Introduction

Top Summary Introduction Procedures Results Discussion References

Products of sphingolipid metabolism can modulate various cellular functions, such as cell growth and DNA synthesis, as well as programmed cell death (for review see Spiegel et al., 1996). An intracellular second messenger role of sphingolipids, particularly of SPP, for cellular Ca²⁺ homeostasis is supported by the demonstration that SPP can mobilize Ca²⁺ from intracellular stores (Ghosh et al., 1994; Mattie et al., 1996) and by the finding that cellular SPP levels increase on activation of platelet-derived growth factor and FcRI antigen receptors (Olivera and Spiegel, 1993; Choi et al., 1996).

In addition to their role as intracellular messengers, sphingolipid compounds apparently bind to and activate PTX-sensitive G protein-coupled plasma membrane receptors in a variety of cell types. The strongest evidence for the existence of a specific sphingolipid receptor emerged from studies on muscarinic K⁺ currents in guinea pig atrial myocytes. Both SPP and SPPC, at similar nanomolar concentrations, activated K⁺ currents in a strictly GTP- and PTX-sensitive manner, but only when applied to the extracellular face (Bünemann et al., 1996; van Koppen et al., 1996a). Furthermore, exogenously added SPP and SPPC have been reported to cause a rapid PTX-sensitive increase in [Ca²⁺]_i in various cell types (Okajima and Kondo, 1995; Meyer zu Heringdorf et al., 1996, 1997; Spiegel et al., 1996; van Koppen et al., 1996a, 1996b). In contrast to the activation of cardiac K⁺ currents by SPP and SPPC, SPP was generally more potent than SPPC to increase [Ca²⁺]_i in intact cells. For example, in HEK 293 and bovine aortic endothelial cells, nanomolar concentrations of SPP led to a rapid increase in [Ca²⁺]_i, whereas micromolar concentrations of SPPC were required to elicit a similar response (Meyer zu Heringdorf et al., 1996; van Koppen et al., 1996a). Finally, in human leukemia HL-60 cells, only SPPC, at micromolar concentrations, elevated [Ca²⁺]_i, whereas SPP was ineffective (van Koppen et al., 1996b). Based on these observations, the existence of distinct sphingolipid receptor subtypes with different ligand profiles has been suggested (Bünemann et al., 1996; Meyer zu Heringdorf et al., 1997).

Because an increase in [Ca²⁺]_i is an important prerequisite for exocytosis and insulin secretion in pancreatic  cells (Sharp, 1997), we investigated whether sphingolipids modulate Ca²⁺ homeostasis in the rat insulinoma cell line RINm5F. For this, various sphingolipids were studied on basal and depolarization-induced [Ca²⁺]_i increase in intact cells and on membrane Ca²⁺ currents. We report that certain sphingolipids markedly inhibit depolarization-induced [Ca²⁺]_i increase in RINm5F cells, without affecting basal [Ca²⁺]_i, and inhibit L-type Ca²⁺ currents in a guanine nucleotide-sensitive manner.

	Experimental Procedures

Top Summary Introduction Procedures Results Discussion References

Materials. SPP was purchased from BIOMOL (Hamburg, Germany). All other sphingolipids were obtained from Matreya (Bad Homburg, Germany). The sphingolipids were dissolved in methanol and stored at 20°. Aliquots were dried in a SpeedVac concentrator before use and redissolved in 1 mg/ml bovine serum albumin. Clonidine, nifedipine, verapamil, GTP, GTPS, and GDPS (trilithium salt) were from Sigma (Deisenhofen, Germany). None of the solvents used for the various compounds influenced measurements of [Ca²⁺]_i or membrane currents. All other materials were from previously described sources (Meyer zu Heringdorf et al., 1996; van Koppen et al., 1996a).

Cell culture. RINm5F insulinoma cells were cultured in RPMI-1640 medium containing 10% fetal calf serum, 100 units/ml penicillin G, and 0.1 mg/ml streptomycin. HEK 293 cells stably expressing the human cardiac L-type Ca²⁺ channel _1C subunit together with a ₃ subunit (Klöckner et al., 1997) were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, 100 units/ml penicillin G, 0.1 mg/ml streptomycin, and 0.6 mg/ml G418. For current measurements, cells were plated at low density onto thin glass coverslips and used on the following day. Pretreatment with PTX was performed for 16-20 hr with 100 ng/ml of the toxin.

[Ca²⁺]_i measurements. [Ca²⁺]_i was quantified spectrofluorimetrically (LS 5 B; Perkin-Elmer Cetus, Norwalk, CT, or F2000; Hitachi, Yokohama, Japan) with the fluorescent dye Fura-2 as described previously (Meyer zu Heringdorf et al., 1996; van Koppen et al., 1996a). RINm5F cells were harvested after mild trypsin treatment, resuspended in culture medium, and incubated with 1 µM Fura-2/AM for 60 min at room temperature. To remove excess Fura-2/AM, cells were washed twice and resuspended at a density of 10⁶ cells/ml. [Ca²⁺]_i was measured at room temperature in a continuously stirred cell suspension, with alternating excitation wavelengths between 340 nm and 380 nm while fluorescence emission was read at 495 nm. For measurements in the absence of extracellular Ca²⁺, cells were resuspended in Ca²⁺-free buffer immediately before start of the experiment, and EGTA (50 µM) was added 30 sec before KCl.

Electrophysiology. Cells were grown on glass coverslips that were mounted in a custom-made chamber placed on the stage of an inverted microscope (Axiovert 10; Carl Zeiss, Oberkochen, Germany). The chamber was continuously perfused at a constant rate (1.2 ml/min). The single-electrode voltage-clamp technique was applied to measure membrane potential and membrane current. Fire-polished pipettes from borosilicate filament glass (Hilgenberg, Malsfeld, Germany; o.d., 1.5 mm) were used to form G seals with gentle suction. The patched membrane then was disrupted by a pulse of suction to establish continuity of the interior of the electrode with the cytosol. Voltage-clamp or current-clamp was achieved using an Axon 200 amplifier. For stimulus protocol design and data acquisition, the Axolab TL-125 interface and pClamp 5.5 software (Axon Instruments, Foster City, CA) were used.

Characterization of the Ba²⁺ current. Because several types of voltage-activated Ca²⁺ channels may coexist in insulin-secreting cells (for review, see Ashcroft and Rorsman, 1989), conditions were chosen such as to maximize currents through L-type Ca²⁺ channels using Ba²⁺ as charge carrier (de Waard et al., 1996). Under these conditions, sizeable Ba²⁺ currents could be measured in 86% of the 250 patched cells. Starting from the holding potential of 80 mV throughout all experiments, voltage ramps in the range of 100 to +80 mV (duration, 400 msec) activated membrane currents with current-voltage relations typical for I_Ca.L. No voltage-activated currents were observed negative to 40 mV. In 17% of the cells, voltage ramp-derived current-voltage relations displayed a "hump" in the negative slope region around 20 mV, suggesting the presence of T-type Ca²⁺ channels. The properties of voltage ramp-derived current-voltage relations were identical to those obtained by voltage steps (range, 60 to +60 mV; increments, 10 mV; duration, 300 msec). Currents activated positive to 40 mV, peaked at +3 ± 1 mV, and reversed at +49 ± 1 mV; conductance was half-maximal at 10 ± 1 mV with a respective slope factor of +6 ± 3 mV (mean ± standard error, 18 experiments). A standard double-pulse protocol was applied to test Ca²⁺ channel availability, which was half-maximal at 17 ± 2 mV (slope factor, 11 ± 1 mV, mean ± standard error; 26 experiments; data not shown).

Membrane potential measurements. Membrane potential measurements in current-clamp mode were performed with a bath solution composed of 150 mM NaCl, 5.4 mM KCl, 1.8 mM CaCl₂, 2.0 mM MgCl₂, 10.0 mM HEPES, and 10.0 mM glucose (pH 7.4 adjusted with NaOH). Electrodes were filled with a solution consisting of 140 mM KCl, 4.0 mM MgCl₂, 5.0 mM CaCl₂, 10.0 mM EGTA, 10.0 mM HEPES, and 4.0 mM Na₂-ATP (pH 7.3 adjusted with KOH). Assuming a total [Ca²⁺] of 10 µM, free [Ca²⁺] and free [Mg²⁺] were 50 nM and 300 µM, respectively (EQCAL; Biosoft).

Data analysis and statistics. Conductance curves for I_Ca were estimated by dividing current amplitude at different potentials (V_m) by the corresponding driving force V_m  V_rev, where V_rev is the reversal potential determined by linear regression from the positive slope region of individual current-voltage curves. A Boltzmann function of the form was fitted to the data points for calculation of potentials of half-maximum activation V_0.5 and slope factors k. Steady state inactivation curves for I_Ca were fitted in a similar manner using normalized current data. Curve fitting was performed with pClamp software (Clampfit) or Prism (GraphPAD Software, San Diego, CA).

	Results

Top Summary Introduction Procedures Results Discussion References

Inhibition of KCl-induced [Ca²⁺]_i increase in RINm5F cells by lysosphingolipids. Depolarization of suspended RINm5F cells by the addition of 30 mM KCl rapidly increased [Ca²⁺]_i by several hundred nanomolar before reaching a new elevated plateau level (Fig. 1A). The addition of SPPC (10 µM) had no effect on basal [Ca²⁺]_i. However, SPPC largely attenuated the increase in [Ca²⁺]_i elicited by KCl depolarization (Fig. 1B). This inhibitory effect on the KCl-induced increase in [Ca²⁺]_i was not confined to SPPC but also was observed with dihydro-SPPC, GPS, and PS (Fig. 2A). On the other hand, N-acylated sphingolipids (i.e., N-acetyl-SPPC, N-acetyl-PS, C2-ceramide, and ceramide-1-phosphate), as well as SPP, had little or no effect on the KCl-induced [Ca²⁺]_i increase. The inhibitory effect of SPPC and GPS was half-maximal at 4.9 ± 1.0 µM (five experiments) and 5.6 ± 0.6 µM (five experiments), respectively (Fig. 2B). Because in other cellular systems, sphingolipid-induced signaling is mediated by G proteins of the G_i family (Okajima and Kondo, 1995; Bünemann et al., 1996; Meyer zu Heringdorf et al., 1996; van Koppen et al., 1996a; 1996b), RINm5F cells were pretreated with PTX (100 ng/ml, 16-20 hr). However, PTX pretreatment affected neither the KCl-induced increase in [Ca²⁺]_i (Fig. 1C) nor its attenuation by SPPC (Fig. 1D). Similar results were obtained with other lysosphingolipids (data not shown). Thus, the sphingolipids tested did not require G_i-type G proteins for inhibition of KCl-induced [Ca²⁺]_i transients. Because lysosphingolipids such as PS can inhibit PKC (Hannun and Bell, 1987), the influence of PKC inhibition on [Ca²⁺]_i transients was tested. Pretreatment of RINm5F cells with the PKC inhibitor staurosporine (100 nM, 30 min) affected neither the KCl-induced [Ca²⁺]_i increase nor its inhibition by the lysosphingolipid GPS. On the other hand, activation of PKC by phorbol 12-myristate 13-acetate (1 µM, 10 min) reduced the KCl-induced [Ca²⁺]_i increase and enhanced the inhibitory effect of GPS (data not shown). Thus, inhibition of PKC apparently was not involved in inhibition of [Ca²⁺]_i transients by lysosphingolipids.

View larger version (34K): [in this window] [in a new window]   Fig. 1.   Influence of SPPC on basal and KCl-induced [Ca2+]i increase in RINm5F cells. Tracings of representative [Ca2+]i measurements in control (A and B) and PTX-treated (100 ng/ml, 16-20 hr) cells (C and D). A and C, Addition of 30 mM KCl only. B and D, Addition of 10 µM SPPC followed 60 sec later by addition of KCl.

View larger version (23K): [in this window] [in a new window]   Fig. 2.   Influence of various sphingolipids on the KCl-induced [Ca2+]i increase in RINm5F cells. A, Effects of various sphingolipids (10 µM each) on the KCl (30 mM)-induced [Ca2+]i increase. Values are mean ± standard error relative to controls, in which KCl-induced [Ca2+]i increase amounted to 491 nM. Bottom of columns, number of experiments. N-Ac, N-acetyl; DH, dihydro, CP, ceramide-1-phosphate; C2-Cer, C2-ceramide or N-acetyl-sphingosine. B, Concentration dependence of the inhibitory effect of SPPC () and GPS () on the KCl-induced [Ca2+]i increase.

View larger version (16K): [in this window] [in a new window]   Fig. 3.   Effects of high K+ and GPS on membrane potential in RINm5F cells. Time course of membrane potential VM in a representative RINm5F cell. With [K+]o = 5.4 mM, VM stabilized around 67 mV 3 min after patch disruption. Elevation of [K+]o to 35.4 mM reversibly depolarized the cell to 30 mV, whereas GPS (20 µM) depolarized by <10 mV.

Inhibition of Ba²⁺ current in RINm5F cells by lysosphingolipids. To test the hypothesis that sphingolipids inhibit Ca²⁺ influx in RINm5F cells by blocking L-type Ca²⁺ channels, the effects of GPS, SPPC, and SPP on membrane Ba²⁺ currents were studied. Bath application of GPS (10 µM) reduced the amplitude of Ba²⁺ currents within seconds (Fig. 4A). Although the reversibility of this effect was generally poor, subsequent current block by Cd²⁺ was fully reversible. The current tracings displayed in Fig. 4, B and C, further illustrate the effect of GPS. Note that GPS did not accelerate or otherwise affect the inactivation time course of Ba²⁺ current during voltage-clamp steps. In nine cells, GPS reduced current amplitude at 0 mV from 13.8 ± 2.1 to 9.3 ± 1.6 pA/pF (Fig. 5A). This current decrease was significant within the voltage range from 30 to +30 mV. The extent of current inhibition by GPS was independent of the holding potential (fraction of control current at 0 mV: 71 ± 5% at V_H 80 mV versus 68 ± 8% at V_H 50 mV; four experiments; data not shown). The reversal potential was slightly shifted from +42.6 ± 2.5 mV (control) to +39.7 ± 2.2 mV (GPS, p < 0.01). Ba²⁺ current was half-activated at 13.7 ± 1.8 mV before and at 10.0 ± 1.7 mV (p < 0.05) after exposure to GPS; the slope of the conductance curves was not changed (6.6 ± 0.3 mV versus 6.9 ± 0.4 mV). Consistent with this shift, we found that the inhibitory action of GPS on Ba²⁺ current was potential dependent (Fig. 5B), being significantly more pronounced at 25 mV than at +10 mV (fraction of block: 51.4 ± 6.6% versus 29.3 ± 3.8%, respectively; p < 0.05). The GPS-induced inhibition of Ba²⁺ current was concentration dependent (Fig. 5C). However, even at the highest concentration tested (20 µM), GPS did not abolish Ba²⁺ current.

View larger version (32K): [in this window] [in a new window]   Fig. 4.   Inhibition of Ba2+ current by GPS in RINm5F cells. A, Time course of voltage ramp-derived current measured at 0 mV in a typical cell. Bars, exposure to extracellular GPS (10 µM) and Cd2+ (200 µM). Arrows, times at which the current tracings displayed in B and C were recorded. B, Superimposed current tracings elicited by voltage steps to 0 mV before (Con) and after exposure to GPS and Cd2+. Arrowhead, zero current level. Calibration bars, 100 pA, 100 msec. C, Superimposed current tracings in response to voltage ramps from 100 to +80 mV (duration, 400 msec) before (Con) and after exposure to GPS and Cd2+. Dashed line, zero current level. Calibration bars, 100 pA, 50 mV.

View larger version (26K): [in this window] [in a new window]   Fig. 5.   GPS reduces Ba2+ current in RINm5F cells in a voltage- and concentration-dependent manner. A, Ramp-derived current-voltage relations in the absence () and presence () of 10 µM GPS. B, Fraction of blocked current as a function of clamp potential (data from A). Block became evident positive to 45 mV, reached a maximum around 25 mV, and declined toward more positive potentials. C, Average current at 0 mV in the presence of the indicated concentrations of GPS. Values are mean ± standard error; number of experiments is indicated.

Mechanism of Ba²⁺ current inhibition by lysosphingolipids in RINm5F cells. Sphingolipids apparently play a dual role in cellular regulation, acting as intracellular messengers and as ligands for plasma membrane receptors (Spiegel et al., 1996; Meyer zu Heringdorf et al., 1997). To distinguish between an intracellular and extracellular site of action, GPS (10 µM) was included in the pipette solution when measuring Ba²⁺ currents (Fig. 6A). As in time-matched controls, current amplitude measured at 0 mV increased rapidly to reach a stable level ~3 min after establishing the whole-cell mode. Thereafter, current remained unchanged for several minutes, regardless of the intracellular presence of GPS. On average, the voltage ramp-derived current-voltage relation obtained after 7.2 ± 1.2 min (nine experiments) of cell dialysis with GPS-containing pipette solution did not differ from that measured under control conditions (no GPS exposure) in a comparable group of cells (Fig. 6B). However, additional exposure of the cell to extracellular GPS led to a rapid current reduction. In all three cells exposed to extracellular GPS in addition to GPS in the pipette solution, current amplitude was reduced to an extent similar to bath-applied GPS only. These results indicate that the target for GPS and possibly other sphingolipids is at the outer face of the plasma membrane in RINm5F cells.

View larger version (30K): [in this window] [in a new window]   Fig. 6.   Intracellularly applied GPS does not reduce Ba2+ current in RINm5F cells. A, Time course of current measured at 0 mV in a typical cell with 10 µM GPS in pipette solution (GPSpip) and after additional bath application of GPS (GPSbath). Arrows, times at which the curves displayed in B were recorded; dashed line, average current of time-matched controls (TMC). B, Average ramp-derived current-voltage relations measured after 7.2 ± 1.2 min of cell dialysis with 10 µM GPS in pipette solution (GPSpip, , nine experiments) and after additional bath application of GPS (+GPSbath, , three experiments). Dotted line, control curve taken from Fig. 5A.

View larger version (23K): [in this window] [in a new window]   Fig. 7.   Influence of intracellular GTPS and GDPS on Ba2+ current inhibition by extracellular GPS in RINm5F cells. Shown are average ramp-derived current-voltage relations in the absence () and presence () of 10 µM extracellular GPS with either GTPS (100 µM, A) or GDPS (10 mM, B) added to the electrode solution.

View larger version (26K): [in this window] [in a new window]   Fig. 8.   Summary of effects of various agents on Ba2+ current in RINm5F cells. Current amplitude is expressed as fraction of control current at 0 mV. Values are mean ± standard error. Bottom of columns, number of experiments. Because control current amplitudes did not vary significantly between different experimental groups (Kruskal-Wallis test), pooled control currents (14.9 ± 1.0 pA/pF, 44 experiments) were taken to calculate the effect of GPS in the electrode solution ([GPS]pip). All groups, except SPP, [GPS]pip, and GPS + [GDPS]pip, were significantly different (p < 0.05) from their respective controls.

Inhibition of Ba²⁺ current in _1C/₃-expressing HEK 293 cells by GPS. The data presented so far suggest an extracellular target and participation of a G protein in the inhibitory effect of GPS in RINm5F cells. For comparison, we investigated the effects of GPS in HEK 293 cells stably expressing the human cardiac L-type Ca²⁺ channel _1C subunit together with a ₃ subunit (Klöckner et al., 1997). Ba²⁺ currents in these HEK 293 cells exhibited a bell-shaped current-voltage relation, with an activation threshold at 30 mV, a maximum between +10 and +20 mV, and current reversal between +50 and +60 mV (Fig. 9), which are properties typical for I_Ca.L. Similar to RINm5F cells, extracellular application of GPS (10 µM) significantly reduced the Ba²⁺ current in HEK 293 cells, by 53 ± 6% at maximum current (Fig. 9A). However, quite distinct from RINm5F cells, _1C/₃ subunit-expressing HEK 293 cells dialyzed with GPS (10 µM for 6 ± 1 min) in the electrode solution also exhibited a significant reduction in maximal current density compared with a control group, from 6.4 ± 1.0 to 3.6 ± 0.8 pA/pF (Fig. 9A). Furthermore, current amplitudes decreased continuously after entering the whole-cell configuration with intracellular GPS, whereas they increased and stabilized on a high level in time-matched controls (data not shown). The difference between RINm5F and HEK 293 cells was even more striking when the effect of the stable GDP analog GDPS was analyzed. Although in RINm5F cells intracellular application of GDPS abolished the inhibitory effect of GPS, it had no effect on GPS-induced inhibition of Ba²⁺ currents in HEK 293 cells. As illustrated in Fig. 9B, when HEK 293 cells were first dialyzed with GDPS (10 mM for 8 ± 1 min), Ba²⁺ current inhibition by extracellular GPS (10 µM) was similar to that in control cells. These results indicate that in HEK 293 cells stably expressing the pore-forming L-type Ca²⁺ channel _1C subunit together with a ₃ subunit, both extracellular and intracellular GPS reduce Ba²⁺ currents and apparently without involvement of a G protein.

View larger version (29K): [in this window] [in a new window]   Fig. 9.   Effects of GPS on Ba2+ current in HEK 293 cells expressing 1C and 3 Ca2+ channel subunits. A, Average step-derived current-voltage relations in the absence () and presence () of 10 µM extracellular GPS (12 experiments) and in cells dialyzed for 6 ± 1 min with 10 µM GPS (GPSpip, , six experiments). B, Average step-derived current-voltage relations in cells (six experiments) dialyzed for 8 ± 1 min with 10 mM GDPS in the electrode solution before () and after () bath application of 10 µM GPS.

	Discussion

Top Summary Introduction Procedures Results Discussion References

The lysosphingolipids SPP and SPPC have been reported to cause a rapid increase in [Ca²⁺]_i in various cell types (Okajima and Kondo, 1995; Meyer zu Heringdorf et al., 1996; 1997; Spiegel et al., 1996; van Koppen et al., 1996a, 1996b), apparently by activating G protein-coupled sphingolipid receptors. Therefore, we examined in the current study whether sphingolipids may regulate Ca²⁺ homeostasis in RINm5F insulinoma cells in a similar manner. The data presented herein demonstrate that sphingolipids elicit an opposite response in RINm5F cells (i.e., they inhibit Ca²⁺ transients, apparently by inhibiting Ca²⁺ entry through L-type Ca²⁺ channels). By testing a variety of different sphingolipids, we could demonstrate that the attenuation of depolarization-induced Ca²⁺ influx was restricted to certain compounds (i.e, GPS, PS, SPPC, and dihydro-SPPC), whereas various N-acylated sphingolipids and SPP had little or no effect. A similar pharmacological profile was obtained when measuring Ba²⁺ currents through L-type Ca²⁺ channels. GPS and SPPC inhibited the current, whereas SPP was without effect. The distinct effectiveness of some, but not all, sphingolipids studied, all of which consist of a long lipophilic hydrocarbon backbone and a hydrophilic head group, strongly argues against a nonspecific effect on the membrane lipid bilayer. Furthermore, preferential insertion of the amphiphilic lysosphingolipids into one leaflet of the lipid bilayer, thereby causing tension in the other leaflet and thus modulating channel activity (Martinac et al., 1990), also is unlikely to be the mechanism by which these compounds inhibit Ca²⁺ currents. Such responses require tens of minutes to develop, whereas sphingolipid-induced inhibition of currents was observed within seconds. Finally, there is no evidence that Ca²⁺ current inhibition is due to a surface charge effect (Ji et al., 1993) of the inhibitory sphingolipids because a small depolarization of membrane potential by GPS was measured (Fig. 3), which in fact may decrease channel availability rather than increasing it as theoretically expected for a positively charged compound.

Inhibition of KCl-induced [Ca²⁺]_i transients in intact RINm5F cells by the sphingolipids was more pronounced than the sphingolipid-induced inhibition of Ba²⁺ currents. For example, 10 µM GPS reduced the KCl-induced [Ca²⁺]_i increase by ~80%, whereas inhibition of maximal Ba²⁺ current was only 33%, suggesting that additional sites of action of GPS should be considered. However, first, GPS did not inhibit ryanodine receptor-mediated Ca²⁺ release, which may contribute to KCl-induced [Ca²⁺]_i increase in intact cells. Second, GPS did not lead to a hyperpolarization (Fig. 3), which could potentially counteract the KCl-induced depolarization and the consequent Ca²⁺ influx. The depolarization from 59 to 54 mV caused by GPS should result in a negligible reduction of Ca²⁺ channel availability (97.8% versus 96.6%; see Experimental Procedures), making a GPS-mediated inactivated state block highly unlikely. Third, a participation of low voltage-activated T-type Ca²⁺ channels in the KCl-induced Ca²⁺ influx and its inhibition by sphingolipids also is unlikely because (1) the L-type Ca²⁺ channel blocker verapamil completely prevented the KCl-induced Ca²⁺ influx and (2) only in a minor fraction of cells (17%) was some evidence obtained for the presence of T-type Ca²⁺ channels. Finally, it has to be considered that at the high extracellular K⁺ concentration (35 mM) used for measurement of [Ca²⁺]_i transients in intact cells, the cells depolarized to only 30 mV (Fig. 3) and not to 0 mV, where the maximal Ba²⁺ current was observed. On the other hand, inhibition of Ba²⁺ current by GPS was shown to be potential dependent (i.e., maximal at ~30 mV) (Fig. 5B). Thus, taking into account the different conditions to measure [Ca²⁺]_i transients and Ba²⁺ currents, the extent of inhibition by GPS (10 µM), ~80% and 50%, respectively, is remarkably similar. Thus, we hypothesize that the sphingolipid-induced inhibition of depolarization-dependent [Ca²⁺]_i transients in RINm5F cells is, at least to a major extent, due to inhibition of L-type Ca²⁺ channels. Sphingolipids also have been shown to inhibit Ca²⁺ currents in two other cell types. In GH₄C₁ rat pituitary cells, SPPC (10 µM) was found to inhibit both Ca²⁺ currents and depolarization-induced Ca²⁺ influx, whereas SPP had no effect (Törnquist et al., 1995). In rat ventricular myocytes, sphingosine (25 µM) greatly reduced Ca²⁺ current amplitude, whereas SPPC (25 µM) had no (McDonough et al., 1994) or only a small inhibitory effect (Yasui and Palade, 1996).

Activation of muscarinic K⁺ currents in atrial myocytes by SPP and SPPC apparently is mediated by a specific sphingolipid receptor coupled to G_i-type G proteins. The sphingolipid activation occurred at low nanomolar concentrations (EC₅₀ = 1-2 nM), was completely blocked by PTX, was strictly GTP dependent, and was observed only when the sphingolipids were applied to the outer face of the membrane (Bünemann et al., 1996; van Koppen et al., 1996a). In comparison, the inhibitory sphingolipid action on Ca²⁺ entry in RINm5F cells was observed at micromolar concentrations and was PTX insensitive. Furthermore, inhibition of Ca²⁺ current by GPS did not require the addition of GTP. However, application of the stable GDP analog GDPS into the cell interior completely abolished the GPS-induced inhibition, suggesting that a guanine nucleotide-sensitive target, most likely a G protein, is involved in the inhibitory sphingolipid action. The data, furthermore, suggest that endogenous GTP required for activation of the responsible G protein or proteins is present in sufficient amounts under the whole-cell condition used to measure Ca²⁺ currents. To study a possible involvement of a membrane receptor, GPS was applied either extracellularly or intracellularly. We demonstrate that only extracellular application of GPS induced inhibition of Ca²⁺ currents in RINm5F cells, whereas intracellularly applied GPS had no effect and also did not alter Ca²⁺ current inhibition induced by extracellular GPS. These data suggest that Ca²⁺ current inhibition caused by GPS in RINm5F cells is not due to direct activation of the responsible G proteins, which should be located on the inner surface of the membrane, but is mediated by a target facing the extracellular space, such as a membrane receptor.

To corroborate the hypothesis that RINm5F cells express a sphingolipid receptor mediating a G protein-dependent inhibition of Ca²⁺ current, we studied the effect of GPS on Ca²⁺ currents in HEK 293 cells stably expressing the human cardiac L-type Ca²⁺ channel _1C subunit together with a ₃ subunit (Klöckner et al., 1997). GPS inhibited Ca²⁺ currents in these cells at least as well as in RINm5F cells. In contrast, however, to the data obtained in RINm5F cells, the inhibitory effect of GPS in HEK 293 cells was observed regardless of which side of the membrane was exposed to GPS. Moreover, the GPS-induced inhibition of Ca²⁺ currents in HEK 293 cells was not affected by GDPS. It must be noted here that the HEK 293 cells express a different pore-forming L-type Ca²⁺ channel subunit (_1C) than the neuroendocrine RINm5F cells (_1D), which in addition are equipped with the complete set of subunits forming the native L-type Ca²⁺ channel (Birnbaumer et al., 1994). Thus, the minimal conclusion derived from these data is that HEK 293 cells do not express a receptor for GPS mediating via a G protein inhibition of the L-type Ca²⁺ channel composed of _1C and ₃ subunits. The Ca²⁺ current inhibition caused by GPS in HEK 293 cells may be due to a direct effect of the sphingolipid on the _1C or ₃ subunits (or both). Thus, to prove this hypothesis, it may be necessary to express the complete neuroendocrine type of L-type Ca²⁺ channel with all of its subunits in a suitable host cell and then study its regulation by sphingolipids.

We can only speculate about the G protein type apparently involved in the inhibitory sphingolipid signaling in RINm5F cells. These cells possess G_i and G_o proteins, which are both activated by ₂-adrenoceptors and couple to L-type Ca²⁺ channels in an inhibitory manner (Schmidt et al., 1991). However, these G proteins are PTX sensitive and therefore not suited to serve as transducers for putative lysosphingolipid receptors in RINm5F cells. Most, if not all, voltage-dependent Ca²⁺ channels underlie G protein regulation via membrane-delimited or second messenger-dependent mechanisms (Wickman and Clapham, 1995). Thus, for example, one might speculate that subunits released from a PTX-insensitive G protein could carry the signal for inhibition of Ca²⁺ current as suggested for various channels (Wickman and Clapham, 1995; Qin et al., 1997).

In conclusion, we demonstrate in the current study that certain lysosphingolipids markedly inhibit influx of Ca²⁺ through L-type Ca²⁺ channels into RINm5F insulinoma cells and provide circumstantial evidence that this inhibitory sphingolipid action involves a PTX-insensitive G protein and an extracellular target structure, probably a membrane receptor.