Mutant IP3 receptors attenuate store-operated Ca2+ entry by destabilizing STIM–Orai interactions in Drosophila neurons

ABSTRACT Store-operated Ca2+ entry (SOCE) occurs when loss of Ca2+ from the endoplasmic reticulum (ER) stimulates the Ca2+ sensor, STIM, to cluster and activate the plasma membrane Ca2+ channel Orai (encoded by Olf186-F in flies). Inositol 1,4,5-trisphosphate receptors (IP3Rs, which are encoded by a single gene in flies) are assumed to regulate SOCE solely by mediating ER Ca2+ release. We show that in Drosophila neurons, mutant IP3R attenuates SOCE evoked by depleting Ca2+ stores with thapsigargin. In normal neurons, store depletion caused STIM and the IP3R to accumulate near the plasma membrane, association of STIM with Orai, clustering of STIM and Orai at ER–plasma-membrane junctions and activation of SOCE. These responses were attenuated in neurons with mutant IP3Rs and were rescued by overexpression of STIM with Orai. We conclude that, after depletion of Ca2+ stores in Drosophila, translocation of the IP3R to ER–plasma-membrane junctions facilitates the coupling of STIM to Orai that leads to activation of SOCE. Summary: In Drosophila neurons, mutant IP3 receptors disrupt store-operated Ca2+ entry by destabilizing interaction of STIM with the Ca2+ channel, Orai. The interactions could coordinate store emptying with Ca2+ entry.


INTRODUCTION
Receptors that stimulate phospholipase C and, hence, formation of inositol 1,4,5-trisphosphate (IP 3 ) typically evoke both release of Ca 2+ from the endoplasmic reticulum (ER) through IP 3 receptors (IP 3 Rs) and Ca 2+ entry across the plasma membrane. The latter is usually mediated by store-operated Ca 2+ entry (SOCE), an almost ubiquitously present pathway through which empty Ca 2+ stores stimulate Ca 2+ entry across the plasma membrane (Putney and Tomita, 2012). The core molecular components of SOCE are stromal interaction molecule (STIM) and Orai (the gene for which is also known as Olf186-F in flies) (Hogan, 2015;Lewis, 2012). Orai forms a hexameric Ca 2+ -selective ion channel in the plasma membrane (Hou et al., 2012) and STIM is the Ca 2+ sensor anchored in ER membranes (Carrasco and Meyer, 2011). Ca 2+ dissociates from the luminal EF-hand of STIM when Ca 2+ is lost from the ER. This causes STIM to oligomerize, unmasking residues that interact with Orai, and allowing STIM to accumulate at ER-plasma-membrane junctions, where the gap between membranes is narrow enough to allow the cytosolic CAD region of STIM to bind directly to Orai (Hogan, 2015). That interaction traps STIM and Orai clusters within ER-plasma-membrane junctions and it stimulates opening of the Orai channel (Wu et al., 2014). Additional proteins also regulate SOCE, often by modulating interactions between STIM and Orai (Srikanth and Gwack, 2012) or by facilitating their interactions by stabilizing ER-plasma-membrane junctions (Cao et al., 2015) or the organization of phosphatidylinositol 4,5-bisphosphate (PIP 2 )enriched membrane domains (Sharma et al., 2013). SOCE can be activated by thapsigargin, which depletes Ca 2+ stores by inhibiting the ER Ca 2+ pump, but for SOCE evoked by physiological stimuli, the Ca 2+ stores are depleted by activation of IP 3 Rs. In the present study, we use genetic manipulations in Drosophila neurons to ask whether IP 3 Rs regulate SOCE solely by mediating Ca 2+ release from the ER or whether they can also play additional roles downstream of store depletion. Drosophila is well suited to this analysis because single genes encode IP 3 R, STIM and Orai, whereas vertebrates have several genes for each of these proteins. Our results demonstrate that in Drosophila, its IP 3 R contributes to assembly of the STIM-Orai complex. Comparison of results from Drosophila and vertebrates suggests that the STIM-Orai complex might assemble in plasmamembrane-ER regions equipped to allow local depletion of Ca 2+ stores.

Mutant IP 3 Rs attenuate SOCE in Drosophila
SOCE evoked by depleting ER Ca 2+ stores with thapsigargin in Drosophila neurons was abolished by RNA interference (RNAi) treatment for STIM or Orai (see Fig. 1D; Venkiteswaran and Hasan, 2009). This is consistent with evidence that STIM and Orai are core components of SOCE. Subsequent experiments examine the role of the IP 3 R, which is encoded by a single gene (itpr) in Drosophila, in regulating SOCE. To characterize SOCE in Drosophila neurons with mutant itpr, we examined five hetero-allelic combinations of a 15-residue C-terminal deletion and three point mutations located in different parts of the IP 3 R (Banerjee et al., 2004;Joshi et al., 2004) (Fig. 1A). We used these combinations because the adults with these mutations are viable with distinct flight phenotypes, whereas homozygotes and other hetero-allelic combinations are lethal (Joshi et al., 2004). We also used neurons heterozygous for each individual mutation. The peak Ca 2+ signals evoked by addition of thapsigargin in Ca 2+ -free medium and the response to restoration of extracellular Ca 2+ (SOCE) were measured in primary neuronal cultures for each genotype (Fig. 1B-D). Our use of fluo 4 fluorescence changes (ΔF/F 0 , see Materials and Methods) to report cytosolic free Ca 2+ concentration ([Ca 2+ ] c ) is vindicated by evidence that [Ca 2+ ] c in unstimulated cells was unaffected by mutant IP 3 R (Fig. S1A) and the peak fluorescence changes evoked by SOCE in wild-type neurons were only 32±14% (mean±s.d., n=9) of those evoked by saturating the indicator with Ca 2+ . SOCE was significantly reduced in all five itpr mutant combinations ( Fig. 1D; Table S1). However, the resting [Ca 2+ ] c and thapsigargin-evoked release of Ca 2+ from intracellular stores were unaffected, confirming that mutant IP 3 R selectively perturbed SOCE ( Fig. 1C; Fig. S1A). The reduced SOCE in itpr mutant neurons was not, therefore, restricted to a single combination of mutant alleles. Mutant combinations in the ligand-binding domain (S224F), modulatory domain (G1891S and G2117E) and Cterminus (Δ2814-2828) of the IP 3 R all inhibited SOCE. SOCE was not significantly affected in neurons heterozygous for the individual mutants ( Fig. 1D; Table S1). In adult flies, these heterozygous itpr mutants also have no significant effect on viability or flight phenotype (Banerjee et al., 2004;Joshi et al., 2004). The results establish that attenuated SOCE in Drosophila neurons is due to a perturbation of IP 3 R function in the recessive heteroallelic mutant combinations. This conclusion is supported by evidence that RNAi-mediated knockdown of IP 3 R also inhibited thapsigargin-evoked SOCE (Fig. 1D). It might seem surprising that so many combinations of four different itpr mutants should inhibit SOCE. However, selection of the original mutant combinations was based on flight phenotypes (Banerjee et al., 2004), and restoring SOCE can rescue these flight defects (Agrawal et al., 2010). The original selection might therefore have preferentially identified itpr mutants that attenuate SOCE. Immunoblots established that expression of IP 3 R, STIM and Orai were similar in the larval central nervous system from wild-type and itpr mutant flies ( Fig. S1B,C).
Similar functional consequences of mutant itpr were observed in cultured haemocytes from Drosophila larvae, where itpr mutants attenuated thapsigargin-evoked SOCE without affecting basal [Ca 2+ ] c or Ca 2+ release from intracellular stores (Fig. S2).
The results so far establish that loss of IP 3 R or mutations within IP 3 R attenuate SOCE without affecting the Ca 2+ content of the intracellular stores. The effects are not due to loss of STIM or Orai.

Over-expressed STIM and Orai restores SOCE in neurons with mutant IP 3 Rs
We used the mutant itpr combination itpr S224F/G1891S to examine the effects of overexpressing STIM and Orai on SOCE in cultured neurons. We chose this combination because it has been the most extensively studied of the heteroallelic mutant itpr combinations (Agrawal et al., 2010;Venkiteswaran and Hasan, 2009). The response to thapsigargin in Ca 2+ -free medium was unaffected by overexpression of STIM and Orai (Fig. 1C,E). Rates of recovery from these [Ca 2+ ] c increases were also unaffected (Fig. 1F). These results demonstrate that the ER Ca 2+ content, passive leak of Ca 2+ from the ER, and rates of Ca 2+ extrusion from the cytosol were similar in neurons with mutant or wild-type IP 3 R, and unaffected by overexpression of STIM and Orai. However, SOCE in neurons expressing mutant IP 3 R was restored by overexpression of STIM (C,D) Summary results for peak responses evoked by thapsigargin (Ca 2+ release) and Ca 2+ restoration (SOCE) for neurons with the indicated genotypes and for WT neurons treated with the indicated siRNA. The box represents the 25-75th percentiles, and the median is indicated. The whiskers show the 10-90th percentiles. Outliers are represented by dots. Results are from >100 cells from at least five independent experiments.*P<0.05, **P<0.01 (Kruskal-Wallis test, followed by Wilcoxon signed-rank post-hoc test). (E) Responses of neurons from WT or itpr mutants (S224F/G1891S) alone and after overexpression of STIM and Orai. Results (mean±s.d., from >100 cells from at least five independent experiments) show Ca 2+ release evoked by thapsigargin (10 µM) in Ca 2+ -free HBM, and SOCE evoked by subsequent addition of extracellular Ca 2+ (2 mM). (F) Rates of Ca 2+ leak and recovery from the thapsigargin-evoked Ca 2+ release, calculated from E. The colour key applies to panels E and F. STIM + , overexpression of STIM; Orai + , overexpression of Orai.
with Orai (Fig. 1D,E). This is consistent with behavioural analyses (Fig. S3A,B) (Agrawal et al., 2010). A similar restoration of SOCE upon overexpression of STIM has been reported in neurons in which IP 3 R expression was reduced by small interfering RNA (siRNA) (Deb et al., 2016). Hence, even though mutant IP 3 Rs do not affect expression of STIM or Orai, the attenuated SOCE can be compensated for by overexpressing STIM and Orai (Fig. 1D). The results so far suggest that the IP 3 R regulates SOCE downstream of ER Ca 2+ release, perhaps by influencing interactions between STIM and Orai.
Mutant IP 3 Rs attenuate the association of STIM with Orai after store depletion We tested whether IP 3 R mutations affect interactions between STIM and Orai using an ectopically expressed dOrai-CFP + transgene with a pan-neuronal driver (Elav C155 ). This allowed immunoprecipitation of Orai with an anti-GFP antibody. Expression of Orai-CFP did not restore SOCE in itpr mutant neurons (Fig. S3C,D). Treatment with thapsigargin enhanced the pulldown of STIM with anti-GFP antibody from lysates of wild-type brain, consistent with enhanced interaction between STIM and Orai after store depletion. However, the pulldown of STIM from thapsigargintreated brains with mutant IP 3 R was much reduced ( Fig. 2A,B). In the reciprocal immunoprecipitation using wild-type brain expressing STIM-YFP, thapsigargin increased the pulldown of Orai with the anti-GFP antibody (Fig. 2C,D). There was no detectable IP 3 R in this immunoprecipitate (data not shown), suggesting that any interaction between IP 3 R and STIM or Orai, whether direct or through other proteins, might be too weak to survive immunoprecipitation. It was impracticable to assess the effects of mutant IP 3 R in these immunoprecipitation experiments because STIM-YFP rescued the mutant IP 3 R phenotypes ( Fig. 1D) (Agrawal et al., 2010). These results suggest that wild-type IP 3 R stabilizes interactions between STIM and Orai after depletion of Ca 2+ stores.
To avoid reversal of attenuated SOCE in neurons with mutant IP 3 Rs after overexpression of STIM (Fig. 1D), we used immunostaining of fixed neurons to examine the effects of store depletion on the distribution of endogenous STIM, Orai and IP 3 R. We quantified the near-plasma-membrane distribution of STIM and IP 3 R by measuring either peripheral fluorescence in confocal sections across a mid-plane of the cell (Fig. 3A,B; Movies 1-4) or total fluorescence within a plane that included mostly plasma membrane (Fig. 3C,D) (see Materials and Methods). In wild-type neurons, thapsigargin significantly increased the amount of STIM detected near the plasma membrane. This redistribution of STIM was attenuated in neurons with mutant IP 3 R (itpr S224F/G1891S ) (Fig. 3C,D). Store depletion also increased the intensity of IP 3 R immunostaining near the plasma membrane ( Fig. 3A-D). There was no significant redistribution of IP 3 R in thapsigargin-treated neurons expressing mutant IP 3 R (Fig. 3A,B). Thapsigargin stimulated formation of STIM puncta in neurons expressing STIM-YFP (Fig. 3E), although the puncta were not detected with endogenous STIM. This is consistent with the effects of store depletion in mammalian cells, where STIM puncta are typically observed after overexpression of tagged STIM. The formation of STIM-YFP puncta after store depletion was significantly attenuated in Drosophila neurons with mutant IP 3 Rs (itpr S224F/G1891S ); and siRNA for the IP 3 R appeared to have a similar effect (Fig. 3E,F). The translocation of STIM and wild-type IP 3 R towards the plasma membrane after store depletion was not due to a general reorganization of the ER because co-staining of neurons for STIM and another ER protein (GFP-tagged protein disulphide isomerase, PDI-GFP) revealed that only STIM redistributed after thapsigargin treatment (Fig. S4). These results demonstrate that IP 3 R and STIM accumulate in peripheral ER near the plasma membrane after store depletion, and loss of IP 3 R or mutations within it inhibits the translocation of STIM.
Mutant IP 3 Rs attenuate formation of Orai puncta at the plasma membrane Using an antibody to endogenous Orai, we observed that depletion of Ca 2+ stores with thapsigargin stimulated formation of Orai puncta in neurons expressing wild-type IP 3 R, but not in neurons expressing mutant IP 3 Rs (itpr S224F/G1891S ) (Fig. 4A). Furthermore, the sparse Orai puncta that did form in neurons with mutant IP 3 Rs were both smaller and less intensely stained than in neurons with wild-type IP 3 Rs (Fig. 4B-D). Overexpression of STIM had no effect on the formation of Orai puncta in neurons with either genotype, although it partially restored SOCE (Fig. 1D). However, after overexpression of both STIM and Orai, store depletion stimulated formation of Orai puncta that were similar in neurons with wild-type or mutant IP 3 Rs ( Fig. 4B-D). The ability of overexpressed STIM and Orai to rescue formation of Orai puncta in neurons with mutant IP 3 Rs coincides with a similar rescue of SOCE in mutant neurons (Fig. 1D) and of flight in flies with mutant IP 3 R (Fig. S3A,B). These results suggest that overexpression of STIM with Orai can override the requirement of IP 3 R for formation of Orai puncta or SOCE after store depletion. However, when STIM and Orai are expressed at native levels in Drosophila, the interaction between them, the formation of Orai puncta and the activation of SOCE are enhanced by IP 3 R.

DISCUSSION
Before STIM was identified as the Ca 2+ sensor that regulates SOCE, IP 3 Rs were speculated to adopt this role (Irvine, 1990). However, thapsigargin evokes SOCE in avian DT40 cells lacking IP 3 Rs (Sugawara et al., 1997) (Fig. 5A) and SOCE can be functionally reconstituted with Orai and STIM (Zhou et al., 2010). IP 3 Rs are not, therefore, essential for empty Ca 2+ stores to activate SOCE. Our results, showing that SOCE is attenuated in Drosophila neurons with mutant IP 3 Rs (Fig. 1), suggest that IP 3 Rs can modulate SOCE. Depletion of intracellular Ca 2+ stores caused STIM and mutant IP 3 Rs to accumulate near the plasma membrane ( Fig. 3A-F), STIM and Orai to associate (Fig. 2), formation of STIM and Orai puncta (Figs 3 and 4), and activation of SOCE (Fig. 1). These responses were attenuated in neurons with mutant IP 3 Rs. The effects of mutant IP 3 Rs were not due to a dominant-negative property of the mutants because SOCE was also attenuated when IP 3 Rs expression was reduced by siRNA ( Fig. 1D) (Agrawal et al., 2010) and the mutant IP 3 Rs reduced SOCE only when both alleles were mutated (Fig. 1D). We suggest that after store-depletion, both STIM and IP 3 R translocate to ER-plasma-membrane junctions, where IP 3 R might stabilize the interaction of STIM with Orai, and thereby promote SOCE (Fig. 5B).
In some mammalian cells, IP 3 Rs have been shown to colocalize with Orai1 (Lur et al., 2011) and to interact with STIM1, Orai1 and transient receptor potential canonical channels (TRPCs) (Hong et al., 2011), but there is no functional evidence that IP 3 Rs directly contribute to SOCE mediated by Orai. Block of SOCE by an antagonist of IP 3 Rs, 2-aminoethoxydiphenyl borate (2-APB), was originally suggested to reflect IP 3 R coupling to a SOCE channel (Ma et al., 2000), but it is now attributed to direct inhibition of STIM and Orai by 2-APB. However, most analyses of SOCE use thapsigargin to completely deplete intracellular Ca 2+ stores, and overexpressed proteins to track movements of Orai and STIM. These exaggerated conditions successfully identify key features of SOCE, but they might override more subtle modulatory influences, including, for example, the contribution of PIP 2 to recruitment of STIM1 to ER-plasma-membrane junctions (Hogan, 2015;Park et al., 2009). We used DT40 cells with and without IP 3 R1 (encoded by itpr1) and examined SOCE after graded depletion of intracellular Ca 2+ stores to assess whether partially depleted stores might more effectively activate SOCE in the presence of IP 3 R. However, the relationship between store depletion and SOCE was unaffected by expression of IP 3 R1 (Fig. 5A). Hence, there is no compelling evidence to suggest that the contribution of IP 3 R to SOCE in Drosophila is a feature shared with vertebrates.
Inhibition of Orai clustering in Drosophila neurons with mutant IP 3 Rs is reminiscent of the effects of septin depletion in mammalian cells. Septin 4 concentrates PIP 2 around Orai1 and facilitates recruitment of STIM1 through its polybasic cytoplasmic tail (Sharma et al., 2013). Assembly of STIM-Orai complexes at PIP 2enriched plasma membrane domains concentrates the complexes at regions best equipped to sustain production of the IP 3 that evokes Ca 2+ release from stores. Such colocalization of Ca 2+ release and SOCE might be important because activation of SOCE by physiological stimuli probably requires substantial local depletion of intracellular Ca 2+ stores (Bird et al., 2009;Luik et al., 2008). However, Drosophila STIM lacks the polybasic tail through which mammalian STIM1 binds to PIP 2 (Huang et al., 2006). Association of Drosophila STIM with IP 3 R, which might itself bind to PIP 2 (Lupu et al., 1998), could serve a function analogous to PIP 2mediated targeting of STIM1 in mammals. Recent evidence demonstrating a link between Septin 7, IP 3 R and SOCE (Deb et al., 2016) suggests that IP 3 R might influence STIM-Orai interactions within a larger macromolecular complex. We speculate that interaction of mammalian STIM1 with PIP 2 might ensure that intracellular stores locally depleted of Ca 2+ by IP 3 are effectively localized to STIM-Orai complexes (Fig. 5B). Translocation of both IP 3 R and STIM to ER-plasma-membrane junctions after store depletion, where IP 3 R facilitates the interaction of STIM with Orai, might fulfil a similar role in Drosophila (Fig. 5B).

Drosophila strains
Single-point mutants of the itpr gene were characterized as described previously (Joshi et al., 2004;Srikanth et al., 2004). UAS transgenic strains were generated by injecting Drosophila embryos with a pUAST construct. All fly strains, including Elav C155 GAL4 ( pan neuronal, Bloomington Stock Center, Indiana University, Bloomington, IN), and RNAi lines for itpr (no. 1063, National Institute of Genetics, Japan), STIM (no. 47073) and Orai (no. 12221) were procured from the Vienna Drosophila Resource Centre, Austria. The Canton-S strain was used as the wild-type control.

Measurements of [Ca 2+ ] c in primary cultures of Drosophila neurons
Materials, unless stated otherwise, were from ThermoFisher Scientific (Waltham, MA). Methods for primary cultures were adapted from Wu et al. (1983). The brain and ventral ganglia from Drosophila third-instar larvae were dissociated by incubation for 20 min at 25°C in Schneider's medium containing collagenase (0.75 µg/µl) and dispase (0.4 µg/µl, Roche, Burgess Hill, UK). After centrifugation (600 g for 5 min), cells were plated onto poly-L-lysine-coated coverslips in HEPES-buffered medium [HBM, in mM: HEPES (30), NaCl (150), KCl (5), MgCl 2 (1), CaCl 2 (1), sucrose (35), pH 7.2] or (for most experiments) Dulbecco's modified Eagle's medium (DMEM) with F12 and Glutamax-I, NaHCO 3 and sodium pyruvate, and supplemented with 20 mM HEPES ( pH 7.2) and 10% fetal bovine serum. This enriched medium substantially reduced the heterogeneity of the Ca 2+ signals between neurons. All culture media contained 50 units/ml penicillin, 50 µg/ml streptomycin and 10 µg/ml amphotericin B. Cells were incubated at 25°C in humidified air with 5% CO 2 . After 14-16 h, cells were loaded with fluo 4 by incubation at 25°C for 30 min with fluo 4-AM (2.5 µM) and Pluronic F-127 (0.02%) in HBM. Medium was then replaced with HBM, and after a further 10-30 min, with Ca 2+ -free HBM containing 0.5 mM EGTA. Cells were immediately imaged at 15-s intervals with excitation at 488 nm and emission at 520 nm using a Nikon TE2000 microscope with a 60×1.4 NA objective, Evolution QEi CCD camera and QED imaging software (Media Cybernetics, Rockville, MD). Background fluorescence (measured from an area without cells) was subtracted from all measurements before calculation of ΔF/F 0 , In parallel wells, the peak increase in [Ca 2+ ] c evoked by ionomycin (1 µM, to determine the Ca 2+ content of the intracellular stores) or restoration of extracellular Ca 2+ (1.5 mM, to determine SOCE) were measured. Results (means±s.e.m., n=3, with three replicates in each) show the relationship between store depletion and SOCE for the two cell lines. (B) Substantial loss of Ca 2+ from the ER causes STIM to cluster and assemble with Orai at ER-plasma-membrane junctions. The polybasic cytoplasmic tail of vertebrate STIM1 binds to PIP 2 within the plasma membrane and contributes to its targeting to junctions. Drosophila STIM lacks a PIP 2 -binding motif, but, after store-depletion, Drosophila STIM moves to ERplasma-membrane junctions, and so too does IP 3 R where it might bind to PIP 2 . Physiological stimuli, through IP 3 , probably trigger the large decrease in luminal [Ca 2+ ] needed to activate STIM1 in only a subset of the ER. Targeting of vertebrate STIM1 to plasma membrane enriched in the PIP 2 , from which IP 3 is synthesized, ensures that the machinery needed to locally deplete Ca 2+ stores remains closely associated with essential components of the SOCE pathway. We speculate that in Drosophila, association of IP 3 R with STIM and Orai at ER-plasma-membrane junctions might fulfil a similar role.
where F 0 is the initial fluorescence and ΔF is the difference between basal and peak fluorescence.

Image analysis
Confocal images were deconvoluted using Huygens 4.5 software (SVI, The Netherlands) as described previously (Deb et al., 2016). To quantify the peripheral fluorescence of immunostained IP 3 R in each neuron, most of which have a near-circular profile (Fig. 3A), an automated algorithm (Matlab) was used to identify the confocal section with the maximal perimeter. Within this section, which we describe as the 'mid-section', the centre of the cell was identified and an average radius calculated (r). Fluorescence intensities were then calculated for the central circular region (with r/2) and for the remaining peripheral annulus. The ratio (peripheral fluorescence to total fluorescence) was then used to report the redistribution of IP 3 R.
To quantify near-plasma-membrane immunostaining of STIM and IP 3 R, we used Orai immunostaining to manually identify the confocal section within which most plasma membrane apposed the coverslip (Fig. 3C). The fluorescence intensity within this optical section relative to that from the entire cell was used to report the near-plasma-membrane distribution.
To quantify the distribution of STIM-YFP (Fig. 3G,H) and Orai (Fig. 4) puncta, every confocal section (∼15 sections/cell) was analysed (Deb et al., 2016). Puncta were identified automatically (Matlab) as fluorescence spots that exceeded average cellular fluorescence by at least 1.9× the s.d. and occupied a square with sides of 1-12 pixels (1 pixel=103 nm×103 nm) with a circularity of 0-0.3. Analysis of sequential sections within the z-stack allowed non-redundant counting of puncta, from which the total number of puncta/cell was determined. The section in which a punctum had the brightest intensity was used for analysis, and then normalized to the mean intensity of Orai for the cell.

Statistical analysis
Most data were analysed using non-parametric methods (Kruskal-Wallis test for variance followed by Wilcoxon signed-rank post-hoc tests). These data are presented as box and whisker plots showing medians, 25-75th percentiles (boxes), 10-90th percentiles (whiskers) and points for values beyond the 10th and 90th percentiles. Student's t-tests were used for statistical analysis of western blots (Fig. 2B) and Ca 2+ signals in DT40 cells (Fig. 5A).