STIM1 promotes migration, phagosomal maturation and antigen cross-presentation in dendritic cells

Antigen cross-presentation by dendritic cells (DC) stimulates cytotoxic T cell activation to promote immunity to intracellular pathogens, viruses and cancer. Phagocytosed antigens generate potent T cell responses, but the signalling and trafficking pathways regulating their cross-presentation are unclear. Here, we show that ablation of the store-operated-Ca2+-entry regulator STIM1 in mouse myeloid cells impairs cross-presentation and DC migration in vivo and in vitro. Stim1 ablation reduces Ca2+ signals, cross-presentation, and chemotaxis in mouse bone-marrow-derived DCs without altering cell differentiation, maturation or phagocytic capacity. Phagosomal pH homoeostasis and ROS production are unaffected by STIM1 deficiency, but phagosomal proteolysis and leucyl aminopeptidase activity, IRAP recruitment, as well as fusion of phagosomes with endosomes and lysosomes are all impaired. These data suggest that STIM1-dependent Ca2+ signalling promotes the delivery of endolysosomal enzymes to phagosomes to enable efficient cross-presentation.

D endritic cells (DC) are phagocytic immune cells that link innate and adaptive immunity by processing and presenting ingested antigens. One of the unique functions of DCs is cross-presentation, which is a specific type of antigen presentation that occurs via major histocompatibility complex class I (MHC-I) molecules to activate CD8 + T cells and help generate antigen-specific immunity to intracellular pathogens, viruses and cancer cells. Cross-presentation of antigens acquired through phagocytosis produces more potent T cell responses than soluble antigens 1 , and DCs are particularly involved in phagocytosis and transport of large particles (> 500 nm) to draining lymph nodes 2 . However, the precise molecular mechanisms by which cross-presentation of phagocytosed antigens occurs are not well understood. Cross-presentation requires a number of proteins normally located in the endoplasmic reticulum (ER), such as tapasin, calreticulin, ERp57 and the translocon Sec61 1 . DC phagosomes are particularly rich in ER proteins 3,4 , but the signalling and trafficking mechanisms regulating the relationship between the ER and the phagosome during cross-presentation is controversial 3,[5][6][7][8] .
Ca 2+ signalling is linked to a variety of DC functions including differentiation, maturation, migration, cytokine secretion, phagocytic ingestion and antigen presentation 9 . However, most studies have relied on the use of non-specific inhibitors, ionophores and chelators, which can have pleiotropic effects. Stromal interaction molecule (STIM) proteins, which include the two isoforms STIM1 and STIM2 each with multiple splice variants, are ER transmembrane proteins that sense the ER Ca 2+ depletion resulting from activation of inositol trisphosphate (InsP 3 ) receptors 10 . They subsequently remodel the ER and promote the formation and expansion of membrane contact sites (MCS) between the ER and plasma membrane (ER-PM MCS), where they directly activate PM-resident Ca 2+ channels of the ORAI and transient receptor potential (TRPC) families in the process termed store-operated Ca 2+ -entry (SOCE) 11 . Electrophysiological studies suggest that SOCE is the major Ca 2+ entry pathway in DCs 12 , and one study suggests that STIM2 is the major isoform regulating DC function in mice 13 . In human peripheral blood monocyte-derived DCs genetic manipulation of ORAI1 and STIM1 suggested that STIM1 is critical for DC maturation 14 , but another study suggests that STIM1 and STIM2 are dispensable for a variety of DC functions in mice 15 .
Although the classic model of cross-presentation postulates that antigens are first partially proteolysed in phagosomes, retrotranslocated from the phagosome to the cytosol where they are further processed by the proteasome, and then reimported into the ER for loading onto ER-resident MHC-I molecules 1 , some studies propose that non-canonical trafficking pathways involving fusion of ERGIC vesicles and recycling endosomes with phagosomes may explain the presence of ER proteins on phagosomes 7,8,16 . However, the signalling and targeting mechanisms that control these pathways are unclear. In neutrophils, we previously showed that STIM1 promotes the formation of contact sites between the ER and phagosomes that allow localized Ca 2+ signalling 17 , raising the question of whether STIM1 may also affect the association between phagosomes and the ER in DCs.
In the present study, we characterize the consequences of genetic ablation of Stim1 on DC functions including differentiation, maturation, migration, phagocytosis and cross-presentation. Our data establish that STIM1 is the major isoform controlling SOCE in mouse DCs and suggest that STIM1 promotes crosspresentation at least in part by increasing Ca 2+ -dependent migration. In addition, STIM1 promotes the formation of contact sites between the ER and phagosomes that in turn produce localized Ca 2+ signals that may potentiate proteolysis and fusion of phagosomes with endosomes and lysosomes.

Results
Stim1 promotes cross-presentation of phagocytosed antigens.
To determine whether STIM1 promotes cross-presentation, PBS solutions with 0, 0.5, or 1% ovalbumin (OVA)-coated beads (OVAb) were injected into footpads of LysM-Cre; Stim1 fl/fl mice bearing the CD45.2 allele and a conditional deletion of the Stim1 gene in myeloid cells, and into control CD45.2 Stim1 fl/fl littermates. After 24 h, CD45.1, H2-K b /OVA(257-264)-reactive CD8α + T cells (OT-I) labelled with carboxyfluorescein succinimidyl ester (CFSE) were injected retro-orbitally. Draining (DL) and non-draining (NDL) lymph nodes were harvested after 72 h, and the total number of CD45.1 + OT-I cells within the CD8α + population, as well as the CFSE dilution as a measure of OT-I proliferation, were determined. The full gating strategy is shown in Fig. 1a. STIM1 deficiency dramatically reduced the total number of CD45.1 + OT-I cells within the CD8α + gate in DL of mice injected with 1 or 0.5% OVAb but not in NDL (Fig. 1a, b) or in lymph nodes from mice injected with PBS ( Supplementary  Fig. 1a). OT-I proliferation was reduced in a dose-dependent manner (Fig. 1a, b), indicating that effectively, cross-presentation was impaired upon Stim1 ablation in myeloid cells.
We next examined whether cell-specific DC cross-presentation was impaired by co-culturing bone-marrow-derived DCs (BMDC) pulsed with OVAb along with OT-I cells in vitro and measuring OT-I proliferation after 72 h by bromodeoxyuridine (BrdU) incorporation. Western blots confirmed that STIM1 protein expression was reduced by > 90% in BMDCs isolated from LysM-Cre; Stim1 fl/fl mice (labelled henceforth Stim1 −/− for simplicity) as compared to BMDCs isolated from Stim1 fl/fl mice (labelled henceforth WT, Supplementary Fig. 1b). Maturation with microbial products CpG oligodeoxynucleotides and lipopolysaccharide (LPS) increased OT-I proliferation by ∼twofold (Fig. 1c), while Ca 2+ chelation with BAPTA-AM or inhibition of InsP 3 -receptors with Xestospongin C (Xesto) decreased crosspresentation, as did the inhibitors of the NADPH oxidase and the vacuolar-ATPase, diphenylpropidium iodide (DPI) and concanamycin A (ConcA) respectively, applied as positive controls (Fig. 1d). This indicates that cross-presentation is Ca 2+ -dependent. Stim1 ablation reduced the relative levels of crosspresentation to similar extents in immature and mature cells, an effect that was most pronounced at intermediate DC:T cell ratios (Fig. 1c). Similar results were obtained when measuring OT-I proliferation by CFSE dilution (Supplementary Fig. 1c, d). The Stim1-dependent defect in cross-presentation was specific for processing of phagocytosed antigens, as STIM1-deficient BMDCs were equally efficient at cross-presenting as WT when pulsed with varying doses of SIINFEKL peptide, an OVA(257-264) fragment that does not require proteolytic processing (Fig. 1e, Supplementary Fig. 1e). To confirm that abrogating STIM1 expression reduces cross-presentation in vitro, a CD8α + DC cell line highly efficient at cross-presentation, DC 211418 , was transduced with either control or shRNA directed against STIM1 (shCTR and shSTIM1 respectively). STIM1 depletion was verified by Western blot (Supplementary Fig. 1b). Similar to primary BMDCs, crosspresentation was reduced by STIM1 depletion in DC 2114 cells (Fig. 1f). DC 2114 induced nearly fourfold higher levels of OT-I proliferation as compared to BMDCs ( Supplementary Fig. 1f), generating large colonies easily discerned by microscopy where the effect of STIM1 knockdown was clearly apparent (Supplementary Fig. 1g). These data indicate that STIM1 promotes the cross-presentation of phagocytosed antigens in vivo and that this effect is recapitulated in vitro in primary or cultured DCs. of STIM1-deficient mice, a measure of functional cross-presentation, could result from defective migration of DCs to and from the site of OVAb injection where antigens are captured, to the lymph nodes where DCs encounter OT-I cells. To test whether DC migration to lymph nodes was impaired in vivo, a 1:1 mixture of BMDCs from CD45.2 WT mice expressing cytosolic GFP (WT-GFP) and CD45.2 Stim1 −/− BMDCs was coinjected into footpads of CD45.1 host mice together with 0.5%

ARTICLE
OVAb. DLs were collected at 24 and 48 h post injection, and the CD45.1 − CD45.2 + population analysed for GFP expression. The percentage of GFP-expressing WT cells in DLs was strongly and significantly higher at 24 h post injection, whereas the initial 1:1 ratio was nearly recovered at 48 h (Fig. 2a, Supplementary Fig 2a). BMDC migration was next examined in vitro using transwell chambers and the chemoattractant fMIFL to mimic microbial peptides found at the site of infection, SDF-1 (also known as CXCL12), a chemokine expressed by lymphatic vessels 19 , or CCL21 (also known as Exodus-2 or SLC), the major chemokine eliciting DC homing to lymph nodes (Fig. 2b, Supplementary  Fig. 2b). No differences between baseline levels of migration between WT and Stim1 −/− cells were observed, and surface expression of the SDF-1 and CCL21 receptors CXCR4 and CCR7 was similar in WT and Stim1 −/− BMDCs matured or not with CpG, LPS, or OVAb ( Supplementary Fig. 2c (Fig. 3a), a defect that could not be accounted for by changes in ORAI1 or STIM2 expression ( Supplementary Fig. 3a). STIM1 knockdown also reduced SOCE by 70% in DC 2114 cells ( Supplementary Fig. 3b), and Ca 2+ responses evoked by platelet activating factor (PAF) in BMDCs were markedly impaired regardless of maturation (Fig. 3b, Supplementary Fig. 3c). In addition to global Ca 2+ signals, localized Ca 2+ hotspots occur during phagocytosis in macrophages, neutrophils and phagocytic fibroblasts 17,22 . In fibroblasts, these signals coincide in time and space with the STIM1-mediated tethering of ER cisternae to phagosomes, reflecting both ER Ca 2+ release as well as the opening of phagosomal Ca 2+ channels 17 . We therefore checked whether periphagosomal Ca 2+ hotspots and ER-phagosome MCS (ER-Ph MCS) were also present in DCs. Indeed, periphagosomal Ca 2+ hotspots were observed in BMDCs exposed to OVAb for 30 min and were reduced by ∼50% upon Stim1 ablation (Fig. 3c, d). Periphagosomal Ca 2+ hotspots were observed even after 90 min of OVAb exposure, although their frequency was reduced and their occurrence was no longer STIM1 dependent (Fig. 3c). SOCE channel inhibition with GSK7975A (GSK, 10 µM) added concomitantly with OVAb, and InsP 3 R inhibition with Xesto (1 µM) 20 min after OVAb addition, reduced hotspot frequencies in WT but not STIM1-deficient cells (Fig. 3c). Although Xesto may affect STIM1 recruitment to phagosomes by preventing ER store depletion, these data suggest that both Ca 2+ release from phagosomes as well as Ca 2+ release from recruited ER stores contribute to local Ca 2+ signals. In addition, in transduced BMDCs, periphagosomal mCherry-STIM1 puncta co-localized with periphagosomal hotspots (Fig. 3e, white arrows). Periphagosomal STIM1 puncta that were not associated with Ca 2+ hotspots were also observed ( Fig. 3e, yellow arrows), but the majority of hotspots co-localized with STIM1 clusters. Immunostainings confirmed that STIM1 puncta were observed near phagosomes ( Supplementary Fig. 4a). Similar to neutrophils, ER cisternae closely apposed (<30 nm) to phagosomes were observed in BMDCs exposed to OVAb for 30 min, as visualized using both classic transmission electron microscopy ( Supplementary Fig. 4b) as well as focused-ion-beam scanning Interestingly, the 3D reconstruction also revealed that the same ER cisterna may display multiple sites of contact with the same phagosome (highlighted in yellow in Fig. 4a and Supplementary Movie 1). Additionally, whereas the phagosomal membrane closely follows the surface of the ingested beads for most of the phagosomal surface, in certain instances the region surrounding these ER-Ph MCS displayed multiple sites of bulging away from the bead that were reminiscent of vesicular trafficking (fusion or fission) activity (Fig. 4b, arrows). Quantification of EM slices revealed that similar to neutrophils 17 , Stim1 ablation significantly reduced the number of ER-Ph MCS in BMDCs (Fig. 4c). Together, these data imply that STIM1 depletion has a major impact on global SOCE in DCs, as well as on localized Ca 2+ signals mediated through ER-Ph MCS.
Stim1 is dispensable for DC maturation and phagocytosis. Previous reports suggest that DC differentiation and maturation are dependent on Ca 2+ signalling 22 . BMDCs differentiated from bonemarrow precursors expressed high levels of CD11c, CD11b and low levels of F4/80, characteristic of conventional-like DCs 23 (Fig. 5a, Supplementary Fig. 5a). In both wild-type and Stim1 −/− BMDC cultures, > 90% of cells expressed CD11c and expression of all three markers were similar in both genotypes, as was surface levels of MHC-I, indicating that DC differentiation was unaffected by Stim1 ablation (Fig. 5a). DC 2114 cells were CD8α + , CD11b − , F4/80 − and B220 − , (characteristic of CD8 + lymphoid-like DCs 23 ), and expressed similar levels of CD11c when transduced with either shCTR or shSTIM1 ( Supplementary Fig. 5b, c). Exposure of BMDCs to TLR ligands induces DC maturation characterized in part by an upregulation of cell-surface markers such as CD40, CD80, CD86 and MHC-II. When BMDCs were exposed to maturation stimuli CpG or LPS for 18 h the cell-surface expression of CD40, CD80, CD86 and MHC-II was similarly increased in both genotypes ( Fig. 5b, Supplementary Fig. 2c). STIM1-silenced DC 2114 cells were also equally capable of upregulating maturation markers in response to CpG ( Supplementary Fig. 5d). We then checked whether acute administration of these TLR ligands evoked Ca 2+ responses in immature cells. LPS evoked small Ca 2+ transients in ∼30% of BMDCs and these signals were of lower amplitude in Stim1 −/− cells, while CpG did not induce any detectable Ca 2+ transient in either genotype ( Supplementary Fig. 5e).
In neutrophils phagocytic ingestion is Ca 2+ -and STIM1dependent 17, 24, 25 , and phagocytosis is Ca 2+ -dependent in DCs 9, 15 . Phagocytic ingestion was therefore quantified by exposing BMDCs to OVA-coated Fluoresbrite-YG beads for 30 min, 4 h and 24 h. While Ca 2+ chelation by pre-incubation of cells with BAPTA-AM reduced phagocytosis by ∼40% in control WT cells, inhibition was less pronounced or not significant in mature cells, and neither Stim1 ablation nor the addition of the non-specific SOCE inhibitor La 3+ significantly affected phagocytosis in either immature or mature BMDCs (Fig. 5c, Supplementary Fig. 2c). Furthermore, phagocytosis was unaltered in STIM1-deficient cells as assessed by microscopy ( Fig. 5d). Together these data suggest that defects in DC differentiation, maturation or rates of phagocytic ingestion are unlikely to contribute significantly to the impaired cross-presentation of STIM1-deficient DCs.
Stim1 ablation does not alter phagosomal ROS or pH. Several phagosomal maturation steps such as actin shedding, ROS production and fusion with granules or lysosomes are Ca 2+ -dependent in neutrophils and macrophages 9,22 . In DCs, lower phagosomal proteolysis, higher phagosomal ROS and reduced phagosomal acidity correlate with higher cross-presentation 26 . In this context, reduced phagosomal ROS production, or increased acidity or proteolytic activity might lead to impaired crosspresentation in STIM1-deficient DCs. We therefore measured intracellular ROS production with dihydroethidium (DHE) in cells exposed to OVAb or zymosan particles (phagocytic targets known to elicit high levels of ROS), with the NADPH oxidase activator phorbol myristate acid (PMA) and inhibitor DPI serving as positive and negative controls. Intracellular ROS levels increased by 10-12-fold upon PMA or zymosan stimulation and by 4−6-fold upon exposure to OVAb, showing the highest level in cells matured with CpG (Fig. 6a). Extracellular ROS levels, measured with Amplex Rex, increased by 4-6-fold upon PMA exposure, whereas OVAb did not produce detectable extracellular ROS ( Supplementary Fig. 6a). Stim1 ablation had no significant effect on ROS production in any condition tested (Fig. 6a, Supplementary Fig. 6a). BMDCs were also exposed to OVAb coupled to OxyBurst and Alexa-568, and imaged at 30 min and 90 min after bead addition. The ratio of OxyBurst to Alexa-568 fluorescence increased progressively following particle ingestion and was inhibited by DPI, yet again, no significant differences were detected (Fig. 6b). We next examined whether phagosomal pH might be affected by Stim1 ablation. Phagosomal pH was measured by exposing BMDCs to either FITC-coupled or to pHrodo-Red-and Alexa-488-coupled zymosan. Zymosan was used instead of OVAb as FITC and pHrodo coupling to amine-functionalized beads was unstable. BMDCs were imaged after 30 and 90 min of exposure to zymosan, and pre-treatment with ConcA used to document V-ATPase-mediated acidification. Phagosomal pH was  Fig. 6b) and measurements with FITC and pHrodo were made in parallel, using cells originating from the same mice. BMDCs showed a broad distribution of phagosomal pH ranging from very acidic to very basic (∼pH 4-9), displaying a roughly bimodal distribution, with an average pH that steadily decreased over the course of the recordings in a ConcA-sensitive manner (Fig. 6c-e). Phagosomes within the same cell had widely differing pH values, indicating that heterogeneity is determined at the level of single phagosomes (Fig. 6f). Regardless of the dye used, the phagosomes of Stim1 −/− BMDCs displayed an average pH similar to WT cells (6.57 ± 0.11 vs. 6.68 ± 0.19 and 5.67 ± 0.34 vs. 5.66 ± 0.21 after 90 min for WT and Stim1 −/− cells using FITC and pHrodo respectively, Fig. 6c, d). Endosomal pH was additionally assessed by loading cells with Alexa-568 and FITC-coupled dextran. Although a trend for a more alkaline endosomal pH in Stim1 −/− cells was apparent, the difference was not significant (Supplementary Fig. 6c).
Stim1 ablation impairs phagosomal proteolysis and fusion. We next measured phagosomal proteolysis, using beads coated with Alexa-568 and DQ-OVA, a protease probe consisting of OVA molecules heavily labelled with self-quenching BODIPY dye, which becomes brightly fluorescent upon OVA hydrolysis 27 . DQ-OVA fluorescence increased progressively, and green fluorescence leaked from the phagosome into the cytosol, indicating that proteolysis occurred (Fig. 7a). Contrary to macrophages 27 , phagosomal proteolysis reported by this assay was only partially inhibited by ConcA (Fig. 7a). Stim1 ablation significantly decreased DQ-OVA fluorescence by ∼30% at 90 min. This effect tended to persist in the presence of ConcA, although the difference here was not significant (p = 0.09) (Fig. 7a).
Pre-loading cells with BAPTA-AM reduced proteolysis in both genotypes to the levels of untreated Stim1 −/− cells, indicating that STIM1 mediates Ca 2+ -dependent proteolysis in BMDCs (Fig. 7a). As an alternative approach, the levels of phagosomeassociated OVA was determined by quantifying anti-OVA immunostaining of isolated BMDC phagosomes 28 . OVA degradation was inhibited in STIM1-deficient cells at 30 and 60 min ARTICLE but recovered after 2 h, indicating a partial or delayed OVA degradation defect (Fig. 7b, Supplementary Fig. 7a).
Since phagosome-to-cytosol transfer of peptides is suggested to be favoured by the delivery of ER proteins to phagosomes 16,29 , we investigated whether endosome-to-cytosol transfer was influenced by Stim1 ablation as a proxy measure of phagosomal antigen transfer 16 . BMDCs were loaded with the β-lactamase FRET reporter CCF4 and then exposed to β-lactamase. Upon transfer to the cytosol, β-lactamase cleaved CCF4, reducing the FRET emission in cells incubated at 37°C with identical kinetics in WT and STIM1-deficient cells (Supplementary Fig. 7b, c). This suggests that defective phagosome-to-cytosol transfer likely cannot account for the decreased DQ-OVA signal and that delivery of the molecular machinery underlying phagosome-to-cytosol transfer may not be dependent on STIM1.
We then tested phago-lysosome fusion. BMDCs were pulsed with the FRET acceptor Alexa-568-HA for 3 h followed by an overnight chase to accumulate the dye in lysosomes 30 . OVAb coupled to the FRET donor Alexa-488 were then added for 30 and 90 min. The phagosome-associated FRET signal increased from 0 to 90 min in a ConcA and BAPTA-dependent manner, confirming that phago-lysosome fusion is Ca 2+ -dependent in BMDCs (Fig. 7c). Interestingly, Stim1 ablation reduced phago-lysosome fusion by ∼30% at 90 min (Fig. 7c). Since OVA degradation was most strongly affected at earlier time points, we additionally tested whether endosome fusion might be defective, by pre-loading cells for 15 min with Alexa-594-dextran before exposing them to Alexa-488-OVAb and measuring the cellular FRET signal. Indeed, phago-endosome fusion was reduced in Stim1 −/− cells at 15 and 30 min, with the effect most pronounced at 15 min (Fig. 7d). Addition of GSK (10 µM) together with OVAb abolished differences between WT and Stim1 −/− cells (Fig. 7d). These data suggest that reduced phago-lysosome and phago-endosome fusion contribute to reduced proteolysis in STIM1-deficient cells. Therefore, we tested whether the proteolytic activity of lysosomal or endosomal proteases might be affected. While there was a trend for decreased activity of the lysosomal hydrolases involved in antigen processing 31 cathepsin B/L (CatB/L), cathepsin S (CatS) and asparagine endopeptidase (AEP), the differences were not significant ( Supplementary Fig. 7d). On the other hand, leucyl aminopeptidase activity, characteristic of the endosomal protease insulinregulated aminopeptidase (IRAP) previously shown to be important for cross-presentation 32 , was reduced by ∼40% in phagosomes isolated from STIM1-deficient cells, but not in whole-cell lysates (Fig. 7e). Quantification of immunostainings revealed that periphagosomal IRAP was reduced by ∼20% (Fig. 7f). Together with the fact that IRAP + endosomes are partially co-labelled with fluorescent dextran internalized for 15 min (Supplementary Fig. 7e), these data indicate that reduced fusion of IRAP-containing endosomes contributes to the crosspresentation defect observed in STIM1-deficient cells, although a direct effect on the activity of phagosomal IRAP or of other leucyl aminopeptidases cannot be excluded.

Discussion
We report here that mice lacking STIM1 in the myeloid lineage fail to effectively cross-present phagocytosed antigens in vivo. The cross-presentation defect is recapitulated in vitro in DCs matured or not with TLR ligands, and is restricted to phagocytosed antigens, pointing to specific defects in phagocytic processing. Both global and localized Ca 2+ signals are reduced in STIM1-deficient DCs. Surprisingly, but in line with a recent study 15 , DC differentiation, maturation and phagocytosis are not affected by Stim1 ablation. Instead, defective cross-presentation is linked to two other impaired DC functions: defective in vivo migration, which correlates with reduced chemotaxis to Ca 2+ -mobilizing chemokines, and altered phagosomal maturation. STIM1 deficiency reduces SOCE by 70% in both BMDCs and DC 2114 cells indicating that STIM1 is the dominant isoform in DCs, and not STIM2 as proposed 13 . Since double Stim1;Stim2 ablation completely abolishes SOCE in DCs 15 however, STIM2 still contributes to Ca 2+ signalling in DCs. Similar to Vaeth et al, and unlike human DCs 14 , we observe that Stim1 is not required for the upregulation of DC maturation markers in mice, nor does it affect differentiation. Indeed, increased ROS production and upregulation of CCR7 and CXCR4, both events associated with DC maturation are additionally intact. This is in stark contrast to neutrophils derived from similar mouse genetic models, where phagocytosis 17,25 and ROS production 25 are both STIM1 dependent. These differences highlight the importance of cell-type dependence of Ca 2+ signalling outcomes. Moreover, in contrast to RAW macrophages 33 and similar to B cells 34 , CpG does not induce detectable Ca 2+ transients, while LPS induces small and variable Ca 2+ transients in only ∼30% of cells. Thus, redundant Ca 2+ -dependent and independent pathways downstream of maturation stimuli appear sufficient to overcome a loss of STIM1 function in DCs.
We expected ROS production to be affected because NADPH oxidase activation is Ca 2+ and STIM1 dependent in neutrophils 35 , and ROS production is required for effective cross-presentation ( Fig. 1d and refs [36][37][38]. Surprisingly ROS production is unaffected in STIM1-deficient DCs suggesting that either sufficient Ca 2+ signalling remains, or that oxidase activation is SOCEindependent in DCs. Phagosomal pH is intimately linked to phagosomal ROS production and depends on the delivery of V-ATPase as well as HVCN1 proton channels 35,39,40 . In DCs phagosomal pH is controversial, with some groups reporting pH levels near 6.5 and others between 7 and 8 within the first 60 min of phagocytosis 36,37 . In all previous reports phagosomal pH was measured on cell populations. Here pH measurements are conducted on single phagosomes, using two independent probes, FITC and pHrodo, comparing cells originating from the same mice. No differences in subpopulations or average pH are detected between STIM1 and wild-type cells, but the two dyes yield an average pH differing by nearly 1.0 unit. This likely reflects the lower in situ pKa of pHrodo (6.1 ± 0.4 vs. 7.6 ± 0.3 for FITC) and the increased sensitivity of pHrodo at low pH levels ( Supplementary Fig. 6b), which indicates that pHrodo reports a more reliable pH estimate. On the other hand, the differences could also be related to the large variability in phagosomal pH that appears to be regulated at the level of single phagosomes. Shifts in the balance between the peaks of this bimodal, non-Gaussian distribution between experimental conditions may explain why different groups report such disparate values for mean population measurements. Why DC phagosomes display such a large variability remains unclear but phagosomal ROS, which can be highly heterogenous 41 , and can inhibit V-ATPase recruitment 35 , is likely involved. Indeed, Hvcn1 −/− neutrophils, which suffer a 50% loss in ROS production, also display a bimodal distribution of phagosomal pH 39 , and thus it could be interesting to correlate ROS and pH measurements at the level of single phagosomes in future studies in DCs. It is interesting to note that ORAI channels are sensitive to extracellular acidity, being activated at alkaline pH while inhibited at pH < 6 42 . The heterogeneity of phagosomal pH observed would thus allow phagosomes with robust ORAI activity to co-exist with phagosomes where ORAI is inhibited, and could explain why only a subset of STIM1-positive periphagosomal puncta coincide with Ca 2+ hotspots. The identical pH profiles of STIM1 and wild-type cells however, indicate that altered pH cannot account for defects in cross-presentation.
Despite the lack of differences in either ROS or pH, STIM1deficient DCs display surprisingly lower levels of proteolysis, measured by two independent assays, as well as reduced leucyl aminopeptidase activity, characteristic of the endosomal protease IRAP previously implicated in cross-presentation 32 . Decreased proteolysis is associated with decreased phago-endolysosome fusion, decreased global and periphagosomal Ca 2+ signals, and decreased frequency of ER-Ph MCS in STIM1-deficient cells. Although we cannot differentiate whether loss of global or local Ca2 + signals cause reduced fusion events, the high vesicular fusion/fission activity near ER-Ph MCS favours a model where local Ca 2+ signals promote local fusion events. It should be noted that decreased cathepsin-dependent proteolysis and reduced phago-lysosome fusion correlated with higher, not lower, levels of cross-presentation in DCs 38,43 . Yet, some proteolysis must occur to generate peptides from whole OVA molecules and inhibiting proteolysis with the broad-spectrum inhibitor leupeptin reduces cross-presentation in some [44][45][46] although not all 47,48 contexts. Indeed, in the companion study published in this same issue which analyses the function of UNC93B1, a novel STIM1 interactor 49 , Maschalidi et al. not only confirm a role for STIM1 in cross-presentation, but also observe that decreased proteolysis correlates with decreased cross-presentation. Proteolysis has therefore a bipartite effect on antigen processing, with both negative and positive effects on the MHC-I loading depending on its extent. Our data suggest that finely regulated levels of proteolysis must be reached to favour cross-presentation, and that this fine-tuning is achieved at least in part through Ca 2 + -dependent regulation of endosome fusion and IRAP delivery.
We also considered that antigen export from phagosomes into the cytosol might be affected by STIM1 deletion. Phagosomal export was reported for some but not all antigens 50 and cytosol transfer requires ER proteins such as Sec61, tapasin and TAP 1 . That STIM1 mediates close contacts between the ER and phagosomes argues that at least some ER proteins found on isolated phagosomes are derived from contacts, which may in turn facilitate the delivery of this ER-derived machinery. That the frequency of ER-Ph MCS was fivefold higher in WT DCs as compared to WT neutrophils, yet the frequency of periphagosomal Ca 2+ signals is only half that of neutrophils 17 , as well as the observation of periphagosomal STIM1 puncta without associated Ca 2+ signals (Fig. 3e), further support the idea that ER-Ph MCS may have roles beyond Ca 2+ signalling in DCs, and that they could partly mediate specialized DC functions that rely on ER proteins. However, no defects in cytosol transfer were observed. On one hand, ER contacts were only decreased by ∼35% upon Stim1 deletion, and potentially, sufficient activity remained to allow effective cytosol transfer. On the other hand, recent data suggests that different ER proteins such as tapasin 16 , MHC-I 8 and Sec61 7 reach phagosomes through unconventional vesicular trafficking pathways involving the ERGIC, recycling endosome and possibly other compartments. These new findings on alternative trafficking raise the intriguing possibility that endomembrane trafficking is more generally non-canonical in DCs and that specialized compartments analogous to secondary and tertiary granules in neutrophils might exists in DCs. Indeed, a lysosome-like Rab34-dependent compartment delivers NADPH oxidase in DCs 48 . Since cross-presentation was more strongly impaired than phago-lysosome fusion, STIM1 might regulate alternative Ca 2+ -dependent trafficking pathways in a way that bypasses the need for cytosol transfer. The generation of better tools to manipulate these pathways will be required to gain a deeper understanding of alternative compartments and the signals that regulate their activity.
In summary, we show that STIM1 has an important and selective regulatory role on Ca 2+ -dependent processes that control phagosome maturation and migration in DCs, thereby impacting the efficiency of cross-presentation. STIM-gated ORAI channels are promising drug targets for treating autoimmune diseases 51 , muscle defects 52 , skin disorders 53,54 and cancer 55 . In addition, activating ORAI1 through a genetically encoded lightinducible STIM1 fragment promotes tumour remission in a mouse model of DC cell-based immunotherapy 56 . Together with our data showing that either pharmacological or genetic manipulation of Ca 2+ signalling can decrease cross-presentation, this suggests that delineating the precise role of SOCE in innate immune cells is highly relevant for the identification of potential risks and benefits associated with the development of SOCE blockers and activators designed for therapeutic use. All lentiviral particles were produced in Lenti-X 293T cells using the Lenti-X HTS Packaging System (Takara, Japan) according to the manufacturer's instructions. Lentiviral titres were determined using the LentiX-p24 Rapid Titre ELISA kit (Takara). All cell culture reagents were obtained from ThermoFisher Scientific, and all chemicals were purchased from Sigma-Aldrich unless otherwise stated. mCherry-STIM1 57 cloned into pENTR1a was purchased from GenScript (USA). mCherry-STIM1 was then cloned into the p2K7 bsd lentiviral vector containing the ubiquitin promoter using the Gateway LR clonase (ThermoFisher Scientific, USA) following the protocol described in ref. 58 . OVA-beads were prepared by washing 3.0 μm unlabelled or YG-Fluoresbrite polystyrene microspheres (Polysciences, USA) in 25 mM sodium citrate, 25 mM sodium phosphate buffer, pH 5.0, and incubating beads with 20 mg mL −1 endotoxin-free chicken ovalbumin (OVA, InvivoGen, USA) overnight at 4°C on an end-over-end rotor, followed by three washes in sterile PBS (ThermoFisher). Bone marrow was isolated from sex and age-matched 6-16 week-old males and females. All animal manipulations were approved by the Geneva canton's Direction Générale de la Santé (authorizations GE/87/15 and GE/142/16) and performed in accordance to the guidelines of the animal research committee at the University of Geneva. BMDCs were generated by culturing bone marrow cells isolated from mouse femurs and tibias, where red blood cells had been removed by lysis in ammonium buffer (155 mM NH 4 Cl, 10 mM KHCO 3 , 0.1 mM EDTA), in DMEM (41965-039) supplemented with 10% endotoxin-free FCS, 50 μM β-mercaptoethanol, 1% sodium pyruvate, 1% penicillin/streptomycin and 20 ng mL −1 murine recombinant GM-CSF. Cells were used between 8 and 15 days of culture. The purity of each culture was determined to be > 85% CD11c + by flow cytometry. The DC 2114 cell line, derived from C57BL/6 mice with an H2-K b haplotype 18 was cultured in IMDM (31980) supplemented with 10% FCS, 50 μM β-mercaptoethanol, and 1% penicillin/streptomycin. Cells were tested for mycoplasma every six months. Maturation was performed by incubating cells in either 1 μg mL −1 LPS or 0.1 μM CpG (ODN 1826, InvivoGen) for 18 h. Seeding cells on coverslips tended to activate them to variable extents, thus unless otherwise indicated cells seeded on coverslips were matured with CpG. Lentivirus transduction was performed by centrifuging cells and viral particles at 5 MOI in complete medium supplemented with 8 μg mL −1 polybrene at 500 g at 37°C for 1 h. To produce DC 2114 cells stably expressing shRNA, cells were cultured in 1 μg mL −1 puromycin as of 2 days after transduction. Splenic OT-I cells were isolated from C57BL/6-Tg(TcraTcrb)1100Mjb/J mice 61 (Jackson) using the CD8α + negative selection mouse T cell isolation kit (Miltenyi Biotec, Germany) according to the manufacturer's instructions. Congenic CD45.1 + OT-I cells were obtained by breeding OT-I mice to the congenic CD45.1 strain B6. SJL-Ptprc a Pepc b /BoyJ 62 (Jackson) and similarly isolating CD8α + cells. Cell purity was verified to be >95% CD8α + , CD4 − by flow cytometry.
In vivo cross-presentation. Three to four-month-old female LysM-Cre ki/+ ; Stim1 fl/fl and otherwise wild-type Stim1 fl/fl littermates were anesthetized with isofluorane and injected with 50 μl of 1.0% OVA-beads/PBS on the left and 0.5% OVA-beads/PBS on the right footpads, or with only PBS on both footpads as control. Twenty four hours later the mice were retro-orbitally injected with 1 × 10 6 CD45.1 + OT-I cells labelled with 3.5 μM CFSE for 10 min and washed and resuspended in 50 μl sterile PBS. After 72 h, the mice were sacrificed and popliteal draining and axial/brachial non-draining lymph nodes were harvested in FACS buffer containing 2% FCS. Lymph nodes were mechanically crushed onto 40 μm cell strainers and the cellular filtrate blocked and stained as described above with anti-CD8α-PE/Cy7 anti-CD45.1-APC antibodies. The total number of CD45.1 + cells as well as their CFSE fluorescence was determined for 50,000 CD8α + cells. % Proliferation is defined at the percentage of cells showing diluted levels of CFSE fluorescence as compared to uninjected controls.
In vitro cross-presentation. BMDCs were seeded at 20,000 cells per well and DC 2114 at 10,000 cells per well in black/clear-bottom 96-well plates (Greiner, for BrDU-based assays) or in round-bottom 96-well plates (Corning, for CFSE assays), and allowed to adhere for 24 h. For 3:1 and 10:1 T cell:DC ratios, BMDCs were diluted accordingly. Cells were exposed to either OVAb at 20:1 ratio for 4 h or SIINFEKL peptide (InvivoGen) for 1 h at the given concentrations, before wells were washed in complete medium, and irradiated with 3000 cGy using a Gammacell 3000 Elan Ce 127 irradiator (Best Theratronics, Canada). 20,000 (for BMDC) or 10,000 (for DC 2114 ) OT-I cells were added to each well, and plates were incubated at 37°C/5%CO 2 for 72 h. T cell proliferation was assessed by BrDU incorporation using a Cell Proliferation ELISA, BrDU (Chemiluminescent) kit (Roche, Switzerland) according to the manufacturer's instructions and using a SpectraMAX Paradigm (Molecular Devices, USA) plate reader. Proliferation of OT-I cells labelled with CFSE was assessed as described above.
In vitro transwell migration assay. BMDCs were washed and resuspended in migration medium (1% BSA, 10% serum in phenol-free DMEM). 5 × 10 5 cells were seeded onto 8.0 μm pore cell culture inserts in a 24-well plate (Corning), and either control medium or medium supplemented with 10 −6 M chemoattractant (fMIFL, SDF-1α or CCL21) was added to the bottom chamber to initiate migration. Cells were incubated at 37°C for 24 h, after which inserts were removed and plates were placed on ice, 5 mM EDTA was added to each well, and nuclei were stained with 2.5 μM DRAQ5 (ThermoFisher). The total number of migrated DRAQ5 + cells was quantified. % Migration is defined as the total number of migrated cells in response to chemoattractant divided by the total number of migrated cells in response to control medium minus 1 times 100%.
Phagocytosis. For flow cytometry-based phagocytosis assays, OVA-coated YG-Fluoresbrite beads were added at a 20:1 ratio in duplicate for the indicated times. Cells were washed in FACS buffer and stained with anti-CD11c-APC antibodies as described above, and the mean fluorescence intensity (MFI) quantified for CD11c + cells. MFIs were normalized to the mean WT 4 h phagocytosis value performed on the same day. For microscopy-based phagocytosis assays cells were matured with CpG, exposed to 20:1 OVAb for 4 h, washed, fixed in 4% paraformaldehyde/PBS, permeabilized and blocked in 0.1% triton-X/1% BSA, stained with 0.6 μM TRITCphalloidin and 10 μg mL −1 Hoechst 33342 and visualized using confocal microscopy.
Ca 2+ imaging. BMDCs and DC 2114 cells seeded on Fluorodishes (WPI, USA) and, where indicated, matured as described above. Cells were loaded with 4 μM Fura-2-AM, 0.01% pluronic (ThermoFisher) in modified Ringer's 17 for 30 min at room temperature (RT). 340/380 nm excitation 510± 40 nm emission ratiometric imaging was performed at 37°C in modified Ringer's where Ca 2+ -free solution contained 1 mM EGTA instead of 2 mM CaCl 2 . Frames were acquired every 3 s. Ca 2+ microdomain/hotspot imaging was performed as described 17 except cells were loaded with 4 μM Fluo-8-AM (AAT Bioquest) for 30 min at 37°C, 30 min at RT and 2.5 μM BAPTA-AM for the last 10 min, in modified Ringer's containing 500 μM sulfinpyrazone. Simultaneous excitation at 488 and 543 nm and emission collection in two separate channels for green (Fluo8) and red (mCherry) was used. For quantification of periphagosomal Ca 2+ hotspots, images were averaged over 6 s and captured between 20 and 30 min after the addition of OVAb. Where applicable Xestospongin C (1 μM) was applied 20 min after exposure of BMDCs to OVAb and hotspots imaged at 30 min, while GSK7975A (a gift from Dr. Martin Lochner, University of Bern, 10 μM) was applied concurrently with OVAb and images taken after 30 min.
Western blotting. Cells were washed with ice cold PBS and lysates prepared as described 17 . Total protein was quantified using a BCA assay (ThermoFisher) according to the manufacturer's instructions. 50 μg per lane were loaded for ORAI1 gels, while 30 μg per lane were loaded for all others. 4-20% Mini-Protean TGX Precast gels (BioRad) and iBlot PVDF kits (ThermoFisher) were used for SDS-PAGE and transfer respectively. All antibodies were diluted in 3% milk/1%Tween/TBS, primary antibodies were incubated overnight and secondary antibodies incubated for 1 h. Membranes were visualized using Immobilon Western Chemiluminescent HRP Substrate (Millipore) and a PXi gel imaging system (Syngene, UK).
Electron microscopy. Classic transmission electron microscopy (TEM) and 3D Focused-ion beam scanning electron microscopy (FIB-SEM) were performed with the help of the Electron Microscopy Core Facility of the University of Geneva. TEM samples were prepared and imaged as described 17 . FIB-SEM was performed on a Helios NanoLab G3 microscope (FEI, Netherlands) with images acquired using the electron beam at 2 kV/0.2 nA, and milling sections with the ion beam at 30 kV/2.5 nA every 10 nm. Specimens were prepared using a protocol modified from 64 . Cells were fixed in 2.5% glutaraldehyde, 2% PFA in 2 mM CaCl 2 in 0.15 M cacodylate buffer (pH 7.4) (Ca-Caco) for 3 h on ice and washed 5× in cold Ca-Caco. Samples were incubated for 1 h on ice in 2% OsO 4 , 1.5% KFCN in Ca-Caco and washed 5× in cold ddH 2 O. Samples were then incubated in a freshly prepared and filtered 1% thiocarbohydrazide for 20 min at RT, washed 5× in RT ddH 2 O, incubated in 2% OsO 4 for 30 min at RT and then incubated in cold 1% uranyl acetate at 4°C overnight. Samples were then washed 5× in RT ddH 2 O and stained using Walton's lead aspartate method for 30 min at 60°C, and rinsed 5× in RT ddH 2 O. Samples were dehydrated in 2 × 7 min steps of cold 20, 50, 70, 90, 100 and 100% anhydrous ethanol, then cold anhydrous acetone for 10 min and RT acetone for 10 min. Samples were then embedded in hard epon resin by infiltrating with 25, 50 and 75% hard epon/acetone at RT for 2 h each, then 100% hard epon overnight. Fresh 100% hard epon was exchanged the next day incubated for 3 h. Epon was once more exchanged before polymerization for 48 h at 60°C. Cut and microtometrimmed blocks were mounted on aluminium pins using Pelco C100 cyanoacrylate glue and Pelco conductive silver paint (Ted Pella, USA). Samples were sputtercoated with gold for 20 s using a Q150T ES coater (Quorum Technologies, UK). pH measurements. Phagosomal pH was measured using FITC-coupled zymosan added at 20:1 ratio as described 65 except that calibrations (pH 4.0-9.0) were performed using five different stacks per condition for each calibration buffer rather than a single time-lapse acquisition. pHrodo Red Zymosan Bioparticles (Ther-moFisher) were washed 3× in sterile filtered 0.1 M NaCO 3 buffer (pH 8.0), 10 μg mL −1 Alexa-488-succinimidyl ester (SE) was added and particles were agitated for 4 h at RT. Unbound dye was removed by washing in 50, 33%, then 25% DMSO/ PBS and 3× in sterile PBS. Zymosan was then incubated in 20 mg mL −1 OVA overnight and washed 3× in PBS. Measurements, calibration and analysis were obtained using an identical protocol to FITC-coupled zymosan with the following exceptions: imaging was performed using alternate 555/590 and 488/530 nm Ex/ Em illumination. Zymosan were added to cells in complete medium and were washed in modified Ringer's after 30 min, and imaged 30 and/or 90 min after addition. Where applicable, 0.2 nM ConcA was added immediately before zymosan. Endosomal pH measurements were made as previously described 66 . Briefly, cells were loaded with with 1 mg mL −1 of FITC-and Alexa-647-labelled 40 kDa dextrans (Thermo) for 10 min at 37°C and extensively washed with cold PBS/1% BSA. Cells were then chased for different times and analysed by flow cytometry.
ROS measurements. Extracellular ROS were measured on cells seeded at 50,000 cells per well in black/clear bottom plates using the Amplex Red Ultra Hydrogen Peroxide/Peroxidase kit (ThermoFisher), according to the manufacturer's instructions. Intracellular ROS were measured on cells similarly plated. Dihydroethidium (DHE) (30 μM) was loaded for 1 h in complete medium at 37°C before washing in modified Ringer's. For both assays fluorescence was measured every 2 min for 90 min at 37°C after addition of the indicated stimuli (100 nM PMA, 20:1 zymosan or 20:1 OVAb) using a SpectraMAX Paradigm plate reader, and is reported as the maximum fluorescence value normalized to the average baseline value over the first 6 min. Intraphagosomal/intracellular ROS was measured on cells plated on Fluorodishes using OxyBurst/Alexa-568-coupled beads. 3.0 μm Polybead Amino Microspheres (Polysciences) were washed in 3× in sterile filtered 0.1 M NaCO 3 buffer (pH 8.0), 20 μg mL −1 OxyBurst-H2DCFDA-SE (ThermoFisher) was added and particles were agitated for overnight at RT in tubes filled with nitrogen. Alexa-568-SE (ThermoFisher) was then added for 1 h. Unbound dye was removed as above; beads were coated with OVA as above and beads were stored under nitrogen. OxyBurst/Alexa568-OVA beads were added to cells at 20:1 in complete medium, were washed in modified Ringer's after 30 min, and imaged using alternate 555/590, 488/530 nm Ex/Em and brightfield illumination 30 and/or 90 min after addition. Where applicable cells were pre-incubated for 10 min with 10 μM DPI prior to stimulus addition. As OxyBurst leaked from phagosomes into the cytosol, brightfield images were used to demarcate cell borders and the ratio of whole-cell OxyBurst to whole-cell Alexa-568 fluorescence minus the basal (external bead) ratio calculated as an index of intracellular/phagosomal ROS.
Measurement of cytosol transfer. Measurement of cytosol transfer was performed as previously described 16 . Briefly, cells were loaded with 4 μM CCF4-AM probe (a FRET-sensitive cytosolic substrate of β-lactamase, ThermoFisher) for 1 h, washed with PBS and incubated at 37°C with 2 mg mL −1 β-lactamase for different time points. Reactions were stopped with cold PBS and live, single, CD11c + cells were analysed on a BD Biosciences Fortessa flow cytometer by monitoring the increase in 450 nm fluorescence and loss of 535 nm emission fluorescence resulting from CCF4 cleavage.
Proteolysis assays. Phagosomal proteolysis was measured using a protocol modified from ref. 30 . 3.0 μm Polybead Amino Microspheres (Polysciences) were labelled with Alexa-568-SE as described above. Labelled beads were then washed in 25 mM sodium citrate, 25 mM sodium phosphate buffer (pH 5.0), and incubated with 20 mg mL −1 DQ-ovalbumin (DQ-OVA, ThermoFisher) overnight at 4°C on an end-over-end rotor, followed by three washes in sterile PBS (ThermoFisher). Cells were exposed to beads, imaged and analysed as described above for OxyBurst assays. Where applicable cells were pre-loaded for 30 min with 40 μM BAPTA-AM, and washed prior to stimulus addition, and 0.2 nM ConcA added just prior to beads.
Phagosomal OVA degradation assay. Phagosomal OVA degradation assay was performed as previously described 66 . Briefly, 3.0 µm Polybead Amino Microspheres (Polysciences) were covalently coated with OVA (0.2 mg mL −1 ) according to the manufacturer's instructions. BMDCs were pulsed for 15 min at 37°C and chased for different times at 37°C. Cells were lysed in 50 mM Tris-HCl (pH 7.4) supplemented with 150 mM NaCl, 1 mM DTT, 0.5% NP-40, and cocktail of protease inhibitors (Roche). For flow cytometry experiments, beads were stained with polyclonal anti-OVA (Sigma) and a secondary anti-rabbit Alexa488 (Thermo), and analysed on a BD Biosciences Fortessa flow cytometer.
Phagosome isolation for leucyl aminopeptidase activity. 1.5 × 10 7 BMDCs were exposed to 3 × 10 8 magnetic 3 µm OVA-coated beads (Polysciences) for 15 min on ice. Warm medium was added and cells were incubated for 20 min at 37°C. Cells were returned to ice and washed twice in cold PBS and once in cold IEB buffer containing protease inhibitor cocktail (PI, Sigma) (IEB: 10 mM HEPES, 1 mM EGTA, 25 mM KCL, 250 mM sucrose). Cells were scraped, centrifuged, and washed once in IEB/PI and counted. Cells were then centrifuged and resuspended in IEB/ PI at a concentration of 4 × 10 6 cells mL −1 , and passed 20x through a 20 µm clearance ball-bearing homogenizer. Homogenates were then transferred to a magnet and washed once in IEB/PI, once in IEB/PI + 4 mM MgCl 2 + 10 mM Na-ATP, and were then incubated on an end-over-end rotor for 15 min at 4°C. Homogenates were washed one final time on a magnet and resuspended in lysis buffer (50 mM Tris; 150 mM NaCl; 1% NP40) without protease inhibitors. Lysates were incubated on ice for 30 min prior to quantification using a BCA kit (ThermoFisher).
Phago-lysosome and phago-endosome fusion assay. Phago-lysosome fusion was measured using a protocol modified from 30 . 3.0 μm Polybead Amino Microspheres (Polysciences) were labelled with Alexa-488-SE and coated with OVA as described above. Twenty four hours prior to the assay cells were pulse labelled with 20 μg mL −1 Alexa-594-HA (ThermoFisher) in complete medium for 3 h, washed, and followed by a chase period of 24 h to allow the dye to accumulate in lysosomes. Alexa488-OVA beads were added to cells at 20:1 in complete medium, were washed in modified Ringer's after 30 min, and imaged using alternate 555/590, 488/ 530, 488/590 nm Ex/Em and brightfield illumination 30 and/or 90 min after addition. When applicable cells were pre-loaded for 30 min with 40 μM BAPTA-AM, and washed prior to stimulus addition and 0.2 nM ConcA added just prior to beads. The P-L fusion index is computed as the ratio of the FRET (488/590)/Alexa-488 (488/530) signal minus the basal (external bead) FRET signal, normalized to the total average cellular Alexa-594-HA fluorescence. Phago-endosome fusion assays were performed in a similar manner except that 1 μg mL −1 Alexa-594dextran of 10,000 MW (ThermoFisher) was added 15 min prior to beads, cells were washed 3× in complete medium prior to Alexa488-OVA bead addition and external beads were washed in modified Ringer's after 15 min. Where applicable, GSK7975A (10 μM) was added concurrently with beads.
Image analysis and statistics. All fluorescence image analyses were performed using ImageJ software (NIH) on maximum projections of at least 5 × 15 μm z-stacks per condition, except for IRAP quantification which were performed on a single confocal plane from each stack, and which were taken with a 100× objective. Semi-automated segmentation based on a single visually determined threshold for the pH/ROS/proteolysis insensitive wavelength did not fully separate all closely clustered-phagosomes particles, and thus single-phagosome measurements include an estimated 20-30% phagosome clusters. External (non-cell-associated) zymosan/ beads were used to estimate time 0 ratio values and then eliminated manually from the analysis. For IRAP quantification, the total periphagosomal IRAP signal (obtained from the bead segmentation mask, and measured in a single plane that captured most phagosomes within a stack at their mid-section) was normalized to the total IRAP cellular fluorescence, calculated from sum projections of the same stack. Alignment, segmentation and 3D reconstruction of FIB-SEM images was performed using Amira software (FEI). Quantification of ER-Ph MCS was conducted in single-blind and performed manually on 10 nm slices, at 30 slice intervals to avoid double counting of MCS. MCS were defined as areas where the ER was < 30 nm away from the phagosomal membrane. Phagosome cross-sections smaller than 1.5 µm in diameter were excluded from the analysis to avoid any undersampling bias of phagosomal membranes. MCS frequency was defined as the number of MCS (also called ER junctions) detected divided by the total number of phagosomes analysed, multiplied by 100 to obtain the average number of ER junctions per 100 phagosomes. Animal studies were not conducted blind, but at least N = 4 animals were tested for each condition, and, within each genotype, were randomly assigned to control and experimental groups. All statistical analyses were performed using Prism 6.0 software (GraphPad). Pairwise comparisons were made using a Mann-Whitney test for experiments based on phagocytosis, as phagocytic events do not follow a Gaussian distribution. A two-sided Student's t-test was used for all other pairwise comparisons, and, where F-test showed significantly difference variances, Welch's correction was applied. A Two-way ANOVA with Sidak's multiple comparisons test was used for repeated measures. N = number of independent experiments, at least three independent experiments were performed for all conditions. Data availability. The data that support the findings of this study are available from the corresponding author upon reasonable request.