LETTERS STIM1 carboxyl-terminus activates native SOC, Icrac and TRPC1 channels Guo N. Huang1,2,5, Weizhong Zeng3,5, Joo Young Kim3, Joseph P. Yuan3, Linhuang Han1, Shmuel Muallem3,6 and Paul F. Worley1,4,6 Receptor-evoked Ca2+ signalling involves Ca2+ release from the mediation of diverse cellular functions1,2. One form of SOC is the 2+ the endoplasmic reticulum, followed by Ca influx across the Icrac channel. The molecular identity of SOC and Icrac channels are not plasma membrane1. Ca2+ influx is essential for many cellular known, although several studies implicate members of the canonical functions, from secretion to transcription, and is mediated by transient receptor potential (TRPC) subfamily of TRP channels in 2+ 2+ 1 Ca -release activated Ca (Icrac) channels and store-operated SOC-channel activity . Recently, STIM1 was identified in screens for 2 calcium entry (SOC) channels . Although the molecular identity molecules that are essential for the activation of SOC and Icrac chan- 5,6 and regulation of Icrac and SOC channels have not been precisely nels . STIM1 possesses a signal sequence and a single transmembrane determined1, notable recent findings are the identification domain, indicating a topology that places an EF-hand domain either of STIM1, which has been indicated to regulate SOC and Icrac within the lumen of the endoplasmic reticulum or exposed on the channels by functioning as an endoplasmic reticulum Ca2+ plasma membrane5,6,9 (Fig. 1a). Mutation of the EF hand results in con- sensor3–6, and ORAI1 (ref. 7) or CRACM1 (ref. 8) — both of stitutively active STIM1 (refs 3, 4, 6), and increases its localization near 4,6 which may function as Icrac channels or as an Icrac subunit. or at the plasma membrane . To examine the molecular basis of STIM1 How STIM1 activates the Ca2+ influx channels and whether function, we established a screening assay based on the translocation STIM1 contributes to the channel pore remains unknown. of a transcription factor, nuclear factor of activated T cells (NFAT), 2+ Here, we identify the structural features that are essential for to the nucleus in response to sustained elevation of cytoplasmic Ca STIM1-dependent activation of SOC and Icrac channels, and (ref. 10). As expected, NFAT–GFP localizes to the cytoplasm in rest- demonstrate that they are identical to those involved in the ing HEK cells (Fig. 1b and see Supplementary Information, Fig. S1). binding and activation of TRPC1. Notably, the cytosolic carboxyl When Ca2+ entry is increased, consequent to depletion of the endoplas- 2+ terminus of STIM1 is sufficient to activate SOC, Icrac and mic reticulum Ca store by either ionomycin or a sarco/endoplasmic TRPC1 channels even when native STIM1 is depleted by small reticulum Ca2+ ATPase (SERCA) Ca2+ pump blocker (thapsigargin), interfering RNA. Activity of STIM1 requires an ERM domain, NFAT–GFP translocates into the nucleus. Nuclear localization is cal- which mediates the selective binding of STIM1 to TRPC1, 2 and cineurin-dependent as it is blocked by FK506 (Fig. 1b). When wild-type 4, but not to TRPC3, 6 or 7, and a cationic lysine-rich region, STIM1 is cotransfected with NFAT–GFP into HEK cells, NFAT–GFP which is essential for gating of TRPC1. Deletion of either region remains in the cytoplasm (Fig. 1b). In contrast, the constitutively active in the constitutively active STIM1D76A yields dominant-negative mutant STIM1D76A (ref. 6) induces the accumulation of NFAT–GFP in mutants that block native SOC channels, expressed TRPC1 the nucleus, and this is blocked by FK506 or by a SOC-channel blocker in HEK293 cells and Icrac in Jurkat cells. These observations SKF96365 (Fig. 1b). implicate STIM1 as a key regulator of activity rather than a Using the NFAT nuclear-localization assay, we examined the struc- channel component, and reveal similar regulation of SOC, Icrac ture–function properties of STIM1. The cytosolic carboxyl terminus and TRPC channel activation by STIM1. of STIM1 (STIM1CT), which lacks a transmembrane domain, induces NFAT localization to the nucleus in the absence of Ca2+ store deple- Receptor and SOC channels are essential for the maintenance tion (Fig. 1c). The STIM1CT contains several discrete regions: a con- of Ca2+ stores within the endoplasmic reticulum at precise levels served ERM (ezrin/radixin/moesin) domain (see Supplementary — for signalling in both non-excitable and excitable tissues, and for Information, Fig. S2), a glutamate-rich region, a serine–proline-rich 1Department of Neuroscience, 2Program in Biochemistry, Cellular and Molecular Biology, The Johns Hopkins University School of Medicine, Baltimore, MD 21205, USA. 3Department of Physiology, University of Texas Southwestern Medical Center, Dallas, TX 75390, USA. 4Department of Neurology, The Johns Hopkins University School of Medicine, Baltimore, MD 21205, USA. 5These authors contributed equally to this work. 6Correspondence should be addressed to S.M. or P.F.W. (e-mail: [email protected]; [email protected]) Received 12 April 2006; accepted 12 June 2006; published online 13 August 2006; DOI: 10.1038/ncb1454 NATURE CELL BIOLOGY VOLUME 8 | NUMBER 9 | SEPTEMBER 2006 1003 © 2006 Nature Publishing Group pprint_ncb1454.inddrint_ncb1454.indd 11003003 116/8/066/8/06 112:35:062:35:06 ppmm LETTERS a d Signal STIM1D76A peptide D76 E-rich + ∆ERM ∆K STIM1 EF SAM TM ERM SP K Thapsigargin 1 23 67 95 132 200 215 234 251 270 336 535 600 629 672 685 D76A − b STIM1 WT STIM1 − NFAT GFP + +FK506 +SKF 100 − Thapsigargin + Thapsigargin 80 100 60 80 40 60 (percentage) 20 40 (percentage) Cells with nuclear NFAT 0 20 −−FK506 FK506 − −−FK506 SKF Cells with nuclear NFAT 0 Ionomycin Thapsigargin D76A: −+∆ERM ∆K − + ∆ERM ∆K STIM1 None WT D76A e c Control CT 5 min STIM1 0.75 STIM1 WT + ∆ERM ∆E ∆SP ∆K D76A Thapsigargin CT D76A−∆ERM − 340:380 ratio 1 Ca2+ 0.25 2 Ba2+ + 0 Ca2+ 10 µM CPA 0.20 * * 200 * − Thapsigargin + Thapsigargin * 100 0.15 160 ratio) * 80 ∆ 0.10 120 influx (% Ctl) 60 influx ( 80 2+ 2+ 0.05 Ba 40 Ca 40 * (percentage) 20 * 0.00 0 l l o T o T Cells with nuclear NFAT 0 r CT 6A r CT 6A W 7 W 7 + ∆ERM ∆E ∆SP ∆K + ∆ERM ∆E ∆SP ∆K nt ERM nt ERM CT: o D o D C −∆ C −∆ D67A D67A Figure 1 The EF-hand mutant and cytosolic fragment of STIM1 activate of scoring results of three wells with at least 100 cells counted per well. endogenous SOC channels in HEK293 cells. (a) Schematic representation (e) HEK293 cells expressing empty vector (control) or the indicated STIM1 of human STIM1 domain structure. The conserved D76 in the EF hand constructs and bathed in medium containing 1 mM Ca2+ were perfused with is shown. (b) Representative live images of HEK293 cells expressing Ca2+-free medium and then with media containing 1 mM Ca2+. The extent NFAT1–GFP and the indicated full-length STIM1 proteins in the presence of of change in 340:380 Fura 2 ratio measured after 2 min of incubation in indicated drugs. Below is the summary of all data quantifying the percentage Ca2+-free media is indicative of the cell permeability to Ca2+. Results from of cells with nuclear NFAT. WT, wild type. (c) Images of NFAT1–GFP in cells 4–7 experiments are plotted in the columns. The cells were then incubated coexpressing the indicated wild-type or deletion mutants of STIM1 cytosolic in Ca2+-free medium containing 10 µM cyclopiazonic acid (CPA) to deplete termini. The cells were imaged and scored before (–) or after (+) thapsigargin the stores and divalent cation permeability was determined from the rate of (thap) treatment, and the summary data is shown below. (d) The effect of Ba2+ influx, which was determined from the first derivative of the Ba2+ influx thapsigargin treatment on NFAT1–GFP localization when coexpressed with traces. All results are the mean ± s.e.m. (Control, n = 7; CT, n = 7; D76A, STIM1D76A full-length or mutants lacking either the ERM domain or the n = 6; D76A−ΔERM, n = 6; WT, n = 4). * donates P < 0.05 relative to control lysine-rich tail. The scale bars represent 10 µm in b–d. Values are the means cells transfected with empty vector. region and a lysine-rich region. To identify regions that are important for thapsigargin in >90% and 70% of the cells, respectively, indicating that SOC-channel activation, we deleted each region separately, and found they act as dominant-negatives (Fig. 1d). In contrast, wild-type STIM1 that STIM1CT mutants lacking either the ERM domain or the lysine-rich lacking the ERM domain or the lysine-rich region are less effective domi- region lose the ability to drive GFP–NFAT into the nucleus (Fig. 1c). nant-negatives that block store-depletion-dependent NFAT translocation Furthermore, STIM1D76A full-length mutants lacking either the ERM in 50% and 40% of cells, respectively (data not shown). These findings domain or the lysine-rich region do not induce nuclear translocation reveal the importance of both the ERM domain and the lysine-rich tail of NFAT, and block GFP–NFAT translocation after store depletion by of STIM1 in activating endogenous SOC channels. 1004 NATURE CELL BIOLOGY VOLUME 8 | NUMBER 9 | SEPTEMBER 2006 © 2006 Nature Publishing Group pprint_ncb1454.inddrint_ncb1454.indd 11004004 116/8/066/8/06 112:35:112:35:11 ppmm LETTERS a b ed Input IP: anti-Myc IP: STIM1 Brain lysates eabsorb HA−TRPC1: +++ +++ +++ r Input P Antibody Myc−STIM1: − WT D76A − WT D76A − WT D76A IP: anti-TRPC1 STIM1 Mr(K) IB: anti-STIM1 62 - − IB: anti-HA HRP HA−TRPC1: ++ ++ (multimer) c 98 - TRPC1 Myc−STIM1: − WT D76A CT (monomer) Mr(K) 62 - 98 - Surface IB: anti-HA−HRP 62 - IB: anti-STIM1 98 - IB: anti-HA−HRP 62 - 62 - Total 98 - IB: anti-STIM1NT + Myc 62 - e d HA−TRPC1 + Myc−STIM1 IP: anti-Myc Input (biotinylatedfraction) IP: anti-Myc Input (biotinylated fraction) HA−TRPC1: +++ +++ − − Myc−STIM1: − WT D76A − WT D76A iono thap iono thap M (K) Mr(K) r (multimer) (multimer) IB: anti-HA−HRP IB: anti-HA−HRP 98 - 98 - TRPC1 TRPC1 (monomer) 62 - (monomer) 62 - IB: anti-STIM1 IB: anti-STIM1 62 - 62 - f g No treatment Thapsigargin: Before After CFP−TRPC1 CFP−TRPC1 YFP−STIM1 WT YFP−STIM1D76A Merged Merged Figure 2 STIM1 and TRPC1 interaction and punctae formation.