Role of the microtubule cytoskeleton in the function of the store-operated Ca²⁺ channel activator STIM1

STIM1 is a pivotal molecular component of the SOCE pathway (Roos et al., 2005; Liou et al., 2005). A major fraction of cellular STIM1 resides within the membrane of the ER (Dziadek and Johnstone, 2007), the primary Ca²⁺ storage compartment of mammalian cells; upon depletion of Ca²⁺ from the ER stores, STIM1 relocates to distinct puncta located in close proximity to the plasma membrane (Liou et al., 2005; Zhang et al., 2005). Once at the plasma membrane, STIM1 directly or indirectly activates members of the Orai class of SOCE channels, resulting in Ca²⁺ influx into the cell (Zhang et al., 2006; Mercer et al., 2006; Peinelt et al., 2006; Xu et al., 2006).

Molecular and functional data indicate that STIM1 acts as the ER Ca²⁺ sensor that responds to Ca²⁺ store depletion. First and foremost, STIM1 contains an N-terminal EF hand domain that is situated within the lumen of the ER (Dziadek and Johnstone, 2007). Mutation of Ca²⁺-binding residues within the EF hand domain render STIM1 constitutively active: that is, EF-hand-mutated STIM1 localizes to plasma membrane puncta even in the presence of replete Ca²⁺ stores, and SOCE is always active (Liou et al., 2005). Furthermore, Orai1 forms clusters within the plasma membrane in response to Ca²⁺ store depletion that colocalize with STIM1 puncta (Luik et al., 2006; Xu et al., 2006), and it has been shown that STIM1 and Orai1 co-immunoprecipitate (Yeromin et al., 2006; Vig et al., 2006) (but see Gwack et al., 2007). However, a direct interaction between STIM1 and Orai channels has not been demonstrated. Thus, although it is clear that STIM1 responds to Ca²⁺ store depletion and subsequently activates SOCE via Orai channels, the mechanism by which this activation occurs, including the mechanism of STIM1 rearrangement, remains unclear.

The precise localization and movements of intracellular signaling proteins is often governed by cytoskeletal elements, the major components being the actin cytoskeleton and the microtubule cytoskeleton. Most studies report that depolymerization of the actin cytoskeleton with cytochalasins does not negatively impact SOCE (Ribeiro et al., 1997; Patterson et al., 1999). Thus, the actin cytoskeleton does not appear to play an obligate role per se in the SOCE pathway. The role of the microtubule cytoskeleton in SOCE has also been investigated. However, as was also shown for the actin cytoskeleton (Ribeiro et al., 1997; Patterson et al., 1999), several studies have reported that nocodazole, a drug that causes depolymerization of the microtubular cytoskeleton, fails to inhibit SOCE or the I_CRAC, the current most often associated with SOCE, in RBL, NIH 3T3 and DT40 cells (Ribeiro et al., 1997; Bakowski et al., 2001; Baba et al., 2006).

Despite these negative reports, we (Mercer et al., 2006) and others (Baba et al., 2006) have shown that, when human STIM1 tagged with an N-terminal EYFP is overexpressed, STIM1 adopts an organization that is strikingly similar to that of the microtubule cytoskeleton. The microtubular cytoskeleton is known to be a major regulator of ER structure and function (Terasaki et al., 1986). Thus, microtubules might direct or influence the organization or movements of STIM1 in some way. In this study, we investigated the role of microtubules in the SOCE signaling pathway, with particular emphasis on the dependence of STIM1 function on microtubules. Our results indicate that microtubules might play a facilitative role in organizing STIM1 for optimal Ca²⁺ sensing and/or communication with Orai channels.

As previously reported (Mercer et al., 2006; Baba et al., 2006), when EYFP-STIM1 is overexpressed in HEK 293 cells, the protein adopts a configuration that is strikingly similar in appearance to a microtubule array. EYFP-STIM1 is organized in long filament-like structures, many of which appear to emanate from a central `organizing center' near the nucleus (Fig. 1A); this EYFP-STIM1 organizing center is highly reminiscent of a centrosome, which functions as the microtubule-organizing center in interphase cells. For comparison, Fig. 1B shows an HEK 293 cell that was fixed and immunostained with an antibody against tubulin. By comparison, as shown in Fig. 1C, the ER in HEK 293 cells has a lattice-like organization throughout the cell, which is quite distinct in appearance from the microtubule or EYFP-STIM1 arrays seen in these cells. We therefore hypothesized that STIM1 might occupy only a subset of the total cellular ER at any given time and that this specialized localization of STIM1 within the ER might be directed by microtubules in some way.

It has previously been reported that STIM1 colocalizes with microtubules in HeLa cells that were fixed and immunostained for α-tubulin (Baba et al., 2006). We initially performed our analysis on live HEK 293 cells co-overexpressing EYFP-STIM1 and an mCherry-α-tubulin construct (mCherry-tub). As shown in Fig. 2A, expression of the mCherry-tub construct resulted in prominent fluorescence in the cytoplasm, and there were only a limited number of well-defined microtubules that were clearly visible above this background. It is clear, however, that, for many of the filaments of mCherry-tub that are clearly discernible, there is an EYFP-STIM1 filament that exactly matches its localization (Fig. 2A, upper panel, arrowheads), indicating an association between the overexpressed mCherry-tub and EYFP-STIM1. We also performed a colocalization analysis on cells overexpressing EYFP-STIM1 that were fixed and immunostained with an antibody against α-tubulin and an antibody against GFP (which recognizes EYFP-STIM1). These cells exhibited a well-defined α-tubulin network, as well as an EYFP-STIM1 localization that was similar to that seen in live cells (Fig. 2B). In the fixed and immunostained samples, there was a high degree of colocalization between EYFP-STIM1 and α-tubulin (Fig. 2B, upper panel, merged image. Note, interestingly, EYFP-STIM1 appears to be excluded from the centrosome region.). Statistical analysis of this colocalization revealed that 85.4±0.02% (n=13 cells) of pixels that were positive for EYFP-STIM1 immunofluorescence were also positive for α-tubulin immunofluorescence. Conversely, 74.1±0.02% of α-tubulin-positive pixels were also positive for EYFP-STIM1. Thus, the majority of EYFP-STIM1 colocalizes with α-tubulin, which supports the idea that STIM1 is organized into a fibrillar distribution through direct or indirect association with microtubules.

We also analyzed the localization of EYFP-STIM1 and α-tubulin in cells in which intracellular Ca²⁺ stores were depleted with thapsigargin (Fig. 2A,B, lower panels). As has been shown previously (Liou et al., 2005; Zhang et al., 2005; Mercer et al., 2006; Baba et al., 2006; Wu et al., 2006), store depletion induces a rearrangement of EYFP-STIM1 from fibrillar structures to puncta that are located in close proximity to the plasma membrane. In the fixed and immunostained cells, it was possible to detect several tracks of α-tubulin along which EYFP-STIM1 puncta appear to associate (Fig. 2B, bottom panel, arrowheads). Thus, microtubules might direct the relocalization of STIM1 into puncta near the plasma membrane.

To test more directly whether STIM1 associates with microtubules, we monitored the localization of EYFP-STIM1 following microtubule depolymerization with nocodazole. As shown in Fig. 3, treatment of cells for 20 minutes with nocodazole (10 μM) caused EYFP-STIM1 to significantly rearrange from a filamentous configuration to one that looks strikingly similar to the configuration of the ER. In the same cell, nocodazole treatment induced only a marginal change in the ER conformation, as measured by the localization of an ER-targeted cyan fluorescent protein (CFP-ER). Accordingly, it is apparent in the merged image that EYFP-STIM1 was fairly evenly distributed throughout the ER following nocodazole treatment. Cells overexpressing EYFP-STIM1 that were fixed and immunostained for α-tubulin demonstrated that the 20 minute treatment with nocodazole was sufficient to induce a significant disruption of the microtubule cytoskeleton (Fig. 3B, compare with Fig. 2B), although some long filaments were still evident in these cells. Interestingly, EYFP-STIM1 did not colocalize to any significant degree with these α-tubulin filaments that remained following nocodazole treatment. Colchicine, which is another drug known to disrupt microtubules (Oka et al., 2005), had similar effects on STIM1 distribution and microtubule structure, although the effects on microtubule structure were somewhat less extensive than those produced by nocodazole (data not shown; and supplementary material Fig. S1).

As microtubule depolymerization had a significant impact on the localization of EYFP-STIM1, it was therefore of interest to evaluate the effect of microtubule-depolymerizing drugs on STIM1-mediated SOCE. SOCE was activated by the sarco/endoplasmic reticulum Ca²⁺-ATPase (SERCA) inhibitor thapsigargin. As shown in Fig. 4, pretreatment of HEK 293 cells with either 100 μM colchicine or 10 μM nocodazole resulted in significant inhibition of the sustained [Ca²⁺]_i elevation following thapsigargin addition, suggesting inhibition of store-operated entry. Consistent with the morphological observations, the extent of inhibition was greater for nocodazole than for colchicine, and thus nocodazole was used for the majority of subsequent experiments.

The inhibition by microtubule-depolymerizing drugs of the thapsigargin-induced sustained [Ca²⁺]_i elevation seen in Fig. 4 might reflect decreased Ca²⁺ entry through SOCE channels, but could also conceivably result from increased Ca²⁺ buffering by intracellular or plasma membrane transporters or from a decreased driving force for Ca²⁺ influx (i.e. by membrane depolarization). Thus, we also tested the ability of nocodazole to inhibit the store-operated current I_CRAC in HEK 293 cells. Whole-cell currents were measured using a pipette solution containing inositol (1,4,5)-trisphosphate [Ins(1,4,5)P₃] to deplete intracellular Ca²⁺ stores and BAPTA to prevent Ca²⁺-dependent inactivation; when HEK 293 cells were patched in the presence of 10 mM extracellular Ca²⁺, a very small inwardly rectifying current on the order of –0.5 pA/pF developed (Fig. 5A). Cells pre-treated with 10 μM nocodazole for 20 minutes before patching exhibited a smaller Ca²⁺ current, although this decrease was not statistically significant (Fig. 5A). Thus, given the very small Ca²⁺ current in HEK 293 cells, it was difficult to ascertain whether nocodazole had an inhibitory effect. We therefore switched to a protocol using a divalent-cation-free extracellular solution, which allows for a more accurate assessment of I_CRAC currents in these cells (DeHaven et al., 2007). When cells are patched in the presence of extracellular Ca²⁺ and then switched to a divalent-free solution, a readily discernible Na⁺ current of approximately –3 pA/pF develops (Fig. 5B). When the divalent-free protocol was applied to cells pre-treated with nocodazole for 20 minutes, the peak Na⁺ current was approximately 50% less compared with that of untreated controls (Fig. 5B,C). Thus, as observed for SOCE, I_CRAC is inhibited in HEK 293 cells by the microtubule-depolarizing drug nocodazole.

The results of the experiments showing inhibition of SOCE by nocodazole are not directly comparable to the morphological findings because, in the latter case, the cells overexpressed EYFP-STIM1. Thus, we carried out experiments comparing the effects of nocodazole on Ca²⁺ entry in wild-type HEK 293 cells with those in which cells overexpressed EYFP-STIM1. For these experiments, SOCE was monitored by emptying intracellular Ca²⁺ stores with thapsigargin in the presence of nominal extracellular Ca²⁺, followed by re-addition of 1.8 mM extracellular Ca²⁺. Consistent with the results in Fig. 4, when HEK 293 cells were treated with 10 μM nocodazole for 20 minutes before store depletion with thapsigargin, Ca²⁺ entry was reduced by approximately 70% in comparison with that of cells treated with DMSO alone (Fig. 6A,C; P<0.01). Surprisingly, in HEK 293 cells overexpressing EYFP-STIM1, nocodazole had relatively little effect on SOCE in these cells (Fig. 6B,C). Furthermore, although the inhibition in EYFP-STIM1 cells was statistically significant compared with that of DMSO-treated controls at the P<0.05 level, the degree of inhibition was only 15%. Thus, overexpression of EYFP-STIM1 appears to significantly rescue the inhibition of SOCE caused by microtubule depolymerization, consistent with the conclusion that the primary cause of inhibition by nocodazole is related to STIM1 dysfunction.

We also carried out experiments whereby re-addition of Ca²⁺ was performed in the absence of store depletion with thapsigargin. Unexpectedly, nocodazole treatment alone for 20 minutes was sufficient to activate a significant Ca²⁺ entry upon Ca²⁺ re-addition in cells overexpressing EYFP-STIM1 in comparison with cells treated with DMSO alone (Fig. 7A,C). As shown in Fig. 7E, all the cells overexpressing EYFP-STIM1 and treated with nocodazole responded with Ca²⁺ entry, although the magnitude of this entry varied widely among the cells; by contrast, only very few of the DMSO-treated cells exhibited an observable Ca²⁺ entry (Fig. 7D). Little or no nocodazole-activated entry was seen in cells overexpressing unconjugated EYFP (Fig. 7B). The Ca²⁺ entry induced by nocodazole in EYFP-STIM1 cells was attributable to activation of physiological SOCE channels by two criteria. First, reduction of Orai1 expression by RNAi significantly reduced nocodazole-activated Ca²⁺ entry (Fig. 7A,C). Second, the increased entry in nocodazole-treated EYFP-STIM1 cells is associated with increased I_CRAC. EYFP-STIM1 cells, control or treated with nocodazole, were patched using a pipette solution in which Ca²⁺ was strongly buffered to low levels to passively deplete stores, and the extracellular solution was switched to a divalent-free solution at intervals of 1 minute (Fig. 7G,H). Because the magnitude of the observed increase in Ca²⁺ entry in nocodazole-treated cells was somewhat variable, we selected nocodazole-treated cells that showed visible EYFP-STIM1 puncta (see below). With this passive depletion protocol [i.e. no Ins(1,4,5)P₃ in the pipette], there is a delay before the onset of development of I_CRAC, permitting evaluation of the constitutive current following break-in. As shown in Fig. 7G,H, in EYFP-STIM1 cells treated with nocodazole (10 μM, 20 min), the Na⁺ current at the time of the first switch to a divalent-free solution was substantially greater than that seen in untreated controls (nocodazole-treated cells: 9/10 had initial Na⁺ currents >4pA/pF; control cells: 1/6 had similarly large currents). Thus, nocodazole-treated cells exhibited an I_CRAC current that was active before store depletion induced by the patch-clamp protocol.

The various effects of nocodazole on SOCE and I_CRAC led us to investigate the influence of microtubule depolymerization on the formation of EYFP-STIM1 puncta near the plasma membrane. Store depletion with thapsigargin in nocodazole-treated cells caused formation of EYFP-STIM1 puncta, as evaluated by confocal microscopy (Fig. 8A). To provide a more quantitative analysis of the formation of EYFP-STIM1 puncta, we performed similar experiments using total internal reflectance fluorescence microscopy (TIRFM). As shown in Fig. 8B, the 20 minute time course of nocodazole treatment caused a marked increase in EYFP-STIM1 puncta near the plasma membrane; formation of puncta was further increased upon store depletion with thapsigargin. DMSO treatment alone had no effect (Fig. 8C). Quantitative analysis of the TIRFM fluorescence over time indicates that the EYFP-STIM1 fluorescence near the plasma membrane began to increase soon after nocodazole application (Fig. 8D); at the culmination of the 20 minute nocodazole treatment (labeled `ii'), there was a statistically significant increase in the fluorescence intensity compared with that of cells treated with DMSO (Fig. 8E; P<0.01). Furthermore, the thapsigargin-induced increase in fluorescence intensity was significantly greater in the nocodazole-treated versus DMSO-treated cells (Fig. 8E; P<0.05). In fact, the nocodazole-induced and thapsigargin-induced increases in fluorescence intensity appeared to be additive, implying that the cause of the nocodazole-induced increase is independent of that of thapsigargin (i.e. store depletion).

These results indicate that disruption of microtubules by nocodazole can induce formation of STIM1 puncta and activate SOCE and I_CRAC without the use of Ca²⁺-store-depleting drugs. However, we noticed that the shape of the Ca²⁺ release response to thapsigargin was altered in nocodazole-treated cells (see, for example, Fig. 6), such that the rate of release appeared faster. This could indicate a subtle effect of nocodazole on the Ca²⁺ permeability of the ER. Thus, we examined the total intracellular ionomycin-releasable Ca²⁺ pool (see Bird et al., 1992; Bird and Putney, Jr, 2005) in HEK 293 cells and found that the ionomycin-induced 340/380 fluorescence peak was 3.59±0.08 in DMSO-treated EYFP-STIM1-expressing cells (n=85 cells, three coverslips) and 2.77±0.09 in nocodazole-treated cells (n=84 cells, three coverslips). Thus, the total ionomycin-releasable intracellular Ca²⁺ stores were reduced by 22.8% by nocodazole (P<0.01). Because of the additive effect of nocodazole and thapsigargin shown in Fig. 8, we thought it unlikely that this small depletion of Ca²⁺ stores could account for the activation of STIM1 redistribution and SOCE. However, we considered the possibility that this release might be required and that the microtubule-disrupting action of nocodazole might in some manner potentiate the response. To address this possibility, we took advantage of the fact that colchicine, like nocodazole, disrupts microtubules but does not appear to cause any significant depletion of Ca²⁺ stores (J.T.S. et al., unpublished). As shown in Fig. 9A, treatment of EYFP-STIM1-expressing HEK 293 cells with colchicine did not result in constitutive activation of Ca²⁺ entry. However, when colchicine-treated cells were stimulated with 1 μM cyclopiazonic acid (CPA), a reversible SERCA pump inhibitor that elicits only partial store depletion when used at low-micromolar concentrations, an augmentation of Ca²⁺ entry was observed compared with that of cells treated with CPA alone (Fig. 9A,B). Thus, in cells overexpressing STIM1, disruption of the microtubule cytoskeleton can actually augment Ca²⁺ entry, but apparently at least some Ca²⁺ store depletion is necessary.

We have presented several lines of evidence indicating that STIM1 localization and function in HEK 293 cells involves the microtubule cytoskeleton. First and foremost is the observation that overexpressed EYFP-STIM1 colocalizes with α-tubulin in HEK 293 cells, as has previously been shown for HeLa cells (Baba et al., 2006). Second, the fibrillar arrangement of EYFP-STIM1 is drastically altered in cells treated with the microtubule-depolymerizing agent nocodazole, resulting in a distribution of STIM1 similar to that of a general marker of the ER. Thus, STIM1 appears to be organized within the ER by the microtubule network. Third, SOCE and I_CRAC were affected in several ways by microtubule depolymerization. In cells that did not overexpress EYFP-STIM1, nocodazole inhibited SOCE and I_CRAC. This inhibition of SOCE was almost completely reversed by overexpression of EYFP-STIM1, implying that the inhibitory effect of SOCE by nocodazole is due to an impairment of STIM1 function. We considered that the disruption of microtubules can lead to a retraction of the ER away from the plasma membrane (Graier et al., 1998). For this reason, we limited pretreatment with the drugs to times wherein no visible retraction was observed by confocal microscopy. As a more rigorous test of this idea, we also examined the near-membrane localization of an ER-targeted GFP construct by TIRF microscopy. Treatment of HEK 293 cells expressing this fluorescent protein with nocodazole or colchicine for 20 minutes did not result in significant retraction of the ER, but instead the TIRF signal was slightly, although not significantly, increased (F/F0 for control, 0.94±0.04, n=4; nocodazole, 1.45±0.28, n=4; colchicine, 1.38±0.12, n=6).

We were surprised to find that nocodazole treatment alone was sufficient to activate SOCE and I_CRAC in cells overexpressing EYFP-STIM1. This increase in Ca²⁺ entry caused by nocodazole is probably initiated by modest Ca²⁺ store depletion by nocodazole, together with a significant potentiation of STIM1 movements and SOCE activation due to microtubule disaggregation. In support of this interpretation, colchicine, which does not cause any Ca²⁺ store depletion, did not activate entry on its own but substantially potentiated entry in response to submaximal store depletion. In aggregate, these findings suggest that the movement of STIM1 to near-membrane puncta in response to store depletion is not a microtubule-dependent process per se. Rather, the organizing effects of the microtubule cytoskeleton on STIM1 localization and movements is required for optimal signaling when STIM1 is limiting or expressed at physiological levels. The microtubule-dependent organization becomes unnecessary at very high levels of expression and even inhibits excessive access of STIM1 to signaling complexes, presumably those involving Orai channels.

Several previous studies have reported discordant effects of nocodazole treatment on SOCE and I_CRAC in a variety of cell types. Bakowski et al. (Bakowski et al., 2001) reported a lack of effect of 5 μM nocodazole treatment for 16-26 hours on I_CRAC in RBL-1 cells, the assumption being that such a treatment should cause retraction of the ER away from the plasma membrane and prevent conformational coupling with the I_CRAC channel. We also observed little or no effects of nocodazole on I_CRAC in RBL cells with the protocol used in the current study (data not shown). A lack of effect of nocodazole treatment (10 μM, 30 minutes) on SOCE has also been reported in NIH 3T3 cells (Ribeiro et al., 1997). However, a more recent study showed a significant inhibitory effect of nocodazole (0.3-3.0 μM, 15 minutes) and colchicine on SOCE in RBL-2H3 cells and bone-marrow-derived mast cells (Oka et al., 2005), although no current measurements were presented. Thus, it appears that any number of factors, including cell type and treatment protocol, might significantly influence the results of experiments intended to assess the effects of microtubule depolymerization on SOCE and/or I_CRAC. One intriguing possibility is that differences in STIM1 expression levels in different cell types could have an impact. In HEK 293 cells, endogenous STIM1 expression appears to be limiting for activation of SOCE as overexpression of STIM1 results in enhanced entry in this cell type (Roos et al., 2005; DeHaven et al., 2007). Furthermore, overexpression of EYFP-STIM1 significantly rescues nocodazole-inhibited SOCE, implying that microtubule depolymerization is far less deleterious under conditions in which STIM1 expression is not limiting. The most likely explanation for these results is that microtubule depolymerization results in mislocalization of endogenous STIM1, as we have shown by confocal imaging with overexpressed EYFP-STIM1 (unfortunately, a lack of antibodies against STIM1 suitable for immunofluorescence has prevented us from monitoring the distribution of endogenous STIM1). Such mislocalization probably prevents STIM1 from efficiently sensing depletion of Ca²⁺ stores and/or from efficiently coupling with the SOCE signaling machinery and possibly members of the Orai family of SOCE channels. When STIM1 is overexpressed, the excess protein permits near full activation even when the efficiency of the process is compromised.

Interestingly, we also observed a nocodazole-induced activation of SOCE and I_CRAC in cells expressing EYFP-STIM1, as well as an increase in EYFP-STIM1 localized near to the plasma membrane. Nocodazole causes minor depletion of Ca²⁺ stores, which is probably necessary for this effect. However, in the TIRFM experiments, the increase in EYFP-STIM1 near to the plasma membrane induced by nocodazole was additive with that induced by full store depletion with thapsigargin, indicating that nocodazole and store depletion also influence rearrangement of EYFP-STIM1 by independent mechanisms. In other words, if nocodazole caused formation of EYFP-STIM1 puncta only because of store depletion, then it would be expected that the extent of EYFP-STIM1 puncta formation in response to full store depletion with thapsigargin in nocodazole-treated cells should not be any greater than that seen in cells in which stores are depleted with thapsigargin alone. An alternative interpretation is that the localization of EYFP-STIM1 is less restricted when microtubules are depolymerized, such that, when limited discharge of Ca²⁺ stores mobilizes EYFP-STIM1, more EYFP-STIM1 is able to interact with and activate the SOCE machinery; this effect is not seen in wild-type cells to any appreciable degree because endogenous STIM1 expression is much less than in the cells overexpressing EYFP-STIM1. This is an interesting possibility as, if true, it implies that the quantitative relationship between STIM1 mobilization and activation of SOCE can be regulated by factors that influence its localization and mobility independently of the degree of depletion of Ca²⁺ stores.

In a recent report, Baba et al. (Baba et al., 2006) also documented a dependence of STIM1 localization on microtubules, wherein it was shown that overexpressed STIM1 exhibits a fibrillar organization in DT40 B-lymphocytes that is lost upon treatment with nocodazole. In that study, nocodazole-treated cells exhibited unimpaired SOCE; however, the experiments of Baba et al. were performed on STIM1-overexpressing cells and the effect of nocodazole on SOCE in wild-type cells was not reported. Interestingly, Baba et al. (Baba et al., 2006) also reported that a mutant STIM1 lacking the C-terminal, cytoplasmic serine/threonine-rich domain exhibited a localization that was not fibrillar but instead was very similar to the localization of the ER; this is reminiscent of effects we have reported of nocodazole on EYFP-STIM1 and ER localization. STIM1 lacking the serine/threonine-rich region did not rearrange into puncta upon Ca²⁺ store depletion and was unable to support SOCE. Thus, this serine/threonine-rich region might be involved in the microtubule association of STIM1 as well as in downstream signaling from STIM1 to Orai channels.

An effect of microtubule depolymerization on SOCE and I_CRAC involving mitochondria transport was reported by Quintana et al. (Quintana et al., 2006). In that study, it was shown that mitochondria are recruited towards the plasma membrane in response to initiation of Ca²⁺ entry and that these translocated mitochondria act as Ca²⁺ buffers that attenuate Ca²⁺-dependent inactivation of SOCE. However, this is unlikely to account for the inhibition that we observed with nocodazole for two reasons. First, in our study, nocodazole inhibited I_CRAC in electrophysiological experiments performed under conditions of high intracellular Ca²⁺ buffering, such that Ca²⁺-dependent inactivation is prevented and mitochondrial effects are not generally seen. Second, the nocodazole-induced inhibition of entry could be rescued by overexpression of EYFP-STIM1, a finding that is not consistent with a mitochondrial effect.

Exactly how STIM1 localization and function depend on microtubules is not clear at this time. The major conclusion from this study is that the microtubular cytoskeleton facilitates and perhaps organizes STIM1 movements, but the translocation and aggregation of STIM1 does not absolutely depend upon a microtubular mechanism. What drives this important translocation step in the signaling mechanism for store-operated Ca²⁺ channels, and how STIM1 acts to open the channels, will be subject to future investigations.

HEK 293 cells were obtained from the ATCC and were cultured and transfected as previously described (DeHaven et al., 2007). The EYFP-STIM1 plasmid was obtained from Tobias Meyer (Stanford University, CA), mCherry-α-tubulin was obtained from Roger Tsien (University of California, San Diego, CA), and ER-targeted GFP and CFP constructs were purchased from Clontech. The amounts of cDNA used in transfections were as follows: EYFP-STIM1, 0.5 μg; mCherry-α-tubulin, 0.05 μg; ER-CFP, 0.5 μg. The Orai1 siRNA used for knockdown of Orai1 expression was purchased from Invitrogen and had the sequence CCCUUCGGCCUGAUCUUUAUCGUCU; 100 nM was used per transfection.

Cells were grown on glass coverslips for 24-48 hours, and coverslips were mounted in Teflon chambers for imaging. Cells were maintained in Hepes-buffered saline solution (HBSS: 120 mM NaCl, 5.4 mM KCl, 1.8 mM CaCl₂, 0.8 mM MgCl₂, 11 mM glucose and 20 mM Hepes, pH 7.4) at room temperature. Confocal imaging was performed using a Zeiss LSM 510 laser scanning system, and either a 40× water-immersion (NA 1.2) or a 63× oil-immersion (NA 1.4) objective was used. All confocal images were collected with the pinhole set at 1 Airy unit. For EYFP-STIM1, 488 nm or 514 nm illumination was provided by an Argon laser and emission was selected with a 530-600 nm bandpass filter. The excitation for mCherry-tubulin was 543 nm from a HeNe laser and emission was selected with a 560 nm low-pass filter. The excitation for ER-CFP was 458 nm from an Argon laser, and emission was selected with a 470-510 bandpass filter. When cells expressing multiple probes were imaged, lack of bleed-through between channels was verified by imaging cells that expressed each of the probes individually. TIRFM was performed essentially as described previously (Smyth et al., 2005).

Cells for fixation were grown in LabTekII chamber slides (Nalge Nunc) for 24 to 48 hours. At the time of fixation, slides were rinsed in phosphate-buffered saline (PBS: 137 mM NaCl, 2.68 mM KCl, 1.47 mM KH₂PO₄, 14.9 mM Na₂HPO₄) and were extracted in Karesnti's buffer (80 mM Pipes, 1.0 mM MgSO₄, 5.0 mM EGTA, 0.5% Triton X-100) for 8 seconds at room temperature. Cells were then fixed in 100% methanol on ice for 10 minutes and rehydrated in PBS containing 0.1% Tween-20 and 3.0 mM NaN₃ (PBS-Tw-Az). Cells were incubated in primary antibodies diluted in PBS-Tw-Az containing 1% bovine serum albumin (BSA) overnight at room temperature and with secondary antibodies in PBS-Tw-Az plus BSA for 45 minutes at room temperature. Cells were washed three times in PBS-Tw-Az after each antibody incubation, and, after the final wash, coverslips were mounted onto the slides using Vectashield hard-set mounting medium (Vector Laboratories). For detection of α-tubulin, the primary antibody used was DM1A (AbCam) at a dilution of 1:100, and the secondary antibody was Alexa Fluor 633 goat anti-mouse IgG₁ (Invitrogen) at a dilution of 1:1000. For detection of EYFP-STIM1, the primary antibody was a polyclonal against GFP (AbCam) used at a dilution of 1:2000, and the secondary antibody was Alexa Fluor 488 goat anti-rabbit IgG (Invitrogen) at a dilution of 1:1000. Slides were imaged by confocal microscopy as described above. Colocalization analysis was performed using the Zeiss LSM 510 software. For each image analyzed, the percentage of pixels that exhibited a signal above a predetermined threshold for each channel was recorded.

Intracellular Ca²⁺ measurements were performed as previously described (Mercer et al., 2006). Briefly, cells were loaded with 1 μM fura-5F/AM (Invitrogen) for 25 minutes at 37°C. Fura-5F fluorescence was measured when cells were excited consecutively at 340 nm and 380 nm, and relative Ca²⁺ concentrations are reported as the ratio of fluorescence emission at the two excitation wavelengths. Cells transfected with EYFP or EYFP-STIM1 were chosen based on their fluorescence when excited at 477 nm. Typically, 20-30 cells were measured on a single coverslip per experiment.

Whole-cell currents were investigated at room temperature (20-25°C) in HEK 293 cells using the patch-clamp technique in the whole-cell configuration. The standard HEPES buffered saline solution contained: 145 mM NaCl, 3 mM KCl, 10 mM CsCl, 1.2 mM MgCl₂, 10.0 mM CaCl₂, 10 mM glucose and 10 mM HEPES (pH to 7.4 with NaOH). The standard divalent-free (DVF) solution contained: 155 mM Na-methanesulfonate, 10 mM HEDTA, 1 mM EDTA and 10 mM HEPES (pH 7.4). Fire-polished pipettes fabricated from borosilicate glass capillaries (WPI, Sarasota, FL) with 3-5 MΩ resistance were filled with: 145 mM Cs-methanesulfonate, 20 mM BAPTA, 10 mM HEPES and 8 mM MgCl₂ (pH to 7.2 with CsOH). In the indicated experiments, the pipette also contained 25 μM Ins(1,4,5)P₃ (hexasodium salt, Sigma) to actively deplete intracellular Ca²⁺ pools. Voltage ramps (–100 mV to +100 mV) of 250 ms were recorded every two seconds immediately after gaining access to the cell from a holding potential of 0 mV, and the currents were normalized based on cell capacitance. Leak currents were subtracted by taking an initial ramp current before I_CRAC developed and subtracting this from all subsequent ramp currents. Access resistance was typically between 5-10 MΩ. The currents were acquired with pCLAMP-10 (Axon Instruments) and analyzed with Clampfit (Axon Instruments) and Origin 6 (Microcal) software. All solutions were applied by means of a gravity based multi-barrel local perfusion system with an extremely low dead volume common delivery port (Perfusion Pencil, Automate Scientific).

Cells were treated with nocodazole or colchicine (both from Calbiochem) in HBSS at room temperature. Because nocodazole treatment caused activation of Ca²⁺ entry (see Results), all nocodazole incubations were performed in nominally Ca²⁺-free HBSS unless otherwise noted. Ionomycin and cyclopiazonic acid were obtained from Calbiochem, and thapsigargin was obtained from Alexis Biochemicals.

We thank Patricia Wadsworth and Nasser Rusan for helpful discussions and advice throughout the completion of this study. We also thank Jeff Reece and Jeff Tucker for technical assistance with TIRF and confocal microscopy, and Rebecca Boyles for plasmid preparations. Steven Shears and David Miller read the manuscript and provided helpful comments. This research was supported by the Intramural Program, National Institute of Environmental Health Sciences, NIH.