Fundamental Ca²⁺ Signaling Mechanisms in Mouse Dendritic Cells: CRAC Is the Major Ca²⁺ Entry Pathway

Although dendritic cells (DC)³ are recognized to play key roles in both initiating and modulating immune responses (1, 2) many fundamental aspects of their biology remain unknown. Ca²⁺ signaling in DC represents one such area, despite the fact that alterations in intracellular Ca²⁺ are known to underlie many immune responses. Indeed, a sustained increase in intracellular Ca²⁺ accompanies T and B cell receptor signaling and is necessary for gene activation, cellular proliferation, and Ab secretion (3, 4).

Similarly, many critical functions in DC appear to involve Ca²⁺ signaling. Apoptotic body engulfment and processing are accompanied by a rise in intracellular Ca²⁺ and are dependent on external Ca²⁺ (5). Chemotactic molecules uniformly produce Ca²⁺ increases in DC (6, 7, 8, 9), suggesting that Ca²⁺ transients regulate DC migration. DC maturation, including the enhanced expression of MHC class II and costimulatory molecules, is inhibited by chelation of external Ca²⁺ (10). Conversely, agents that mobilize intracellular Ca²⁺ can promote DC maturation in the absence of normal cytokine stimulation (10, 11).

However, the Ca²⁺-signaling pathways involved in these DC functions are not well defined. Chemokine-induced Ca²⁺ mobilization likely occurs via intracellular inositol trisphosphate (IP₃) receptors, because in many cell types the G protein-coupled chemokine receptors are known to activate phospholipase C β₂, and in turn, generate IP₃. Less is known about Ca²⁺ entry pathways. Previous studies have suggested the presence of dihydropyridine (DHP)-sensitive Ca²⁺ channels and ATP-gated channels, although these have not been functionally characterized. Here we have examined Ca²⁺ entry in DC using electrophysiological and calcium imaging techniques. We show that mouse, myeloid DC express neither functional voltage- nor DHP-gated channels; instead, DHPs mobilize Ca²⁺ from internal stores. Similarly, ATP signaling leads predominantly to Ca²⁺ mobilization rather than entry via plasma membrane channels. We show that the major Ca²⁺ entry pathway in DC is through the Ca²⁺ release-activated Ca²⁺ channel (CRAC) (12, 13, 14), a plasma membrane channel expressed in many cell types and activated by the depletion of intracellular Ca²⁺ stores. Furthermore, we show that CRAC is activated during physiologic DC signaling and that activation of CRAC promotes DC maturation.

C57BL/10J (C57) mice, 6–12 wk old, were obtained from The Jackson Laboratory (Bar Harbor, ME) and maintained in the specific pathogen-free facility of the University of Pittsburgh Medical Center.

Recombinant (r) GM-CSF and IL-4 were gifts of S. K. Narula (Schering-Plough, Kenilworth, NJ). Bay K8644, nifedipine, and Na₂ATP were obtained from Sigma (St. Louis, MO). Thapsigargin, SKF 96365, and IP₃ were obtained from Calbiochem (San Diego, CA).

DC were cultured using the method initially reported by Inaba et al. (15) with the following modifications. Bone marrow cells were prepared from the femurs and tibias of normal C57 mice and cultured at a density of 3 × 10⁵ cells/ml in RPMI 1640 (Life Technologies, Gaithersburg, MD) supplemented with 10% FCS (Nalgene, Miami, FL), nonessential amino acids, l-glutamine, sodium pyruvate, penicillin-streptomycin, and 2-ME (all obtained from Life Technologies). Cultures were supplemented with GM-CSF and IL-4 (at 4 ng/ml and 1000 U/ml, respectively). DC were harvested after a total of 3–4 days of culture for immature cells or 5–6 days of culture for mature cells and purified by metrizamide density (16.5 or 14.5%, respectively) centrifugation. Free [Ca²⁺] in the medium was varied by the addition of a pH-buffered EGTA stock (final concentration 0.4–0.5 mM) or CaCl₂. The actual free [Ca²⁺] with EGTA was verified with a Ca²⁺-selective electrode. The free [Ca²⁺] in normal RPMI 1640 medium (0.44 mM total Ca²⁺) and supplemented medium (5.44 mM total Ca²⁺) was estimated to be ∼0.36 and 4.6 mM, respectively, using the software Bound and Determined (16) (available online at http://superior.carleton.ca/∼kbstorey).

Flow cytometric analysis was undertaken using a Beckman Coulter EPICS Elite flow cytometer (Beckman Coulter, Hialeah, FL), and data were analyzed using either WinMDI or EXPO32 software (Applied Cytometry Systems, Sheffield, U.K.). Monoclonal Abs specific for MHC class II (IA^b), CD11c, CD80, and CD86 (PharMingen, San Diego, CA) were used as FITC and PE conjugates. Cells stained with species-specific, isotype-matched irrelevant mAbs were used as negative controls. Bone marrow cells cultured for 3–4 days were CDllc⁺, MHC class II⁺, CD80^low, and CD86^low/−, consistent with the immature or “Ag-processing” phenotype reported for DC both in situ or freshly isolated from peripheral tissues. On longer in vitro culture, DC up-regulated their expression of MHC class II Ags and costimulatory molecules consistent with mature DC (15, 17, 18). CD86 expression (vs CD11c) was used to distinguish immature and mature DC in this study (see Fig. 1).

DC were plated on coverslips in culture medium and loaded with Fluo-3AM (3–5 μM) for 20 min at 25 or 37°C degree. Cells were then washed with several volumes of bathing solution and left for another 20 min before recording. Standard bathing solution was (in mM) 130 NaCl, 4 KCl, 10 HEPES, 10 glucose, 2 CaCl₂, 2 MgCl₂ pH 7.3. Fluorescence measurements were made with either a Zeiss Axiovert 100 TV confocal microscope, or a Deltascan fluorometer (Photon Technology International, South Brunswick, NJ) coupled to a Diastar microscope (Leica, Deerfield, IL). Fluo-3 was excited at 488 nm, and emitted fluorescence was filtered with a 535 ± 25 nm bandpass filter. The fluorescence signal was calibrated in ATP experiments by measuring the maximal fluorescence after treatment with ionomycin (10–50 μM). Absolute estimates of [Ca²⁺] were then obtained by the expression [Ca²⁺] = K_D × (F − F_min)/(F_max − F), where K_D is the Ca²⁺ dissociation constant for Fluo3, and F_max and F_min are the maximal and minimal fluorescence, respectively. F_min was assumed to be negligible.

Whole cell patch clamp recordings were made using an EPC-7 amplifier interfaced to a Macintosh Power PC running IgorPro software (Wavemetrics, Lake Oswego, OR). Patch pipettes with resistances between 2 and 4 MΩ were prepared from aluminosilicate glass (Garner Glass, Claremont, CA). Series resistance compensation was routinely set at ∼50%. Data were filtered at 1 kHz and sampled at 5 kHz. For I_CRAC recording the bathing solution was (in mM): 145 NaCl, 2 KCl, 2 MgCl₂, 5 Glu, 5 HEPES, and 10 either Ca²⁺/Ba²⁺/Mg²⁺/Sr²⁺. The pipette solution contained (in mM): 128 CsAsp, 10 CsBAPTA, 0.1 CaCl₂, 3.16 MgCl₂, 10 mM HEPES, pH 7.4. I_CRAC was recorded with 200-ms voltage ramps from −130 to +60 or +90 mV from a holding potential of 0 mV.

For simultaneous patch clamp and fluorescence measurements with ATP and Bay K8644 the bathing solution contained (in mM): 140 N-methyl-d-glucamine chloride (NMDG-Cl), 4 KCl, 10 HEPES, 2 CaCl₂, 2 MgCl₂, pH 7.3, and the patch pipette contained (in mM): 130 CsCl, 10 NaCl, 10 HEPES, 0.1 EGTA, 4 MgATP, 0.1 GTP, and 50 μM Fluo-3 pentapotassium salt, pH 7.3. Junction potentials of ∼10–15 mV (calculated with PClamp software) were corrected offline.

To explore immature and mature DC (Fig. 1) for Ca²⁺ entry pathways, we first tested for functional voltage-gated channels. Voltage-clamped cells were depolarized with pulses from −90 to 0 mV. In some experiments we also made simultaneous Ca²⁺ fluorescence recordings. Fig. 2 A shows that a 5-s depolarization failed to activate an inward Ca²⁺ current or produce any change in intracellular [Ca²⁺]. In summary, no detectable Ca²⁺ currents were observed in either immature or mature DC (<0.1 pA/pF, n = 25).

Next we tested for ATP-gated ion channels (19, 20), which are expressed in many leukocytes including macrophages (21) and T lymphocytes (22, 23, 24). We found that ATP (10–500 μm) produced similar large Ca²⁺ transients in the majority of immature and mature DC tested (109/121). From calibration experiments with the calcium ionophore, ionomycin (see Materials and Methods), we estimated that 100 μM ATP increased free Ca²⁺ to 2.6 ± 0.9 μM (n = 4). The ATP-evoked responses desensitized both during ATP application (Fig. 2,C) and with repeated applications. In addition, ADP evoked a similar response to ATP (data not shown). Dual ATP/ADP sensitivity is characteristic of the metabotropic P2Y class of receptors (20). Indeed, simultaneous voltage clamping and Ca²⁺ imaging in NMDG⁺ based medium confirmed that the ATP-evoked Ca²⁺ rise was largely independent of inward Ca²⁺ current (Fig. 2,B). In some experiments small inward currents were observed (20 ± 4 pA, n = 5), but these currents occurred at variable times following ATP application and did not always correlate with the Ca²⁺ transient. Furthermore, we found that the magnitude of ATP-induced Ca²⁺ rises was unaffected when Ca²⁺ was removed from the external medium (Fig. 2 C; control 277 ± 20%, n = 24, zero Ca²⁺, 275 ± 28%, n = 18), although the duration was considerably shortened (τ = 89 ± 20 s vs τ = 22 ± 3 s, p < 0.01) due to elimination of a late Ca²⁺ plateau or hump. This Ca²⁺ hump may be the result of store-operated Ca²⁺ entry as described below. These results suggest that mouse myeloid DC predominantly express metabotropic purinoceptors, which mobilize Ca²⁺ by formation of IP₃ and not via plasma membrane ATP-gated ion channels. This is consistent with the recent report of functional P2Y type receptors in human myeloid DC (25).

Next, we examined whether a presumed voltage-insensitive L-type Ca²⁺ channel, recently identified in human myeloid DC (26), was present in mouse myeloid DC. Similarly we found that 10–25 μM Bay K8644 (an L-type channel agonist) evoked large Ca²⁺ rises (Fig. 3,A, n = 20). However, inward currents did not accompany these Ca²⁺ rises. In addition, we found that nifedipine (an L-type channel antagonist) also elicited Ca²⁺ rises, and these Ca²⁺ responses persisted in Ca²⁺-free medium (Fig. 3,B). In contrast, pretreatment with thapsigargin to deplete Ca²⁺ stores occluded the Bay K8644-evoked response (Fig. 3 C). These results indicate that 1,4 DHPs do not induce Ca²⁺ entry through a plasmalemmal channel, but rather mobilize Ca²⁺ from internal stores.

The presence of store-operated channels (SOCs) in DC was investigated by treatment of Fluo-3-loaded cells with the microsomal Ca²⁺-ATPase inhibitor, thapsigargin. This is a commonly used method for activating SOCs in other cell types (4, 27). In Fig. 4 A, the upper trace shows the Ca²⁺ fluorescence of a single cell, whereas the lower trace shows the mean fluorescence of 10 cells from the same experiment. Application of thapsigargin in <10 nM bathing Ca²⁺ (0.2 mM EGTA and no added Ca²⁺) produced a small increase in cytosolic Ca²⁺ due to depletion of internal stores, and this slowly declined over 3 min. Reapplication of 2 mM external Ca²⁺ induced a large and sustained increase in intracellular Ca²⁺. Similar responses to thapsigargin treatment were observed in the majority of immature (38/39) and mature (25/26) DC tested. No responses were seen when cells were incubated in zero Ca²⁺ without thapsigargin. This dependence of Ca²⁺ entry on prior Ca²⁺ mobilization suggests that SOCs mediate the entry. We tested several common pharmacological blockers of SOCs. SKF 96365 (10 μM) completely inhibited the response (n = 20), as did 1 mM Cd²⁺ (n = 20). In contrast, 100 μM Cd²⁺ (n = 10), a concentration sufficient to inhibit voltage-gated Ca²⁺ channels but not SOCs, and nimodipine (n = 10), a specific L-type Ca²⁺ channel blocker, had no significant effect (data not shown). These results suggest that the Ca²⁺ influx following thapsigargin treatment was via SOCs.

To investigate the properties of these SOCs further, and to determine whether these channels carried a nonspecific cation current or alternatively a Ca²⁺ selective current (I_CRAC), we conducted whole-cell voltage clamp experiments. In these experiments Ca²⁺ stores were depleted by the inclusion of IP₃ together with the Ca²⁺ chelator BAPTA in the patch pipette solution. Fig. 5,A shows the whole cell currents elicited by voltage ramps from −120 to +100 mV in the presence of external solutions containing different divalent cations. In Fig. 5,B (from the same experiment as in Fig. 5,A) the I_CRAC develops after break-in. The current is robust and reproducibly altered with different divalent cations. Inward currents were activated at hyperpolarized potentials with either 10 mM Ca²⁺, Ba²⁺, or Sr²⁺. Most of the inward current was inhibited when Mg²⁺ replaced Ca²⁺. The remaining inward current and the outward current in Mg²⁺ most likely represents leak current. Subtracting this leak current from the other currents revealed the pure I_CRAC. Consistent with I_CRAC in other cells (12, 28) this current exhibits a characteristic inward rectification (inset). The relative conductivity was Ca²⁺ > Ba²⁺ > Sr²⁺ with Ca²⁺ conductivity roughly 2-fold greater than Ba²⁺ and Sr²⁺. A similar permeability sequence was reported for I_CRAC in T lymphocytes (27). The Ca²⁺ current density at −80 mV was ∼0.7 pA/pF (n = 3) and again is similar to values reported for T lymphocytes of ∼1 pA/pF (12, 29). I_CRAC is known to be highly selective for divalent over monovalent cations. In agreement with this, we found that I_CRAC in DC was essentially independent of the external [Na⁺] (Fig. 6,A). In addition, the current was reversibly inhibited by 1 μM SKF 96365 (Fig. 6 B). This inhibition by SKF 96365 is consistent with the block of Ca²⁺ entry observed in our imaging experiments. These results indicate that SOC in DC are Ca²⁺-selective channels and similar to the channels that mediate I_CRAC.

We next considered whether CRAC is activated under physiological conditions. ATP is a putative DC chemotactic factor (25) and may attract DC to sites of cell injury and inflammation and stimulate DC maturation (30, 31). ATP-evoked Ca²⁺ responses exhibited two components: a fast Ca²⁺ transient that was independent of external Ca²⁺ and a slower Ca²⁺-dependent plateau (Fig. 2,C). This slower ATP-evoked Ca²⁺ response suggests that a SOC may be activated in DC during purinergic receptor signaling. To explore this further we studied the dependence of this delayed ATP-induced Ca²⁺ entry on both external Ca²⁺ and voltage. In these experiments DC were either perfused with the standard saline solution (4 mM K⁺) to generate a normal resting potential of −50 to −60 mV, or a solution containing 140 mM K⁺ to clamp the membrane potential close to 0 mV. In separate experiments where DC were held under current-clamp, we confirmed that high K⁺ does depolarize cells to ∼0 mV. This dependence of membrane potential on external [K⁺] may arise from the presence of leak or voltage-activated K⁺ currents (32). Fig. 7 shows that application of 100 μM ATP in Ca²⁺-free medium evoked a rapid Ca²⁺ transient in a DC. The readdition of Ca²⁺ induced a smaller, long lasting Ca²⁺ rise but only when the DC was held at a negative resting potential; no Ca²⁺ rise was seen when the cell was depolarized to 0 mV, and no Ca²⁺ rise was seen when cells were incubated in zero Ca²⁺ (up to 3 min) without ATP (F/F₀ = 0.98 ± 0.04, n = 20), ruling out the possibility that CRAC was activated by a passive loss of Ca²⁺. In a total of 10 cells, ATP evoked a normalized (F/F₀) Ca²⁺ plateau of 1.3 ± 0.2. The voltage dependence of the delayed ATP-evoked Ca²⁺ entry is consistent with it being mediated by CRAC.

Next we assessed whether CRAC participates in DC maturation. A heterogenous culture of immature and mature mouse myeloid DC were incubated overnight (18 h) with 50 nM thapsigargin to activate CRAC, and then double immunostained as described in Materials and Methods to detect their surface expression of MHC class II and costimulatory molecules. Fig. 8 shows that untreated controls were a heterogenous mix of immature (MHC class II^−/dim, CD80^−/dim, CD86^−/dim) and mature (MHC class II^high, CD80^high, CD86^high) DC. In contrast, DC exposed to thapsigargin overnight were homogenously mature, and expressed high levels of MHC class II, CD80, and CD86. These results complement those previously described for thapsigargin in myeloid leukocytes, including transformed cell lines, monocytes, and cultured bone marrow cells (10). Thapsigargin also mobilizes Ca²⁺ from intracellular stores. To test that stimulation by thapsigargin depended on Ca²⁺ entry via CRAC, we repeated the experiment in a low Ca²⁺ medium (1 μM free); however, under these conditions DC viability was reduced by ∼50%. As an additional test of the involvement of CRAC in DC maturation, we cultured DC overnight in medium containing different free Ca²⁺ concentrations (0.001, 0.36, and 4.6 mM) without any other stimuli. Because our results above indicate that CRAC is the major Ca²⁺ entry pathway, then varying the transmembrane Ca²⁺ gradient should mainly affect current via CRAC. Fig. 9 shows that the percentage of CD86⁺ DC (right-hand peak) increased in direct proportion to [Ca²⁺]. Thus, this result suggests that modulating the passive Ca²⁺ entry via CRAC is capable of influencing the spontaneous, in vitro maturation of DC.

This study constitutes the first in-depth investigation of Ca²⁺ signaling in DC. Our results demonstrate that mouse myeloid DC (both immature and mature) express SOCs with the properties of CRAC, but express neither functional voltage-dependent Ca²⁺ channels nor DHP-gated channels and few if any ATP-gated ion channels. Importantly, this suggests that CRAC is likely to be the major Ca²⁺ entry pathway in DC and thus an intrinsic component of the Ca²⁺ signaling processes that drive DC maturation and migration. The absence of voltage-dependent channels is not surprising given that myeloid DC, like other leukocytes, are essentially nonexcitable, and voltage-gated Ca²⁺ channels have not been clearly demonstrated in leukocytes (4). In contrast, ATP-gated cation channels are expressed in macrophages (closely related myeloid-lineage cells) (21), mast cells (33), and T cells (22, 23, 24). ATP-gated channels are Ca²⁺ permeable. In macrophages they are important in lipopolysaccharide-activated inflammatory responses (34, 35), in mast cells they modulate histamine secretion (33), and in T lymphocytes they may play a role in cell differentiation (24) or death (22, 23). Our results indicate that ATP signaling in DC predominantly occurs via the metabotropic (P2Y) class of receptors. In support of this, ATP responses were recorded without accompanying membrane current and in zero external Ca²⁺, and similar responses were seen with ADP. These results agree with a previous patch-clamp study using human myeloid DC (25). Several types of P2Y receptor have been identified in human DC (30, 36). The P2Y receptor is structurally similar to the chemokine receptor family; both are seven-transmembrane G protein-coupled receptors and they share identical intracellular signal transduction cascades (37). Therefore, P2Y-mediated signaling may contribute to chemotaxis, attracting DC to sites of inflammation. Consistent with this idea, ATP has been shown to alter DC shape and dendrite orientation (25). Although mRNA for the P2X channels (P2X_1,4,5,7) have been identified in DC by RT-PCR (30, 36, 38) functional evidence for channel expression is weak. For example, although Berchtold et al. observed that ATP (100 μM) evoked rapid Ca²⁺ rises, these were not affected by the buffering of external Ca²⁺. In contrast, there are reports that high ATP concentrations (0.75–5 mM) can permeabilize DC to low molecular mass dyes (36, 38, 39), indicating functional P2X₇ channels. The high concentrations required may explain why no P2X₇ currents were seen here with 10–500 μM ATP. Whether millimolar extracellular ATP levels play a biological role in DC function is unclear. Moreover, the fact that large Ca²⁺ transients can be evoked by low ATP (∼10 μM) via P2Y receptors suggests strongly that this latter pathway is more physiological. It is also significant that DC express high levels of membrane ATPase activity (30, 40) and that extracellular ATP is very rapidly hydrolyzed by DC (41). Thus, this would limit the activation of P2X₇ channels and serve to protect DC, relatively rare leukocytes, from ATP-induced apoptosis (22, 23, 41).

An important feature of this study is that we have clarified whether DC express functional DHP-gated channels. Poggi et al. (26) reported that human myeloid DC express the β subunit of L-type Ca²⁺ channels and that the DHP, Bay K8644, but not membrane depolarization, induced increases in intracellular [Ca²⁺]. These data were interpreted as evidence for the presence of nonvoltage-gated L-type Ca²⁺ channels. In this study we found that Bay K8644 could indeed produce Ca²⁺ increases in mouse myeloid DC. However, we further found that these Ca²⁺ increases were independent of external Ca²⁺ and changes in membrane conductance. These data are not inconsistent with Poggi et al. because they did not report whether Bay K8644 responses were dependent on external Ca²⁺. Our data clearly indicate that DHP-induced Ca²⁺ rises are not due to Ca²⁺ entry but rather the result of Ca²⁺ mobilization from internal stores. Accordingly, we found that responses to DHPs were occluded by emptying intracellular Ca²⁺ stores with thapsigargin. Moreover, we found that Ca²⁺ mobilization was also induced by nifedipine, an L-type channel antagonist. Thus, this strongly suggests that DHPs do not act via L-type channels. Just how DHPs mobilize Ca²⁺ is unclear. Nevertheless, the signaling pathway deserves further attention because nifedipine has been found to modulate numerous DC functions, including inhibition of Ag processing (42), apoptotic body engulfment, and IL-12 secretion (26).

The biophysical properties of CRAC in DC are similar to those reported in mast cells and Jurkat T cells (12, 29, 43). The relative conductance sequence is Ca²⁺ > Ba²⁺ ≥ Sr²⁺ ≫ Mg²⁺, with Ca²⁺ permeability approximately twice that of Ba²⁺ and Sr²⁺. The Ca²⁺ current density at −80 mV is ∼0.7 pA/pF, similar to that in Jurkat T cells (∼1 pA/pF). The current exhibits inward rectification such that the current at −60 mV is 3- to 4-fold larger than at 0 mV (Fig. 5 A), more than what would be expected from the difference in driving force. Again this is similar to that reported in T cells (44). Thus, modulation of resting membrane potential is likely to have marked effects on the degree of Ca²⁺ entry. Interestingly, the resting membrane potential of human myeloid DC becomes more hyperpolarized following activation with TNF-α (P.J.O. and G.P.A., unpublished observations), and this may serve to augment Ca²⁺ entry through SOCs.

It is notable that store-operated Ca²⁺ entry was observed in the majority of both immature and mature DC, indicating its importance in a range of DC functions. Ca²⁺ signaling is involved in DC maturation, chemotaxis, and migration to secondary lymphoid tissue (6, 7, 8, 9), thus, it is likely that CRAC plays an important role in all these processes. Our data are consistent with this proposal. First, ATP, a putative physiological activator of DC, induced store-operated Ca²⁺ entry with properties similar to CRAC (Figs. 2,C and 7). Second, activation of CRAC with thapsigargin induced marked DC maturation (Fig. 8). This finding in mouse myeloid DC is consistent with the previously reported observations for thapsigargin using human myeloid cell lines, monocytes, and DC (10, 11). Third, spontaneous DC maturation was directly proportional to the extracellular Ca²⁺ concentration (Fig. 9). Thus, a CRAC signaling pathway is likely to be involved in DC maturation. On a practical note, these results suggest that it may be useful to supplement DC culture medium (such as RPMI 1640), containing only 0.4 mM total Ca²⁺, with extra Ca²⁺ to optimize maturation.

In summary, this study has shown that a SOC with properties similar to CRAC plays a dominant role in DC Ca²⁺ signaling, with little contribution from the other Ca²⁺ entry pathways that are common in leukocytes.

We thank Alison J. Logar for expert assistance with flow cytometric data collection and analysis.