Cytosolic Ca2+ signals are often amplified by massive calcium release from the endoplasmic reticulum (ER). This calcium-induced calcium release (CICR) occurs by activation of an ER Ca2+ channel, the ryanodine receptor (RyR), which is facilitated by both cytosolic- and ER Ca2+ levels. Caffeine sensitizes RyR to Ca2+ and promotes ER Ca2+ release at basal cytosolic Ca2+ levels. This outcome is frequently used as a readout for the presence of CICR. By monitoring ER luminal Ca2+ with the low-affinity genetic Ca2+ probe erGAP3, we find here that application of 50 mM caffeine rapidly reduces the Ca2+ content of the ER in HeLa cells by ∼50%. Interestingly, this apparent ER Ca2+ release does not go along with the expected cytosolic Ca2+ increase. These results can be explained by Ca2+ chelation by caffeine inside the ER. Ca2+-overloaded mitochondria also display a drop of the matrix Ca2+ concentration upon caffeine addition. In contrast, in the cytosol, with a low free Ca2+ concentration (10−7 M), no chelation is observed. Expression of RyR3 sensitizes the responses to caffeine with effects both in the ER (increase in Ca2+ release) and in the cytosol (increase in Ca2+ peak) at low caffeine concentrations (0.3–1 mM) that have no effects in control cells. Our results illustrate the fact that simultaneous monitoring of both cytosolic- and ER Ca2+ are necessary to understand the action of caffeine and raise concerns against the use of high concentrations of caffeine as a readout of the presence of CICR.

Introduction

In many excitable cells, the cytosolic Ca2+ signals generated by Ca2+ entry through the plasma membrane are amplified by Ca2+ release from the endoplasmic reticulum (ER) via an autocatalytic process known as Ca2+-induced Ca2+ release (CICR) [1]. In this process, Ca2+ itself is a potent activator of endomembrane Ca2+ channels, mainly the ryanodine receptors (RyRs). Mammalian tissues express three isoforms: RyR1, RyR2, and RyR3, each encoded by a different gene. RyR1 and RyR2 are predominantly expressed in the sarcoplasmic reticulum of skeletal muscle and heart, respectively, where they play an essential role to trigger muscle contraction [2,3]. RyR3 was originally identified in the brain [4], but it is also expressed in many other cell types, including non-excitable ones [5,6]. CICR can also be triggered by ER release induced by inositol trisphosphate [1] and can also amplify [Ca2+]C (cytosolic-free Ca2+ concentration) signals generated by Ca2+ release from other intracellular organelles, such as Golgi apparatus or the endolysosomal system [1,7].

RyRs are able to sense sudden increases of cytosolic Ca2+ concentration ([Ca2+]C) and transduce them into an increase in ER Ca2+ permeability that produces a massive Ca2+ release. Increased Ca2+ concentration in the ER lumen ([Ca2+]ER) does also facilitate activation of RyRs and Ca2+ release [8]. The second messenger cyclic ADP ribose (cADPR) increases the accumulation of Ca2+ into the ER and facilitates Ca2+ release from the store by sensitization of the RyR [7]. Caffeine sensitizes RyRs to both [Ca2+]C and [Ca2+]ER, and facilitates CICR [8]. At millimolar concentrations, caffeine produces ER Ca2+ release at basal [Ca2+]C, and this action has been used as a readout of the presence of RyRs in a given cell or tissue.

Caffeine is not a clean tool, as it may interfere with fluorescence of Ca2+ probes [9]. In addition, loading the cells with Ca2+ probes may decrease the responses to caffeine [10]. This has been attributed to the increase in Ca2+ buffering by the cytoplasm and to subtle perturbations of Ca2+ diffusion in the cytosol that result in dissipation of high Ca2+ microdomains [1013]. In the present paper, the role of Ca2+ chelation by caffeine in the ER lumen is examined.

Experimental procedures

Cell culture and gene transfection

The stably transfected HeLa clones expressing ER- targeted GAP3 (erGAP3) or ER-targeted GAP1 (IgGAP1) have been previously described [14,15]. Stable HeLa clones expressing mutated mitochondrially targeted (D119A) GFP (green fluorescent protein)-aequorin (mitmutGA) were generated by Lipofectamine transfection of HeLa cells followed by 0.8 mg/ml G-418 selection. Single-cell clones were selected by limited dilution. All the HeLa clones were maintained in DMEM-GlutaMAX medium (GIBCO) supplemented with 100 μg/ml streptomycin and 100 U/ml penicillin, 10% (v/v) fetal bovine serum and 0.2 mg/ml G-418, at 37°C under 5% CO2. For imaging experiments, erGAP3-expressing cells were seeded on poly-l-lysine-coated 12 mm-diameter coverslips at a density of 4 × 104 cells. For bioluminescence experiments, IgGAP1- or mitmutGA-expressing cells were seeded on poly-l-lysine-coated four-well plates at a density of 7 × 104 cells/well. In experiments shown in Figures 5 and 6, erGAP3 HeLa cells were transfected with 1 μg rabbit smooth muscle RyR3 cDNA cloned into pcDNA3 (kindly donated by Dr S.R.W. Chen, University of Calgary, Canada) using Lipofectamine 2000 (Invitrogen). In some experiments (Figure 6), RyR3 (1 µg) was cotransfected with the pmCherry-N3 (Clontech, 0.01 µg).

Western blot analysis

HeLa cells transfected wit RyR3- or control empty vector pcDNA were grown for 24–36 h, washed with ice-cold phosphate-buffered saline (PBS), and extracted with RIPA buffer (50 mM Tris–HCl, pH 7.4; 150 mM NaCl; 1% Triton X-100; and 0.1% SDS and 0.5% deoxycholate) for 30 min. After centrifugation at 13 000 rpm for 5 min, the supernatant was diluted in Laemmli sample buffer and 50 µg of protein samples were separated by PAGE. The membrane was first probed overnight with a mouse anti-RyR1 antibody (1 : 5000; ThermoFisher) followed by a horseradish peroxidase-labelled secondary antibody (1 : 1000; Bio-Rad) and incubated for 1 h. To control for protein loading, membrane was probed with a mouse anti-tubulin antibody (1 : 5000; Sigma). Sample proteins were quantified by the Bradford assay.

Immunofluorescence

Stable HeLa clone-expressing erGAP3 (see below) were seeded on 12 mm coverslips and transiently transfected with RyR3 cDNA. Cells were fixed for 24–36 h with 4% PFA in PBS for 20 min; 10% normal goat serum was added for blocking nonspecific binding sites. Expression of RyR3 was detected by incubating the mouse anti-RyR1 antibody (1 : 200; ThermoFisher) diluted in PBS and containing 10% goat serum, overnight at 4°C. After washing with PBS, the secondary Alexa Fluor 568-conjugated antibody (1 : 200; Molecular Probes) was added and incubated for 60 min at 22°C. Cultures were washed with PBS three times and mounted in Vectashield (Vector). GAP was detected as green fluorescence (excited at 470/40 nm and filtered at 540/50 nm) in a Zeiss Axioplan Z microscope equipped with a 63×/1.2w Korr objective and an AxioCam MR camera. The red fluorescence was excited at 560/40 and filtered at 605/50 nm. The Zeiss ApoTomeR system was used for optical sectioning and images were analyzed with AxioVision and Image J software.

Fluorescence Ca2+ imaging

For simultaneous imaging [Ca2+]ER and [Ca2+]C, HeLa cells stably expressing erGAP3 Ca2+ indicator were loaded with Rhod-3 by incubating with Rhod-3/AM (2 µM, Molecular Probes) for 1 h at 22°C. Cells were imaged in a Zeiss axioplan upright microscope equipped with a water immersion 25× objective (Plan-Neofluar, Zeiss; NA = 0.8) and sequentially excited at 402, 470 and 546 nm; fluorescence was read at >515 nm (535DF35) and >590 nm (LP590). GAP was excited using the two filters, ET402/15x and 470/35 DF, and a 505DRLP dichroic mirror. Cells were alternately epi-illuminated at 402 and 470 nm, and light emitted above 505 nm was recorded using a Carl Zeiss AxioCam 12 bit camera handled by the AxioVision 4.6.3 software. Output images were background-subtracted and ratioed pixel-to-pixel using ImageJ software. The ratio F470/F402 was used as an index of [Ca2+]ER. The fluorescence excited at 546 nm (F546) was read using a FT580 dichroic mirror and a LP509 emission filter. F546 was expressed as F/F0, as an index of [Ca2+]C. F0 was the average of the fluorescence values obtained during the first 5–10 frames. The cells were under continuous perfusion (5–6 ml/min) at 22–25°C, with extracellular-like solution (ELS) with the following composition (in mM): NaCl, 145; KCl, 5; MgCl2, 1; CaCl2, 1; glucose, 10; sodium-HEPES, 10, pH 7.4. If not specified otherwise, all stimuli were diluted in ELS. Other details were as described previously [14,16,17].

Aequorin luminescence Ca2+ measurements

For measuring [Ca2+]ER by bioluminescence, cells expressing the ER-targeted low-affinity Ca2+ probe IgGAP1 [15] were incubated for 10 min at 22°C in Ca2+-free ELS containing 0.5 mM EGTA and 10 µM of the reversible sarco–endoplasmic reticulum Ca2+-ATPase (SERCA) inhibitor 2,5-di-tert-butylhydroquinone (TBH) [18]. GAP1 aequorin was then reconstituted by incubation with 1 µM coelenterazine n in the same medium during 1 h, prior to measurements. Cells were then placed in a purpose-built luminometer (Cairn Research, U.K.) and perfused with ELS containing 1 mM CaCl2, at 5 ml/min, to load the ER Ca2+ stores. At the end of each run, cells were lysed with 0.1 mM digitonin dissolved in 10 mM CaCl2, in order to release all the residual aequorin luminescence. The total luminescence (LTOTAL) was calculated by adding up all the L-values from each time point up to the end of the experiment. Data are shown as the ratio between luminescence (L) to the total luminescence at each time point (L/LTOTAL). The [Ca2+]ER was estimated by interpolation in the calibration curve [15].

In the experiments of mitochondrial Ca2+ (Figure 4), HeLa cells expressing mitochondrially targeted mutated (D119A)-GFP-aequorin (mutmitGA) [19] were reconstituted with coelenterazine n as described for the ER. The reconstitution medium also contained 10 µM of the SERCA inhibitor TBH in order to prevent refilling of the Ca2+ stores. Intracellular-like solution (ILS) was used in the permeabilized experiments. It had the following composition (in mM): KCl, 140; K2HPO4/KH2PO4, 1; MgCl2, 1; Mg-ATP, 1; sodium succinate, 2; sodium-pyruvate, 1; K-HEPES, 20; pH 7.2. The cells were first permeabilized by perfusion with 50 µM digitonin in ILS containing 0.5 mM EGTA and 10 µM TBH during 1 min. The solution was then switched to ILS without digitonin containing 50 µM free Ca2+. This solution was made by blending titrated solutions of EDTA-Mg2+ and EDTA-Ca2+ in the required amounts, calculated using the program MaxChelator [20]. The calibration of L/LTOTAL into [Ca2+]M (Ca2+ concentration inside mitochondria) was calculated according to the published constant's values previously published [21]. All the measurements were performed at 22°C.

Results

The activity of RyRs visualized by the cytosolic transient induced by caffeine is variable in different cell types [10]. For example, chromaffin cells [22] respond with a fast and sharp [Ca2+]C increase to a caffeine challenge (Figure 1A), whereas cerebellar granule neurons show a smaller response, which can be detected with Fluo-3 (Figure 1B), but not with Fura-2 (Figure 1C) [10]. Even though RyRs are mainly expressed in excitable cells, some non-excitable cells are also able to express RyR [5]. Figure 1D illustrates the [Ca2+]C peak to caffeine in Human Embryonic Kidney 293 (HEK 293) cells loaded with fluo-3. In contrast, when these cells were loaded with Fura-2, they do not show an increase in [Ca2+]C in response to caffeine stimulation (Figure 1E), similarly to granule neurons (compare to Figure 1C). The [Ca2+]C response to the IP3 (inositol 1,4,5-trisphosphate)-producing physiological agonist ATP was larger and faster than that to caffeine. Direct measurements of the changes in [Ca2+]ER with ER-targeted aequorin (Figure 1E) were consistent with the [Ca2+]C results, demonstrating that the decrease in [Ca2+]ER on stimulation with ATP is faster than the one induced by caffeine.

Comparison of cytosolic Ca2+ ([Ca2+]C) increases induced by caffeine in different cell types.

Figure 1.
Comparison of cytosolic Ca2+ ([Ca2+]C) increases induced by caffeine in different cell types.

(A) Bovine chromaffin cells (CRF); (B,C) cerebellar granule neurons (GN); (DF) HEK293T cells (HEK). Cells were stimulated with: depolarizing high K (70 mM; K); 50 mM caffeine (CAF); or 10 µM ATP (ATP), as shown. [Ca2+]C was measured either with Fura-2 (A, C, E, and F; loading with 4 µM Fura-2/AM), Fluo-3 (B, D; loading with 1 µM Fluo-3/AM), or aequorin (F; cells expressing the ER-targeted aequorin were reconstituted with 1 µM coelenterazine n in Ca2+-free medium for 60–90 min. Experiment was started by refilling the ER with 1 mM Ca2+). Temperature, 22°C. F/F0: Fluo-3 fluorescence expressed as fraction of the baseline value. Reproduced with permission from ref. [10].

Figure 1.
Comparison of cytosolic Ca2+ ([Ca2+]C) increases induced by caffeine in different cell types.

(A) Bovine chromaffin cells (CRF); (B,C) cerebellar granule neurons (GN); (DF) HEK293T cells (HEK). Cells were stimulated with: depolarizing high K (70 mM; K); 50 mM caffeine (CAF); or 10 µM ATP (ATP), as shown. [Ca2+]C was measured either with Fura-2 (A, C, E, and F; loading with 4 µM Fura-2/AM), Fluo-3 (B, D; loading with 1 µM Fluo-3/AM), or aequorin (F; cells expressing the ER-targeted aequorin were reconstituted with 1 µM coelenterazine n in Ca2+-free medium for 60–90 min. Experiment was started by refilling the ER with 1 mM Ca2+). Temperature, 22°C. F/F0: Fluo-3 fluorescence expressed as fraction of the baseline value. Reproduced with permission from ref. [10].

We have recently developed a new family of fluorescent genetically encoded Ca2+ indicators dubbed GAP (GFP-Aequorin Protein), optimized for [Ca2+] measurements in organelles [23,24]. One of the main advantages of the GAP family is that it permits double-wavelength ratiometric measurements. This ability makes possible not only detecting Ca2+ transients but also estimations of the basal Ca2+ concentrations. In addition, the affinity of the GAP sensors for Ca2+ can be tailored for environments with different Ca2+ concentrations. For example, GAP3, with a Kd = 489 µM and scarcely sensitive to Mg2+ and pH, is optimal for measurements in high [Ca2+] environments such as the lumen of ER [14] or the Golgi apparatus [25].

Measurements in a HeLa cell line expressing erGAP3 stimulated with a maximal concentration of ATP are illustrated in Figure 2. The upper traces (A and B) show the time-course of the two individual fluorescence traces, at 402 and 470 nm. The fluorescence excited at the second wavelength increases with [Ca2+], while the first one decreases [14,24]. Note the specular behavior of both wavelengths when ER is emptied by stimulation with the IP3-producing agonist ATP (Figure 2A,B). The signal is calibrated by a complete discharge of the ER Ca2+ store, by adding a second ATP pulse in Ca2+-free medium containing the SERCA inhibitor TBH. Trace C in Figure 2 illustrates the changes in the F470/F402 ratio, which is proportional to the [Ca2+]ER, and has been normalized here by dividing the values by the baseline fluorescence (R/R0, Figure 2C). Cytosolic Ca2+ concentration ([Ca2+]C) can be simultaneously measured by loading the erGAP3-expressing cells with a red cytosolic Ca2+ probe such as Rhod-3 [26]. The trace, normalized against the baseline fluorescence (F/F0), is shown in Figure 2D. Simultaneous measurement of Ca2+ in the cytosol and in the ER allows precisely studying the temporal relationship between [Ca2+]C and [Ca2+]ER in order to draw mechanistic conclusions. We observe, for example, that the onset of the decrease in [Ca2+]ER coincides with the increase in [Ca2+]C, consistent with the cytosolic Ca2+ being released from the ER. It is also noticeable that the [Ca2+]C increase reaches its peak prior to [Ca2+]ER reaches its minimum (Figure 2C,D). This result may seem surprising at first glance, but it can be easily rationalized by considering the fact that the [Ca2+]C peak only reveals that Ca2+ clearance from the cytosol exceeds Ca2+ release from the stores, and this may occur as soon as the rate of release begins to decrease, even though net Ca2+ exit from the ER continues.

Simultaneous fluorimetric measurements of [Ca2+]ER and [Ca2+]C in HeLa cells.

Figure 2.
Simultaneous fluorimetric measurements of [Ca2+]ER and [Ca2+]C in HeLa cells.

Cells expressing erGAP3 and loaded with Rhod-3 were challenged first with 100 µM ATP. The same stimulus was applied at the end of the experiment in Ca2+-free medium containing 10 µM of the SERCA blocker TBH to empty completely the ER calcium stores. Traces are the average values of 39 cells present in the same field and are representative of 10 similar experiments. (A,B) The changes of the individual fluorescences excited at 402 and 470 nm, respectively. The fluorescence values have been normalized by expressing them as F/F0. (C) Ratio of the two GAP fluorescences (F470/F402), which is proportional to [Ca2+]ER. These values are normalized by expressing them as fraction of the baseline value (R/R0). (D) Rhod-3 fluorescence, normalized as F/F0. This value is proportional to [Ca2+]C.

Figure 2.
Simultaneous fluorimetric measurements of [Ca2+]ER and [Ca2+]C in HeLa cells.

Cells expressing erGAP3 and loaded with Rhod-3 were challenged first with 100 µM ATP. The same stimulus was applied at the end of the experiment in Ca2+-free medium containing 10 µM of the SERCA blocker TBH to empty completely the ER calcium stores. Traces are the average values of 39 cells present in the same field and are representative of 10 similar experiments. (A,B) The changes of the individual fluorescences excited at 402 and 470 nm, respectively. The fluorescence values have been normalized by expressing them as F/F0. (C) Ratio of the two GAP fluorescences (F470/F402), which is proportional to [Ca2+]ER. These values are normalized by expressing them as fraction of the baseline value (R/R0). (D) Rhod-3 fluorescence, normalized as F/F0. This value is proportional to [Ca2+]C.

Figure 3 illustrates the effects of caffeine in HeLa cells on [Ca2+]C and [Ca2+]ER. The dynamics of the individual GAP3 wavelengths, F402 and F470, are shown in the upper panel (Figure 3A), whereas the F470/F402 ratio (proportional to [Ca2+]ER) and the Rhod-3 fluorescence (F/F0, proportional to [Ca2+]C) are shown in the lower panel (Figure 3B). The traces of the individual GAP3 fluorescences were specular, indicating that the changes are not artefactual. Moreover, bioluminescence measurements with aequorin also confirmed the [Ca2+]ER drop induced by caffeine (Supplementary Figure S1). Surprisingly, the caffeine-induced [Ca2+]ER decrease does not go along with changes in the [Ca2+]C (Rhod-3 trace in Figure 3B). Only an overshoot in [Ca2+]C due to the SERCA inhibition by TBH is detected. In contrast, stimulation with the IP3-producing agonist ATP/carbachol triggered a decrease in GAP3 fluorescence (ER Ca2+ release) along with a Rhod-3 fluorescence increase ([Ca2+]C increase). The simplest explanation for this discrepancy would be a chelation of Ca2+ inside the ER. This would not produce Ca2+ fluxes through the ER membrane and hence no changes in [Ca2+]C. Also, if caffeine chelated ER Ca2+ independently of activating RyRs, then the Ca2+ responses should be insensitive to RyR blockers. Indeed, ryanodine (10 µM) did not block the caffeine-induced [Ca2+]ER decrease (Supplementary Figure S2).

Comparison of the effects of caffeine (CAF, 50 mM) or IP3-producing agonists (ATP/CCh) on [Ca2+]ER (GAP3 fluorescence) and [Ca2+]C (Rhod-3 fluorescence) in HeLa cells.

Figure 3.
Comparison of the effects of caffeine (CAF, 50 mM) or IP3-producing agonists (ATP/CCh) on [Ca2+]ER (GAP3 fluorescence) and [Ca2+]C (Rhod-3 fluorescence) in HeLa cells.

(A) Individual GAP3 fluorescences. Stimulation protocol as shown: ATP/CCh stands for 100 µM ATP + 100 µM carbachol. (B) Comparison of the effects of caffeine and ATP on [Ca2+]ER (GAP3) and [Ca2+]C (Rhod-3). Values are the average traces of 29 cells present in the same field and are normalized to the baseline value as R/R0 or F/F0. Other details as in Figure 2. Results are representative of two independent experiments with the same protocol where a total of 66 cells were studied. The standard error of the mean (SEM) values of the GAP3 fluorescence in the different peaks ranged from 0.01 to 0.02. For the Rhod-3 fluorescence, the SEM in the different peaks oscillated between 0.04 and 0.07. The peak following stimulation by ATP/CCh expressed as delta F/F0 was (mean ± SEM, n = 66) 0.50 ± 0.07, whereas after the repeated stimulation with caffeine, the delta F/F0 was only 0.08 ± 0.04.

Figure 3.
Comparison of the effects of caffeine (CAF, 50 mM) or IP3-producing agonists (ATP/CCh) on [Ca2+]ER (GAP3 fluorescence) and [Ca2+]C (Rhod-3 fluorescence) in HeLa cells.

(A) Individual GAP3 fluorescences. Stimulation protocol as shown: ATP/CCh stands for 100 µM ATP + 100 µM carbachol. (B) Comparison of the effects of caffeine and ATP on [Ca2+]ER (GAP3) and [Ca2+]C (Rhod-3). Values are the average traces of 29 cells present in the same field and are normalized to the baseline value as R/R0 or F/F0. Other details as in Figure 2. Results are representative of two independent experiments with the same protocol where a total of 66 cells were studied. The standard error of the mean (SEM) values of the GAP3 fluorescence in the different peaks ranged from 0.01 to 0.02. For the Rhod-3 fluorescence, the SEM in the different peaks oscillated between 0.04 and 0.07. The peak following stimulation by ATP/CCh expressed as delta F/F0 was (mean ± SEM, n = 66) 0.50 ± 0.07, whereas after the repeated stimulation with caffeine, the delta F/F0 was only 0.08 ± 0.04.

Effects of caffeine (50 mM) on the [Ca2+]M in permeabilized HeLa cells.

Figure 4.
Effects of caffeine (50 mM) on the [Ca2+]M in permeabilized HeLa cells.

The plasma membrane of cells expressing GFP-aequorin targeted to the mitochondrial matrix (mitGA) [19,34] was permeabilized by treatment with 50 µM digitonin for 1 min and perfused with an ILS containing 100 nM Ca2+. The SERCA blocker TBH was added at 10 µM to avoid interferences with ER. At the time shown, perfusion of the cells was switched to ILS containing 50 µM Ca2+ to provoke mitochondrial calcium uptake through the mitochondrial calcium uniporter [17]. At the times shown, caffeine (50 mM) was added (CAF). Results are shown as luminescence/total luminescence (left ordinate axis), which is proportional to [Ca2+]M (approximate calibration at right ordinate axis). Experiment representative of three similar ones.

Figure 4.
Effects of caffeine (50 mM) on the [Ca2+]M in permeabilized HeLa cells.

The plasma membrane of cells expressing GFP-aequorin targeted to the mitochondrial matrix (mitGA) [19,34] was permeabilized by treatment with 50 µM digitonin for 1 min and perfused with an ILS containing 100 nM Ca2+. The SERCA blocker TBH was added at 10 µM to avoid interferences with ER. At the time shown, perfusion of the cells was switched to ILS containing 50 µM Ca2+ to provoke mitochondrial calcium uptake through the mitochondrial calcium uniporter [17]. At the times shown, caffeine (50 mM) was added (CAF). Results are shown as luminescence/total luminescence (left ordinate axis), which is proportional to [Ca2+]M (approximate calibration at right ordinate axis). Experiment representative of three similar ones.

Comparison of the responses to caffeine in control (A) and RyR3-expressing HeLa cells (B).

Figure 5.
Comparison of the responses to caffeine in control (A) and RyR3-expressing HeLa cells (B).

HeLa cells were transiently transfected with RyR3 cDNA. ER (GAP3, scale at left) and cytosolic Ca2+ transients (Rhod-3, scale at right). Responses to different caffeine concentrations (CAF; 0.3, 1, 10, and 50 mM) are shown. The effects of ATP (100 µM) or the same stimulus in Ca2+-free medium + TBH (to obtain maximum ER Ca2+ release) are included for comparison. The single-cell traces shown are representative of 25 untransfected cells (A) or 14 RyR3-transiently transfected cells (B). The SEM values oscillated between 0.02 and 0.04. Results are representative of seven (control) and five (RyR3-transfected) independent experiments with a total number of 274 cells (control) and 70 cells (RyR3-transfected) studied. Detailed quantification is summarized in Supplementary Table S1. (C) RyR3 expression in the transfected HeLa cells. Whole extracts obtained from control cells transfected with empty vector (control) or with RyR3 cDNA (+RyR) were subjected to Western blot with either a specific anti-RyR3 or an anti-tubulin antibody as a control of protein loading. (D) Inmunofluorescence of stably expressing erGAP3 HeLa cells transiently transfected with RyR3. Cells were fixed, permeabilized and immunostained with anti-RyR antibody; erGAP is shown in green and the RyR in red. Calibration mark, 10 µm.

Figure 5.
Comparison of the responses to caffeine in control (A) and RyR3-expressing HeLa cells (B).

HeLa cells were transiently transfected with RyR3 cDNA. ER (GAP3, scale at left) and cytosolic Ca2+ transients (Rhod-3, scale at right). Responses to different caffeine concentrations (CAF; 0.3, 1, 10, and 50 mM) are shown. The effects of ATP (100 µM) or the same stimulus in Ca2+-free medium + TBH (to obtain maximum ER Ca2+ release) are included for comparison. The single-cell traces shown are representative of 25 untransfected cells (A) or 14 RyR3-transiently transfected cells (B). The SEM values oscillated between 0.02 and 0.04. Results are representative of seven (control) and five (RyR3-transfected) independent experiments with a total number of 274 cells (control) and 70 cells (RyR3-transfected) studied. Detailed quantification is summarized in Supplementary Table S1. (C) RyR3 expression in the transfected HeLa cells. Whole extracts obtained from control cells transfected with empty vector (control) or with RyR3 cDNA (+RyR) were subjected to Western blot with either a specific anti-RyR3 or an anti-tubulin antibody as a control of protein loading. (D) Inmunofluorescence of stably expressing erGAP3 HeLa cells transiently transfected with RyR3. Cells were fixed, permeabilized and immunostained with anti-RyR antibody; erGAP is shown in green and the RyR in red. Calibration mark, 10 µm.

Dose–response plots of the caffeine-induced [Ca2+]ER decreases in HeLa cells.

Figure 6.
Dose–response plots of the caffeine-induced [Ca2+]ER decreases in HeLa cells.

Cells were cotransfected with RyR3 and Cherry-Fluorescent Protein (Cherry). Unicellular ROIs selected for Cherry-positive or -negative cells were averaged separately. (A) Average traces of 26 RyR3-transfected cells (red line) and 19 control cells (black line) in a representative experiment. Protocol includes sequential stimulation with 0.3, 1, 10, and 50 mM caffeine, followed by stimulation with ATP (100 µM), removal of external Ca (Ca0), and maximal ER Ca2+ discharge with ATP + Ca0 + TBH. (B) ER Ca2+ emptying (expressed as the decrease in R/R0) plotted against caffeine concentration. Max stands for maximal ER Ca2+ discharge. The mean values in RyR3-expressing (red) or control (black) cells correspond to experiment shown in A. The vertical bars stand for SEM values. Similar results were obtained in four independent experiments with different transfections.

Figure 6.
Dose–response plots of the caffeine-induced [Ca2+]ER decreases in HeLa cells.

Cells were cotransfected with RyR3 and Cherry-Fluorescent Protein (Cherry). Unicellular ROIs selected for Cherry-positive or -negative cells were averaged separately. (A) Average traces of 26 RyR3-transfected cells (red line) and 19 control cells (black line) in a representative experiment. Protocol includes sequential stimulation with 0.3, 1, 10, and 50 mM caffeine, followed by stimulation with ATP (100 µM), removal of external Ca (Ca0), and maximal ER Ca2+ discharge with ATP + Ca0 + TBH. (B) ER Ca2+ emptying (expressed as the decrease in R/R0) plotted against caffeine concentration. Max stands for maximal ER Ca2+ discharge. The mean values in RyR3-expressing (red) or control (black) cells correspond to experiment shown in A. The vertical bars stand for SEM values. Similar results were obtained in four independent experiments with different transfections.

If changes of [Ca2+]ER were due to chelation by caffeine, then a similar decrease in luminal [Ca2+] should occur in other organelles with high Ca2+ content. We tested this prediction in mitochondria overloaded with Ca2+ by incubation of permeabilized cells with 50 µM [Ca2+]C (Figure 4). Under these conditions, the mitochondrial Ca2+ concentration ([Ca2+]M) reached 500–600 µM and perfusion with 50 mM caffeine decreased [Ca2+]M by ∼50%. This outcome indicates that caffeine is also able to chelate Ca2+ within the mitochondrial matrix.

We next compared the effect of caffeine in cells expressing RyRs. Figure 5 illustrates the dose–response relationship for caffeine concentrations between 0.3 and 50 mM in control or in RyR3-expressing cells. In control cells, there was little or no effect on [Ca2+]ER at low concentrations of caffeine (0.3 or 1 mM), whereas higher concentrations (10 or 50 mM) produced a graded [Ca2+]ER decrease (Figure 5A). In no case, [Ca2+]C was increased, as demonstrated by the Rhod-3 trace recorded simultaneously (scale at right). The experiment was terminated with stimulation with the IP3-producing agonist ATP, alone or in the presence of the SERCA inhibitor TBH, to demonstrate that ATP stimulation produced a decrease in [Ca2+]ER concomitant with an increase in [Ca2+]C, as expected. The last ATP stimulation (maximum release) allows calibration of the [Ca2+]ER signal. Note that [Ca2+]ER decrease induced by 50 mM caffeine was ∼50% of the maximum.

When the same stimulation protocol was applied to cells expressing RyR3 (Figure 5B), the action of caffeine was very much sensitized. Low caffeine concentrations (0.3 and 1 mM) produced now a clear [Ca2+]ER release (see GAP3 trace), and the 1 mM stimulus triggered [Ca2+]ER oscillations. In addition, the drops of [Ca2+]ER were in all cases associated with synchronic [Ca2+]C increases (see Rhod-3 trace). Moreover, the ER Ca2+ release was already maximum at 10 mM caffeine, as no further decrease in [Ca2+]ER was seen at 50 mM. Finally, these high caffeine concentrations discharged ER completely, whereas the [Ca2+]ER decrease in control cells was only 50% at 50 mM caffeine. A detailed quantification of the cytosolic- and ER Ca2+ responses in both types of cells is provided in the figure legend and in Supplementary Table S1. Expression of the RyR3 protein in the transfected cells was confirmed by Western blot (Figure 5C) and by immunofluorescence (Figure 5D).

To assess more precise and quantitatively the effect of caffeine, we marked the RyR-expressing cells with mCherry-Fluorescent Protein (Cherry) and stimulated them with increasing caffeine concentrations (0.3–50 mM). Figure 6A compares the averaged responses, grouped separately for Cherry-positive or -negative cells. Note that the non-synchronic oscillatory behavior of individual cells is hindered by the averaging. The RyR3-expressing cells (red trace) responded better to caffeine at all the concentrations tested. The response was almost maximal at 10 mM caffeine, whereas in the non-expressing cells (black trace) [Ca2+]ER was decreased by only 10–20% of the total ER Ca2+ (Figure 6A). Figure 6B summarizes the results. It is clear that the apparent affinity for caffeine was much higher in the RyR3-expressing cells. Cells transfected with the empty plasmid behaved as the untransfected cells (results not shown).

Finally, we tested the effect of theophylline, another methylxanthine able to sensitize CICR [8], on [Ca2+]ER, and compared it with that provoked by caffeine. Results are shown in Supplementary Figure S3. Both xanthic acid derivatives had similar effects at 20 mM, although the theophylline-induced [Ca2+]ER decrease was somewhat smaller and slower than the one induced by caffeine.

Discussion

RyR-mediated CICR is important for amplifying cytosolic Ca2+ signals in both excitable and non-excitable cells. Detection of CICR mechanisms can be performed on the basis of immunoreactivity, ryanodine binding or functional tests measuring modifications of Ca2+ signaling. One of the easiest protocols assesses the increase in [Ca2+]C in response to caffeine. Caffeine sensitizes RyRs to cytosolic Ca2+ and releases this cation from the ER at resting [Ca2+]C. The response is vigorous in some cell types (Figure 1A) and weaker in others (Figures 1B,D). The own Ca2+ probes, which are mobile Ca2+ chelators, can hinder small [Ca2+]C peaks making them practically undetectable (see examples in Figures 1C,E). This effect has been attributed to buffering of the [Ca2+]C microdomain by the Ca2+ probe and to an increase in Ca2+ diffusion velocity by mobile Ca2+ buffers, thus accelerating dissipation of high Ca2+ microdomains [1012,2729].

The above interferences illustrate the problems and limitations of indirect measurements, which can be circumvented by direct measurement of [Ca2+]ER. These direct [Ca2+]ER measurements were first made possible in intact cells with ER-targeted aequorin [30], although low-affinity Ca2+ probes are required for quantitative measurements [22,31]. Suitable fluorescent probes have also been recently developed [14,23,24].

Direct measurements of [Ca2+]ER, either with luminescent (Figure 1F) or with fluorescent probes (Figure 3) in cells stimulated with caffeine, revealed that this xanthine produces a rapid reduction in [Ca2+]ER, suggesting a release of Ca2+ from the ER to the cytosol. Unexpectedly, this [Ca2+]ER decrease did not go along with a [Ca2+]C increase (Figure 3B). This intriguing result can be explained by chelation of Ca2+ in the ER lumen. Caffeine is quite lipid-soluble and diffuses through cell membranes, including the ER membrane. On the other hand, it has been reported that caffeine binds Ca2+, although with very low affinity (association constant, 30 M−1) [32,33]. Although the interaction of caffeine with Ca2+ at the cytosol, where [Ca2+]C is ∼10−7 M, should be negligible, the interplay inside compartments with high Ca2+ would increase in proportion to the calcium concentration. Thus, at 1 mM Ca2+ and 50 mM caffeine, near 50% of the total calcium in the ER should be bound to caffeine. The decrease in the ER GAP3 signal in HeLa cells on adding caffeine is consistent with this computation. Note that chelation of Ca2+ at the stores could also activate store-operated Ca2+ entry. Theophylline is also able to bind Ca2+ with similar affinity as caffeine [33], and this is consistent with our results (Supplementary Figure S3).

As mentioned in the introduction, a decrease in [Ca2+]ER elicited by caffeine is often used for functional detection of the presence of RyRs in cells and tissues. Our present results alert against this practice, as reduction in [Ca2+]ER can reflect binding of Ca2+ by caffeine into the ER lumen rather than a real ER Ca2+ release. The range of caffeine concentrations used is also relevant, as binding is evident at concentrations of 10–50 mM, whereas activation of CICR by binding to RyR occurs at concentrations of 1 mM or less (Figures 5 and 6). Our results evidence that measurements of either the cytosolic or the ER Ca2+ concentration both have severe limitations to explain CICR, so that simultaneous monitoring of [Ca2+]ER and [Ca2+]C is necessary to draw founded conclusions.

Caffeine does also bind other divalent metals with more affinity than Ca2+ [32]. These metals are present at trace amounts in living organisms, but are often concentrated inside the intracellular organelles of living cells. This opens up the possibility that binding of these trace elements within intracellular calcium stores such as ER, Golgi apparatus or mitochondria might be involved in the action mechanism of caffeine.

Abbreviations

     
  • [Ca2+]C

    cytosolic-free Ca2+ concentration

  •  
  • [Ca2+]ER

    Ca2+ concentration inside ER

  •  
  • [Ca2+]M

    Ca2+ concentration inside mitochondria

  •  
  • CCh

    carbachol

  •  
  • CICR

    calcium-induced calcium release

  •  
  • ELS

    extracellular-like solution

  •  
  • ER

    endoplasmic reticulum

  •  
  • GAP

    GFP-Aequorin protein

  •  
  • GFP

    green fluorescent protein

  •  
  • HEK 293

    Human Embryonic Kidney 293

  •  
  • ILS

    intracellular-like solution

  •  
  • IP3

    inositol 1,4,5-trisphosphate

  •  
  • PBS

    phosphate-buffered saline

  •  
  • RyR

    ryanodine receptor

  •  
  • SERCA

    sarco/endoplasmic reticulum Ca2+ ATPase

  •  
  • TBH

    2,5-di-tert-butylhydroquinone

Author Contribution

J.R.-R., M.R.-P. and A.D.-L. did most of the experimental work. M.T.A. and J.-G.-S. provided conceptual input and designed the experiments. All authors participated in analysis, discussion and interpretation of data, revised the article and gave final approval. J.G.-S. put together all data and wrote the manuscript.

Acknowledgments

We thank Jesús Fernández and Miriam García Cubillas for expert technical help.

Funding

This work was supported by grants from The Spanish MINECO [BFU2017-83066-P] and the Junta de Castilla y León [GR175]. J.R.-R., M.R.-P., and A.D.-L. were supported by fellowships from the Spanish MINECO.

Competing Interests

The Authors declare that there are no competing interests associated with the manuscript.

References

References
1
Roderick
,
H.L.
,
Berridge
,
M.J.
and
Bootman
,
M.D.
(
2003
)
Calcium-induced calcium release
.
Curr. Biol.
13
,
R425
2
Zucchi
,
R.
and
Ronca-Testoni
,
S.
(
1997
)
The sarcoplasmic reticulum Ca2+ channel/ryanodine receptor: modulation by endogenous effectors, drugs and disease states
.
Pharmacol. Rev.
49
,
1
51
PMID:
[PubMed]
3
Sutko
,
J.L.
and
Airey
,
J.A.
(
1996
)
Ryanodine receptor Ca2+ release channels: does diversity in form equal diversity in function?
Physiol. Rev.
76
,
1027
1071
4
Hakamata
,
Y.
,
Nakai
,
J.
,
Takeshima
,
H.
and
Imoto
,
K.
(
1992
)
Primary structure and distribution of a novel ryanodine receptor/calcium release channel from rabbit brain
.
FEBS Lett.
312
,
229
235
5
Bennett
,
D.L.
,
Cheek
,
T.R.
,
Berridge
,
M.J.
,
De Smedt
,
H.
,
Parys
,
J.B.
,
Missiaen
,
L.
et al. 
(
1996
)
Expression and function of ryanodine receptors in nonexcitable cells
.
J. Biol. Chem.
271
,
6356
6362
6
Querfurth
,
H.W.
,
Haughey
,
N.J.
,
Greenway
,
S.C.
,
Yacono
,
P.W.
,
Golan
,
D.E.
and
Geiger
,
J.D.
(
1998
)
Expression of ryanodine receptors in human embryonic kidney (HEK293) cells
.
Biochem. J.
334
,
79
86
7
Patel
,
S.
,
Ramakrishnan
,
L.
,
Rahman
,
T.
,
Hamdoun
,
A.
,
Marchant
,
J.S.
,
Taylor
,
C.W.
et al. 
(
2011
)
The endo-lysosomal system as an NAADP-sensitive acidic Ca2+ store: role for the two-pore channels
.
Cell Calcium
50
,
157
167
8
Kong
,
H.
,
Jones
,
P.P.
,
Koop
,
A.
,
Zhang
,
L.
,
Duff
,
H.J.
and
Chen
,
S.R.
(
2008
)
Caffeine induces Ca2+ release by reducing the threshold for luminal Ca2+ activation of the ryanodine receptor
.
Biochem. J.
414
,
441
452
9
Muschol
,
M.
,
Dasgupta
,
B.R.
and
Salzberg
,
B.M.
(
1999
)
Caffeine interaction with fluorescent calcium indicator dyes
.
Biophys. J.
77
,
577
586
10
Alonso
,
M.T.
,
Chamero
,
P.
,
Villalobos
,
C.
and
Garcia-Sancho
,
J.
(
2003
)
Fura-2 antagonises calcium-induced calcium release
.
Cell Calcium
33
,
27
35
11
Neher
,
E.
(
1998
)
Vesicle pools and Ca2+ microdomains: new tools for understanding their roles in neurotransmitter release
.
Neuron
20
,
389
399
12
Xu
,
T.
,
Naraghi
,
M.
,
Kang
,
H.
and
Neher
,
E.
(
1997
)
Kinetic studies of Ca2+ binding and Ca2+ clearance in the cytosol of adrenal chromaffin cells
.
Biophys. J.
73
,
532
545
13
Neher
,
E.
and
Augustine
,
G.J.
(
1992
)
Calcium gradients and buffers in bovine chromaffin cells
.
J. Physiol.
450
,
273
301
14
Navas-Navarro
,
P.
,
Rojo-Ruiz
,
J.
,
Rodriguez-Prados
,
M.
,
Ganfornina
,
M.D.
,
Looger
,
L.L.
,
Alonso
,
M.T.
et al. 
(
2016
)
GFP-aequorin protein sensor for ex vivo and in vivo imaging of Ca2+ dynamics in high-Ca2+ organelles
.
Cell Chem. Biol.
23
,
738
745
15
Rodríguez-Prados
,
M.
,
Rojo-Ruiz
,
J.
,
Aulestia
,
F.J.
,
García-Sancho
,
J.
and
Alonso
,
M.T.
(
2015
)
A new low-Ca2+ affinity GAP indicator to monitor high Ca2+ in organelles by luminescence
.
Cell Calcium
58
,
558
564
16
Villalobos
,
C.
,
Nunez
,
L.
,
Frawley
,
L.S.
,
Garcia-Sancho
,
J.
and
Sanchez
,
A.
(
1997
)
Multi-responsiveness of single anterior pituitary cells to hypothalamic-releasing hormones: a cellular basis for paradoxical secretion
.
Proc. Natl Acad. Sci. U.S.A.
94
,
14132
14137
17
Villalobos
,
C.
,
Nuñez
,
L.
,
Montero
,
M.
,
García
,
A.G.
,
Alonso
,
M.T.
,
Chamero
,
P.
et al. 
(
2002
)
Redistribution of Ca2+ among cytosol and organella during stimulation of bovine chromaffin cells
.
FASEB J.
16
,
343
353
18
Moore
,
G.A.
,
Kass
,
G.E.
,
Duddy
,
S.K.
,
Farrell
,
G.C.
,
Llopis
,
J.
and
Orrenius
,
S.
(
1990
)
2,5-Di(tert-butyl)-1,4-benzohydroquinone--a novel mobilizer of the inositol 1,4,5-trisphosphate-sensitive Ca2+ pool
.
Free Radic. Res. Commun.
8
,
337
345
19
Manjarrés
,
I.M.
,
Chamero
,
P.
,
Domingo
,
B.
,
Molina
,
F.
,
Llopis
,
J.
,
Alonso
,
M.T.
et al. 
(
2008
)
Red and green aequorins for simultaneous monitoring of Ca2+ signals from two different organelles
.
Pflugers Arch.
455
,
961
970
20
Bers
,
D.M.
,
Patton
,
C.W.
and
Nuccitelli
,
R.
(
2010
)
A practical guide to the preparation of Ca2+ buffers
.
Methods Cell Biol.
99
,
1
26
21
Alvarez
,
J.
and
Montero
,
M.
(
2002
)
Measuring [Ca2+] in the endoplasmic reticulum with aequorin
.
Cell Calcium
32
,
251
260
22
Alonso
,
M.T.
,
Barrero
,
M.J.
,
Michelena
,
P.
,
Carnicero
,
E.
,
Cuchillo
,
I.
,
García
,
A.G.
et al. 
(
1999
)
Ca2+-induced Ca2+ release in chromaffin cells seen from inside the ER with targeted aequorin
.
J. Cell Biol.
144
,
241
254
23
Rodriguez-Garcia
,
A.
,
Rojo-Ruiz
,
J.
,
Navas-Navarro
,
P.
,
Aulestia
,
F.J.
,
Gallego-Sandin
,
S.
,
Garcia-Sancho
,
J.
et al. 
(
2014
)
GAP, an aequorin-based fluorescent indicator for imaging Ca2+ in organelles
.
Proc. Natl Acad. Sci. U.S.A.
111
,
2584
2589
24
Alonso
,
M.T.
,
Rojo-Ruiz
,
J.
,
Navas-Navarro
,
P.
,
Rodríguez-Prados
,
M.
and
García-Sancho
,
J.
(
2017
)
Measuring Ca2+ inside intracellular organelles with luminescent and fluorescent aequorin-based sensors
.
Biochim. Biophys. Acta
1864
,
894
899
25
Aulestia
,
F.J.
,
Alonso
,
M.T.
and
García-Sancho
,
J.
(
2015
)
Differential calcium handling by the cis and trans regions of the Golgi apparatus
.
Biochem. J.
466
,
455
465
26
Cooley
,
L.A.
,
Martin
,
V.V.
and
Gee
,
K.R.
(
2013
)
Development and in vitro characterization of ratiometric and intensity-based fluorescent ion sensors
.
Methods Mol. Biol.
995
,
133
145
27
Naraghi
,
M.
,
Müller
,
T.H.
and
Neher
,
E.
(
1998
)
Two-dimensional determination of the cellular Ca2+ binding in bovine chromaffin cells
.
Biophys. J.
75
,
1635
1647
28
Zhou
,
Z.
and
Neher
,
E.
(
1993
)
Mobile and immobile calcium buffers in bovine adrenal chromaffin cells
.
J. Physiol.
469
,
245
273
29
Sala
,
F.
and
Hernández-Cruz
,
A.
(
1990
)
Calcium diffusion modeling in a spherical neuron. Relevance of buffering properties
.
Biophys. J.
57
,
313
324
30
Montero
,
M.
,
Brini
,
M.
,
Marsault
,
R.
,
Alvarez
,
J.
,
Sitia
,
R.
,
Pozzan
,
T.
et al. 
(
1995
)
Monitoring dynamic changes in free Ca2+ concentration in the endoplasmic reticulum of intact cells
.
EMBO J.
14
,
5467
5475
31
Barrero
,
M.J.
,
Montero
,
M.
and
Alvarez
,
J.
(
1997
)
Dynamics of [Ca2+] in the endoplasmic reticulum and cytoplasm of intact HeLa cells. A comparative study
.
J. Biol. Chem.
272
,
27694
27699
32
Kolaylı
,
S.
,
Ocak
,
M.
,
Küçük
,
M.
and
Abbasoǧlu
,
R.
(
2004
)
Does caffeine bind to metal ions?
Food Chem.
84
,
383
388
33
Nafisi
,
S.
,
Shamloo
,
D.S.
,
Mohajerani
,
N.
and
Omidi
,
A.
(
2002
)
A comparative study of caffeine and theophylline binding to Mg(II) and Ca(II) ions: studied by FTIR and UV spectroscopic methods
.
J. Mol. Struct.
608
,
1
7
34
Chamero
,
P.
,
Manjarres
,
I.M.
,
García-Verdugo
,
J.M.
,
Villalobos
,
C.
,
Alonso
,
M.T.
and
García-Sancho
,
J.
(
2008
)
Nuclear calcium signaling by inositol trisphosphate in GH3 pituitary cells
.
Cell Calcium
43
,
205
214