Selective Activation of the Transcription Factor NFAT1 by Calcium Microdomains near Ca2+ Release-activated Ca2+ (CRAC) Channels*

NFATs are a family of Ca2+-dependent transcription factors that play a central role in the morphogenesis, development, and physiological activities of numerous distinct cell types and organ systems. Here, we visualize NFAT1 movement in and out of the nucleus in response to transient activation of store-operated Ca2+ release-activated Ca2+ (CRAC) channels in nonexcitable cells. We show that NFAT migration is exquisitely sensitive to Ca2+ microdomains near open CRAC channels. Another Ca2+-permeable ion channel (TRPC3) was ineffective in driving NFAT1 to the nucleus. NFAT1 movement is temporally dissociated from the time course of the Ca2+ signal and remains within the nucleus for 10 times longer than the duration of the trigger Ca2+ signal. Kinetic analyses of each step linking CRAC channel activation to NFAT1 nuclear residency reveals that the rate-limiting step is transcription factor exit from the nucleus. The slow deactivation of NFAT provides a mechanism whereby Ca2+-dependent responses can be sustained despite the termination of the initial Ca2+ signal and helps explain how gene expression in nonexcitable cells can continue after the primary stimulus has been removed.

Cytoplasmic Ca 2ϩ is a universal signal that activates a myriad of spatially and temporally distinct cellular responses (1). Intrinsic to the use of such a multifarious intracellular messenger is the question of specificity. How can selective responses be induced by a signal capable of influencing numerous cellular activities simultaneously? It has now been firmly established that spatially restricted local Ca 2ϩ signals (Ca 2ϩ microdomains) provide a rapid, robust, and reliable route for selective activation of co-localized targets (2,3). Well understood examples utilizing Ca 2ϩ microdomains include the tight functional coupling between voltage-gated Ca 2ϩ channels and the rapidly releasable pool of secretory vesicles in presynaptic nerve terminals (2) and between inositol trisphosphate-sensitive Ca 2ϩ channels in the endoplasmic reticulum and the uniporter Ca 2ϩ channels in the mitochondrial inner membrane (4). A more complex scenario arises when the target is located at a considerable distance from the realm of the local Ca 2ϩ signal, as illustrated by stimulus-transcription coupling. Here, local Ca 2ϩ sig-nals accompanying the opening of either voltage-gated Ca 2ϩ channels or store-operated CRAC 2 channels in the plasma membrane activate Ca 2ϩ -dependent gene expression in the nucleus (5)(6)(7). This form of signaling is remarkable in two ways. First, the local Ca 2ϩ signal extends just a few nanometers from the channel pore yet influences events located several micrometers away (3). Second, a burst of channel activity for just a few seconds to minutes leads to gene expression several hours later (7). Hence, the impact of the Ca 2ϩ microdomain is expanded over several orders of magnitude in both the spatial and temporal domains.
One well studied Ca 2ϩ -dependent transcription factor is the nuclear factor of activated T cells (NFAT). Four members of the NFAT family (NFAT1-4) are stimulated by a rise in cytoplasmic Ca 2ϩ , and activated NFAT regulates numerous inducible genes that are essential for synaptic plasticity, axonal growth, and neuronal survival (8,9) as well as T cell development and the generation of effective immune responses (10). The Ca 2ϩ rise required for NFAT activation occurs through the opening of voltage-gated L-type Ca 2ϩ channels in hippocampal neurons (8) and CRAC channels in the plasma membrane of nonexcitable cells (11,12). In both cases, Ca 2ϩ influx leads to stimulation of the Ca 2ϩ -calmodulin-dependent protein phosphatase calcineurin, which dephosphorylates multiple residues on cytosolic NFAT (13,14). This exposes a nuclear localization sequence, resulting in NFAT import to the nucleus. Nuclear confined NFAT leads to gene expression, often in tandem with the AP-1 transcription factor complex of c-Fos and c-Jun (10). An essential role for CRAC channels in activation of NFAT in T cells has been established from studies on patients with a single point mutation (R91W) in the pore-forming Orai1 subunit (15). The R91W mutant CRAC channel fails to conduct Ca 2ϩ , and NFAT activation is suppressed.
In cultured hippocampal neurons, several pieces of evidence suggest that Ca 2ϩ microdomains near the L-type Ca 2ϩ channels selectively couple to NFAT activation. First, depolarization with high K ϩ solution triggers NFAT migration to the nucleus, and despite the presence of other types of voltage-gated Ca 2ϩ channel, this is prevented by L-type channel blockers (8,9). Second, loading the cytoplasm with the slow Ca 2ϩ chelator EGTA fails to impede NFAT movement (9). Third, calcineurin is tethered next to L-type channels through the anchoring protein AKAP79/150 (9) the Ca 2ϩ -NFAT interaction to nonexcitable cells, several questions remain. Is NFAT activated by Ca 2ϩ microdomains near CRAC channels or is a bulk rise in cytoplasmic Ca 2ϩ sufficient? If Ca 2ϩ acts locally, how local is local? How rapidly does NFAT migrate in and out of the nucleus? Using an NFAT1 construct tagged with GFP, we have addressed these issues using high resolution single cell live imaging. We find that NFAT1 activation is tightly coupled to Ca 2ϩ microdomains near open CRAC channels. We find evidence for strong short term memory in that NFAT1 movement into the nucleus outlasts the cytoplasmic Ca 2ϩ signal by several minutes. Our results also reveal that NFAT1 exits from the nucleus slowly and long after the Ca 2ϩ signal has been terminated. This relatively slow deactivation rate of NFAT ensures that a pulse of CRAC channel activity for a few minutes is sufficient to maintain NFAT in the nucleus for considerably longer, thereby promoting gene expression while obviating the detrimental effects of a sustained global Ca 2ϩ rise. Long lasting Ca 2ϩ -dependent responses can therefore be achieved through the slow deactivation (off rate) of an effector.

MATERIALS AND METHODS
Cell Culture-HEK293 cells were purchased from ATCC (via UK supplier LGC) and were cultured at 37°C with 5% CO 2 in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin. RBL-1 cells were also purchased from ATCC and cultured as described previously (16).
cDNA Constructs and Transfection-NFAT1-GFP was a kind gift from Dr. Paul Worley (The Johns Hopkins School of Medicine). The EGFP-based reporter plasmid (pNFAT-TA-EGFP) was a kind gift from Dr. Yuriy Usachev (University of Iowa). Orai1 was from OriGene, and STIM1-YFP was a generous gift from Dr. Tobias Meyer (Stanford University). HEK293 cells were transfected using the Lipofectamine method. All plasmids were used at 1 g, and experiments commenced 24 -48 h after transfection. Two sources of HEK293 cells stably expressing TRPC3 channels were used, one kindly provided by Dr. James Putney (NIEHS, National Institutes of Health) and the other kindly supplied by Dr. Michael Zhou (Ohio State University).
NFAT1-GFP Movement-NFAT1-GFP levels in the cytosol and nucleus were measured using the IMAGO charge-coupled device camera-based system from TILL Photonics, with a ϫ100 oil immersion objective (numerical aperture 1.3). Regions of interest of identical size were drawn in the cytosol and nucleus of each cell, and fluorescence was computed. Nuclear localization was confirmed by co-staining with a nuclear dye (DAPI; Fig. 1). Unless otherwise indicated, we calculated the nuclear/ cytosolic ratio of NFAT-GFP.
Gene Reporter Assay-24 -36 h following transfection with the EGFP-based reporter plasmid that contained an NFAT promoter, cells were stimulated with leukotriene C 4 for 40 min, and the percentage of cells expressing EGFP was measured. Gene expression was defined as fluorescence 3 ϫ S.D. Ͼ cell autofluorescence, measured in nontransfected cells. Cells were stimulated in culture medium and maintained in the incubator.
Ca 2ϩ Measurements-Cytoplasmic Ca 2ϩ imaging experiments were carried out at room temperature using the IMAGO charge-coupled device camera-based system from TILL Photonics, as described previously (17). Cells were alternately excited at 356 and 380 nm (20-ms exposures), and images were acquired every 2 s. Images were analyzed off line using IGOR Pro for Windows. Cells were loaded with Fura-2/AM (1 M) or Fura-5F/AM (Fig. 4G) for 40 min at room temperature in the dark and then washed three times in standard external solution composed of 145 mM NaCl, 2.8 mM KCl, 2 mM CaCl 2 , 2 mM MgCl 2 , 10 mM D-glucose, 10 mM HEPES, pH 7.4, with NaOH. Cells were left for 15 min to allow further de-esterification. Ca 2ϩ -free solution had the following composition: 145 mM NaCl, 2.8 mM KCl, 2 mM MgCl 2 , 10 mM D-glucose, 10 mM HEPES, 0.1 mM EGTA, pH 7.4, with NaOH. Ca 2ϩ signals are plotted as R, which denotes the 356/380 nm ratio.
Western Blotting-Total cell lysates (60 g) were analyzed by SDS-PAGE on a 10% gel. Membranes were blocked with 5% nonfat dry milk in PBS plus 0.1% Tween 20 (PBST) buffer for 1 h at room temperature (18). Membranes were washed with PBST three times and then incubated with primary Ab for 24 h at 4°C. Total ERK2 and GFP antibodies were from Santa Cruz Biotechnology and Cell Signaling, respectively, and used at dilutions of 1:5000 (ERK2) and 1:2000 (GFP), respectively. The membranes were then washed with PBST again and incubated with 1:2500 dilutions of peroxidase-linked anti-rabbit IgG from Santa Cruz Biotechnology for 1 h at room temperature. After washing with PBST, the bands were detected by an enhanced chemiluminescence (ECL) plus Western blotting detection system (Amersham Biosciences). Blots were analyzed by UNI-Scan software.
Statistical Analysis-Data are presented as the mean Ϯ S.E. Statistical significance was determined using Student's t test, where * denotes p Ͻ 0.05 and ** denotes p Ͻ 0.01.

RESULTS
NFAT Accumulates in the Nucleus following Opening of CRAC Channels-We initially investigated whether Ca 2ϩ entry through CRAC channels drives NFAT1 movement to the nucleus. Stimulation with the sarco/endoplasmic reticulum Ca 2ϩ -ATPase pump inhibitor thapsigargin raises cytoplasmic Ca 2ϩ by depleting intracellular Ca 2ϩ stores. In the absence of external Ca 2ϩ , thapsigargin (2 M) generated a transient Ca 2ϩ rise ( Fig. 1A; trace labeled 0 Ca 2ϩ ). However, this consistently failed to drive NFAT1 movement into the nucleus for up to 40 min after stimulation (Fig. 1, B and F). On the other hand, a more prolonged Ca 2ϩ rise occurred when thapsigargin was applied in the presence of external Ca 2ϩ (Fig. 1A), and this was associated with prominent NFAT1 movement into the nucleus (Fig. 1, C and F). Nuclear migration was confirmed by the strong co-localization between NFAT1-GFP and the nuclear dye DAPI after stimulation with thapsigargin in the presence of external Ca 2ϩ (Fig. 1D). Hence, Ca 2ϩ influx through CRAC channels leads to NFAT1 activation and translocation to the nucleus. Consistent with this, pretreatment with the CRAC channel inhibitor Synta compound In many cell types, CRAC channel activity is controlled by mitochondria (19). In RBL-1 mast cells, mitochondrial depolarization reduces Ca 2ϩ influx (16,20) and therefore subsequent downstream Ca 2ϩ -dependent responses (21). Consistent with this, mitochondrial depolarization following the combination of antimycin A and oligomycin reduced store-operated Ca 2ϩ entry (the global Ca 2ϩ rise upon readmission of external Ca 2ϩ to cells treated with thapsigargin in Ca 2ϩ -free solution was reduced by ϳ60%, data not shown) and suppressed NFAT1 movement to the nucleus (Fig. 1, E and F). These results complement recent studies in sensory neurons and T cells that found a prominent role for mitochondria in regulating NFAT movement into the nucleus in response to voltage-gated Ca 2ϩ channel and CRAC channel activation (22,23).
To obtain a better estimate of the kinetic relationship between Ca 2ϩ entry through CRAC channels and NFAT1 movement, we stimulated cells with thapsigargin in Ca 2ϩ -free solution to deplete the stores and, once cytoplasmic Ca 2ϩ had returned to resting levels, we readmitted external Ca 2ϩ ( Fig.  2A) and measured NFAT1 movement to the nucleus as a function of the duration of Ca 2ϩ influx. No resolvable movement occurred within 4 min of Ca 2ϩ readmission, but migration was discernible after 8 min (Fig. 2B). Aggregate data from several cells are depicted in Fig. 2C. The relationship was best fitted by a sigmoidal function, yielding a lag phase of ϳ4 min followed by an increase in nuclear migration with time. The lag phase likely reflects NFAT dephosphorylation in the cytosol (Fig. 5), a process that precedes NFAT movement.
NFAT1 Movement Shows Short Term Memory-NFAT1 translocation to the nucleus did not correlate tightly with the time course of the cytoplasmic Ca 2ϩ signal. Although cytoplasmic Ca 2ϩ was elevated within 1 min of stimulation with thapsigargin (Fig. 1A), NFAT1 migration to the nucleus was not detectable until several minutes later (Fig. 1C). Furthermore, the Ca 2ϩ signal in the presence of external Ca 2ϩ decreased with time ( Fig. 1A), due to Ca 2ϩ -dependent inactivation of CRAC channels, yet NFAT1 movement continued. For example, there was little difference in cytoplasmic Ca 2ϩ between 20 and 40 min of thapsigargin stimulation (Fig. 1A), yet NFAT1 movement increased over this time period (Fig. 1C). Hence, NFAT1 continues to migrate to the nucleus even after Ca 2ϩ returns to basal levels, indicating a form of memory, as has been described previously (24). To obtain a more quantitative measure of this phenomenon, we applied a pulse of Ca 2ϩ entry for 10 min and measured subsequent NFAT1 movement for a further 40 min. Stores were depleted following exposure to thapsigargin in Ca 2ϩ -free solution, and then external Ca 2ϩ was readmitted before cells were perfused with Ca 2ϩ -free solution once more (Fig. 2D). The cytoplasmic Ca 2ϩ signal because of Ca 2ϩ influx fell rapidly upon exposure to Ca 2ϩ -free solution (Fig. 2D), but NFAT1-GFP continued to migrate to the nucleus for several minutes (Fig. 2E). Kinetic analysis of the movement showed that NFAT continued to accumulate within the nucleus for 20 min after cytoplasmic Ca 2ϩ had recovered to resting levels (Fig. 2F), indicating prominent short term memory. NFAT1 Movement and Local Ca 2ϩ Influx-We designed experiments to explore whether NFAT1 movement was activated by local Ca 2ϩ entry through CRAC channels. A hallmark of processes driven by Ca 2ϩ microdomains is that they are much more sensitive to the Ca 2ϩ flux through single channels than the bulk cytoplasmic Ca 2ϩ rise (3). To manipulate the single channel Ca 2ϩ current, we varied external Ca 2ϩ concen-tration over the range 0.25-2 mM. The Ca 2ϩ response in 0.25 mM external Ca 2ϩ was significantly smaller than that in 2 mM Ca 2ϩ (Fig. 3A) and was completely ineffective in driving NFAT1 movement to the nucleus (Fig. 3, B and C). Stimulation with thapsigargin in 0.5 mM external Ca 2ϩ produced similar bulk Ca 2ϩ signals to those seen in 2 mM Ca 2ϩ (Fig. 3A). However, despite this, stimulation with thapsigargin in 0.5 mM Ca 2ϩ  failed to generate any significant movement of NFAT1 movement into the nucleus (Fig. 3, B and C).
A further prediction of dependence on Ca 2ϩ microdomains is that a slow Ca 2ϩ chelator such as EGTA should have little effect on NFAT1 movement in response to CRAC channel opening. Because of its slow on-rate for binding Ca 2ϩ , EGTA has no impact on the extent of the Ca 2ϩ microdomain within a few tens of nanometers of the CRAC channel pore (2, 3). We therefore loaded cells with EGTA (by incubation in 25 M EGTA-AM for 45 min) and then stimulated them with thapsigargin in the presence of 2 mM Ca 2ϩ . The bulk Ca 2ϩ signal was substantially reduced in EGTA-loaded cells (Fig. 3D), as expected from an increase in cytoplasmic Ca 2ϩ buffering. However, NFAT1 migration was largely unaffected (Fig. 3, E and F).
NFAT1 Activation Is Tightly Coupled to CRAC Channels-We asked if NFAT1 could be activated by a generalized rise in subplasmalemmal Ca 2ϩ or whether it was tightly coupled to CRAC channels. A prediction of the former mechanism is that other Ca 2ϩ -permeable ion channels should be capable of activating NFAT1. We expressed TRPC3 channels, which are nonselective cation channels permeable to Ca 2ϩ (25). Activation of these channels with the diacylglycerol analog 1-oleoyl-2-acetylsn-glycerol (OAG, 50 M) evoked robust bulk Ca 2ϩ signals that were similar in size to those generated following endogenous CRAC channel activation ( Fig. 4A; OAG signals are shown in gray). However, NFAT1 failed to move to the nucleus upon activation of TRPC3 ( Fig. 4B; aggregate data is shown in Fig.  4C). Nontransfected HEK293 cells failed to respond to OAG, confirming that the Ca 2ϩ signals were indeed a consequence of TRPC3 expression (data not shown). NFAT1 movement out of the nucleus (using a protocol described in Fig. 5) was not accelerated by OAG exposure (data not shown), suggesting movement into the nucleus was not being masked by accelerated nuclear exit. Overexpression of Orai1 (with STIM1) increased NFAT1 migration significantly (Fig. 4, D and E), along with a faster rate of Ca 2ϩ entry (Fig. 4F), compared with the corresponding rates with endogenous CRAC channel activity. Collectively, these results reveal that a generalized rise in subplasmalemmal Ca 2ϩ is not sufficient for NFAT movement. Rather, local Ca 2ϩ signals close to CRAC channels are critical.
Calculations estimate the local Ca 2ϩ concentration near CRAC channels to be in the low micromolar range (3). One might therefore expect that if a large and sustained bulk increase in cytoplasmic Ca 2ϩ , arising exclusively from Ca 2ϩ release, raises subplasmalemmal Ca 2ϩ to a similar extent, then this should also activate NFAT1 movement. To test this, we used a method described previously (7,26,27) that produces a large and prolonged Ca 2ϩ signal through the combined exposure to thapsigargin in Ca 2ϩ -free external solution (to prevent any Ca 2ϩ entry) supplemented with La 3ϩ . La 3ϩ inhibits the plasma membrane Ca 2ϩ ATPase pump, reducing Ca 2ϩ extrusion across the plasma membrane. Stimulation with thapsigargin in the presence of Ca 2ϩ -free solution containing 1 mM La 3ϩ led to a robust and prolonged Ca 2ϩ signal, considerably larger than that seen in the absence of La 3ϩ or in response to thapsigargin in 2 mM Ca 2ϩ (Fig. 4G), and this larger signal was associated with strong NFAT1 movement to the nucleus (Fig. 4, H  and I). Hence, NFAT1 migration is not absolutely dependent on Ca 2ϩ entry; Ca 2ϩ release also drives movement provided it is of sufficient amplitude and duration.
Kinetic Features of NFAT1 Activation and Nuclear Dynamics-Accumulation of NFAT1 within the nucleus is a multistep process involving the following: (i) NFAT1 dephosphorylation in the cytoplasm by calcineurin; (ii) NFAT1 movement into the nucleus, and (iii) NFAT1 exit from the nucleus. We measured the kinetics of each of these steps following CRAC channel activation, to compare the Ca 2ϩ signal with NFAT dynamics. NFAT dephosphorylation can be monitored by tracking the gel . NFAT1 movement is tightly coupled to CRAC channels. A, Ca 2ϩ signals evoked by CRAC channels and TRPC3 channels (activated by OAG) are compared. B, images show the lack of effect of OAG on NFAT1 movement. C, aggregate data are shown comparing the basal nuclear/cytoplasmic (N/C) NFAT ratio with that seen after 40 min of stimulation with OAG (n ϭ 16 cells). D, images show the effect of overexpression of STIM1 and Orai1 (S1/O1) on NFAT movement. E, time course of NFAT movement is compared between control cells (with endogenous CRAC channel activity) and cells transfected with STIM1 and Orai1 (n ϭ 6 -9 cells for each condition). F, histogram compares the rate of rise of the Ca 2ϩ signal, obtained upon readmission of external Ca 2ϩ to cells with depleted stores, for the two conditions shown (n ϭ 16 for control and n ϭ 11 for S1/O1). G, inhibition of Ca 2ϩ extrusion with 1 mM La 3ϩ increases the size and prolongs the time course of the Ca 2ϩ rise to thapsigargin (Thap) in Ca 2ϩ -free external solution (n ϭ 21 for 0Ca 2ϩ , n ϭ 20 for 0Ca 2ϩ ϩ La 3ϩ , n ϭ 21 for 2 Ca 2ϩ ).
In these experiments, we used the low affinity dye Fura-5F because Fura-2 would have been saturated by the larger Ca 2ϩ signal. H, images show NFAT1 movement following stimulation with thapsigargin in Ca 2ϩ -free solution supplemented with La 3ϩ . I, aggregate data are compared for the conditions shown (n ϭ 4 cells).
shift that occurs upon dephosphorylation of NFAT (Fig. 5A). As shown in Fig. 5B, dephosphorylation proceeded relatively rapidly, reaching completion within 10 min. This is in good agreement with previous studies that reported substantial dephosphorylation within 5 min (24,28). To measure migration out of the nucleus, we initially triggered NFAT1-GFP movement into the nucleus by stimulation with thapsigargin in 2 mM external Ca 2ϩ for 20 min and then suppressed Ca 2ϩ influx by removing external Ca 2ϩ and simultaneously inhibited calcineurin with cyclosporin A. NFAT migrated slowly out of the nucleus ( Fig.  5C; aggregate data from several cells is summarized in Fig. 5D). The schematic in Fig. 5E summarizes the half-times for each of the key steps in NFAT1 dynamics. This is not a quantitative model and suffers from some limitations. First, NFAT1 movement has been measured at the single cell level, whereas the kinetics of cytoplasmic dephosphorylation was derived from Western blots and is therefore an ensemble of thousands of cells. Individual cellular dephosphorylation kinetics, which could vary significantly between cells, cannot be related to NFAT1 movement in the same cell. Second, NFAT1 exit from the nucleus is not a true unidirectional measurement because it occurs in the presence of NFAT1 migration into the nucleus, although precautions were taken to abolish the latter process by abolishing Ca 2ϩ flux through CRAC channels together with inhibition of calcineurin. The presence of cyclosporin A prevented the continued movement of NFAT-GFP upon removal of external Ca 2ϩ that was seen in experiments of the type described in Fig. 2 (data not shown). Third, we were unable to separate the two steps involved in NFAT1 nuclear exit, namely nuclear phosphorylation and efflux. The t 1 ⁄ 2 of 24.2 min in Fig. 5 combines both together. Finally, it is conceivable that the GFP tag on NFAT1 impairs its transport into and out of the nucleus, and to different extents. Nuclear import and export of an NFAT-GFP construct in sensory neurons in response to voltage-gated Ca 2ϩ channel activity are both significantly faster (ϳ2-and 5-fold, respectively) than the values we report here, suggesting the presence of the GFP tag per se might not compromise NFAT dynamics significantly. Notwithstanding these issues, the model reveals NFAT migration out of the nucleus as the step with the slowest kinetics.
Local Ca 2ϩ Entry Evoked by Agonist Drives NFAT1 to the Nucleus-To examine whether local Ca 2ϩ entry through CRAC channels stimulates NFAT1 movement to the nucleus in response to physiological levels of agonist, we expressed NFAT1-GFP in a mast cell line and stimulated endogenous cysteinyl leukotriene receptors with a sub-maximal concentration (120 nM) of LTC 4 . This concentration of agonist triggers repetitive Ca 2ϩ oscillations in the presence of external Ca 2ϩ (Fig. 6A,  black trace) or in the absence of external Ca 2ϩ provided La 3ϩ is present extracellularly to inhibit Ca 2ϩ extrusion via the plasma membrane Ca 2ϩ ATPase pump (Fig. 6A, gray traces) (7). In the former case, Ca 2ϩ entry through CRAC channels refills the stores and sustains the oscillatory response. In the presence of La 3ϩ , Ca 2ϩ released from the endoplasmic reticulum is pumped back into the stores to support repetitive Ca 2ϩ release. The oscillations have identical amplitudes and frequencies, but the underlying spatial Ca 2ϩ gradients are different (7). Only those oscillations that occur in the presence of external Ca 2ϩ generate local Ca 2ϩ signals near open CRAC channels, which selectively couple to gene expression (7). Stimulation with LTC 4 in the presence of 0 Ca 2ϩ plus La 3ϩ consistently failed to trigger an increase in NFAT1 migration to the nucleus (Fig. 6, B and C). However, significant movement occurred when the same dose of agonist was applied in the presence of external Ca 2ϩ , despite evoking Ca 2ϩ oscillations of the same amplitude and frequency. Hence, local Ca 2ϩ entry through CRAC channels drives NFAT1 migration to the nucleus when low doses of a physiological agonist are used. Consistent with this, loading the cytoplasm with the slow Ca 2ϩ chelator EGTA (through preincubation with EGTA-AM) failed to prevent NFAT1 movement into the nucleus following stimulation with LTC 4 (Fig. 6C).
Local Ca 2ϩ Entry by Agonist Drives NFAT-dependent Gene Expression-To examine whether NFAT1 migration to the nucleus in response to Ca 2ϩ microdomains near CRAC channels led to gene expression, we transfected cells with an EGFP plasmid driven by an NFAT promoter. Whereas very few cells expressed EGFP in the absence of stimulation (Fig. 6, D and F), exposure to 120 nM LTC 4 (for 40 min) led to a significant increase in the number of EGFP-positive cells (Fig. 6, E and F). Blocking CRAC channels with the Synta compound (5 M; 10 min pretreatment) completely abolished gene expression in response to LTC 4 (Fig. 6F). As we have shown recently, LTC 4 still evokes Ca 2ϩ oscillations in the presence of CRAC channel FIGURE 6. Local Ca 2؉ entry evoked by an agonist drives NFAT1 movement. A, stimulation of Fura-5F-loaded cells with 120 nM LTC 4 in the presence of external Ca 2ϩ (black trace) or absence of external Ca 2ϩ with 1 mM La 3ϩ evokes Ca 2ϩ oscillations of similar amplitude and frequency. B, images show distribution of NFAT1-GFP (left panels) and DAPI (right panels) for rest (nonstimulated), 0Ca 2ϩ ϩ La 3ϩ (exposed to LTC 4 for 10 min), and 2Ca 2ϩ (LTC 4 was applied for 10 min). C, aggregate data from several cells is shown. N/C, nuclear/cytoplasmic ratio. D, EGFP gene expression is compared between nonstimulated (control) cells and those exposed to 120 nM LTC 4 (40 min). E, histogram compares the number of EGFP-positive cells for the different conditions. Data are the aggregate from six fields of view from three coverslips each. blockers (7). However, these oscillations are similar to those obtained in Ca 2ϩ -free solution and run down more quickly than those obtained in 2 mM external Ca 2ϩ in the absence of CRAC channel blockers. Whereas Ca 2ϩ oscillations to 120 nM LTC 4 in 2 mM Ca 2ϩ continue for 10 min of stimulation, those in the presence of CRAC channel blockers are lost after ϳ7 min (supplemental Fig. 1). Nevertheless, 4 Ϯ 1 robust Ca 2ϩ oscillations occurred in the presence of the Synta compound, yet these failed to evoke any detectable gene expression. Our attempts to evoke sustained Ca 2ϩ oscillations by stimulating with LTC 4 in Ca 2ϩ -free solution supplemented with 1 mM La 3ϩ were thwarted by the fact that La 3ϩ was precipitated in the culture medium.
Concluding Remarks-As with activation of NFAT in neurons following the opening of L-type Ca 2ϩ channels, several independent pieces of evidence reveal that NFAT1 activation is acutely tuned to Ca 2ϩ microdomains near open CRAC channels in the nonexcitable cells we have examined. First, NFAT1 migration to the nucleus was minimal when cells were challenged in reduced external Ca 2ϩ (0.5 mM), despite global Ca 2ϩ being similar to that obtained in 2 mM Ca 2ϩ . Second, the slow Ca 2ϩ chelator EGTA had little effect on NFAT1 movement despite reducing bulk Ca 2ϩ . Third, activation of other Ca 2ϩpermeable pathways, including TRPC3 channels, failed to elicit NFAT1 movement despite a similar rise in bulk Ca 2ϩ to that evoked by CRAC channel opening. This latter result reveals that not all Ca 2ϩ -permeable pathways are equally effective in activating NFAT1, despite raising bulk Ca 2ϩ to similar extents. By contrast, overexpression of Orai1 greatly accelerated NFAT1 movement into the nucleus. Hence the level of CRAC channel expression will impact significantly on the rate and extent of NFAT1 movement. How can NFAT1 be tightly linked to local Ca 2ϩ signals? The Ca 2ϩ sensor in the NFAT activation pathway must presumably be co-localized with the CRAC channel. The sensor is probably calmodulin, which, when occupied by Ca 2ϩ , activates calcineurin. Recent work has revealed that calmodulin can bind to the amino terminus of Orai1 (29), providing a mechanism whereby the sensor for the NFAT1 pathway can detect local Ca 2ϩ signals near CRAC channels. In hippocampal neurons, calcineurin is held close to the L-type Ca 2ϩ channel through AKAP79/150 (9). It will be interesting to see whether such scaffolding proteins play a similar role in nonexcitable cells.
Although our findings reveal an important role for local Ca 2ϩ entry through CRAC channels in driving NFAT movement to the nucleus and subsequent gene expression, they do not rule out a role for global Ca 2ϩ under certain conditions. Our experiments using stimulation with thapsigargin, 0 Ca 2ϩ /La 3ϩ demonstrate global Ca 2ϩ can drive NFAT movement, provided it is high enough to elevate subplasmalemmal Ca 2ϩ near CRAC channels and thus occupy the Ca 2ϩ sensor. It has been established that a sustained Ca 2ϩ rise of more than ϳ400 nM for ϳ2 h is required for NFAT-dependent expression of genes that commit T cells to activation (30,31). Although it is not clear whether NFAT activation is dependent on local or global Ca 2ϩ signals in these cells, it is interesting to note that the commonly used methods for raising Ca 2ϩ in such experiments involve either activation of the T cell receptor or application of the ionophore ionomycin. Both approaches open CRAC channels and therefore will produce local Ca 2ϩ signals below the membrane that will diverge significantly from bulk Ca 2ϩ measurements. These spatial gradients will be accentuated by the substantial increase in plasma membrane Ca 2ϩ ATPase pump activity that occurs following CRAC channel opening in T cells (32). It is noteworthy that stimulation with thapsigargin in 0.5 mM Ca 2ϩ elicited a similar bulk Ca 2ϩ rise and of similar duration to that triggered by thapsigargin in 2 mM Ca 2ϩ , but only the latter led to NFAT1 movement. Hence, the duration of the bulk Ca 2ϩ rise does not seem as important as the properties of the local Ca 2ϩ signal in activating NFAT1, at least under our experimental conditions.
Local Ca 2ϩ signals arising from focal release of Ca 2ϩ into the nucleus by perinuclear inositol trisphosphate receptors have been reported to activate the calcineurin/NFAT pathway in atrial myocytes (33). NFAT activation can therefore occur in response to local Ca 2ϩ gradients arising from either Ca 2ϩ release directly into the nucleus or through Ca 2ϩ entry via plasma membrane CRAC channels. It is conceivable that these two components reinforce one another in response to physiological stimuli. Alternatively, they could operate independently, enabling selective activation of the NFAT pathway through recruitment of different phases of the Ca 2ϩ signal.
At the immunological synapse, which forms on a T cell membrane where it apposes an antigen-presenting cell, receptors, signaling molecules, and Orai1 and STIM1 can accumulate (34). It is tempting to speculate that local Ca 2ϩ signals through CRAC channels confined to the immunological synapse might selectively drive NFAT activation and subsequent NFAT-dependent gene expression.
The marked temporal dissociation between the time course of the cytoplasmic Ca 2ϩ rise and NFAT1 migration to the nucleus reveals a form of short term memory that maintains gene expression long after the Ca 2ϩ stimulus has been removed. A pulse of Ca 2ϩ influx for ϳ10 min was sufficient for NFAT1 to continue to accumulate in the nucleus even 20 min later. Similar temporal uncoupling between the Ca 2ϩ signal and NFAT movement has been seen in cultured hippocampal neurons, where a depolarization for a few minutes was sufficient for continued NFAT accumulation within the nucleus (8,9). Such temporal uncoupling between the Ca 2ϩ signal and NFAT1 activation provides a mechanism for sustaining Ca 2ϩ -dependent nuclear events without the obvious disadvantages that arise from a protracted bulk Ca 2ϩ signal (35,36). NFAT-dependent gene expression can be maintained by repetitive cytoplasmic Ca 2ϩ oscillations (37), with a periodicity of Ͻ400 s in T lymphocytes (38). NFAT movement out of the nucleus in studies on T cell populations is considerably slower than the decay of the Ca 2ϩ signal, with almost 50% of the NFAT pool remaining within the nucleus 10 min after cessation of the Ca 2ϩ spike (39). This is in good agreement with our single cell analysis and suggests each pulse of Ca 2ϩ tops up the nuclear NFAT pool sufficiently to maintain gene expression.
Finally, our kinetic analysis of the three main stages involved in NFAT1 nuclear accumulation reveals that NFAT1 activation (arising from cytoplasmic dephosphorylation) and migration into the nucleus are both considerably faster (ϳ10-and 3-fold, respectively) than NFAT movement out of the nucleus. Nuclear export of NFAT in hippocampal neurons is also relatively slow, with a reported half-time of ϳ10 min (8). Our results establish that the slow off-rate for NFAT1 (exodus from the nucleus) is a major factor in determining its nuclear buildup and thereby gene expression. Slow off-rates in Ca 2ϩ -dependent signaling might therefore be an effective general strategy for ensuring long lasting responses following termination of the Ca 2ϩ trigger.