Cooperative and alternate functions for STIM1 and STIM2 in macrophage activation and in the context of inflammation

Calcium (Ca2+) signaling in immune cells, including macrophages, controls a wide range of effector functions that are critical for host defense and contribute to inflammation and autoimmune diseases. However, receptor-mediated Ca2+ responses consist of complex mechanisms that make it difficult to identify the pathogenesis and develop therapy. Previous studies have revealed the importance of the Ca2+ sensor STIM1 and store-operated Ca2+-entry (SOCE) for Fcγ-receptor activation and IgG-induced inflammation. Here, we identify the closely related STIM2 as mediator of cell migration and cytokine production downstream of GPCR and TLR4 activation in macrophages and show that mice lacking STIM2 are partially resistant to inflammatory responses in peritonitis and LPS-induced inflammation. Interestingly, STIM2 modulates the migratory behavior of macrophages independent from STIM1 and without a strict requirement for Ca2+ influx. While STIM2 also contributes in part to FcγR activation, the C5a-induced amplification of IgG-mediated phagocytosis is mainly dependent on STIM1. Blockade of STIM-related functions limits mortality in experimental models of AIHA and LPS-sepsis in normal mice. These results suggest benefits of Ca2+-inhibition for suppression of exacerbated immune reactions and illustrate the significance of alternate functions of STIM proteins in macrophage activation and in the context of innate immune inflammation.

Macrophage activation can assume a pathogenic role, which may necessitate activation via distinct receptors. The pathogenic relevance of the cross-talk between macrophage C5aR and activating FcgRs has been demonstrated in several animal models of disease, such as autoimmune hemolytic anemia (AIHA), anti-glomerular basement membrane nephritis, and immune complex (IC)-induced tissue injury [9,[14][15][16]. Excessive production of TNFa and other toxic cytokines downstream of TLR4 activation can cause septic shock [10,17]. Moreover, evidence on direct or indirect interactions between FcgRs and TLR4 has been provided [18,19]. Little is known about the pathways that directly impact C5aR, FcgR, and TLR signaling and their relevance in immunological diseases.
Ca 2þ is a crucial second messenger involved in signaling downstream of several classes of macrophage receptors, including C5aR, FcgRs, and TLRs [20][21][22][23]. Signaling cascades involving phospholipase C activation result in inositol 1,4,5triphosphate (InsP 3 )-mediated release of Ca 2þ from the endoplasmic reticulum (ER) through InsP 3 receptor channels yielding a rapid but transient increase of cytosolic Ca 2þ [24][25][26]. ER Ca 2þ store depletion can induce the opening of plasma membrane-expressed store-operated Ca 2þ (SOC) channels, also known as calcium release-activated Ca 2þ (CRAC) channels, resulting in more sustained Ca 2þ signals [27,28]. Such a mode of store-operated Ca 2þ entry (SOCE) is regulated by the ER-resident Ca 2þ sensor stromal interaction molecule 1 (STIM1) in many cells including mast cells, T and B cells, as well as platelets [29,30]. The closely related STIM2 has been also suggested to activate SOC influx, but appears less effective than STIM1 in T and B cells [31][32][33]. Furthermore, STIM2 may have an additional distinct role in regulating basal cytosolic and ER Ca 2þ concentrations [34,35]. The relevance of STIM2 for Ca 2þ -induced activation of macrophages, however, has not been addressed so far, and the effect of STIM proteins on the regulation of major macrophage effector functions and their contribution to disease pathogenesis is not fully understood.
In the present study, we investigated the requirement for STIM2 in macrophage activation and inflammation. Our data suggest a previously unidentified function of STIM2 for TLR-activated cytokine production and GPCR-mediated cell migration independent from STIM1. STIM2, like STIM1, also contributes to SOCE and phagocytosis, but only STIM1 is essential for C5aR-mediated amplification of FcgR activation in vitro and in a model of AIHA in vivo. Thus, STIM1 and STIM2 are critical but distinctly acting mediators of different macrophage effector functions and, as shown for septic shock and fatal hemolysis, inhibition of Ca 2þ signaling may be beneficial in the treatment of severe inflammation and autoimmune injury.

Results
Differential requirement of extracellular Ca 2þ for phagocytosis, chemotaxis, and TLR4-mediated activation of macrophages To assess the relative contribution of different sources of Ca 2þ in key aspects of macrophage biology, we used freshly isolated PMs from C57Bl/6 mice in the presence of the two Ca 2þ -specific chelators EGTA and Bapta/AM. Bapta/AM removes intracellular Ca 2þ , whereas EGTA blocks the influx of extracellular Ca 2þ . Incubation with Bapta/AM (50 mM) and EGTA (2 mM) had no effect on the viability (Fig. 1A) and adhesive capacity (Fig. 1B) of PM cells. We first investigated the effects of the two Ca 2þ chelators on FcgRdependent phagocytosis. PM cells were incubated with MRBCs opsonized with anti-MRBC IgG and measured for phagocytosis of MRBCs by light microscopy as described [36].
Macrophages showed strongly diminished IgG-mediated phagocytosis upon removal of either extracellular Ca 2þ (by EGTA) or intracellular Ca 2þ (by Bapta/AM) (Fig. 1C). Simultaneous application of EGTA and Bapta/AM completely abolished phagocytosis, providing evidence that FcgRmediated phagocytosis requires sustained mobilization of Ca 2þ from both the intracellular stores and the extracellular milieu. We also examined LPS-induced cytokine production. Strikingly, similar levels of TNFa and IL-6 were detected in the presence and absence of EGTA in cell-culture supernatant of PMs stimulated with LPS, whereas inhibition of intracellular Ca 2þ by Bapta/AM resulted in reduced secretion of TNFa and IL-6 ( Fig. 1D). Next, we tested the role of Ca 2þ for the migratory capacity of PMs in response to CCL2 and C5a in standard Transwell chemotaxis assays. Similar to TLR4-mediated activation and the expression of cytokines, inhibition of intracellular Ca 2þ (but not of Ca 2þ influx) yielded substantially decreased chemotaxis (Fig. 1E). Taken together, these results indicate that FcgR-mediated phagocytosis, but not TLR4-mediated activation or CCL2-/C5ainduced migration is dependent on extracellular Ca 2þ in macrophages.
STIM1 is essential for C5a-induced upregulation of FcgR-mediated phagocytosis but not for cytokine production and macrophage migration in vitro and in vivo We have previously shown that FcgR-mediated phagocytosis and IgG IC-inflammation are mediated by STIM1dependent mechanisms and that STIM1 is an essential regulator of SOCE in PMs [37]. The role of STIM1 in other Ca 2þ -dependent, but EGTA-insensitive (Ca 2þ influx-independent) responses of macrophages, however, has not been addressed by these previous studies. Here, we induced TLR4mediated cytokine production and agonist (CCL2/C5a)induced migration in STIM1-deficient PM cells in vitro. Similar to the EGTA results obtained before (see Fig. 1), neither LPS-induced secretion of TNFa and IL-6 ( Fig. 2A) nor CCL2/C5a-induced chemotaxis (Fig. 2B) was reduced in Stim1 À/À PMs compared with WT controls. Importantly, FcgR-mediated phagocytosis and its further enhancement by C5a were almost completely absent in Stim1 À/À macrophages (Fig. 2C), demonstrating that STIM1-although not necessary for chemotaxis and TLR4-mediated cytokine production-is essential for phagocytosis and its C5ainduced regulation.
To further corroborate the in vitro findings we subjected Stim1 À/À bone marrow chimeric mice to different in vivo models of LPS-induced sepsis, Thg-induced peritonitis, and AIHA. Immunoblot analysis and RT-qPCR confirmed the absence of STIM1 in PM preparations from the mutant chimeras, whereas the mRNA and protein were strongly  Normal cytokine release and chemotaxis, but defective IgG-induced and C5a-regulated phagocytosis in Stim1 À/À macrophages. (A) PM cells of Stim1 À/À and Stim +/+ chimeric mice were stimulated with 100 ng/mL LPS and analyzed at 24 h for the release of TNFa and IL-6 by ELISA. (B) PMs were stimulated with C5a and CCL2 for chemotaxis in Transwell migration assays, and migrated cells were counted under light microscopy. (C) PM cells were incubated with uncoated (w/o IgG) or IgG-coated MRBCs and percentage of phagocytosis was assessed (left panel). Phagocytosis of IgG-opsonized MRBCs by Stim1 +/+ PMs (indicated by the intermittent line) was increased in the presence of 50 ng/mL of C5a (IgG þ C5a; significance is shown in bold) in Stim1 +/+ but not Stim1 À/À cells (right panel). All results are expressed as mean AE SEM of 3-4 independent experiments (ns, not significant; **P < 0.01; ***P < 0.001). Stim1 À/À and Stim1 +/+ PM cells only differed significantly for IgG-induced and C5a-regulated phagocytosis. expressed in control PM (Fig. 3A, B). Furthermore, the mRNA levels of Stim2 and Orai isoforms (Orai1, Orai2, and Orai3) were almost unaltered in Stim1 À/À PM cells (Fig. 3B). A hallmark of the early response in LPS-induced acute inflammation is the massive production of proinflammatory cytokines, such as TNFa, IL-6, and IL-1b [38]. Serum levels of these cytokines were studied at 2 h after i.p. injection with LPS (10 mg/ kg). No difference was detected between Stim1 À/À chimeric mice and WT controls (Fig. 3C). In the second model, Thg-induced cell recruitment into the peritoneal cavity was examined on day 4 by counting macrophage numbers obtained after peritoneal lavage. Again, Stim1 À/À and WT chimeras each showed normal macrophage elicitation (Fig. 3D). In the model of AIHA [15,36], however, STIM1 deficiency provides reduction from anemia induced by pathogenic anti-MRBC 34-3C autoantibodies of IgG2a and IgG2b subclasses (Fig. 3E).
STIM2 contributes to macrophage SOCE, FcgRmediated Ca 2þ , and phagocytosis but is not required for C5a-induced regulation of FcgRdependent MRBC cell destruction and AIHA Analysis of STIM2-deficient mice has recently suggested that STIM2 may co-activate SOCE in T and B cells [32,33]. In contrast, the importance of STIM2 in relation to STIM1 for SOCE and FcgR-mediated Ca 2þ mobilization in macrophages remained unknown. Thus, we compared the roles of STIM1 and STIM2 for SOCE. SOC influx in Stim1 À/À , Stim2 À/À , and matched WT PMs was triggered by stimulating them with thapsigargin. TG-induced Ca 2þ store release was reduced approximately 30% in Stim2 À/À PMs compared with wild-type controls (Fig. 4A). Subsequent TG-dependent SOC influx, however, was substantially more decreased in Stim1 À/À compared to Stim2 À/À PMs. We also examined the capacity of STIM-deficient macrophages to respond to FcgR activation. Stim2 À/À PMs showed only a modest reduction in the FcgRIII-and FcgRIV-mediated elevation of Ca 2þ , whereas cytosolic Ca 2þ levels were much lower when STIM1 was absent (Fig. 4B). These results show that both STIM1 and (to a lesser extent) STIM2 contribute to SOCE and FcgR-mediated Ca 2þ responses and indicate that STIM2 is also required in part for proper control of store content in macrophages.
To determine the consequences of STIM2 deficiency for FcgR-mediated macrophage effector functions, we then studied erythrophagocytosis and its associated C5a-induced regulation both in vitro and in vivo. Stim2 À/À PMs showed reduced IgG-induced phagocytosis. This defect, however, could be rescued and phagocytosis efficiency could be increased by the addition of C5a (Fig. 5A). Interestingly and in contrast to the case of STIM1 deficiency, macrophage exposure to liver supernatant derived from AIHA-subjected B6 mice-previously reported to contain C5a [14]-also reinforced defective phagocytosis in Stim2 À/À PMs (Fig. 5B). This suggests that STIM2-while contributing to some extent to phagocytosis in vitro-may not play a critical role in AIHA. Indeed, STIM2-deficient mice showed no negative effect on 34-3C mAb-induced anemia in the clearance of MRBC as indicated by Ht levels (Fig. 5C) and the percentage of liver cells containing ingested erythrocytes (Fig. 5D).

STIM2 is selectively required for effective CCL2and C5a-induced macrophage migration and TLR4-mediated cytokine production in vitro and in vivo
The above results indicated that both STIM1 and STIM2 contribute to phagocytosis and Ca 2þ responses downstream of FcgRs. However, only STIM1 plays a critical role in C5ainduced regulation of FcgR-mediated phagocytosis and, as a consequence, in the pathogenesis of AIHA. On the other hand, STIM1 appears not to be involved in EGTAinsensitive macrophage Ca 2þ responses, as exemplified for CCL2/C5a-induced migration and TLR4-mediated cytokine production. To address the role of STIM2 in these functions, we analyzed STIM2-deficient macrophages in vitro and Stim2 À/À mice in vivo. In contrast to the results obtained for Stim1 À/À PMs, CCL2-and C5a-induced chemotaxis ( Fig. 6A) and TLR4-induced secretion of TNFa and IL-6 ( Fig. 6B) were reduced in Stim2 À/À PM cells. Furthermore, STIM2-deficient animals showed decreased macrophage recruitment in the model of Thg-induced peritonitis ( Fig. 6C) and reduced TNFa, IL-6, and IL-1b serum levels in LPS-induced inflammation (Fig. 6D).
In order to restrict STIM2 deficiency to the hematopoietic system, we transplanted lethally irradiated wild-type mice with bone marrow from Stim2 À/À or WT control mice and analyzed them after 10-16 weeks. We confirmed the absence of STIM2 in PM cells from Stim2 À/À chimeras by Western blot analysis and RT-qPCR (Fig. 7A, B). Transcript levels of STIM1 and Orai1, 2, 3 were almost unaltered in Stim2 À/À PM cells (Fig. 7B). Similar to STIM2-deficient mice (see Figs. 5, 6), Stim2 À/À chimeras showed no negative effect on 34-3C mAb-induced AIHA (Fig. 7C) but exhibited reduced Thg-peritonitis (Fig. 7D) and LPS inflammation (Fig. 7E). Moreover, Stim2 À/À chimeras injected with a lethal dose of LPS showed improved survival (Fig. 7F). The results suggest a role of STIM2 in hematopoietic cells in the control of cell migration and cytokine production that appears of relevance in severe inflammation.
Silencing of STIM proteins by RNA interference confirm the STIM-isoform specific role of STIM2 for EGTA-insensitive Ca 2 þ responses in macrophage activation To further strengthen our findings on alternate functions of STIM proteins in calcium activation, we silenced STIM1 and STIM2 in RAW 264.7 macrophages by STIM1 and STIM2 shRNA expression. The obtained KD cells exhibited strongly decreased STIM mRNA and protein levels with no compensatory up-or down-regulation of STIM1 and STIM2 in the absence of the other STIM isoform (Supplemental Fig. S1). In line with the effect of EGTA-or Bapta/AM-mediated Ca 2þ depletion/chelation in PM cells (Fig. 1), phagocytosis, but not CCL2/C5a-induced migration is dependent on extracellular Ca 2þ in RAW 264.7 macrophages (Supplemental Fig. S2).
SOCE was detectable in mock transfected RAW 264.7 cells and KD of STIM1 caused a more reduced influx of extracellular Ca 2þ as compared with Stim2 KD cells (Fig. 8A). Importantly, the TG-induced Ca 2þ ER store release was reduced in Stim2 KD but not Stim1 KD or mocktransfected cells (Fig. 8A). These findings reveal that STIM1 is most critical for SOCE and confirm the additional function of STIM2 in regulating Ca 2þ store content. We also evaluated the effects of silenced STIM expression in phagocytosis and chemotaxis. Comparable to the situation of Stim1 À/À and Stim2 À/À PMs, C5a-induced regulation of FcgR-mediated phagocytosis is defective in Stim1 KD cells  The results are expressed as mean AE SEM of three to four independent experiments (*P < 0.05; **P < 0.01). Note the significantly induced phagocytosis by C5acontaining AIHA LS in Stim2 À/À but not Stim1 À/À PMs. (C, D) Experimental AIHA was induced by 150 mg of the anti-MRBC 34-3C IgG2a mAb in Stim2 À/À and Stim2 +/+ mice. The results are shown as the mean AE SEM of (C) hematocrit and (D) liver phagocytosis at days 0 and 2 from 5 mice per group (ns, not significant).  ( Fig. 8B), whereas chemotactic migration induced by C5a (Fig. 8C) and CCL2 (Fig. 8D) was reduced in Stim2 KD cells, confirming STIM protein selectivity for distinct Ca 2þtriggered macrophage effector responses.
The Ca 2þ flux inhibitor BTP2 further affects STIM2-regulated functions independent from Orai1 We next tested whether the inhibitor BTP2, previously reported to inhibit Ca 2þ entry in T cells [39,40], blocks STIM1-mediated SOCE and maybe other Ca 2þ signals in macrophages. Interestingly, we found that BTP2 not only reduced SOCE but also abrogated the STIM2-sensitive part of TG-induced Ca 2þ ER store release in RAW 264.7 cells (Supplemental Fig. S3A). Moreover, IgG-induced phagocytosis (Supplemental Fig. S3B), C5a-elicited chemotaxis (Supplemental Fig. S3C) and LPS-induced secretion of TNFa (Supplemental Fig. S3D) were suppressed by BTP2. The inhibitory effect of BTP2 on cytokine production was dependent on the presence of STIM2. Silencing of STIM2 and treatment with BTP2 each reduced LPS-induced TNFa (Supplemental Fig. S3E) and NF-kB activity (Supplemental Fig. S3F) in RAW 264.7 cells which, however, was not synergistic when Stim2 KD cells were treated with BTP2. We also observed that Orai1 was not required for STIM2regulated function. In contrast to Stim2 À/À PMs and Stim2 KD RAW 264.7 cells, LPS-induced secretion of TNFa was neither reduced in Orai1 KD cells (Supplemental Fig. 4A, B) nor in Orai1 À/À PMs (Supplemental Fig. S4C, D). BTP2 inhibition of TNFa was equally effective in Orai1-sufficient and Orai1-deficient PM cells (Supplemental Fig. S4D).

BTP2 limits lethality in AIHA and LPS-induced sepsis
To examine the therapeutic potential of Ca 2þ -inhibition for suppression of exacerbated immune responses in vivo, we finally tested BTP2 in sublethal and lethal forms of AIHA and sepsis. In the model of AIHA, C57Bl/6 mice were injected with 150 and 300 mg of the pathogenic 34-3C IgG2a and received increasing BTP2 concentrations (5-30 mg/kg per os) or carrier along with 34-3C mAb challenge. Similar to the patterns observed in STIM1-deficiency [37], mice treated with 30 mg/kg of BTP2 displayed significant reduction/ protection with respect to Ht levels and lethality (Fig. 9A, B). Moreover, BTP2 not only decreased the transient drop in Ht in a dose-dependent manner, but also reduced in vivo erythrophagocytosis (Fig. 9C). Macrophage recruitment in Thg-induced peritonitis (15.52 AE 1.68 Â 10 6 cells, n ¼ 8) was also found to be significantly decreased after BTP2 treatment (8.44 AE 1.93 Â 10 6 cells, n ¼ 9, P ¼ 0.0023). These data suggest that BTP2 can interfere with systemic inflammatory responses that involve activation of STIM1 and STIM2. So, we next evaluated whether BTP2 also limits mortality in LPS-induced fatal sepsis. Administration of 15 mg/kg of BTP2 markedly reduced the elevated TNFa, IL-6, and IL-1b cytokine response after 10 mg/kg LPS challenge (Fig. 9D) and led to significant improvement of survival in mice treated with a lethal dose of LPS (30 mg/kg i.p.) (Fig. 9E). Together, the results show that BTP2 is effective in reducing exacerbated lethal responses in AIHA as well as sepsis.

Discussion
STIM1 is a crucial activator of store-operated Ca 2þ entry and the function of CRAC channels in immune cells [29,41]. In the absence of STIM1, T, and B cells showed defective SOCE in response to TCR and BCR stimulation [32,33]. STIM1mediated SOCE is also a central mechanism of Ca 2þ entry downstream of FceRI and FcgR activation in mast cells and macrophages [37,42]. STIM2, a related homolog of STIM1, may additionally activate SOCE [43]. However, the loss of STIM2 in T and B cells had a smaller negative effect on SOC influx than STIM1 [32,33]. STIM2 is also suggested to act as feedback regulator that stabilizes ER Ca 2þ levels [34], while STIM1 seems to play a less critical but variable role [34,37,42].
Here, we have examined the role of STIM2 in relation to STIM1 in various Ca 2þ -dependent effector functions of macrophages. We used STIM protein-deficient mice, RNAmediated interference, and Ca 2þ chelation approaches to demonstrate that STIM2 is a critical regulator of GPCRmediated cell migration and TLR4-dependent cytokine production, two macrophage responses shown here to be compromised by inhibition of intracellular Ca 2þ but not to depend on outside Ca 2þ , or STIM1. We also noted reduced Ca 2þ release from intracellular stores in the absence of STIM2. This was also reflected by Stim2 KD cells. No alteration in the expression levels of STIM1 was detected in the absence of STIM2 (Supplemental Fig. S1), suggesting distinct requirements for STIM1 and STIM2 in the regulation of certain Ca 2þ signals. Although further investigation is needed to understand the molecular basis of the functional difference between STIM1 and STIM2, our findings clearly indicate that STIM2 is the essential STIMprotein in triggering chemotactic cell migration and TLR4activation, two processes that also contribute to inflammatory reactions in mouse models of peritonitis and LPSinduced sepsis. It is important to note, however, that it is at present not clear whether macrophages alone or together with other cell types, most notably neutrophils, mediate the STIM2-specific response in vivo.
Despite identification of STIM1 as a signal mediator of FcgR-mediated functions, the mechanisms by which Ca 2þ is mobilized and controls phagocytosis are still not fully resolved [37]. For example, residual FcgR-induced Ca 2þ influx previously detected in STIM1-deficient macrophages might be explained by STIM2-regulated SOCE [44]. Indeed, STIM2-deficient macrophages showed reduced FcgR-dependent Ca 2þ influx as well as SOCE. However, these defects were more pronounced in STIM1 as compared to STIM2 deficiency, which is similar to previous analyses of STIM1and STIM2-deficient T and B cells [32,33,45]. In addition, FcgR-mediated phagocytosis, which depends on the availability of Ca 2þ from both the outside and internal stores, is almost abolished or reduced in the absence of STIM1 and STIM2, respectively. These data indicate a cooperative function of STIM1 and STIM2 for efficient SOCE and FcgR-mediated phagocytosis in macrophages.
IgG autoantibody-mediated phagocytosis contributes many autoimmune conditions and is specifically causal in AIHA. In line with the defects in FcgR-mediated Ca 2þ -influx and phagocytosis, Stim1 À/À chimeric mice were significantly protected from IgG-induced elimination of red blood cells in anemia (Fig. 3) [37]. Surprisingly however, IgG-induced in vivo erythrophagocytosis was STIM2-independent, contrasting the situation in vitro. A major difference might be the contribution of complement receptor signals. Recently, it was reported that the production of C5a in the liver positively regulates FcgRs and phagocytosis in AIHA autoimmune disease [14]. We here found that the phagocytosis promoting effect of C5a-either in recombinant form or used as biological fluid from anemic mice-strictly depends on the presence of STIM1, whereas C5a was sufficient to counteract the basal defects of phagocytosis in Stim2 À/À macrophages. Thus, while both STIM1 and STIM2 mediate phagocytosis in response to FcgR-activated Ca 2þ increase in vitro, the results of FcgR activation in conjunction with C5aR suggest STIM1 as the main signal mediator of the regulatory loop of C5a and FcgRs in AIHA in vivo.
BTP2 is considered a calcium influx inhibitor affecting the function of STIM1 and several different calcium channels, especially CRAC and transient receptor potential channels, in various cell types [39][40][41][46][47][48][49]. Moreover, BTP2 has been successfully used to reduce exacerbated immune responses in animal models of allergy [40], graft versus host disease [39], and vascular inflammation [48]. These data combined with our finding that BTP2 is a potent inhibitor of FcgR-induced activation and SOCE in macrophages suggest that STIM1 may be inhibited by BTP2. In accordance, BTP2 prevents lethality in severe AIHA. Interestingly however, we also observed significant reduction of migration and Ca 2þ ER store release in BTP2-treated cells, indicating an immunosuppressive function of BTP2 that may affect STIM1 as well as STIM2regulated functions. As shown for LPS-induced NF-kB activation and TNFa production, the inhibitory effect of BTP2 requires the presence of STIM2, indicating that BTP2 can influence STIM proteins (in this case STIM2) most likely through direct inhibition.
Here, we identified STIM2 as an important mediator of LPS-induced cytokine production in macrophages. In the absence of STIM2, the increase of circulating TNFa and other cytokines is reduced in LPS-challenged mice. Low serum cytokine levels are also seen in BTP2-treated mice. Furthermore, BTP2 resulted in enhanced survival of mice receiving a high dose of LPS. The selective absence of STIM2 in hematopoietic cells also results in an improved survival after LPS challenge. The results concerning chemotaxis of macrophages revealed that STIM2 is required for CCL2 and C5a-dependent migration in vitro and Thg-induced macrophage recruitment in vivo. Since these reactions are not largely dependent on STIM1 but are reduced in BTP2treated mice, STIM2 may be a possible target through which BTP2 suppresses cell migration and provides protection against cytokine-mediated effects in sepsis.
Our work extends previous reports that describe a dominant contribution of STIM1 in SOCE-mediated immune cell activation [32,33,37,42]. The results presented here provide evidence that STIM2 serves as critical and selective regulator of TLR4-activated cytokine production and GPCR-mediated cell migration. We also note that loss of STIM2 correlates with reduced functions of SOCE and FcgRmediated phagocytosis, indicating that STIM2 may contribute to some degree to cell destructive events in autoimmune diseases. In the experimental model of AIHA, however, autoantibody-induced cellular destruction occurs via multiple FcgR and C5aR signals favoring a dominant contribution of STIM1. The pyrazole derivative BTP2 is effective to block several cell responses related to STIM activation and this may account for the improved capacity of mice to withstand lethal challenges with hemolysis-inducing antibodies and LPS. In summary, our study suggests the potential of Ca 2þ -inhibition for the treatment of septic inflammation and autoimmune injury and identifies a STIM-isoform-specific role for STIM2 in certain macrophage effector responses that do not strictly rely on extracellular Ca 2þ influx. Future work has to elucidate the molecular basis of the alternate and cooperative functions of the STIM1/2 isoforms. For example, it remains to be investigated whether structural differences of STIM proteins in the interaction with plasma membrane Ca 2þ channels and/or ER-resident Ca 2þ pumps determine their distinct behavior in macrophages and maybe other innate immune effector cells.

Macrophage preparation and characterization
Peritoneal macrophages (PMs) were collected after flushing out the peritoneal cavity of mice with RPMI, without any supplements. Cells were washed twice and suspended in RPM1 1640 medium containing 10% FCS. The PM cells were allowed to adhere for 4 h at a density of 3 Â 10 6 cells/ well of 6-well plate cell culture dishes (Costar, Munich, Germany), followed by the removal of nonadherent cells, and were then used in various macrophage activation assays as described below. Stim1 À/À and Stim2 À/À F4/80-and Mac1-positive (> 95%) PM were microscopically indistinguishable from wild-type controls, exhibited unaltered mRNA expression of Orai1, Orai2, and Orai3 (Figs. 3B, 7B) and showed normal adhesion and IgG-MRBC rosette formation (data not shown).

Macrophage Ca 2þ depletion/chelation and STIM inhibition
PMs or RAW264.7 cells were incubated with Bapta/AM (Calbiochem, Darmstadt, Germany) at 50 mM for 30 min. Cells were subsequently washed and given 30 min as esterification time. This was followed by two final washing steps. EGTA (Sigma-Aldrich, Munich, Germany) was applied in assay medium at 2 mM immediately prior to cell stimulation. The bistrifluoromethyl pyrazole (BTP) derivative BTP2 (Sigma-Aldrich) was used at 20 mM for 1 h prior to stimulation.

Analysis of NF-kB activity in Stim2 KD cells
Stim2 KD and mock control cells 5 Â 10 5 were transfected with 1 mg of the pGL4.32 [luc2P/NF-kB-RE/Hygro] and 0.1 mg of the reference Renilla luciferase plasmid, which was used as a transfection control. The cells were recovered after 24 h, cultured for 24 h in RPM1 1640 medium containing 1% FCS. and treated with 100 ng/mL LPS for 3 h or left untreated. To measure luciferase activity we used the Dual-Luciferase reporter assay system (Promega). The firefly luciferase activity was normalized to Renilla luciferase activity and the activity of stimulated cells to the one of untreated cells to yield the relative promoter activity driven by NF-kB.

Functional analysis of macrophage effector responses
IgG-induced phagocytosis Freshly isolated mouse RBCs (MRBCs) from C57Bl/6 mice were washed and processed for IgG opsonization. Hereby, 100 mL of MRBCs were incubated with an equal volume of PBS supplemented with 50 mg/mL anti-MRBC IgG2a for 1 h. MRBCs were then washed and suspended at the original volume. PM cells were exposed to a 4% MRBC solution in RPMI 1640 medium/10% FCS for 3 h. In some experiments, PM cells were additionally incubated with C5a (50 ng/mL) or AIHA-derived liver supernatants. Non-ingested RBCs were removed by hypotonic lysis, PMs were washed with PBS, fixed with 4% paraformaldehyde, stained, and phagocytosis was determined by light microscopy. A slightly modified assay was used for the RAW264.7 macrophage cell line, where cells were first allowed to adhere overnight in RPMI 1640 medium/1% FCS at a density of 1.5 Â 10 5 cells/well before washing and exposure to a 4% MRBC buffer in RPMI 1640 medium/10% FCS for 90 min.
C5a/CCL2-induced chemotactic migration PMs or RAW264.7 macrophages 5 Â 10 5 in 100 ml RPMI 1640 medium/0.1% BSA (fatty acid free) were placed into the insert of a Transwell chemotaxis chamber (8-mm pore diameter; Greiner, Frickenhausen, Germany), and the bottom well was filled with 500 ml of RPMI 1640 medium/ 0.1% BSA (fatty acid free) or the same medium supplemented with C5a (50 ng/mL) or CCL2 (100 ng/mL). Inserts were transferred to the lower chambers and incubated at 378C for 4 h. For quantification of chemotaxis, Transwell filters were fixed in methanol, stained with May-Gr€ unwald-Giemsa. Cells on the upper side of the filter were removed and filters were mounted on a glass slide. Migrated macrophages that attached to the filter were counted by light microscopy as described [52].

Experimental sepsis
Mice were injected i.p. with a dose of 10 mg/kg LPS/5% FBS (Sigma-Aldrich). Two hours post injection, blood was collected from the retro-orbital plexus. Serum was collected after blood was left at RT for 45 min and spun at 3000 rpm for 10 min. Serum cytokine levels were assayed by TNFa-, IL-6-, and IL-1b-specific ELISA kits (R&D Systems). In BTP2-inhibition experiments, C57Bl/6 mice received single doses of 5 and 15 mg/kg BTP2 i.p. or vehicle control 1 h before LPS challenge (10 and 30 mg/kg), survival was monitored, and serum cytokines were determined by ELISA.

Thg-induced peritonitis
Peritoneal recruitment of leukocytes was induced using Thg medium (PBS/4% brewer thioglycollate; BD-BBL, Heidelberg, Germany). Thg medium was initially produced at 8%, left to mature for 4 weeks, and diluted to 4% with 2Â PBS immediately before i.p. injection. In experiments evaluating the effect of BTP2, mice received together with Thg a daily dose of 30 mg/kg BTP2 or vehicle control per osintragastric, via gavage for 3 days. On day 4 post Thg injection, mice were killed, and the peritoneal cavity was lavaged two times with 5 mL of PBS/5mM EDTA. Total cell count of collected fluids was determined in a hemocytometer (Neubauer Z€ ahlkammer, Gehrden, Germany). For quantification of macrophage influx, differential cell counts were performed on cytospins (10 min at 50 Â g).
Experimental AIHA Mouse anti-MRBC IgG2a monoclonal autoantibody 34-3C and its IgG2b genetic variant [15,55] were purified from tissue-culture supernatants by protein G affinity chromatography. Mice were injected i.p. with 150-300 mg of the pathogenic mAbs, survival was monitored, and daily hematocrit (Ht) was determined with heparinized microhematocrit capillary tubes using blood samples obtained from the retro-orbital plexus. In BTP2-inhibition experiments, mice further received daily doses of 5, 15, and 30 mg/ kg BTP2 or vehicle control per os for 3 days. H&E-stained formalin-fixed sections of liver were prepared from mice killed at days 0 and 2 after 34-3C mAb treatment and examined for histopathologic changes. To produce liver supernatants (from day 2 of AIHA), livers were perfused, excised, minced into small pieces, and treated with PBS þ 0.025% Collagenase IV (Sigma-Aldrich). This was followed by passages through a stainless steel gauze (mesh, 500 and 100 mm), so that single cell suspensions are obtained. Hepatocytes and the nonparenchymal cell fraction were removed by centrifugation and the obtained supernatant was stored at À80 8C for later use as described [14].

Statistics
Statistical analysis was performed using the Prism 5 statistical software package (GraphPad Software). Comparisons between groups were analyzed with 2-tailed Student's t-test. Differences in means among multiple groups were analyzed using one-way ANOVA with the Tukey's post-test.

Supporting Information
Additional supporting information may be found in the online version of this article at the publisher's web-site. Figure S1. Generation of macrophage Stim1 and Stim2 KD cells. Figure S2. Different effects of intracellular vs. extracellular Ca 2þ chelation/inhibition on RAW264.7 cell mediated functions. Figure S3. Defective TG-induced Ca 2þ ER store release, SOCE, phagocytosis, chemotaxis, and TLR4 activation in BTP2-treated RAW264.7 macrophages. Figure S4. LPS-induced TNFa release occurs normally in Orai1 KD cells and Orai1 -/-PM.