Journal of Neurophysiology

Ca2+-Induced Ca2+ Release Mediates a Slow Post-Spike Hyperpolarization in Rabbit Vagal Afferent Neurons

Kimberly A. Moore, Akiva S. Cohen, Joseph P. Y. Kao, Daniel Weinreich


Moore, Kimberly A., Akiva S. Cohen, Joseph P. Y. Kao, and Daniel Weinreich. Ca2+-induced Ca2+ release mediates a slow post-spike hyperpolarization in rabbit vagal afferent neurons. J. Neurophysiol. 79: 688–694, 1998. The relation between Ca2+-induced Ca2+ release (CICR) elicited by action potentials (APs) and a Ca2+-dependent slow post-spike hyperpolarization (AHPslow) in acutely dissociated adult rabbit nodose neurons was studied using microfluorimetric calcium measurements in conjunction with standard intracellular current- and voltage-clamp recording techniques. The magnitude of the AP-induced transient increase in [Ca2+]i (ΔCat) was used to monitor CICR. There was a close correlation between the magnitude of the ΔCat and the AHPslow current over the range of 1–16 APs (r = 0.985). Functional CICR blockers, ryanodine (10 μM), thapsigargin (100 nM), 2,5-di(t-butyl)hydroquinone (10 μM) or cyclopiazonic acid (10 μM), selectively reduced the peak amplitude of the AHPslow ≥91%. In five neurons, simultaneous recordings of the ΔCat and the AHPslow revealed that both responses were blocked in parallel. These findings indicate that CICR is necessary for the generation of the AHPslow in rabbit nodose neurons. The ΔCat rises and decays significantly faster than the AHPslow. This temporal disparity suggests that activation of the AHPslow by Ca2+ may require additional signal transduction steps.


Afferent information in the peripheral nervous system is encoded by modulation of action potential (AP) frequency. Activation of distinct classes of potassium channels can dramatically affect the frequency and the pattern of neuronal firing. In ∼35% of the ∼16,000 vagal somata (inferior or nodose ganglion neurons) of the rabbit, the pattern of AP firing can be effectively modified by a Ca2+-dependent K+ current. This current produces a slowly developing and persistent post-spike hyperpolarization (AHPslow) that plays a significant role in the regulation of membrane excitability, and is responsible for spike frequency accommodation in these neurons (Fowler et al. 1985; Higashi et al. 1984; Weinreich and Wonderlin 1987). Following a single AP, theAHPslow displays a slow rise time-to-peak (0.3–0.5 s) and a long duration (3–15 s). The critical importance of the AHPslow in mediating spike frequency accommodation is exemplified by observations that its inhibition by various endogenous autacoids results in an increase in firing frequency from <0.1 to >10 Hz (Undem and Weinreich 1993; Weinreich and Wonderlin 1987). Consistent with a functional role of the AHPslow, some of these autacoids also produce an augmentation in impulse activity recorded in vagal afferents in vivo (Coleridge and Coleridge 1984). The AHPslow is not restricted to vagal afferent neurons; analogous slow afterpotentials have been characterized in sympathetic, myenteric, and CNS neurons (reviewed by Sah 1996).

Observations of the AHPslow recorded in rabbit nodose neurons, and analogous data from other nerve cells, suggest that Ca2+ discharged from a Ca2+-induced Ca2+ release (CICR) pool may contribute to the generation of theAHPslow. Intracellular calcium chelation with ethylene glycol-bis(β-aminoethyl ether)-N,N,N′,N′-tetraacetic acid (EGTA) (Higashi et al. 1984) or bis-(o-aminophenoxy)-N,N,N′,N′-tetraacetic acid (BAPTA) (Cohen et al. 1994) abolishes the AHPslow. Manipulations that increase [Ca2+]i, such as intracellular injection of Ca2+, or applications of ouabain or low concentrations of caffeine, augment and prolong the AHPslow (Higashi et al. 1987). CICR channel modulators (e.g., ryanodine) or sarcoplasmic/endoplasmic reticulum Ca2+-ATPase (SERCA) inhibitors [e.g., thapsigargin or 2,5-di(t-butyl)hydroquinone (DBHQ)] block an analogous AHPslow in sympathetic (Jobling et al. 1993; Kawai and Watanabe 1989, 1991), parasympathetic (Yoshizaki et al. 1995), and vagal dorsal motor nucleus neurons (Sah and McLachlan 1991).

In rabbit nodose neurons, Ca2+ influx produced by 1–7 APs is amplified 5- to 10-fold by CICR (Cohen et al. 1997). The time courses of the CICR-dependent transient increases of intracellular free Ca2+ and the AHPslow appear qualitatively similar (Cohen et al. 1994), suggesting that mobilization of Ca2+ from a CICR pool may underlie the AHPslow. In the current work we describe experiments using physiological stimuli, in conjunction with manipulations of CICR, to demonstrate that CICR is essential for the development of the AHPslow.


Cell dissociation

New Zealand white rabbits of either sex weighing 1–2 kg were obtained from Robinson Services (Winston-Salem, NC) and killed by pentobarbital sodium overdose (100 mg/kg), as approved by the Institutional Animal Care and Use Committee. Dissociated nodose neurons were prepared as described previously (Leal-Cardoso et al. 1993). After a 3- to 12-h incubation at 37°C, neurons were maintained at room temperature (22–24°C) to minimize neurite outgrowth and were suitable for experimental use for at least 3–4 days (Magee and Schofield 1991; and our own observations). There were no observable differences in visual appearance or basic electrophysiological properties of the cells based on the sex of the animals or length of time neurons were kept in culture.

Electrode fabrication, recording chamber, and drug delivery

Intracellular recording microelectrodes were fabricated on a Flaming/Brown model P-97 micropipette puller (Sutter Instrument, San Francisco, CA). The aluminosilicate micropipettes (Sutter) had resistances ranging from 30 to 70 MΩ when filled with 4 M potassium acetate or 3 M potassium chloride.

A custom recording chamber provided superfusion (3–5 ml/min) of a 25-mm coverslip with physiological salt solution via a gravity-flow system. It was mounted on the stage of an inverted microscope (Zeiss IM35) equipped with a ×40 phase-contrast oil-immersion objective (Fluor, NA 1.3, Nikon) to allow direct visualization of neurons for intracellular recording and fluorescence measurements. In some experiments where only electrical measurements were made, we used a compound microscope equipped with Hoffman optics (×400).

Electrophysiological recording

Standard intracellular stimulating and recording techniques were used to monitor electrical activity with “sharp” micropipettes (see Christian et al. 1989 for details). Current-clamp recordings were made with an Axoclamp-2A amplifier (Axon Instruments, Foster City, CA) either in bridge (filtering at 10 kHz) or in the discontinuous mode (sample rate 5 kHz, filtering at 3 kHz). AP-induced AHPslow currents (I AHP) were recorded using a hybrid voltage-clamp technique. Varying numbers of APs were evoked by transmembrane depolarizing current pulses (3 nA, 3 ms, 10 Hz). One hundred milliseconds after the final depolarizing current pulse, the amplifier was electronically switched from current-clamp to voltage-clamp mode to record the I AHP. Current and voltage signals were digitized with a Neurocorder (Neurodata Instruments, Delaware Water Gap, NJ) for storage on videocassette tapes for off-line analysis. The membrane input resistance (R in) of the cell was monitored by measuring the magnitude of the electrotonic voltage transient produced by hyperpolarizing current pulses (100 pA, 150 ms). Analysis of electrophysiological data were performed using pClamp 6.2 software (Axon Instruments).

During experiments the cells were superfused with 22–24°C Locke solution that had the following composition (in mM): 136 NaCl, 5.6 KCl, 1.2 NaH2PO4, 14.3 NaHCO3, 1.2 MgCl2, 2.2 CaCl2, and 10 dextrose (continuously equilibrated with 95% O2-5% CO2; pH 7.2–7.4).


Reagents were procured from the following vendors: thapsigargin from LC-laboratories (Woburn, MA), ryanodine and cyclopiazonic acid from Calbiochem (La Jolla, CA), and fura-2/acetoxymethyl ester (AM), BAPTA/AM, and BAPTA sodium salt from Molecular Probes (Eugene, OR). Inorganic salts were obtained from VWR (Piscataway, NJ).

Unless otherwise noted, drug solutions were prepared daily from concentrated (>10 mM) stock solutions that were stored frozen. Drugs were delivered via the superfusate by switching a three-way valve to a reservoir containing a known concentration of the drug in oxygenated Locke solution.

Ca2+ measurements

To measure Ca2+, cells on coverslips were incubated for 45–60 min at room temperature (22–24°C) in a solution containing 1 μM fura-2/AM as previously described (Cohen et al. 1994). After incubation the coverslip was placed in the recording chamber and superfused with Locke solution. Fura-2 fluorescence measurements were performed with a DeltaScan Illumination System [Photon Technology International (PTI), South Brunswick, NJ] coupled to the microscope through a fiberoptic cable. Each neuron under study was alternately illuminated with 340-nm and 380-nm light and the fluorescence emission, after passing through a 510-nm band-pass filter, was sampled by a photomultiplier tube, the output from which was digitized and stored for subsequent analysis. Instrument control, data acquisition, and analysis were performed using FELIX 1.1 software (PTI) running on a dedicated microcomputer.

[Ca2+]i calibration

All fura-2 fluorescence records were corrected for background fluorescence by subtracting the light intensity measured after cell lysis with digitonin. Values of intracellular [Ca2+] ([Ca2+]i) were calculated using the equation of Grynkiewicz et al. (1985) [Ca2+]i=Kd×[(RRmin)/(RmaxR)]×[(Sf2)/(Sb2)] where R is the ratio F 340/F 380, R min and R max are the minimum and maximum values of the ratio, attained at zero and saturating Ca2+ concentrations, respectively. F 340 is the fluorescence emitted by the dye when excited at 340 nm, and F 380 is the fluorescence emitted by the dye when excited at 380 nm. S f2/S b2 is the ratio of fluorescence intensities for Ca2+-free and Ca2+-bound indicator measured with 380-nm excitation. R min, R max, and S f2/S b2 were determined from six acutely dissociated neurons used specifically for calibration purposes.

Data are expressed as means ± SE. Analysis of variance(ANOVA) and Student's t-test were used to assess significant differences between calculated means; P < 0.05 was considered significant. Unless specified, results were replicated in at least three different neurons.


Effect of BAPTA on the AHPslow

AP-induced elevations of [Ca2+]i (ΔCat) recorded in rabbit nodose neurons are strictly dependent on an intact CICR process (Cohen et al. 1997). Additionally, superfusion of acutely isolated rabbit nodose neurons with the AM ester form of the Ca2+ chelator BAPTA (25 μM) concomitantly abolishes both the AHPslow and ΔCat (Cohen et al. 1994). In the current work, we have extended these results and tested whether BAPTA treatment affected Ca2+ influx. In 11 neurons incubated with 10–30 μM BAPTA/AM, the AHPslow was completely blocked within 2–7 min (Fig. 1 A). By contrast, incubating neurons with the membrane-impermeant sodium salt of BAPTA (20–30 μM) for 8 min produced no measurable effect on the AHPslow (n = 3). Although the reduction of the AHPslow by intracellular BAPTA is most likely due to Ca2+ buffering, BAPTA could conceivably alter Ca2+ influx through voltage dependent calcium channels (VDCCs). However, because the fast Ca2+-dependent afterhyperpolarization (AHPfast) following each AP was unaffected by intracellular BAPTA loading, it appears that Ca2+ influx through VDCCs is not compromised (cf. Fig. 1, Ba and Bb; n = 10). When the superfusate was switched to one containing 100 μM CdCl2, a blocker of Ca2+ influx, the AHPfast was reduced in amplitude (Fig. 1 Bc). The small remaining component of the AHPfast is presumably a reflection of the delayed rectifier. In this neuron, abolition of the AHPslow was accompanied by a marked increase in the excitability of the neuron as evidenced by the lowering of the threshold for AP firing, and an increase in the average frequency of firing, measured during a depolarizing ramp, from 1 to 5.5 Hz (Fig. 1 C).

Fig. 1.

Effect of bis-(o-aminophenoxy)-N,N,N′,N′-tetraacetic acid/acetoxymethyl ester (BAPTA/AM) on slow post-spike hyperpolarization (AHPslow) and on excitability in an isolated nodose neuron. A: AHPslow elicited by 2 action potentials (APs) in control Locke solution and in Locke solution containing 10 μM BAPTA/AM. APs were evoked by transmembrane depolarizing current pulses (4 nA, 1.5 ms, 10 Hz) and are truncated. Dashed line represents the resting membrane potential (−60 mV). Resting membrane input resistance was 70 MΩ. Bath application of BAPTA/AM blocks the AHPslow within 5 min without changing the resting membrane potential or membrane input resistance. B: responses digitized at a higher rate. The AHPfast, which precedes theAHPslow, is unaffected by BAPTA/AM (compare Ba with Bb). The Ca2+ dependence of the AHPfast is illustrated in Bc, where the neuron is superfused with 100 μM CdCl2 for 30 s, which blocks most of the AHPfast, and all of the AHPslow. The residual component of the AHPfast recorded in CdCl2 is presumably mediated by the delayed rectifier current. C: depression of the AHPslow markedly increases neuronal excitability. The average number of APs induced by a current ramp protocol (1 nA, 2 s) increased from 1 to 5.5 Hz when the AHPslow was blocked. Scale bar represents 3 mV, 2 s in A; 15 mV, 0.25 s in B; 15 mV, 0.5 s in C.

Relation between the number of APs and the magnitude of the AHPslow current

We tested the dependence of the AHPslow on CICR by examining whether the magnitude of the AHPslow current (I AHP) saturates with increasing numbers of APs in a manner parallel to that observed for the ΔCat (see Fig. 1 in Cohen et al. 1997) (and see Fig. 2). Neurons were current clamped at their resting membrane potential (approximately −60 mV) while varying numbers of APs were evoked by suprathreshold transmembrane depolarizing current pulses (2 ms, 10 Hz). One hundred milliseconds after the last AP, the neuron was voltage clamped to approximately −60 mV to measure the magnitude of the resulting I AHP. Over the range of 1–8 APs, the amplitude of the I AHP increased steeply. Beyond eight APs, the I AHP amplitude began to level off, almost reaching a plateau by the 40th AP. Over the range of 1–40 APs, the relation between I AHP amplitude and number of APs was well fit by a rectangular hyperbola (χ2 = 2.35, r = 0.975, n = 10; Fig. 2 B). These data are remarkably similar to those relating the amplitude of the ΔCat to the number of APs (dashed curve in Fig. 2 B, redrawn from Cohen et al. 1997). When the I AHP elicited by a given number of APs is plotted against the amplitude of the ΔCat evoked by the same number of APs, there was a close correlation (r = 0.985) between the magnitudes of ΔCat and I AHP (Fig. 2 C), suggesting a potential mechanistic linkage between the two.

Fig. 2.

Effect of varying numbers of APs on the amplitude of the I AHP recorded in acutely isolated adult rabbit nodose neurons. A: I AHPs were evoked by APs using a hybrid voltage-clamp technique (see Methods). APs were produced by transmembrane depolarizing current pulses (3 nA, 3 ms, 10 Hz) from a holding potential of −60 mV. One hundred ms following the last AP, the amplifier is switched to voltage-clamp mode (−60 mV holding potential) to record the I AHP. The number of APs is indicated to the left of each trace. B: filled circles represent the mean I AHP amplitude (±SE) elicited by varying numbers of APs. The number of observations is indicated near each data point. I AHPs are normalized to the maximal response recorded in a given neuron. The normalized amplitudes of the I AHPs depicted in A are also shown (open circles). The continuous curve is a rectangular hyperbola fit to the mean data (χ2 = 2.35, r = 0.975). Dashed curve is the least-squares rectangular hyperbola relating the normalized ΔCat amplitude to the number of APs (originally presented in Cohen et al. 1997). C: I AHP amplitude elicited by a given number of APs (1, 2, 4, 8, and 15) is plotted against the amplitude of the Ca2+ transient elicited by the same number of APs (Ca2+ transient data from Cohen et al. 1997). Values are mean ± SE. There is a linear relation between the amplitude of the Ca2+ transient and the I AHP over the range of 1–15 APs (r = 0.985).

Effects of modulators of CICR on the magnitude of the AHPslow

We have previously reported that ryanodine and SERCA inhibitors, such as thapsigargin, DBHQ, and cyclopiazonic acid (CPA) (reviewed by Inesi and Sagara 1994) can abolish caffeine- and AP-induced ΔCats in rabbit nodose neurons (Cohen et al. 1997), where a robust process of CICR underlies AP-induced ΔCats. In the current experiments, we investigate the effects of these reagents on the AHPslow.

When neurons were superfused with Locke solution containing ryanodine (10 μM), thapsigargin (100 nM), DBHQ (10 μM), or CPA (10 μM), the amplitude of the AHPslow was reduced >90% (Table 1 and Fig. 3). Figure 3, A and B, illustrates experiments where neurons were superfused with either ryanodine or thapsigargin. Although these compounds disable CICR by entirely different mechanisms, in these experiments they both abolished the AHPslow completely. The time required to abolish the AHPslow by these CICR antagonists (20 and 18 min, respectively) was similar to the time required to block the ΔCat (Cohen et al. 1997). At the concentrations used, ryanodine and thapsigargin do not alter either the AP waveform or the amount of Ca2+ influx per AP (Cohen et al. 1997).

View this table:
Table 1.

Effects of CICR inhibitors on the AHPslow

Fig. 3.

Effects of Ca2+-induced Ca2+ release (CICR) inhibitors on the AHPslow and the AP-induced Ca2+ transient. A: superimposed traces of an AHPslow elicited by 4 APs, recorded in control Locke solution and 20 min after switching to Locke solution containing ryanodine (10 μM). Ryanodine completely abolished the AHPslow. B: analogous experiment with thapsigargin (100 nM) recorded in another nodose neuron. Eighteen min after switching to Locke solution containing thapsigargin, the AHPslow was completely blocked. C: effect of 2,5-di(t-butyl)hydroquinone (DBHQ) on the AP-induced Ca2+ transient and on the AHPslow. Top traces: superimposed Ca2+ transients evoked by a train of 4 APs (10 Hz) recorded in control Locke solution and 5 min after switching to Locke solution containing 10 μM DBHQ. Bottom traces: AHPslows recorded simultaneously with the Ca2+ transients. DBHQ treatment completely blocked both the Ca2+ transient and the AHPslow. Resting [Ca2+]i was 91 nM. Fluorescence data were acquired at 10 Hz. D: time course of block of the AHPslow and the Ca2+ transient for the neuron shown in C. Resting membrane potentials in A–C were −56, −58, and −67 mV, respectively. AP amplitudes are truncated.

In five neurons, we simultaneously recorded the ΔCat and the AHPslow in normal Locke solution and in Locke solutions containing DBHQ (n = 2), ryanodine (n = 2), or thapsigargin (n = 1). The average values for control ΔCat andAHPslow, evoked by four APs, were 28 ± 4.3 nM and 7 ± 2.0 mV, respectively. Drug treatment completely abolished both the ΔCat and the AHPslow in all five neurons. The data in Fig. 3 C illustrate the effects of DBHQ on the ΔCat and the AHPslow recorded in the same neuron. Five minutes after switching to a superfusate containing DBHQ, the ΔCat and the AHPslow were concomitantly blocked. Parallel inhibition of the ΔCat and the AHPslow in this neuron is illustrated in Fig. 3 D. Collectively, these data support the conclusion that CICR is an important determinant of the AHPslow in rabbit nodose neurons.

Comparison between the time course of the AHPslow and the ΔCat

There are two mechanisms by which Ca2+ from AP-induced CICR could regulate the AHPslow. First, Ca2+ may initiate signaling cascades that ultimately control the K+ channels underlying the AHPslow. Alternatively, Ca2+ itself could activate the K+ channels directly, and thus exert moment-by-moment control of the AHPslow. If the AHPslow is directly dependent on Ca2+ released from the CICR pool, the AHPslow and the rise in [Ca2+]i elicited by an AP might display similar kinetics. Quantitative kinetic comparisons between these two variables are, unfortunately, subject to some uncertainty, because the time course of the ΔCat reflects global changes in [Ca2+]i, whereas the kinetics of the AHPslow are determined by events at the plasma membrane. Nonetheless, we determined the time-to-peak and 10 to 90% decay time for both the AHPslow and the ΔCat elicited by 1–8 APs (Table 2). The time-to-peak for ΔCats elicited by 1–8 APs was 1.0 ± 0.06 s (n = 32). The AHPslow evoked by 1–8 APs peaked in 1.9 ± 0.13 s (n = 25), and was significantly slower in reaching peak than the ΔCat (P < 0.0001). The ΔCat also decayed much more rapidly than the AHPslow (2.8 ± 0.42 s, n = 24, versus 6.7 ± 0.38 s, n = 26; P < 0.0001). These quantitative differences are graphically illustrated by Fig. 4, in which corresponding ΔCats and AHPslows are scaled and overlaid. Although our preliminary observations (Cohen et al. 1994) suggested a similar time course for the AHPslow and the ΔCat, quantitative analysis of a much larger data set now reveals significant temporal differences between these parameters.

View this table:
Table 2.

Time-to-peak and decay time of the AHPslow current and the Ca2+ transient evoked by varying numbers of action potentials

Fig. 4.

Comparison of the temporal profiles of AP-evoked Ca2+ transients and the corresponding AHPslows. A: superposition of Ca2+ transient (thick line) and AHPslow (thin line) evoked by 4 APs. The AHPslow has been inverted and scaled for ease of visual comparison with the Ca2+ transient. Arrows mark the peaks of the Ca2+ transient and AHPslow. B and C: superpositions analogous to that shown in A, but for 8 and 14 APs, respectively. Resting membrane potential and [Ca2+]i were −55 mV and 70 nM, respectively. All traces were recorded in the same neuron. Population data for these experiments are presented in Table 2.


Our principal finding is that Ca2+ release from the CICR pool is necessary for the development of the AHPslow in nodose neurons. In virtually all rabbit nodose neurons, there exists a CICR pool that can be activated by Ca2+ influx resulting from a single AP (Cohen et al. 1997). By contrast, neurons with AHPslows are randomly distributed and occur in only ∼35% of the population (Fowler et al. 1985). Thus expression of the AHPslow is dependent on a neuronal property independent of the CICR pool. The most parsimonious explanation is that the expression of the AHPslow K+ channel determines whether a nodose neuron exhibits a AHPslow. The recent cloning of a small conductance, Ca2+-activated K+ channel (Köhler et al. 1996) with characteristics similar to the AHPslow current may soon allow direct testing of this hypothesis.

In rabbit nodose neurons, APs induce transient increases in [Ca2+]i (ΔCats) that are dependent on an intact CICR pool (Cohen et al. 1997). If the Ca2+ trigger is increased by progressively increasing the number of APs from 1 to 64, Ca2+ influx per AP remains constant, but Ca2+ release from the CICR pool (i.e., the magnitude of the ΔCat) first increases sharply with the number of APs but then plateaus. This is reflected by the hyperbolic relation between the magnitude of the AP-evoked ΔCat and number of APs (dashed curve in Fig. 2 B, redrawn from Cohen et al. 1997). In the current work, a similar hyperbolic relation was observed when the magnitude of I AHP was plotted against a varying number of APs (Fig. 2 B). Over the range of 1–16 APs, there was a close correlation (r = 0.985) between the magnitudes of the ΔCat and the I AHP (Fig. 2 C). This strong correlation suggests that CICR could underlie the AHPslow.

There are several interpretations of the hyperbolic relation between the I AHP and the number of APs. First, the amount of Ca2+ influx per AP might decrease progressively during a train of APs. However, we have demonstrated previously that over the range of 1–32 APs, the Ca2+ influx per AP does not change (Cohen et al. 1997). Thus the plateauing of the I AHP-AP relation with increasing numbers of APs must reflect another process. A second possibility is that the number of AHPslow potassium channels in nodose neurons is limiting. This too is unlikely because when the concentration of extracellular Ca2+ is doubled, the plateau of the I AHP-AP relation is elevated (unpublished observations). Finally, saturation of the I AHP amplitude could reflect exhaustion of the releasable content of the CICR pool. In principle, application of caffeine, the standard CICR agonist, before and during the plateau phase could test whether the CICR pool was exhausted. Unfortunately, caffeine activates not only CICR but also a novel calcium influx pathway in these neurons (Hoesch et al. 1997) thereby complicating interpretation of such experiments.

Pharmacological evidence that supports the involvement of CICR in the generation of the AHPslow was obtained by using two functionally distinct classes of reagents to abolish CICR. Ryanodine disrupts CICR by interfering with intracellular Ca2+ release channels (the ryanodine receptors) (Fill and Coronado 1988). On the other hand, SERCA inhibitors (thapsigargin, DBHQ, and CPA) inhibit the ATPases that transport Ca2+ into stores, including the CICR pool (Inesi and Sagara 1994), thus allowing dissipation of the stores through leakage. When neurons were treated with SERCA inhibitors or with ryanodine, at concentrations and incubation times previously shown to block caffeine- and AP-induced ΔCats (Table 2) (Fig. 5 in Cohen et al. 1997; also unpublished observations), the AHPslow was consistently abolished (Fig. 3 and Table 1). Because these reagents do not affect Ca2+ influx (Cohen et al. 1997), these pharmacological data support the contention that, under physiological conditions, Ca2+ release from the CICR pool is necessary for development of the AHPslow. What remains unresolved, however, is whether Ca2+ released from the CICR pool is sufficient, by itself, to activate and sustain this persistent afterpotential. The time course of elevated [Ca2+]i near the plasma membrane can be approximated by the duration of the Ca2+-dependent fast post-spike afterpotential (AHPfast, 150 ms) that precedes the AHPslow (Fowler et al. 1985). Assuming that Ca2+ influx gates the AHPslow potassium channels directly (Köhler et al. 1996; Sah 1996), it seems unlikely that AP-induced Ca2+ influx alone could sustain this protracted afterpotential.

The significant temporal disparity between the AHPslow 1 and the underlying ΔCat (Fig. 4 and Table 2) also argues against any model where Ca2+ ions alone are sufficient to activate the AHPslow directly. Rather, the incommensurate time courses of the AHPslow and the ΔCat favor a model in which activation of the AHPslow by Ca2+ is mediated by additional signal transduction steps. In this regard, it is notable that Lasser-Ross et al. (1997) have inferred a similar mechanism for Ca2+ activation of the AHPslow in vagal motoneurons, where they suggest that additional second-messenger systems may be involved. Our results and those of Lasser-Ross et al. (1997) concerning the temporal characteristics of the AHPslow and the ΔCat are concordant in all but one important respect. In vagal motoneurons, the peak of the ΔCat coincides with the last AP. In contrast, in vagal afferent (nodose) neurons, there is a delay between the last AP and the peak of the ΔCat. This delay probably reflects the contribution of CICR to the calcium transient.

Several processes could contribute to the slow time-to-peak and long duration of the AHPslow. These include slow diffusion of Ca2+ ions, CICR, an enzymatic signal transduction cascade, or a combination of these factors. In CA1 pyramidal neurons of the hippocampus, there exists a well-characterized AHPslow with slow kinetics and physiological and pharmacological properties similar to the AHPslow recorded in nodose neurons (reviewed by Sah 1996). When intracellular photolabile Ca2+ chelators (“caged” calciums) were used to rapidly increase [Ca2+]i in these neurons, the latency between Ca2+ photorelease and the Ca2+-induced membrane hyperpolarization was no longer than the membrane time constant (Lancaster and Zucker 1994). These results suggest that in CA1 pyramidal neurons Ca2+ directly gates AHPslow K+ channels and that the slow onset kinetics could arise from Ca2+ channels residing on the proximal dendrites, distal to the AHPslow potassium channels that are uniformly distributed in the somata. Acutely isolated nodose neurons are essentially spherical structures lacking dendritic processes. Thus, in these neurons, the delayed onset of the AHPslow is unlikely to be the result of slow diffusion of Ca2+ from distal sources. The high temperature coefficient (Q 10 ≈ 3.5) for the rising phase of the AHPslow in rabbit nodose neurons (Fowler et al. 1985) also argues against simple Ca2+ diffusion (Q 10 ≈ 1.3) as an explanation for the slow kinetics (see also Sah and McLachlan 1991). One possibility is that the CICR system imposes a time dependence on the AHPslow; another may be that opening of the AHPslow potassium channels requires an enzymatic signal transduction cascade. In this regard, it will be useful to learn whether rapid photorelease of Ca2+ in nodose neurons produces an instantaneous or delayed membrane hyperpolarization.

In summary, a CICR pool in nodose neurons releases Ca2+ in response to Ca2+ influx evoked by single APs. One physiological role for CICR in vagal afferents is to activate and control the AHPslow that is responsible for spike frequency adaptation.


The authors thank Dr. Brad Undem for constructive suggestions on an earlier draft of this manuscript. The first two authors contributed equally to this work.

This work was supported by National Institutes of Health Grants NS-22069 to D. Weinreich, GM-46956 to J.P.Y. Kao, and training grant NS-07375 to K. A. Moore.


  • Address for reprint requests: D. Weinreich, Dept. of Pharmacology and Experimental Therapeutics, Rm. 522 Health Science Facility, 685 West Baltimore St., University of Maryland, School of Medicine, Baltimore, MD 21201-1559.

  • 1 Fowler et al. (1985) demonstrated that the AHPslow and its conductance follow an identical time course.


View Abstract