 °C) and rapidly decapitated (Cunningham & McKay, 1993). The brainstem was subsequently removed and placed onto a vibratome (PelcoVibratome 1000). Coronal brainstem slices (300 μm) were cut into artificial cerebrospinal fluid (aCSF; for composition see below) at 4-6 °C. Slices were then incubated at room temperature in aCSF equilibrated with 95 % O₂-5 % CO₂ (pH ∼7.45) until needed. At the time of the experiment, slices were placed in a perfusion chamber and held with a nylon grid. Brain slices were continuously superfused through stainless steel tubing via a gravity fed system at a rate of 3-5 ml min⁻¹. Solutions were maintained at 40 °C in a reservoir in a water bath and bubbled with the equilibrating gas. Solution flowed from the reservoir through stainless steel tubing (o.d. 1.6 mm), a glass bubble trap and finally through a thermoelectric Peltier assembly (Kelso et al. 1983) to maintain temperature in the perfusion chamber at 35 ± 2 °C.

Solutions

The aCSF was composed of (mm): KCl 5, NaCl 124, MgSO₄ 1.3, NaHCO₃ 26, KH₂PO₄ 1.24, glucose 10, CaCl₂ 2.4 and equilibrated with 95 % O₂-5 % CO₂, pH ∼7.45. The composition of the HA solution was the same as that of the aCSF solution with the exception that it was bubbled with 85 % O₂-15 % CO₂, pH ∼6.8. For the IH solution, NaHCO₃ was increased to 77 mm (replacing NaCl isosmotically) and equilibrated with 85 % O₂-15 % CO₂, pH ∼7.45. For the N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid (Hepes)-buffered aCSF, NaHCO₃ was replaced with 26 mm Hepes and the pH titrated to 7.45 or 6.8 with NaOH or HCl. The NH₄Cl solution was similar to the aCSF solution with the exception that 30 mm NaCl was replaced isosmotically with 30 mm NH₄Cl. As with the NH₄Cl solution, the propionate solution consisted of aCSF solution with 50 mm NaCl being isosmotically replaced with 50 mm sodium propionate. Finally, for the isocapnic acidotic (IA) solution, NaHCO₃ was lowered to 7 mm and NaCl increased to 143 mm to obtain an isosmotic solution (pH ∼6.8). The osmolarity of all solutions was measured and adjusted to 300 mosmol l⁻¹ with the major salt of the solution. The pH calibration solution consisted of (mm): KCl 104, MgSO₄ 1.3, KH₂PO₄ 1.24, N-methyl-d-glucamine (NMDG)-Hepes 25, K-Hepes 25, glucose 10, CaCl₂ 2.4 and nigericin 0.004 and was titrated with either KOH or HCl to pH values of 6.7, 7.2 or 7.7.

The change in pH_o upon changing to different solutions was monitored in separate experiments. Perfusion was achieved as described above. A micro-combination pH electrode (MI-414, Microelectrode, Inc., Bedford, NH, USA) was placed in the perfusion chamber and pH_o was recorded every 20 s before, during and after switching from aCSF to HA, IH, IA, propionate and acidified-Hepes (AcHepes) solution.

Electrophysiological recordings and dye-loading

We have developed a new technique for loading the pH-sensitive dye BCECF (2′,7′-bis(2carboxyethyl)-5-(and 6)-carboxyfluorescein) using a perforated patch pipette. Perforated patch recordings were performed using a technique modified from Rae et al. (1991). Briefly, pipettes were made with thin-walled filament borosilicate glass (o.d. 1.5 mm, i.d. 1.12 mm) pulled to a tip resistance of ∼5 MΩ. The tip of the pipette was filled with an internal solution free of amphotericin B. The rest of the pipette was backfilled with the same internal solution containing amphotericin B (240 μg ml⁻¹). The composition of the intracellular pipette solution was (mm): 130 CH₄O₃S, 130 KOH, 20 KCl, 5 Hepes and 1 EGTA, pH ∼7.35. An amphotericin B stock was prepared with 6 mg amphotericin B dissolved in 100 μl DMSO (Rae et al. 1991). Dye-loading into a patched neurone was achieved by addition of the membrane-permeant, acetoxymethyl ester (AM) form of BCECF, BCECF AM. The internal solution at the tip of the pipette lacked BCECF AM but the rest of the electrode was backfilled with a solution containing 100-150 μM BCECF AM. LC neurones were visualized (× 720 magnification) with an upright microscope (Nikon Eclipse E600) using a × 40 water-immersion objective (NA 0.8, 3.0 mm working distance). A slight positive pressure was constantly applied to the pipette while it was moved towards a neurone to prevent the tip from becoming obstructed and to deflect debris from the neuronal surface. Gigaseals were obtained using a slight negative pressure applied at the time the pipette touched the neuronal cell body. Experiments were started at the point where a stable V_m of −45 to-60 mV was established.

Electrophysiological recordings were conducted in ‘current-clamp mode’. The criteria for selecting a healthy LC neurone were that it had a V_m (measured with a Dagan BVC-700 Cornerstone amplifier) ranging from −45 to −60 mV and an initial integrated firing rate < 4 Hz. If a neurone was not spontaneously active it was not used to study chemosensitivity. Access resistance varied from neurone to neurone, ranging from 10 to 60 MΩ. Signals were filtered (1 kHz) and stored on tape for further analysis. Spikes and integrated firing rate were recorded with a multichannel slope/ height window discriminator (FHC Model 700B, Bowdoinham, ME, USA). Membrane potential and firing rate were analysed using pCLAMP software version 8.0.2 (Axon Instruments). We minimized possible junction potentials by using a 3 m KCl agar reference electrode in the bath attached to a chlorided silver wire.

pH imaging

pH was measured with fluorescence imaging microscopy, as described previously (Ritucci et al. 1996). All pH_i measurements were recorded from the cell soma. Briefly, a dye-loaded cell was alternately excited with light at 440 and 500 nm (Omega Optical filters; 440 ± 10 nm and 500 ± 10 nm) using a xenon lamp and a Sutter Lambda 10-2 filter wheel. Emitted fluorescence at 530 nm was directed to a Nikon multi-image port module and then directed to a GenIIsys image intensifier and a CCD 100 camera (both from Dage-MTI). To minimize light-induced changes in the electrophysiological properties of LC neurones, neutral density filters of 4, 8 or 16 were used, depending on the fluorescence intensity of the loaded neurone. Image acquisition took ∼2 s and was repeated at 30-60 s intervals, with no excitation light between intervals to further reduce photobleaching. Fluorescence images were acquired by a Gateway 2000 E-3100 computer, collected and processed using Metafluor 4.0 software (Universal Imaging), and the 500/440 fluorescence ratio (R_Fl) determined. Intensified data at each excitation were corrected by subtracting a background image (collected with no excitation light) prior to calculation of the ratio. Calibration was achieved with the high K⁺-nigericin technique (Thomas et al. 1979).

Data analysis

Values are expressed as means ± 1 s.e.m. The maximum change in firing rate and pH_i were determined as the difference between control values (average value for the 2 min before solution change) and the maximum (firing rate) or minimal (pH_i) value after solution change. The rate of pH_i recovery in the maintained presence of an acid challenge (IH and propionate) was estimated by a least-squares regression line (pH units s⁻¹) through at least five points, starting with the minimum value of pH_i. To compare the initial rates of fall in pH_i (pH units s⁻¹) and the initial rates of increase in firing rate (impulses s⁻¹ min⁻¹) between HA and IA, least-squares regression lines were fitted from the point just before solution change to the point 4 min after solution change (regressions with 5 points for pH_i and 9 points for firing rate). Values of percentage change in pH_i were calculated by dividing the difference between the average pH_i value before acid challenge and each pH_i value during acidification by the maximum change in pH_i during the acid challenge (this ratio was multiplied by 100 to obtain percentage change). Similarly, values of percentage change in firing rate were calculated by dividing the difference between each firing rate value during an acid challenge and the average firing rate before acidification by the maximum change in firing rate during the acid challenge (this ratio was also multiplied by 100 to obtain percentage change). All differences between two means were determined using Student's t tests and were considered significant at P < 0.05. All differences among three or more means were determined by ANOVA. For the curves of percentage change vs. time, Dunnett tests were run to determine the first time at which the percentage change was significantly different (P < 0.05) from control values before an acid challenge. Multiple paired comparisons were made using Tukey's pairwise comparison tests, with a level of significance of P < 0.05. For the correlations between the change in pH_i and firing rate with different acid challenges, r values were tested for significance using a test for independence (Dixon & Massey, 1969).

RESULTS

Technique for simultaneous measurement of pH_i and V_m

We have developed a new technique to measure pH_i and V_m simultaneously using the perforated patch configuration. The basis for this technique is the loading of the dye into the cell through the patch pipette. BCECF AM is membrane permeant and readily diffuses into the neuronal cell body across the membrane through the tip of the perforated patch pipette. Once inside the cell the dye becomes de-esterified to its charged and fluorescent form (BCECF). Using this technique we achieved excellent dye-loading of neuronal cell bodies (Fig. 1). The dye concentration in the cell body at any given time depends on the balance between continued loading of BCECF from the pipette, loss of the BCECF dye into neuronal processes as well as to the extracellular space, and photobleaching. A major advantage of this technique is that fluorescence can be measured using a patch configuration that prevents washout of cytosolic factors that form part of the mechanism under study. In addition, this method of dye-loading makes visualizing the dye-loaded neurone much easier since that cell is virtually the only cell loaded.

Figure 1. BCECF-loaded locus coeruleus (LC) neurone.

Fluorescence microscopic image (500 nm excitation) of a LC neuronal cell body loaded with BCECF using a perforated patch pipette. Note the faintly loaded neuronal process. Inset, a BCECF-loaded LC neurone surrounded by loaded neighbouring cells. Scale bar, 20 μm.

In fact, we often saw adjacent cells loaded to some degree (Fig. 1, inset). We considered whether this was due to leakage of some BCECF AM from the electrode before formation of a seal, to diffusion of uncleaved BCECF AM from the loaded cell to adjacent cells, or to diffusion of BCECF from the loaded cell to adjacent cells through gap junctions (Dean et al. 1997; Huang et al. 1997). Although not all of these hypotheses were tested, we noted that when the pipette was brought down to a neuronal cell body for a few seconds (without patching), that cell, as well as some adjacent cells, were loaded to some degree. The fact that the pipette had a slight positive pressure as it was moved towards the cell suggests that there may be some leakage of BCECF AM before formation of a seal, which could result in dye-loading of neurones in the vicinity of the one being patched. Over a period of about 10 min, however, the fluorescence signal of neighbouring cells decreased or disappeared leaving the patched neurone the only loaded, brightly fluorescent neurone.

Some of the disadvantages encountered when developing this technique, at least in LC neurones, included changes in the electrophysiological properties of these cells. We initially loaded neurones with a high concentration of BCECF (∼100-200 μM BCECF AM in the pipette) and used full excitation light (no neutral density filters) and found that neurones loaded well and were easy to visualize. However, within a few minutes of acquiring a seal, we started noticing changes in the electrophysiological properties of the neurones, characterized by a switch in the neuronal firing pattern. The normal firing pattern of LC neurones involved repeated spontaneous action potentials with an irregular firing pattern, including occasional doublets (e.g. Fig. 2). In light-damaged neurones, the firing pattern switched progressively and irreversibly to one involving doublets, triplets, quadruplets or more, and finally to multiple action potentials that lacked a complete repolarization phase (sustained depolarization) followed by a long interspike interval (data not shown). Our first approach to this problem was to reduce the dye concentration in the loading pipette (30-50 μM), assuming that high dye concentration was altering the electrophysiological properties of LC neurones. As a result, there was less dye-loading but no alteration in the electrophysiological properties of the cells. Even though reducing the dye concentration prevented changes in firing rate patterns it also resulted in a low value of the Fl₅₀₀ and Fl₄₄₀ signals, which limited the duration of the experiments. Low dye-loading also made calibration at the end of the experiments difficult to achieve. Thus, a second approach was taken. The dye concentration in the patch pipette was once again increased to 100-150 μM and the excitation light intensity was reduced (using neutral density filters of 4-16). As a result, within ∼5 min of acquiring a seal we got a stable and sufficiently intense fluorescence reading with no alteration in the electrophysiological properties of the neurone, indicating that the BCECF dye appears to be damaging at high levels of excitation light.

Figure 2. Effects of hypercapnic acidosis (HA) on pH_i, integrated firing rate (IFR) and V_m of a LC neurone.

A, pH_i response to HA. HA exposure resulted in a rapid and sustained intracellular acidification and a return towards the initial value upon return to normal aCSF. B, IFR response to HA. IFR increased in response to HA and then returned towards the initial value upon return to normal aCSF. C, a brief sequence of truncated action potentials taken several minutes into the exposure to normal aCSF (left), to HA (middle) and to normal aCSF again (right).

With this new approach, we were able to load sufficient dye to obtain calibration values. A linear calibration curve of R_Flvs. pH_i was obtained. This curve was converted to the final calibration curve by normalizing R_Fl values (N_Fl) at pH 6.7 and 7.7 to the R_Fl value at pH 7.2. This resulted in a linear calibration curve of N_Flvs. pH (pH_i = 1.70 × N_Fl + 5.47, r = 0.99, n = 12). In all further experiments, unknown R_Fl values were converted to N_Fl by dividing them by a calibration R_Fl at pH 7.2 obtained at the end of each experiment. These N_Fl values were then converted to pH_i using the above calibration equation.

Properties of LC neurones

The mean control values for LC neurones after establishing a seal under normal conditions (5 % CO₂) were V_m, −48.6 ± 0.5 mV; firing rate, 1.95 ± 0.12 Hz (n = 72); and pH_i, 7.20 ± 0.01 (n = 44). These values are comparable to previously reported values for LC neurones (Williams et al. 1984; Wang & Aghajanian, 1987; Alreja & Aghajanian, 1995; Oyamada et al. 1999; Dean et al. 2001).

Effect of hypercapnic acidosis on firing rate and pH_i in LC neurones

We measured firing rate and pH_i responses to HA simultaneously, where CO₂ was increased (15 %) with a decrease in pH_o to ∼6.8. The pH_i response of LC neurones exposed to HA consisted of an initial fall in pH_i by 0.31 ± 0.01 pH units (n = 24) that was maintained for the duration of the HA exposure (Fig. 2A). The electrophysiological response of LC neurones to HA was characterized by an increase in firing rate of 1.38 ± 0.15 Hz from 1.81 ± 0.17 to 3.19 ± 0.25 Hz (n = 24), around a 93 % increase (Fig. 2B and C). Due to the spontaneous activity of LC neurones it was difficult to quantitatively define the changes in resting V_m, but HA exposure appeared to result in a small membrane depolarization in all 24 neurones (Fig. 2C). Our results are consistent with previously reported data in which HA resulted in an increased firing rate in LC neurones (Oyamada et al. 1999). Upon removal of the HA solution, pH_i and firing rate returned towards baseline values of 7.17 ± 0.03 and 2.02 ± 0.18 Hz, respectively.

Effects of isohydric hypercapnia on firing rate and pH_i in LC neurones

In order to evaluate the role of pH_o in CO₂-induced increased firing rate in LC neurones, we exposed them to IH, with increased CO₂ (15 %) and normal pH_o (∼7.45), and HCO₃⁻ elevated to 77 mm. The pH_i response of LC neurones exposed to IH consisted of an initial fall in pH_i by 0.17 ± 0.02 pH units (n = 12) that was followed by a pH_i recovery of 0.0123 ± 0.0024 pH units s⁻¹ (Fig. 3A). Upon removal of the IH solution, pH_i overshot its initial value of 7.20 ± 0.01 to an alkaline mean value of 7.28 ± 0.03 (n = 12; Fig. 3A), consistent with pH_i recovery during IH exposure as seen in other chemosensitive areas (Ritucci et al. 1997). The electrophysiological response of LC neurones to IH was characterized by a significant increase in firing rate of 1.10 ± 0.15 Hz from 1.87 ± 0.29 to 2.97 ± 0.39 Hz (n = 12), around a 76 % increase (Fig. 3B and C). Again, resting V_m was difficult to quantify but was not depolarized and indeed appeared to be hyperpolarized in response to IH in all 12 neurones (Fig. 3C). Upon removal of the IH solution, pH_i and firing rate returned to near baseline values of 7.21 ± 0.02 and 2.18 ± 0.31 Hz, respectively.

Figure 3. Effects of isohydric hypercapnia (IH) on pH_i, integrated firing rate and V_m of a LC neurone.

A, pH_i response to IH. IH exposure resulted in a rapid fall in pH_i followed by pH_i recovery and overshoot upon removal of IH. B, IFR response to IH. IFR increased in response to IH and then returned towards the initial value upon return to normal aCSF. C, a brief sequence of truncated action potentials taken several minutes into the exposure to normal aCSF (left), to IH (middle) and to normal aCSF again (right).

A subset of the above cells (n = 8) was exposed to both IH and HA, and pH_i and firing rate were measured. The pattern of response in these neurones was similar to that reported above. The decrease in pH_i was twice as large when induced by HA (0.32 ± 0.02 pH units) as compared to IH (0.16 ± 0.02 pH units). The pH_i fall in IH was followed by pH_i recovery of 0.0132 ± 0.0028 pH units s⁻¹. No pH_i recovery was seen with HA. The increase in firing rate was also larger in response to HA (1.55 ± 0.37 Hz, going from 1.78 ± 0.38 to 3.32 ± 0.52 Hz) than in response to IH (1.04 ± 0.22 Hz, going from 1.95 ± 0.44 to 2.99 ± 0.59 Hz), or an increase of 121 ± 42 % and 84 ± 28 %, respectively. These data suggest that under conditions of unchanged pH_o there was a decrease in both the degree of acidification as well as in the firing rate response to acid challenges, suggesting that pH_o might modulate the response either by altering the magnitude and duration of the pH_i change or by affecting membrane potential changes through alterations of various membrane conductances.

Effects of hypercapnic acidosis and isohydric hypercapnia in LC neurones in the presence of TTX

To try to clearly visualize possible changes in V_m during exposure to hypercapnia, action potentials were inhibited with the fast Na⁺ channel blocker tetrodotoxin (TTX). We first verified that the patched LC neurone was CO₂/H⁺ sensitive by showing an increased firing rate in response to HA (data not shown). In the presence of 1 μM TTX, all neurones showed a loss of action potentials. The pH_i response to HA in the presence of TTX was similar to the response in the absence of TTX (data not shown). Six out of the eight neurones exposed to 1 μM TTX expressed subthreshold membrane oscillations (SROs), characteristic of LC neurones from neonatal rats (Williams & Marshall, 1987; Travagli et al. 1995; Oyamada et al. 1998). In the presence of 1 μM TTX, exposure to HA (in neurones that had SROs) resulted in a membrane depolarization and the appearance of small spike activity (n = 6; Fig. 4A). The other two neurones that lacked SROs did not exhibit spiking activity in response to HA but did exhibit a HA-induced membrane depolarization that reversed upon return to the normocapnic solution (Fig. 4B).

Figure 4. Effects of hypercapnic acidosis and isohydric hypercapnia on V_m and pH_i of a LC neurone in the presence of 1 μM TTX.

A, V_m response to HA in a LC neurone that expressed subthreshold membrane oscillations (SROs). Exposure to HA in the presence of 1 μM TTX resulted in a small reversible depolarization and the appearance of spikes. B, V_m response to HA in a LC neurone that lacked SROs. Exposure to HA in the presence of 1 μM TTX resulted in a slight, reversible membrane depolarization but no spikes. C, V_m response to IH in a LC neurone that expressed SROs. Exposure to IH resulted in an apparent small hyperpolarization and the appearance of spikes in the presence of 1 μM TTX. D, overlapped action potentials from a HA trace before (thin trace) and during (thick trace) TTX exposure. Note the expanded time scale.

We also exposed LC neurones to IH in the presence of 1 μM TTX. The pH_i response to IH in the presence of TTX was also similar to the response in the absence of TTX (data not shown). Within 2 min of exposure to IH, LC neurones appeared to hyperpolarize and showed small spike activity (Fig. 4C). Upon removal of the IH solution, V_m returned to normal (n = 6). These data support the observation that HA leads to membrane depolarization while IH does not (i.e. Fig. 3C). IH may in fact result in a small membrane hyperpolarization.

We suspect that the spikes observed in the presence of TTX under HA and IH represent the activity of Ca²⁺ channels, known to be expressed in LC neurones (Horvath et al. 1999; Ivanov & Aston-Jones, 2000). We have no direct evidence that these are calcium spikes or what the relative magnitudes of the currents are in HA vs. IH, but the fact that the spikes seen in the presence of TTX have a higher threshold, a slower depolarizing upswing, and a slower repolarization (Fig. 4D) is consistent with these spikes being due to Ca²⁺ channel activity.

Effects of acidified Hepes-buffered solution on firing rate and pH_i in LC neurones

In order to test whether molecular CO₂ is necessary for increased firing rate, LC neurones were exposed to AcHepes solution (equilibrated with 100 % O₂, nominally CO₂ free) with an extracellular pH similar to that of the hypercapnic acidotic solution (pH_o ∼6.8). Brain slices were placed in the recording chamber and perfused with Hepes-buffered aCSF (pH_o 7.45) for 15 min to allow them to adjust. Control V_m (-48.7 ± 1.3 mV) and pH_i values (7.22 ± 0.01; n = 5) were obtained and were comparable to values for neurones in normal CO₂-containing aCSF. Neurones were then exposed to Hepes-buffered aCSF with a pH_o of ∼6.8 (similar to the pH_o in HA). This resulted in an acidification of 0.43 ± 0.03 pH units (n = 5; Fig. 5A). LC neurones exposed to AcHepes solution responded with a significant increase in firing rate of 2.03 ± 0.32 Hz from 1.87 ± 0.28 to 3.90 ± 0.11 Hz (n = 5), about a 126 % increase (Fig. 5B and C). Membrane potential appeared to depolarize in response to the AcHepes solution in all five neurones (Fig 5C). Removal of the AcHepes solution resulted in only a partial return of pH_i (6.88 ± 0.03) and firing rate (2.50 ± 0.17 Hz) towards their initial values (Fig. 5). It is not clear why pH_i exhibits only a partial return towards initial values after exposure to an AcHepes solution but such a pattern has previously been seen in other cells (Putnam & Grubbs, 1990). It is clear from these data that LC neurone firing rate can be increased under conditions of marked decreases in pH_i and pH_o even in the nominal absence of CO₂.

Figure 5. Effects of acidified Hepes-buffered aCSF on pH_i, integrated firing rate and V_m of a LC neurone.

A, pH_i response to acidified Hepes (AcHepes). AcHepes exposure resulted in a continuous fall in pH_i for the duration of the pulse followed by only a partial return of pH_i towards its initial value. B, IFR response to AcHepes. AcHepes resulted in a large increase in IFR that only partially returned towards its initial value upon return to aCSF. C, a brief sequence of truncated action potentials taken several minutes into the exposure to the various conditions.

Effects of propionate on firing rate and pH_i in LC neurones

To evaluate whether intracellular acidification (at constant pH_o and CO₂) could result in an increased firing rate, LC neurones were exposed to aCSF containing 30 mm NH₄Cl. NH₄Cl exposure induced a large (∼20 mV) depolarization, which resulted in a depolarization block and cessation of firing (data not shown), which precluded its use for studies of the relationship between changes in pH_i and firing rate. We therefore used aCSF (normal pH_o ∼7.45 and 5 % CO₂) containing 50 mm propionate (isosmotically replacing 50 mm NaCl). The pH_i response of LC neurones exposed to propionate consisted of an initial fall in pH_i by 0.18 ± 0.03 pH units followed by pH_i recovery of 0.0118 ± 0.0051 pH units s⁻¹ (n = 6) during the maintained presence of propionate (Fig. 6A). Upon removal of the propionate solution, pH_i overshot the initial mean value of 7.20 ± 0.20 to an alkaline mean value of 7.25 ± 0.04 (n = 6). The electrophysiological response of LC neurones to 50 mm propionate was characterized by no significant change in firing rate (0.21 ± 0.22 Hz, n = 6; Fig. 6B and C). Exposure to propionate appeared to be accompanied by a small membrane hyperpolarization in all six neurones (Fig. 6C). Interestingly, upon removal of propionate (re-perfusion with normal aCSF), we often saw an increase in firing rate (Fig. 6B and C). We do not know the basis for this effect but it suggests that propionate may be affecting neuronal electrical properties in ways not related to its ability to alter pH_i.

Figure 6. Effects of 50 mm propionate on pH_i, integrated firing rate and V_m of a LC neurone.

A, pH_i response to propionate and HA. Exposure to 50 mm propionate resulted in a rapid fall in pH_i followed by pH_i recovery. Exposure of the same neurone to HA resulted in a rapid fall in pH_i of greater magnitude, which was maintained during the hypercapnic pulse. B, IFR response to propionate and HA. IFR during 50 mm propionate exposure was unchanged. Note, however, the slight increase in IFR upon removal of the propionate solution. Exposure of the same neurone to HA resulted in a significant increase in IFR. C, a brief sequence of truncated action potentials taken several minutes into the exposure to the various conditions.

The goal of this experiment was to determine whether pH_i acidification by itself could stimulate LC neurones to increase their firing rate. The small transient intracellular acidification induced by propionate resulted in no significant change in the firing rate of LC neurones. This lack of increase in firing rate was not due to these cells lacking a chemosensitive response since they showed a robust increase in firing rate in response to HA (Fig. 6).

Effects of isocapnic acidosis on pH_i and firing rate in LC neurones

In order to evaluate whether extracellular acidification could increase firing rate in LC neurones, we exposed them to isocapnic acidosis (IA), with normal CO₂ (5 %) and HCO₃⁻ lowered to 7 mm (pH_o ∼6.8). The aim of this experiment was to determine, upon exposure to rapid extracellular acidification in a way that should induce a slower intracellular acidification, whether LC neurones increased their firing rate rapidly or at a slower rate. The relationship between the rate of intracellular acidification and the rate of change in firing rate observed under IA vs. HA could yield valuable insights into the potential roles of changes in pH_i and pH_o as signals for increased firing rate in LC neurones. The rate of fall in pH_i (pH units min⁻¹) and the rate of increase in firing rate (impulses s⁻¹ min⁻¹) were obtained for IA and HA from the initial 4 min following a change in the solution from control to IA or HA, respectively. The pH_i response of LC neurones exposed to IA consisted of an initial fall in pH_i at a rate of 0.025 ± 0.005 pH units min⁻¹ (r = 0.88 ± 0.08) to a maximum acidification of 0.26 ± 0.03 pH units (n = 5; Fig. 7A). Upon removal of the IA solution, pH_i returned to its initial value of ∼7.18. The electrophysiological response of LC neurones to IA was characterized by a significant increase in firing rate of 1.00 ± 0.20 Hz from 1.55 ± 0.19 to 2.55 ± 0.20 Hz (n = 5), a 72 % increase (Fig. 7B and C). The initial increase in firing rate occurred at a rate of 0.12 ± 0.047 impulses s⁻¹ min⁻¹ (r = 0.84 ± 0.10). Following IA all neurones were exposed to HA to test whether the patched LC neurone was ‘chemosensitive’ and also to compare the initial rate of the pH_i and firing rate responses to IA vs. HA. The pH_i response of LC neurones exposed to HA consisted of an initial rate of fall in pH_i of 0.061 ± 0.0047 pH units min⁻¹ (r = 0.93 ± 0.02), significantly (P < 0.001) faster than the rate of acidification seen with IA. The maximum acidification with HA was 0.24 ± 0.02 pH units (n = 5) and was maintained for the duration of the HA exposure (Fig. 7A). Exposure of LC neurones to HA resulted in a firing rate increase of 1.31 ± 0.20 Hz from 1.74 ± 0.26 to 3.05 ± 0.35 Hz (n = 5), an 83 % increase (Fig. 7B and C). The initial increase in firing rate occurred at a rate of 0.28 ± 0.040 impulses s⁻¹ min⁻¹ (r = 0.89 ± 0.10), also significantly (P < 0.05) faster than the rate during IA exposure. These data show that both the rate of fall in pH_i and the rate of rise in firing rate were faster in LC neurones in response to HA than in response to IA.

Figure 7. Effects of isocapnic acidosis (IA) on pH_i, integrated firing rate and V_m of a LC neurone.

A, pH_i effects of IA and HA. Note that exposure to IA caused a slow fall in pH_i, whereas exposure of the same neurone to HA resulted in a rapid and sustained intracellular acidification. B, IFR effects of IA and HA. The slow increase in IFR with IA correlated with the slow fall in pH_i, whereas the rapid increase in IFR with HA correlated with the rapid pH_i fall under this condition. C, a brief sequence of truncated action potentials taken several minutes into the exposure to the various conditions.

Exposure of LC neurones to IA (rapid decrease in pH_o, slow decrease in pH_i) does not result in as rapid an increase in firing rate as that seen with HA (rapid decrease in both pH_o and pH_i). Thus our data suggest that the slower rate of increase in firing rate correlates better with the slower rate of intracellular acidification, suggesting that pH_i (and not pH_o) is the primary signal in the chemosensitive response of LC neurones.

Kinetics of changes in pH_o, pH_i and firing rate in response to various conditions of acid challenge in LC neurones

If changes in pH_i serve as a signal for the increased firing rate in LC neurones, we would expect that the changes in pH_i should precede, or at least occur simultaneously with, the changes in firing rate. By simultaneously measuring changes in pH_i and firing rate in the same neurone, we can compare the kinetics of these changes. It is evident from such an analysis that the fall in pH_i occurs with a similar time course to the rise of firing rate in four of the acid challenges tested (i.e. HA, IH, AcHepes and IA; Fig. 8A-D). We were not able to do this analysis for the propionate exposure since there was no significant increase in firing rate. We compared the rate of change in pH_i and firing rate in two ways. Firstly, we ran Dunnett tests on each curve to determine the first time point at which a parameter became significantly different from control for each type of acid challenge. In all cases, the time at which the parameter first changed significantly was either earlier for pH_i than for firing rate (AcHepes: pH_i ∼2 min, firing rate ∼3 min; IA: pH_i ∼4 min, firing rate ∼5 min) or the changes occurred simultaneously (HA: pH_i and firing rate ∼2 min; IH: pH_i and firing rate ∼2 min). Secondly, we determined t₅₀ values for each neurone, by interpolating the time at which the parameter changed by 50 % of its maximum. For all conditions, the average t₅₀ value for the fall in pH_i was statistically indistinguishable from that for the rise in firing rate (Table 1), indicating that the fall in pH_i occurred simultaneously with the rise in firing rate. Comparing across conditions, the fall in pH_i occurred significantly more rapidly for IH (t₅₀ ∼2 min), and as expected, the change in pH_i and firing rate occurred significantly more slowly for IA (t₅₀ ∼5. 5 min), compared to HA and AcHepes (t₅₀ ∼3-3.5 min) (Table 1). The change in firing rate was not significantly different for IH vs. HA, suggesting that factors in addition to decreased pH_i may be involved in determining the firing rate in response to IH. Interestingly, the changes in pH_i and firing rate upon return to initial conditions were no longer linked. Thus, firing rate appeared to return towards initial values more rapidly than pH_i upon removal of acid challenge with HA, AcHepes and IA, while pH_i returned towards initial values more rapidly than firing rate upon removal of IH (Fig. 8).

Figure 8. Kinetic representation of the percentage changes in pH_i and integrated firing rate in the presence of different acidotic conditions.

A-D, percentage change in pH_i and IFR over time in the presence of HA, IH, acidified-Hepes and IA, respectively. The exposure began at time t = 0 and lasted for the period indicated by the filled horizontal bar. Each value represents the mean ± 1 s.e.m. (n = 24 for HA, 12 for IH and 5 for AcHepes and IA). E, absolute pH_o changes in the presence of HA, IH, AcHepes, propionate and IA over time (n = 4 for each condition). The exposure period is indicated by the filled horizontal bar.

Table 1.

The time for a change of 50 % of maximum (t₅₀; in min) in pH_o, pH_i and firing rate in response to different acid challenges in LC neurones

	HA	IH	IA	AcHepes
pHo	1.05 ± 0.13 (4) *	—	1.61 ± 0.14 (4)*	1.73 ± 0.04 (4) *
pHi	3.01 ± 0.23 (24)	1.80 ± 0.10 (12)†	5.79 ± 0.41 (5)‡	3.87 ± 0.23 (5)
Firing rate	2.63 ± 0.24 (24)	2.26 ± 0.46 (12)	5.17 ± 0.38 (5)‡	3.38 ± 0.18 (5)

Values are presented as means ± 1 s.e.m. (number of neurones). HA, hypercapnic acidosis; IH, isohydric hypercapnia; IA, isocapnic acidosis; AcHepes, acidified Hepes; pH_o, extracellular pH; pH_i, intracellular pH.

^*t₅₀ value for the change in pH_o is significantly faster than that for changes in pH_i and firing rate (P < 0.05).

^†t₅₀ value for change in pH_i in response to IH is significantly faster than that for the change in pH_i in response to the other conditions (P < 0.01).

^‡t₅₀ values for the changes in pH_i and firing rate in response to IA are significantly slower than the values for the other conditions (P < 0.05).

The changes in pH_i and firing rate were always much slower than the rate of change in pH_o. We verified that pH_o did not change significantly upon switching to IH or propionate-containing solution (Fig. 8E). The change in pH_o with HA, IA and AcHepes was rapid (Fig. 8E), with t₅₀ values of ∼1-1.75 min (Table 1). Thus, both pH_i and firing rate changed more slowly in response to various acid challenges than did pH_o.

Relationship between changes in CO₂, pH_o, pH_i and firing rate

It is clear from our data that firing rate could increase without the presence of CO₂ (AcHepes) or without an increase in CO₂ (IA). It is also clear that changes in pH_o were poorly correlated with the increase in firing rate. Firing rate increased without any change in pH_o (IH) and increased to different extents even though pH_o changed by the same amount (HA, IA and AcHepes). To examine the relationship between pH_i and firing rate, we correlated the magnitude of the change in pH_i with the magnitude of the change in firing rate induced by various acid challenges. Since pH_i and firing rate appear to change at the same rate for each condition (Fig. 8), we did a correlation of the magnitude of change in pH_ivs. that in firing rate at 5 min after solution change for all acid challenges. At this time, the changes were maximal for IH, nearly maximal for HA and AcHepes, and less than 50 % of maximum for IA. The effects of propionate, which did not result in a significant increase in firing rate, were also determined at 5 min. At this time, the increase in firing rate was significantly (P < 0.05) correlated with the change in pH_i (Fig. 9A). We also analysed each set of data for the same degree of change. To do this, we determined the first time point at which pH_i had decreased by at least 80 % of its maximum for each condition and determined the corresponding change in firing rate at that time. These data also yielded a significant (P < 0.05) correlation between the magnitude of the change in pH_i and the change in firing rate (Fig. 9B). Thus, it appears that upon exposure to various acid challenges, the magnitude of the increase in firing rate is positively correlated with the magnitude of the fall in pH_i.

Figure 9. Relationship between the magnitudes of the changes in firing rate and pH_i of LC neurones induced by different acid challenges.

Correlation between the magnitude of the change in pH_i induced by various acid challenges and the resulting magnitude of the change in firing rate. A, the magnitude of all changes was measured at 5 min after solution change. The equation and r for the fitted line are given on the figure. B, the magnitude of the change in pH_i was determined at the time that it first exceeded 80 % of its maximum change for each condition and the magnitude of the change in firing rate was determined at the same time. That time was 5 min for HA and for IH, 4 min for propionate, and 8 min for AcHepes and for IA. The equation and r for the fitted line are given on the figure. Lines were fitted by a least-squares linear regression to the points. Each point represents the mean ± 1 s.e.m. of between 5 and 24 individual values. The various points are labelled: HA, hypercapnic acidosis; IH, isohydric hypercapnia; AcHepes, acidified-Hepes; propionate; and IA, isocapnic acidosis.

DISCUSSION

Two major findings are presented in this study: (1) a new technique for the use of perforated patch pipettes to load fluorescent dyes in order to simultaneously measure V_m and pH_i; and (2) the marked correlation of pH_i changes with changes in firing rate in LC neurones in response to various acid challenges. Our data imply that a change in pH_i is a major component of activation of LC neurones, but they also suggest the involvement of signals other than pH_i in determining the ultimate firing rate in response to chemosensitive stimuli.

New technique for simultaneous measurements of pH_i and V_m

We used a novel technique to simultaneously measure pH_i and V_m responses to various acid challenges. Due to the observation of washout of the chemosensitive response with the use of whole-cell recordings (Dean & Reddy, 1995; Richerson, 1995; Ritucci et al. 1999) we developed a technique that allows us to simultaneously measure pH_i and firing rate using the perforated patch configuration (which eliminates washout) with fluorescence imaging microscopy.

One of the major advantages of our technique is that a single neurone is loaded with dye and thus background fluorescence is significantly reduced. Loading of the neuronal cell body is dependent on dye leakage from the pipette onto the cell surface (before acquiring a seal) as well as dye diffusion from the pipette across the cell membrane into the cytosol (once a seal is established). The degree of loading depends on the concentration of the dye in the pipette and the access resistance. The dye concentration in the soma is a function of the rate of continued dye-loading from the pipette, photobleaching and diffusion of the dye into neuronal cell processes and the extracellular space. Cell viability, measured as the rate of decrease in the emitted fluorescence with excitation light of 440 nm (Bevensee et al. 1995), could not be used with our technique due to continuous loading of the cell with dye and to the diffusion of the dye from the soma into neuronal processes. For our study, cell viability was determined based on the observation of normal values of pH_i, V_m and firing rate in the patched, fluorescent neurone. We note that our dye-loaded cells gave similar electrophysiological responses to non-loaded LC neurones (Pineda & Aghajanian, 1997; Oyamada et al. 1998, 1999) indicating no apparent BCECF damage to the neurones at the concentrations employed, provided illumination intensities were restricted to relatively low levels. Cell viability was also determined from the normal pH_i responses seen upon exposure of the neurone to different acid/base disturbances.

Our technique should be useful for loading membrane-permeant fluorescent dyes into individual cells when the perforated patch configuration must be used to prevent washout of key cytosolic factors (Dean & Reddy, 1995; Richerson, 1995) or the rundown of ion channels (Becq, 1996). Some adjustments to the technique may be needed in other preparations. Although we do not know the basis for changes in the electrophysiological properties of LC neurones, we obtained excellent fluorescence images without disturbing the neuronal electrophysiological properties using high dye concentrations in the pipette in combination with low excitation light intensities. In other preparations, proper loading may require altering the dye concentration within the pipette or using pipettes with different access resistances. Loading of a cell may be more rapid and complete in a preparation where the surface of the cell has been ‘cleaned’ of overlying tissue (Gray & Johnston, 1985; Edwards et al. 1989), thereby reducing the series resistance, or in cultured cells, requiring lower dye concentrations in the patch pipette. While our technique should be readily applicable for the use of pH-sensitive and Ca²⁺-sensitive fluorescent dyes, the size and permeability of a given membrane-permeant dye might be a limiting factor for the use of this technique. Finally, our technique might prove to be a useful way to load pH-sensitive dyes into neuronal processes (e.g. Fig. 1) and thereby study the regulation of pH_i in these processes.

Stimulus for increased firing rate

The general model for chemosensitivity assumes that hypercapnia induces a fall in pH_i which inhibits a K⁺ channel, resulting in membrane depolarization and a subsequent increase in firing rate (Dean et al. 1989; Wang et al. 1998; Wellner-Kienitz et al. 1998; Oyamada et al. 1999). Our data clearly suggest a more complex interaction between the different signals (pH_i and pH_o) present during the stimulus and their effect on the different ion channels expressed in LC neurones.

In this study, we present evidence that pH_i is a major signal in the chemosensitive response of LC neurones. This conclusion is supported by the strong correlation between the magnitude of the pH_i change and that of the increase in firing rate (Fig. 9). That an increase in CO₂per se is not required for increased firing rate of LC neurones is demonstrated by the increase in firing rate in response to AcHepes or in response to IA where CO₂ is either absent or unchanged, respectively. Decreased pH_o is also unlikely to be the predominant signal for increased LC neurone firing rate for two reasons. Firstly, an increase in firing rate was seen in the absence of any change in pH_o (IH). Secondly, the pH_o change was the same with HA, IA and AcHepes, and yet each of these conditions elicited a different firing rate response in LC neurones. Thus, the factor that best correlates with increased firing rate is the change in pH_i, and not the change in pH_o or CO₂.

The role of pH_i as the major signal in the chemosensitive response of LC neurones is further supported by the kinetic analysis shown in Fig. 8. In all cases, the rate of fall in pH_i occurred simultaneously with or preceded the rate of rise in firing rate. This is especially clear when comparing the chemosensitive response of IA with that of HA, where pH_o decreased rapidly with both (t₅₀ 1.61 and 1.05 min, respectively), but pH_i decreased slowly with IA compared to HA (t₅₀ 5.79 vs. 3.01 min, respectively). In both cases, the rate of change in firing rate was compatible with the rate of change in pH_i and not in pH_o.

From the proposed chemosensitive model one might expect pH_i changes to precede firing rate changes. There are two possible reasons why this was not observed. Firstly, it is possible, given the time scale used in our study (tens of seconds), that pH_i-induced firing rate changes occur rapidly (within seconds). Thus, when a change in pH_i occurs, changes in parameters such as K⁺ conductances or spike threshold could possibly occur within seconds, resulting in a rapid increase in firing rate. To test this possibility we will need a determination of the kinetics of changes in pH_i with a much greater time resolution. Secondly, it is also possible that chemotransduction occurs at a site distal to the soma, such as the dendrites. In these small regions, the rate of pH_i change in response to an acid challenge would probably be more rapid than the rate of somatic pH_i changes measured in our study due to the more favourable surface to volume ratio. In contrast, a spike initiation signal established in these regions should be rapidly conveyed to the soma, resulting in an increased firing rate. For instance, a dendritic potential change measured at 500 μm from the soma required only 1 ms to reach the soma (Stuart et al. 1997). Therefore, our measurements of the time course of the change in firing rate in the soma might more accurately reflect the kinetics of possible dendritic spike initiation whereas our measurements of the time course of somatic pH_i changes would be considerably slower than dendritic pH_i changes. An evaluation of this possibility must await simultaneous measurements of V_m and pH_i in the dendrites of LC neurones, but a dendritic site of chemoreception has previously been proposed (Ballantyne & Scheid, 2000).

Evidence for other factors affecting LC neuronal firing rate

Although our data point to pH_i as the major signal in the chemosensitive response of LC neurones, other factors appear to have a role in determining firing rate. One line of evidence for additional signalling factors is that the correlation of pH_i changes with firing rate changes (Fig. 9) appears to be better under conditions in which a decrease in pH_o accompanies a decrease in pH_i (e.g. AcHepes, HA, IA). When pH_o is held constant, the response of LC neurones to intracellular acidification is more variable, with either a substantial increase in firing rate (IH) or virtually no increase in firing rate (propionate) (Fig. 9). These findings suggest that changes in pH_o may modulate the LC neurone response to decreased pH_i. pH_o modulation of the chemosensitive response of LC neurones might result from the inhibition of pH_i regulatory systems, which leads to a sustained fall in pH_i (Figs 1, 5 and 7). This is in contrast to the observed recovery from intracellular acidification seen when pH_o is held constant (Ritucci et al. 1997, 1998). Thus, changes in pH_o appear to play at least a modulatory role in chemosensitive signalling in LC neurones.

The involvement of signalling components other than pH_i is also suggested by the firing rate response to propionate. A small, but distinct, increase in firing rate is seen in LC neurones upon removal of propionate, and this increase occurs at a time when pH_i is somewhat alkaline compared to normal (Fig. 6). We do not know the signalling mechanism involved in this response but the recent demonstration that propionate removal elevated intracellular Ca²⁺ in snail neurones (Willoughby et al. 2001) suggests that Ca²⁺ may play a role.

The poor correlation between the rates of return of pH_i and firing rate towards initial values upon removal of acid challenge (Fig. 8) further suggests that the change in pH_i is not the only factor determining firing rate. For instance, upon the removal of IH, pH_i exhibits an alkaline overshoot of its initial value, but firing rate returns only to its initial value (Fig. 3). Although we have no direct evidence for the involvement of other factors in the chemotransduction pathway, several possibilities exist. Changes in intracellular Ca²⁺ have been proposed to play a role in chemotransduction (Williams & Marshall, 1987). Our findings that HA and IH (in the presence of TTX) induce spike activity, possibly due to opening of Ca²⁺ channels, support a potential role for Ca²⁺ in the chemosensitive response (Fig. 4). Another potential secondary signal is the release of neurotransmitters (Spyer & Thomas, 2000). This is of especial interest with respect to the off response to IH and propionate, since neurotransmitter release has been shown to be affected by the alkaline rebound after pH_i recovery from acidification (Trudeau et al. 1999). Polyamines (Pineda & Aghajanian, 1997), carbonic anhydrase (Vovk et al. 2000) and cell volume changes could also play a role in the overall chemosensitve response of LC neurones.

Finally, we propose that, contrary to the original chemosensitive model, acidification and changes in pH_i do not just affect a single target (e.g. a specific type of K⁺ channel) but rather affect multiple targets. We found that when pH_o decreased with pH_i (e.g. HA, IA or AcHepes), firing rate increased and V_m appeared to depolarize (Figs 2, 5 and 7). However, when pH_o was held constant (e.g. IH) we observed that decreased pH_i resulted in increased firing rate with no apparent change in resting V_m and possibly a V_m hyperpolarization (Fig. 3). The finding of a possible small hyperpolarization with an increased firing rate is unusual. We suggest that such a result could arise if ion channels responsible for determining membrane potential were different from those responsible for initiating and setting the firing rate, and if these channels were affected differently by the conditions of the various acid challenges. For instance, LC neurones are known to contain TWIK-related acid-sensitive K⁺ (TASK) channels (Talley et al. 2001), which are active at normal pH_o but are inhibited by physiological decreases in pH_o. Thus, these channels would be active in response to IH and propionate (where we see hyperpolarization) but inhibited by HA, AcHepes and IA (where we see depolarization). Along the same lines, assuming that Ca²⁺ enters LC neurones upon acid challenges, accumulation of intracellular Ca²⁺ could activate Ca²⁺-dependent K⁺ channels (Osmanovi'c & Shefner, 1993), and thus contribute to a V_m hyperpolarization. At the same time, intracellular acidification should inhibit inwardly rectifying K⁺ channels and could increase the slope of the interspike depolarization (Pineda & Aghajanian, 1997), resulting in a faster rise to threshold potential and an increase in firing rate. Thus, it is possible that by having multiple effects of changes in pH_o and pH_i, one could observe an increased firing rate as a result of an increase in the interspike depolarization in the presence of an underlying hyperpolarization. In fact, a hyperpolarization with increased firing rate has previously been observed with increased temperature in Aplysia neurones (Carpenter, 1967), and was attributed to temperature reducing neuronal resting V_m (reduced excitability) while increasing the processes involved in spike initiation. Our findings suggest the possibility of multiple targets for decreased pH_i in the presence of a chemosensitive stimulus, and that some of these targets (e.g. ion channels) may be affected by signals other than pH_i (e.g. pH_o, Ca²⁺, neurotransmitters). It is the combination of these effects that most probably leads to the ultimate firing rate and change in V_m in LC neurones in response to any given chemosensitive signal.

Relationship to studies in other chemosensitive cells

LC neurones are just one of many types of cells that respond to an increase in CO₂/H⁺. Neurones from a number of other brainstem regions respond to elevated CO₂/H⁺ with an increased firing rate. There is evidence that changes in pH_i play a major signalling role in these neurones as well. In VLM and NTS neurones, HA-induced acidification is maintained with no pH_i recovery whereas pH_i recovery is present in neurones from non-chemosensitive regions (Ritucci et al. 1997, 1998). Further, the increase in bioelectric activity of VLM neurones in organotypic cell culture is most apparent in conditions that induce a fall in pH_i (Wiemann & Bingmann, 2001). Finally, inhibitors of a major pH_i regulatory exchanger, the Na⁺-H⁺ exchanger, result in an intracellular acidosis and an increase in the bioelectric activity of VLM neurones in vitro (Wiemann et al. 1999) and in an increase in respiration when administered in vivo (Kiwull-Schöne et al. 2001). A change in pH_i as a chemosensitive signal has also been proposed for peripheral chemoreceptors, i.e. the glomus cells of the carotid body (Buckler & Vaughan-Jones, 1994), central chemosensitive neurones of the snail, Helix aspersa (Goldstein et al. 2000), and avian intra-pulmonary chemoreceptors (Hempleman et al. 2000). Finally, in acid-sensitive taste receptor cells, a change in pH_i has been proposed to be the proximal signal for sour taste reception (Lyall et al. 2001). Thus, based on these various findings and on the significant correlation between the magnitudes of the changes in pH_i and firing rate that we observed with our simultaneous measurements, it appears that a change in pH_i is likely to be a major part of a generalized signalling pathway in a wide variety of cells that are sensitive to CO₂/H⁺.

In summary, our results demonstrate that the major signal in the chemosensitive response of LC neurones is pH_i rather than pH_o or molecular CO₂. The chemosensitive response of LC neurones appears to involve other signals in addition to changes in pH_i and these various signals are likely to affect multiple targets. The ultimate firing rate response of LC neurones to any chemosensitive signal will reflect the summed response of all the induced signals and the responses of the various targets to these signals.