Ca²⁺ influx in resting rat sensory neurones that regulates and is regulated by ryanodine-sensitive Ca²⁺ stores

Abstract

1. Store-operated, voltage-independent Ca²⁺ channels are activated by depletion of intracellular Ca²⁺ stores and mediate Ca²⁺ influx into non-excitable cells at resting membrane potential. We used microfluorimetry, patch-clamp and Mn²⁺-quench techniques to explore the possibility that a similar mechanism exists in rat dorsal root ganglion (DRG) neurones in primary culture.

2. Following caffeine-induced depletion, ryanodine-sensitive Ca²⁺ stores refilled with Ca²⁺ at resting membrane potential. The refilling process required extracellular Ca²⁺, was blocked by 2 mM Ni²⁺, and was facilitated by membrane hyperpolarization from −55 to −80 mV, indicating a key role for Ca²⁺ influx. This influx of Ca²⁺ was not affected by the voltage-operated Ca²⁺ channel (VOCC) antagonists nicardipine (10 μM), nimodipine (10 μm) or ω-grammotoxin SIA (1 μm).

3. When ryanodine-sensitive Ca²⁺ stores were depleted in Ca²⁺-free media, a return to 2 mM external Ca²⁺ resulted in a pronounced [Ca²⁺]_i overshoot, indicating an increased permeability to Ca²⁺. Depletion of Ca²⁺ stores also produced a 2-fold increase in the rate of Mn²⁺ influx. The [Ca²⁺]_i overshoot and Mn²⁺ entry were both inhibited by Ni²⁺, but not by VOCC antagonists.

4. Caffeine induced periodic Ca²⁺ release from, and reuptake into, ryanodine-sensitive stores. The [Ca²⁺]_i oscillations were arrested by removal of extracellular Ca²⁺ or by addition of Ni²⁺, but they were not affected by VOCC antagonists. Hyperpolarization increased the frequency of this rhythmic activity.

5. These data suggest the presence of a Ca²⁺ entry pathway in mammalian sensory neurones that is distinct from VOCCs and is regulated by ryanodine-sensitive Ca²⁺ stores. This pathway participates in refilling intracellular Ca²⁺ stores and maintaining [Ca²⁺]_i oscillations and thus controls the balance between intra- and extracellular Ca²⁺ reservoirs in resting DRG neurones.

In non-excitable cells, capacitative Ca²⁺ influx is the principal route of Ca²⁺ entry (Berridge, 1995; Clementi & Meldolesi, 1996; Parekh & Penner, 1997). Store-operated Ca²⁺ channels (SOCCs) that are activated by depletion of intracellular Ca²⁺ stores mediate this influx. The single channel conductance of SOCCs is estimated to be below 1 pS (Hoth & Penner, 1993). SOCCs are selectively permeable to Ca²⁺ and blocked by a number of inorganic cations (Hoth & Penner, 1993), but no organic antagonist has proven to be selective for these channels (Lo & Thayer, 1995; Clementi & Meldolesi, 1996). SOCCs are voltage independent; thus, SOCC-mediated Ca²⁺ influx persists at resting membrane potential. Hyperpolarization increases Ca²⁺ influx through these channels due to the increased driving force for Ca²⁺ (Parekh & Penner, 1997). In non-excitable cells that lack voltage-gated Ca²⁺ channels, SOCCs are the primary pathway for the Ca²⁺ influx that is needed to replenish intracellular stores, to maintain [Ca²⁺]_i oscillations and to establish long-lasting increases in [Ca²⁺]_i following agonist-induced Ca²⁺ release (Berridge, 1995; Clementi & Meldolesi, 1996).

Voltage-operated Ca²⁺ channels (VOCCs) are the primary pathway for Ca²⁺ entry during excitation in neurones (Lipscombe et al. 1988; Thayer & Miller, 1990). These channels have been studied extensively in a variety of neuronal tissues (Bertolino & Llinás, 1992; Tsien et al. 1995). However, few studies have addressed the question of whether alternative routes of Ca²⁺ influx might contribute to [Ca²⁺]_i homeostasis in neurones at rest. The evidence for Ca²⁺ influx pathways active in quiescent neurones emerges from the observation that the [Ca²⁺]_i baseline is sensitive to the extracellular Ca²⁺ concentration (Lipscombe et al. 1988; Nohmi et al. 1992; Gomez et al. 1995). Furthermore, ryanodine-sensitive stores in sympathetic (Friel & Tsien, 1992), sensory (Usachev et al. 1993) and hippocampal neurones (Garaschuk et al. 1997) replenish spontaneously at resting membrane potential by a process that requires Ca²⁺ influx. In neuroblastoma cells, Ca²⁺ influx was activated by agonist- or thapsigargin-induced depletion of Ca²⁺ stores (Takemura et al. 1991; Mathes & Thompson, 1995). Store depletion-activated Ca²⁺ influx has also been reported for other excitable cells such as phaeochromocytoma (PC12) cells (Clementi et al. 1992; Bennett et al. 1998) and pancreatic β-cells (Miura et al. 1997). Thus, store-operated Ca²⁺ influx may be a common feature of excitable cells.

In this report, we show that a mechanism similar to capacitative Ca²⁺ entry maintains Ca²⁺ influx in mammalian neurones at resting membrane potential. We used indo-1-based microfluorimetry in combination with whole-cell patch-clamp and Mn²⁺-quench techniques to study Ca²⁺ entry into cultured rat dorsal root ganglion (DRG) neurones before and after depletion of intracellular Ca²⁺ stores. We found that Ca²⁺ influx persisted in resting neurones, was not prevented by selective antagonists of voltage-gated Ca²⁺ channels and increased with hyperpolarization. Depletion of Ca²⁺ stores increased Ca²⁺ influx. Ca²⁺ influx was needed to replenish ryanodine-sensitive stores and maintain caffeine-induced [Ca²⁺]_i oscillations. This pathway may regulate cellular Ca²⁺ levels between periods of electrical activity in neurones.

Preliminary reports of this work have been published previously (Usachev & Thayer, 1997; Thayer et al. 1998).

METHODS

Cell culture

Rat DRG neurones were grown in primary culture as described previously (Thayer & Miller, 1990). In brief, 1- to 3-day-old Sprague-Dawley rats were killed by decapitation with sharp scissors, according to a protocol approved by the University of Minnesota Institutional Animal Care and Use Committee (IACUC). DRG were dissected from the thoracic and lumbar segments of the spinal cord and incubated at 37°C in collagenase-dispase (0.8 and 6.4 U ml⁻¹, respectively) for 45 min. Ganglia were dissociated by trituration through a flame-constricted pipette and then plated onto laminin-coated (50 μg ml⁻¹) glass coverslips (25 mm diameter). Cells were grown in Ham's F12 medium supplemented with 5 % heat-inactivated horse serum and 5 % fetal bovine serum, 50 ng ml⁻¹ nerve growth factor (NGF), 4.4 mM glucose, 2 mM L-glutamine, modified Eagle's medium vitamins, and penicillin- streptomycin (100 U ml⁻¹ and 100 μg ml⁻¹, respectively). All cell culture reagents were purchased from Sigma. Cultures were maintained at 37°C in a humidified atmosphere of 5 % CO₂. Cells were used within 2–6 days after plating.

Microfluorimetric measurements

[Ca²⁺]_i was recorded from cultured DRG neurones by using indo-1-based microfluorimetry (Grynkiewicz et al. 1985). The instrumentation has been described in detail previously (Werth et al. 1996). Cells were placed in a flow-through chamber (Thayer & Miller, 1990) (10 s solution exchange) that was mounted on the stage of an inverted epifluorescence microscope equipped with a × 70 objective (Leitz, NA = 1.15). Indo-1 was loaded into the cells by incubation in medium containing 5 μM indo-1 AM and 0.02 % (w/w) Pluronic F-127 for 30 min at room temperature or via the patch pipette when the [Ca²⁺]_i measurements were combined with patch-clamp recordings. The dye was excited at 350 nm (10 nm bandpass) and emission was detected at 405 (20) and 490 (20) nm. Fluorescence was monitored by a pair of photomultiplier tubes (Thorn EMI, Fairfield, NJ, USA) operating in photon-counting mode. The 5 V output signals were then integrated by 8-pole Bessel filters and digitized at 1 or 10 Hz with an analog-to-digital converter (Indec Systems, Sunnyvale, CA, USA). Data were stored and analysed on an IBM-compatible computer.

[Ca²⁺]_i and Mn²⁺ influx were recorded simultaneously using fura-2-based microfluorimetry. The procedure for dye loading and the microscope configuration were similar to that used for indo-1. Fura-2 fluorescence was alternately excited at 360 (10) nm and 380 (10) nm by using a computer-controlled wheel (Sutter Instruments, Inc., Novato, CA, USA). Fluorescence images (510 nm, 40 nm bandpass) were projected onto a cooled charge-coupled device camera (Photometrics, Inc., Tucson, AZ, USA; 384 pixels × 576 pixels binned to 192 pixels × 288 pixels) controlled by an IBM-compatible computer. Image pairs were collected every 6 s and spatially averaged for further analysis.

Changes in the fluorescence of indo-1 (fura-2) were converted to [Ca²⁺]_i using the formula:

where R is the 405 nm/490 nm (360 nm/380 nm) fluorescence intensity ratio (Grynkiewicz et al. 1985). The dissociation constants used were 250 and 224 nM for indo-1 and fura-2, respectively. β is the ratio of fluorescence emitted at 490 nm for indo-1 or excited at 360 nm for fura-2 measured in the absence and presence of Ca²⁺. R_min, R_max and β were determined in intact cells by applying 10 μM ionomycin in Ca²⁺-free buffer (1 mM EGTA) and saturating Ca²⁺ (5 mM Ca²⁺). Values for R_min, R_max and β, respectively, were 0.25, 2.3 and 3.5 for indo-1 and 1.0, 3.5 and 4.2 for fura-2. Mn²⁺ influx was measured as the change in fura-2 fluorescence excited near the isobestic point (360 nm). Background light levels were determined in an area that did not contain a cell.

Data are presented as means ± s.e.m.; n, number of cells tested. P < 0.05 was considered significant.

Electrophysiology

Whole-cell patch-clamp recordings (Hamill et al. 1981) were obtained using a patch-clamp amplifier (PC501; Warner Instrument Corporation, Hamden, CT, USA) and an analog-to-digital converter (Indec Systems). The whole-cell currents were filtered at 1 kHz and sampled every 200 μs, or 100 ms when recorded in combination with measurements of [Ca²⁺]_i. Patch pipettes were pulled from borosilicate glass (Narishige; 2–4 MΩ) on a Sutter Instruments P-87 micropipette puller and filled with the following solution (mM): potassium gluconate, 125; KCl, 10; Mg-ATP, 3; MgCl₂, 1; Hepes, 10; indo-1, 0.1; pH 7.25 with KOH, 290 mosmol kg⁻¹ with sucrose. The extracellular recording solution contained (mM): NaCl, 140; KCl, 5; CaCl₂, 2; MgCl₂, 1; Hepes, 10; glucose, 10; pH 7.35 with NaOH, 310 mosmol kg⁻¹ with sucrose. Ca²⁺-free solution was obtained by substituting 0.1 mM EGTA for Ca²⁺, unless otherwise indicated. To isolate Ca²⁺ currents from other currents, Cs⁺ was substituted for K⁺ in the pipette solution and extracellular Na⁺ and K⁺ were replaced with TEA⁺; 10 mM BAPTA replaced indo-1 in the pipette solution when patch-clamp recordings were not accompanied by [Ca²⁺]_i measurements.

For the ‘perforated’ modification (Horn & Marty, 1988) of the patch-clamp technique the following solution was used in the patch pipettes (mM): KCl, 55; K₂SO₄, 70; MgCl₂, 7; Hepes, 10; pH 7.25 with KOH. Amphotericin B was added to filtered pipette solution from a DMSO stock (2 mg amphotericin B in 40 μl DMSO) to a final concentration of 240 μg ml⁻¹. The solution was kept in the dark and used within 2 h. The access resistance became < 20 MΩ within 5–10 min after seal formation and was further compensated by 30–40 %.

To evoke action potentials in intact neurones, extracellular field stimulation was employed (Werth et al. 1996). Field potentials were generated by passing current between two platinum electrodes via a Grass S44 stimulator and a stimulus isolation unit (Quincy, MA, USA). Trains of 1 ms pulses were delivered at a rate 10 Hz. The stimulus voltage sufficient to elicit a detectable increase in [Ca²⁺]_i from a cell was determined before beginning an experiment, and subsequent stimuli were 20 V over this threshold.

Reagents

Indo-1, indo-1 AM, fura-2 AM and Pluronic F-127 were obtained from Molecular Probes (Eugene, OR, USA). ω-Grammotoxin SIA (GsTX) was a gift from Dr Richard A. Keith at Zeneca Pharmaceuticals. All other reagents were purchased from Sigma.

RESULTS

Ca²⁺ influx is required to refill ryanodine-sensitive Ca²⁺ stores

We have shown previously that ryanodine-sensitive Ca²⁺ stores refill spontaneously in resting DRG neurones (Usachev et al. 1993). However, a significant portion of released Ca²⁺ is extruded immediately from the cell by active transport (Benham et al. 1992; Usachev et al. 1993; Werth et al. 1996) reducing the level of intracellular Ca²⁺ in the cytosol available for subsequent replenishment of the stores. Therefore, an external source of Ca²⁺ is probably involved in the refilling process in resting neurones. We tested this hypothesis by studying the role of Ca²⁺ entry in refilling ryanodine-sensitive stores.

In this series of experiments, [Ca²⁺]_i was monitored in intact DRG neurones loaded with indo-1 AM. Caffeine (5 mM), a modulator of ryanodine receptors, was employed in a paired-pulse protocol in which the first (control) application of caffeine was used to deplete the stores, then after a delay, a second (test) caffeine stimulation was applied to probe the amount of Ca²⁺ that had reaccumulated in the stores (Fig. 1A). During depolarization, Ca²⁺ influx through VOCCs supplies Ca²⁺ to reload intracellular Ca²⁺ stores (Usachev et al. 1993; Garaschuk et al. 1997). In these experiments, a complete replenishment of the Ca²⁺ stores was provided before each pair of caffeine applications by a series of action potentials (10 Hz for 5 s) using extracellular field stimulation. Caffeine was added to Ca²⁺-free media to prevent Ca²⁺ influx during the release activation. The amplitude of the test [Ca²⁺]_i response was normalized to the amplitude of the control response and used as an index of the efficiency of the refilling process. This paired-pulse protocol renders the peak [Ca²⁺]_i dependent on the refilling state of the stores because basal [Ca²⁺]_i was constant (± 20 nM) for a given pair of stimuli and [Ca²⁺]_i buffering processes were much slower (the time from peak to basal [Ca²⁺]_i was 165 ± 14 s) than the upstroke of the [Ca²⁺]_i transient (the time from basal to peak [Ca²⁺]_i was 24 ± 3 s; n = 16), providing a clear temporal separation of release from recovery. The Ca²⁺ stores refilled in a time-dependent manner. For example, the amplitude of the test response after 1 min of refilling was only 24 ± 9 % (n = 6) of control, whereas after 20 min the test response recovered to 86 ± 6 % of control (n = 8). The refilling was completely blocked (n = 4, data not shown) by 1 μM cyclopiazonic acid (CPA), a selective antagonist of sarco-endoplasmic reticulum Ca²⁺-ATPases (SERCAs) (Thomas & Hanley, 1994). This is consistent with the role of SERCAs in actively transporting Ca²⁺ into intracellular stores (Pozzan et al. 1994).

Figure 1. Ca²⁺ influx is required to refill ryanodine-sensitive Ca²⁺ stores.

A, Ca²⁺ release was activated by 5 mM caffeine in Ca²⁺-free extracellular solution, as indicated by the horizontal bars. The paired [Ca²⁺]_i transients are presented in order of increasing interstimulus interval as shown above the [Ca²⁺]_i traces, although they were intermixed in the actual experiment. In B and C, the effect of extracellular Ca²⁺ (B) and Ni²⁺ (C) on the refilling process was studied using a paired-pulse protocol. The interval between two subsequent caffeine applications was 5 min. The horizontal bars indicate the duration of the treatments.

Next, we tested whether replenishment of ryanodine-sensitive stores depended on extracellular Ca²⁺. Removal of extracellular Ca²⁺ during the interval between caffeine applications blocked refilling, as indicated by the failure of the second application of caffeine to evoke a [Ca²⁺]_i response (n = 4; Fig. 1B). The response did not recover if the cell remained in Ca²⁺-free buffer (data not shown), suggesting that Ca²⁺ influx is needed to refill the stores. Furthermore, if the cell was treated with 2 mM Ni²⁺ during the refilling interval, as shown in Fig. 1C, the amplitude of the test [Ca²⁺]_i response decreased from 69 ± 5 to 28 ± 5 % of control (n = 18; P < 0.001, Student's paired t test). The inhibitory effect of Ni²⁺ was fully reversible.

Ca²⁺ influx at resting membrane potential is not mediated by voltage-gated Ca²⁺ channels

Because the refilling of Ca²⁺ stores required Ca²⁺ influx, we tried to determine which channels mediated Ca²⁺ influx in unstimulated neurones. In non-excitable cells this influx is conducted by SOCCs (Berridge, 1995; Parekh & Penner, 1997). In neurones, VOCCs are the principal Ca²⁺ entry pathway during electrical activity (Lipscombe et al. 1988; Thayer & Miller, 1990). However, the contribution of VOCCs to Ca²⁺ entry at resting membrane potential is not clear. We investigated whether Ca²⁺ influx in resting DRG neurones was mediated by VOCCs.

DRG neurones express T-, L-, N- and P/Q-type Ca²⁺ channels (Fox et al. 1987; Rusin & Moises, 1995). We studied the effects of VOCC antagonists on the refilling of Ca²⁺ stores by using a paired-pulse protocol in which the caffeine stimuli were separated by 5 min intervals (Figs 2 and 3). Nimodipine and GsTX were used in combination to block L-type channels and N- and P/Q-type channels, respectively (McCarthy & TanPiengco, 1992; Piser et al. 1995). In these experiments, the normalized amplitude of the test [Ca²⁺]_i response was 66 ± 5 % (n = 8) in untreated cells and was dependent on Ca²⁺ influx, as indicated by sensitivity to Ni²⁺ (2 mM). However, the combination of 10 μM nimodipine and 1 μM GsTX did not affect refilling of the stores (Fig. 2A) and the normalized amplitude of the test response after this treatment was 71 ± 6 % (n = 8). The same combination of drugs completely blocked the [Ca²⁺]_i response elicited by depolarization with 40 mM K⁺ (n = 5; Fig. 2B) and inhibited by 94 ± 3 % (n = 6) high-threshold voltage-gated Ca²⁺ currents elicited by step depolarization from −60 to +10 mV (Fig. 2C).

Figure 2. Refilling of ryanodine-sensitive Ca²⁺ stores is not dependent on high-threshold voltage-gated Ca²⁺ channels.

A, [Ca²⁺]_i transients were elicited by 5 mM caffeine in Ca²⁺-free media. The cells were treated with a mixture of 10 μM nimodipine (Nim) and 1 μM ω-grammotoxin SIA (GsTX) during the 5 min period between two subsequent applications of caffeine. The duration of the drug application is indicated by the horizontal bars below the [Ca²⁺]_i traces. B, [Ca²⁺]_i transients were elicited by 40 mM K⁺. The combination of 10 μM nimodipine and 1 μM GsTX completely blocked depolarization-induced increases in [Ca²⁺]_i. The break in the [Ca²⁺]_i trace corresponds to 20 min. C, whole-cell Ca²⁺ currents were evoked by depolarization from −60 to +10 mV for 100 ms. Voltage pulses were applied every 30 s. The amplitude of the Ca²⁺ current is plotted versus time. The traces (right) show currents for the same experiments obtained before treatment (control) and after block with a combination of 10 μM nimodipine and 1 μM GsTX.

Figure 3. Low-threshold voltage-gated Ca²⁺ channels do not mediate refilling of ryanodine-sensitive Ca²⁺ stores.

A, [Ca²⁺]_i transients were elicited by 5 mM caffeine in Ca²⁺-free media. Nicardipine (10 μM; Nic) was applied during the 5 min period between the two caffeine applications as indicated by the filled horizontal bar. B, whole-cell Ca²⁺ currents were evoked by depolarization from −90 to −30 mV for 200 ms. Voltage pulses were applied every 30 s. The amplitude of the Ca²⁺ current is plotted versus time. The current traces (right) represent Ca²⁺ currents before (control) and during treatment of the cell with 10 μM nicardipine.

We next examined the contribution of T-type Ca²⁺ channels to the refilling of ryanodine-sensitive Ca²⁺ stores. We found that amiloride (500 μM), which is often used to block low-threshold voltage-gated Ca²⁺ channels, strongly interfered with indo-1 fluorescence (n = 4; data not shown) and therefore could not be used for [Ca²⁺]_i measurements. Certain dihydropyridines have been reported to block T-type channels at high concentrations (Akaike et al. 1989). In DRG neurones, nicardipine was shown to be effective (Richard et al. 1991). In our system, 10 μM nicardipine inhibited by 97 ± 3 % (n = 5) low-threshold voltage-gated currents elicited by step depolarization from −90 to −30 mV (Fig. 3B). The effect of nicardipine reversed slowly with time. For comparison, 500 μM amiloride and 100 μM Ni²⁺ reduced low-threshold Ca²⁺ currents by 64 ± 3 % (n = 4) and 88 ± 3 % (n = 5), respectively (data not shown). However, 10 μM nicardipine did not slow the refilling process (Fig. 3A). In these experiments, the normalized amplitude of the test [Ca²⁺]_i response was 71 ± 13 % (n = 4) in untreated cells and 77 ± 11 % (n = 4) after treatment with 10 μM nicardipine.

Voltage dependence can reveal important information about the nature of ionic fluxes. Hyperpolarization would be predicted to decrease Ca²⁺ influx mediated by VOCCs due to decreased channel activation (Fox et al. 1987; Magee et al. 1996). In contrast, SOCC-mediated Ca²⁺ influx would be greater at more negative potentials because the driving force for Ca²⁺ would increase without a decrease in channel opening (Parekh & Penner, 1997). The effects of changes in membrane potential during the refilling process are described in Fig. 4. The paired-pulse protocol was applied to neurones clamped at −55 or −80 mV using the whole-cell configuration of the patch-clamp technique. Because the stability of the electrical recording was impaired in Ca²⁺-free media (0.1 mM EGTA), extracellular Ca²⁺ was not removed during caffeine applications in these experiments. When cells were clamped at −55 mV during the interval between two caffeine applications, the normalized amplitude of the test [Ca²⁺]_i response was 65 ± 6 % of the control (n = 8). Hyperpolarization to −80 mV significantly facilitated filling of the Ca²⁺ stores and the amplitude of the test response increased to 90 ± 7 % (n = 8; Fig. 4B). This facilitation was probably a result of enhanced Ca²⁺ influx. Indeed, 40 s hyperpolarization from −55 to −80 mV following depletion of the stores with 10 mM caffeine produced a small elevation in [Ca²⁺]_i that could be detected in the presence of 5 mM Ca²⁺ in the extracellular solution (Fig. 4C). Ni²⁺ (2 mM) inhibited this elevation, reducing the amplitude of the response from 44 ± 11 to 15 ± 2 nM (n = 5). Similar results were obtained using the amphotericin B-based patch-clamp technique (n = 4; data not shown). These data demonstrate that in DRG neurones significant Ca²⁺ influx occurs at resting membrane potential that is independent of VOCCs. We next explored the possibility that SOCCs might mediate this Ca²⁺ influx. If SOCCs are present in these cells then Ca²⁺ influx might be increased by depletion of intracellular Ca²⁺ stores.

Figure 4. Hyperpolarization increases Ca²⁺ influx and facilitates the refilling process.

DRG neurones were loaded with 100 μM indo-1 using the patch pipette. A, [Ca²⁺]_i was measured in neurones clamped to −55 or −80 mV following stimulation with 5 mM caffeine, as indicated below the [Ca²⁺]_i traces. For this paired-pulse protocol, the interstimulus time was 5 min. B, histogram displaying the amplitude of the second [Ca²⁺]_i response normalized to the first for a paired-pulse protocol such as that described in A (n = 8). The membrane potential in the interval between the two subsequent caffeine applications is indicated below the bars. ** P < 0.01, Student's paired t test. C, hyperpolarization induced an elevation in [Ca²⁺]_i that was inhibited by 2 mM Ni²⁺. In both instances, hyperpolarization was applied after the stores were depleted with 10 mM caffeine. The extracellular Ca²⁺ concentration was increased to 5 mM.

Depletion of intracellular Ca²⁺ stores facilitates Ca²⁺ influx in resting neurones

In non-excitable cells, store-operated Ca²⁺ influx is commonly exhibited as a sustained plateau phase following agonist-evoked Ca²⁺ release from intracellular stores. This plateau phase depends on extracellular Ca²⁺ and is thought to be mediated by store-operated Ca²⁺ channels (Jacob, 1990; Parekh & Penner, 1997). We found that in DRG neurones a rapid caffeine-induced [Ca²⁺]_i rise was followed by a sustained [Ca²⁺]_i elevation above the baseline (Fig. 5A). The amplitude of the initial [Ca²⁺]_i increase did not depend on extracellular Ca²⁺ (172 ± 16 nM, n = 10, with 2 mM Ca²⁺ in the media, and 186 ± 16 nM, n = 10, in Ca²⁺-free media) consistent with Ca²⁺ release from intracellular stores. In contrast, the plateau phase could be abolished by removal of extracellular Ca²⁺ (n = 10), suggesting that it was mediated by Ca²⁺ influx. In the absence of extracellular Ca²⁺, caffeine-induced [Ca²⁺]_i transients recovered completely to the baseline within 3 min, whereas in control (2 mM Ca²⁺ in the media) [Ca²⁺]_i was elevated 47 ± 13 nM (n = 10) and 9 ± 3 nM (n = 10) above the baseline when measured 3 and 8 min, respectively, after the beginning of the response (Fig. 5B). Caffeine quenches indo-1 fluorescence, although this effect is wavelength independent and does not influence the ratiometric measurements (O'Neill et al. 1990). Thus, the observed plateau phase was unlikely to result from an indo-1-caffeine interaction. Furthermore, a similar sustained [Ca²⁺]_i elevation was observed with another Ca²⁺-sensitive dye, fura-2 (n = 6, data not shown), the fluorescence of which increases in the presence of caffeine (Nohmi et al. 1992). The plateau phase displayed a graded inactivation. This may result from Ca²⁺-dependent inhibition of Ca²⁺ influx (Zweifach & Lewis, 1995; Madge et al. 1997), inactivation of ryanodine receptor-mediated Ca²⁺ release with subsequent store refilling (Pozzan et al. 1994), or an upregulation of Ca²⁺ efflux processes (Miller, 1991). Depletion of Ca²⁺ stores with CPA also produced a sustained elevation in [Ca²⁺]_i; it decreased at a slower rate (plateau amplitude at 8 min = 63 ± 15 nM; n = 7), suggesting that inactivation of Ca²⁺ release contributed to the decline in [Ca²⁺]_i during the plateau phase.

Figure 5. Caffeine-induced [Ca²⁺]_i response exhibits a Ca²⁺ influx-dependent plateau phase.

A, [Ca²⁺]_i was measured in indo-1 AM-loaded DRG neurones during caffeine (10 mM) application. Representative [Ca²⁺]_i traces obtained in the presence of 2 mM extracellular Ca²⁺ (2 mM Ca²⁺) or in Ca²⁺-free media were superimposed. B, histogram displaying the difference between [Ca²⁺]_i and the baseline measured at the peak of the [Ca²⁺]_i response, and at 3 and 8 min after the beginning of the response (n = 10). The baseline was measured before drug application. Data obtained in the presence or absence of extracellular Ca²⁺ were compared using Student's paired t test (*** P < 0.001). Negative values correspond to [Ca²⁺]_i levels below baseline.

Store-operated Ca²⁺ influx in non-excitable cells can also be observed as a pronounced [Ca²⁺]_i increase elicited by switching from Ca²⁺-free to Ca²⁺-containing media following depletion of intracellular Ca²⁺ stores (Clementi et al. 1992; Berridge, 1995; Bennett et al. 1998). Depletion of the stores opens plasmalemmal channels that allow Ca²⁺ to rush into the cell upon its return to the media. We found that readdition of external Ca²⁺ to the bath after caffeine-induced discharge of the stores resulted in a [Ca²⁺]_i overshoot in 68 % of DRG neurones tested (n = 47; Fig. 6), which is consistent with the presence of store-operated Ca²⁺ entry in neurones. During this transient elevation, [Ca²⁺]_i peaked at 37 ± 6 nM above resting [Ca²⁺]_i (n = 28) and the maximal rate of [Ca²⁺]_i rise was 4 ± 1 nM s⁻¹ (n = 28). [Ca²⁺]_i then recovered to the basal level. The recovery could be fitted by a monoexponential function with a time constant of 123 ± 7 s (n = 28). Ni²⁺ (2 mM), but not the selective VOCC antagonists, inhibited the overshoot by 62 ± 6 % (n = 8; Fig. 6C, see also Figs 1C, 2A and 3A). The overshoot was not observed if the switch from Ca²⁺-free to Ca²⁺-containing media (2 mM Ca²⁺) was not preceded by depletion of the stores (n = 6; Fig. 6A).

Figure 6. Depletion of intracellular Ca²⁺ stores increases Ca²⁺ influx.

[Ca²⁺]_i was measured in intact neurones loaded with indo-1 AM. A, a [Ca²⁺]_i overshoot was observed after depletion of ryanodine-sensitive stores with caffeine. The breaks between the traces correspond to 15 min during which the cell was stimulated with 3 trains of action potentials (4 s at 10 Hz) elicited by extracellular field stimulation. B, for the recordings shown in A, changes in [Ca²⁺]_i evoked by addition of 2 mM Ca²⁺ following store depletion with caffeine (empty) or with ‘full’ stores are superimposed. The caffeine-elicited [Ca²⁺]_i transient is indicated (Caff). In C, the [Ca²⁺]_i overshoots elicited by store depletion in the absence (control) or presence (Ni²⁺) of 2 mM Ni²⁺ are compared. Application of 5 mM caffeine is indicated on the trace (Caff). D, a [Ca²⁺]_i overshoot was elicited by depletion of the stores with 5 mM caffeine or 5 μM cyclopiazonic acid (CPA), but not by 0.5 mM 3-isobutyl-1-methylxanthine (IBMX). Horizontal bars below the trace indicate the duration of treatment.

In addition to its action on ryanodine receptors, caffeine is also known to inhibit phosphodiesterases and will thus increase the intracellular concentrations of cyclic AMP and cyclic GMP (Daly, 1993). These cyclic nucleotides may affect the Ca²⁺ permeability of the plasma membrane (Clementi & Meldolesi, 1996). To address the question of whether the action of caffeine resulted from an increase in cyclic nucleotides, we treated cells with 0.5 mM 3-isobutyl-1-methylxanthine (IBMX), which is an approximately 30-fold more potent inhibitor of phosphodiesterases than caffeine (Daly, 1993), but will not release Ca²⁺ from stores at this concentration (Usachev & Verkhratsky, 1995). As shown in Fig. 6D, treatment with 0.5 mM IBMX did not produce a [Ca²⁺]_i overshoot (n = 3). Furthermore, the induction of the overshoot did not depend on the delay between removal of caffeine and addition of Ca²⁺ to the external media. Thus, the overshoot resulted from depletion of Ca²⁺ stores and subsequent activation of Ca²⁺ influx. An alternative means of depleting Ca²⁺ stores is to inhibit the SERCAs with agents such as CPA (Thomas & Hanley, 1994). [Ca²⁺]_i transients elicited by CPA (5 μM) developed more slowly and had a smaller amplitude than those induced by caffeine (Figs 6D and 9A). To ensure an effective discharge of the stores we applied CPA for 4–8 min (compare with 2–3 min for caffeine). After treatment with 5 μM CPA in Ca²⁺-free media, addition of 2 mM Ca²⁺ elicited a pronounced overshoot in [Ca²⁺]_i in five of six neurones tested (Fig. 6D). The amplitude of the overshoot relative to the [Ca²⁺]_i baseline was 75 ± 23 nM (n = 5).

Figure 9. Action potential-induced [Ca²⁺]_i transients are modulated by the level of Ca²⁺ in ryanodine-sensitive Ca²⁺ stores.

A, [Ca²⁺]_i transients were elicited by extracellular field stimulation (10 Hz, 4 s) in intact neurones loaded with indo-1 AM, as indicated by triangles above the [Ca²⁺]_i trace. Horizontal bars indicate application of 5 mM caffeine or 5 μM CPA. A comparison of corresponding [Ca²⁺]_i transients before (1) and after (2) caffeine application, or in the presence of CPA (3), is shown in the inset. The traces were offset along the [Ca²⁺]_i axes. In B and C, the changes in time constant that characterize the [Ca²⁺]_i recovery process (B) and the amplitude of the [Ca²⁺]_i elevation (C) for the experimental protocol described in A are analysed for 5 cells. Each point represents the mean ± s.e.m. of the corresponding parameter for the action potential-induced [Ca²⁺]_i transients. A monoexponential function was fitted to the [Ca²⁺]_i recovery process and the corresponding time constant was calculated by using a non-linear, least-squares curve fitting algorithm (Origin software). The data are plotted versus time, which was set to zero during the caffeine treatment (vertical dotted line).

These data suggest that the overshoot reflects facilitation of Ca²⁺ influx after depletion of Ca²⁺ stores that, in turn, should accelerate replenishment of the stores. We compared the rate of refilling of the Ca²⁺ stores in cells that displayed an overshoot (Fig. 7A) with that in cells in which the overshoot could not be detected (Fig. 7B). As summarized in Fig. 7C and D, the replenishment occurred significantly faster in cells for which Ca²⁺ influx was facilitated by store depletion. Using a paired-pulse protocol (Fig. 1A), we found that the refilling process could be described by a monoexponential function with a time constant of 137 s for cells with the overshoot and 414 s for those without it (Fig. 7C). Furthermore, the amplitudes of the caffeine-induced [Ca²⁺]_i response after 5 min of refilling recovered to 85 ± 4 % (n = 21) and 36 ± 8 % (n = 12) for the cells with and without overshoot, respectively (Fig. 7D).

Figure 7. Ryanodine-sensitive stores refill faster in neurones that exhibit a [Ca²⁺]_i overshoot.

A and B, [Ca²⁺]_i transients were elicited by 5 mM caffeine in DRG neurones with (A) or without (B) an overshoot, to test the rate of refilling of the stores using a paired-pulse protocol. The interstimulus time was 5 min. In C, the time dependence of the refilling process was studied using a paired-pulse protocol in neurones with (•; n = 4) or without (▴; n = 5) a [Ca²⁺]_i overshoot. An overshoot was defined as a net [Ca²⁺]_i increase greater than 10 nM that recovered to the basal level within 5–10 min. Ryanodine-sensitive Ca²⁺ stores were depleted by the first (control) application of 5 mM caffeine in Ca²⁺-free media. The level of replenishment of the Ca²⁺ stores at a given time was evaluated by applying a second (test) stimulus with 5 mM caffeine in Ca²⁺-free media. Each point represents the mean ± s.e.m. of the test response normalized to the control for various interstimulus time intervals. Data points were fitted with a single exponential function (smooth curves) using a non-linear, least-squares curve fitting algorithm (Origin software, Microcal). In D, the normalized amplitudes of the test response for neurones with (n = 21; overshoot) and without (n = 12; not detected) a [Ca²⁺]_i overshoot are compared for experiments such as those described in A and B. *** P < 0.001; Student's unpaired t test.

Mn²⁺ permeates many Ca²⁺ pathways and, because it quenches the fluorescence of the Ca²⁺ indicator fura-2, it can be used as a probe for divalent cation entry (Jacob, 1990). We measured Mn²⁺ influx in intact neurones as a decrease in fura-2 fluorescence excited near the isobestic point for Ca²⁺ (360 nm). [Ca²⁺]_i was monitored simultaneously by taking the ratio of the fluorescence signals excited at 360 and 380 nm (F₃₆₀/F₃₈₀). Figure 8A shows an example of such a recording. In untreated cells, F₃₆₀ decreased slowly with time (0.07 ± 0.01 arbitrary units per second (a.u. s⁻¹); n = 15) presumably as a result of fura-2 bleaching. To maximize Ca²⁺ influx, Ca²⁺ stores were depleted by application of 5 mM caffeine in Ca²⁺-free buffer. Addition of 300 μM Mn²⁺ increased significantly the rate of decay of F₃₆₀ (0.39 ± 0.04 a.u. s⁻¹; n = 11). A mixture of the VOCC antagonists GsTX (1 μM) and nimodipine (10 μM) did not change the Mn²⁺ influx, whereas it was inhibited by 2 mM Ni²⁺ (Fig. 8A); the F₃₆₀ decay rate was slowed to 0.12 ± 0.02 a.u. s⁻¹ (n = 11). Nicardipine (10 μM) was also without effect (n = 4; data not shown). These data are consistent with our previous observations (Figs 2A, 3A and 6C) and suggest that Mn²⁺ can be used as a probe for Ca²⁺ influx in neurones. Next we compared the rate of Mn²⁺ influx before and after depletion of Ca²⁺ stores by 5 mM caffeine (Fig. 8B and C) or 10 μM CPA (Fig. 8D and E). Both treatments significantly increased the F₃₆₀ decay rate (P < 0.01 and < 0.05, respectively; Student's paired t test). After correcting for background decay of the F₃₆₀ signal caused by fura-2 bleaching, we found that depletion of the stores with 5 mM caffeine or 10 μM CPA increased the rate of Mn²⁺ influx by 2.3 ± 0.5-fold (n = 6) and 2.0 ± 0.4-fold (n = 5), respectively.

Figure 8. Depletion of Ca²⁺ stores increases the rate of Mn²⁺ influx in DRG neurones.

[Ca²⁺]_i and Mn²⁺ (300 μM extracellular concentration) influx were measured simultaneously in fura-2 AM-loaded DRG neurones. Mn²⁺ influx was detected as a quench of 360 nm fluorescence (F₃₆₀) shown in arbitrary units (a.u.). A, to maximize Mn²⁺ influx, Ca²⁺ stores were depleted with 5 mM caffeine (Caff) in Ca²⁺-free media. Mn²⁺ influx was slowed significantly by 2 mM Ni²⁺ but not by combined application of the VOCC antagonists GsTX (1 μM) and nimodipine (10 μM; Nim). Horizontal bars above the traces indicate the duration of the drug treatments. B-E, depletion of Ca²⁺ stores with 5 mM caffeine (B and C) or 10 μM CPA (D and E) increased the influx of Mn²⁺. The corresponding traces before (control) and after (C, Caff; E, CPA) depletion of the stores were offset along the fluorescence intensity axis and compared in C and E.

The state of ryanodine-sensitive stores controls the size and the shape of [Ca²⁺]_i transients elicited by trains of action potentials

We have shown that store-regulated Ca²⁺ influx is the principal source for refilling ryanodine-sensitive Ca²⁺ stores at the resting membrane potential in rat DRG neurones. When Ca²⁺ influx was blocked, intracellular Ca²⁺ stores remained empty. Ca²⁺ stores can regulate [Ca²⁺]_i responses evoked by K⁺ application in neurones (Friel & Tsien, 1992; Usachev et al. 1993). We compared Ca²⁺ signals in response to physiological stimuli for full and empty stores (Fig. 9A). Trains of action potentials were elicited every 60 or 120 s (indicated by the triangles) by extracellular field stimulation (10 Hz) to produce [Ca²⁺]_i transients in intact cells. For a given cell, the intensity of the stimulus remained constant during the recording. The amplitude (Δ[Ca²⁺]_i) and the time constant for the monoexponential function that described the recovery process (τ) for each [Ca²⁺]_i transient were analysed. Depletion of the stores by 5 mM caffeine significantly reduced the time constant from 12 ± 3 to 5 ± 1 s (n = 5; Fig. 9B) and it decreased the amplitude of the response from 181 ± 20 to 89 ± 12 nM (n = 5; Fig. 9C). This was probably the consequence of the increased capability of the emptied stores to buffer cytosolic Ca²⁺. Both parameters slowly recovered as the stores recharged during subsequent stimulation. Inhibition of the SERCAs with 5 μM CPA excluded Ca²⁺ stores from the buffering process and the resulting [Ca²⁺]_i transients increased in amplitude to 167 ± 21 nM (n = 5) and recovery slowed more than 3-fold (τ = 28 ± 5 s; n = 5) relative to the last [Ca²⁺]_i transient before CPA application (Δ[Ca²⁺]_i = 125 ± 11 nM and τ = 9 ± 2 s; n = 5). The effect of CPA reversed completely in 5–8 min (Fig. 9). This is the first demonstration that intracellular stores contribute significantly to the amplitude and recovery kinetics of action potential-elicited [Ca²⁺]_i transients. These findings corroborate previous observations (Friel & Tsien, 1992; Usachev et al. 1993; Toescu, 1998) that depletion of ryanodine-sensitive Ca²⁺ stores enhances Ca²⁺ uptake into the stores.

Caffeine-induced [Ca²⁺]_i oscillations at resting membrane potential are controlled by store-regulated Ca²⁺ influx

We found that in 37 % (n = 134) of large (28–34 μm) DRG neurones 5 mM caffeine initiated [Ca²⁺]_i oscillations (Fig. 10). These oscillations resulted from periodic Ca²⁺ release from and reuptake into ryanodine-sensitive stores (Lipscombe et al. 1988; Nohmi et al. 1992). The oscillations were maintained in neurones voltage clamped at −55 mV. However, they required a persistent Ca²⁺ influx, because addition of 2 mM Ni²⁺ to the buffer (Fig. 10A) or removal of extracellular Ca²⁺ (Fig. 10B) abolished the oscillations. The influx was not mediated by VOCCs because nimodipine (10 μM) and GsTX (1 μM) were without effect (n = 4). As shown in Fig. 10B-D, hyperpolarization from −55 to −80 mV increased significantly the frequency of oscillations from 0.6 ± 0.3 to 0.8 ± 0.3 min⁻¹ (n = 5). This probably resulted from an increase in Ca²⁺ influx upon hyperpolarization and faster recharge of the Ca²⁺ stores, which subsequently accelerated the onset of the next [Ca²⁺]_i spike.

Figure 10. [Ca²⁺]_i oscillations are modulated by store-regulated Ca²⁺ influx in DRG neurones.

A-C, [Ca²⁺]_i oscillations were induced by 5 mM caffeine in neurones clamped at −55 or −80 mV as indicated. The cells were loaded with 100 μM indo-1 through the patch pipette. Removal of extracellular Ca²⁺, treatment with a combination of 1 μM GsTX and 10 μM nimodipine (GsTX + Nim) and application of 2 mM Ni²⁺ are indicated by horizontal bars below the [Ca²⁺]_i traces. In D, the frequency of [Ca²⁺]_i oscillations for membrane holding potentials of −55 and −80 mV are compared for 5 cells in experiments similar to those described in B and C. * P < 0.05; Student's paired t test.

DISCUSSION

In this report, we describe a Ca²⁺ influx pathway present in quiescent DRG neurones. The influx was not mediated by VOCCs, but rather shared certain features with capacitative Ca²⁺ influx. Hyperpolarization and depletion of ryanodine-sensitive Ca²⁺ stores enhanced the influx. This modest but persistent Ca²⁺ influx appeared to regulate the rate at which Ca²⁺ stores refilled with Ca²⁺, and produced marked effects on store-regulated processes such as Ca²⁺ buffering and [Ca²⁺]_i oscillations.

Routes of Ca²⁺ entry in resting neurones

Neurones express multiple types of VOCCs (Bertolino & Llinás, 1992; Tsien et al. 1995). Upon excitation, Ca²⁺ enters the cell through these channels to initiate a number of processes including neurotransmitter release and transcription (Ghosh & Greenberg, 1995). This depolarization-induced Ca²⁺ influx also potentiates replenishment of intracellular Ca²⁺ stores, and thus depolarization is commonly used to refill Ca²⁺ stores in neurones (Usachev et al. 1993; Garaschuk et al. 1997). However, at the resting membrane potential, we found that VOCCs contributed little to Ca²⁺ influx or to refilling of Ca²⁺ stores in rat DRG neurones. Ca²⁺ influx was the rate-limiting step in the refilling of Ca²⁺ stores as indicated by the requirement for extracellular Ca²⁺, its block by Ni²⁺ and the enhanced refilling produced by hyperpolarization. In spite of this exquisite sensitivity to factors that influence Ca²⁺ influx, pharmacological block of VOCCs with dihydropyridine drugs and Ca²⁺ channel toxins had no effect on the refilling of the stores, resting Ca²⁺ influx, or Mn²⁺ influx. These drugs were clearly effective under our recording conditions (Figs 2B and C and 3B). The action of dihydropyridines is voltage dependent; depolarization enhances their inhibitory effect on L-type channels. However, at concentrations above 100 nM these drugs are effective even at hyperpolarized membrane potentials (McCarthy & TanPiengco, 1992). Therefore, it is unlikely that a lack of effect of 10 μM nimodipine on the refilling process could be explained by its voltage dependence. In fact, short pre-treatment of DRG neurones with 10 μM nimodipine at the resting membrane potential completely abolished high-K⁺-induced Ca²⁺ flux through L-type channels (Fig. 2B). Our observations are in agreement with the observation that dihydropyridines did not affect Ca²⁺ influx at rest in a neuronal cell line (Takemura et al. 1991), sympathetic (Lampe et al. 1995) or hippocampal neurones (Garaschuk et al. 1997) and are in contrast to another study that described a steady-state current mediated by dihydropyridine-sensitive channels in central neurones (Magee et al. 1996). This discrepancy might result from differences in dihydropyridine-sensitive (L-type) channels expressed in brain relative to peripheral tissues (Tsien et al. 1995) or the different methods used in these studies. The fact that VOCCs do not contribute to Ca²⁺ influx in DRG neurones at rest is in agreement with the voltage dependence of activation and inactivation for VOCCs in these cells (Fox et al. 1987). Despite the narrow ‘window current’ for T-type channels near the resting membrane potential (between −70 and −60 mV), their involvement is unlikely because hyperpolarization to −80 mV did not reduce, but instead enhanced, Ca²⁺ influx in resting DRG neurones (Figs 4 and 10). Similarly, the Na⁺-Ca²⁺ exchanger, which could potentially carry Ca²⁺ into the cell in reverse mode (Miller, 1991), can also be excluded from consideration because hyperpolarization would favour Ca²⁺ extrusion from the cell. Thus, in this study we describe a novel route of Ca²⁺ influx into resting DRG neurones.

The Ca²⁺ influx described here shares a number of features with SOCCs found in non-excitable cells (Berridge, 1995; Parekh & Penner, 1997). Ca²⁺ influx was: (1) enhanced by Ca²⁺ store depletion, (2) enhanced by hyperpolarization, (3) blocked by Ni²⁺, and (4) not sensitive to selective VOCC antagonists. The idea that SOCCs might participate in neuronal function is consistent with the high level of expression of mammalian homologues of the Drosophila trp gene in brain; TRP channels are believed to mediate capacitative Ca²⁺ entry (Mori et al. 1998; Philipp et al. 1998). Ca²⁺ influx induced by store depletion was most easily detected as a sustained [Ca²⁺]_i plateau phase during the caffeine-induced response (Fig. 5), and as a [Ca²⁺]_i overshoot following the switch from Ca²⁺-free to 2 mM Ca²⁺-containing media (Fig. 6). In the latter case, a similar result might occur if Ca²⁺ buffering mechanisms were downregulated during the Ca²⁺-free treatment, but this was not the case because the overshoot was dependent on depletion of the store by caffeine (Fig. 6), a treatment that actually enhanced buffering of electrically induced [Ca²⁺]_i increases (Fig. 9). Furthermore, the influx of Mn²⁺, a cation that permeates SOCCs (Jacob, 1990), was enhanced by approximately 2-fold after depletion of the stores. The recovery of the overshoot was approximately 10-fold slower than that of electrically induced [Ca²⁺]_i transients suggesting that termination of the overshoot was determined by inactivation of Ca²⁺ influx rather than by Ca²⁺ buffering processes. Once activation of SOCCs is initiated by depletion of the stores, these channels inactivate as the consequence of refilling of the store (Jacob, 1990; Zweifach & Lewis, 1995). In DRG neurones, the time course for refilling the stores with Ca²⁺ was characterized by virtually the same kinetics (τ = 137 s) as the recovery from the overshoot (τ = 123 s), suggesting that inactivation of Ca²⁺ influx in these cells was also controlled by replenishment of intracellular Ca²⁺ stores.

Store-operated Ca²⁺ influx in non-excitable cells has been mainly attributed to depletion of inositol 1,4,5-trisphosphate (IP₃)-sensitive Ca²⁺ stores (Parekh & Penner, 1997). Here we demonstrate a similar link between depletion of ryanodine-sensitive Ca²⁺ stores and Ca²⁺ influx. Both types of store are associated with the endoplasmic reticulum and there is a high level of structural and functional similarity between IP₃ and ryanodine receptors (Pozzan et al. 1994). The fact that Ca²⁺ influx can be activated by depletion of Ca²⁺ stores with SERCA inhibitors (Thomas & Hanley, 1994) indicates that formation of IP₃ is not required for this activation. Therefore, it is plausible that depletion of ryanodine-sensitive stores might initiate mechanisms similar to those proposed for the IP₃-sensitive stores (Berridge, 1995; Parekh & Penner, 1997). Indeed, activation of Ca²⁺ influx by depletion of ryanodine-sensitive Ca²⁺ stores has been reported recently in PC12 cells (Clementi et al. 1992; Bennett et al. 1998). This influx was similar to that evoked by IP₃ or the SERCA inhibitor thapsigargin in the same cells, suggesting that different intracellular Ca²⁺ stores may utilize common pathways to induce capacitative Ca²⁺ entry.

Physiological role of the store-regulated Ca²⁺ influx in neurones

While voltage-gated Ca²⁺ channels are the major pathway for Ca²⁺ influx during excitation in neurones, the novel mechanism of Ca²⁺ influx described in this study may be important between periods of electrical activity. We estimate that the electrical current corresponding to this influx, based on the maximal rate of [Ca²⁺]_i rise during the [Ca²⁺]_i overshoot (Fig. 6), is approximately 3 orders of magnitude smaller than the Ca²⁺ current activated during an action potential (Fox et al. 1987; Thayer & Miller, 1990). However, because this influx persists at resting membrane potential, in 1–2 s it carries as much Ca²⁺ as that carried during an action potential. The sustained Ca²⁺ influx elicited by store depletion in sympathetic neurones protected these cells from apoptosis induced by nerve growth factor deprivation (Lampe et al. 1995).

The Ca²⁺ influx described here served to replenish intracellular Ca²⁺ stores. The Ca²⁺ level within the stores is important for several reasons. The stores participate in removal of Ca²⁺ from the cytosol during cell excitation. We found that the efficiency of this transport was significantly increased when the stores were depleted. Depletion of the stores with 5 mM caffeine resulted in a more than 2-fold decrease in the amplitude and duration of electrically induced [Ca²⁺]_i responses (Fig. 9), both of which recovered as the stores refilled. Caffeine (10 mM) reduced Ca²⁺ currents by approximately 25 % in sympathetic neurones, although the current recovered within a few seconds after removal of the drug (Lipscombe et al. 1988). An effect of caffeine on VOCCs was unlikely to influence the experiments described here because the first post-caffeine electrical stimulation was applied 60 s after the drug was completely washed from the bath (Fig. 9A), as confirmed by recovery from the caffeine-induced quench of indo-1 fluorescence (O'Neill et al. 1990). Thus, our observations indicate that the Ca²⁺ level within the store determines the size and duration of the [Ca²⁺]_i transients elicited by a given pattern of electrical activity, which in turn would control triggering of various Ca²⁺-dependent cell functions (Ghosh & Greenberg, 1995). Chronic depletion of Ca²⁺ stores has been shown to disturb gene expression by inhibiting molecular transport across the nuclear envelope (Perez-Terzik et al. 1997) and to block protein synthesis (Thomas & Hanley, 1994).

Ca²⁺ influx was required to maintain caffeine-induced [Ca²⁺]_i oscillations at resting membrane potential in neurones (Fig. 10). Store-regulated Ca²⁺ entry enabled changes in membrane potential to tune the frequency of [Ca²⁺]_i oscillations. The exact role of [Ca²⁺]_i oscillations in neurones is not clear, although some observations suggest that certain patterns of periodic [Ca²⁺]_i spikes are important for regulation of gene expression (Itoh et al. 1995) and cell growth and differentiation (Gu & Spitzer, 1995; Gomez et al. 1995).

We have shown that mammalian sensory neurones possess a distinct Ca²⁺ entry pathway that is active at rest and enhanced by depletion of Ca²⁺ stores. This observation is consistent with studies in non-excitable cells that describe specialized channels that are somehow activated by store depletion and play an important role in refilling the store with Ca²⁺ and maintaining sustained Ca²⁺ influx. Why neurones, with their extensive complement of voltage- and receptor-operated Ca²⁺ channels, need yet another Ca²⁺ entry pathway is not entirely clear. However, the pathway is present and appears to maintain the Ca²⁺ balance between intracellular and extracellular pools in DRG neurones at resting membrane potential.