Effects of the immunosuppressant FK506 on intracellular Ca²⁺ release and Ca²⁺ accumulation mechanisms

Abstract

1. The immunophilin FKBP12 associates with intracellular Ca²⁺ channels and this interaction can be disrupted by the immunosuppressant FK506. We have investigated the effect of FK506 on Ca²⁺ release and Ca²⁺ uptake in permeabilized cell types.

2. Changes in medium free [Ca²⁺] were detected by the fluorescent Ca²⁺ indicator fluo-3 in digitonin-permeabilized SH-SY5Y human neuroblastoma cells, DT40 and R23-11 (i.e. triple inositol 1,4,5-trisphosphate (IP₃) receptor knockout cells) chicken B lymphocytes and differentiated and undifferentiated BC₃H1 skeletal muscle cells. ⁴⁵Ca²⁺ fluxes were studied in saponin-permeabilized A7r5 rat smooth muscle cells.

3. Addition of FK506 to permeabilized SH-SY5Y cells led to a sustained elevation of the medium [Ca²⁺] corresponding to ∼30% of the Ca²⁺ ionophore A23187-induced [Ca²⁺] rise. This rise in [Ca²⁺] was not dependent on mitochondrial activity.

4. This FK506-induced [Ca²⁺] rise was related to the inhibition of the sarcoplasmic/endoplasmic reticulum Ca²⁺-Mg²⁺-ATPase (SERCA) Ca²⁺ pump. Oxalate-facilitated ⁴⁵Ca²⁺ uptake in SH-SY5Y microsomes was inhibited by FK506 with an IC₅₀ of 19 μm.

5. The inhibition of the SERCA Ca²⁺ pump was not specific since several macrocyclic lactone compounds (ivermectin > FK506, ascomycin and rapamycin) were able to inhibit Ca²⁺ uptake activity.

6. FK506 (10 μm) did not affect IP₃-induced Ca²⁺ release in permeabilized SH-SY5Y and A7r5 cells, but enhanced caffeine-induced Ca²⁺ release via the ryanodine receptor (RyR) in differentiated BC₃H1 cells.

7. In conclusion, FK506 inhibited active Ca²⁺ uptake by the SERCA Ca²⁺ pump; in addition, FK506 enhanced intracellular Ca²⁺ release through the RyR, but it had no direct effect on IP₃-induced Ca²⁺ release.

FK506 is a macrocyclic lactone immunosuppressant, which is used for the prevention of allograft rejection in organ transplantation (Spencer et al. 1997). An important cellular target of FK506 is the ubiquitous FK506-binding protein, FKBP12, a 12 kDa member of the immunophilin family (Marks, 1996). The FK506-FKBP12 complex binds to calcineurin, a Ca²⁺-calmodulin-dependent protein phosphatase and suppresses the interleukin2-dependent proliferation of T cells (Schreiber, 1991).

FKBP12 does not only play an important role in T cells, but it is present in many cell types and it is very well conserved from plants and yeast to humans (Kay, 1996). FKBP12 can participate in other cellular processes, including neurotransmitter release, neural nitric oxide production, nerve growth, signal transduction via the type 1 receptor for transforming growth factor-β and intracellular Ca²⁺ release via the ryanodine receptor (RyR) and the inositol 1,4,5-trisphosphate receptor (IP₃R) (Snyder et al. 1998).

The RyR and IP₃R are tetrameric intracellular Ca²⁺ release channels, which mediate mobilization of stored Ca²⁺ from the sarcoplasmic reticulum (SR) and the endoplasmic reticulum (ER) into the cytoplasm (Berridge, 1993). The RyR plays an important role in excitation-contraction coupling in skeletal muscle cells (RyR1) and in heart muscle cells (RyR2). The IP₃R is a ubiquitous Ca²⁺ release channel activated by IP₃ produced from phosphatidylinositol 4,5-bisphosphate by phospholipase C in response to several extracellular stimuli (Berridge, 1993).

The interaction of FKBP12 with the RyR is well characterized. FKBP12 binds stoichiometrically to the RyR1, i.e. four FKBP12 molecules bind per tetrameric channel (Jayaraman et al. 1992). Moreover, the RyR can be affinity purified from terminal cisternae based on its tight association with FKBP12 (Xin et al. 1995). FK506 disrupts the interaction of FKBP12 with the RyR1, thereby increasing the open probability and mean open time of the channel (Ahern et al. 1994). Furthermore, associated FKBP12 stabilizes the RyR channel (Brillantes et al. 1994) and induces a rectification of the Ca²⁺-release channel, which favours unidirectional Ca²⁺ flux from the lumen of the ER to the cytosol (Chen et al. 1994). FKBP12 also stabilizes the interaction between different RyRs, thereby enhancing coupled gating of a cluster of channels (Marx et al. 1998). Recent studies revealed that this interaction is conserved, as FKBP12 forms an integral part of the RyR1 and modulates the RyR1 in each of the five classes of vertebrates (Qi et al. 1998). These biochemical and functional data suggest that FKBP12 can function as a modulator of the Ca²⁺ flux properties of the RyR1 and that interaction with FKBP12 can influence Ca²⁺ signalling by the RyR.

Recently, Cameron et al. (1995b) demonstrated that FKBP12 could also interact directly with the type 1 IP₃ receptor (IP₃R1), thereby modulating the Ca²⁺ flux through this channel. Dissociation of FKBP12 from cerebellar microsomes using FK506 potentiated Ca²⁺ release via IP₃R1. In contrast to this observation, intracellular Ca²⁺ oscillations mediated by the IP₃R in intact tracheal epithelial cells were strongly inhibited by FK506 (Kanoh et al. 1999). This effect of FK506 was accompanied by a decrease of the Ca²⁺ content of the ER Ca²⁺ stores. Furthermore, FKBP12 could target calcineurin to the IP₃R1. The associated calcineurin could rapidly modulate the phosphorylation state of the IP₃R1 and might induce a fine-tuning of the Ca²⁺ flux properties of this channel (Cameron et al. 1995a). Besides these direct effects, FK506 attenuated IP₃R expression over more extended periods by inhibiting the calcineurin pathway (Genazzani et al. 1999).

Although the IP₃R is the most prevalent intracellular Ca²⁺ release mechanism, little is known to what degree IP₃-induced Ca²⁺ release is affected by FK506 in a cellular context. In order to clarify the potential role of immunophilins in cellular Ca²⁺ handling, we have investigated the functional effects of FK506 on intracellular Ca²⁺ release in permeabilized cells.

The SH-SY5Y neuroblastoma cell line was chosen since it expresses predominantly IP₃R1 (Wojcikiewicz, 1995). Cell lines that express high levels of RyR (differentiated BC₃H1 muscle cells; De Smedt et al. 1991) or triple IP₃R knockout B cells (R23-11) (Sugawara et al. 1997) were used for comparison purposes.

Our results indicate that FK506 inhibited the intracellular SERCA Ca²⁺ pumps. In addition, FK506 did not affect IP₃-induced Ca²⁺ release, but enhanced Ca²⁺ flux through the RyR. The effects of FK506 on intracellular Ca²⁺ signalling may be relevant for the potential side effects of this drug which is widely applied in allograft transplantation surgery.

METHODS

Materials

FK506 was kindly provided by Dr A. Friedrich, Fujisawa GmbH, München, Germany. Rapamycin, cyclosporin A, midecamycin, ivermectin and A23187 were purchased from Sigma (St Louis, MO, USA). Ascomycin and polyclonal rabbit anti-FKBP12 antibody were purchased from Affinity Bioreagents, Inc. (Golden, CO, USA). Fluo-3 was obtained from Molecular Probes (Junction City, OR, USA) and ⁴⁵Ca²⁺ was obtained from Amersham Pharmacia Biotech AB (Uppsala, Sweden). IP₃ was from Boehringer (Mannheim, Germany).

Cell culture and preparation of cell homogenates and microsomes

The SH-SY5Y human neuroblastoma cell line (9403034) was obtained from the European Collection of Animal Cell Cultures (Wiltshire, UK). The BC₃H1 mouse muscle cell line (CRL-1443) and the A7r5 rat aorta smooth muscle cell line (CRL-1444) were obtained from the American Type Culture Collection (Rockville, MD, USA). The DT40 and R23-11 chicken B lymphocyte cell lines were a generous gift from Dr T. Kurosaki (Tokyo, Japan). Undifferentiated and differentiated BC₃H1 (De Smedt et al. 1991), A7r5 (Missiaen et al. 1992a) and SH-SY5Y cell monolayers (Wojcikiewicz, 1995) as well as DT40 and R23-11 cell suspensions (Miyakawa et al. 1999) were cultured as described. Cell harvest and microsomal preparations were performed according to earlier published procedures (Parys et al. 1995). SH-SY5Y homogenates for immunoblot analysis were prepared by a brief sonication step (4 times 10 s with a probe sonicator MSE Ltd, Westminster, UK) in 50 mM Tris-HCl, 300 mM sucrose and 0.5 % Triton X-100 at pH 7.4. Mouse heart homogenates (Ji et al. 1999) and cardiac SR (Kirchberger & Antonetz, 1982) were prepared as described. All preparations were frozen in liquid nitrogen and stored at -80°C until use. Protein concentrations were measured by the method of Lowry et al. (1951) with bovine serum albumin as standard after precipitation of the proteins with 10 % (w/v) trichloroacetic acid.

Ca²⁺ release studies in permeabilized cells

Cell pellets were resuspended in intracellular medium (120 mM KCl, 30 mM Hepes pH 7.4, 1 mM MgCl₂) in the presence of 1 mM ATP, 25 mM creatine phosphate, 50 i.u. creatine kinase and 2.5-5 μM fluo-3 and transferred to a 37°C thermostatically controlled 4 ml fluorescence quartz cuvette. Cell density was 15 × 10⁶ cells ml⁻¹ for SH-SY5Y, 50 × 10⁶ cells ml⁻¹ for DT40 and R23-11 and 3.5 × 10⁶ cells ml⁻¹ for BC₃H1. Mild treatment of the cells with digitonin (50 μM) disrupted the plasma membrane. Free [Ca²⁺] was estimated from the fluorescence signal upon binding to fluo-3. The fluorescence (excitation: 503 nm, emission: 530 nm) was measured with a Perkin Elmer LS50 fluorimeter and recorded as a function of time by the Fluorescence Luminescence Data Manager (version 2.50, Perkin Elmer) software. A23187 (8 μM) was added at the end of each experiment to control store loading and to measure the total releasable Ca²⁺. The fluorescence signal (F) was calibrated by first adding 0.5 mM Ca²⁺ (F_max) and then 5 mM EGTA (F_min).

The free [Ca²⁺] was calculated by the equation:

(Grynkiewicz et al. 1985), using a dissociation constant of 864 nM (K_d), as determined in cytosol-like medium at 37°C (Missiaen et al. 1991).

⁴⁵Ca²⁺ uptake experiments

⁴⁵Ca²⁺ uptake was determined using a rapid filtration technique and was basically performed as previously described (Ji et al. 1999). Cardiac homogenates (100 μg protein ml⁻¹) or SH-SY5Y microsomes (200 μg protein ml⁻¹) were incubated at 37°C in Ca²⁺ uptake medium (1.5 ml) containing 40 mM imidazole-HCl pH 7.0, 100 mM KCl, 5 mM MgCl₂, 5 mM NaN₃, 5 mM potassium oxalate, 0.5 mM EGTA, 1 μM ruthenium red and a free [Ca²⁺] of 3.16 μM as determined by CaBuf software (Dr G. Droogmans, K.U.Leuven, Belgium). ⁴⁵Ca²⁺ (10 μCi ml⁻¹) uptake was started by the addition of 5 mM ATP and the initial uptake rate was determined by sampling after 30, 60 and 120 s for cardiac homogenates and after 6, 12 and 18 min for SH-SY5Y microsomes and performing a linear regression analysis. The non-specific ⁴⁵Ca²⁺ uptake was determined before the addition of ATP and subtracted from the sample values. Microsomes were pre-incubated for 5 min with drug or vehicle prior to the addition of ATP, which initiates ⁴⁵Ca²⁺ uptake. Longer pre-incubation times (up to 30 min) did not affect the inhibitory effect, indicating that an equilibrium distribution was obtained.

Ca²⁺-Mg²⁺-ATPase activity of cardiac SR vesicles

The Ca²⁺-Mg²⁺-ATPase activity was measured at 37°C in 1 ml of medium containing 20 mM Mops pH 6.8, 100 mM KCl, 1 mM EGTA, 5 mM NaN₃, 5 mM ATP, 5.77 mM MgCl₂, 100 μM ouabain, 40 i.u. pyruvate kinase, 36 i.u. lactate dehydrogenase, 1.5 mM phosphoenolpyruvate, 0.63 mM disodium NADH and 10 μM A23187. The decrease in absorbance at 340 nm was recorded with a Beckman DU-1 spectrophotometer. After thermal equilibration of the cuvette, 25 μg of cardiac SR vesicles were added. A final free [Ca²⁺] of 10 μM was obtained by adding 0.97 mM Ca²⁺. To determine the rate of the Ca²⁺-Mg²⁺-ATPase activity, the difference between the ATPase activities in the presence and absence of Ca²⁺ was measured. FK506 or ethanol (vehicle) was added to the thermostatically controlled cuvette and incubated for 5 min before recording the ATPase activity.

⁴⁵Ca²⁺ fluxes on saponin-permeabilized A7r5 cells

⁴⁵Ca²⁺ fluxes on monolayers of saponin-permeabilized A7r5 cells at 25°C were measured exactly as described previously (Missiaen et al. 1992a).

Sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) and Western blot analysis

Cell homogenates and microsomes were separated on 13 % SDS-PAGE followed by semidry electrophoretic transfer to a polyvinylidene difluoride membrane (Immobilon-P, Millipore Corp., Bedford, USA). Transfer was performed for 2 h in transfer buffer, containing 25 mM Tris, 192 mM glycine, 0.0375 % SDS and 20 % methanol. After blocking the membrane with PBS-T (phosphate-buffered saline with 1 % Tween-20) containing 5 % non-fat dried milk powder, the membrane was incubated overnight with the rabbit anti-FKBP12 antibody (diluted 1/1000) at 4°C. The membrane was further incubated with an alkaline phosphatase-conjugated anti-rabbit IgG antibody for 1 h at room temperature. The immunoreactive bands were visualized by conversion of the Vistra ECF substrate (Amersham Pharmacia Biotech AB) into a fluorescent probe and analysed and quantified as described previously (Vanlingen et al. 1997).

Statistics

Values are given as means ±s.d., unless otherwise stated. Statistical comparisons between two groups were performed by Student's t test. P < 0.05 was considered significantly different.

RESULTS

FK506-induced Ca²⁺ responses in SH-SY5Y cells

Permeabilization of the plasma membrane of SH-SY5Y cells in the uptake medium with digitonin caused an immediate decrease of the free [Ca²⁺] in the cuvette, due to Ca²⁺ uptake in the ER and in the mitochondria. The steady-state [Ca²⁺] after Ca²⁺ accumulation was typically 106 ± 39 nM (n= 20). Addition of 100 μM FK506 resulted in a sustained elevation of [Ca²⁺] to about 500 nM (Fig. 1A). After reaching the new Ca²⁺ level, 100 nM IP₃ still released Ca²⁺ from the Ca²⁺ stores (Fig. 1A). At the end of the experiment, the Ca²⁺ ionophore A23187 (8 μM) was given to measure the medium [Ca²⁺] after maximal discharge of the intracellular stores, which was set as 100 % in further calculations. The FK506-induced [Ca²⁺] rise did not occur when the vehicle (0.1 % ethanol) was given alone and FK506 had no effect on the fluo-3 fluorescence in the absence of cells (data not shown). This indicates that the rise in [Ca²⁺] was neither due to an effect of the solvent nor to autofluorescence properties of FK506.

Figure 1. FK506-induced [Ca²⁺] rise in permeabilized SH-SY5Y cells and FKBP12 expression in SH-SY5Y homogenates and microsomes.

A, a representative recording of the free [Ca²⁺] measured by fluo-3 fluorescence. Calibration was performed as described in Methods. After permeabilization of the plasma membrane with digitonin (50 μM), cells accumulated Ca²⁺ in their Ca²⁺ stores. After stabilization of the signal, FK506 (100 μM) and IP₃ (0.1 μM) were successively added. At the end of each experiment, A23187 (8 μM) was added to estimate the total releasable amount of Ca²⁺. B, dose-response curve for the FK506-induced [Ca²⁺] rise, normalized to the A23187-induced [Ca²⁺] rise, which was set as 100 %. Each data point is the mean ±s.d. of at least three independent experiments. Sigmoidal fitting was performed by using Microcal Origin (version 6.0) software. C, Western blot analysis (anti-FKBP12, diluted 1/1000) of SH-SY5Y cell homogenates (lane 1, 75 μg), SH-SY5Y microsomes (lane 2, 75 μg) and human recombinant FKBP12 (lane 3, 50 ng) after separation on 13 % SDS-PAGE and electroblotting to Immobilon-P.

The FK506-induced [Ca²⁺] rise, calculated as a percentage of the A23187-induced [Ca²⁺] rise, was dose dependent (Fig. 1B). The concentration dependence was fitted by a sigmoidal curve and characterized by an apparent EC₅₀ of 49 ± 6 μM FK506. The maximum [Ca²⁺] rise induced by FK506 was reached at about 150 μM FK506. This concentration led to a sustained [Ca²⁺] elevation corresponding to 31 ± 6 % of the A23187-induced [Ca²⁺] rise. No detectable [Ca²⁺] increase was induced by FK506 at a concentration lower than 10 μM. A more pronounced [Ca²⁺] increase could not be achieved, since it is technically impossible to use concentrations higher than 200 μM FK506, due to its low solubility in an aqueous environment and due to potential solvent (ethanol) effects on intracellular Ca²⁺ signals. It should also be pointed out that the effective dose of lipophilic compounds such as FK506 is determined by the partition of the drug between aqueous and lipid compartments (Heirwegh et al. 1992). The EC₅₀ and maximum level of the FK506-induced [Ca²⁺] rise will therefore be dependent on the amount of membrane material used in the assays. For this reason, it will be difficult to make a quantitative comparison between different cell types or membrane preparations.

FKBP12, the best characterized cellular target for FK506, could be immunologically detected not only in a total homogenate of SH-SY5Y cells but also in microsomes of these cells, suggesting a molecular link between FKBP12 and an ER component (Fig. 1C). The microsomal fraction contained 20 ± 4 % (n= 4) of the total FKBP12 expressed in SH-SY5Y cells. These results confirmed the previous observations that FKBP12 is a ubiquitously expressed protein and is abundant in cells of neural origin (Steiner et al. 1992).

IP₃R-dependent Ca²⁺ mobilization pathway

FK506 may dissociate FKBP12 from IP₃R1 and increase its sensitivity to IP₃ (Cameron et al. 1995b). We therefore investigated whether the [Ca²⁺] rise induced by FK506 was mediated by the IP₃R. We compared the effect of FK506 on permeabilized DT40 cells that express the three IP₃R isoforms and on R23-11 cells, which are genetically manipulated DT40 cells deficient for all three IP₃R isoforms (Sugawara et al. 1997; Miyakawa et al. 1999). Addition of 50 μM FK506 after permeabilization of both cell strains led to a similar sustained [Ca²⁺] elevation amounting to 28 ± 6 % (n= 5) and 27 ± 7 % (n= 5) of the A23187-induced [Ca²⁺] rise for the DT40 and R23-11 cell strains, respectively (Fig. 2). As illustrated in Fig. 2C, IP₃ could not induce Ca²⁺ release in permeabilized R23-11 cells. This indicates that other pathways different from the IP₃R mainly mediated the FK506-induced [Ca²⁺] rise. The very similar [Ca²⁺] elevation observed for R23-11 cells and the parent DT40 cells seems to exclude a major contribution of Ca²⁺-induced Ca²⁺ release mechanisms via intracellular Ca²⁺ release channels, particularly since these cells also did not show caffeine-induced Ca²⁺ release (data not shown). On the other hand, thapsigargin could empty the R23-11 Ca²⁺ stores and provoked store-operated Ca²⁺ entry (Sugawara et al. 1997). This may indicate that a pathway for rapid store depletion is still operative in these triple IP₃R knockout cells.

Figure 2. FK506-induced [Ca²⁺] rise in permeabilized DT40 and R23-11 cells.

FK506 (50 μM)-induced [Ca²⁺] rise in permeabilized DT40 (A) and R23-11 (B) cells. Typical traces out of 5 experiments are shown. C shows the absence of IP₃-induced Ca²⁺ release in R23-11 cells. The experimental procedure was identical to that described in Fig. 1A.

Identity of the FK506-sensitive Ca²⁺ store

There are different intracellular Ca²⁺ stores, which could be modulated by FK506. We first investigated the effect of FK506 on mitochondrial Ca²⁺ accumulation in permeabilized SH-SY5Y cells. First, preferential Ca²⁺ uptake in the mitochondria was obtained by application of thapsigargin (10 μM), which blocked any Ca²⁺ uptake in the ER and addition of spermine (1 mM), which activated the mitochondrial Ca²⁺ uniporter (Lenzen et al. 1992). The mitochondrial set point for Ca²⁺ is about 1 μM, which correlates with the low affinity (K_d∼1-5 μM) of the mitochondrial Ca²⁺ uniporter (Duchen, 1999). According to the characteristics of the mitochondrial Ca²⁺ uniporter, optimal Ca²⁺ uptake in the mitochondria occurred in the presence of 1 mM spermine. In these experimental conditions with thapsigargin blocking the high-affinity ER Ca²⁺ pump, the equilibrium [Ca²⁺] never dropped below 1 μM. From Fig. 3A, it is evident that the [Ca²⁺] rise induced by FK506 (50 μM) did not originate from the mitochondria. Second, FK506 was still able to induce a sustained [Ca²⁺] elevation, when Ca²⁺ was accumulated in the absence of thapsigargin and when the mitochondrial Ca²⁺ accumulation was inhibited by mitochondrial inhibitors, such as azide (10 mM). In Fig. 3B, intact cells were pre-treated during 10 min with azide (10 mM NaN₃) in the extracellular medium. After permeabilization of the plasma membrane and reaching the steady-state [Ca²⁺], azide (10 mM) was added during the experiment to confirm that the mitochondria did not contain any Ca²⁺. The ‘hump’ in the free [Ca²⁺] after application of azide could be due to a small amount of contaminating Ca²⁺ in the azide solution and was also observed in earlier reports (Missiaen et al. 1992b). FK506 (50 μM) still induced a [Ca²⁺] rise. Similar results were obtained with ruthenium red (4 μM) and carbonyl cyanide p-(trifluoromethoxy)phenylhydrazone (FCCP, 20 μM; data not shown).

Figure 3. Source of the FK506-induced [Ca²⁺] rise.

A, accumulation of Ca²⁺ in the mitochondrial compartment of permeabilized SH-SY5Y cells in the presence of 10 μM thapsigargin and 1 mM spermine. Addition of FK506 (50 μM), IP₃ (0.1 μM) and A23187 (8 μM) are indicated. B, accumulation of Ca²⁺ in non-mitochondrial Ca²⁺ stores of permeabilized SH-SY5Y cells, pre-treated with 10 mM NaN₃ 10 min before digitonin addition. As a control on its inhibitory effect, NaN₃ (10 mM) was again administered in the course of the experiment, 2 min before FK506 (50 μM). A23187 (8 μM) was given at the end of the experiment. Typical for 3 experiments.

Taken together, these results suggest that the ER, but not the mitochondria, was the most likely source for the FK506-induced [Ca²⁺] rise.

Effect on SERCA Ca²⁺ pumps

To probe the effect of FK506 on Ca²⁺ pump activity, cells were permeabilized with digitonin in the presence of 20 mM oxalate. Oxalate immobilizes luminal Ca²⁺ and thus enables the measurement of unidirectional Ca²⁺ uptake properties of permeabilized SH-SY5Y cells. After reaching the steady-state [Ca²⁺], a 10 nmol pulse of CaCl₂ was added to the medium (Fig. 4A). This caused an immediate increase in fluorescence, followed by an exponential decrease in resting [Ca²⁺] due to the Ca²⁺ pump activity. After four Ca²⁺ pulses, either ethanol (0.2 %), FK506 (10-100 μM) or thapsigargin (1 μM) was added. Three minutes after this addition the cells were again challenged with four pulses of 10 nmol CaCl₂. These four pulses were averaged and normalized. The mean rate of Ca²⁺ uptake (n= 4) was compared before and after addition of FK506.

Figure 4. Effect of FK506 on the uptake of exogenously added Ca²⁺ in permeabilized SH-SY5Y cells under oxalate-facilitated uptake conditions.

A, the effect of FK506 (50 μM) on Ca²⁺ uptake was measured by adding 4 pulses of exogenous Ca²⁺ (10 nmol) before and after FK506 application. The fluorescence signal is indicated in arbitrary units (A.U.), since it was difficult to estimate the [Ca²⁺] accurately in the presence of 20 mM oxalate. B, the 4 pulses before and after FK506 were normalized, averaged and superimposed. The mean decay (n= 4) of the [Ca²⁺] in control conditions and after addition of FK506 (50 μM) is plotted. This Ca²⁺ decay can be fitted by a monoexponential function (y= e^−t/τ), which resulted in a significant increase in the time constant τ of ≈35 % (P < 0.01 by Student's paired t test). Typical for 3 experiments. The inset shows the effect of the vehicle (0.1 % ethanol) on the uptake of exogenously added Ca²⁺ (average of 4 pulses of 10 nmol Ca²⁺). C, the effect of thapsigargin (4 nM or 1 μM) on the uptake of exogenously added Ca²⁺ (average of 4 pulses of 10 nmol Ca²⁺).

The normalized uptake of the exogenously added Ca²⁺ (10 nmol) was fitted by a monoexponential function (y= e^−t/τ). Up to 10 μM FK506 did not significantly inhibit the uptake rate of exogenously added Ca²⁺, but 50 μM FK506 had a significant inhibitory effect on the uptake rate (τ⁻¹) of the exogenously added Ca²⁺ (Fig. 4B) and caused an increase in τ of 35 ± 5 % (mean ±s.e.m., n= 3). At 100 μM, FK506 caused an increase in τ of 46 ± 7 % (mean ±s.e.m., n= 3). A similar inhibition of the SERCA pump occurred in R23-11 cells (data not shown). Furthermore, these changes in transient kinetics by FK506 were clearly absent in the presence of the vehicle alone (see inset Fig. 4B), which indicates that FK506, but not the steady increase of the ER Ca²⁺ content, was responsible for the reduced Ca²⁺ uptake rate. Moreover, 4 nM thapsigargin caused a similar increase in τ of ∼36 % as observed for 50 μM FK506 (Fig. 4C). This concentration of thapsigargin elevated the free [Ca²⁺] to a value corresponding to ∼16 % of the A23187-induced [Ca²⁺] rise, which is also very similar to the [Ca²⁺] rise induced by 50 μM FK506 (Fig. 1B). In the presence of 1 μM thapsigargin, accumulation of the exogenously added Ca²⁺ into the ER Ca²⁺ stores was completely inhibited (Fig. 4C).

For a quantitative assessment of the inhibitory effect of FK506 on the initial Ca²⁺ uptake rate mediated by the SERCA pump, we also performed ⁴⁵Ca²⁺ uptake experiments in SH-SY5Y microsomes in the presence of oxalate and mitochondrial inhibitors using a rapid filtration technique (Fig. 5A). The Ca²⁺ uptake rate at a free [Ca²⁺] of 3.16 μM in control conditions was 1.3 ± 0.2 nmol mg⁻¹ min⁻¹ (n= 17), determined by linear regression. The thapsigargin-insensitive Ca²⁺ uptake activity represented 10 ± 6 % (n= 9) of the Ca²⁺ uptake in control conditions. Figure 5A also shows the time-dependent uptake in the presence of 100 μM FK506 (0.22 nmol mg⁻¹ min⁻¹) or 10 μM thapsigargin (0.08 nmol mg⁻¹ min⁻¹). No ⁴⁵Ca²⁺ uptake was detected in the absence of ATP. FK506 inhibited the rate of the thapsigargin-sensitive part of the Ca²⁺ uptake in SH-SY5Y microsomes with an IC₅₀ of about 19 μM (Fig. 5B).

Figure 5. Effect of FK506 on oxalate-facilitated ⁴⁵Ca²⁺ uptake in SH-SY5Y microsomes.

A, the ATP-driven ⁴⁵Ca²⁺ uptake at a free [Ca²⁺] of 3.16 μM was measured in control conditions (in the presence of 0.2 % ethanol; •), in the presence of 100 μM FK506 (▪), 10 μM thapsigargin (▴) or without ATP (▾). The vehicle, FK506 or thapsigargin was added 5 min prior to ATP, which initiates ⁴⁵Ca²⁺ uptake. A typical experiment out of 4 experiments is shown. B, dose-response curve for the effect of FK506 on the thapsigargin-sensitive part of the initial rate of Ca²⁺ uptake (% of control). The initial Ca²⁺ uptake rate was obtained by linear regression through 4 time points each performed in duplicate. Each result is the mean of at least three independent experiments. s.e.m. is indicated, unless smaller than the symbol. Sigmoidal fitting was performed by using Microcal Origin (version 6.0) software.

A similar inhibitory effect of FK506 on Ca²⁺ uptake activity was found for cardiac homogenates, which contain the related SERCA2a Ca²⁺ pump. The dose-response curve in these cells was characterized by an IC₅₀ of about 39 μM (data not shown).

To verify whether the inhibition of the Ca²⁺ pump activity was accompanied by a lower activity of the Ca²⁺-Mg²⁺-ATPase, we measured the effect of FK506 on this Ca²⁺- Mg²⁺-ATPase activity in cardiac microsomes. We observed a partial inhibition (∼30 %) of the Ca²⁺-Mg²⁺-ATPase activity by 100 μM FK506. The Ca²⁺-Mg²⁺-ATPase activity in SH-SY5Y microsomes was below the detection limit of the assay (data not shown).

In order to elucidate whether the inhibition of Ca²⁺ uptake by FK506 represented a specific effect, we tested several related compounds with an immunosuppressive action (rapamycin, cyclosporin A and ascomycin) or with a macrocyclic lactone structure (rapamycin, ascomycin, ivermectin and midecamycin) (Fig. 6). Ivermectin (30 μM) was the most potent inhibitor of ATP-dependent Ca²⁺ uptake. At the same concentration, FK506, rapamycin and ascomycin inhibited the SERCA pump to a similar extent, while cyclosporin A and midecamycin were not effective.

Figure 6. Effect of functional and structural analogues of FK506 on ⁴⁵Ca²⁺ uptake in SH-SY5Y microsomes.

The initial Ca²⁺ uptake rate was obtained by linear regression as explained in Methods. The Ca²⁺ uptake rate in control conditions (0.1 % ethanol) was set as 100 %. All drugs were tested at a concentration of 30 μM. Each result is the mean of at least three independent experiments, each performed in duplicate. s.e.m. is indicated.

This inhibition of Ca²⁺ uptake activity by FK506 depended on the microsomal concentration, since 30 μM FK506 reduced the Ca²⁺ uptake rate to 45 ± 2 % of control (mean ±s.e.m., n= 6) when 200 μg ml⁻¹ SH-SY5Y microsomes were used. However, the same concentration of FK506 reduced the Ca²⁺ uptake rate to only 68 ± 5 % of control (mean ±s.e.m., n= 3) when 400 μg ml⁻¹ SH-SY5Y microsomes were used.

Effect of FK506 on IP₃-induced Ca²⁺ release

Although the major effect of FK506 on Ca²⁺ signalling was due to an inhibition of the SERCA pumps, we cannot exclude that FK506 may in addition alter Ca²⁺ flux through the IP₃R (Cameron et al. 1995b). We therefore investigated whether low FK506 concentrations that did not affect the SERCA pumps could still modulate the intracellular Ca²⁺ channels. We measured the effect of FK506 on IP₃-induced Ca²⁺ release in permeabilized SH-SY5Y cells. After permeabilization of the cells and a 5 min pre-incubation with FK506 (or 0.1 % ethanol) IP₃ (0.1 μM) was added to the intracellular medium. IP₃ caused an immediate, transient increase in the free cytosolic [Ca²⁺]. We have determined in a large number of observations that the IP₃-induced peak [Ca²⁺] in control conditions (19 ± 3 % of the A23187-releasable Ca²⁺, n= 11) did not differ from the IP₃-induced peak [Ca²⁺] released in the presence of 10 μM FK506 (19 ± 4 % of the A23187-releasable Ca²⁺, n= 19) (Fig. 7A). Similar results were obtained in permeabilized undifferentiated BC₃H1 cells, which also exhibit predominantly IP₃-induced Ca²⁺ release (De Smedt et al. 1991; data not shown).

Figure 7. Effect of FK506 on IP₃-induced Ca²⁺ release in permeabilized SH-SY5Y or permeabilized A7r5 cells.

A, SH-SY5Y permeabilization and Ca²⁺ uptake were as described in Fig. 1A. After reaching a stable steady-state level for Ca²⁺, either FK506 (10 μM) or ethanol (0.1 %) were added (first arrow), 5 min before administration of IP₃ (0.1 μM; second arrow). Two separate experiments out of more than 10 are superimposed. B, permeabilization and loading of A7r5 monolayers was as described in Methods. The non-mitochondrial Ca²⁺ stores, loaded to steady state with ⁴⁵Ca²⁺, released Ca²⁺ passively when incubated in efflux medium in the continuous presence of 0.1 % ethanol (○ and □) or 10 μM FK506 (• and ▪) for 10 min, at which time 1 μM IP₃ (○ and •) or 10 μM A23187 (□ and ▪) were added for 2 min, as indicated by the horizontal bar. Ca²⁺ release was plotted as the fractional loss, i.e. the amount of Ca²⁺ leaving the stores in 2 min divided by the total store Ca²⁺ content at that time. The result is the mean of three experiments.

Permeabilized monolayers of A7r5 cells are a well-characterized model system which allows a very accurate analysis of IP₃-induced Ca²⁺ release in unidirectional flux conditions (Missiaen et al. 1992a). As shown in Fig. 7B, the increase in fractional loss in the presence of 1 μM IP₃ (i.e. the amount of Ca²⁺ released by IP₃ in 2 min divided by the total store Ca²⁺ content at that time) did not differ significantly between control cells and 10 μM FK506- treated A7r5 cells.

Taken together, these data indicate that FK506 did not affect IP₃-induced Ca²⁺ release in permeabilized SH-SY5Y cells, undifferentiated BC₃H1 cells or A7r5 cells.

Effect of FK506 on caffeine-induced Ca²⁺ release

To validate our assay, we controlled whether FK506 could indeed potentiate caffeine-induced Ca²⁺ flux through the RyR, as described in several reports (Timerman et al. 1993; Brillantes et al. 1994). Neither SH-SY5Y cells nor A7r5 cells demonstrate caffeine-induced Ca²⁺ release (data not shown). We therefore investigated the effect of FK506 in differentiated BC₃H1 cells, which predominantly express the skeletal muscle isoform type 1 RyR as a major Ca²⁺ release channel (De Smedt et al. 1991). In Fig. 8, caffeine-induced Ca²⁺ release in permeabilized differentiated BC₃H1 cells is shown in control conditions and in the presence of 10 μM FK506. FK506 potentiated the caffeine-induced Ca²⁺ release by a factor of 1.8 ± 0.2 (n= 3). This increase in Ca²⁺ release was not due to an inhibition of the SERCA pump, since under these conditions 10 μM FK506 had no significant effect on the Ca²⁺ uptake of exogenously added Ca²⁺ in the presence of oxalate (data not shown). Furthermore, the caffeine-induced Ca²⁺ release, both in the presence and in the absence of FK506, could be largely blocked by 100 μM ryanodine (see Fig. 8, inset) and 20 μM ruthenium red (data not shown), both inhibitors of the RyR.

Figure 8. Effect of FK506 on caffeine-induced Ca²⁺ release in permeabilized differentiated BC₃H1 cells.

Permeabilization and Ca²⁺ loading of BC₃H1 cells was similar to that described for SH-SY5Y cells in Fig. 1A. After reaching a stable steady-state level for Ca²⁺, either FK506 (10 μM) or ethanol (0.1 %) was added (first arrow), 3 min before administration of caffeine (7.5 mM; second arrow). The results were not normalized to the total releasable Ca²⁺, since it is known that caffeine can interact with fluorescent Ca²⁺ dyes (Muschol et al. 1999). This interaction distorts the accurate estimation of the total releasable Ca²⁺. Two separate experiments are superimposed. The figure is representative of three experiments. The inset shows the effect of 10 μM FK506 or vehicle (first arrow) on caffeine-induced Ca²⁺ release in the presence of 100 μM ryanodine.

DISCUSSION

Effects of FK506 on Ca²⁺ uptake mechanisms (SERCA) in permeabilized cells

FK506 induced a sustained rise in the free [Ca²⁺] in permeabilized SH-SY5Y cells. A similar [Ca²⁺] elevation was also observed in permeabilized R23-11 cells, which do not possess any functional IP₃R. This indicates that FK506 could induce a [Ca²⁺] rise independently of the IP₃R. This FK506-induced [Ca²⁺] rise was also independent of the RyR, since neither R23-11 nor SH-SY5Y cells expressed a functional RyR. Furthermore, the effect of FK506 on the sustained elevation of [Ca²⁺] was not dependent on Ca²⁺ sequestration in the mitochondria, but required Ca²⁺ uptake in the endoplasmic reticulum.

On the other hand, we clearly demonstrated that FK506 caused an inhibition of the SERCA Ca²⁺ pump activity. This inhibition of the Ca²⁺ uptake activity by FK506 was observed in permeabilized SH-SY5Y cells, SH-SY5Y microsomes and cardiac microsomes.

In order to maintain the steady-state cytosolic [Ca²⁺], the intracellular Ca²⁺ stores have to continuously resequester the Ca²⁺ that passively leaks out of these stores. Interrupting this cycle of passive Ca²⁺ leak and active Ca²⁺ uptake by inhibiting the Ca²⁺ pumps with FK506 or thapsigargin leaves the Ca²⁺ efflux temporarily uncompensated, leading to a partial discharge of the stored Ca²⁺ from the ER. As a consequence, there will be a net Ca²⁺ efflux out of the Ca²⁺ stores until the new and higher steady-state cytosolic [Ca²⁺] is reached. This was observed as a sustained elevation of the medium [Ca²⁺]. Similarly, Thastrup et al. (1990) demonstrated that thapsigargin caused a sustained [Ca²⁺] rise by specifically inhibiting Ca²⁺ uptake through the SERCA pump. The pathway for Ca²⁺ efflux out of the intracellular Ca²⁺ stores is as yet unidentified, but is generally described as ‘passive leak’. It is important to note that experiments in permeabilized A7r5 monolayers, where the ‘passive leak’ can be measured in the absence of Ca²⁺ uptake activity, showed no increase of this ‘passive leak’ pathway by FK506 (authors’ unpublished results).

We have to point out that the concentration of FK506 required for inhibition of the Ca²⁺ uptake activity was rather high (>10 μM) compared with the effects of FK506 on other cellular processes. Concentrations in the nanomolar range are required for the inhibition of calcineurin and subsequent suppression of the interleukin2-dependent T cell activation (Schreiber, 1991) and stoichiometrical concentrations (<5 μM) are required for the dissociation of FKBP12 from the RyR (Timerman et al. 1993). Nevertheless, several reports demonstrated the effects of higher concentrations (10-50 μM) of FK506 on cellular Ca²⁺-dependent mechanisms, including the effects on voltage-operated Ca²⁺ channels in the porcine coronary artery (Yasutsune et al. 1999), elementary Ca²⁺ events in cardiac myocytes (Xiao et al. 1997), cardiac intracellular [Ca²⁺] and contraction (Valdivia, 1998) and voltage-operated Ca²⁺ channel-dependent long-term potentiation in the hippocampus (Onuma et al. 1998).

Furthermore, FK506 (up to 20 μM) increased the cytosolic [Ca²⁺] of intestinal smooth muscle cells (Bielefeldt et al. 1997). The Ca²⁺ originated from the ryanodine-sensitive stores, since depletion of these stores with caffeine abolished the effect. These data indicate that FK506 had at least two effects resulting in a cytosolic [Ca²⁺] increase. Besides the effect of low concentrations of FK506 (<5 μM) on the RyR, there might be an additional effect on the SERCA Ca²⁺ pump activity, since the [Ca²⁺] increase did not saturate at the stoichiometrical concentrations of FK506 (<5 μM) needed for dissociation of the RyR from FKBP12. Instead, higher concentrations of FK506 (20 μM) still enhanced the FK506-induced cytosolic [Ca²⁺] increase.

Our observations are also compatible with the attenuating effect of FK506 on ATP-induced Ca²⁺ oscillations in tracheal epithelium (Kanoh et al. 1999). If FK506 would potentiate Ca²⁺ flux through the IP₃R (Cameron et al. 1995b), an enhancement of the Ca²⁺ oscillations should have been observed. A reduced Ca²⁺ uptake activity by the ER in contrast would suppress the ATP-induced Ca²⁺ oscillations.

A recent report showed that compounds with a macrocyclic lactone ring structure (e.g. ivermectin) could reduce SERCA-mediated Ca²⁺ uptake in terminal cisternae of skeletal muscle in an FKBP12-independent way (Ahern et al. 1999). Taken together, the relative high concentrations needed and the observation that other macrocyclic compounds also inhibit Ca²⁺ uptake, suggest that the observed inhibition of the SERCA pump by FK506 is non-specific and FKBP12 independent.

Effects of FK506 on intracellular Ca²⁺ release channels (IP₃R and RyR) in permeabilized cells

Evidence has been presented for a potentially important role of the immunophilin FKBP12 in the regulation of intracellular Ca²⁺ release through interaction with both the RyR and the IP₃R (Marks, 1996).

To study functional effects of FKBP12 on intracellular Ca²⁺-release channels, we investigated the effects of FK506 on Ca²⁺ release in permeabilized SH-SY5Y cells. This cell type expresses almost exclusively the neuronal IP₃R1 isoform and is therefore used as a model system for IP₃-induced Ca²⁺ release via this IP₃R isoform (Wojcikiewicz, 1995). We demonstrated that FKBP12 is also expressed in this cell type and that a significant fraction of the FKBP12 protein is retained in microsomal membranes.

FK506 had no effect on the IP₃-induced Ca²⁺ release in permeabilized SH-SY5Y, A7r5 cells and undifferentiated BC₃H1 cells which expressed the IP₃R as the predominant intracellular Ca²⁺ release channel. On the other hand, FK506 clearly stimulated caffeine-induced Ca²⁺ mobilization in differentiated BC₃H1 cells which expressed the RyR. This observation is in agreement with previous data indicating that dissociation of FKBP12 from the RyR by FK506 enhanced Ca²⁺ flux through the RyR (Brillantes et al. 1994). In addition, a recent report showed that compounds with a macrocyclic lactone ring structure could activate RyRs directly independently of FKBP12 (Ahern et al. 1999).

Cameron et al. (1995b) showed a reduced accumulation of Ca²⁺ into IP₃-sensitive stores when cerebellar microsomes were treated with FK506. Dissociation of FKBP12 from the IP₃R was found to change the phosphorylation state of this receptor and increase its sensitivity towards IP₃ (Cameron et al. 1995a). We did not find a potentiating effect of FK506 on IP₃-induced Ca²⁺ release in the investigated cell types whereas such an effect was clearly evident for caffeine-induced Ca²⁺ release in RyR-expressing cells. It should be pointed out, however, that the proposed FKBP12 interaction site on IP₃R1 is located in close proximity to a number of other potential regulatory sites, such as the SII-splice site, a Ca²⁺-binding site, a calmodulin-binding site, a PKA-phosphorylation site and an ATP-binding site (Patel et al. 1999). It can therefore not be excluded that more subtle effects, potentially dependent on other modulatory compounds of the IP₃R, were not present in our assay. It is, however, clear that the major effects of FK506 on intracellular Ca²⁺ signalling could not be attributed to a direct or indirect effect on the IP₃R.

Pharmacological and physiological implications

FK506 is an immunosuppressant widely used in primary and rescue therapy for various organ transplantations (e.g. liver, kidney, heart, lung, pancreas and intestine; Spencer et al. 1997). However, treatment of patients with FK506 is often associated with side effects, e.g. renal, hepatic and metabolic toxicity (Spencer et al. 1997).

The concentrations needed for inhibition of calcineurin activity (Schreiber, 1991) and subsequent suppression of allograft rejection in organ transplants are in the nanomolar range (1-25 nM) (Venkataramanan et al. 1987). The concentrations required for affecting Ca²⁺ uptake activity are at least 1000-fold higher than the concentrations needed for immunosuppression. Nevertheless, due to the lipophilic structure of FK506 (Tanaka et al. 1987) and its repetitive application during therapy in allograft patients (Spencer et al. 1997), it may accumulate in cellular membranes. It was reported that FK506 has the capacity to accumulate in T lymphocytes and concentrate up to 900-fold the extracellularly added concentration (Dumont et al. 1994). This intracellular accumulated FK506 can lead to a reduction in Ca²⁺ accumulation in the ER of cells and subsequently to dysfunction.

For instance, a hypertrophic cardiomyopathy was associated with application of FK506 in paediatric patients (Atkison et al. 1995). Another report about paediatric liver transplantation demonstrated heart failure during application of FK506, which was resolved when treatment with FK506 was discontinued (Whitington et al. 1994). It is interesting to note that animal models, with dilated cardiomyopathy, had a reduced Ca²⁺ uptake and binding activity of the SR (Edes et al. 1991) and humans with idiopathic dilated cardiomyopathy had a decreased SR Ca²⁺-uptake rate (Limas et al. 1987) and Ca²⁺-Mg²⁺-ATPase activity.

In conclusion, our findings demonstrate that the immunosuppressant FK506 reduced Ca²⁺ uptake in the endoplasmic Ca²⁺ stores via an FKBP12-independent inhibition of the SERCA pump. Furthermore, FK506 did not affect Ca²⁺ flux through the IP₃R in permeabilized SH-SY5Y and A7r5 cells. On the other hand, FK506 potentiated caffeine-induced Ca²⁺ release through the RyR in permeabilized differentiated BC₃H1 cells. Our findings may explain some side effects of FK506 during treatment of allograft transplant patients.