Identification of CO 2 Chemoreceptors in Helix pomatia 1

Transcription

1 AMER. ZOOL., 37:54-64 (1997) Identification of CO 2 Chemoreceptors in Helix pomatia 1 JOSEPH S. ERLICHMAN AND J. C. LEITER 2 Departments of Physiology and Medicine, Dartmouth Medical School, Lebanon, New Hampshire Tel.: (603) , FAX: (603)-650-6l30, james.c.leiter@dartmouth.edu SYNOPSIS. Gas exchange in pulmonate snails of the family Helicidae occurs through a highly vascularized diffusion lung known as the mantle. The extent of ventilation of the mantle depends upon the duration and size of opening of an occlusible pore known as the pneumostome. In Helix aspersa and Helix pomatia, pneumostomal size and frequency of opening are exquisitely sensitive to CO 2. Respiratory CO 2 chemosensitivity resides in a discrete region of the subesophageal ganglia. The discharge pattern of many neurons in the chemoreceptor area changes during stimulation with CO 2. However, the electrophysiological response to CO, stimulation alone does not discriminate between CO 2 chemoreceptor cells and CO 2 -insensitive neurons active in the pneumostomal response to CO 2. We identified a subset of CO 2 -sensitive neurons from the larger population of neurons active during CO 2 stimulation. The action potential discharge frequency of CO 2 chemosensory neurons increased in response to CO 2 stimulation. An increased discharge frequency of CO 2 -sensitive neurons was associated with increased pneumostomal opening, and both the size and the frequency of pneumostomal opening increased during CO 2 stimulation. Injecting depolarizing current into individual chemosensory neurons elicited opening of the pneumostome in the absence of CO 2. Action potential generation in response to CO 2 was independent of synaptic transmission. Removal of individual CO 2 -sensitive cells or inhibition of action potential generation in CO 2 -sensitive cells reduced or eliminated pneumostomal responses to CO 2. CO 2 sensitivity in chemoreceptor cells required extracellular calcium, but not sodium. Substituting barium for calcium supported chemoreceptor activity. In summary, we have identified respiratory related, chemosensory neurons that are CO 2 sensitive in the absence of synaptic input. INTRODUCTION The ventilatory response to CO 2 was first described in mammals by Pfliiger (1868). Although the ventilatory effects and response to CO 2 have been well studied in the intervening years, the nature of the stimulus (extracellular ph, the transmembrane ph gradient, or intracellular ph) and the exact location of the sensor(s) remains unknown. CO 2 -sensitive regions have been identified on or below the ventral surface of 1 From the Symposium Control of Arterial Blood Gases: Cardiovascular and Ventilatory Perspectives presented al the Annual Meeting of the Society for Integrative and Comparative Biology, December 1995, at Washington, D.C. * Corresponding author. the medulla, but specific chemoreceptor cells within these regions have not been identified (Millhorn and Eldridge, 1986). CO 2 -sensitive cells have been identified in many regions of the brain (Dean et al., 1990; Dillon and Waldrop, 1992), but the function of these cells is not known. Some investigators have suggested that CO 2 chemosensory cells, per se, will never be found; these investigators have suggested that CO 2 detection may be a property of neural networks, and the "sensor" resides in ph modulation of synaptic receptorneurotransmitter interactions (Brassfield and Gesell, 1942; Dev and Loeschke, 1979). The structural and functional complexity of mammalian brains present major difficulties in studying chemoreceptor mechanisms, 54

2 CO-, CHEMORECEPTORS IN HELIX POMATIA 55 and proving that the CO 2 responses of particular neurons drive the ventilatory response to CO 2 has been impossible to date. We previously studied the respiratory response to CO 2 in a terrestrial pulmonate snail, Helix aspersa. The role of CO 2 in ventilatory control and acid-base balance is similar in active terrestrial molluscs and airbreathing vertebrates even though molluscs and vertebrates have no shared air-breathing evolutionary ancestor (Barnhart, 1992). Exposure of the whole snail to CO 2 augments opening of the pneumostome, a muscular aperture that controls access to the gas exchange surface of the mantle cavity. We found a discrete central chemosensory area in the subesophageal ganglia capable of eliciting pneumostomal opening when stimulated focally with CO 2 (Erlichman and Leiter, 1993). The central ganglia of invertebrates have fewer cells than mammalian brains, and the cell bodies of neurons within the central nervous system are superficial and accessible to study. Given these neuroanatomical advantages of the molluscan nervous system, we tried to isolate CO 2 - sensitive respiratory chemoreceptor neurons in the pulmonate snail Helix pomatia. METHODS Helix pomatia were purchased in the spring (Mavad Hungarian Game Management and Trading Company, Ltd., Budapest, Hungary) and kept in aquaria at C and 80% relative humidity as described previously (Erlichman and Leiter, 1993). Isolated central nervous system-pneumostome preparation Each snail was pinned to a Sylgard slab; the skin was reflected away from the underlying organs; and the nervous tissue pinned down. The subesophageal ganglionic mass was exposed, and the central ganglia were isolated by sectioning all neural connectives except the anal nerve and right parietal nerve; both of these nerves innervate the mantle cavity and pneumostome. Separate wells were constructed around the ganglia and around the combination of the pneumostome and mantle using petroleum jelly. The subesophageal ganglia were manually desheathed after treatment with Pronase (1 mg/ml) (Sigma, St. Louis, MO) for approximately 7 min. Only the ganglia or parts of the ganglia were exposed to test solutions during the experiments. The pneumostome and mantle were independently perfused with control saline equilibrated with 2-3% CO 2 at ph 7.8; this CO 2 concentration range and ph correspond to the in vivo CO 2 concentration and ph at C. Control and test solutions Control saline contained the following salts: 80 mm NaCl, 4 mm KC1, 5 mm MgCl 2, 7 mm CaCl 2, 20 mm NaHCO 3, and 0.2 mm Na 2 PO 4. Control solutions were equilibrated with 2-3% CO 2, buffered with 20 mm N-[2-hydroxyethyl] piperazine-n 1 - [2-ethanesulfonic acid] (HEPES-free acid; Sigma, St. Louis, MO), and titrated to a ph of 7.8 with NaOH. Hypercapnic solutions were equilibrated with 6% CO 2, buffered with 20 mm 3-[N-morpholino] propanesulfonic acid (MOPS-free acid; Sigma, St. Louis, MO), and titrated with NaOH to a ph of 7.2. We determined in preliminary experiments that addition of HEPES or MOPS to the saline solutions did not affect the pneumostomal response to CO 2. In low Ca 2+ /high Mg 2+ saline, the calcium was reduced to 1.7 mm, and the magnesium was increased to 20 mm. In high Ca 2+ /high Mg 2+ solutions, the calcium was increased to 42 mm, and magnesium was increased to 30 mm. In zero Ca 2+ /high Mg 2+ saline, calcium in the normal saline was replaced with magnesium, which increased the magnesium concentration from 5 to 12 mm. In Ca 2+ -free/ba 2+ substituted solutions, barium replaced calcium mm for mm. In Na + -free solutions, the impermeant cation, N-methyl-D-glucamine (Sigma, St. Louis, MO), was substituted for sodium, mm for mm. Microperfusion pipette and microperfusion solutions The microperfusion pipette was constructed from two pipettes in which a small, inner pipette (OD 1.0 mm, ID 0.75 mm) fitted within the lumen of a larger, outer pipette (OD 1.9 mm, ID 1.35 mm). The inner pipette was pulled on a vertical microelec-

3 56 JOSEPH S. ERLICHMAN AND J. C. LEITER trode puller, and the tip was beveled back to a diameter of approximately 100 j.m. The inner pipette was advanced beyond the end of the outer pipette. A vacuum was applied to the outer pipette so that solutions forced through the inner pipette were immediately drawn up through the shank of the outer pipette and into a vacuum trap. A syringe pump delivered the high CO 2 solution to the neurons of the central nervous system through the smaller, inner pipette at a flow rate of approximately 120 nl/sec. This combined pipette system produced a perfusion sphere, the diameter of which was controlled by altering either the amount of suction provided by the vacuum source or the rate at which the test solution was forced through the inner pipette. The size and location of the sphere were visualized by adding dye to the delivery system. Pneumostomal area analysis Images of the pneumostome were recorded through a video camera (Panasonic, Japan) mounted on a dissecting microscope (Olympus, Japan). A counter/timer (Thalner Electronic Laboratories, Inc., MI) placed numbers seriatim on the video images of the pneumostome and simultaneously sent an electronic tick coincident with each numbered image to the video tape recorder on which electrophysiological measurements were stored. The pneumostomal area was measured on digitized images obtained from the video recordings. The area of the pneumostome was determined by a cubicspline fitting of points marked on the edge of the pneumostome (Erlichman and Leiter, 1993). The size of the pneumostome was measured every 10 sec. Each image measured was the largest pneumostomal area observed within each 10 sec period. The numbered video images were matched to the appropriate electrophysiological data by counting the ticks on the electrophysiological record. Electrophysiological measurements Glass microelectrodes (A-M Systems, Seattle, WA) were filled with either 3 M KC1 or 4M K-Acetate and had resistances of 7-15 Mfl and Mfl, respectively. Potentials were measured relative to a KC1 ground (WPI Dri-ref, Sarasota, FL) and amplified using a dual microelectrode amplifier (WPI, Sarasota, FL). Membrane potential and current were displayed on an oscilloscope (Tektronix, Beaverton, OR) and recorded on digital VCR tape (Vetter, Harrisburg, PA). The VCR recorder digitized signals at approximately 5.5 khz. Cell input resistance was determined by injecting hyperpolarizing current ( na) as square waves after balancing the bridge circuit of the amplifier and measuring the change in voltage of the membrane potential. The amplifier bridge was balanced before each determination of input resistance. Five levels of hyperpolarizing current were injected to estimate neuronal input resistance. The average input resistance was calculated from five trials of current injection. Statistics Comparisons between chemoreceptor and non-chemoreceptor cells during normocapnic and hypercapnic tests were made using a 2-way ANOVA (cell type by level of CO 2 ). Correlations between interspike interval and pneumostomal area were made using a Quasi-Newton algorithm for nonlinear curve fitting (SYSTAT, Inc., Evanston, IL). RESULTS To locate the central CO 2 chemoreceptor region in H. pomatia, we used the microperfusion pipette to stimulate small, discrete regions on the surface of the snail central nervous system with hypercapnic saline. We identified two sites on the dorsal surface of the subesophageal ganglia of//, pomatia capable of opening the pneumostome when stimulated with CO 2. Focal CO 2 stimulation on the medial, dorsal surface of the visceral and right parietal ganglia resulted in large increases in pneumostomal area (Fig. 1; n = 43 preparations), whereas focal stimulation of other sites on the ventral and dorsal surface of the subesophageal ganglia failed to result in opening of the pneumostome. The sites sensitive to CO 2 in H. pomatia were anatomically homologous with the site previously described in H. aspersa.

4 CO 2 CHEMORECEPTORS IN HELIX POMATIA 57 left parietal ganglion visceral ganglion dorsal surface - subesophageal ganglia right parietal ganglion FIG. 1. The chemoreceptor area has been shaded on a schematic of the dorsal surface of the subesophageal ganglia. The letters indicate the location of chemoreceptor cells described in Figure 2A and B. Electrophysiological responses To identify CO 2 chemoreceptor cells with a respiratory function within the CO 2 chemosensory areas of the subesophageal ganglia, intracellular electrophysiological responses and video images of the pneumostome were recorded simultaneously during focal hypercapnic stimulation applied with the microperfusion pipette. Pneumostomal responses were recorded without focal hypercapnic stimulation, but with current injection designed to duplicate the electrophysiological response of each cell during focal hypercapnic stimulation. The criteria for a putative respiratory chemoreceptor neuron were that the cell had an electrophysiological response to hypercapnia that was associated with a pneumostomal response and that current injection in the absence of hypercapnia duplicated the pneumostomal response to focal hypercapnia. In this way, single cell electrophysiological responses were coupled to respiratory function. Neurons failing to meet either of these criteria were not studied further. The specificity of single cell responses to CO 2 was tested by examining cells both within the chemosensitive area of the subesophageal ganglia and cells outside this area. Cells on the dorsal surface of the subesophageal ganglia outside the chemoreceptor area were impaled with glass microelectrodes, stimulated focally with hypercapnic saline from the microperfusion pipette and the membrane potential recorded (n = 18 neurons). Hypercapnia had a variety of effects on neuronal activity: action potential generation increased in some cells, spike frequency decreased in others, and some cells showed no effect. Focal hypercapnic stimulation of cells within the CO 2 - sensitive region of the central nervous system demonstrated a similar variety of neuronal responses, but by definition, the pneumostome opened only during CO 2 stimula-

5 58 JOSEPH S. ERLICHMAN AND J. C. LEITER tion of the chemoreceptor region of the subesophageal ganglia. We were interested only in those cells within the chemosensitive area that met our criteria for chemoreceptor cells. Over 180 cells were studied in the chemoreceptor areas of the subesophageal ganglia in approximately 100 snails. We were rarely able to complete the identification process in more than two cells per snail. Based on the early results, the only cells meriting further study were those cells in which CO 2 stimulation increased generation of action potentials. Other patterns of activity were not consistently associated with pneumostomal opening. The firing frequency of many cells within the CO 2 -sensitive region of the central nervous system increased during hypercapnic stimulation (n = 30); however, injecting depolarizing current in the absence of hypercapnia failed to open the pneumostome in 11 of the cells. In the remaining 19 cells, injecting pulses of depolarizing current during normocapnia to mimic the frequency of action potentials observed during hypercapnia alone opened the pneumostome (Fig. 2). In single cells that met the foregoing criteria for chemoreceptor cells, constant hyperpolarizing current was injected to oppose action potential generation during focal CO 2 stimulation. This treatment had variable effects on pneumostomal area. In two cells, eliminating CO 2 -induced action potentials reduced pneumostomal area slightly (Fig. 2A is an example of one of these cells). In other cells, injecting constant hyperpolarizing current had no effect on pneumostomal area (n = 4). In two of the four preparations in which constant hyperpolarizing current had no effect on the hypercapnic increase in pneumostomal area, we identified at least one other chemoreceptor cell within the region of CO 2 stimulation. Therefore, inhibition of action potential generation in one chemoreceptor cell was inadequate to eliminate the pneumostomal response to CO 2 while another chemoreceptor cell in the same region was being stimulated. In those instances in which injection of hyperpolarizing current into a single chemoreceptor cell closed or reduced the size of the pneumostome, no other chemoreceptor cell was identified in the region stimulated by the microperfusion pipette. The relationship between chemoreceptor activity and pneumostomal area was not one-to-one, but the pneumostomal area was correlated with chemoreceptor cell activity. In Fig. 3, the correlation between interspike interval and pneumostomal area is shown for the two cells depicted in Fig. 2. The correlation coefficients between interspike interval and pneumostomal area were significant in both cases, but the correlation coefficients indicate that no more than ~50% of the variation in pneumostomal area can be attributed to activity of the particular chemoreceptor cell being studied. The CO 2 and single cell stimulation studies used to identify respiratory chemoreceptor cells were buttressed by studies in which FIG. 2. Two examples of simultaneous measurements in single chemosensory cells of membrane potential, current injection, and pneumostomal area. The data are plotted as a function of time. In each panel, the first set of hyperpolarizing current pulses was used to balance the bridge circuit, and the second set of hyperpolarizing pulses was used to calculate the input resistance of the cell. In panel A, the neuron was located in the chemosensitive area at point A shown on the schematic of the subesophageal ganglia in Fig. 1. Constant hyperpolarizing current injection kept the pneumostome almost completely closed during normocapnia. Depolarizing current injection (middle of current record) caused the pneumostome to open. Spiking frequency in the cell increased during focal CO 2 stimulation (CO 2 stimulation started at the arrow and persisted throughout the remainder of the tracing), and injection of constant hyperpolarizing current during CO 2 exposure reduced the pneumostomal area markedly. Breaks in the current and voltage records are due to the removal of portions of the record that were obscured by transient electrical noise induced by the microperfusion pipette. In panel B, the neuron was located at point B within the chemosensitive area shown on the schematic of the subesophageal ganglia in Figure 1. Focal CO 2 stimulation was associated with an increased spike frequency compared to normocapnic conditions, and the pneumostome was open more of the time during hypercapnic stimulation than during normocapnia. When depolarizing current was injected at a frequency similar to that observed during focal hypercapnic stimulation (right side of current tracing), the pneumostome opened. CO 2 sensitivity was evident in these cells even during synaptic blockade with modified divalent saline solutions (data not shown).

6 pneumostomal area absolute pixels -t IO b> O O O O O O O 103 pneumostomal area absolute pixels 3 CD 3 O" CD O. CD o o oi m 2O a: 1

7 60 JOSEPH S. ERLICHMAN AND J. C. LEITER 160 * y 120*exp(-0.041*x) re ~ EQ. I Q r =.7O interspike interval (sec) ft Ma * :J y It 1 m» = 287*exp(-0.386*x) r=.75 * interspike interval (sec) FIG. 3. Pneumostomal area has been plotted as a function of interspike interval for both cells shown in Figure 2. Panel A in this figure corresponds to panel A in Figure 2, and panel B corresponds to panel B in Figure 2. Data from normocapnic and hypercapnic treatments were combined. The correlation was significant in both cases {P < 0.01). neurons were accidentally ablated. In four preparations, the cell bodies of the chemoreceptor cell and other surrounding cells (approximately 10 cells) exposed to focal CO 2 stimulation were accidentally pulled from the ganglia by the recording microelectrode. This is a crude anatomical equivalent to the studies in which we injected hyperpolarizing current into single CO 2 -sensitive cells. Following removal of these cell clusters, the microperfusion pipette was used to re-apply hypercapnic CO 2 stimulation at the site of removal. In all snails tested, the removal of cell bodies at a CO 2 -sensitive site eliminated pneumostomal movements during subsequent CO 2 stimulation at that site. However, other sites in the same preparation remained responsive to CO 2. Evidently, the neurons that were removed were not essential components of the pattern generating system of pneumostomal movement: the pneumostome continued to open and close and pneumostomal responses to CO 2 were intact when hypercapnic stimulation was applied at other sites within the chemosensitive areas. Electrophysiological response during synoptic blockade To determine whether the respiratory CO 2 -sensitive neurons were intrinsically sensitive to CO 2 or whether the electrical activity of the cells was dependent upon synaptic input, the central nervous system was perfused with either low Ca ++ /high Mg ++ or high Ca ++ /high Mg ++ saline solutions. These modified divalent solutions block chemical and electrical synaptic transmission (Asada and Bennett, 1971; Pappas et al., 1971) and have been used previously in snails (Prior and Gelperin, 1977; Syed et al., 1991). Synaptic blockade with these solutions applied to the central nervous system only caused all spontaneous pneumostomal movements to cease. Excitatory and inhibitory post-synaptic potentials were frequent in all cells studied, but postsynaptic potential activity ceased after synaptic blockade. In the subset of cells in the subesophageal ganglia that generated action potentials in response to CO 2 and were capable of opening the pneumostome during current injection (n = 19), 11 cells no longer generated action potentials when CO 2 was applied, either focally or in the bath, during perfusion with modified divalent solutions. The remaining eight cells continued to generate action potentials in response to hypercapnia following synaptic blockade by perfusion of modified divalent solutions, suggesting to us that these cells were intrinsically CO 2 sensitive (Fig. 4). Ion substitution experiments were done to identify the ionic species carrying the action potential in CO 2 chemosensitive cells. The

8 CO, CHEMORECEPTORS IN HELIX POMATIA 61 A normocapnia control saline hypercapnia B normocapnia hypercapnia control saline C control saline CO, on high Ca high Ca~/high Mg*> high Ca+Vhigh Mg<- CO,on low Ca+ Na+-free Ca^-free/Ba^-substituted - zero Ca+Vhigh Mg«CO, on 50m VI 50 mv 50 mv L 5 sec 2 sec 5 sec FIG. 4. In the ion substitution experiments shown in this figure, the entire central nervous system was perfused with each solution, but hypercapnic stimulation was applied focally. Panel A: Intracellular recordings from a CO 2 chemoreceptor cell located in the chemoreceptor area of the visceral ganglion. Top tracing: The neuron was relatively quiet in normal saline and increased firing frequency when stimulated focally with hypercapnic saline. Middle tracing: The cell was quiet during perfusion of normocapnic, high Ca 2+ /high Mg 2+ saline, but continued to generate action potentials in response to CO 2. Arrow indicates the injection of a small, hyperpolarizing current. Bottom tracing: Spiking persisted in low Ca 2+ /high Mg 2+ saline during focal, hypercapnic stimulation. Panel B: Intracellular recordings from a CO 2 chemoreceptor cell located in the medial margin of the right parietal ganglion. First tracing: Action potential generation in control saline under normocapnic and hypercapnic conditions. Second tracing: Spiking continues following exposure to hypercapnic saline containing low Ca 2+ / high Mg 2+ saline. Third tracing: Cell was quiet during perfusion with Na + -free saline, but action potentials were still generated in response to depolarizing current injection (arrows). Focal hypercapnic stimulation still elicited action potentials in Na + -free saline. Fourth tracing: During focal hypercapnic stimulation, action potentials persisted during perfusion of the entire central nervous system with Ba 2+ substituted, Ca 2+ free saline. Panel C: Intracellular recordings from a CO 2 chemoreceptor cell located in the medial margin of the right parietal ganglion. Top tracing: CO 2 induced action potentials under control conditions. Middle tracing: Action potentials persisted during perfusion with high Ca 2+ /high Mg 2+ substituted saline, which blocked synaptic transmission. Bottom tracing: Ca 2+ -free perfusion eliminated hypercapnic responses. saline solutions in which sodium and calcium were replaced by N-methyl-Dglucamine and barium, respectively, were used only after demonstrating that stimulation of the cell by focal hypercapnic saline and current injection caused pneumostomal opening, and the cell was intrinsically CO 2 sensitive. Two cells were studied in each condition. During perfusion of the entire subesophageal ganglia with Na + -free solutions, hypercapnic sensitivity persisted, and injecting depolarizing current elicited action potentials morphologically similar to those generated in control solutions (Fig. 4B.). The chemoreceptor cells were completely quiescent during perfusion with Ca ++ -free, Mg ++ substituted solutions (Fig. 4C). Hypercapnic sensitivity persisted during perfusion with Ca ++ -free, Ba ++ substituted solutions (Fig. 4B).

9 62 JOSEPH S. ERLICHMAN AND J. C. LETTER TABLE 1. Effects of CO2 on membrane potential and input resistance in chemoreceptors and other cells within the chemosensitive areas.* Chemoreceptors Non-chemoreceptors Control 22.1 ± ± 3.4 Input resistance (MO) CO ± ± 4.2 Resting membrane potential (mv) Control CO; ± ± ± ± 2.4 * "Non-chemoreceptors" indicates that the CO 2 responsiveness of these neurons was dependent upon synaptic input. Values are means and standard errors for samples numbering 8 and 11 for chemoreceptors and non-chemoreceptors, respectively. Interactions between cell type and level of CO 2 were not significant in the ANOVAs, and main effects are reported here. Hypercapnia caused the input resistance to increase in both chemoreceptors and non-chemoreceptors (F,, 7 = 8.43, P < 0.01). Hypercapnia had no effect on the resting membrane potential which was less negative in chemoreceptors than in non-chemoreceptors (F,, 7 = 5.23, P = 0.01). Electrophysiological characterization The resting membrane potential and input resistance of chemoreceptor and nonchemoreceptor cells within the chemosensitive areas of the subesophageal ganglia are shown in Table 1. The average resting membrane potential of cells that met the criteria for chemoreceptors was 48.6 mv during normocapnia, whereas the average resting membrane potential of other cells in which CO 2 sensitivity was dependent upon synaptic transmission was slightly more negative, 54.1 mv. CO 2 stimulation did not have any dramatic or consistent effect on the resting membrane potential of any of the cells within the CO 2 -sensitive region from which we recorded (Table 1). The input resistance was not different between chemoreceptor and non-chemoreceptor cells, and CO 2 increased input resistance in chemoreceptors and non-chemoreceptors alike. DISCUSSION We identified individual CO 2 chemoreceptor cells within the population of cells in the CO 2 -sensitive area of the subesophageal ganglia. These cells met the following criteria: 1) hypercapnic stimulation elicited an electrophysiological response associated with pneumostomal opening; 2) duplication of hypercapnia-induced electrical activity by current injection in the absence of hypercapnia resulted in opening of the pneumostome; and 3) CO 2 responses were independent of synaptic input. We believe that these criteria are necessary and sufficient to define chemoreceptor cells. Moreover, hyperpolarizing current injected into cells meeting the foregoing criteria reduced the size of the pneumostomal area, antagonizing the effect hypercapnic stimulation, when only one chemoreceptor cell was in the area stimulated by focally applied hypercapnic saline. Removing the soma of CO 2 chemosensory cells eliminated the pneumostomal response to hypercapnia at the site of cell removal. These findings demonstrate that stimulation of a single cell can modify the pneumostomal response in the snail and impart functional significance to the neuron being recorded in a way not possible in mammals. Chemoreceptor cells meeting similarly stringent criteria have not been identified in mammals. Chemoreceptor cells do not seem to be essential elements in the neural circuit controlling pneumostomal activity. Pneumostomal responses to CO 2 could still be elicited after removing small clumps of cells containing individual chemoreceptor cells and after hyperpolarizing current injection into single chemoreceptor cells if another CO 2 -sensitive site remained functional. Neither the pneumostomal area nor the frequency of pneumostomal opening and closing was coupled one-to-one with electrical activity of particular chemoreceptor cells (Fig. 2, 3); chemoreceptor activity promoted pneumostomal opening, but the pattern and size of the pneumostome appeared to be controlled by a pattern generator for which CO 2 was one among many inputs. A variety of afferents contribute to pneumostomal control. Electrical or mechanical stimulation of body wall afferents elicits reflex pneumostomal responses independent of CO 2 responsiveness (Erlichman and Leiter, 1994). Hypoxia also promotes pneumostomal opening in the aquatic snail Lymnea. stagnalis, apparently through mantle affer-

10 CO, CHEMORECEPTORS IN HELIX POMATIA 63 ents (Janse et al., 1985). We believe that CO 2 chemoreceptors contribute information to a central pattern generator that integrates many afferents and determines a final pattern of pneumostomal activity. This belief is consistent with the pattern of activity of the chemoreceptor cells we identified, the level of correlation between chemoreceptor activity and pneumostomal area (Fig. 3), the effect on pneumostomal area of stimulating other afferents, and the persistence of pneumostomal activity after clumps of cells containing one or more chemoreceptor cells were removed from the central nervous system. The firing frequency of neurons meeting our criteria for chemoreceptor cells increased during CO 2 stimulation by focal application of hypercapnic saline within the chemoreceptor region. However, this response was not unique to chemoreceptor cells. The variety of effects of CO 2 on the membrane potential demonstrates that particular changes in neuronal electrical activity during hypercapnia (depolarization, hyperpolarization, increase in firing frequency, etc.) bear no unique relationship to the functional role of a given neuron: depolarization during CO 2 stimulation was not uniformly associated with either the chemoreceptor area of the subesophageal ganglia or the ability to control pneumostomal activity. That depolarization during hypercapnia is not confined to chemoreceptor cells is contrary to hypotheses and chemoreceptor identification schemes proposed by others (Fukuda and Loescke, 1979; Dean et al., 1990; Jarolimek et al., 1990). Simply noting the electrophysiological response of a cell to CO 2 was not useful in identifying respiratory chemoreceptor cells. Many CO 2 -sensitive cells had no role that we could detect in pneumostomal function; identification of chemoreceptor cells depended on coupling the non-specific electrophysiological response to CO 2 to a specific physiological function. The ion substitution experiments indicate that CO 2 sensitivity may depend on a Ca 2+ conductance but not on a sodium conductance. Ba 2+ can substitute for Ca 2+ as a charge carrier but cannot support many secondary cellular events mediated by Ca 2+ (Hille, 1992). The overshoot of the action potential of CO 2 chemoreceptor cells exceeded E Na+ (approximately +54 mv) and approached E Ca++ (approximately mv). Hence, the action potential in CO 2 chemoreceptor cells seems to be carried by calcium. Purely calciumdependent action potentials have been reported previously in Helix (Eckert and Lux, 1976). In retrospect, the only identifying electrophysiological feature of chemoreceptor cells was a less negative membrane potential. We are reluctant to place too much emphasis on this finding at this time. The membrane potential can vary more than 4 mv depending on the season, and we did not systematically study CO 2 -sensitive and insensitive cells outside the chemoreceptor area. We prefer to emphasize those features that did not provide useful discrimination between chemoreceptor cells and nonchemoreceptor cells: the normocapnic input resistance, the change in input resistance during CO 2 exposure, and the pattern of electrical activity during hypercapnic stimulation. The respiratory CO 2 chemoreceptor cells in the snail were intrinsically CO 2 - sensitive. Therefore, CO 2 chemosensitivity need not be a network or synaptic property but may be the result of stimulation of discrete CO 2 responsive sensory neurons within the central nervous system. These findings are not consistent with the long standing hypothesis that CO 2 chemoreception resides in ph modulation of neurotransmitter-receptor interactions (Brassfield and Gesell, 1942; Fukuda et al., 1978; Dev and Loeschke, 1979; Nattie et al., 1989) because chemoreceptor activity could be elicited in the absence of synaptic input from neurons with a clear respiratory function. CO 2 may modulate synaptic transmission, and synaptic events may modulate the function of the CO 2 chemosensory cells (Erlichman et al., 1994); but the primary sensory event in the transduction of CO 2 chemosensory information in the pulmonate snail seems to be an intrinsic property of the chemosensory cells that is independent of synaptic input.

11 64 JOSEPH S. ERUCHMAN AND J. C. LEITER ACKNOWLEDGMENTS This work was supported by grant no. HL Dr. Leiter is the recipient of an American Lung Association Career Investigator Award. As always, we thank Dr. F. V. McCann for lending equipment to us. REFERENCES Asada, Y. and M. V. L. Bennett Experimental alteration of coupling resistance at an electrotonic synapse. J. Cell Biol. 49: Barnhart, M. C Acid-base regulation in pulmonate molluscs. J. Exp. Zool. 263: Brassfield, C. R. and R. Gesell Examples of the acid-neurohumoral mechanism of nervous integration. Fed. Proc. l:a10. Dean, J. B., P. A. Bayliss, J. T. Erickson, W. L. Lawing, and D. E. Millhorn Depolarization and stimulation of neurons in the nucleus tractus solitarii by carbon dioxide does not require chemical synaptic input. Neuroscience 36: Dev, N. B. and H. H. Loeschke Topography of the respiratory and circulatory responses to acetylcholine and nicotine on the ventral surface of the medulla oblongata. Pflugers Arch. 379: Dillon, G. H. and T. G. Waldrop In vitro responses of caudal hypothalamic neurons to hypoxia and hypercapnia. Neuroscience 51: Eckert, R. and H. D. Lux A voltage-sensitive persistent calcium conductance in neuronal somata of Helix. J. Physiol. (London) 254: Erlichman, J. S. and J. C. Leiter CO 2 chemoreception in the pulmonate snail, Helix aspersa. Respir. Physiol. 93: Erlichman, J. S. and J. C. Leiter Central chemoreceptor stimulus in the terrestrial, pulmonate snail, Helix aspersa. Respir. Physiol. 95: Erlichman, J. S., J. C. Leiter, and F. V. McCann Nitric oxide modulates CO 2 chemoreception in the pulmonate snail. Neuroscience Abstr. 20:A957. Fukuda, Y, Y Honda, M. Schlafke, and H. H. Loeschke Effect of H+on the membrane potential of silent cells in the ventral and dorsal surface layers of the rat medulla in vitro. Pflugers Arch. 376: Fukuda, Y. and H. H. Loeschke Cholinergic mechanism involved in the neuronal excitation by H+ of the respiratory chemosensitive structures in the ventral medulla oblongata of rats in vitro. Pflugers Arch. 379: Hille, B Ionic channels of excitable membranes. Sunderland, MA, Sinauer Associates, Inc. pp Janse, C, J. van der Wilt, J. van der Plas, and M. van derroest Central and peripheral neurones involved in oxygen perception in the pulmonate snail Lymnaea stagnalis (Mollusca, Gastropoda). Comp. Biochem. Physiol. 82A:459-^69. Jarolimek, W., U. Misgeld, and H. D. Lux Neurons sensitive to ph in slices of the rat ventral medulla oblongata. Pflugers Arch. 416: Millhorn, D. E. and F. L. Eldridge Roleofventrolateral medulla in regulation of respiratory and cardiovascular systems. J. Appl. Physiol. 61: Nattie, E. E., J. Wood, A. Mega, and W. Goritski Rostral ventrolateral medulla muscarinic receptor involvement in central ventilatory chemosensitivity. J. Appl. Physiol. 66: Pappas, G. D., Y. Asada, and M. V. L. Bennet Morphological correlates of increased coupling resistance at an electrotonic synapse. J. Cell Biol. 49: Pfliiger, E Uber die Ursache der Atembewegungen sowie der Dyspnoe und Apnoe. Pflugers Arch. 1: Prior, D. J. and A. Gelperin Autoactive molluscan neuron: Reflex function and synaptic modulation during feeding in the terrestrial slug, Limax maximus. J. Comp. Physiol. A 114: Syed, N. I., D. Harrison, and W. Winlow Respiratory behavior in the pond snail Lymnaea stagnalis. I. Behavioral analysis and the identification of motor neurons. J. Comp. Physiol. A 169: