GFP-expressing locus ceruleus neurons from Prp57 transgenic mice exhibit CO₂/H⁺ responses in primary cell culture

Abstract

The locus ceruleus (LC) contains neurons that increase their firing rate (FR) in vitro when exposed to elevated CO₂/H⁺ and have been proposed to influence the respiratory network to make compensatory adjustments in ventilation. Prp57 transgenic mice express green fluorescent protein (GFP) in the LC and were used to isolate, culture, and target LC neurons for electrophysiological recording. We hypothesized that GFP-LC neurons would exhibit CO₂/H⁺ chemosensitivity under primary culture conditions, evidenced as a change in FR. This is the first study to quantify CO₂/H⁺ responses in LC neuron FR in cell culture. Neurons were continuously bathed with solutions containing antagonists of glutamate and GABA receptors, and the acid-base status was changed from control (5% CO₂; pH ∼7.4) to hypercapnic acidosis (9% CO₂; pH ∼7.2) and hypocapnic alkalosis (3% CO₂; pH ∼7.6). FR was quantified during perforated patch current clamp recordings. Approximately 86% of GFP-LC neurons were stimulated, and ∼14% were insensitive to changes in CO₂/H⁺. The magnitude of the response of these neurons depended on the baseline FR, ranging from 155.9 ± 6% when FR started at 2.95 ± 0.49 Hz to 381 ± 55.6% when FR started at 1.32 ± 0.31 Hz. These results demonstrate that cultured LC neurons from Prp57 transgenic mice retain functional sensing molecules necessary for CO₂/H⁺ responses. Prp57 transgenic mice will serve as a valuable model to delineate mechanisms involved in CO₂/H⁺ responsiveness in catecholaminergic neurons.

central respiratory chemoreceptors detect fluctuations in CO₂ and/or pH, alter their neuronal activity, and subsequently influence the respiratory network to appropriately adjust ventilation. Chemoreceptors were first proposed to be located in distinct cellular aggregations at or near the ventrolateral medullary surface (37, 57). However, it is now well established that there are many neurons widely distributed along the neuraxis that increase their firing rate (FR) in response to acidosis, including cells within the ventrolateral medullary surface (36, 37, 57), medullary raphé (6, 49, 51, 70, 71, 74), pre-Bötzinger complex (34, 52, 61), retrotrapezoid nucleus (RTN) (20, 27, 56), hypothalamus (15, 16, 75), nucleus tractus solitarius (NTS), and locus ceruleus (LC) (12, 18, 22, 23, 28, 33, 40, 42, 53, 55). It has been proposed that all of these groups of pH-sensitive neurons are central respiratory chemoreceptors, each one contributing to the increased respiratory motor output to ensure that the pH of the blood and cerebrospinal fluid is restricted to normal physiological limits (9, 10, 41). The relative importance of each one to the overall response, and the conditions under which each one contributes, is not known.

Considerable evidence suggests that the LC contains chemosensitive neurons that participate in the ventilatory response to CO₂/H⁺ (22, 23, 33, 35, 44–46, 54, 56, 63). This dorsal pontine region bilaterally situated at the floor of the fourth ventricle, contains the greatest concentration of catecholaminergic neurons in the central nervous system and is known to exhibit arousal state-dependent activity and play a modulatory role in the processing of sensory information, arousal, feeding behaviors, nociception, and cardiovascular control (3, 32, 45). The relative importance of LC neurons in central chemoreception is unclear. However, retrograde labeling studies have provided evidence for direct connections from the LC to phrenic motoneurons (17). Focal acidosis within the LC causes an increase in respiratory frequency and phrenic nerve discharge in cats (9). Furthermore, LC neurons have been suggested to play an essential role in the proper development of the respiratory network (68) and exhibit a modulatory action on respiratory rhythm (2, 8, 26, 31, 43, 65). Targeted elimination of the A6/A5 noradrenergic cell groups with antidopamine beta hydroxylase conjugated to saporin administered into the fourth ventricle (35) or focal deletion of LC cells with 6-hydroxydopamine (7) results in a 28 and 64% reduction of the hypercapnic ventilatory response, respectively. Haxhiu and coworkers have shown that prolonged exposure to elevated CO₂ induces c-fos expression in the LC in vivo (29). Hypercapnic acidosis (HA) leads to an increase in FR in >80% of neurons in the LC in vitro (4, 5, 45, 46). This response is due in part to a decrease in intracellular pH (22) but can also occur in response to a decrease in extracellular pH alone (28).

Most of the aforementioned studies were conducted in whole animals, brain slices, or en bloc brain stem spinal cord preparations. However, the chemosensory properties of neurons from other regions have been studied in dissociated cell cultures. Studies were conducted on respiratory neurons cultured from the upper medulla (24), and respiratory pacemaker-like cells exhibited responses to hypercapnia (52). Richerson and coworkers quantified CO₂/H⁺ chemosensitive responses of medullary raphé respiratory chemoreceptors in primary culture (71, 72, 74) and found that the FR of 73% of serotonergic neurons was increased by small elevations in CO₂/H⁺ (74). More recently, Su et al. (64) studied CO₂ chemosensitivity in cultured neurons (from the medulla and pons) using microelectrode array recordings and found that 20% of the neurons were stimulated by CO₂. Although culture conditions provide much better recording stability than is possible in slices, lead to better isolation of intrinsic responses, and allow optimal access and control over the external environment of the cells, it was necessary in these studies to use immunohistochemistry after recording to confirm the phenotype of each neuron.

Additional studies are needed to define the intrinsic properties of putative central chemoreceptive neurons to determine the mechanisms that induce changes in neuronal FR during CO₂/pH challenges. We have employed primary cell culture and patch-clamp recordings to study catecholaminergic neurons of the LC phenotypically identified by endogenous expression of green fluorescent protein (GFP). The present investigation focuses on characterization of the CO₂/H⁺ responsiveness of cultured GFP-LC neurons obtained from the Prp57 transgenic mouse (67) and assesses the feasibility of the transgenic model for future studies. A portion of the data presented in this paper was previously published in abstract form (33).

MATERIALS AND METHODS

Prp57 transgenic mice expressing GFP in the LC.

Previous reports have described the methods used to generate the Prp57 transgenic mouse strain, which exhibits strong expression of GFP in the LC (66, 67). In brief, a mouse prion promoter was used to drive a cDNA construct of an enhanced, red-shifted, mammalian codon-corrected GFP. Greater than 98% of GFP-expressing LC neurons found in the Prp57 mouse also express the enzymatic marker, tyrosine hydroxylase (TH), which is used to identify noradrenergic neurons of the LC (67). There were no significant differences found in basic electrophysiological properties between GFP-expressing LC neurons from slices of transgenic mice to non-GFP-expressing neurons from nontransgenic mice of the same strain (67). All animal procedures were conducted in accordance with experimental protocols approved by the Yale University Animal Care and Use Committee.

Primary cell culture.

Using aseptic technique, GFP-LC primary cell cultures were prepared using protocols previously described for raphé neurons by Richerson and coworkers (71–74). Briefly, Prp57 neonatal mouse pups (P0–P3) were rapidly decapitated, and a craniotomy was performed to expose the dorsal surface of the brain and spinal cord. Transverse cuts were made rostral to the midbrain-pontine border and at the level of the cervical spinal cord (C1–C3). The brain stem-spinal cord was sliced rostrocaudally with a McIlwain tissue chopper (Frederick Haer, Bowdoin, ME) in ∼500-μm increments or manually with a microknife until the level was reached containing the LC. GFP expression was initially verified in pontine slices using a fluorescence microscope (Fig. 1), and tissue samples including the region of the LC were bilaterally excised from three to four neonatal mice. There was a limited possibility that we obtained GFP-expressing cells from an area outside of the LC, but slices were cut in a manner that minimized the possibility of contamination from the closest area that also expressed GFP, the A5 medullary region. The LC tissue samples were initially stored in HEPES-buffered solution containing (in mM) 130 NaCl, 4 KCl, 1 MgCl₂, 1.5 CaCl₂, 10 HEPES, 10 dextrose, and 3 NaOH, pH 7.3. Samples were then transferred to HEPES digestion solution containing papain (100 units/10 ml) (Worthington-Biochem, Lakewood, NJ) and placed on a rotator within an incubator with 5% CO₂ and 95% air at 37°C. Subsequently, tissue was triturated in modified Eagle's medium and plated onto poly-l-ornithine-coated glass coverslips at a density of ∼50,000 cells/ml. Approximately 1 h after plating, a single coverslip was examined with a fluorescence microscope equipped with GFP filters to verify GFP expression. Cultures were maintained in glial-conditioned medium (10% fetal bovine serum in 70% modified Eagle's medium + 30% Neurobasal medium with B27 supplement + penicillin-streptomycin). Basic fibroblast growth factor (0.1–1 ng/ml) and fibroblast growth factor 5 (1–10 ng/ml) were also added to enhance neuronal survival. On days 4–7, a one-half volume change was made with Neurobasal/B27 supplemented with cytosine α-d-arabinofuranoside hydrochloride (0.3–3 μM) to reduce glial growth. Subsequent maintenance feedings were made every 1–2 wk with a one-half volume change of Neurobasal/B27. Cell cultures established under these conditions were characterized by a glial bed with neurons dispersed on its surface and were maintained for periods ≥8 wk.

Fig. 1.

Expression of green fluorescent protein (GFP) was used to target locus ceruleus (LC) neurons. A: fluorescence photomicrograph of a pontine slice from a P2 Prp57 transgenic mouse pup that expressed GFP in the LC. Fluorescently labeled neurons appear as a densely packed elongated nucleus at the level of the fourth ventricle (4V; arrowheads) lateral to the midline (scale bar, 200 μm). Inset: the region bilaterally excised for primary cell culture. B–C: the dendritic processes and cell bodies of individual neurons (arrows) are seen at high magnification (scale bars, 100 μm and 20 μm, B and C, respectively). D: neurons in culture were targeted for patch-clamp recordings using visible light Nomarski differential interference contrast (DIC) and fluorescence (scale bar, 50 μm).

Perforated patch electrophysiological recordings.

Coverslips were transferred to a recording chamber (E. W. Wright, Guilford, CT). The chamber was mounted on a fixed-stage upright microscope (Axioskop 2 FS, Zeiss) equipped with visible light Nomarski differential interference contrast and fluorescence capabilities. Candidate cells for chemosensitivity tests were identified by expression of GFP and subsequently targeted for patch-clamp recordings. Our goal was to obtain a homogeneous data set from GFP-LC neurons. The cultures did include some cells from areas adjacent to the LC that were likely to be heterogeneous; however, we specifically focused on GFP-LC neurons and did not study cells that did not express GFP. However, earlier experiments from cultured rat neurons indicated that the majority of neurons cultured from this region (most of which are not catecholaminergic) were not chemosensitive. All experiments were conducted at room temperature while extracellular pH was continuously monitored with an electrode inserted into an inflow port to the recording chamber (MI-414; Microelectrodes, Londonderry, NH).

Under control conditions, neurons were bathed with Ringer solution containing (in mM) 124 NaCl, 26 NaHCO₃, 3 KCl, 2 MgCl₂, 2 CaCl₂, 1.3 NaH₂PO₄, and 10 dextrose equilibrated with medical-grade-certified 95% O₂ and 5% CO₂ (Airgas Northeast, Cheshire, CT). To eliminate fast glutamatergic and GABAergic synaptic transmission, all bath solutions included kynurenic acid (1 mM) and picrotoxin (100 μM) during recordings (71). Electrodes (3–10 MΩ) were fabricated from borosilicate glass using a micropipette puller (Sutter P-97, Sutter Instruments) and filled with an intracellular solution containing (in mM) 135 KOH, 135 methanesulphonic acid, 10 KCl, 5 HEPES, and 1 EGTA; pH 7.2; osmolarity 275 ± 10 mosM. Perforated patch recordings were made using gramicidin (50–100 ng/ml) (69). All recordings were made in current clamp mode. Neurons were considered healthy when the resting membrane potential was less than or equal to −40 mV and action potential height was ≥60 mV. Cells were also considered healthy with maintenance of continuous spiking, either spontaneous or with depolarizing current injection under control conditions (5% CO₂, pH ∼7.4). Signals were amplified (Axopatch 1D, Axon Instruments), filtered (10-kHz low pass), and acquired at 10 kilosamples/s with a computerized data acquisition board (PCI-MIO-16E-1, National Instruments, Austin, TX) using custom-written software.

Once a stable membrane potential was achieved, non-spiking neurons were depolarized with current until a consistent baseline FR was obtained. In most cases, the level of depolarizing current was then maintained constant throughout the rest of the experiment to enable monitoring of CO₂/H⁺ induced changes in neuronal firing patterns. Neurons requiring a hyperpolarizing current to stabilize FR and membrane potential were not included in the analysis of chemosensitivity since this can be an indication of an unhealthy recording.

Solutions used to alter pH and CO₂.

FR was measured under three conditions; control (5% CO₂, pH ∼7.4), HA (9% CO₂, pH ∼7.2), and hypocapnic alkalosis (3% CO₂, pH ∼7.6). Each analyzed neuron was subjected to a minimum of four CO₂-induced pH transitions to establish the consistency of its response. Prior to any acid-base perturbation, a baseline FR was established at 5% CO₂, pH ∼7.4. In most neurons, this was followed by a change to 9% CO₂ with washout back to control, followed subsequently by exposure to 3% CO₂ and return to 5% CO₂. For stable recordings, multiple changes were then made between 5 and 9% CO₂. In other cases, the baseline FR was established at 5%, and then cells were exposed to 3% CO₂ followed by return to 5% CO₂, and then multiple changes between 5 and 9% CO₂. Based on these protocols, neurons were classified either as CO₂/H⁺ stimulated or insensitive. We did not encounter any neurons inhibited by CO₂/H⁺.

Data analysis and quantification of response to CO₂/H⁺.

GFP-LC neurons were evaluated to assess their chemosensitive properties based on the criteria previously used to define responsiveness of medullary raphé neurons in culture (49, 71–74). We defined GFP-LC neurons as chemosensitive if they met the following criteria: 1) reproducible responses to a minimum of four pH transitions; 2) consistency in time course for each pH response; 3) opposite responses to acidosis and alkalosis if exposed to both; and 4) a statistically significant increase or decrease in mean FR of at least 20% for a 0.2-unit change in pH (P ≤ 0.05, Student's t-test). Cells that exhibited consistent increases in FR during acidosis and reductions in FR during alkalosis were classified as acidosis-stimulated neurons. GFP-LC cells that did not consistently respond to CO₂/H⁺ challenges were considered CO₂/H⁺ insensitive. Our approach was to establish whether a neuron had a reproducible response and to avoid false classification due to random changes in neuronal FR that might be mistaken for chemosensitivity. However, in general, if a neuron responded once it would respond every time it was exposed to acidosis.

Neuronal responses were quantified from data obtained with a software-based spike threshold detection algorithm that generated FR plots using a moving average (bin size, 1–2 s, 20 bins per point). The steady-state mean FR and pH were calculated from the last 60 s of each epoch at each pH level as previously described (74). The mean FR during steady state was determined for each neuron under a given test condition and compared with the control period immediately before and after the pH change to reduce the effect of FR variability. Values for percent change in FR were calculated by dividing the steady-state FR during challenge by the control FR. For example, a neuron's FR would be 150% of control when the FR was 3 Hz during acidosis and 2 Hz during the control period. We also calculated the chemosensitivity index (CI) for each neuron, as previously described (71). Neurons exhibited FRs between 1 and 3 Hz under control conditions. We did, in some cases, encounter neurons that had a baseline FR that was <0.2 Hz. This would lead to a deceptively large percent change in FR. For those neurons, we substituted a baseline FR of 0.2 Hz to produce a more conservative estimate of the percentage change in FR, as was previously done for neurons of the medullary raphé (71). Data are presented means ± SE, unless otherwise noted. A Student's t-test (two-tailed with unequal variance; Microsoft Excel) was used to determine whether means were significantly different (P < 0.05, unless otherwise noted). NS indicates that results were not statistically different.

TH immunohistochemistry.

We performed immunofluorescence experiments to verify that GFP-LC neurons produced TH and to characterize their morphological features. Coverslips were fixed with 4% paraformaldehyde for 1 h, rinsed with phosphate-buffered saline (PBS; pH ∼7.35), and stored in cryoprotectant solution (1% polyvinylpyrrolidone, 30% sucrose, 30% ethylene glycol in 0.05 M phosphate buffer) until processed. Coverslips were washed with 0.1 M PBS three times for 10 min each. Nonspecific binding was reduced by incubating cultures in blocking solution containing 5% normal goat serum (NGS) in 0.1 M PBS + 0.3% Triton-X for 30 min. Coverslips were incubated (18–24 h at 4°C) with a rabbit polyclonal antiserum against TH (1:2,000, Chemicon) diluted in PBS with 0.3% Triton-X and 2% NGS and then washed with 0.1 M PBS for 30 min. Next, coverslips were exposed to goat anti-rabbit rhodamine conjugated secondary antibody (1:200) at room temperature for 2 h to produce red fluorescent labeling of TH-positive neurons. For negative controls, primary antibody was excluded from the protocol.

Coverslips were mounted onto glass slides with Vectashield mounting medium (Vector Laboratories) and observed with a fluorescence microscope using filters for GFP and rhodamine (Olympus Provis). We identified and counted GFP-LC neurons from two litters of Prp57 mice cultured on different dates and determined how many of these GFP-LC neurons also expressed TH.

RESULTS

Endogenous GFP and TH immunofluorescence in cultured LC neurons.

Unlike previous studies that required the use of immunostaining to determine the phenotype of neurons after recording the response to changes in CO₂/H⁺, the expression of GFP by catecholaminergic LC neurons allowed us to target a specific cell type before electrophysiological recording. To validate this approach, we examined the expression of GFP in neurons cultured from the LC of Prp57 mouse pups and determined that GFP-LC neurons coexpressed TH (Fig. 2, A–C). Total cell counts of 337 GFP-expressing neurons from two independent culture dates revealed that 98.89% (333 of 337) were immunoreactive for TH (Fig. 2D).

Fig. 2.

Morphology of LC neurons in culture. A: GFP-LC neurons were immunoreactive (IR) for TH (scale bar, 20 μm). B: the cell bodies, nuclei, and dendritic processes of GFP-LC neurons were clearly discernable (scale bars, 20 μm). Left two images are TH immunofluorescence, and right two images are GFP fluorescence. C: this neuron was identified by visualizing expression of GFP (inset; scale bar, 20 μm), and TH immunohistochemistry allowed visualization of the dendritic field (scale bar, 100 μm). Although multiple processes were TH-IR, it is possible that some of the dendrites emanated from nearby overlapping cells. D: most LC neurons were found to coexpress both GFP and TH. Bar graph shows the number of cells that expressed GFP (green), were TH immunofluorescent (red), and were both GFP+ and TH-IR (yellow).

Neurons that coexpressed GFP and TH immunoreactivity were most commonly characterized by large cell bodies with multiple dendrites. The somata of these neurons ranged from ∼20 to 30 μm in diameter with a prominent nucleus. TH immunoreactivity revealed elaborate labeling of neuronal processes that was better visualized than with GFP fluorescence (Fig. 2B). Some of the processes had projections with en passant swellings that were detectable.

Effect of pH changes on GFP-LC neuron FR.

We quantified responses to changes in CO₂/H⁺ from LC neurons cultured for 12–56 days (32.9 ± 12 days; mean ± SD). This corresponds to previous studies conducted on raphé neurons where chemosensory responses were shown to develop in culture and in brain slices at ∼12 days (72), and other neurons studied in dissociated cultures (24, 52, 64). We did not attempt to make patch recordings before this time.

We targeted 36 GFP-LC neurons, exposed them to 9% CO₂, and determined that 31 of 36 GFP-LC cells (86.1%) were stimulated by HA and 5 of 36 GFP-LC cells (13.9%) were insensitive to HA. HA stimulated neurons exhibited a reproducible increase in FR during exposure to 9% CO₂/91% O₂ and a bath pH of ∼7.2. As the bath pH decreased, FR increased within the first 30–60 s of the pH change reaching steady-state FR within 2–4 min (Figs. 3–5). The FR was sustained above baseline throughout exposure to HA and decreased toward baseline on switching back to control solution. The FR during HA ranged from 122 to 927% of the baseline FR with a mean of 250.6 ± 30.2% (n = 31). The confidence internal (CI) ranged from 116 to 535%, with a mean of 220 ± 19.5% (n = 31). The morphology and CO₂/H⁺ response for a pH-sensitive GFP-LC neuron is shown in Fig. 3, where the mean FR increased from 1.82 Hz at baseline to 3.02 Hz during HA exposure or an increase to 165.9% of control. Subsequent pH tests revealed that hypocapnic alkalosis exposure caused the FR of this cell to decrease to 80.2% of the baseline FR.

Fig. 3.

GFP-LC neuron stimulated by hypercapnic acidosis (HA). A: membrane potential (Em) of a GFP- LC neuron stimulated by HA. Top: hypocapnic alkalosis (3% CO₂, pH ∼7.6). Middle: control (5% CO₂, pH ∼7.4). Bottom: HA (9% CO₂, pH 7.2). The mean FRs (in Hz) were 1.01 during alkalosis, 1.82 in control solution, and 3.02 during acidosis. B: simultaneous recordings of FR (top) and extracellular bath pH (bottom) induced by alterations in CO₂. C: this LC neuron was visualized with both fluorescence microscopy (top) and Normaski DIC optics (bottom). Scale bar, 50 μm.

Fig. 5.

Changes in FR of high response GFP-LC neurons during HA. Shown are two examples of high response neurons, one with a baseline FR that was >2 Hz (A) and the other with a baseline FR that averaged <2 Hz(B). Both cells exhibited a large increase in FR during HA.

Some GFP-LC neurons were more sensitive to small fluctuations in pH than others (Figs. 4 and 5). Based on their CO₂ response, we established two categories of HA-stimulated neurons: low responders (<200% of baseline FR during HA; n = 18; 58%) and high responders (>200% baseline FR during HA; n = 13; 42%) with group means of 155.9 ± 6% and 381 ± 55.6%, respectively. The CI for low- and high-response neurons was 146 ± 5.8 and 332 ± 26.4%, respectively. Examples of recordings of high responder GFP-LC neurons are shown in Figs. 4 and 5. The cell in Fig. 4 exhibited FRs (in Hz) of 0.39 (alkalosis), 1.37 (baseline), and 4.86 (acidosis) and increased its FR to a mean of 354.7% of baseline during HA exposure. The CI was 273%. The magnitude of the response was not dependent on the order of presentation of pH perturbation. In this case, the cell shown in Fig. 4 was exposed to hypocapnic alkalosis after establishing a baseline FR. Here, we observed a reduction in FR during exposure to alkalosis followed by a progressive increase in FR as the CO₂ increased from 3 to 5 to 9%. Subsequently, the neuron was exposed to control solution followed by multiple transitions between 5 and 9% CO₂. The FRs during HA and hypocapnic alkalosis were statistically different from the baseline (P < 0.001). High response neurons shown in Fig. 5 consistently increased and decreased FR as expected during shifts between acidosis and alkalosis.

Fig. 4.

Membrane potential of a high response GFP-LC acidosis-stimulated neuron. A: membrane potential of a cultured LC neuron recorded during hypocapnic alkalosis (3% CO₂, pH ∼7.6); control conditions (5% CO₂, pH ∼7.4); and HA (9% CO₂, pH ∼7.2). B: changes in FR and pH induced by changes in CO₂. Exposure to 3 or 9% CO₂ caused a reproducible inhibition or stimulation of spontaneous FR, respectively. Traces in A were taken from the first three episodes shown in B.

A low response HA-stimulated neuron is shown in Fig. 6. This neuron received a depolarizing current injection of 130 pA to establish a baseline FR and this was maintained during acid challenges. The neuron exhibited a reproducible increase in FR in response to acidosis during the first three trials. At the end of the third trial, the FR was lower than the baseline. Subsequently, the depolarizing current was increased from 130 to 140 pA to reestablish the baseline FR at 5% CO₂, and the CO₂/H⁺ challenge was repeated. The cell continued to respond to acid challenge with a similar responsiveness. This GFP-LC HA-stimulated neuron also responded to alkalosis with a reduction in FR during transition to 3% CO₂.

Fig. 6.

Changes in FR of a low-response GFP-LC neuron during HA. Simultaneous recordings of FR and pH from a low-response GFP-LC acidosis-stimulated neuron. The vertical line marks a point at which the level of depolarizing current was increased to adjust the baseline FR. The magnitude of the response was similar before and after the increase in depolarizing current.

All of the HA-stimulated neurons tested for responses to alkalosis exhibited a reproducible reduction in FR with a mean reduction of 30.2 ± 5.11% so that cells fired at 69.8 ± 5.36% of baseline FR.

The magnitude of CO₂/H⁺ responsiveness was dependent on baseline FR.

We found that the likelihood of a neuron being classified as a low or high response cell was dependent on the baseline FR (Fig. 7; Tables 1 and 2). The majority of high response cells (11 of 13, 85%) had a baseline FR of <2 Hz (mean 1.04 ± 0.17 Hz). During HA, these neurons increased their FR on average to 392.7 ± 82.1% of the baseline FR. In contrast, high response cells that started out with a baseline FR of >2 Hz only increased their FR to 243 ± 38.2% of control during HA exposure (P = 0.056).

Fig. 7.

The response of GFP-LC neurons was dependent on baseline FR. Neurons that had a lower baseline FR (<2 Hz) had a greater increase in FR than neurons with higher baseline FRs (>2 Hz).

Table 1.

CO₂/H⁺ responsiveness for GFP-LC neurons during hypercapnic acidosis and hypocapnic alkalosis

Condition	n	pH	Firing Rate, % of Control
Acidosis stimulated
Control (5% CO2)	31	7.43±0.03	100.0±0.00
Acidosis (9% CO2)	31	7.19±0.04	250.6±30.2*
Alkalosis (3% CO2)	14	7.64±0.03	69.8±5.11*
Acidosis insensitive
Control (5% CO2)	5	7.42±0.01	100.0±0.00
Acidosis (9% CO2)	5	7.21±0.01	105.9±1.34

Values are means ± SE.

^*Values are statistically different from control (P < 0.001).

Table 2.

Effect of baseline firing rate on the response to hypercapnic acidosis

Baseline Firing Rate	<2 Hz	>2 Hz	All Neurons
High response	392.7±82.1 (n=11)	243±38.2 (n=2)	381.6±55.6*(n=13)
Mean baseline firing rate	1.04±0.17	2.85±0.05	1.32±0.31†
% of neurons	85%	15%
Low response	166±7.9 (n=9)	149.6±12.7 (n=9)	155.9±6.0*(n=18)
Mean baseline firing rate	1.14±0.17	4.04±0.76	2.95±0.49†
% of neurons	50%	50%

Values are means ± SE. The firing rate during hypercapnic acidosis (expressed as a % of control) shown for high-response and low-response neurons, was statistically different

^*(P < 0.001) as were the baseline firing rates

^†(P = 0.006).

As a group, low-response neurons had a higher baseline FR than high-response neurons. The mean baseline FR was 2.95 ± 0.49 Hz (n = 18) for low-response neurons compared with 1.32 ± 0.31 Hz (n = 13) for high-response neurons. This difference in baseline FR was statistically significant (P = 0.006). However, among low-response neurons, there was no correlation between baseline FR and CO₂/H⁺ responsiveness; r² = 0.014. Thus 50% of low-response cells had a baseline FR of <2 Hz and 50% had a baseline FR of >2 Hz. The mean increase in FR during HA for these two groups was 166 ± 7.9% (<2 Hz) and 149.6 ± 12.7% (>2 Hz), respectively (NS, P = 0.31).

Acidosis-insensitive GFP-LC neurons.

Of GFP-LC neurons, 13.9% (5 of 36) did not change their FR by >20% in response to HA. The action potentials of insensitive neurons were stable and repetitive with membrane potential trajectories that were similar to stimulated neurons. There were no distinguishing morphological or electrophysiological features of these neurons. The average baseline FR for HA-stimulated neurons was 2.26 ± 0.32 Hz (n = 31) and 2.16 ± 0.73 for insensitive neurons (P = 0.89). The average membrane potential obtained after formation of a perforated patch was −58.75 ± 1.21 and −55.6 ± 3.2 mV for HA-stimulated and -insensitive neurons, respectively (P = 0.40). The mean FR of insensitive neurons during HA was 105.9 ± 1.34% of baseline FR. An example is shown in Fig. 8, where the FR remained ∼1 Hz throughout the CO₂/H⁺ protocol.

Fig. 8.

Membrane potential and firing rate of a CO₂-insensitive GFP-LC neuron. A: membrane potential of a CO₂/H⁺-insensitive GFP-LC neuron during exposure to hypocapnic alkalosis (top), control conditions (middle), and HA (bottom). B: the FR of the same neuron did not change in response to acidosis or alkalosis.

Effect of age in culture on responsiveness to hypercapnia.

The number of days that a neuron was maintained in culture did not have a significant effect on CO₂/H⁺ responsiveness. We compared the responses to HA of cells maintained in culture for 12–28, 29–34, and 35–56 days in vitro (DIV) (Fig. 9). There was no significant difference in the response to acidosis of cells 12–28 DIV (P = 0.85) or 35–56 DIV (P = 0.31) compared with 29–34 DIV. There was also no difference in the response of cells 12–28 DIV compared with those 35–56 DIV (P = 0.18).

Fig. 9.

The number of days in culture did not influence CO₂/H⁺ responsiveness. Cells in culture <29 days and >35 days were compared with neurons in culture 29–34 days. There was no statistical difference in CO₂/H⁺ responsiveness of neurons over this age range.

DISCUSSION

We have used perforated patch recordings from the Prp57 transgenic mouse that expresses GFP in catecholaminergic neurons of the LC (66, 67) to quantify changes in the FR of LC neurons in primary cell culture during CO₂/H⁺ challenge. Consistent with previous studies from slices, we found that cells from the LC exhibited chemosensitivity in culture after pharmacological block of fast excitatory and inhibitory synaptic transmission. Our results show that >85% of GFP-positive LC neurons are stimulated by elevated CO₂/H⁺.

Comparison to previous work on chemosensitivity of LC neurons.

In vitro investigations have provided an extensive body of information regarding LC responsiveness to changes in CO₂. The first report of LC chemosensitivity (46) revealed that 38 of 40 cells reduced their FR to ∼64% of baseline when CO₂ was reduced from 5% to 0%. When CO₂ was increased from 2.5% CO₂ to 10% CO₂ or 15% CO₂, LC neurons increased their FR by 53 and 54%, respectively (46). Since then, additional studies have reported similar results, with LC neurons having relatively low sensitivity to pH fluctuations during recordings from brain slices (22, 28, 48, 54, 56) and the in vitro brain stem spinal cord preparation (45).

The responses we observed here from LC neurons in cell culture (a mean increase in FR to 251% of baseline in response to 9% CO₂) were considerably larger than those previously reported in brain slices (up to ∼193% of baseline in response to 15% CO₂) (22, 45, 46, 54, 56). The reasons for the larger response are not clear. One possibility is that LC neurons studied in cell culture are older than those studied in slices, since this leads to larger responses in raphé neurons (see below). However, others have not found a significant effect of age on chemosensitivity of LC neurons (47, 63), and we did not find a significant effect of age on the response of LC neurons in culture after P12 (Fig. 9). Another possibility is that, in culture, the entire cell body and all dendritic arborizations are exposed to changes in bath pH, and/or there is less pH buffering by surrounding cells, as opposed to slices where many dendritic processes are cut or are beneath the surface of the slice where they are protected from bath pH changes. However, previous work demonstrated that a full increase in LC neuron FR induced with HA was only seen with somatic stimulation, and that stimulation of dendrites alone evoked no FR response (54). An additional possibility is that the recordings in brain slices were from LC neurons with a higher baseline FR (1.95 ± 0.12 Hz) (22) than our recordings from cultured neurons (1.04 ± 0.17 Hz). Indeed, the responses of LC neurons reported in brain slices are consistent with our low response group in culture. It will be important to consider this possibility in the future and will be useful to present data in more ways than just the percent change from baseline, as done here and in recent work from raphé neurons (70).

Possible mechanisms for enhanced HA responsiveness in a subpopulation of chemosensitive GFP-LC neurons.

More than 90% of LC neurons in slices or brain stem spinal cord preparations are CO₂ responsive, representing the largest concentration of pH-sensitive neurons for any identified putative central chemoreceptive site (46). Similarly, we observed that 86.1% of cultured GFP-LC neurons were stimulated by HA during recordings where we blocked glutamate and GABA receptors. These conditions eliminate the possibility of fast glutamatergic or GABAergic synaptic influences of chemosensitive neurons on non-chemosensitive cells (25, 49, 71, 72, 74). Thus our data are consistent with studies from slice and en bloc brain stem spinal cord preparations with regard to the percentage of neurons exhibiting responses to CO₂/H⁺.

There is evidence that LC neuron chemosensitivity is highly dependent on gap junctions (28). There may be differences in gap junction expression in our culture system compared with brain slices. Gap junctions may play a role in the regulation of ventilation during development before the maturation of synaptic connections (11, 44, 58–60, 62). Gap junctions have been shown to enhance or amplify the CO₂ response observed in younger animals in the RTN and NTS (30). In a recent study, Hartzler et al. (28) provided evidence that gap junctions influence the percentage of LC neurons with intrinsic chemosensitivity at different ages during development from P0 up to P18. They found that 100% of LC neurons at P0–P4, 80% at P5–P9, and 42% at P10–P18 were stimulated by HA with intact gap junctions. However, when gap junctions were blocked, the percentage of LC neurons activated by HA fell to 50% at P5–P9 and 25% at P10–P18; the P0–P4 group was unaffected. These data suggest that LC chemosensitivity and CO₂ responsiveness changes during development and is dependent on gap junction coupling within the region. We did not explore the possible role of gap junctions in this study, nor did we attempt to block other forms of chemical synaptic transmission, which may have occurred between neurons within our cultures.

There are other aspects of our experimental approach that could lead to differences in the degree of chemosensitivity of LC neurons compared with previous work. For example, there may be differences in glial influences in culture compared with brain slices (20). Our use of 95% O₂ could also alter excitability of LC neurons. Recent studies by Dean and coworkers (13, 14, 38, 39) indicate that some central chemoreceptive neurons in the solitary complex are stimulated by high levels of O₂ and reactive oxygen species. It is not known whether increased O₂ levels can lead to an increase in the response of chemosensitive neurons to CO₂/H⁺, although this is a possibility. Since most in vitro studies and all of the previous studies on raphé neurons were conducted using 95% O₂, we followed the same protocol for studies of cultured LC neurons. This enabled a direct comparison of LC responses to previous data from cultured raphé neurons.

Another important consideration is that we performed our experiments at room temperature. This limits comparison to data from studies that were conducted on LC neurons at higher temperatures, since it is possible that temperature modulates responsiveness to CO₂/H⁺.

Comparison to previous studies of chemosensitive 5-HT neurons in culture.

Dissociated cultures have previously been used to characterize the electrophysiological properties of chemosensitive neurons in the NTS and nucleus ambiguus (24, 52), medulla and pons (64), and medullary raphé (71). These studies employed different approaches to test for and quantify CO₂ responsiveness. To allow direct comparisons, we used the same data acquisition and analysis protocols as used previously to study medullary raphé cells in culture (71, 73). Those previous studies were done on unidentified raphé neurons that were then subsequently verified to be serotonergic using immunohistochemistry for tryptophan hydroxylase. In those experiments, only 15–30% of recorded cells were HA stimulated. However, 73% (22 of 30) of those neurons later identified to be serotonergic neurons were HA stimulated (74). Similarly, we initially began by recording from cultured LC neurons from rats, in which we randomly recorded from neurons and identified LC neurons using TH immunohistochemistry after recording. In that case, most of the cells we recorded were TH negative and unresponsive to changes in pH.

A greater percentage (86.1%) of GFP-LC neurons responded to CO₂ with at least a 20% increase in FR, but LC neurons were slightly less chemosensitive on average than medullary raphé neurons. However, there is overlap between these two cell types, with some chemosensitive LC neurons increasing their FR to >392% of baseline and having an average CI of 332%, a response that is equivalent to some medullary raphé neurons.

Unlike our recordings from cultured LC neurons, raphé neurons increase their chemosensitivity with age in culture and in brain slices (72). Since it is difficult to make stable recordings from raphé neurons in brain slices older than P21, all the published data on chemosensitivity of raphé neurons in slices are from these younger animals, and as such the responses are not as large as many published examples of chemosensitive raphé neurons in culture. However, when raphé neurons of the same age are compared between P5 and P21, the degree of chemosensitivity is approximately the same in culture and slices. Thus there is no evidence that raphé neurons are more chemosensitive in culture than in slices.

In prior experiments, it was found that the morphological features of HA-activated raphé neurons distinguished them from HA inhibited cells from that region (71). Our TH immunofluorescence allowed detailed morphological analysis of LC neurons. Insensitive cells as well as low and high response HA activated cells all expressed GFP, and showed no obvious differences in appearance. Therefore, there was no means to predict based on morphological traits whether a cell would respond, and if it did whether it would be classified as a low or high response neuron or insensitive.

Why are some LC neurons more sensitive than others?

All cultured GFP-LC neurons did not respond with the same robustness. Forty-two percent of HA-stimulated neurons were highly stimulated (Figs. 4 and 5), whereas other GFP-LC neurons responded with a less robust response to the same pH challenge. There are several possible explanations for this. One possibility is that GFP-LC neurons have a heterogeneous composition of pH-sensitive mechanisms. This hypothesis is supported by the recent work by Su et al. (64), which suggests that chemosensitive cells cultured from the medulla and pons may express a broad spectrum of pH-sensitive molecules. Another possibility is simply that all CO₂-sensitive neurons have more or less a similar capability for chemosensitivity, but their response depends on the baseline conditions such as the baseline FR, as shown here. A third possibility is that LC neurons may have non-glutamatergic/non-GABAergic modulatory inputs that lead to differences in the responsive to pH. For example, the LC has intracoerulear inhibitory connections mediated by α₂-adrenoceptors that contribute to the overall level of activity, possibly in response to HA (1, 19, 45). It is also presently unclear whether there are other cell types within our culture system that could lead to release of neuromodulators that influence the response of LC neurons to pH/CO₂.

Functional significance of chemosensitivity of LC neurons.

There are numerous groups of central nervous system neurons that have been proposed to be central respiratory chemoreceptors (21, 48, 50). Currently, it remains controversial which of these pH-sensitive neuronal groups are involved in modulation of respiratory output in response to hypercapnia in vivo, and if so under what conditions this occurs. The presence of pH sensitivity in vitro does not prove that a neuron is a central respiratory chemoreceptor. However, support for this conclusion also comes from in vivo studies, which have shown that cells in the LC increase their spike frequency in response to elevated levels of CO₂, before and after peripheral chemoreceptor deafferentation (18). In other reports, focal acidification of the A6 region caused an increase in phrenic nerve activity (9). In addition, lesions of the LC lead to a decrease in the ventilatory response to hypercapnia in vivo (7, 35). Despite this evidence, it remains unclear how important LC neurons are for control of breathing, and the possibility should be considered that they may play their most important role in cortical arousal by hypercapnia.

Primary cultures of GFP-LC neurons as a model for studying intrinsic pH mechanisms.

The approach we use here offers advantages in the study of LC neuron chemosensitivity. First, GFP expression allows targeting of catecholaminergic neurons specifically. Second, primary culture provides optimal stability for patch-clamp recordings. Third, culture allows optimal access of solutions and pharmacological agents of neurons, since they exist in a monolayer. We conclude that the Prp57 transgenic mouse will serve as a useful experimental model for delineating CO₂/H⁺ chemosensory mechanisms of phenotypically identified catecholaminergic neurons.

GRANTS

Research supported by National Institute of Neurological Disorders and Stroke (NINDS) Grant NS-047422 to S. M. Johnson, by National Institute of Child Health and Human Development and Veterans Affairs Medical Center to G. B. Richerson, and in part by NINDS Grant NS-039407.