A critical role for STIM1 in cardiac hypertrophy

Abstract

Background

Cardiomyocytes (CM) utilize Ca²⁺ not only in excitation-contraction coupling (ECC), but also as a signaling molecule promoting for example cardiac hypertrophy. It is largely unclear how Ca²⁺ triggers signaling in CM in the presence of the rapid and large Ca²⁺ fluctuations that occur during ECC. A potential route is store-operated Ca²⁺ entry (SOCE), a drug-inducible mechanism for Ca²⁺ signaling that requires stromal interaction molecule 1 (STIM1). SOCE can also be induced in cardiomyocytes, which prompted us to study STIM1-dependent Ca²⁺-entry with respect to cardiac hypertrophy in vitro and in vivo.

Methods and Results

Consistent with earlier reports, we found drug-inducible SOCE in neonatal rat cardiomyocytes, which was dependent on STIM1. While this STIM1-dependent, drug-inducible SOCE was only marginal in adult cardiomyocytes isolated from control hearts, it significantly increased in cardiomyocytes isolated from adult rats that had developed compensated cardiac hypertrophy after abdominal aortic banding. Moreover, we detected an inwardly rectifying current in hypertrophic cardiomyocytes that occurs under native conditions (i.e. in the absence of drug-induced store depletion) and is dependent on STIM1. By manipulating its expression, STIM1 was found to be both sufficient and necessary for cardiomyocyte hypertrophy both in vitro and in the adult heart in vivo. Stim1 silencing by AAV9-mediated gene transfer protected rats from pressure overload-induced cardiac hypertrophy.

Conclusions

STIM1 promotes cardiac hypertrophy by controlling a previously unrecognized sarcolemmal current.

Introduction

Cardiomyocytes exhibit dramatic fluctuations of cytosolic Ca²⁺ in alternating cycles of excitation, contraction and relaxation. In each cycle, depolarization of the plasma membrane causes an influx of Ca²⁺ into the cytoplasm through the sequential opening of plasma membrane L-type Ca²⁺ channels (LTCC) and sarcoplasmic reticulum (SR) ryanodine receptors (RyR)^{1, 2}. The coupling of RyR activity to LTCC gating is structurally based on their clustering in diads, i.e. small cellular compartments where these molecules are in close proximity^{1, 2}. Originating from this compartment, Ca²⁺ then rapidly diffuses to adjacent sarcomeres, where it binds to troponin C, thus facilitating actin-myosin interaction for muscle contraction. In addition to cardiomyocyte contraction, Ca²⁺ is fundamental for transcriptional activation that occurs during the cardiac growth response (hypertrophy) to various stressors. The two major Ca²⁺-dependent signaling pathways in this respect result in the activation of nuclear factor of activated T-cells (NFAT) through calcineurin, or in the regulation of histone deacetylase activity through CaM-kinase II (CamKII) and/or protein kinase D (PKD)^3–5.

For the Ca²⁺-dependent regulation of nuclear HDAC activity, a mechanism involving IP₃-dependent release of Ca²⁺ from the nuclear envelope and/or CamKII-dependent activation of PKD has been demonstrated^{6, 7}. By contrast, it is largely unclear how the CaN-NFAT-axis is activated in the presence of the Ca²⁺ fluctuations that occur during cardiomyocyte excitation-contraction coupling⁸. The Ca²⁺ involved here seems to be released independent from the myofilament-activating Ca²⁺ transients. Candidate gates for Ca²⁺ that induce downstream signaling are the LTCC, the RyR, or dysfunctional SR Ca²⁺ ATPase (SERCA).

A potential mechanism for Ca²⁺-dependent NFAT activation in cardiomyocytes is store-operated Ca²⁺ entry (SOCE), which has been largely described in non-excitatory cells, but also observed in cardiomyocytes upon drug-induced inhibition of the ER/SR Ca²⁺ ATPase (SERCA)^{9, 10}. In non-excitable cells, two groups of proteins, termed Stromal interaction molecules (STIM) and Ca²⁺ release-activated channel modulators (CRACM, also termed ORAI), have been identified as key mediators of SOCE-mediated Ca²⁺ signaling^{11, 12}. The characterization of STIM1 has unraveled a complex mechanism for Ca²⁺-dependent signal transduction. Upon depletion of the endoplasmic reticulum (ER) from Ca²⁺, STIM1 forms homo-oligomers and relocalizes within the ER to regions that are in close vicinity to the cytoplasma membrane. STIM1 thereby promotes the opening of ORAI1 in the plasma membrane to induce the entry of extracellular Ca^{2+ 13}, resulting in transcriptional activation through NFAT^{14, 15}. In addition, STIM1 has also been reported to interact with the canonical transient receptor protein channel family (TRPC) in various cell types^16–20 and contacts between STIM1 and TRPC1 have been mapped in both proteins^{16, 21}. This is one of the arguments why TRPC channels have likewise been regarded as store-operated channels²⁰, despite the fact that they are less selective for Ca²⁺ than ORAI proteins^{22, 23}. It seems currently unresolved whether store-operated currents activated by STIM1 are based on its interaction with one or several ORAI proteins, TRPC proteins, or even a combination of both protein families^{19, 24}.

Even less is known about the potential function of STIM1 in the heart. Two recent studies carried out in neonatal rat cardiomyocytes suggested that it translates pro-hypertrophic stimuli into a growth response^{25, 26}, but the underlying mechanism remained elusive and it remained unclear whether STIM-dependent Ca²⁺ entry occurs in the adult heart.

In the present study, we have expanded the experimental scope to cardiomyocytes isolated from an in vivo model for cardiac hypertrophy (transaortic constriction in adult rats) and ultimately to the rat heart in vivo. We analyzed transmembrane currents and intracellular Ca²⁺ concentrations in the presence and in the absence of drugs that induce SR store depletion, and investigated their dependence on expression of STIM. Our data provide evidence for an inwardly rectifying current which occurs in the absence of store-depleting drugs, which increases significantly in cardiomyocyte hypertrophy and which requires STIM1. Together with our finding that reduced expression of STIM1 in vivo partially protects from cardiac hypertrophy, we conclude that STIM1 is a key player in cardiac hypertrophy.

Methods

Please refer to the Expanded Methods sections for a detailed description.

Cardiomyocyte isolation and cellular electrophysiology

Whole-cell patch-clamp experiments were performed at room temperature (~24 °C) with an Axopatch 200B (Axon Instruments, Burlingham, CA, USA). Patch pipettes had a resistance of 2–3 MΩ. Currents were normalized to the cell membrane capacitance and presented as current density (pA/pF). Patch pipettes were filled with a solution containing 137 mM cesium aspartate, 2 mM CsCl, 8 mM MgSO₄, 15 mM HEPES, 5 mM EGTA, adjusted to pH 7.2 (with CsOH) and 310 mOsm (with d-Mannitol). The external solution consisted of 150 mM NaCl, 2 mM CaCl₂, 1 mM MgCl₂, 10 mM HEPES, 10 mM glucose, 20 mM sucrose, adjusted to pH 7.4 (with NaOH) and 320 mOsm (with d-mannitol); in the NMDG solution, Na⁺ was replaced with an equimolar amount of N-Methyl-D-glucamin (pH 7.4 adjusted with HCl); in the Ba²⁺ solution, Ba²⁺ replaced Ca²⁺; the DVF solution was divalent cation-free Ca²⁺ and contained 10 mM EDTA. To block the L-type Ca²⁺ channel, verapamil (10 μM) was added in external solutions; K⁺ channels were blocked by Cs⁺ in the internal solution; and the voltage-dependent Na⁺ channel was inactivated by the stimulation protocol. Currents were recorded using an Axonpatch-200A amplifier with Digidata 1200 interface and analyzed with pCLAMP software. Currents were induced by 1 s voltage ramp protocols every 2 s (from 50 mV to −120 mV), at a holding potential −80 mV. As quality controls of the patch-clamp configuration, access resistance was required to stay below 6.5 MΩ, be stable throughout analysis, and leak current should not exceed 100 pA at −80 mV in external standard solution (with Ca²⁺ and Na²⁺) during 5 min before switching solution. In this condition, 17% of cardiomyocytes from sham-operated cells were excluded and 23% of cells from AAB animals.

Transgenic mice

The generation of Stim1^−/− mice has been described recently²⁷. Wild-type or heterozygous Stim1^+/− mice served as controls.

Statistical analysis

All quantitative data are reported as means ± SEM. Statistical analysis was performed with the Prism software package (GraphPad v4). One-way analysis of variance (ANOVA) was used to compare each parameter. Post hoc t-test comparisons were performed to identify which group differences accounted for significant overall ANOVA results. Differences were considered significant when p<0.05.

Results

Store-operated Ca²⁺ entry is STIM1-dependent in both neonatal and adult cardiomyocytes and is strongly enhanced in CM from hypertrophic hearts

We initially analyzed the effects of manipulated STIM1 expression on intracellular Ca²⁺ concentrations in isolated neonatal rat cardiomyocytes (NRCM) and in cardiomyocytes from Stim1^−/− mice. In each case, cells were exposed to Ca²⁺-free buffer and thapsigargin, a SERCA inhibitor that causes depletion of the SR from Ca²⁺. Then, CaCl₂ was added as an extracellular source for store-operated Ca²⁺ entry, typically resulting in Ca²⁺ influx, as measured by Fura-2 (Figure 1A and B, black tracings). Partial silencing of Stim1 (by adenoviral infection with a short hairpin RNA matching Stim1 mRNA) significantly reduced this response, as also observed in Stim1^−/− CM (Figure 1B, green tracings). A reciprocal effect was obtained upon adenovirus-driven Stim1 overexpression in NRCM (Figure 1A, red tracing). As detected by an immunostaining with an anti-STIM1 antibody, thapsigargin treatment of wild-type rat and mouse CM enhanced the clustering of STIM1 into puncta, a typical phenomenon of SOCE (Figures 1C and supplemental Figure 1A). Given the short time until cell fixation (10 min), we attribute the increased signal to the clustering and redistribution of STIM1 rather than to upregulated Stim1 expression. A similar pattern was obtained using an alternative antibody raised against the N-terminus of STIM1 (supplemental Figure 1B). We also controlled for a potential contribution of non-cardiomyocyte cells to these results. Our isolates typically contain <20% of non-cardiomyocyte cells, the majority of which being cardiac fibroblasts (Supplemental Figure 2A). Ca²⁺ fluxes in these cells were smaller and not affected by Stim1 silencing (Supplemental Figure 1C), excluding a significant contribution of non-myocyte cells to the above findings. Together, these data recapitulate and expand the existing knowledge that STIM1 is required for SOCE in the presence of a SERCA inhibitor.

Figure 1. Role of STIM1 in store depletion-induced Ca²⁺ entry in neonatal cardiomyocytes.

(A) Fluorescence analysis of Ca²⁺ entry in neonatal rat cardiomyocytes (NRCM). Left, representative recordings of the Fura-2 emission ratio (F340/380) in cardiomyocytes after SR Ca²⁺ depletion (by Ca²⁺ removal and thapsigargin addition) and subsequent switch to Ca²⁺-containing buffer. 72 h before measurement, cells were infected with adenovirus encoding STIM1 (Ad-Stim1), a short hairpin to reduce STIM1 (Ad-shStim1) and controls expressing β-galactosidase (Ad-LacZ) and scrambled shRNA (Ad-shScr).

(B) Ca²⁺ entry in cardiomyocytes isolated from Stim1^−/− mice. Upon extracellular addition of Ca²⁺ (2 mM), store-operated Ca²⁺ entry occurred in control (n=53) cardiomyocytes, but not in Stim1^−/− cardiomyocytes (n=11).

Data are mean ± s.e.m. and * P < 0.05, ** P < 0.01, *** P < 0.005 in this and other figures.

(C) Microscopic detection (TIRF) of endogenous STIM1 in isolated NRCM. Immunofluorescent staining for STIM1 (red) and α-actinin (green) before and after Ca²⁺ depletion of the SR. Scale bar represents 5 μm.

We then analyzed whether SOCE likewise exists in adult rat CM (ARCM), and whether it is affected by prohypertrophic conditions. Because Stim1^−/− mice die perinatally²⁷, a knock-down strategy to reduce expression of Stim1 was chosen instead. We used a rat model of compensated cardiac hypertrophy due to pressure overload by abdominal aortic banding. Sham-operated rats served as controls. 24 days after surgery, both groups were infected with an adenovirus that carries a short hairpin RNA (shRNA) directed against Stim1, in addition to cDNA encoding the dsRed fluorescent protein (Ad-shStim1-dsRed), and ARCM were isolated 4 days later (at day 28) (Figure 2A). This strategy allows distinguishing unaltered from reduced Stim1 expression in CMs isolated from the same rat heart. Echocardiography was performed on all animals before adenovirus injection, confirming left ventricular hypertrophy with preserved ejection fraction in the pressure-overloaded group (Supplemental Table 1). Suitability of the adenoviral constructs was validated in NRCM (residual Stim1 mRNA in dsRed-positive CM was 10–15% of mean control value, see also Ref.¹⁴). The shStim1 sequence was selected from several sequences tested, which were all found to efficiently prevent hypertrophy of CM (data not shown). Marked differences between CMs from adult healthy or hypertrophic rats were observed in patch clamp recordings of sarcolemmal store-operated currents (I_SOC). As depicted in Figure 2B, thapsigargin induced a divalent cation current that was small in CM from sham-treated adult rats, but markedly increased in CM isolated from banded adult rats (Figure 2B, black tracings). In both cases, experimentally reduced STIM1 expression dramatically decreased this current (green tracings in Figure 2B). This is further supported by Indo-1 measurement of intracellular Ca²⁺ upon store-depletion. Whereas ARCM from sham-treated animals allowed only for a moderate entry of extracellular Ca²⁺ after SR depletion (not shown), a strong increase was observed in the analogous experiment with CM from banding-treated animals (Figure 2D, black tracing). Reduced expression of STIM1 potently diminished this response (Figure 2D, green tracing and column).

Figure 2. STIM1-dependent cation currents in adult cardiomyocytes.

(A)Schematic timescale and experimental strategy to analyze SOC currents in adult rat cardiomyocytes (ARCM). Cells were isolated 28 days after abdominal aortic constriction or sham treatment. The Ad-shStim1-dsRed vector was administered on day 24 after surgery. Isolated ARCM were assigned to the shStim or the control group (=non-infected) based on dsRed fluorescence, thus allowing to compare the effects of reduced versus normal Stim1 expression in a collective of cells from the same animal.

(B) Whole cell patch clamp recordings in ARCM before and after SR Ca²⁺ store depletion by thapsigargin (TG). Left, Recordings of Ba²⁺ current (I_SOC) in the presence or absence of thapsigargin-induced SR store depletion in cardiomyocytes from sham-treated rats. Cells in which STIM1 expression had been reduced by Ad-shStim1-dsRed (green symbols) are shown next to tracings of non-silenced cardiomyocytes (black symbols). Right, Same recordings conducted on ARCM isolated after pressure overload. Recordings were performed at −80 mV. Analyses are from 3 animals/group with >10 cells/animal.

(C) I_soc current-to-voltage relation after Ba²⁺ (asterisk) and Ba²⁺/thapsigargin perfusion as indicated in (A) in non-infected (black) or Ad-shStim1-dsRed infected (green) ARCM from pressure-overloaded hearts. Tracings display the average of 12 cells (banding) and 6 cells (banding shStim1).

(D) Fluorescence analysis of Ca²⁺ entry in hypertrophic ARCM. Cardiomyocytes were isolated from hypertrophic hearts as in (A), and loaded with the Ca²⁺-sensor Indo-1. Store-operated entry of Ca²⁺ was induced as described in Figure 1A, and was detected as increased Indo-1 emission ratio (F405/480). Left, representative tracings recorded from hypertrophic cardiomyocytes with or without silencing of Stim1. Right, quantitation from 3 independent experiments, with 3 animals/group and >10 cells analyzed/animal.

Caffeine-induced Ca²⁺ release, which allows to evaluate SR Ca²⁺ concentrations, suggested that the different currents of the four groups of cardiomyocytes did not originate from different SR Ca²⁺ loads (Supplemental Figure 3C). Together, these data demonstrate that STIM1 is crucially involved in thapsigargin-induced SOCE in adult cardiomyocytes.

STIM1 also promotes a current in the absence of drug-induced store depletion

Intriguingly, we also observed a current in the absence of drug-induced store depletion. Although marginal in CM from healthy adult rats, this current is clearly visible and enhanced in CM from hypertrophic hearts (Figure 2B, compare black tracings in section marked by asterisk). This current is completely abolished in shStim1-infected cells from the same CM population (Figure 2B, green tracing in this section), but did not affect excitation contraction coupling, as neither cellular Ca²⁺ transients nor cell shortening were altered in Ad-shStim1-dsRed infected cells (Supplemental Figure 3C and D). Both, the currents recorded in the absence or in the presence of thapsigargin share I_SOC features, such as the amplification in a divalent cation-free environment or the inhibition by SKF or by La³⁺ (supplemental Figure 3A and B). By contrast, differences became evident upon plotting the current-voltage relation that underlied these currents. Perfusion of hypertrophic ARCM with thapsigargin induced a double rectifying current (Figure 2C, black tracing). The SOC current was similarly permeable to barium as to calcium (Figure 2B and Supplemental Figure 3B).

In contrast to the above, the current-voltage curve recorded in the absence of thapsigargin is predominantly an inward rectifying current (Figure 2C, black tracing, see asterisk). With Ba²⁺ (Figure 2B) or a Na⁺-free Ca²⁺ solution (Supplemental Figure 3A), it developed slowly and reached steady state after ~2 min. This curve shape displays characteristics of gating by ORAI proteins (see²⁸), but the current did not show the pronounced selectivity for Ca²⁺ over Ba²⁺ that has been reported for ORAI1 but not for ORAI 2 and 3 ²⁹ (Supplemental Figure 3A).

Effects of STIM1 overexpression or silencing on the growth of isolated cardiomyocytes

Our finding that STIM1 deficiency reduces sarcolemmal Ca²⁺ fluxes prompted us to ask whether altered STIM1 expression affects cardiomyocyte growth and intracellular signaling, and whether this would involve store-operated channels. Indeed, NRCM overexpressing STIM1 were significantly larger than LacZ-overexpressing controls, as detected by α-actinin staining and by automated size detection (Figure 3A). In presence of SKF96365, cell sizes were comparable to LacZ controls, suggesting that STIM1 confers hypertrophy through an interaction with ORAI or TRP channel proteins. In line with STIM1 being an activator of the NFAT pathway, a reporter assay indicated that STIM1 overexpression enhanced NFAT activity (Figure 3B).

Figure 3. Role of STIM1 in cardiomyocyte growth and signaling.

(A) Effects of STIM1 overexpression on the growth of neonatal rat cardiomyocytes. Left, immunofluorescence analysis of NRCM infected with adenoviral vectors (MOI 10) encoding β-galactosidase (Ad-LacZ) or STIM1 (Ad-Stim1) and treated with the SOCE inhibitor SKF96365 or control (DMSO). Immunofluorescence analysis was carried out 48h after stimulation, using an antibody against α-actinin (green). Nuclei were stained with DAPI. Scale bar represents 5 μm. Right, quantitation of cardiomyocyte area.

(B) Determination of NFAT activity by luciferase reporter assay in NRCM that have been treated as in (A).

(C) Effects of Stim1 silencing on cardiomyocyte hypertrophy. Left, Western blot detecting STIM1 in cardiomyocytes 72h after infection with an adenoviral silencing vector (Ad-shStim1) or a control (Ad-shScr), and quantitative analysis of the results. Center, surface area determination of the cardiomyocytes. Right, ³H-leucine incorporation during 48 h of stimulation with phenylephrine (50 μM), three days after infection.

(D) Effects of STIM1-silencing on phenylephrine-induced NFAT activation and expression of atrial natriuretic factor (ANF) and modulatory calcineurin-interacting protein 1 (MCIP1).

(E) Membrane capacitance of cardiomyocytes as a measure of cell size. ARCM infected with Ad-shStim1 were isolated as depicted in Figure 2. n ≥6 for all groups.

Data are from ≥3 independent experiments with ≥3 replicates each, except that two independent experiments were performed for ³H-leucine detection in (C), ANF and MCIP mRNA determination in (D). >500 cells were analyzed in each independent experiment in (A) and (C).

Additional parameters were assessed to test for the functional impact of STIM1 on NFAT activation and hypertrophy. Reduction of Stim1 expression in NRCM (to approximately 25% of control, see Figure 3C left) partially prevented phenylephrine-induced hypertrophy (as determined by size or ³H-leucine integration, Figure 3C). Furthermore, NFAT activity and mRNAs encoding atrial natriuretic factor (ANF) or modulatory calcineurin interacting protein 1 (MCIP1) were reduced (Figure 3D). Finally, cellular capacitance as a means of cardiomyocyte size was higher in CM from pressure-overloaded hearts compared to sham, and Stim1 silencing (see also Figure 2A) resulted in significantly lower values (Figure 3E). These data indicate that STIM1 is both sufficient and necessary for the cardiomyocyte hypertrophic response.

Adeno-associated virus-mediated silencing of STIM1 in vivo prevents cardiac hypertrophy

We also asked whether endogenous STIM1 levels change under hypertrophic conditions. We found increased amounts of Stim1 mRNA and protein in NRCM 48 h after stimulation by endothelin 1 (ET1) or phenylephrine (PE) (Supplemental Figure 4A). Stim1 expression was, observed to be moderate in the adult heart supplemental Figure 4B), but was significantly upregulated in left ventricular myocardium from rats after pressure overload (Supplemental Figure 4C). To determine the effects of silencing Stim1 on cardiac hypertrophy in vivo, we generated recombinant cardiotropic adeno-associated viruses of serotype 9 (AAV9), allowing for cardiomyocyte-targeted RNAi against Stim1 (AAV9-shStim1) under control of the U6 promoter (Supplemental Figure 5). An AAV9 encoding an shRNA directed against luciferase served as a negative control (AAV9-shLuc). Vectors were injected into rats (n=9 per group, 5×10¹¹ genomes per animal) and 28 days later, abdominal aortic constriction was applied (Figure 4A). 8 out of 9 animals survived in both groups.

Figure 4. Silencing of STIM in a model of pressure-induced cardiac hypertrophy in vivo.

(A) Left, experimental strategy for the analysis of STIM1 silencing in living rats. Right, immunoblot analysis of STIM1 and GAPDH in lysates from AAV9-shLuc- and AAV9-shStim1-treated hearts. For quantitation, STIM1 data were normalized to the GAPDH signal in each lane. n=8 per group.

(B) Serial echocardiographic measurements of left ventricular anterior and posterior wall dimensions in AAV9-shStim1 or AAV9-shLuc treated rats. n=8/group.

(C) Left, postmortem determination of the ratio of left ventricle to body weight (LV/BW) in AAV9-shStim1 or AAV9-shLuc treated rats (n=8/group). Middle, average cardiomyocyte diameter in each group (> 20 cells/animal from 4 animals/group). Right, sirius red staining for interstitial fibrosis in sections of left ventricular tissue, and quantitative analysis from 5 animals per group.

(D) Detection and quantitation of NFAT in subcellular fractions by immunoblot analysis.

Consistent with the cardiotropic serotype of AAV9³⁰, cardiac expression of a AAV9-GFP control construct led to exclusive GFP detection in cardiomyocytes, opposed to endothelial cells or interstitial cells which we presume to be mainly fibroblasts (Supplemental Figure 5C). Furthermore, we determined whether our viral constructs would impact on lymphocyte infiltration or capilary density (rather than STIM1 activity in CM). For this, we analyzed tissue sections of rat hearts for the presence of CD8⁺ and CD45⁺ cells and for vessels stained positive for von-Willebrand-factor (vWF). Moderately enhanced infiltration by CD45⁺ lymphocytes and a decrease in the capillary density was observed in hypertrophic hearts, but irrespective of manipulated Stim1 expression (Supplemental Figure 6).

Reduced Stim1 expression, occurring after infection with AAV9-shStim1 (Figure 4A), significantly prevented cardiac hypertrophy, as delineated from serial echocardiographic measurements of LV wall thickness (Figure 4B and Supplemental Table 2) and the ratio of left ventricular weight to body weight (Figure 4C). Concordantly, histological analysis showed reduction of both cardiomyocyte size and myocardial fibrosis in AAV9-shStim1-treated rats (Figure 4C). None of these effects occurred in the control group. Moreover, Stim1-silenced rats exhibited a reduction of Ca²⁺ signaling, as nuclear translocation of NFATC3 was reduced in cardiomyocytes from Stim1-silenced rats (Figure 4D).

Discussion

The cardiac muscle responds to mechanical and humoral stress by hypertrophic growth of individual myocytes³¹. While some degree of cardiac hypertrophy serves to reduce wall stress and helps to compensate for increased load on the myocardium, sustained pro-hypertrophic signaling within cardiomyocytes is clearly detrimental and a major factor contributing to the progression into failure^{31, 32}. Cardiac hypertrophy is typically accompanied by the activation of Ca²⁺-dependent signaling pathways and the reinduction of a fetal gene expression program^{31, 33}.

Among the Ca²⁺-dependent signaling pathways that have been implicated in cardiac growth control, calmodulin-dependent activation of the serine-threonine phophatase calcineurin and subsequent NFAT translocation to the nucleus are particularly important³⁴. While disturbances of cardiomyocyte SR Ca²⁺ release and SR uptake (leading to e.g. an increase in diastolic Ca²⁺) clearly become dominant in more advanced disease stages³⁵, calcineurin activation occurs at early stages of cardiac hypertrophy, where excitation contraction coupling and SR Ca²⁺ load are still normal and diastolic Ca²⁺ concentrations are in the physiologic range⁸.

It is increasingly understood how global impairment of cardiomyocyte Ca²⁺ handling, as seen in advanced cardiac disease, elicits certain disease-related signaling pathways. By contrast, we know little about the mechanisms that drive the simultaneous activation of various Ca²⁺-dependent signaling pathways observed at early disease stages¹. In contrast to recent advances in understanding IP₃-mediated Ca²⁺ release from the nuclear envelope⁷, it remains unclear how the activation of the CaN-NFAT-axis in cardiac disease occurs¹.

In non-excitable cells, an important mechanism towards Ca²⁺ signaling involves Ca²⁺ release from the ER and subsequent influx of extracellular Ca²⁺ into the cytosol. The key protein involved in this store-operated Ca²⁺ entry (SOCE) is STIM1, which activates ORAI1, the pore-forming subunit of a Ca²⁺ release-activated Ca²⁺ channel (CRAC)^{13, 36, 37}. In contrast to non-excitable cells, the role of STIM1 in muscle cells is barely understood. Despite the early findings by Hunton et al.^{9, 10} that SOCE also occurs in cardiomyocytes, it remained unclear whether SOCE is causatively involved in cardiac hypertrophy and whether STIM1 ties SOCE to hypertrophy. Although two recent reports^{25, 26} could show that reduced expression of Stim1 interferes with the response of cardiomyocytes to prohypertrophic receptor agonists, the restriction to neonatal cardiomyocytes and drug-induced SOCE in these studies left unanswered whether SOCE in adult cardiomyocytes indeed triggers disease and whether this involves a Ca²⁺-sensing activity of STIM1.

We believe the most important aspects of our study are the identification of a sarcolemmal current in the absence of drug-induced SERCA inhibition, its stronger amplitude in hypertrophic cardiomyocytes, and its dependence on STIM1. Remarkably, a previous study on thapsigargin-induced currents in lymphocytes and Jurkat T cells also reported on a current in the absence of SERCA inhibition³⁸. Although that current was not the focus of this study, its inward rectificying characteristic shares similarity with the current we found to be STIM1-dependent in cardiomyocytes. The existence of such a current in lymphocytes justifies the presumption that it may likewise depend on STIM1.

A question that remains is whether the STIM1-dependent current in the absence of SERCA inhibition mirrors a true independence from SR calcium store depletion. Opposed to this, one may envision that store depletion also occurs under physiological conditions, but is obscured by STIM1-mediated refilling of the SR with calcium. The latter has similarly been proposed to occur in HeLa cells³⁹. By contrast, the observation that SR-based Ca²⁺ stores remain unchanged in cardiac hypertrophy⁴⁰ and our finding that silenced STIM1 expression did not alter the SR Ca²⁺ (Supplemental Figure 3C) argue against such a scenario. Evidence that STIM1 may indeed function in a store-independent manner comes from studies on arachidonic acid-stimulated Ca²⁺ signaling in HEK cells⁴¹. Although there are clear differences with respect to cell type (non-exitable vs. exitable) and receptor activation, in both cases, STIM1 appears to function independently of ER/SR Ca²⁺ store depletion.

To the best of our knowledge, our data are the first, which demonstrate a role for STIM1 in cardiomyocytes independent from drug-induced store depletion and suggest a critical role for STIM1 in the adult heart. Yet, several important questions remain: If STIM1 functions in the absence of Ca²⁺ depletion from the SR, then what are the upstream regulatory mechanisms that lead to its activation? Further, does STIM1 in this pathway interact exclusively with the recently identified Ca²⁺ release activated channels of the ORAI protein family, or are other channel proteins are involved?

In the years before ORAI1 was discovered, the store-operated Ca²⁺ channel was expected to be found in another protein family, termed transient receptor proteins type C (TRPC). TRPC channels are less selective for cations than ORAI1^{22, 23}, a fact which has been cited as evidence that the STIM1-operated channel is ORAI1 ¹⁹. However, several studies have shown interactions between STIM1 and TRPC proteins^17–20, which justifies the hypothesis that SOCE has more than one origin¹⁹. Indeed, aside from ORAI1, which was recently proposed to function in cardiac hypertrophy, analogous correlations were established for TRPC1, TRPC3 and TRPC6^{36, 42}. Given this, the currents we had measured in the presence and in the absence of SERCA inhibitors deserve a more detailed discussion. Both currents meet criteria of I_SOC currents, that is, their susceptibility to the channel-blocking drug SKF or to La³⁺ ions. As in case of the thapsigargin-dependent current, its inability to discriminate between Ba²⁺ and Ca²⁺, together with double rectification, are well consistent with currents mediated by TRPC channels, as reported by Yuan et al.¹⁶, Stiber et al.⁴³, Huang et al.¹⁶ or in studies that specifically addressed ion gating by TRPC1⁴² or TRPC3 and TRPC6³⁶. On the other hand, the identity of the channel that promoted a current in the absence of thapsigargin raises a new question. While this current appears to be mainly inward rectifiying and thus compliant with ORAI gating, it appears to lack the profound ion selectivity of this channel often stated for this protein family ⁴⁴. However, the extent to which ORAI1, ORAI2 and ORAI3 discriminate cations differs (in descending order) ²⁹. In addition, the cardiac expression of the three ORAI isoforms has yet to be determined and ORAI1 may be functionally replaced by either ORAI2 or ORAI3⁴⁵. While this supports the idea that the STIM1-dependent current we observe without SERCA inhibition is mediated by one or both of these channel proteins, we may also envisage the participation of heteromeric channels formed by ORAI and TRPC members, as suggested by Liao et al²⁴. Future studies should identify the channel that is activated by STIM1 in the absence of SR Ca²⁺ store depletion, test whether STIM1 oligomerization within the SR and its interaction with plasma membrane ORAI is intact under conditions of cardiac hypertrophy, but also investigate the role of TRPC channel proteins under such conditions. Interestingly, STIM1 was recently reported to inhibit LTCC in neurons and vascular smooth muscle cells ^{46, 47} and the authors of both studies speculate that this mechanism promotes the decision which Ca²⁺-signaling pathways are specifically activated. In our study, we inhibited the LTCC in order to resolve the role of STIM1 on store-operated channels and cardiac hypertrophy. However, since LTCC blockers are known to function anti-hypertrophic ⁴⁸, we have to assume that STIM1-mediated regulation of LTCC activity does not promote CM hypertrophy.

At present, it is unclear, whether additional Ca²⁺-dependent signaling mechanisms that have been implicated in CM hypertrophy involve STIM1. These include the direct coupling of the CaM/CaN-axis to LTCC-induced Ca²⁺ entry that has been described in neurons⁴⁹ and the Ca²⁺ and integrin–binding protein 1 (CIB1) identified as a prohypertrophic calcineurin-interacting protein⁵⁰. It remains to be seen if, and how, STIM1 contributes to these pathways. The answer to these crucial questions will eventually lead to a detailed picture of how STIM1 is activated and what protein it recruits at the plasmalemma to mediate NFAT-activation and thereby cardiomyocyte hypertrophy in the presence of the large fluctuations of intracellular Ca²⁺ that occur during EC coupling. Taken together, our data demonstrate an important role for STIM1 in the progression of cardiac hypertrophy and suggest a possible role in cardiac disease.