• Amperometric recordings were conducted to investigate the ability of hypoxia and anoxia to evoke quantal catecholamine secretion from isolated type I cells of the rat carotid body.
• Hypoxia (P_O2 8–14 mmHg) consistently failed to evoke catecholamine secretion from type I cells, when cells were perfused either at room temperature (21-24 °C) or at 35–37 °C, and regardless of whether Hepes- or HCO₃⁻/CO₂-buffered solutions were used.
• Elevating extracellular [K⁺] caused concentration-dependent secretion from individual type I cells, with a threshold concentration of approximately 25 mM. In the presence of this level of extracellular K⁺, hypoxia (P_O2 8–14 mmHg) caused a marked enhancement of secretion which was fully blocked by 200 μM Cd²⁺, a non-specific blocker of voltage-gated Ca²⁺ channels.
• Anoxia (N₂-equilibrated solution containing 0·5 mM dithionite) evoked exocytosis from type I cells when extracellular [K⁺] was 5 mM. This secretion was completely inhibited by removal of extracellular Ca²⁺, but was not significantly affected by Cd²⁺ (200 μM), Ni²⁺ (2 mM), Zn²⁺ (1 mM) or nifedipine (2 μM). Secretion was also observed when 0·5 mM dithionite was added to air-equilibrated solutions.
• Anoxia also evoked secretion from chemoreceptive phaeochromocytoma (PC12) cells, which was wholly Ca²⁺ dependent, but unaffected by Cd²⁺ (200 μM).
• Our results suggest that hypoxia can evoke catecholamine secretion from isolated type I cells, but only in the presence of elevated extracellular [K⁺]. This may be due to the cells being relatively hyperpolarized following dissociation. In addition, we have shown that dithionite evokes catecholamine release regardless of P_O2 levels, and this release is due mainly to an artefactual Ca²⁺ influx pathway activated in the presence of dithionite.

The mechanism(s) by which hypoxia evokes neurotransmitter release from type I cells has received much attention over the past decade (Gonzalez et al. 1994; Peers & Buckler, 1995; Peers, 1996; Lopez-Barneo, 1996). Several workers have used patch-clamp recordings to show that hypoxia causes rapid and reversible inhibition of K⁺ channels in isolated type I cells (Lopez-Barneo et al. 1988; Delpiano & Hescheler, 1989; Peers, 1990; Stea & Nurse, 1991; Buckler, 1997) which leads to membrane depolarization sufficient to activate voltage-gated Ca²⁺ channels (Buckler & Vaughan-Jones, 1994a). The subsequent Ca²⁺ influx is then presumed to trigger neurosecretion. However, there is a notable lack of direct evidence demonstrating that hypoxia can evoke secretion from isolated type I cells of the rat. To date, only one group has reported hypoxia-evoked release of catecholamines, using type I cells isolated from carotid bodies of the rabbit (Urena et al. 1994; Montoro et al. 1996).

Most electrophysiological studies aimed at elucidating the mechanisms of chemotransduction have used type I cells isolated either from rabbit or rat carotid bodies. There are clear and important differences in the electrophysiological properties of cells from these two species (reviewed by Lopez-Lopez & Peers, 1997). Therefore, results gained from rabbit type I cells cannot necessarily be assumed to be representative of the chemoreceptive process in other species.

In the present study, we have investigated the ability of hypoxia to evoke catecholamine release from isolated rat type I carotid body cells. To this end, we have used amperometric recordings which allow the resolution of released contents of individual secretory vesicles (Wightman et al. 1991; Chow & Von Ruden, 1995). Our results support the idea that hypoxic stimulation of catecholamine release occurs via voltage-gated Ca²⁺ entry, but also suggest that secretory responses to hypoxia of type I cells may be influenced by their isolation. In addition, our findings indicate that the use of dithionite to achieve anoxic solutions generates artefactual responses.

Phaeochromocytoma (PC12) cells were maintained in continuous culture and plated onto poly-D-lysine-coated coverslips for electrochemical recordings as previously described (Taylor & Peers, 1999).

Figure 1 Hypoxia does not evoke catecholamine release from isolated rat type I carotid body cells

The same amperometric equipment was also used to monitor P_O2 levels in the recording chamber, except that the polarity of the microelectrode was reversed to -800 mV (see Mojet et al. 1997). The time course of fall of P_O2 in the recording chamber was highly reproducible for any given degree of hypoxia or anoxia.

All results are presented as individual examples and means ±s.e.m. (each mean value determined from cells isolated from at least 4 rats) and statistical comparisons were made using Student's unpaired t test.

Figure 1 illustrates representative amperometric recordings from individual rat carotid body type I cells. Our initial studies examined the ability of hypoxia (P_O2 8–14 mmHg) to evoke release when cells were perfused with a Hepes-buffered solution at room temperature (21-24°C). In 40 cells examined (e.g. Fig. 1A), hypoxia failed to evoke secretion. A similar lack of secretory response was found when cells were perfused with Hepes-buffered solution at 35–37°C (e.g. Fig. 1B, representative of 9 cells examined). Previous studies have suggested that transduction of hypoxic stimuli is dependent on the presence of HCO₃⁻ in the perfusate (Shirahata & Fitzgerald, 1991). We therefore examined the ability of hypoxia to evoke secretion whilst cells were perfused with a solution buffered with HCO₃⁻/CO₂ at both room temperature (Fig. 1C, representative of 15 cells tested) and 35–37°C (Fig. 1D, representative of 13 cells tested). Again, this level of hypoxia consistently failed to evoke secretion. This lack of response to acute hypoxia was not due to cells being unable to undergo exocytosis, since when they were subsequently exposed to a solution of 50 mM K⁺ after returning to normoxia, secretion was always observed (Fig. 1A–D; 50 mM K⁺ applied at the time point indicated by the arrow).

Since the levels of hypoxia used here failed to evoke secretion, but have previously been shown by us to cause inhibition of K⁺ currents and membrane depolarization in rat type I cells (Wyatt et al. 1995; Wyatt & Peers, 1995), we speculated that hypoxia-induced depolarization might be insufficient to cause significant Ca²⁺ influx required for triggering catecholamine release. To investigate this further, we firstly examined the relationship between membrane depolarization (caused by raising extracellular [K⁺]) and exocytosis. Results are plotted in Fig. 2. Secretion was only detected at [K⁺]_o levels of 25 mM or higher, and appeared to peak at 75 mM, with a slight decline in the secretory response at 100 mM. This secretion was entirely due to Ca²⁺ influx through voltage-gated Ca²⁺ channels, since bath application of the non-selective Ca²⁺ channel blocker Cd²⁺ (200 μM) completely abolishes high [K⁺]_o-evoked secretion (Hatton & Peers, 1997). In order to determine how membrane potential changed with varying extracellular [K⁺], we also recorded membrane potential over the same range of [K⁺]_o levels, using the perforated-patch technique. As illustrated in Fig. 3, the relationship between membrane potential and [K⁺]_o deviated significantly from that predicted if cells were purely K⁺ permeable (indicated by dashed line), suggesting that other ionic conductances exert important influences on membrane potential. Since our previous voltage-clamp studies indicated that type I cells display a ‘window’ Ca²⁺ channel current (i.e. there may be significant, tonic Ca²⁺ channel activity; Peers et al. 1996), the measured membrane potential values could be positive to that predicted for a purely K⁺-permeable membrane because of Ca²⁺ influx through voltage-gated Ca²⁺ channels. To test this idea, we examined the effects of 200 μM Cd²⁺ (which completely blocks Ca²⁺ channel currents in these cells) on membrane potential ([K⁺]_o 5 mM). As illustrated in Fig. 4, Cd²⁺ most commonly (4 out of 6 cells) caused membrane depolarization (mean depolarization 8.9 ± 1.2 mV), and in two other cells was without effect. If Cd²⁺ selectively blocked tonic Ca²⁺ influx at the resting membrane potential, a hyperpolarization would have been expected. This observation therefore suggests that tonic Ca²⁺ influx through voltage-gated Ca²⁺ channels does not contribute to resting membrane potential. The effects of Na⁺ removal (equimolar replacement with N-methyl-D-glucamine) were also investigated. As Fig. 4 exemplifies, removal of Na⁺ consistently caused a marked, reversible hyperpolarization (mean -21.0 ± 2.8 mV, n = 6 cells). This observation is in accordance with the study of Buckler & Vaughan-Jones (1994b), and indicates that a tonic Na⁺ influx is the major factor underlying the positive deviation of measured membrane potential in type I cells from values predicted for a purely K⁺-selective membrane.

Figure 2 Elevated [K+]o evokes graded secretion from rat type I carotid body cells

Figure 3 The relationship between [K+]o and membrane potential in isolated rat type I carotid body cells

Figure 4 Effects of Cd2+ application and Na+ removal on membrane potential of type I cells

To examine further our speculation that isolated type I cells were unable to depolarize sufficiently to activate voltage-gated Ca²⁺ entry, we examined the ability of hypoxia to evoke secretion in the presence of an elevated [K⁺]_o level of 25 mM. As indicated in Fig. 5, this level of [K⁺]_o was sufficient in itself to evoke a small amount of secretion (see also Fig. 2), and when hypoxia was co-applied, secretion was increased nearly 3-fold (Fig. 5A and B) as compared with the secretion evoked by 25 mM K⁺ alone. This enhancement of secretion caused by hypoxia was most likely to be due to Ca²⁺ influx through voltage-gated Ca²⁺ channels, since it was completely abolished by Cd²⁺ (200 μM; Fig. 5A and B).

Figure 5 Hypoxia evokes exocytosis from type I cells in the presence of elevated [K+]o

Integration of individual, clearly resolved exocytotic events was performed, in order to compare the quantal size of events evoked under normoxic and hypoxic conditions (see Methods). Results are presented in Fig. 6 for events evoked by 25 mM K⁺ in normoxia (Fig. 6A), 25 mM K⁺ in hypoxia (Fig. 6B; P_O2 8–14 mmHg), and 50 mM K⁺ in normoxia (Fig. 6C). Clearly, the distributions were approximately normal in each case, and mean values were similar also (means ±s.d. 0.29 ± 0.06, 0.28 ± 0.08 and 0.30 ± 0.06 pC^1/3, respectively). Thus, increasing release rate under normoxic conditions by application of 50 mM K⁺ instead of 25 mM K⁺ did not affect quantal size, and release evoked by hypoxia in the presence of 25 mM K⁺ (the frequency of which was intermediate between 25 mM and 50 mM K⁺ under normoxic conditions) was also similar. These findings suggest that hypoxia and raised K⁺ levels evoke release of the same intracellular pool of vesicles.

Figure 6 Comparative distributions of quantal size determined from events evoked by 25 mM K+ and hypoxia

In addition to examining the effects of hypoxia in the presence of raised [K⁺]_o, we also examined the effects of an anoxic stimulus, achieved by addition of sodium dithionite (0.5 mM) to N₂-equilibrated solutions containing the control K⁺ level of 5 mM. As illustrated in Fig. 7A, the presence of dithionite itself interfered with the polarized carbon fibre electrodes, giving rise to a substantial, sustained offset current which returned to control levels when the solution was changed back to a normoxic one. Nevertheless, when the electrode was placed adjacent to a type I cell, quantal events could be resolved, superimposed on the large offset current (Fig. 7B). Thus, anoxia was able to evoke catecholamine secretion in the presence of 5 mM , and this was a consistent observation (n = 19 cells), with a mean exocytotic rate of 25 ± 6 events per minute. We did not analyse these events for determination of Q, since the reactivity of dithionite itself with the electrodes may introduce inaccuracies.

Figure 7 Amperometric detection of anoxia-evoked catecholamine release from type I carotid body cells

We next investigated the Ca²⁺ dependency of the secretory response of type I cells to anoxia, and this response was consistently abolished by removal of extracellular Ca²⁺ (replaced with 1 mM EGTA; e.g. Fig. 8A). However, in contrast to the effects of either raised [K⁺]_o (Hatton & Peers, 1997), or hypoxia in the presence of raised [K⁺]_o (see Fig. 5), application of Cd²⁺ (200 μM) did not significantly affect anoxia-evoked secretion (Fig. 8B). This observation raised the question of whether release was being evoked by anoxia (achieved by dithionite addition), or by the presence of dithionite itself. To investigate this, we applied an air-equilibrated solution containing 0.5 mM dithionite to type I cells, and found that this solution (which had a measured P_O2 of 130 mmHg in the recording chamber) was also capable of evoking exocytosis (Fig. 8C). Indeed, the number of exocytotic events evoked per minute was not significantly different from that evoked by anoxia (Fig. 8D). Furthermore, anoxia-evoked secretion was not significantly affected by 2 mM Ni²⁺ or 1 mM Zn²⁺ (Fig. 8D) or by nifedipine (2 μM; Fig. 8D). Thus, although anoxia-evoked secretion was entirely dependent on the presence of extracellular Ca²⁺, known blockers of voltage-gated Ca²⁺ channels could not prevent such secretion.

Figure 8 Anoxia-evoked catecholamine release from type I carotid body cells is Ca2+ dependent but unaffected by blockers of voltage-gated Ca2+ channels

Given that previous studies had suggested that hypoxia-evoked secretion is dependent on voltage-gated Ca²⁺ entry (see Introduction and Discussion), results presented in Fig. 8 were unexpected, and suggested a possible effect of dithionite which has not previously been reported. To investigate this possibility further, we also examined anoxia-evoked catecholamine release from phaeochromocytoma (PC12) cells. These cells have been proposed as model chemoreceptor cells (Zhu et al. 1996; Conforti et al. 1999), and our previous studies have shown that hypoxia evokes release of catecholamines in a graded manner which is entirely dependent on Ca²⁺ influx through voltage-gated Ca²⁺ channels (Taylor & Peers, 1998). Anoxia also evoked secretion from PC12 cells (e.g. Fig. 9A and D), and this was fully abolished by removal of extracellular Ca²⁺ (Fig. 9B and D). However, Cd²⁺ (200 μM) failed to suppress this secretory response (Fig. 9C and D). This finding was comparable to that seen in type I cells (Fig. 8), and suggested that under anoxic conditions (in the presence of dithionite) catecholamine release could be evoked in both cell types via Ca²⁺ influx which does not appear to be through Cd²⁺-sensitive channels. This is in contrast to hypoxia-evoked secretion in the presence of 25 mM K⁺, which was wholly dependent on Cd²⁺-sensitive channels (see Fig. 5).

Figure 9 Anoxia-evoked catecholamine release from PC12 cells is Ca2+ dependent but unaffected by Cd2+

The main aim of the present study was to investigate directly whether hypoxia could evoke catecholamine release from isolated type I cells of the rat carotid body and, if so, whether evoked release was dependent on Ca²⁺ influx through voltage-gated Ca²⁺ channels. A wealth of previous studies have provided indirect evidence that hypoxia-evoked catecholamine release from type I cells is a requisite step in carotid body chemotransduction (reviewed by Gonzalez et al. 1994), but direct evidence of such release from type I cells in the absence of other cell types is lacking. Recently, such direct evidence for hypoxia-evoked dopamine release from rabbit type I cells was published (Urena et al. 1994; Montoro et al. 1996), providing convincing support for the widely held idea that release requires membrane depolarization and Ca²⁺ influx through voltage-gated Ca²⁺ channels (Gonzalez et al. 1994; Peers & Buckler, 1995). However, marked species differences exist when comparing the biophysical properties of rabbit and rat type I cells (reviewed by Lopez-Lopez & Peers, 1997) and, perhaps most importantly, rabbit type I cells spontaneously generate action potentials in normoxia (Lopez-Barneo, 1996). The frequency of these action potentials increases during hypoxia, owing to selective inhibition of a specific, voltage-gated K⁺ channel, termed the K_O2 channel (Ganfornina & Lopez-Barneo, 1992). By contrast, rat type I cells do not fire spontaneous action potentials in normoxia but under hypoxic conditions they show a persistent depolarization (Buckler & Vaughan-Jones, 1994a; Wyatt & Peers, 1995), termed a receptor potential (Buckler & Vaughan-Jones, 1994a; Buckler, 1997), superimposed on which may appear Ca²⁺-dependent action potentials (Buckler & Vaughan-Jones, 1994a; Buckler, 1997). Owing to these species differences, it was important to demonstrate hypoxia-evoked secretion from rat type I cells, and to investigate whether such release could be accounted for by voltage-gated Ca²⁺ entry.

Our finding that raised [K⁺]_o could evoke secretion in a graded manner (Fig. 2) and also caused graded membrane depolarization (Fig. 3) supports the idea that voltage-dependent exocytosis does occur in rat type I cells, as previously described (Hatton & Peers, 1997). In order to investigate the magnitude of depolarization required to evoke secretion in response to elevating [K⁺]_o, we also studied the relationship between [K⁺]_o and membrane potential. This relationship deviated markedly from that predicted from a purely K⁺-selective membrane, such that at lower [K⁺]_o levels, the same change in [K⁺]_o produces a smaller depolarization that at higher [K⁺]_o levels. At present, other ionic conductances which contribute to membrane potential in the presence of a physiological K⁺ gradient are not fully determined. Type I cells also possess a significant Cl⁻ conductance, but only under either a swelling osmotic gradient or when intracellular cAMP levels increase (Carpenter & Peers, 1997), so this conductance is not likely to contribute to determining resting potential. In the present study, we also found no evidence to suggest that a tonic ‘window’ Ca²⁺ current contributed a depolarizing influence (Fig. 4), but we did find that removal of extracellular Na⁺ caused a marked hyperpolarization (Fig. 4), indicating that Na⁺ influx was a major factor in setting resting membrane potential in these cells, as previously suggested (Buckler & Vaughan-Jones, 1994b).

Results presented in Fig. 2 indicate that secretion is only detectable under normoxic conditions when [K⁺]_o is elevated to 25 mM or higher concentrations. Elevation of [K⁺]_o to 25 mM causes a membrane depolarization of approximately 15 mV (Fig. 3). Our previous studies have shown that hypoxia (P_O2 12–20 mmHg, a similar range to the levels of hypoxia used in the present study) causes a membrane depolarization of approximately 8 mV (Wyatt & Peers, 1995; Wyatt et al. 1995). According to data presented in Fig. 2, this is unlikely to be a sufficiently large depolarization to cause exocytosis and, in accordance with this idea, similar levels of hypoxia consistently failed to evoke secretion (Fig. 1). This lack of secretory effect was found regardless of temperature, or whether the extracellular solution was buffered with Hepes or with HCO₃⁻/CO₂ (Fig. 1). Thus, this level of hypoxia, which is known to excite the carotid body in situ (see Gonzalez et al. 1994), and cause depolarization and a detectable rise of [Ca²⁺]_i in isolated rat type I cells (Buckler & Vaughan-Jones, 1994a; Wyatt & Peers, 1995), is insufficient to evoke catecholamine release. However, when the same level of hypoxia was applied to cells in the presence of 25 mM [K⁺]_o secretion was significantly enhanced (Fig. 5). Furthermore, this hypoxia-evoked secretion was abolished by the non-selective Ca²⁺ channel blocker, Cd²⁺ (Fig. 5A). The simplest interpretation of these findings is that the degree of depolarization caused by hypoxia in cells already exposed to the depolarizing influence of 25 mM K⁺ is sufficient to evoke catecholamine release. Thus our results would support previous suggestions from studies in rat type I cells (Buckler & Vaughan-Jones, 1994a; Wyatt & Peers, 1995) and more direct evidence from intact (Obeso et al. 1992) and dissociated (Urena et al. 1994; Montoro et al. 1996) rabbit carotid body tissue that hypoxia-evoked catecholamine release requires Ca²⁺ influx through voltage-gated Ca²⁺ channels. It is conceivable that 25 mM K⁺ provides an influx of Ca²⁺ that in some way ‘primes’ type I cells so that they release transmitter in response to hypoxia in a Cd²⁺-resistant manner. We cannot discount this possibility at present, but consider it unlikely, and it is not supported by previous studies, detailed above.

In common with many other workers (Biscoe & Duchen, 1990; Pang & Eyzaguirre, 1992; Buckler & Vaughan-Jones, 1994a; Buckler, 1997), we intensified the hypoxic stimulus applied to type I cells by addition of dithionite to the N₂-equilibrated solution, giving rise to solutions which were near anoxic (P_O2 < 2–3 mmHg). This severe hypoxic/anoxic stimulus (applied with perfusate [K⁺] of 5 mM) consistently evoked secretion from type I cells and, as predicted if such release required Ca²⁺ influx through voltage-gated Ca²⁺ channels, such secretion was entirely dependent on the presence of extracellular Ca²⁺ (Figs 7 and 8). However, anoxia-evoked exocytosis was not significantly affected by any of the pharmacological agents known to inhibit most or all voltage-gated Ca²⁺ influx (Ni²⁺, Cd²⁺, nifedipine; also Zn²⁺; Fig. 8). This suggested that under anoxic conditions in the presence of dithionite, a Cd²⁺-resistant Ca²⁺ influx pathway was activated. To examine this further, we measured catecholamine secretion from PC12 cells, which have previously been shown to undergo exocytosis in response to hypoxia in a graded manner due to Ca²⁺ influx through voltage-gated Ca²⁺ channels (Taylor & Peers, 1999). Anoxia was capable of producing a robust secretory response in these cells (Fig. 9) but, as was the case in type I cells, this could only be prevented by removal of extracellular Ca²⁺ and not by application of Cd²⁺. Thus, dithionite appears to induce an unidentified Ca²⁺ influx pathway which can contribute to the secretory response in both cell types. The mechanisms underlying such an effect of dithionite are unknown at present, but may be associated with free radical damage to the plasma membrane (Archer et al. 1995). Clearly, these findings serve as a caution against the use of dithionite in studies of the effects of severe hypoxia or anoxia in chemoreceptive cells.

Although the present study has demonstrated hypoxia-evoked secretion from type I carotid body cells, this was only observed when extracellular K⁺ was elevated (Fig. 5). This finding appears to contrast with previous studies which have shown hypoxia to evoke catecholamine release from intact carotid body preparations in vitro (e.g. Obeso et al. 1992). We cannot account for this discrepancy at present, but suggest that it may be due to type I cells being relatively hyperpolarized following the isolation procedure. In the intact carotid body, type I cells form tight clusters which are in synaptic contact with afferent sensory nerve endings as well as autonomic nerve endings (e.g. Verna, 1997). This raises the likely possibility that they are under the tonic, paracrine influence of basal levels of neurotransmitters (such as catecholamines, acetylcholine and numerous others; Gonzalez et al. 1994; Prabhakar, 1994) released from neighbouring cells. Certainly, type I cells possess receptors for many of these transmitter types (e.g. Benot & Lopez-Barneo, 1990; Wyatt & Peers, 1993; Dasso et al. 1997). In support of this idea, the membrane potential observed in the present study (-51 mV; [K⁺]_o 5 mM; in good agreement with other studies; Buckler & Vaughan-Jones 1994a, b,) is more negative than values reported in more intact preparations (e.g. Eyzaguirre et al. 1989), although a note of caution should be added when making such comparisons of values determined by patch-clamp versus intracellular microelectrode techniques.

In summary, the present study indicates that hypoxia can evoke catecholamine release from isolated type I cells of the rat carotid body. This release is dependent on Ca²⁺ influx through voltage-gated Ca²⁺ channels, but was only observed in the presence of elevated [K⁺]_o. This may be due to type I cells being artificially relatively hyperpolarized when isolated. Anoxia could evoke release of catecholamines from both type I cells and PC12 cells, but this was at least partly due to stimulation of an unidentified Ca²⁺ influx pathway which was not sensitive to known blockers of voltage-gated Ca²⁺ channels. We suggest that this pathway should be considered as an experimental artifact.

This work was supported by The Wellcome Trust and The British Heart Foundation.