PIP₂ and septin control STIM1/Orai1 assembly by regulating cytoskeletal remodeling via a CDC42-WASP/WAVE-ARP2/3 protein complex

Abstract

Store-operated calcium entry (SOCE) is triggered by assembly of Orai1 with STIM proteins in ER-PM junctions. Plasma membrane PIP₂ as well as PIP₂-binding protein, SEPT4, significantly impact Orai1-STIM1 interaction. While septins and PIP₂ can organize the actin cytoskeleton, it is unclear whether the status of actin within the junctions contributes to SOCE. We report herein that actin remodeling modulates STIM1 clustering. Our findings show that a PIP₂- and SEPT4-dependent mechanism involving CDC42, WASP/WAVE, and ARP2 regulates actin remodeling into a ring-like structure around STIM1 puncta. CDC42 localization in the ER-plasma membrane region is enhanced following ER-Ca²⁺ store depletion. PIP₂ depletion or knockdown of SEPT4 attenuate the recruitment of CDC42 to the ER-PM region. Importantly, knockdown of SEPT4, or CDC42+ARP2, disrupts the organization of actin as well as STIM1 clustering. Consequently, Orai1 recruitment to STIM1 puncta, SOCE, and NFAT translocation to the nucleus are all attenuated. Ca²⁺ influx induced by STIM1-C terminus is not affected by CDC42 knockdown. In aggregate, our findings reveal that PIP₂ and SEPT4 affect Orai1/STIM1 clustering by coordinating actin remodeling within ER-PM junctions. This dynamic reorganization of actin has an important role in regulation of SOCE and downstream Ca²⁺-dependent effector functions.

Graphical Abstract

Introduction

Store-operated calcium entry (SOCE) is activated in response to depletion of Ca²⁺ stores within the endoplasmic reticulum (ER) and is mediated by the plasma membrane Ca²⁺ channel, Orai1[1]. Stromal Interaction Molecule 1 (STIM1) senses the decrease in ER-[Ca²⁺] and undergoes a conformational change that not only facilitates STIM1-STIM1 interactions, but also leads to clustering of STIM1 at ER-plasma membrane (ER-PM) junctions where it recruits and activates Orai1 [2, 3]. Interaction between the C-terminal polybasic domain of STIM1 with plasma membrane phosphatidyl 4,5-bisphosphate (PIP₂) is critical for the assembly of Orai1 with STIM1 [4, 5]. An increasing number of studies demonstrate that Orai1/STIM1 assembly as well as stability and architecture of the ER-PM junctions per se are determined by specific protein and lipid components such as PIP₂, septins, and other proteins scaffolding the ER to the plasma membrane. For example pertubations in plasma membrane PIP₂ adversely impact SOCE and STIM1/Orai1 clustering [6–9]. Furthermore, several PIP₂-binding ER-localized proteins, including extended syntaptotagmin (E-Syt) proteins and lipid transfer protein families like Nir and ORP, contribute to the stablity and expansion of these junctions [10, 11].

Septins are a large family of GTP-binding proteins that multimerize into higher-order structures, which associate directly with membranes and contribute to membrane stability. These structures can serve as diffusion barriers for membrane proteins. In addition, septins bind other proteins and can function as platforms to co-ordinate assembly of molecular components involved in cell signaling pathways. Septins have been reported to associate with plasma membrane phospholipids, as well as distinct subsets of actin filaments and microtubules [12–14]. Depleting plasma membrane lipid alters the organization of septins, which can dramatically influence membrane organization and shape [13, 14]. Septins might also be involved in maintenance and positioning of cytoskeletal networks. The Septin 2 subgroup, which consists of Septins 1, 2, 4 and 5, plays a role in mediating SOCE in Jurkat T-cells, HeLa cells and Drosophila neurons [15, 16]. Septins 4 and 5 were shown to modulate Orai1/STIM1 clustering by re-organizing plasma membrane PIP₂ within the junctions [16]. However, Septin 4 does not display stable association with the junctions during assembly of Orai1/STIM1, which led to the suggestion that Septins 4/5 might exert their modulation of Orai1/STIM1 via intermediary components [17]. Septins also associate with PIP2-dependent cytoskeletal remodeling proteins such as cell division cycle 42 (CDC42) that mediates local actin re-organization via activation of the Wiskott-Aldrich syndrome protein (WASP) family of proteins and actin-related proteins 2/3 (ARP2/3) [18, 19]. Finally, remodeling of cortical actin could lead to reorganization of plasma membrane PIP₂ [20–23]. Thus, PIP₂ and cytoskeletal remodeling can reciprocally impact each other and potentially also affect SOCE.

In this study, we investigated the role of actin within the ER-PM junctions in Orai1/STIM1 assembly and SOCE. We found that actin assembles in a ring-like structure around STIM1 puncta within these junctions. PIP₂ depletion, knockdown of Septin4 (SEPT4) or actin remodeling proteins, such as CDC42, WASP, verprolin-homologous protein (WAVE) and ARP2/3, disrupted the actin assembly and attenuated STIM1 clustering and consequently the recruitment of Orai1. This resulted in a reduction of SOCE and translocation of Nuclear Factor Activated Transcription Factor 1 (NFAT1) to the nucleus. Interestingly, SEPT4 and CDC42+ARP2/3 exerted similar, non-additive contributions to SOCE, suggesting that these proteins may act on the same signaling pathway. Further, constitutive Ca²⁺ influx induced by STIM1-C terminus was not affected by siCDC42. Finally, we show that SEPT4 and PIP₂ control the recruitment of CDC42 to the plasma membrane region that is triggered by cell stimulation with thapsigargin (Tg). Together our findings reveal that SEPT4 and PIP₂ coordinate remodeling of actin within ER-PM junctions that is mediated by CDC42, N-WASP/WAVE, and ARP2/3 proteins. The resulting ring-like organization of actin supports STIM1 clustering and assembly of Orai1/STIM1 complex. This suggests that dynamic reorganization of protein and lipid components within ER-PM junctional region plays an important role in the regulation of SOCE and downstream Ca²⁺-dependent effector functions.

Results

PIP₂ and SEPT4 contribute to regulation of SOCE.

Plasma membrane PIP₂ was depleted using rapamycin-inducible phosphatase where the cytosolic 5-phosphatase domain of the type IV phosphoinositide 5-phosphatase enzyme is fused to FKBP12 and plasma membrane-targeted FRB fragment of mTOR [24, 25]. Tg (1μM)-induced clustering of YFP-STIM1 with Orai1-CFP in the ER-PM junctions was assessed using total internal reflection fluorescence (TIRF) microscopy (TIRFM). STIM1/Orai1 clustering was significantly decreased in cells where PIP₂ depletion was triggered by rapamycin addition as compared to control cells (Fig. 1A and B; 58% and 69%, respectively). PIP₂ depletion also decreased YFP-STIM1 clustering in cells even when it was expressed alone (Figs. S1A and B), suggesting that the reduction of Orai1 clustering is due reduced STIM1 clustering. Consequently, Tg-induced SOCE, but not ER-Ca²⁺ store release, was decreased by PIP₂ depletion (Fig. 1C and D). The decrease in plasma membrane PIP₂ was confirmed by measuring fluorescence of YFP-Tubby (a high affinity PIP₂-binding protein; [24]) expressed in cells together with the phosphatase (Figs. S1C and D).

Figure 1. Plasma membrane PIP₂ and SEPT4 contribute to STIM1/Orai1 clustering and function.

(A, B) Orai1-CFP and YFP- STIM1 clustering in control cells and those with PIP₂ depletion. The TIRFM images in A show cells in unstimulated conditions (resting) and after stimulaion with thapsigargin (Tg, 1μM; +Tg). Quantitation of clustering is shown in B. (C, D) 1μM Tg-induced SOCE (Ca²⁺ release and influx) in control cells and those with PIP₂ depletion.The histogram shows Ca²⁺ influx components (SOCE with PIP₂ depletion compared to that in control cells). PIP₂ depletion in A-D was induced by treating cells co-expressing PM-FRB-mRFP and mRFP-FKBP-5-ptase-dom with 200nM rapamycin for 3 min before the addition of Tg (also see Figs. S1A and B). (E, F) Effect of SEPT4 knockdown (siSEPT4) on Tg-induced clustering of Orai1-CFP and YFP-STIM1 in control cells. Quantitation of clusters is shown in F. (G, H) Effect of siSEPT4 and FCF (50μM) treatments on Tg-induced SOCE. The histogram shows Ca²⁺ influx components (SOCE with FCF and siSEPT4 treatment are compared with that in control cells. (I, J) IV curve showing the Tg-induced Ca²⁺ currents in control cells (black) and cells treated with siSEPT4 (red). The histogram shows the magnitude of peak Ca²⁺ currents in the two sets of cells. (K, L) Effect of plasma membrane PIP₂ depletion and siSEPT4 treatment on nuclear translocation of NFAT. The histogram shows the fold-change in nuclear fluorescence of NFAT in cells that were either treated with siSEPT4 or treated with rapamycin to induce PIP₂ depletion. For TIRFM images, scale bar is 10 μm. Data were obtained from n = 3 independent experiments with 10–15 cells used for TIRFM and 5–10 cells for patch clamp experiments, 75–132 cells for NFAT experiments, and 120–196 cells used for calcium imaging experiments. Statitistical significance: P < 0.001 is ***; P < 0.01 is **.

Knockdown of endogenous SEPT4 with siSEPT4 caused significant reduction in STIM1/Orai1 clustering within ER-PM junctions (Figs. 1E and F; c.f. control cells treated with scrambled siRNA, Control). SEPT4 knockdown, as well as forchlorfenuron (FCF; 50μM, which inhibits assembly of the septin network) reduced Tg-induced SOCE (Figs. 1G and H). Measurement of I_CRAC current showed a decrease in the amplitude (measured in HEK293 cells expressing Orai1+STIM1, Figs. 1I and J) in cells treated with siSEPT4. These cells also displayed reduced and transient activation of the current when compared to that in control cells (Fig S1E). Both siSEPT4 and PIP₂ depletion decreased nuclear translocation of NFAT1 that is triggered by Orai1-mediated Ca²⁺ entry (Figs. 1K and L). Loss of SEPT4 reduced co-immunoprecipitation (co-IP) of Orai1 with STIM1. Following Tg stimulation, co-IP of Orai1 and STIM1 increased in control cells (treated with scrambled siRNA) and this was greatly reduced in cells treated with siSEPT4 (Fig. S1F, blot shows co-IP of endogenous proteins).

Remodelling of actin around STIM1 clusters in ER-PM junctions.

The data described above are consistent with previous reports and demonstrate that plasma membrane PIP₂ as well as SEPT4 are important determinants of Orai1/STIM1 assembly within ER-PM junctions. It was suggested that SEPT4 does not directly control Orai1/STIM1 clustering but rather that it induces its effects through an intermediary mechanism [17]. Since both PIP₂ and septins can affect the organization of the cytoskeleton network [16, 22], we visualized the actin cytoskeleton within ER-PM junctional region by expressing pCMVLifeAct-TagRFP (LifeAct) together with YFP-STIM1. While unstimulated cells showed an apparently linear arrangement of actin, Tg treatment resulted in recruitment and re-arrangement of LifeAct into a ring-like structure around STIM1 puncta (see representative image montages obtained from control cells before and after Tg stimulation, Fig. 2A, time between each frame was 5sec and Tg addition at 10^th frame is indicated). As the STIM1 cluster appeared, actin (green signal) initially decreased and then appeared in a ring-like structure around the STIM1 puncta (red signal). Enlarged images shown in Fig.2C illustrate remodeling of actin around a single STIM1 puncta. Select frames starting at different time points are shown - top row starting at 5s (before Tg addition), middle at 75s (15 sec after Tg addition), last at 110s (50 secs after Tg addition) (5s interval/frame). Within this actin ring, there was an increase in the size and fluorescence intensity of the STIM1 cluster (Fig. 2C). Further enlargement of the frames indicated in Fig. 2C (i and ii) are shown in Fig 2E (top panel). Line scans of the two frames (Fig. 2F, left graph) show actin (red peaks) corralling the STIM1 puncta (black peak). The line scans for all frames starting at 75s and 110s in Fig. 2C (iii and iv) are shown Fig. S2A panel, highlighting the development of STIM1 puncta and re-arrangement of actin over time.

Figure 2. SEPT4-dependent actin remodeling in ER-PM junctions contributes to STIM1 clustering and SOCE.

(A, B) Montages showing images of a YFP-STIM1 (STIM1) puncta and pCMVLifeAct-TagRFP (LifeAct) surrounding the puncta in a cell expressing both proteins, without (Control) and with knockdown of SEPT4 (siSEPT4). Time frame for each individual image in the montage is 5s, with the addition thapsigargin (Tg, 1μM) as shown by the arrow. (C, D) View of selected frames from the montages in (A, B) for time points starting at 5s, 75s and 110s. (i, ii) denote individual frames enlarged in (E) and used in line scans shown in (F). (iii, iv) denote the series of frames starting at time points 75s and 110s used in the line scans shown in Fig. S2. Overlay images were created using images of LifeAct and STIM1 pseudo-colored green and red respectively. All images shown were obtained using TIRFM. Images and scans are representative of data obtained from 10–12 cells in n = 3 independent experiments. Effect of treatment with Latrunculins A or B on (G) Tg-induced Ca²⁺ release and influx and (I)nuclear translocation of NFAT. The histograms show the magnitude of (H) Ca²⁺ influx and (J) fold-changes in nuclear fluorescence of NFAT in control cells (black) and cells treated with Lantrunculins A (red) and B (blue). Data were obtained from 131–150 cells in n = 3 independent experiments. Statitistical significance: P < 0.001 is ***.

To determine whether SEPT4 is involved in the restructuring of actin, LifeAct was imaged in cells treated with siSEPT4. As shown in montages for siSEPT4-treated cells in Figs. 2B, D and E, knockdown of SEPT4 caused dramatic changes in actin assembly as well as formation of STIM1 puncta. Unlike control cells, actin was not recruited into a defined ring-like structure in cells with SEPT4 knockdown. Importantly, although small STIM1 punctae were detected, the appearance was more delayed than in control cells and they did not increase in size or fluorescence intensity over time. As shown in the enlargement of Fig. 2D (i and ii) in Fig. 2E (bottom panel), as well as the line scans in Fig. 2F (right graph) and Fig. S2B, STIM1 puncta were not formed in siSEPT4-treated cells. Furthermore, actin did not re-arrange into a ring-like structure. Thus, remodeling of actin was associated with the clustering of STIM1 in ER-PM junctions and both were dependent on SEPT4.

We next assessed whether the effect of siSEPT4 on actin structure and STIM1 puncta were coincident events or whether actin remodeling contributed to STIM1 clustering within ER-PM junctions. Lantrunculin A or B, which block actin remodeling by sequestering G actin and preventing actin polymerization, caused significant decreases in SOCE (Figs. 2G and H). Corresponding to the lantrunculin-induced inhibition of SOCE, activation of NFAT1 following the addition of Tg was significantly reduced following treatment with these agents (Figs. 2I and J). Similar decrease of SOCE was induced by jasplakinolide, which attenuates actin remodeling by stabilizing and enhancing actin polymerization (Figs. S3A and B). Together, these data strongly suggest that the organization of actin within ER-PM junctions contributes to STIM1 clustering in ER-PM junctions.

Role for PIP₂ in regulation of actin remodeling.

PIP₂-dependent mechanisms control remodeling of cortical actin [22, 26, 27]. Several proteins involved in actin remodeling can bind to plasma membrane PIP₂. We assessed the involvement of the PIP₂-dependent actin-remodeling protein, CDC42, in SOCE. As shown in Fig. 3A and B, depletion of ER-Ca²⁺ stores by 1μM Tg induced a time-dependent increase in GFP-CDC42 in the sub-plasma membrane region. Depletion of plasma membrane PIP₂ almost completely suppressed this enhancement, consistent with previous suggestions that CDC42 is scaffolded by plasma membrane PIP₂ [28] Interestingly, a decrease in CDC42 recruitment was also seen by knockdown of SEPT4 (Fig. 3C and D), although the extent was less than that seen in PIP_2-depleted cells. Functional significance of CDC42 in SOCE was tested by knocking down the protein using siCDC42, either alone or in combination with siSEPT4. Loss of CDC42 alone caused a significant reduction of SOCE although to a lesser degree than that caused by siSEPT4 alone. When CDC42 and SEPT4 were knocked down simultaneously (siCDC42+siSEPT4), the decrease was not significantly different from that with siSEPT4 alone (Fig. 3E). Note that knocking down CDC42 did not decrease spontaneous Ca²⁺ entry induced by expression of STIM1 C-terminus (Fig. 3F).

Figure 3. Role for PIP₂- and SEPT4-dependent recruitment of CDC42 in SOCE.

Effects of rapamycin-induced (A, B) PIP₂ depletion and (C, D) knockdown of SEPT4 (siSEPT4) on localization of GFP-CDC42 in the TIRF plane, before and after stimulation with thapsigargin (Tg, 1μM). Scale bar, 10μm. Line graphs show whole-cell fluorescence of GFP-CDC42 at the TIRF plane following (B) PIP₂ depletion and (D) siSEPT4 treatment, with the arrow showing the additions of rapamycin (200nM) and Tg (1μM) where applicable. (E, F). Effect of knocking down CDC42 (siCDC42) and SEPT4 (siSEPT4), either singly or in combination, on Tg-induced Ca²⁺ release and influx. The histogram shows the magnitude of Ca²⁺ influx for control cells (black) and cells treated with either siCDC42 (red), siSEPT4 (blue) or both (magenta). (F) Effect of siCDC42 on Ca²⁺ influx in cells expressing the soluble form of STIM1 C-terminus (STIM1-CT). The histogram shows the magnitude of Ca²⁺ influx for control cells (black) and cells treated with siCDC42 (red). Data were obtained from n = 3 independent experiments with 20–25 cells used for TIRFM experiments, and 125–200 cells used for calcium imaging experiments. Statistical significance: P < 0.001 is ***.

A number of proteins work downstream of CDC42 to control actin remodeling; e.g. N-WASP, WAVE2 and ARP2/3. Knockdown of WAVE1 or WAVE2 (siWAVE2) individually induced a small but significant decrease in Tg-induced SOCE. Knockdown of both together did not cause further decrease in SOCE (Fig. 4A). Knockdown of N-WASP (siNWASP) alone also caused a similar small decrease in SOCE (Fig. 4B). However, knocking N-WASP with either WAVE1 or WAVE2, caused additional decrease in SOCE (siN-WASP+siWAVE1, siN-WASP+siWAVE2) (Fig. 4B). This suggests that while either WAVE protein can exert small effects on SOCE, each can amplify the effect of N-WASP. The ARP2/3 protein complex mediates nucleation of the actin filaments and branching of actin [23, 29–31]. Knockdown of ARP2 significantly reduced both Tg-induced SOCE (Fig. 4C) and I_CRAC (Fig. 4D). Magnitude of decrease in the amplitude of I_CRAC was similar in cells treated with either siARP2 or siCDC42 (Fig. 4D). Moreover, SOCE in cells with simultaneous knockdown of both ARP2 and CDC42 was slightly less than that in cells where CDC42 alone was knocked down (Fig. 4E). Interestingly, knockdown of ARP2 together with SEPT4 caused further attenuation of SOCE than that induced by SEPT4 knockdown alone (Fig. 4F). Consistent with the significant decrease in SOCE following the knockdown of either CDC42 or ARP2 alone, or both proteins together, Tg-induced nuclear translocation of NFAT was also reduced under these conditions (Fig. 4G).

Figure 4. PIP₂-dependent cytoskeleton-remodeling proteins, WASP/WAVE and ARP2, have a role in regulating SOCE via the same pathway as CDC42.

Effects of knocking down WAVE1 alone (A, red), WAVE2 alone (A, blue) or both WAVEs together (A, green) and N-WASP alone (B, red) or together with either WAVE1 (B, blue) or WAVE2 (B, green) on thapsigargin (Tg, 1 μM)-induced Ca²⁺ release and influx. Control is shown as black line graph in both figures. The histograms show the magnitude of Ca²⁺ influx for the conditions described in A and B. (C, D) Tg-induced Ca²⁺ responses in control cells (black) and siARP2-treated cells (red). The magnitude of Ca²⁺ influx for both conditions is shown by the histogram. (D) Tg-induced Ca²⁺ currents in control cells (black) and cells treated with either siARP2 (blue) or siCDC42 (red), with the histogram showing the magnitude of peak current. (E) Effects of knocking down CDC42 either alone (siCDC24, red) or together with ARP2 (siCDC24+siARP2, blue) on Tg-induced Ca²⁺ responses (c.f. control cells, black). (F) Effect of knocking down SEPT4 either alone (siSEPT4, red) or together with ARP2 (siSEPT4+siARP2, blue) on Tg-induced Ca²⁺ responses (c.f. control cells, black). For both E and F, the histograms show the magnitude of Ca²⁺ influx for the same conditions shown on the line graphs. (G) Tg-induced nuclear translocation of NFAT in control cells (black) and cells treated with either siCDC42 (red), siARP2 (blue) or siCDC42+siARP2 (green), with the histogram showing the relative fold-change in nuclear fluorescence. Data were obtained from n = 3 independent experiments, with 10–12 cells used for patch clamp experiments, 77–250 cells used for calcium imaging experiments and 88–180 cells used for NFAT experiments. Statitistical significance: P < 0.001 is ***; P < 0.01 is **.

We also examined the possible contribution of other proteins that modulate remodeling of the actin cytoskeleton. Knockdown of cortactin (cortical actin binding protein), a relatively weak modulator of actin [30] induced a small but significant decrease of SOCE, while knocking down cofilin, which fragments actin without inhibiting remodeling [32, 33], did not have any effect (Figs. S3C and D). Additionally, knockdown of siSEPT4 or siARP2 alone also attenuated agonist-induced SOCE (Figs. S3E and F), consistent with the effects on Tg-induced SOCE.

Role of CDC42, ARP2 and SEPT on the recruitment of STIM1/Orai1 clusters.

The magnitude of Tg-induced SOCE responses depends on the clustering of STIM1 in ER-PM junctions where it recruits and activates the Orai1 channel. STIM1 and Orai1 clustering in response to Tg stimulation of cells was significantly reduced in cells where CDC42 was knocked down either alone (siCDC42) or together with ARP2 (siCDC42+siARP2) (images are shown in Figs. 5A–C, bar graphs in Fig 5D show the relative changes in STIM1 and Orai1 puncta intensity). CDC42 and ARP2 also contribute to actin remodeling in ER-PM junctions, as knockdown of the two proteins simultaneously disrupted the ring-like actin that forms around STIM1 puncta (Figs. 5E and F), reminiscent of what was seen when SEPT4 was knocked down (Fig. 2). These findings substantiate the suggestion that SEPT4, CDC42, and ARP2/3 work concertedly to faciliate actin remodeling around STIM1 puncta. Importantly, this remodeling is required for optimal clustering of STIM1 and Orai1. Figs. S4A and B show that CDC42 is recruited to where STIM1 puncta are formed after stimulation of cells with Tg. Functional relevance of CDC42 and ARP2 are shown in Figs. S4C and D, where knockdown of CDC42 and ARP2 individually in cells expressing Orai1+STIM1 (O1+S1) caused reductions in SOCE and when both proteins were knocked down there was further decrease in SOCE.

Figure 5. CDC42 and ARP2 are involved in STIM1/Orai1 clustering as well as actin remodeling.

TIRFM images showing Orai1/STIM1 clustering in (A) control cells and cells treated with either (B) siCDC42 or (C)siCDC42+siARP2, before and after stimulation with thapsigargin (Tg, 1μM). Scale bar, 10μm. (D) Histograms show the magnitude of STIM1 and Orai1 puncta fluorescence for the conditions described in A to C. (E) View of selected frames from the montages showing images of a YFP-STIM1 (STIM1) puncta and pCMVLifeAct-TagRFP (LifeAct) surrounding the puncta in a cell expressing both proteins, without (Control) and with knockdown of both CDC42 and ARP2 (siCDC24+siARP2) for time points starting at 5s, 75s and 110s. Time frame for each individual image in the montage is 5s. Images shown for the 75s and 110s panels were obtained after stimulation with Tg. (i to iii) denote individual frames used in line scans shown in (F). Overlay images were created using images of LifeAct and STIM1 pseudo-colored green and red respectively. Data were obtained from n = 3 independent experiments with 9–12 cells used for TIRFM experiments. Statitistical significance: P < 0.001 is ***.

Discussion

The data we have presented above demonstrate that SEPT4 and PIP₂ co-ordinate remodeling of actin in ER-PM junctions. Specifically, actin reorganizes to form a ring-like structure around STIM1 clusters that assemble within ER-PM junctions in response to stimulation of cells with agents that induce ER-Ca²⁺ store depletion. This remodeling of actin is an important determinant for the accumulation of STIM1 clusters within ER-PM junctions and thus, for the assembly of Orai1/STIM1 and activation of SOCE. Our findings also reveal the underlying mechanism which controls actin remodeling. We report that SEPT4 and PIP₂ promote recruitment of PIP₂-dependent cytoskeletal modifying protein, CDC42, which together with N-WASP/WAVE and the ARP2/3 complex, controls the actin reorganization around STIM1 puncta which cluster in ER-PM junctions after depletion of the ER-Ca²⁺. We present several key findings to support these suggestions. Depletion of plasma membrane PIP₂ or knockdown of SEPT4 attenuates STIM1-Orai1 clustering and SOCE in addition to disrupting re-organization of actin around the STIM1 clusters. Further, CDC42 is recruited to the plasma membrane region following ER-Ca²⁺ store depletion. Knockdown of SEPT4 decreases the recruitment of CDC42 together with loss of actin reorganization and reduction in STIM1/Orai1 clustering. Knockdown of CDC42 alone or CDC42+ARP2 together cause similar reductions in actin remodeling, STIM1 clustering and SOCE. Collectively, the data support a model suggesting that functional communication between plasma membrane PIP₂, septin and actin regulate STIM1/Orai1 clustering to control SOCE in response to ER-Ca²⁺ store-depletion. Our findings highlight a novel role for actin remodeling in STIM1 and Orai1 clustering within ER-PM junctions, which directly impacts activation of SOCE and regulation of Ca²⁺-dependent cell functions, e.g. NFAT1 activation.

There are a large number of studies that suggest close association between remodeling of actin and SOCE. Specifically, cortical actin that resides immediately below the plasma membrane and likely interacts with both the plasma membrane and ER through actin-binding proteins and has been shown to have a role in modulating SOCE. Actin modifying reagents that either stabilize or depolymerize actin have been shown to either have no effect on SOCE or enhance or decrease, respectively, Ca²⁺ entry [34–36]. The data we show herein reveal an important link between dynamic local remodeling of actin and clustering of STIM1 within ER-PM junctions. Further, our findings demonstrate that as STIM1 puncta appear in the ER-PM region, actin initially decreases and then reorganizes to form ring-like structures that appear to corral the STIM1 clusters. Treatment with agents that disrupt the remodeling or knockdown of proteins that mediate actin remodeling cause attenuation of STIM1 clustering, but do not eliminate the appearance of STIM1 clusters in the ER-PM junctional region. Therefore, although store-depletion per se initiates the translocation of STIM1 to the cell periphery, aggregation of STIM1 within ER-PM junctions is controlled by the architecture of actin within these junctions. It has been previously reported that interaction of the C-terminal polybasic domain of STIM1 with the plasma membrane lipids stabilizes the STIM1 clusters and also contributes to the formation and expansion of the junctions [37–39]. Thus, reorganization of this membrane lipid could also directly affect assembly of STIM proteins with Orai1 in ER-PM junctions. A number of proteins have been demonstrated to contribute to the formation and stability of ER-PM junctions. These include regulators of plasma membrane PIP₂ (e.g. Nir2 [40]) and scaffold proteins that link ER with the PM (e.g. E-Syt1, E-Syt2 [10]). Further studies will be needed to delineate the specific role of these proteins on PIP₂ and PIP₂-dependent actin remodeling.

In conclusion, SEPT4, which appears to modulate plasma membrane PIP₂ domains and contribute to STIM1/Orai1 clustering is not present within ER-PM junctions where STIM1 puncta assemble [16]. This led to the suggestion that SEPT4 might promote STIM1/Orai1 clustering by enhancing the stability of ER-PM junctions via regulating other intermediary proteins [17]. Our findings reveal a concerted role for plasma membrane PIP₂ and Septin in regulating a PIP₂-dependent actin remodeling complex, consisting of CDC42, WASP/WAVE, and ARP2, which mediate remodeling of the actin locally within ER-PM junctions. The architecture of actin contributes to the aggregation of STIM1 clusters in the ER-PM junctions, recruitment of Orai1 by STIM1, and activation of SOCE.

Materials and Methods

Cel culture, RNAi transfection and reagents.

HEK293 cells were cultured at 37°C under 10% CO₂ in EMEM media supplemented with 10% heat-inactivated fetal bovine serum, 1% glutamine and 1% penicillin/streptomycin. All the sequences used for RNAi transfections were obtained from Dharmacon (Lafayette, CO, USA). Sequences for the siRNAs targeting the human SEPT2 subgroup (siSEPT #3 and #4) had been previously described [16]. Smartpools of siRNAs against CDC42, N-WASP, WAVE1, WAVE2, ARP2/3, cofilin and cortactin were obtained from Dharmacon (Lafayette, CO, USA). Lipofectamine RNAiMAX (Life Technologies, Carlsbad, CA, USA) was used for siRNA transfections. Mock transfections of control cells were conducted using siRNA Negative Control (Life Technologies). Cells were typically used 48h post-transfection. All other reagents used were of molecular biology grade obtained from Sigma-Aldrich (St Louis, MO, USA) unless mentioned otherwise.

Plasmids and Transient Transfections.

Orai1-CFP, PM-FRB-mRFP, mRFP-FKBP-5-ptase-dom and YFP-Tubby were kind gifts from Tamas Balla (NICHD, NIH, Bethesda, MD, USA), while YFP-STIM1 was from Tobias Meyer (Stanford University, Stanford, CA, USA), respectively. GFP-CDC42 plasmid was obtained from Addgene (Cambridge, MA, USA). pCMVLifeAct-TagRFP was obtained from ibidi USA (Fitchburg, WI, USA). For these plasmids, cells were transfected using Lipofectamine 2000 (Life Technologies). Experiments were performed 24h post-transfection.

[Ca²⁺]_i Measurements.

Fura-2 fluorescence was measured in single cells cultured for 24h in glass-bottomed MatTek tissue culture dishes (MatTek Corporation, Ashland, MA, USA) and transfected as required; experiments were done 24–48 h post-transfection. Cells were loaded with 1μM Fura-2 (Life Technologies) for 30min at 37°C. Fluorescence was recorded using a Till Photonics Polychrome V spectrofluorimeter (FEI, Hillsboro, OR, USA) and MetaFluor imaging software (Molecular Devices, Sunnyvale, CA, USA) on an Olympus IX51 microscope (Olympus, Center Valley, PA, USA).

Electrophysiology.

Coverslips with HEK293 cells expressing Orai1 and STIM1 were transferred to the recording chamber and perfused with Ca²⁺ containing standard external solution (Ca²⁺-SES) with the following composition (in mM): NaCl-145; KCl-5; MgCl₂-1; CaCl₂-1; HEPES-10; glucose-10; pH 7.4 (NaOH). Cells expressing both proteins were selected for I_CRAC measurements. The patch pipette had resistances between 3–5mΩ after filling with a standard intracellular solution that contained the following (in mM): CsMS-145; NaCl-8; MgCl₂-10; HEPES-10; EGTA-10; pH 7.2 (CsOH). Osmolarity for all the solutions was adjusted with mannose to 300mOsm using a vapor pressure Osmometer (Wescor, Logan, UT, USA). All electrophysiological experiments were performed in the tight-seal whole cell configuration at room temperature (22–25°C) using an Axopatch 200B amplifier (Molecular Devices). Development of the current was assessed by measuring the current amplitudes at a potential of −80mV, taken from high resolution currents in response to voltage ramps ranging from −90 to 90mV over a period of 100ms imposed every 2s (holding potential was 0mV), and digitized at a rate of 1kHz. Liquid-junction potentials were less than 8mV and were not corrected. Capacitative currents and series resistance were determined and minimized. For analysis, the current recorded during the first ramp was used for leak subtraction of the subsequent current records.

TIRF microscopy.

HEK293 cells were plated on collagen-coated glass-bottomed tissue culture dishes (MatTek Corporation), transfected as required and TIRF microscopy (TIRFM) experiments were performed 24h post-transfection. TIRFM was conducted using an Olympus IX81 motorized inverted microscope (Olympus) using 447, 514 and 568 nm lasers for excitation of CFP, YFP and mCherry respectively, a TIRF-optimized Olympus Plan APO 60x (1.45 NA) oil immersion objective and Lambda 10–3 filter wheel (Sutter Instruments, Novato, CA, USA) containing 480-band pass (BP 40 m), 540-band pass (BP 30 m) and 605-band pass (BP 52 m) filters for emission. Images were collected using a Hamamatsu ORCA-Flash4.0 camera (Olympus) and the MetaMorph imaging software (Molecular Devices). These arbitrary fluorescence values were then plotted using the Origin software (OriginLab, Northampton, MA).

Drug treatments and PIP₂ depletion.

Treatments with Forchlorfenuron (FCF, 50μM, 4h [41]), Jasplakinolide (10μM, 1h), LatrunculinA (0.5μM, 30 min) and LatrunculinB (0.5μM, 1h) were carried out in cells incubated at 37°C at the concentrations and time durations as described above. HEK293 cells were transiently transfected with PM-FRB-mRFP and mRFP-FKBP-5-ptase-dom [25] and used 24h later. Rapamycin (200nM) was added to the cells to induce recruitment of the phosphatase to the plasma membrane.

Immunoprecipitation (IP) and Western Blotting.

Where indicated in the figure legend, cells were first stimulated for 5min with 1μM Tg. Cells were then washed with 1x Phosphate-Buffered Saline and lysed in Pierce IP lysis buffer supplemented with protease inhibitors (both from Thermo Fisher Scientific, Waltham, MA, USA). Cell lysates were centrifuged (10,000g, 10min at 4°C) and quantified by Bradford protein assay using Bio-Rad protein assay solution (Bio-Rad, Hercules, CA^, USA). Protein concentrations in the lysate was adjusted to 2mg/ml and incubated with 10μg/ml IP antibody. IP experiments were done using anti-Orai1 antibody (custom made; 1:1000) and Protein A Sepharose CL-4B (GE Healthcare Life Sciences, Piscataway, NJ, USA) as described earlier unless indicated otherwise. Rabbit polyclonal antibody to human Orai1 was produced against C-terminal epitope ELAEFARLQDQLDHRGD and purified by Lofstrand Labs Limited (Gaithersburg, MD, USA). Immunoprecipitates were released by heating by heating at 95°C for 10min in SDS-sample buffer and resolved in 4–12% NuPAGE gels (both from Life Technologies), followed by Western blotting. Immunoblotting was done using anti-STIM1 (Cell Signaling Technology, Danvers, MA, USA; 1:1000) antibody.

Measurement of NFAT Translocation into the Nucleus.

Translocation of NFAT in transfected cells was observed using an Olympus IX81 motorized inverted microscope (Olympus). Excitation was achieved using the 488 nm laser for excitation of GFP, and emission detected using a Lambda 10–3 filter wheel (Sutter Instruments, Novato, CA) containing the 525-band pass (BP50m) filter. Images were collected using a CoolSNAPHQ₂ camera (Photometrics, Tucson, AZ, USA) and the MetaMorph imaging software (Molecular Devices). ImageJ (NIH, Bethesda, MD, USA) was used to measure the fluorescence intensity in the nucleus before and after stimulation with 1μM Tg. Regions of interest (ROIs) were selected to obtain the values for their fluorescence intensities during a time course experiment. These values were then plotted using the Origin 2017 (OriginLab).

Statistics.

Data analysis was performed using Origin 2017 (OriginLab) and GraphPad Software (La Jolla, CA, USA). Statistical comparisons between two groups were made using the Student’s t-test, whereas comparisons of multiple (> two) groups were made using one-way ANOVA followed by Sidak multiple comparisons tests. Experimental values are expressed as mean ± SEM. Differences in the mean values were considered to be significant at P < 0.05.

Highlights.

• Store-operated calcium entry (SOCE) requires assembly of Orai1 with STIM proteins

• Remodeling of actin in ER-PM junctions promotes STIM1/Orai1 clustering

• PIP2 and SEPT4 control actin remodeling via regulating CDC42, WASP/WAVE, and ARP2

• Actin remodeling provides dynamic regulation of Orai1/STIM1 assembly and SOCE.