Junctate boosts phagocytosis by recruiting endoplasmic reticulum Ca²⁺ stores near phagosomes

Phagocytosis, the engulfment of foreign particles into a membrane-enclosed vacuole known as the phagosome, is a fundamental cellular process that is essential for bacterial killing and antigen presentation by innate immune cells. The phagocytic process comprises two phases, an ingestion phase that is characterized by the recognition of the particle and its actin-driven engulfment into a new membrane-derived organelle, as well as a maturation phase during which the phagosome acquires oxidative and lytic properties through fusion with endosomes and lysosomes (Aderem and Underhill, 1999). The kinetics and efficiency of phagosome maturation is promoted by both global as well as localized Ca²⁺ elevations occurring during phagocytosis that drive the shedding of the actin coat (Bengtsson et al., 1993), the assembly and activation of the phagocytic NAPDH oxidase (Dewitt et al., 2003), and the fusion with lysosomes (Jaconi et al., 1990), reviewed by Nunes and Demaurex (2010). Global Ca²⁺ elevations occur during phagocytosis as a result of phagocytic receptor ligation and phospholipase-C- or phospholipase-D-driven store-operated Ca²⁺ entry (SOCE) (Nunes and Demaurex, 2010). Localized periphagosomal Ca²⁺ elevations can be generated in two ways – either by the opening of Ca²⁺ release channels on intracellular Ca²⁺ stores located in the vicinity of phagosomes or by the opening of Ca²⁺ channels present on phagosomes through mechanisms that have only recently begun to be uncovered. We have previously shown that the transmembrane endoplasmic reticulum (ER) Ca²⁺ sensor stromal interaction molecule 1 (STIM1) recruits ER cisternae to phagosomes and promotes the opening of phagosomal Ca²⁺ channels (Nunes et al., 2012). STIM1 mediates the ubiquitous SOCE mechanism (Liou et al., 2005; Roos et al., 2005; Zhang et al., 2005) by acting as an intracellular ligand for plasma membrane Ca²⁺-permeable channels of the Orai (Feske et al., 2006; Vig et al., 2006; Zhang et al., 2006) and transient receptor potential canonical (TRPC) families (Huang et al., 2006; reviewed in Hogan et al., 2010), together referred to as store-operated Ca²⁺ (SOCE) channels. Upon ER Ca²⁺ depletion, STIM1 oligomerizes and accumulates in cortical contact sites between the ER and plasma membrane (Luik et al., 2008; Lur et al., 2009; Orci et al., 2009; Wu et al., 2006), where its channel activating domain (CAD) directly interacts with Orai1 to promote channel opening (Luik et al., 2006; Park et al., 2009; Xu et al., 2006; Zhou et al., 2010). Three independent synergistic mechanisms promote STIM1 localization at ER–plasma-membrane junctions: (1) STIM1 binding to phosphoinositides through the exposed polylysine C-terminal tail (Walsh et al., 2010), (2) binding of the STIM1 cytoplasmic CAD domain to Orai1 (Huang et al., 2006; Muik et al., 2009; Park et al., 2009; Yuan et al., 2009), and (3) Ca²⁺-regulated binding of STIM1 through its luminal domain to junctate (Srikanth et al., 2012)].

Junctate is a ubiquitously expressed 33-kDa single-pass transmembrane ER protein bearing a luminal EF-hand motif, generated by alternative splicing of the gene encoding junctin and aspartate-β-hydroxylase (ASPH isoform 8) (Dinchuk et al., 2000; Hong et al., 2001; Treves et al., 2000). Junctate and STIM1 interact through their luminal domains, thereby facilitating STIM1 recruitment to membrane contact sites (Srikanth et al., 2012), and interacts with the inositol 1,4,5 trisphosphate receptor (InsP3R) and with TRPC channels through its cytosolic N-terminus (Treves et al., 2004, 2010). Junctate also interacts with sarco-endoplasmic reticulum Ca²⁺ ATPase 2a (SERCA2a) in cardiomyocytes to regulate sarcoplasmic reticulum Ca²⁺ cycling (Kwon and Kim, 2009), and mice overexpressing junctate develop myocardial hypertrophy (Hong et al., 2008), highlighting the importance of junctate in Ca²⁺ homeostasis. Junctate populates plasma membrane clusters in resting cells, indicating that it is a structural component of ER–plasma-membrane contact sites, and junctate overexpression elongates and stabilizes ER–plasma–membrane junctions and increases the amount of Ca²⁺ released from intracellular Ca²⁺ stores (Treves et al., 2004, 2010). The N-terminus of junctate is devoid of lipid-binding motifs but is required for junctate accumulation in plasma membrane clusters, suggesting that junctate provides an alternative mechanism of STIM1 recruitment that is independent of phosphoinositides and Orai1(Srikanth et al., 2012). Such a mechanism might be important for the targeting of STIM1 to membranes that have distinct phosphoinositide compositions, or importantly, for the delivery of Ca²⁺ stores to target membranes independently of STIM1.

Phagosomes, although derived from the plasma membrane, have a very distinct lipid and protein composition because extensive sorting occurs during phagosome engulfment. In a previous study, we detected a significant number of ER–phagosome junctions, as well as residual Ca²⁺ hotspots (also called microdomains), in primary neutrophils as well as in mouse embryonic fibroblasts (MEFs) that had been engineered to be phagocytic and isolated from mice in which STIM1 had been genetically ablated. These results suggest that proteins other than STIM1 recruit ER cisternae and mediate localized Ca²⁺ signaling near phagosomes. We therefore postulated that junctate cooperates with STIM1 to recruit ER cisternae to phagosomes, thereby providing an alternative mechanism for generating periphagosomal Ca²⁺ hotspots. We tested this hypothesis by expressing junctate in fibroblasts, which had been rendered phagocytic through expression of FcγRIIA receptors (also known as FCGR2A), from STIM1-knockout mice. Our data show that junctate recruits ER Ca²⁺ stores to phagosomes and promotes localized Ca²⁺ elevations that sustain high-efficiency phagocytosis.

To assess whether junctate is recruited to phagosomes, we quantified the percentage of phagosomes that were decorated by yellow fluorescent protein (YFP)–junctate, by green fluorescent protein (GFP)–KDEL as a control ER marker, or by mCherry–STIM1, which we have previously shown to be recruited to phagosomes (Nunes et al., 2012). The proteins were expressed in wild-type, Stim1^−/− and Stim1^−/−; Stim2^−/− MEFs that had been rendered phagocytic through co-expression with FcγRIIA–c-Myc receptors, and cells were then exposed to IgG-opsonized red blood cells (RBCs). To synchronize phagocytosis, RBCs were centrifuged onto cells that had been seeded on coverslips, and cells were allowed to phagocytose for 10 min before fixation. Coverslips were then immunostained to reveal c-Myc tagged Fc receptors, and imaged by using confocal microscopy. Quantification through visual inspection of confocal z-stacks revealed that both GFP–KDEL and YFP–junctate fluorescent clusters (or puncta) were observed around ∼40% of phagosomes found in wild-type cells (Fig. 1A). In Stim1^−/− cells, no discernible periphagosomal GFP–KDEL clusters could be detected, whereas YFP–junctate and mCherry–STIM1 clusters decorated 31% and 57% of phagosomes, respectively (Fig. 1B). When the proteins were expressed in Stim1^−/−; Stim2^−/− cells, YFP–junctate and mCherry–STIM1 clusters decorated 57% and 68% of phagosomes compared with 16% of phagosomes for GFP–KDEL (Fig. 1C). The absence of GFP–KDEL clusters in Stim1^−/− but not Stim1^−/−; Stim2^−/− cells might reflect the difficulty to identify clusters owing to the low and more finely reticular expression pattern of the ER marker in Stim1^−/− cells (compare Fig. 1B left panel and Fig. 1C lower left panel). Alternatively, an altered nature of residual contact sites in cells expressing only STIM2 could account for this difference. Interestingly, when co-expressed in Stim1^−/− MEFs, YFP–junctate and STIM1–mCherry both localized in the same periphagosomal clusters (Fig. 1D). These data indicate that junctate is recruited to phagosomes independently of STIM proteins but might interact with STIM1 at ER–phagosome contact sites.

To establish whether junctate recruits ER cisternae to the vicinity of phagosomes and/or alters the morphology of ER–phagosome-membrane contact sites, we quantified the extent of ER cisternae recruitment to phagosomes by using electron microscopy in 50-nm slices. To normalize for the cross-sectional sampling, junctional ER was quantified as the average contact length divided by the perimeter of the phagosome (Fig. 2, arrows). Contacts were defined as the contiguous length of an ER cisterna that discernibly remained ≤30 nm from the membrane of a phagosome. These juxtaposed ER structures were detected in 23–29% of phagosomes forming in Stim1^−/− and wild-type phagocytic MEFs, regardless of junctate expression (Fig. 2), and the length of the recruited ER structures averaged ∼250 nm (268±51 nm for YFP–junctate and 231±58 nm for GFP–KDEL, corresponding to 2.84±0.54% and 2.45±0.62% of the phagosome perimeter, n=20 and 28, respectively; mean±s.e.m.; Fig. 2). By contrast, YFP–junctate expression doubled the length of the recruited cisternae in Stim1^−/− cells (231±36 nm vs 115±19 nm for GFP–KDEL, corresponding to 2.45±0.38% vs 1.22±0.20% of the phagosome perimeter, n=24 for each; Fig. 2). Qualitative differences were also apparent as the juxtaposed ER cisternae were thinner in Stim1^−/− cells expressing YFP–junctate (Fig. 2, arrows). These data indicate that junctate, when expressed in the absence of STIM1, can promote the elongation of periphagosomal ER cisternae.

Junctate has been previously reported to populate STIM1–Orai1 membrane contact sites and to interact with TRPC channels, suggesting that this protein might modulate the activity of Ca²⁺ entry channels. To test this hypothesis, we measured Ca²⁺ entry rates following store depletion with the SERCA inhibitor thapsigargin in cells expressing either YFP–junctate, GFP–KDEL or mCherry–STIM1 as controls. Junctate expression had no impact on Ca²⁺ entry rates in wild-type or Stim1^−/− MEFs, but surprisingly, junctate increased Ca²⁺ entry rates in Stim1^−/−; Stim2^−/− MEFs by twofold. In comparison, STIM1 re-expression in Stim1^−/− and Stim1^−/−; Stim2^−/− MEFs increased Ca²⁺ entry rates by fivefold and eightfold, respectively, to values near to 70% of wild-type levels (Fig. 3A,C). The small levels of Ca²⁺ entry mediated by junctate alone, in Stim1^−/−; Stim2^−/− MEFs, was abolished in the presence of the non-specific Ca²⁺ channel inhibitor lanthanum chloride (La³⁺), similar to the STIM1-mediated entry, indicating that the source of the signal was extracellular (Fig. 3A,C). In addition, no significant change in intracellular Ca²⁺ was detected during Ca²⁺ removal and add-back in the absence of thapsigargin (Fig. 3B), indicating that the junctate-mediated entry was dependent on the emptying of ER Ca²⁺ stores. Taken together, these data indicate that junctate can have a small impact on global SOCE that is only revealed when both STIM proteins are absent.

We next tested whether junctate expression impacts the ability of cells to phagocytose, using a high target:cell ratio (10:1) to increase the reliance of the phagocytic process on Ca²⁺ signals (Nunes et al., 2012). To synchronize the time of exposure to targets, RBCs were centrifuged onto cells that had been seeded onto coverslips, and cells were allowed to phagocytose for 30 min before fixation. This single timepoint snapshot provides an estimate of the phagocytic rate as cells reached their maximum phagocytic capacity, ∼2 h after target centrifugation at this target:cell ratio. Coverslips were immunostained to reveal c-Myc-tagged Fc receptors to facilitate phagosome identification, and imaged by using confocal microscopy. Quantification of confocal z-stacks through visual inspection revealed that the average number of phagocytosed particles per cell (the phagocytic index) during this 30-min time interval doubled in cells that expressed junctate compared to cells that expressed GFP–KDEL [wild type, 2.05±0.14 (GFP–KDEL) and 3.76±0.47 (YFP–junctate) phagosomes per cell, mean±s.e.m. n=4 for each; Stim1^−/−, 1.58±0.21 and 3.03±0.35 phagosomes per cell, n=6 for each; Stim1^−/−; Stim2^−/−, 1.45±0.2 and 4.27±0.65 phagosomes per cell, n=6 for each; Fig. 4]. mCherry–STIM1 re-expression further boosted the phagocytic index (Stim1^−/−, to 4.15±0.79 phagosomes per cell, n=6; Stim1^−/−; Stim2^−/−, to 5.97±1.08 per cell, n=6; Fig. 4). These data suggest that junctate can increase the phagocytic capability of cells in the absence of STIM proteins despite having only a marginal effect on SOCE. The fact that Stim1^−/−; Stim2^−/− cells showed similar or higher phagocytic rates and recruitment to phagosomes upon junctate overexpression indicates that junctate does not simply recruit STIM2 as an alternative mechanism to gate SOCE channels, and that, if anything, interaction with STIM2 could hinder junctate function, possibly by sequestering junctate elsewhere.

Because STIM1 boosts phagocytosis by increasing the frequency and duration of local Ca²⁺ elevations around phagosomes, we assessed whether junctate also promotes periphagosomal Ca²⁺ elevations by using live confocal imaging and the Ca²⁺-sensitive dye Fluo-8 (Fig. 5A). Ca²⁺ hotspots (arrows) were detected within a 750-nm perimeter (red region of interest, middle panel) in 6-s snapshots taken 25–30 min after the addition of phagocytic targets. Changes in hotspot occurrence in these snapshots might reflect either an increased frequency or duration of these events, both of which would reflect increased signaling. Although we did not observe changes in the temporal dynamics of hotspots (see Movie 1), approximately 39% of phagosomes in Stim1^−/− cells that expressed red fluorescent protein (RFP)–KDEL as a control ER protein displayed periphagosomal Ca²⁺ hotspots, and this percentage increased to 76% and 82% upon expression of RFP–junctate and mCherry–STIM1, respectively (n=5 for all conditions; Fig. 5A, bottom panel). Thus, junctate is as efficient as STIM1 in promoting periphagosomal Ca²⁺ elevations. Interestingly, although the hotspots promoted by junctate covered a similar fraction of the phagosomal surface as the STIM1-generated hotspots (28% vs 27%, n=47 and 64 phagosomes, respectively), they extended more frequently beyond the 750-nm wide periphagosomal space (84±5.0% vs 59±4.5%, n=3), prompting us to further explore the source of the Ca²⁺ signal. The intracellular Ca²⁺ hotspots that form around phagosomes can reflect either Ca²⁺ release from the ER or Ca²⁺ release from phagosomes, which initially contain a high concentration of Ca²⁺ that is derived from the extracellular space. To distinguish between these two sources of Ca²⁺, we repeated the experiment either in Ca²⁺-free medium or in the presence of SOCE channel inhibitor La³⁺, which abolished global SOCE in Stim1^−/− and Stim1^−/−; Stim2^−/− cells (Fig. 3A,C). This generic Ca²⁺-channel blocker is internalized into the lumen of forming phagosomes and minimizes the contribution of Ca²⁺ flowing across both plasma-membrane and phagosomal channels, allowing us to isolate the Ca²⁺ signals originating from internal stores. Ca²⁺ removal from the external medium decreased the percentage of phagosomes that were associated with local Ca²⁺ elevations to 14% and 22% in RFP–KDEL and mCherry–STIM1 cells, respectively, but to only 32% in cells that expressed RFP–junctate, which exhibited significantly more periphagosomal Ca²⁺ microdomains than cells that expressed the other two proteins (Fig. 5B,C). This difference was further increased in the presence of La³⁺, which had only a mild effect on the activity recorded in RFP–junctate-expressing cells but decreased the activity of mCherry–STIM1-expressing cells to control levels (55% for RFP–junctate versus 34% and 36% for RFP–KDEL and mCherry–STIM1, respectively; Fig. 5B,C).

To confirm that junctate regulates periphagosomal Ca²⁺ hotspots by recruiting functional Ca²⁺ stores, periphagosomal Ca²⁺ was measured in the presence of the InsP3R inhibitor 2-APB. Because 2-APB also blocks Orai channels, La³⁺ was simultaneously applied to eliminate the contribution of any SOCE channel, including Orai-family proteins. This treatment decreased the frequency of periphagosomal Ca²⁺ hotspots to control levels in both RFP–junctate- and mCherry–STIM1-expressing cells (12% for RFP–junctate vs 19% for RFP–KDEL and 13% for mCherry–STIM1; Fig. 5B,C). Finally, to control for the contribution of residual STIM2-mediated channel gating, periphagosomal Ca²⁺ hotspots were also quantified in Stim1^−/−; Stim2^−/− MEFs. Similar to Stim1^−/− cells, YFP–junctate and mCherry–STIM1 expression increased hotspot frequency by twofold, to 51% and 53% respectively, compared to 25% for RFP–KDEL (Fig. 5B). Xestospongin-C, a more specific InsP3R blocker that suppressed ATP-induced Ca²⁺ signals in Stim1^−/−; Stim2^−/− MEFs (Fig. S1), had no effect on hotspot frequency in cells that expressed RFP–KDEL but did abrogate the increase in periphagosomal Ca²⁺ hotspots evoked by junctate expression and decreased the frequency of the STIM1-induced hotspots to 38% (Fig. 5B). The residual Ca²⁺ hotspots persisting in the presence of inhibitors might reflect incomplete inhibition of Ca²⁺ influx and release channels, or release from InsP3R-independent Ca²⁺ stores. Taken together, these data confirm that STIM1-mediated hotspots are generated by both phagosomal and ER-store release components, as previously suggested, and show that the store release component is InsP3R-dependent. Moreover, these results indicate that junctate promotes periphagosomal Ca²⁺ elevations primarily by recruiting functional Ca²⁺ stores to phagosomes rather than by interacting with Ca²⁺-permeable channels that are present in the phagosome membrane.

Periphagosomal Ca²⁺ hotspots have been previously shown to promote shedding of the actin coat that forms during phagosomal ingestion, which is considered to be an important stage during early phagosome maturation (Nunes and Demaurex, 2010; Nunes et al., 2012). Reduced actin shedding might limit global phagocytic rates by hindering phagosome–lysosome fusion as well as fission events that allow recycling of Fc receptors back to the cell surface. The availability of actin, and particularly rapid actin dynamics, could be additionally important for higher speeds of subsequent actin-driven membrane remodeling as the cell internalizes new targets. We thus further investigated whether junctate, like STIM1, influences periphagosomal actin structures. Both in Stim1^−/− and Stim1^−/−; Stim2^−/− MEFs, the overexpression of RFP–junctate and mCherry–STIM1 decreased periphagosomal filamentous F-actin rings compared to that in RFP–KDEL controls (Fig. 6). These data indicate that junctate regulates phagocytic ingestion rates by promoting periphagosomal actin shedding.

In this study, we show that junctate is an important signaling protein that can determine the efficiency of the phagocytic process. We provide morphological evidence that junctate is recruited to phagosomes and increases the length of juxtaposed ER structures, and functional evidence that junctate significantly increases the efficiency of phagocytosis independently of STIM1. We could link the pro-phagocytic effects of junctate to an increased frequency of periphagosomal Ca²⁺ elevations occurring near phagosomes, as well as to increased actin shedding during early phagosome maturation, and show that the junctate-associated Ca²⁺ hotspots or microdomains are much less affected by Ca²⁺ removal and by the SOCE channel inhibitor La³⁺ than STIM1-associated Ca²⁺ microdomains. We further show that the junctate-associated Ca²⁺ microdomains are almost abrogated by the InsP3R inhibitor 2-APB in the presence of La³⁺, as well as by the more specific InsP3R inhibitor xestospongin-C alone, indicating that they essentially reflect Ca²⁺ release from inositol-trisphosphate-sensitive stores. This suggests that, unlike STIM1, junctate does not promote the opening of phagosomal Ca²⁺ channels but instead recruits InsP3R-containing ER Ca²⁺ stores near phagosomes. Junctate is more efficient than STIM1 in recruiting functional ER Ca²⁺ stores, whereas STIM1 is more efficient in gating phagosomal Ca²⁺ channels, allowing the two proteins to cooperate to promote the generation of local Ca²⁺ elevations that boost phagocytosis.

Junctate has previously been shown to play an important role in Ca²⁺ homeostasis and to interact with several components of the Ca²⁺ ‘toolkit’. Initial studies reported that junctate interacts through its cytosolic N-terminal domain with the InsP3R and TRPC channels, and that junctate overexpression increases agonist-induced Ca²⁺ elevations, whereas silencing has the opposite effect (Treves et al., 2004). Those authors have subsequently shown that junctate stabilizes ER–plasma-membrane junctions and promotes agonist-activated Ca²⁺ entry across TRPC3 channels (Treves et al., 2010). Junctate also interacts with SERCA2a in cardiomyocytes (Kwon and Kim, 2009), and with TRPC2 and TRPC5 channels in rodent sperm (Stamboulian et al., 2005). More recently, junctate has been shown to interact through its luminal domain with STIM1 and to provide an alternative mechanism for STIM1 recruitment to Ca²⁺ entry sites at the plasma membrane (Srikanth et al., 2012). In this study, junctate overexpression in primary T cells caused a marginal increase in SOCE, whereas a junctate EF-hand mutant promoted the accumulation of STIM1 in ER–plasma-membrane clusters and caused a marked increase in SOCE. These data indicate that junctate acts as an ER sensor that facilitates STIM1 clustering at ER–plasma-membrane junctions, but they do not exclude the possibility that junctate acts as a ligand for membrane channels on its own in response to store depletion. We observed that when SOCE is maximally activated, junctate overexpression does not further increase global Ca²⁺ entry rates in wild-type MEFs, nor does it rescue the residual SOCE of Stim1^−/− cells. By contrast, it can induce a small but significant SOCE in the complete absence of STIM proteins. This surprising result suggests that junctate can gate SOCE channels independently of STIM proteins. This observation not only lends further insight into the function of junctate, but adds an additional layer to the complex regulation of store-operated Ca²⁺ entry by identifying an additional ER Ca²⁺ sensor, beyond the three known STIM protein isoforms STIM1, STIM1L and STIM2. Because junctate is known to bind to TRPC-family channels, and because the non-specific Ca²⁺ channel blocker La³⁺ abolished junctate-mediated influx, TRPCs are the prime candidates for serving as junctate partners, and further study will be required to identify the exact partner channels as well as whether other physiological functions are linked to this additional influx mechanism. Junctate–TRPC interactions could explain the fact that, in RFP–junctate-expressing cells, La³⁺ slightly decreased the fraction of phagosomes that had Ca²⁺ microdomains (from 76% to 57%; Fig. 5A,B) in Stim1^−/− cells. However, this could also reflect reduced Ca²⁺ release from stores, because La³⁺ could prevent store refilling by inhibiting plasma membrane Ca²⁺ channels during the phagocytic process. In any case, La³⁺ was much more effective in STIM1-expressing cells, impacting the majority of the Ca²⁺-signaling around those phagosomes (from 82% to 36%; Fig. 5B). Thus, the STIM1–Orai interactions that take place at the ER–phagosome interface are much more efficient than the putative junctate–TRPC interactions in promoting the opening of phagosomal Ca²⁺ channels.

Our Stim1^−/− phagocytic cells provided a useful model system to highlight the similarities and differences between STIM1 and junctate. Both STIM1 and junctate were recruited to phagosomes in Stim1^−/− cells, demonstrating that junctate is recruited independently of STIM1, and the two proteins populated the same periphagosomal structures when co-expressed. Junctate is localized at ER–plasma-membrane contact sites before store depletion through its N-terminal domain (Srikanth et al., 2012; Treves et al., 2010), but whether junctate is ‘pre-addressed’ to contact sites through this targeting motif or responds to smaller or localized store depletion compared to STIM1 is unclear. The STIM1 luminal domain has been estimated to have a Ca²⁺-binding K_d of 200–600 µM in vitro (Stathopulos et al., 2006), and 169–210 µM by using fluorescence and patch-clamp recordings (Brandman et al., 2007; Luik et al., 2008). The K_d of junctate has been estimated to be around 217±20 µM by using fluorescence imaging (Treves et al., 2000).We cannot establish whether junctate is recruited to phagosomes independently of store depletion because the Ca²⁺ concentration within the ER lumen decreases as soon as phagocytic receptors are engaged, which could initiate the recruitment of junctate to its target membranes. Whether an additional signal triggers the accumulation of junctate at the ER–phagosome interface or alternatively, whether phagosomes form preferentially around pre-existing junctate-containing ER–plasma-membrane contact sites therefore remains to be determined.

Our electron microscopy data also establish that junctate is an ER-shaping protein that can elongate ER cisternae that are apposed to phagosomes, causing a doubling in their length (from 115 to 231 nm; Fig. 2). This effect was most apparent in the absence of STIM1 because junctate expression in wild-type fibroblasts only marginally increased the already larger size of the recruited ER structures (from 231 to 268 nm; Fig. 2). Previous studies have reported that junctate expression extends ER–plasma-membrane contacts by up to 55% (from 218 to 339 nm) in HEK-293 cells (Treves et al., 2004). We confirm here that junctate is indeed able to elongate ER structures and further establish that this effect is independent of STIM1. In wild-type MEFs, juxtaposed ER might already be near to maximal length owing to high levels of endogenous STIM1, but differences in junctate localization and overexpression efficiency might also explain the lack of lengthening effect in these cells. Compared to STIM1, junctate was less efficient in remodeling the ER because its expression did not promote the appearance of new ER structures that were apposed to phagosomes, only their extension. It should be noted however that we only quantified ER structures that were located less than 30 nm from the phagosomal membrane. Junctate might, in fact, be more efficient than STIM1 in recruiting ER structures at some distance from the phagosomes, as suggested by the fluorescence pattern of periphagosomal YFP–junctate. The physiological role of the ER extension mediated by STIM1 and junctate is unclear. The enlargement of ER–plasma-membrane junctions that is mediated by different STIM1 isoforms does not correlate with an increase in SOCE or in the efficiency of ER Ca²⁺ refilling (Saüc et al., 2015), suggesting that elongation is not related to the Ca²⁺-signaling function of STIM1 molecules. The enlargement of juxtaposed ER structures might provide more stable signaling platforms for interactions of ER proteins with their ligands at the plasma or phagosomal membrane, or for lipid exchange between the two juxtaposed membranes (Toulmay and Prinz, 2011). At the ER–phagosome interface, the extension might promote the cross-presentation of antigens processed within the phagosomes, which relies on the communication between phagosomes and the ER-resident antigen loading machinery in a manner that is still poorly understood (Blum et al., 2013).

To conclude, our data further the understanding of junctate function as an important component of ER membrane contact sites, platforms that allow cells to precisely localize Ca²⁺ signals within cells. We showed that junctate complements STIM1 function, independently increasing contact lengths and Ca²⁺ release from stores near to phagosomes, thereby increasing phagocytic efficiency. Although our work is focused within the context of an important process of immune cell function – that of phagocytosis – in a more general sense, our data illustrate that, when global Ca²⁺ signals are abrogated, increasing localized Ca²⁺ signals through Ca²⁺ release from stores can provide an effective compensatory mechanism when SOCE is naturally, pathologically or therapeutically inhibited. This might be a concept with repercussions beyond the field of phagocytosis because Ca²⁺ signaling is important for many cell types and for a variety of cellular functions, including cell growth, apoptosis, secretion, contraction and motility. Furthermore, because SOCE channels have been proposed as drug targets in therapeutic avenues for diseases including cancer and autoimmune disorders, a better understanding of complementary or compensatory mechanisms of Ca²⁺ signaling will undoubtedly lead to more intelligent therapeutic designs.

Stim1^−/− MEFs, generated through targeted gene disruption (Prins et al., 2011), and wild-type MEF control cells were a kind gift from Dr Marek Michalak (University of Alberta, Canada). YFP-tagged junctate was a kind gift from Dr Susan Treves (University of Basel, Switzerland). Dulbecco's modified Eagle's medium (DMEM; high glucose, catalog number 31966), heat-inactivated fetal calf serum (FCS), penicillin–streptomycin mixture (pen–strep, catalog number 15140), Fura-2-AM, BAPTA-AM, Lipofectamine 2000 transfection reagent, goat-anti-mouse conjugated to Alexa-Fluor-647 (1:1000) and SlowFade mounting medium were obtained from Life Technologies. Quest Fluo-8-AM was purchased from AAT Bioquest (Sunnyvale, CA). Mouse anti-Myc-tag antibody (9B11, 1:100) was purchased from Cell Signaling. Gluteraldehyde-stabilized sheep red blood cells (sRBCs), rabbit anti-sRBCs (1:200) and all other chemicals were obtained from Sigma-Aldrich. To reduce autofluorescence, gluteraldehyde-stabilized sRBCs were treated with bubbling 0.6% NaBH₄ in PBS with agitation for 1 h, washed three times in PBS and stored frozen.

MEFs were grown in DMEM containing 10% FCS and 0.5% pen–strep at 37°C under 5% CO₂ and were passaged twice a week. Cells were used between passages 5 and 50. Transfections were performed with Lipofectamine 2000 with cells at 50–60% confluence in high glucose (4.5 g/l) DMEM. Transfected cells were allowed to recover for 24 h before manipulation.

Opsonization was performed the same day as phagocytosis experiments. RBCs were opsonized by rabbit-anti-sheep RBCs at 37°C for 1 h and washed three times in PBS. Targets were added directly to cells that had been seeded on 12-mm coverslips in 24-well plates at 10:1 target:cell ratio in serum-containing medium. Plates containing coverslips were centrifuged at 600 g for 1 min. Cells were then incubated at 37 °C under 5% CO₂ for 10 min in the case of ER recruitment experiments and 30 min for phagocytosis experiments, before fixation and immunolabeling.

Cells were fixed in 4% paraformaldehyde in PBS. Fixed cells were permeabilized and blocked in 0.3% Triton X-100 with 1% BSA in PBS, and incubated overnight at 4°C in primary antibody (mouse antibody against Myc tag, 1:100). The following day, cells were incubated for 1 h in secondary antibody (goat anti-mouse, conjugated to Alexa-Fluor-647, 1:1000) and then washed three times with PBS. Coverslips were mounted in SlowFade mounting medium containing 1 µg/µl Hoechst 33342.

Cells were fixed for 1 h in 2% glutaraldehyde with 0.1 M NaPO₄, pH 7.4, and scraped, and the pellets were washed once in 0.1 M sodium phosphate buffer, pH 7.4. En bloc staining with uranyl acetate, postfixation with osmium tetroxide, dehydration in ethanol, embedding in Epon and sectioning were performed by the Pole Facultaire de Microscopie Electronique core facility at the University of Geneva. Two grids per independent sample containing 8–10 50-nm sections were observed using a Tecnai transmission electron microscope (FEI, Eindhoven, The Netherlands). Of these, all phagosomes that could be identified in two to three randomly selected intact sections were examined. Quantification was performed using XT Pro software (Soft Imaging System GmbH, Germany).

All fixed-cell imaging was performed using a confocal laser scanning microscope (LSM 700, Carl Zeiss AG) equipped with a 60× objective. Fura-2 imaging was performed using a wide-field fluorescence microscope equipped with a 40× objective, polychromator illumination, 430DCLP dichroic and 510WB40 emission filter (Visitron Systems, GmbH). Here, the fluorescence of YFP and GFP, or of RFP tags in transfected cells was simultaneously examined, and cells expressing very high levels of transfected proteins or showing signs of pre-activation or toxicity were eliminated from the analysis. Fluo-8 imaging was performed using a Nikon A1R inverted confocal microscope system equipped with 60× objective and a resonant scanner, and maintained at 37°C by a microscope temperature control system (Life Imaging Services, Basel, Switzerland). Fluo-8 fluorescence was imaged using the 488-nm laser line, whereas the autofluorescence of RBCs was imaged using the far-red 633-nm laser. Experiments were performed in physiological buffer – 140 mM NaCl, 5 mM KCl, 1 mM MgCl₂, 2 mM CaCl₂, 20 mM Hepes, 10 mM glucose, adjusted to pH 7.4, with NaOH. Ca²⁺-free medium contained 1 mM EGTA instead of 2 mM CaCl₂. Fura-2-AM (2 µM) was loaded in physiological buffer with 0.02% pluronic and visualized using 340/380-nm alternate excitation and 510±40-nm emission. Frames were acquired every 2 s. Fluo8-AM (4 µM) was loaded in physiological buffer with 250 µM sulfinpyrazone at 37°C for 30 min. This was followed by incubation for 20 min at room temperature and subsequent addition of BAPTA-AM (5 µM) for another 10 min. For quantification of phagocytosis, ER protein recruitment and periphagosomal Ca²⁺ hotspots, the same protocol that has been previously described for STIM1 was used (Nunes et al., 2012), except for Ca²⁺-free experiments, where instead of using 3 mM EDTA to chelate extracellular 2 mM Ca²⁺, the experiment was performed in Ca²⁺-free medium. Briefly, for phagocytosis and ER protein recruitment studies, at least three confocal z-stacks of equal dimensions (∼15 µm thickness at 0.5 µm steps) per coverslip were imaged, and at least n=3 independent experiments were quantified for each condition. RBCs completely enclosed by cellular borders were defined as phagosomes. For RBCs found near the cell periphery, in the case of incomplete enclosure, only RBCs with ≥75% of their surface surrounded by Fc receptor immunoreactivity were counted as phagosomes. ER proteins were considered to be recruited to the vicinity of phagosomes when bright fluorescent clusters (puncta) could be detected by visual inspection in the space within 3 pixels (∼750 nm) of phagosomes. Ca²⁺ hotspots were defined as regions of an area ≥500 nm² (4 pixels) within a distance of ∼750 nm (3 pixels) from the phagosomal border displaying fluorescence at least 2 s.d. higher than the average cytoplasmic Fluo-8 intensity. At least five snapshots per coverslip and at least n=3 independent coverslips were quantified for each condition.

All statistical analyses were performed using Prism software (GraphPad). Significance between two sets of experiments was determined using an unpaired Student's t-test.

We are grateful to Mr Cyril Castelbou for excellent technical assistance, to the bioimaging core facility (Geneva Medical Center) and to Dr Maud Frieden and Dr Claes Wollheim for advice and for critical reading of the manuscript.