FM dyes enter via a store-operated calcium channel and modify calcium signaling of cultured astrocytes

Abstract

The amphiphilic fluorescent styryl pyridinium dyes FM1-43 and FM4-64 are used to probe activity-dependent synaptic vesicle cycling in neurons. Cultured astrocytes can internalize FM1-43 and FM4-64 inside vesicles but their uptake is insensitive to the elevation of cytosolic calcium (Ca²⁺) concentration and the underlying mechanism remains unclear. Here we used total internal reflection fluorescence microscopy and pharmacological tools to study the mechanisms of FM4-64 uptake into cultured astrocytes from mouse neocortex. Our data show that: (i) endocytosis is not a major route for FM4-64 uptake into astrocytes; (ii) FM4-64 enters astrocytes through an aqueous pore and strongly affects Ca²⁺ homeostasis; (iii) partitioning of FM4-64 into the outer leaflet of the plasma membrane results in a facilitation of store-operated Ca²⁺ entry (SOCE) channel gating; (iv) FM4-64 permeates and competes with Ca²⁺ for entry through a SOCE channel; (v) intracellular FM4-64 mobilizes Ca²⁺ from the endoplasmic reticulum stores, conveying a positive feedback to activate SOCE and to sustain dye uptake into astrocytes. Our study demonstrates that FM dyes are not markers of cycling vesicles in astrocytes and calls for a careful interpretation of FM fluorescence.

In astrocytes the strong expression of the plasma membrane (PM) store-operated Ca²⁺ entry (SOCE) (1–4) contrasts with a relatively small expression of voltage-gated channels and AMPA/NMDA ligand-gated channels that are mostly expressed by neurons. In response to neuronal activity and neurotransmitter release, the activation of astroglial metabotropic receptors induces a reduction of the endoplasmic reticulum (ER) Ca²⁺ concentration ([Ca²⁺]_ER) that leads to the activation of the SOCE and to an elevation of the intracellular Ca²⁺ concentration ([Ca²⁺]_i) (4) that is critical for the electrically silent astrocytes to release gliotransmitters and control nearby neurons, glia, and blood vessels (5). The type 1–2 stromal interaction molecules (STIM1–2) and Orai1–3 membrane proteins are key components of the SOCE, being the ER Ca²⁺ sensor and the PM pore-forming subunit, respectively (6, 7). A possible participation of the classical transient receptor potential (TRPC) channels to the protein complex generating the SOCE is debated in several cell types (8), including astrocytes (3).

Styryl pyridinium FM dyes are amphiphilic molecules that reversibly partition in lipid membranes. Their positively charged pyridinium head prevents them from diffusing through the PM. Only weakly fluorescent in aqueous solution their quantum yield increases in a lipid environment. Their activity-dependent uptake and destaining has made them common probes for synaptic vesicle cycling at the nerve terminals (9–11). Cultured astrocytes internalize FM dyes (3, 12–18). However, this uptake is not modified by the activation of metabotropic glutamate receptors (19), and its mechanism remains elusive.

Here, we show that incubating astrocytes with FM4-64 for 2 to 5 min leads to an FM4-64 uptake by permeation through a channel involved in SOCE. We also show that: (i) insertion of FM4-64 in PM facilitates SOCE; (ii) like other small FM dyes (FM1-43 and FM5-95, but not FM3-25), FM4-64 permeates through a cationic aqueous pore, and acts as a permeant SOCE blocker competing with Ca²⁺ influx; (iii) FM4-64 increases inositol trisphosphate receptor (IP₃R) channel opening and blocks the Ca²⁺ ATPase activity, thus emptying the ER Ca²⁺ store and thereby further activating SOCE and dye entry. Our data demonstrate the existence of complex interactions between FM4-64 and SOCE in astrocytes.

Results

Endocytosis Is Not a Major Mechanism for Astrocytic FM4-64 Uptake.

Short exposure of mouse cortical astrocytes in culture to micromolar concentrations of FM4-64 (6.7 μM, 2 min) at room temperature (RT) followed by a 10-min wash showed the expected FM4-64 puncta (14, 18, 19) associated with a diffuse labeling. Rupturing the PM in zero extracellular Ca²⁺ concentration ([Ca²⁺]_o) reduced the diffuse labeling without affecting the puncta (Fig. 1A), suggesting that background FM4-64 staining is not a result of the distribution of the dye inside small vesicles and that endocytosis and vesicle trafficking do not contribute to this background. This was confirmed by the weak co-localization found between several FM dyes and fluorescent endocytic markers (Fig. 1B) (20). Five-minute co-application of green FM1-43 and 3,000-MW Texas Red dextran (Tx-d) resulted in a lower correlation coefficient (r₁₂ = 0.16 ± 0.07, n = 8; see SI Appendix) than that found when coapplying green FM1-43 and red FM4-64 (Fig. S1A in SI Appendix; r₁₂ = 0.90 ± 0.03, n = 7). Similarly, the correlation between FM4-64 and 10,000-MW fluorescein-dextran (FL-d) was weak (Fig. S1B in SI Appendix; r₁₂ = 0.13 ± 0.05, n = 18). Both endocytic markers labeled few puncta (0.007 puncta/μm² ± 0.003, n = 9 cells for TX-d; 0.009 puncta/μm² ± 0.004, n = 8 for FL-d) compared with FM1-43 and FM4-64 (0.04 puncta/μm² ± 0.01, P < 0.01 for FM1-43; 0.04 puncta/μm² ± 0.01, P < 0.01 for FM4-64). The low density of Tx-d- and FL-d-labeled puncta was not a result of impaired endocytosis, because a 3-h incubation with either dextran produced significant vesicle labeling (0.07 puncta/μm² ± 0.02, n = 6 cells for Tx-d; 0.06 puncta/μm² ± 0.03, n = 5 cells for FL-d; Fig. S1C in SI Appendix). These data indicate that FM4-64 and FM1-43 uptake in astrocytes operates on time scale and labels subcellular compartments different from endocytic pathway.

Fig. 1.

Endocytosis is not a major route for rapid FM dye uptake in cultured astrocytes. (A) Epifluorescence image of a subregion of a cortical astrocyte loaded with FM4-64 (6.7 μM, 2 min incubation at RT, 10-min wash). (Scale bar, 5 μm.) Traces (gray, individuals; color, mean ± SD) display the evolution with time of the fluorescence intensity upon membrane rupture, measured in the extracellular (turquoise, 11 ROIs from 5 cells), cytoplasmic (red; 13 ROIs), and vesicular ROIs after local background subtraction (blue, 14 ROIs). (B) Top: Pearson correlation coefficient, calculated for subcellular ROIs between FM dyes and endocytic markers, coloaded for 5 min (Fig. S1B in SI Appendix). Bottom: Labeled puncta density (per μm²). (C) TIRFM images of astrocytes loaded with FM4-64 (6.7 μM, 5 min) at RT (control; CTR) and 4 °C (T_chamber, 3.7 ± 0.4 °C, n = 25). (Scale bar: 10 μm.) (D) FM4-64 labeling in control and after pretreatment with brefeldin A (Bfa, 5 μg/mL, 30 min) or dynasore (Dyn, 50 μM, 30 min; P = 0.2–0.77, n = 8–15 cells per condition). (E) FM4-64 labeling in control and after pretreatment with cytochalasin D (Cyt, 5 μM, 60 min) or jasplakinolide (Jasp, 2 μM, 60 min; P = 0.16–0.69, n = 5–18 cells). *, P < 0.05; **, P < 0.01.

Performing FM4-64 loading at 4 °C reduced the dye uptake [single-cell mean fluorescence (MF), 13.2 a.u. ± 4.0, 39% of control; 0.028 puncta/μm² ± 0.02, 45% of control, n = 13 cells; Fig. 1C), suggesting that endocytosis is not a prevalent mechanism for FM4-64 entry (21). Likewise, using brefeldin A to impair vesicle trafficking (22), dynasore to inhibit the GTPase activity of dynamin (23), cytochalasin D to inhibit actin depolymerization, and jasplakinolide to stabilize actin filaments (24, 25) had no effect on astrocytic FM4-64 labeling (P = 0.16–0.69, n = 7–13; Fig. 1 D and E). Finally, 24 h preincubation of astrocytes with tetanus toxin (TeNT, 2 μg/mL) to inhibit astroglial SNARE-mediated exocytosis (26) modified neither FM4-64 uptake (MF, 180.9 a.u. ± 57.8, n = 15, in TeNT vs. 178.3 a.u. ± 30.3, n = 12, in control; P = 0.51) nor puncta density (0.093 puncta/μm² ± 0.014 in TeNT vs. 0.087 puncta/μm² ± 0.03 in control; P = 0.39). These data indicate that exocytic/endocytic vesicle cycling does not contribute to FM4-64 uptake by astrocytes.

FM4-64 Permeates Through an Aqueous Pore.

A diffuse non-endocytic FM1-43 fluorescence was already reported in various sensory cells (27–34). It was attributed to FM1-43 permeation through a cationic channel, the exact nature of which remained elusive.

To test if the FM dye uptake involves permeation through an aqueous pore, we compared the uptake of the small FM5-95 and the large FM3-25 with the corresponding uptake of medium-sized FM4-64 and FM1-43 dyes (10) (see SI Appendix). The whole-cell fluorescence can be used as a direct readout of FM5-95 and FM4-64 uptake because, when mixed with lysophosphatidylcholine (LPC) at equal concentration, both dyes have similar fluorescence intensity (Fig. S2A in SI Appendix) and excitation/emission spectra (10) (Fig. S2B in SI Appendix). The smaller FM5-95 produced a higher fluorescence of astrocytes than the medium-sized FM4-64 (MF, 534 a.u. ± 138, n = 25 cells, vs. 244 a.u. ± 139, n = 21 cells; P < 0.01), and larger puncta density (0.09 puncta/μm² ± 0.03 vs. 0.07 puncta/μm² ± 0.02; P < 0.05; Fig. 2A). Similar initial fluorescence was obtained for FM3-25 and FM1-43 (Fig. S2C in SI Appendix), but the large FM3-25 could not be washed out from the PM (Fig. S2D in SI Appendix); therefore, we quenched extracellular FM3-25 with bromophenol blue (BPB) (19) to allow the quantification of FM3-25 uptake. After BPB treatment, the whole-cell fluorescence was smaller for FM3-25 than FM1-43 (MF, 32.9 a.u. ± 14.7 vs. 167.2 a.u. ± 60; n = 7; P < 0.01), as was the density of labeled puncta (0.01 puncta/μm² ± 0.01 vs. 0.1 puncta/μm² ± 0.02; P < 0.01; Fig. 2B). Thus, the astrocytic dye uptake during the 2- to 5-min incubation time is size-selective.

Fig. 2.

FM dyes uptake in astrocytes involves a cationic channel. (A) Epifluorescence images of astrocytes labeled with either FM4-64 or its spectral analogue FM5-95 (6.7 μM, 5 min each). Loading is enhanced with the smaller FM5-95. (B) Similar labeling to FM4-64 is seen with short lipophilic tail FM1-43 (12 μM), but FM3-25 (20 μM), having a longer lipophilic tail, is virtually impermeant. Quenching extracellular FM1-43 and FM3-25 with 2 mM BPB permitted quantification of the intracellular labeling. (C) Incubation of the astrocytes in high [K⁺]_o solutions reduced FM4-64 uptake. (D) Incubation of the astrocytes with FM4-64 at high [Ca²⁺]_o (20 mM, h-Ca) diminished FM dye uptake compared with control (1.8 mM) medium [control (CTR); n = 7–25 cells per condition]. (Scale bars: 10 μm.)

At physiological pH, permeation of the divalent FM cation (35) through an aqueous pore should be affected by the driving force across the PM. As the resting membrane potential of astrocytes is dominated by K⁺ channel conductance (36) and follows the K⁺ equilibrium potential, we changed the driving force on the FM dye by incubating astrocytes in FM4-64 solutions containing 5.5 to 120 mM extracellular K⁺ ([K⁺]_o), lowering extracellular Na⁺ concentration accordingly. FM4-64 uptake and the density of labeled puncta correlated with [K⁺]_o (MF, 42.2 a.u. ± 15.4, 17.4 a.u. ± 6.1, and 14.9 a.u. ± 5.9 for 40, 60, and 120 mM [K⁺]_o; 56.8%, 23.5%, and 20.1% of control, respectively; P < 0.01; n = 14–20; 0.037 puncta/μm² ± 0.01, 0.027 puncta/μm² ± 0.012, and 0.02 puncta/μm² ± 0.01; 62%, 43.6%, and 32.3% of control, respectively; P < 0.01; Fig. 2C). Interestingly, high [K⁺]_o, evoked a Co²⁺-sensitive [Ca²⁺]_i elevation (Fig. S3A in SI Appendix), but had no effect on vesicular FM4-64 labeling in astrocytes (Fig. S3C in SI Appendix). This contrasts with observations made with cultured neurons, in which the facilitation of synaptic release by high [K⁺]_o increased FM4-64 uptake and destaining (Fig. S3B–C in SI Appendix). These results strongly suggest that FM4-64 enters into astrocytes by permeating through an aqueous cationic channel rather than by endocytosis.

FM4-64 Uptake Relies on a SOCE Pathway.

To identify the FM4-64-permeable channel, we first tested if FM4-64 uptake depends on the extracellular divalent cation concentration. We found that labeling the astrocytes in elevated (20 mM) [Ca²⁺]_o reduced both the whole-cell fluorescence and the density of FM4-64-labeled puncta to 15% and 32% of control in 1.8 mM Ca²⁺, respectively (P < 0.01, n = 9–12 cells; Fig. 2D). Increasing the external Mg²⁺ concentration ([Mg²⁺]_o) from 1.0 to 20 mM had no effect on the dye uptake or the density of labeled puncta (MF, 106.2 a.u. ± 35.8, 0.066 puncta/μm² ± 0.03, n = 11, vs. control, 128.7 a.u. ± 31.4, 0.071 puncta/μm² ± 0.03, n = 10; P > 0.32), suggesting that FM4-64 competes with Ca²⁺ for permeation through a Ca²⁺-permeable cationic channel. A possible involvement of Ca²⁺-permeable ligand-gated channels (TRPV, AMPA, NMDA, P2X, acetylcholine receptors), mechanotransducer channels, voltage-gated Ca²⁺ channels, inward rectifier Kir4.1 channels, and the PM Na⁺-Ca²⁺ exchanger was excluded by testing the effect of specific agonists and antagonists on FM4-64 uptake (Table S1 in SI Appendix). We also excluded a possible role of hemichannels and volume-sensitive anion channels.

Astrocytes in culture generate spontaneous [Ca²⁺]_i elevation (Fig. S4A in SI Appendix), which is sustained by Ca²⁺ influx via the SOCE pathway (4), suggesting SOCE as a possible route for FM dye entry. Recent work identified the STIM and Orai proteins and the TRPC channels as key proteins of SOCE (4, 6–8). We found that depleting the astroglial ER Ca²⁺ stores with thapsigargin (TG, 2 μM, 30 min) increased FM4-64 uptake and the density of FM dye-labeled puncta (Fig. 3A; see Table S2 in SI Appendix for detailed quantification). To further test a possible involvement of SOCE, we used calyculin A (200 nM, 60 min), which up-regulates SOCE (37), and found that it facilitates FM dye uptake; however, it reduces the density of FM dye-labeled puncta (Fig. 3B and Table S2 in SI Appendix). Anisomycin (30 μM, 12 h), which reduces the Ca²⁺ leak from ER and the SOCE (38), reduced FM dye uptake and the puncta density (Fig. 3C and Table S2 in SI Appendix). Gd³⁺, a nonspecific blocker of SOCE (Fig. S4 in SI Appendix), abolished FM4-64 uptake (Fig. 3D and Table S2 in SI Appendix). By using 3,5 bistrifluoromethyl pyrazole 2 (BTP2; 40 μM) (39) to interfere with SOCE by blocking TRPC channels, we observed a reduced FM dye uptake and puncta density (Fig. 3E and Table S2 in SI Appendix). SKF96365 (20 μM), another TRPC blocker (8), similarly decreased both parameters (Fig. 3F and Table S2 in SI Appendix). Finally, 2-aminoethoxy diphenylborate (2-APB) applied at a concentration that blocks Orai1 (100 μM, 10 min) (40) increased the FM4-64 uptake and the puncta density (Table S1 in SI Appendix). These results suggest that the permeant FM4-64 enters astrocytes via SOCE channels and that a TRPC-like rather than a Orai1 channel forms the pore for the dye entry.

Fig. 3.

FM4-64 uptake involves a SOCE pathway. FM4-64 labeling (6.7 μM, 5 min) was increased by TG treatment (2 μM, 30 min) (A), and calyculin A (Caly, 200 nM, 60 min) (B), and inhibited by anisomycin (Aniso, 30 μM, 12 h) (C), Gd³⁺ (100 μM, 10 min) (E), BTP2 (40 μM, 10 min) (D), and SKF96365 (20 μM, 10 min) (F).

FM4-64 Activates SOCE by Interacting with the PM Lipid Bilayer.

The insertion of lipids in the PM alters the lipid rafts (41), the activity of ionic channels (42), SOCE (4), as well as TRPC/Orai interaction (6). Amphiphilic FM dyes alter the lipid arrangement (35) and the PM mechanical properties (43), suggesting that their insertion in the PM could affect the activity of the store-operated channels.

This possibility was tested by studying first the effect on FM4-64 uptake of the spider Grammostola spatulata mechanotoxin peptide 4 (GsMTx-4), an inhibitor of stretch-activated mechanosensitive channels and SOCE (44) that modifies the PM lipid/channel interaction. We found that 5 μM GsMTx-4 reduced FM4-64 uptake and puncta density (Fig. 4A and Table S2 in SI Appendix). Likewise, LPC (5 μM), which facilitates SOCE (4), enhanced the FM4-64 uptake and puncta density (Fig. 4B and Table S2 in SI Appendix). Arachidonic acid (10 μM), a lipid with no effect on SOCE (4), did not alter FM4-64 uptake (Fig. 4B). Finally, methyl-β-cyclodextrin (MβCD), which disrupts lipid rafts by sequestering cholesterol, reduced astrocytic FM4-64 uptake and puncta density (Fig. 4C and Table S2 in SI Appendix). Our results show that FM4-64 insertion in the outer leaflet of the PM, and the subsequent change of the membrane mechanical properties that can change the SOCE, might interfere with FM4-64 uptake.

Fig. 4.

FM dye insertion in the PM outer lipid leaflet activates a SOCE. (A) Preincubating astrocytes with GsMTx-4 (5 μM, 10 min) reduced the FM4-64 uptake (6.7 μM, 5 min). (B) Treating astrocytes with LPC (5 μM, 10 min) increased the FM4-64 uptake. Arachidonic acid (AA, 10 μM, 10 min) had no effect. (C) Treating astrocytes with MβCD (10 mM, 1 h) diminished FM4-64 labeling. (D) The local application of green impermeant FM3-25 facilitated SOCE. The ER store was depleted by treating cells with TG (1 μM, 10 min) in Ca²⁺-free buffer. The [Ca²⁺]_i was monitored with the red Ca²⁺ indicator Xrhod-1 (n = 7–16 cells per condition). (Scale bars: 10 μm.)

To more directly test the lipid-mediated effect of styryl dyes on SOCE, we studied the effect of the non-permeant (Fig. 2B) green FM3-25 on [Ca²⁺]_i using a red-fluorescent Ca²⁺ indicator Xrhod-1, acetoxymethyl ester (AM). We used total internal reflection fluorescence (TIRF) microscopy to monitor near-membrane [Ca²⁺]_i (Fig. 4D). FM3-25 was applied with a small (250 μm ID) silica pipette (Fig. S5A1 in SI Appendix) positioned close to the astrocyte. We verified that the local application of control buffer on intact astrocytes did not trigger an appreciable [Ca²⁺] change [dF/F₀, 0.01 ± 0.03, n = 7 cells, vs. control, 0.01 ± 0.02, n = 9 cells (P = 0.93), measured in static solution over the same period]. When astrocytes were pretreated with TG (1 μM, 10 min) in a Ca²⁺-free buffer to deplete the ER store, locally applying 1.8 mM [Ca²⁺]_o induced a SOCE-mediated [Ca²⁺]_i elevation (dF/F₀, 0.37 ± 0.2, n = 7 cells) that only slowly returns to baseline, as a result of the partial mitochondrial localization of Xrhod-1 (SI Methods in SI Appendix). Simultaneously applying 10 μM FM3-25 and 1.8 mM [Ca²⁺]_o induced a larger [Ca²⁺]_i elevation (dF/F₀, 0.9 ± 0.6, n = 11 cells; P < 0.05), as expected if FM3-25 recruits additional SOCE channels. These results indicate that the FM3-25 insertion into the PM of astrocytes facilitates the SOCE channel gating .

Effect of Permeant FM4-64 on the Ca²⁺ Signaling in Astrocytes.

FM4-64 competes with Ca²⁺ and therefore should act as a permeant store-operated channel blocker, like FM1-43 for the mechanotransducer channels (28). Using the same protocol as in Fig. 4D, we studied the effect of 6.7 μM FM4-64 on the [Ca²⁺]_i measured with Oregon green BAPTA–1 [2,2′-(ethylenedioxy)dianiline-N,N,N′,N′-tetraacetic acid; OGB-1; Fig. 5A]. Unlike the nonpermeant FM3-25, FM4-64 reduced SOCE (dF/F₀, 0.22 ± 0.2, n = 9, with FM4-64 vs. 0.97 ± 0.12, n = 6, in control; P < 0.01). Thus, in addition to facilitating the gating of the SOCE channels, the permeant FM4-64 behaves as a channel blocker of the SOCE in cortical astrocytes.

Fig. 5.

Permeant FM dyes modifies Ca²⁺ signaling in cultured astrocytes. (A) After depletion of ER Ca²⁺ store with TG in 0 mM Ca²⁺, an SOCE was induced by local application of control (CTR) solution (1.8 mM Ca²⁺, Upper) or by CTR solution containing 6.7 μM FM4-64 (Lower). Monitoring [Ca²⁺]_i with the green Ca²⁺ indicator OGB-1, we show that FM4-64 reduces the SOCE (mean ± SD, black and gray traces). (B) We locally applied control solution (1.8 mM Ca²⁺; CTR) first, followed by the application of FM4-64-containing solution (6.7 μM FM4-64 and 1.8 mM Ca²⁺; FM), before returning to the control. FM4-64 evokes a [Ca²⁺]_i elevation (green trace) that follows the FM dye insertion inside the outer PM lipid leaflet (red trace), measured with TIRFM. Inset: Same traces magnified to show the onset of the response to FM4-64. (C) Gd³⁺ (100 μM) and GsMTx-4 (2.5 μM) both reduced FM4-64 uptake (ΔF_FM measured at the end of the 2 min FM4-64 wash, B) and the FM-triggered [Ca²⁺]_i transient (dF/F₀ measured at the peak of the Ca²⁺ response, B). Removing extracellular Ca²⁺ (0 Ca²⁺, 5 mM EGTA) facilitated both the FM4-64 uptake and the Ca²⁺ response. (D) Pretreatment of astrocytes with TG (2 μM, 20 min) enhanced FM4-64 uptake and suppressed the FM dye-induced mobilization of intracellular Ca²⁺ store. (E) In astrocytes transfected with an ER-targeted esterase, local application of FM4-64 (6.7 μM) induced a reduction of the ER Ca²⁺ concentration ([Ca²⁺]_ER) monitored with the low-affinity Ca²⁺ indicator fluo5N-AM. Top: Dual-color Fluo5N and FM4-64 TIRFM kymographs. Bottom: FM4-64 insertion into the PM (red trace) was associated with a reduction of [Ca²⁺]_ER (green trace; n = 5–10 cells per condition).

We then studied the effect of the permeant FM4-64 on the astroglial Ca²⁺ signaling in standard 1.8 mM [Ca²⁺]_o to directly correlate the dye insertion in the PM and its cellular uptake with astroglial Ca²⁺ signaling in nondepleted Ca²⁺ store condition. We used a double-pulse protocol in which control and FM4-64-containing solutions were sequentially applied to OGB-1-loaded cells through the local perfusion pipette as used for Fig. 4D (see SI Text and Fig. S5A in SI Appendix). When applying control solution, no [Ca²⁺]_i changes were observed (Fig. S6 in SI Appendix). Upon addition of FM4-64, the insertion of the dye in the PM was detected as a rapid red fluorescence increase, followed by a slower increase as expected from dye uptake. Even the small leak of FM4-64 observed upon actuating the application of control buffer triggered a [Ca²⁺]_i increase (dF/F₀, 0.13 ± 0.05, n = 5 cells). The FM4-64 application itself evoked a robust [Ca²⁺]_i increase (peak dF/F₀, 0.43 ± 0.1, n = 10 cells) that followed the FM dye red fluorescence by 5.7 s ± 2.9 (n = 6) measured as the half-maximal delay (Fig. 5B Inset). This [Ca²⁺]_i elevation in astrocytes contrasts with the absence of response by neurons [Ca²⁺]_i following the same FM4-64 application (Fig. S8 in SI Appendix). Washing FM4-64 leading to an early fast reduction of the red fluorescence as a result of the rapid departitioning of the dye from the PM was followed by a slower reduction, probably as a result of the permeation of the dye through an SOCE channel. A residual red fluorescence (ΔF_FM) was attributed to dye uptake. The FM4-64-induced [Ca²⁺]_i elevation was concentration-dependent and saturated at relatively low concentration (approximately 1 μM; Fig. S7A in SI Appendix).

Using the same protocol (Fig. 5C), we showed that, in astrocytes, the FM4-64-induced [Ca²⁺]_i elevation was suppressed by GsMTx-4 (dF/F₀, 0.15 ± 0.06, n = 5, in GsMTx-4 vs. 0.51 ± 0.2, n = 7, in control; P < 0.01) and Gd³⁺ (dF/F₀, 0.05 ± 0.04, n = 5, in Gd³⁺ vs. 0.4 ± 0.1, n = 7, in control; P < 0.01). Dye uptake was also reduced by GsMTx-4 (ΔF_FM, 19.6 a.u. ± 5.4 in GsMTx-4 vs. 54.9 a.u. ± 16.3 in control; P < 0.01) and by Gd³⁺ (ΔF_FM, 14.03 a.u. ± 3.2 in Gd³⁺ vs. 31.3 a.u. ± 14.9 in control; P < 0.01). In 0 mM [Ca²⁺]_o, the local application of FM4-64 induced a larger [Ca²⁺]_i increase (279% of control; dF/F₀, 1.2 ± 0.2; n = 10; P < 0.01) and a larger FM4-64 uptake to 423% of control (ΔF_FM, 132.5 a.u. ± 64.3; P < 0.01; Fig. 5C). This is consistent with our previous conclusion (see Fig. 2D) that FM dye and Ca²⁺ compete for the same channel and that Ca²⁺ removal facilitates FM entry (and vice versa). Depleting the ER Ca²⁺ store with TG (2 μM, 30 min) enhanced the FM4-64-triggered FM dye uptake (ΔF_FM, 89.5 a.u. ± 31.8 in TG vs. 31.3 a.u. ± 14.9 in control; P < 0.01) and reduced the evoked Ca²⁺ response (dF/F₀, 0.07 ± 0.03, n = 9, in TG vs. 0.43 ± 0.10, n = 10 cells, in control; P < 0.01; Fig. 5D). These data suggest FM dye entry as a prerequisite for the trigger of Ca²⁺ response and the internal store as a major Ca²⁺ source for the response.

To directly monitor the possible release of Ca²⁺ from the ER by FM dyes, we measured the [Ca²⁺]_ER by transfecting the astrocytes with a recombinant ER-targeted carboxyl esterase (CES-2) to facilitate the accumulation of the low-affinity Ca²⁺ indicator Fluo-5N AM (K_d, approximately 90 μM) to the ER (45). The local application of 6.7 μM FM4-64 on the transfected astrocytes evoked a rapid reduction (12 ± 4% in 60 s) of Fluo-5N fluorescence (dF/F₀, −0.21 ± 0.1, n = 5; Fig. 5E), confirming that FM4-64 facilitates the Ca²⁺ release from the ER. Finally, we tested the effect of 10 mM caffeine, a blocker of IP₃R (46), which abolished most FM dye-evoked Ca²⁺ signal (dF/F₀, 0.08 ± 0.03, 35% of its control value, n = 5; P < 0.01; Fig. S9A in SI Appendix). Ryanodine (1 μM, 1 h) had no effect on the FM4-64-evoked Ca²⁺ response (n = 6–7; P = 0.62; Fig. S9B in SI Appendix). Together, these results indicate that FM dye-evoked Ca²⁺ responses are caused by Ca²⁺ release through IP₃R channels from a TG-sensitive ER store.

Discussion

We demonstrate that astroglial FM dye uptake does not involve endocytosis but an aqueous pore, and that it profoundly affects Ca²⁺ homeostasis of astrocytes. We show that both dye uptake and the associated [Ca²⁺]_i elevation involve a SOCE pathway, suggesting that the permeant FM dyes enter astrocytes via a SOCE channel, i.e., a STIM/Orai/TRPC complex. We also show that FM dyes modify astroglial Ca²⁺ signaling by acting at several levels (Fig. S10 in SI Appendix): (i) the FM dye insertion into the outer leaflet of the PM facilitates the gating of the SOCE; (ii) the FM dyes compete with Ca²⁺ for permeating across the SOCE channel; (iii) intracellular FM dye mobilizes Ca²⁺ from the ER store, conveying a positive feedback signal to activate SOCE and sustain dye uptake into astrocytes.

FM Dye Within Astrocytes Contributes to a Diffuse Background in the Brain Slices.

Early studies introducing FM dyes as fluorescent probes to investigate secretory vesicle cycling in nerve terminals reported a stimulation-insensitive FM dye labeling in Schwann cells (9), but this observation received little attention. In brain slice, FM dye labeling produces a strong background that can be quenched by sulforhodamine 101 (47, 48), a selective marker of astrocytes (49). Our results suggest that the FM1-43 background seen in the slice is due to the astroglial activity-independent FM dye uptake which is quenched by sulforhodamine 101. It suggests that the FM dye distribution within astrocyte in the slice preparation shares similarities with the distribution we found here in astrocytes in culture. However, whether all our findings apply to the astrocytes in situ will need further investigation.

FM Dye Enters Through a TRP-Like Channel.

Our study does not support endocytosis as a major route for the rapid FM dye entry into cortical astrocytes (Fig. 1). Rather, the dependence of the dye uptake on its molecular size and on the electrical driving force across the PM (Fig. 2) indicate that the FM dyes enter through a cationic aqueous pore. Earlier reports have already suggested that FM dyes permeate through cationic channels (28, 30, 31). The temperature sensitivity of the dye uptake (Fig. 1C) is compatible with the thermosensitivity of ion channels in general (50) and of TRP channels (51). The dye uptake is sensitive to TG, calyculin A, anisomycin, Gd³⁺, and SKF96365 (Fig. 3), which interfere with SOCE; and to GsMTx-4 and LPC, which affect the interaction between the PM lipid bilayer and the SOCE (Fig. 4). The implication of SOCE is also evidenced by showing that the FM dye insertion into the PM interferes directly with Ca²⁺ permeation through a SOCE channel (Fig. 4D) and depletes the ER Ca²⁺ stores (Fig. 5D). Finally, our findings that the FM dye uptake is sensitive to BTP2, a blocker of TRPC channels (8), and to 2-APB, a blocker of the Orai proteins (40), indicate that the FM dyes permeate through a STIM1/Orai/TRPC complex. The implication of TRP channels is compatible with their expression by astrocytes (2, 3, 52, 53). The sensitivity of the FM dye uptake to Gd³⁺, BTP2, GsMTx-4, MβCD, and LPC is compatible with the involvement of a TRPC-like channel. The effect of 2-APB that inhibits several TRPC channels at 5 to 20 μM, activates TRPV1–3 channels, and has no effect on TRPV4 (54), suggests that the astroglial FM dye uptake might involve also TRPV1–3 channels. Our earlier report that the mechanical stimulation of astrocytes triggered astroglial [Ca²⁺]_i elevation and a loss of a diffuse intracellular FM dye signal (19) is also compatible with the permeation of the FM dye through mechanosensitive TRP-like channels (55). In conclusion, the permeant FM dyes appear to enter the astrocytes via a TRPC/TRPV-like channel, but the exact stoichiometry of the channel is still uncertain and will require more specific tools such as siRNAs.

Permeant FM Dyes Facilitate ER Store Depletion and FM Dye Entry.

Our data show that, when in the cytosol, the permeant FM dyes facilitate ER store depletion (Fig. 5D), either by enhancing Ca²⁺ efflux through the IP₃R channel as indicated by the effect of caffeine (Fig. S8A in SI Appendix) or by reducing the sarcoplasmic/endoplasmic reticulum Ca²⁺ ATPase activity (56). Such a depletion of the ER Ca²⁺ store should act as a positive feedback loop to enhance the FM dye uptake through the STIM/Orai/TRPC complex and overcome the partial block of the cationic channel by the permeant FM dyes. Previous observations indicate that FM dye accumulate inside lysosomes (18, 19). Our finding of a correlation between the whole-cell fluorescence and the puncta density suggests that the permeant FM dyes traffic from the cytosol to the lysosomal lumen but the mechanism is unknown.

The absence of diffuse neuronal FM dye labeling is surprising given that SOCE, Orai1, STIM1, and TRPC1 are expressed by neurons (57). One reason might be a different channel density, because in neurons Ca²⁺ influx is mostly caused by voltage-gated Ca²⁺ channels, whereas the astroglial Ca²⁺ signaling is dominated by SOCE and relies on the near-membrane ER to rapidly replenish the ER stores (2). The repertoire of TRP channels expressed by astrocytes might also be uniquely favorable to FM dye permeation.

Materials and Methods

For detailed description of experimental procedures, see SI Text in SI Appendix.

Cell Culture and Dye Loading.

We used low-density cortical astrocytes from newborn mice maintained in culture as previously described (19). Control experiments were done on cortical neurons from embryonic mice. Dye loading and recordings were made at RT in solution containing (in mM): 140 NaCl, 5.5 KCl, 1.8 CaCl₂, 1 MgCl₂, 20 glucose, 10 Hepes (pH 7.3, NaOH). Astrocytes were labeled with 6.7 μM FM1-43, FM4-64, or FM5–95 in static bath for 2 to 5 min, followed by a 10-min wash. During recording, cells were constantly perfused at 0.5 to 1 mL/min with standard solution.

Ca²⁺ Imaging and Co-Localization Studies.

A custom-built inverted microscope was used for bright-field, polychromatic epi- and through-the-objective TIRF microscopy. 476-, 488- or 568-nm lines from the beam of an Ar⁺/Kr⁺ multiline laser were used for TIRF microscopy. The near-membrane [Ca²⁺]_i transients are shown after photobleaching correction as dF/F₀ with TIRFM using nonratiometric Ca²⁺ indicators, OGB-1-AM together with FM4-64/FM5-95 or Xrhod-1-AM together with FM3-25. The ER Ca²⁺ change was monitored using the low-affinity Ca²⁺ indicator Fluo-5N AM selectively targeted to the ER lumen by transfecting a ER-targeted recombinant CES-2 (provided by Robert Blum, Münich, Germany) 24 to 36 h before imaging. All combinations of excitation wavelengths, dichroics, and filters used are listed in Table S3 in SI Appendix. Colocalization was assessed by calculating the Pearson correlation coefficient r₁₂.

The density of fluorescent organelles was measured after high-pass filtering, thresholding, and watershed segmentation. Particle number was normalized with total area, estimated for epifluorescence images, from the threshold contour of the cell and, for TIRF images, from the size of the region of interest. MF was measured within their contour, after subtraction of the average mean autofluorescence measured in the corresponding color channel from nonlabeled cells of the same preparation. For each pharmacological condition, control and treated measurements of MF and particle density were performed on paired cells from the same preparation.