Syntaxin 3, but not syntaxin 4, is required for mast cell–regulated exocytosis, where it plays a primary role mediating compound exocytosis

Mast cells (MCs) participate in allergy, inflammation, and defense against pathogens. They release multiple immune mediators via exocytosis, a process that requires SNARE proteins, including syntaxins (Stxs). The identity of the Stxs involved in MC exocytosis remains controversial. Here, we studied the roles of Stx3 and -4 in fully developed MCs from conditional knockout mice by electrophysiology and EM, and found that Stx3, and not Stx4, is crucial for MC exocytosis. The main defect seen in Stx3-deficient MCs was their inability to engage multigranular compound exocytosis, while leaving most single-vesicle fusion events intact. We used this defect to show that this form of exocytosis is not only required to accelerate MC degranulation but also essential to achieve full degranulation. The exocytic defect was severe but not absolute, indicating that an Stx other than Stx3 and -4 is also required for exocytosis in MCs. The removal of Stx3 affected only regulated exocytosis, leaving other MC effector responses intact, including the secretion of cytokines via constitutive exocytosis. Our in vivo model of passive systemic anaphylaxis showed that the residual exocytic function of Stx3-deficient MCs was sufficient to drive a full anaphylactic response in mice.


Mast cells (MCs) participate in allergy, inflammation, and defense against pathogens. They release multiple immune mediators via exocytosis, a process that requires SNARE proteins, including syntaxins (Stxs). The identity of the Stxs involved in MC exocytosis remains controversial.
Here, we studied the roles of Stx3 and -4 in fully developed MCs from conditional knockout mice by electrophysiology and EM, and found that Stx3, and not Stx4, is crucial for MC exocytosis. The main defect seen in Stx3-deficient MCs was their inability to engage multigranular compound exocytosis, while leaving most single-vesicle fusion events intact. We used this defect to show that this form of exocytosis is not only required to accelerate MC degranulation but also essential to achieve full degranulation. The exocytic defect was severe but not absolute, indicating that an Stx other than Stx3 and -4 is also required for exocytosis in MCs. The removal of Stx3 affected only regulated exocytosis, leaving other MC effector responses intact, including the secretion of cytokines via constitutive exocytosis. Our in vivo model of passive systemic anaphylaxis showed that the residual exocytic function of Stx3-deficient MCs was sufficient to drive a full anaphylactic response in mice.
Mast cells (MCs) 2 are key effectors of adaptive and innate immunity, and modulators of local inflammation (1). Armed with prominent electron-dense granules loaded with inflammatory mediators (2,3), and strategically located in the host peripheral tissues, they degranulate upon exposure to a variety of stimuli (4 -6). MC degranulation is one of the best examples of regulated exocytosis, where the preformed contents are stored in secretory granules and released only after a specific stimulus (7,8). The MC constitutes a premier system to study regulated exocytosis at high resolution due to their large granules and the predictable and almost complete degranulation after stimulation (9). Regulated exocytosis in MCs can use single-vesicle and compound exocytosis (10). In single-vesicle exocytosis, each secretory vesicle fuses independently with the plasma membrane. In sequential compound exocytosis, vesicles lying deeper within the cell fuse with vesicles already fused with the plasma membrane. In multigranular compound exocytosis, vesicles fuse homotypically with each other before fusing with the plasma membrane (11). MCs secrete products via constitutive exocytosis too. This involves the continuous traffic of vesicles from the Golgi apparatus to the plasma membrane, and the amount of secreted product depends on the rate of synthesis of the vesicle cargo (12). MCs also have secretory responses that are independent of exocytosis, such as secretion of eicosanoids, which are exported via plasma membrane transporters (13).
As any other form of exocytosis, the molecular machinery that mediates MC degranulation should involve SNARE (soluble N-ethylmaleimide-sensitive factor attachment receptor) proteins, including syntaxins (Stxs) (14). Stxs are a subset of SNARE proteins located on the target membrane, in this case the plasma membrane. Among all of the members of the Stx family, Stx1a, -1b, -2, -3, -4, and -11 mediate exocytosis in different mammalian cells (15,16). The helical SNARE domain of an exocytic Stx associates with the SNARE domains of SNAP23/25 (synaptosomal-associated protein 23 or 25) and VAMP (vesicle-associated membrane protein) to form the SNARE complex, which is essential for fusion of the vesicle membrane and plasma membrane (17). The formation and function of this complex requires the coordinated participation of members of the Munc (mammalian isoform of the uncoordinated gene of Caenorhabditis elegans) 13, Munc18, complexin, and synaptotagmin families (18).
We have described that in mature MCs, regulated exocytosis depends on synaptotagmin-2 (19), Munc13-2 and -4 (20), and Munc18-2 (21,22). About the Stxs involved in this process, it has been accepted for more than a decade (23,24) that Stx4 is an essential participant in MC exocytosis, based on multiple studies on RBL-2H3 cells (25)(26)(27)(28)(29), rat peritoneal MCs (30), and human intestine-derived MCs (31). However, these findings have been questioned by others (32)(33)(34). Stx3, a SNARE protein that is localized both in the plasma and granule membranes (35), has also been shown to play an important role in exocytosis in RBL-2H3 cells (36 -39) and human intestine-derived MCs (40). The main caveats with these studies, some with opposite findings, are that the interference with expression or function of Stx was only partial, few tested Stx3 and -4 in the same experimental paradigms, none dissected the effects on compound exocytosis, and most relied on cell lines. We found that interfering with expression of synaptotagmin-2 (41) and Munc18-1 (42,43) in RBL-2H3 cells had different effects on exocytosis than their elimination in mature MCs (19,22). Thus, we chose to interrogate the impact of complete removal of Stx3 and Stx4 on single-vesicle and compound exocytosis in fully developed MCs.
To study the roles of Stx3 and Stx4 in vivo, we created conditional KO (cKO) mice for both genes. We could not find any role for Stx4 in MC exocytosis in any of our assays. Removal of Stx3 decreased the amount and rate of exocytosis mainly because multigranular compound exocytosis was almost completely eliminated, leaving single-granule secretion mostly intact. This defect affected exclusively the secretion of granule contents and not the secretion of cytokines or eicosanoids. Unlike what we have seen in the absence of Munc13-4 (20) and Munc18-2 (22), elimination of Stx3 in MCs did not affect the anaphylactic response.

Expression of Stxs in MCs and generation of conditional KO mice
Given the importance of MC degranulation in inflammatory responses, we decided to study the Stxs that mediate this process. We used the expression pattern of the exocytic Stxs in C57BL/6J (B6) MCs to choose the Stxs for our first study. We noted that the main isoforms present in peritoneal MCs were Stx3 and -4 ( Fig. 1A) and targeted both for deletion.
For Stx3, we used an ES clone with a gene trap in intron 1 of Stx3 (Fig. S1A). This cassette, which contains a splice acceptor site, interrupts the transcription of Stx3. Heterozygous mice generated from ES cells with this mutant allele were crossed with FLP-expressing mice to invert the cassette, allowing Stx3 expression but retaining two flanking loxP sites ("floxed" or F allele) that render it susceptible to another inversion by Cre recombinase. This final inversion blocks the expression of Stx3 and "locks" the allele (i.e. it cannot be inverted again). To target Stx4 (Fig. S1B), we introduced by homologous recombination two loxP sites flanking exons 1-4 of Stx4 (F allele). Cre recombination removes the translation initiation codon. In both cases, Cre recombination was predicted to eliminate expression of the targeted gene. We crossed Stx3 F/F and Stx4 F/F mice with B6.C-Tg(CMV-cre)1Cgn/J mice to generate germ line (global or constitutive) deletants (Ϫallele) or with Tg(Cma1-cre) ARoer mice to remove expression only in MCs (⌬ allele). We have shown the efficacy of this strategy before (20).
Immunoblots from peritoneal cell-derived MCs (PCMCs) and other tissues confirmed the reduced expression of the targeted Stx in ϩ/Ϫ animals and their normal expression in F/F mice. They also confirmed that recombination by Cre activity removed expression of the targeted gene and demonstrated the specificity of the deletion in ⌬/⌬ mice only in MCs (Fig. 1, B and C). Because it A, quantitative RT-PCR of B6 peritoneal MCs for Stx1a, -1b, -2, -3, -4, and -11 relative to ␤-actin. ND, not detected; n ϭ 3; bar, mean; error bar, S.E. Representative immunoblots of PCMC and kidney lysates are shown for Stx3 (B) and Stx4 (C). ␤-Actin was used as loading control. Bar graph insets, densitometry relative to ␤-actin and ϩ/ϩ.

Syntaxin 3 in mast cell degranulation
has been shown that sometimes the expression of an Stx can affect the expression of its cognate Munc18 protein (44), we tested how expression of Stx3 affected the expression of Munc18Ϫ2 and found no differences among MCs from all Stx3 mutant mice (Fig.  S2).

Absence of Stx3, but not Stx4, hinders the extent and kinetics of exocytosis in MCs
To determine the effects of lacking Stx3 or Stx4 on MC exocytosis at high resolution, we measured plasma membrane capacitance (C m ) in single peritoneal MCs using the whole-cell patch clamp configuration (9,10,45,46). In this assay, the intracellular dialysis of GTP␥S and Ca 2ϩ through the patch pipette induces almost complete MC degranulation, and the resolution is such that single fusion events can be identified (9,45). C m is proportional to the area of the plasma membrane. During exocytosis, a secretory vesicle incorporates its membrane into the plasma membrane, and this increase in area of the plasma membrane is recorded as an increase in C m (⌬C m ) (45). In all of the studied MCs, the mean baseline C m was 5.7 Ϯ 0.5 pF (mean Ϯ S.E.), and the average final intracellular [Ca 2ϩ ] measured using Fura-2 fluorescence was 434 Ϯ 36 nM, and we found no differences in these two values among all Stx3 and Stx4 genotypes. C m recordings in Stx3 ϩ/ϩ MCs with a pipette loaded with an intracellular solution with no GTP␥S showed that intracellular access by itself did not induce exocytosis ( Fig. 2A, black trace). The ⌬C m among Stx3 ϩ/ϩ , Stx3 ϩ/Ϫ , and Stx3 F/F MCs was almost identical, but it was significantly decreased in Stx3 ⌬/⌬ MCs (Fig. 2, A and B). Curves were normalized (Fig. 2C) to calculate the maximum rate of exocytosis (rate between 40 and 60% of total ⌬C m ). Although they achieved the same total ⌬C m , Stx3 ϩ/Ϫ MCs did so at a significantly lower rate compared with Stx3 ϩ/ϩ MCs, a gap that was even larger between Stx3 F/F and Stx3 ⌬/⌬ MCs (Fig. 2D).
In contrast to the clear phenotype we found in Stx3-deficient MCs, deletion of Stx4 had no effects on the magnitude and speed of exocytosis (Fig. 2, E-H). frequency distribution of C m step sizes between 1 and 15% of total ⌬C m . Signals Ͻ 4 fF have been removed for clarity. White line, mean; box, 25th-75th percentile; whiskers, 5th-95th percentile. #, p Յ 0.05; †, p Յ 0.01; *, p Յ 0.001; all compared with ϩ/ϩ MCs unless otherwise specified.

Syntaxin 3 in mast cell degranulation
We also measured the time between cell access and initial rise in C m to test whether absence of expression of Stx3 or Stx4 affected the interval between stimulus and response. Unlike the significant delay we observed in Munc13-4 -deficient MCs (20), we found no alterations in the absence of these Stxs (Fig. 2, I and J).
The slower rate of exocytosis in Stx3-deficient MCs could be due to a reduction of the number of granules being exocytosed per unit of time, to a decrease in the size of the vesicles fusing with the plasma membrane, or both. Each "step" in the C m curve represents the fusion of a single vesicle, and the size of each "step" is proportional to the surface area (and thus the volume) of the fusing vesicle (45). By comparing MCs from Stx3 F/F and Stx3 ⌬/⌬ mice, we noted a significant shift to exocytosis of smaller-sized vesicles in the absence of Stx3 (Fig. 2K). We noted that ϳ70% of steps in Stx3 ⌬/⌬ MCs were Յ12 fF compared with ϳ35% in Stx3 F/F MCs. In absolute numbers, the number of steps Յ8 fF in recordings from Stx3-deficient MCs (41 Ϯ 7; mean Ϯ S.E.; n ϭ 36 cells from 8 animals) was very close to what we found in Stx3-sufficient MCs (35 Ϯ 6; n ϭ 36 cells from 6 animals; p ϭ 0.5), whereas the numbers of events Ն40 fF were 4 Ϯ 3 and 49 Ϯ 11, respectively (p Ͻ 0.001).
The frequency distribution of step sizes was unaffected by the absence of Stx4 (Fig. 2L). We did not study the haploinsufficient Stx4 ϩ/Ϫ MCs in any experiment, given that full deficiency of Stx4 did not result in any electrophysiological abnormality. These results show that Stx3 is partially responsible for MC exocytosis and that the residual exocytosis after its deletion proceeds at a slower pace and is composed mainly of singlevesicle fusion events.

Absence of Stx3, but not Stx4, alters the ultrastructural changes associated with exocytosis in MCs
Measurement of C m is blind to the intracellular events that take place during MC degranulation. Consequently, we studied EM profiles of peritoneal MCs stimulated with a combination of the diacylglycerol-analog phorbol 12-myristate 13-acetate (PMA) and the Ca 2ϩ ionophore ionomycin. Qualitatively, stimulated Stx3-deficient MCs did not undergo the same amount of exocytosis as stimulated Stx3-sufficient MCs (Fig. 3A), whereas Stx4-sufficient or -deficient MCs were indistinguishable (Fig.  3B). To quantify those alterations, we studied cell profiles using stereology and also measured the electron lucency of the MC granules. When MCs are activated, they lose the electron den-

Syntaxin 3 in mast cell degranulation
sity of their granules secondary to hydration and loss of granule contents (47,48), and we quantified this change as gain in electron lucency (20). Compared with unstimulated controls, stimulated Stx3 ϩ/ϩ MCs have a significant increase in electron lucency of their granules (right shift), whereas the signal from stimulated Stx3 ⌬/⌬ MCs was intermediate between those two groups (Fig. 3C). On the other hand, the response to activation was indistinguishable between Stx4 ⌬/⌬ and Stx4 ϩ/ϩ MCs (Fig.  3F). The loss of electron-dense granules during degranulation can be quantified as the fraction of cytoplasm occupied by dense granules (volume density (Vv)) (49,50). Once again, stimulated Stx3deficient MCs behaved more like naive Stx3-sufficient MCs than stimulated Stx3-sufficient MCs (Fig. 3D), whereas the presence or absence of Stx4 made no difference (Fig. 3G).
The most remarkable morphological alteration in Stx3-deficient MCs was the almost complete absence of intracellular multigranular compartments after stimulation. In fact, this is the only measurement in which we found a partial defect in haploinsufficient Stx3 ϩ/Ϫ MCs (Fig. 3E). Deletion of Stx4 had no effect on this structural outcome (Fig. 3H). Because of the absence of a phenotype in Stx4-deficient MCs, Stx4 ϩ/Ϫ MCs were not studied. Altogether, these data suggest that although only a fraction of total MC exocytosis depends on Stx3, almost all MC compound exocytosis requires this protein.

Absence of Stx3, but not Stx4, affects selectively MC-regulated exocytosis
We wanted to address whether the defective MC degranulation we observed in Stx3-deficient MCs was linked to abnormalities in other MC effector responses and whether Stx4 was important for any MC secretory response at all. We measured the secretion of products that depend on regulated exocytosis, constitutive exocytosis, and nonexocytic transport (19,20). PCMCs were stimulated via Fc⑀RI or with PMA/ionomycin (PI). We quantified the amount of histamine and the fraction of total cell ␤-hexosaminidase secreted after stimulation. These two compounds are preformed, are stored in MC secretory granules, and are secreted via regulated exocytosis. To address constitutive exocytosis of newly synthesized cytokines, we measured the secretion of TNF␣ and IL4, and for responses independent of exocytosis, we measured the secretion of PGD 2 and LTC 4 . Given that we had not detected any differences between Stx4 ϩ/ϩ and Stx4 F/F mice in all previous experiments, we included only Stx4 F/F and Stx4 ⌬/⌬ mice in this section.
In the ␤-hexosaminidase secretion assay, we detected the expected bell-shaped response curve to increasing amounts of antigen (19,20,22) in Stx3-sufficient PCMCs. Haploinsufficiency or homozygosity for the "floxed" allele of Stx3 did not affect this response, whereas Stx3 deficiency significantly impaired ␤-hexosaminidase release (Fig. 4, A and C). This defect was corrected once we used PI as a stronger stimulus (Fig. 4C). The defect in secretion of MC mediators released by regulated exocytosis and its correction under strong stimuli were confirmed when we measured histamine release in MCs activated via IgE-Fc⑀RI (Fig. 4D) and PI (Fig. 4E). Despite that, the secretion of mediators released by constitutive exocytosis (Fig. 4, F and G) or transmembrane transporters (Fig. 4, H and I) was not affected by the deletion of Stx3. TNF␣, IL4, PGD 2 , and LTC 4 were undetectable in supernatants of unstimulated cells (not shown).
Independent of the mediator measured or the stimuli employed, we could not detect any secretory abnormality induced by the removal of Stx4 (Fig. 4, B-I). Because we did not detect any differences between Stx4-deficient and Stx4-sufficient MCs in any of the electrophysiology, EM, and secretion assays, we did not perform any in vivo experiments with the Stx4 mutant animals.

Stx3 deficiency does not affect MC numbers, distribution, differentiation, or structure
Defective exocytosis could also be caused by morphological or developmental anomalies in MCs (e.g. improper granule biogenesis or failure to exteriorize surface receptors) (51,52). We studied MCs from all Stx3 mutant mice ( Fig. 5 and Table 1). We found almost identical densities of dermal MCs and similar fractions, numbers, and metachromasia of MCs in peritoneal lavages. Flow cytometry assays showed a similar fraction of CD117 ϩ /Fc⑀RI␣ ϩ double-positive cells in peritoneal lavages and in PCMCs after 2 weeks in medium enriched with IL3 and stem cell factor. In these two assays, the fluorescence intensity for CD117 and FcεRI␣ in nonpermeabilized cells is proportional to the surface expression of these two receptors, and they were unchanged. Stereology of EM MC profiles revealed similar cell surface density (Sv cell) and cell profile area in all genotypes, indicating that the surface complexity, size, and shape of the cell profiles were not affected. Also, the fact that both Vv granule and Sv granule were similar indicates that there were no changes in the number, size, and shape of the MC granules. All of these parameters were also unchanged in Stx4 ϩ/ϩ , Stx4 ϩ/Ϫ , Stx4 F/F , and Stx4 ⌬/⌬ mice (not shown).
Therefore, the changes in the preceding ex vivo assays cannot be explained by a structural abnormality in the absence of Stx3, and any phenotype detected in subsequent in vivo experiments cannot be attributed to changes in the number or distribution of MCs in the mutant mice.

Stx3 deficiency does not alter MC-dependent anaphylactic responses
We have shown in mice with defective MC exocytosis that there is a correlation between the severity of the exocytic defect and protection in a model of passive systemic anaphylaxis (20,22). In this model, the degree of core body heat loss is proportional to the severity of the anaphylaxis (53)(54)(55). The lack of response in MC-deficient (Wsh) mice confirmed the MC dependence of this reaction (Fig. 6A, gray line), and the antigen challenge by itself did not induce hypothermia (Fig. 6A, black line). We found no differences in the hypothermic response among all Stx3 genotypes (Fig. 6, A and B), despite histologic evidence that degranulation of Stx3 ⌬/⌬ MC in connective tissues was partially deficient (Fig. 6, C and D).

Discussion
The Stx responsible for MC degranulation was originally identified as Stx4 (23,

Syntaxin 3 in mast cell degranulation
RBL-2H3 cells (25). Then Stx4 was shown to form SNARE complexes with SNAP23 in RBL-2H3 cells (25,26,29,56), mouse bone marrow-derived MCs (BMMCs) (57), and human intestine-derived MCs (31). In this last study, Sander et al. (31) found that treatment of permeabilized cells with anti-Stx4 antibodies decreased significantly their release of histamine after stimulation via Fc⑀RI or with PI (31). A similar experiment was performed by Salinas et al. (30) in permeabilized rat peritoneal MCs. They found that incubation with anti-Stx4 antibodies diminished histamine secretion induced by exposure to Ca 2ϩ and GTP␥S (30). Woska et al. knocked down Stx4 expression in RBL-2H3 cells using siRNAs and observed a significant reduction in IgE-dependent ␤-hexosaminidase release (27). Also, using an Stx4 siRNA in RBL-2H3 cells, Liu et al. (28) showed that reductions in Stx4 expression induced impairments in IgEand thapsigargin-dependent release of histamine and ␤-hexosaminidase, but not of IL4 and LTC 4 . More recently, Yang et al. (58) showed that a synthetic peptide based on the N terminus of Stx4 was able to inhibit lipid mixing in a liposome fusion assay that used the SNAREs Stx4, SNAP23, and VAMP2 or -8. The same Stx4 peptide inhibited degranulation when introduced into RBL-2H3 cells (58). Here we studied fully mature MCs in which expression of Stx4 was not partially, but completely, eliminated (Fig. 1).
We assessed exocytosis using three different methods, two of them extremely sensitive (C m measurements and EM). We also activated these MCs using Fc⑀RI-dependent, and receptor-independent (PI and Ca 2ϩ -GTP␥S), pathways. Despite all that, we found no abnormalities in exocytosis in Stx4-deficient MCs, whether single-vesicle or compound, regulated or constitutive (Figs. 2-4). These findings correlate with our previous observation that in the absence of Munc18-3, perhaps the exclusive functional partner of Stx4 (59 -61), MC exocytosis was normal (22). Thus, Stx4 could be added to the list of exocytic proteins (including synaptotagmin-2 and Munc18-1) in which there is a poor correlation between findings in cell lines and mature MCs. Another factor that could explain our disagreement with previous reports is that we used deletion of Stx4 instead of relying on antibodies, peptides, or siRNAs, all of which could have significant off-target effects on other Stxs or proteins that interact with them. Finally, our results suggest that efforts to target Stx4 expression or function to therapeutically control pathologic manifestations precipitated by MC degranulation (28,58) ought to be viewed with high skepticism.

Syntaxin 3 in mast cell degranulation
Although overexpression of Stx3 in RBL-2H3 cells did not affect degranulation (25), treatment of permeabilized human intestine-derived MCs with anti-Stx3 antibodies inhibited the stimulated release of chemokines (40). Others found that partial knockdown of Stx3 in RBL-2H3 cells and mouse BMMCs hindered degranulation but not chemokine secretion (37,38). In RBL-2H3 cells, the effects of Stx3 on degranulation depended on a specific interaction with Munc18-2 (38,39).
We removed expression of Stx3 in mature MCs (Fig. 1B). Based on our previous finding that deletion of Munc18-2 in fully developed MCs eliminated MC degranulation almost completely (22) and the proven physical and functional interaction between Munc18-2 and Stx3 in cultured cells (37,38), we were expecting a similarly severe exocytic defect in Stx3 ⌬/⌬ MCs. Applying the high sensitivity of C m measurements to peritoneal MCs, we identified a significant reduction in stimulated MC exocytosis. Nonetheless, despite what the results in cultured cells suggested, the defect in Stx3-deficient MCs was not absolute, as we recorded ϳ40% of residual exocytosis in Stx3-deficient MCs (Fig. 2B). The difference in exocytic failure between our Munc18-2-deficient and Stx3-deficient MCs indicates that another Stx should also mediate this process. We have now shown definitely that it is not Stx4. Stx1a and Stx1b function almost exclusively in neurons, and we have been unable to detect their expression by immunoblots in mature MCs (not shown). We speculate that, among the Stxs known to interact with Munc18-2, Stx2 and Stx11 are the best candidates to explain the residual exocytosis. In other immune cells, Stx3 and Stx11 are redundant to each other in mediating Munc18-2-dependent exocytosis (62). In MCs, the effects of Stx2 or Stx11 deficiency on exocytosis have not been studied at high resolution yet.
The residual exocytosis we recorded in MCs lacking Stx3 happened at a markedly slower rate compared with controls (Fig. 2D). It has been postulated that one mechanism used by MCs to accelerate exocytosis is compound exocytosis (63). Sequential compound exocytosis shortens the distance (and thus the time required for fusion) between granule and target membrane, and multigranular compound exocytosis allows the content from multiple granules to be released in a single fusion event with the plasma membrane (11). It is estimated that the membrane from a single MC granule would increase C m by ϳ7 fF (64). When we measured the contribution of single-capacitance steps to ⌬C m , we found that MCs from Stx3 ⌬/⌬ mice had a clear deviation toward small steps, whereas large steps were almost completely absent (Fig. 2K). These findings had a perfect correlate in our morphologic studies, in which we observed that activated Stx3 ⌬/⌬ MCs had only a partial defect in quantitative EM markers of exocytosis (Fig. 3, C and D), whereas there was virtually an absence of multigranular compartments (Fig. 3E). There was also a close correlation between the only two manifestations of haploinsufficiency in Stx3 ϩ/Ϫ MCs; a decrease in

Syntaxin 3 in mast cell degranulation
the Vv of compound compartments (Fig. 3E) was associated with slower kinetics of exocytosis (Fig. 2D). These results confirm that the long-held hypothesis that MCs use compound exocytosis to accelerate their rate of exocytosis (10) is correct. The possible role of Stx3 in granule-to-granule fusion was suggested by its localization on the granule membrane (34 -36, 57). Here, we show that removal of Stx3 results in a nearly complete failure of this process. The severe defect in MC compound exocytosis is not unique to the absence of Stx3. We have seen the same in MCs deficient in Munc18-2 (22) and . Thus, it seems that the homotypic fusion step of multigranular compound exocytosis requires Stx3, Munc13-4, and Munc18-2.
On the other hand, whereas deletion of Munc18-2 removed almost all traces of single-vesicle exocytosis as well as compound exocytosis, removal of Munc13-4 and Stx3 left a residual response composed almost entirely of single-vesicle fusion events. In the case of Munc13-4, the additional deletion of Munc13-2 eliminated the remaining exocytosis (20). For Stx3, the number of steps reflecting single-vesicle exocytosis (Յ8 fF) was similar between Stx3-deficient and Stx3-sufficient MCs, but fusion of single vesicles alone was insufficient to achieve normal levels of total exocytosis, even at a slower rate. Therefore, the heterotypic fusion between granule membrane and plasma membrane requires mainly Munc18-2 and Munc13-4 and probably an Stx other than Stx3. Moreover, we show for the first time that the importance of compound exocytosis in MCs is not only to help achieve faster secretory kinetics, but also that it is required for full degranulation (Fig. 2).
Although we saw a severe impairment in compound exocytosis in the absence of Munc18-2, the effects of removing Stx3 on this process are not indirectly due to alterations in the expression of Munc18-2 (Fig. S2). The lack of an effect of absence of Stx3 on Munc18-2 expression could be due to the presence of other Munc18-2-interacting Stxs in MCs. For example, in cytotoxic T-lymphocytes, disruption of expression of Stx11 does not alter expression of Munc18-2 (65). On the other hand, the fact that removing Stx3 affects compound but not single-vesicle exocytosis, whereas deleting Munc18-2 affects both, mitigates against the possibility that all of the effects of Munc18-2 are mediated by Stx3.
Similar to what we observed after removing synaptotagmin-2 (19) and , Stx3 deficiency affected mainly regulated exocytosis, leaving other responses intact, including constitutive exocytosis (Fig. 4). This makes sense, given that compound exocytosis has been described as a mechanism in regulated exocytosis but not in constitutive exocytosis (66). The normal secretion of PGD 2 and LTC 4 obtained from Stx3 ⌬/⌬ MCs confirms that these cells can be stimulated and rules out faulty activation as an explanation for the secretory defect. The secretion of histamine and ␤-hexosaminidase was reduced in Stx3-deficient MCs as we observed in Munc13-4 -and Munc18-2-deficient MCs (20,22), but unlike in the MCs from the two mutants, the secretion of these two mediators was rescued in Stx3-deficient MCs by using a stronger stimulus (Fig. 4). This correlates perfectly with the fact that the defects observed in ⌬C m (Fig. 2) and EM stereology (Fig. 3) in MCs lacking Stx3 were also only partial, whereas those we reported in the absence of Munc13-4 (20) and Munc18-2 (22) were almost complete, demonstrating once again the complementarity of electrophysiology, stereology, and secretion assays in MCs.

Syntaxin 3 in mast cell degranulation
The defective exocytosis we recorded in Stx3 ⌬/⌬ MCs had no impact on the development and distribution of MCs in the mutant animals ( Fig. 5 and Table 1). This allowed us to proceed with whole-animal experiments. The partial exocytic defect we identified in isolated Stx3-deficient MCs tightly correlated with the histologic assessment of MC degranulation in Stx3 ⌬/⌬ mice subjected to a model of systemic anaphylaxis (Fig. 6, C and D). Despite that, this partial defect did not affect the hypothermic response, our main outcome (Fig. 6, A and B). This agrees with our findings on Munc13-4 and Munc18-2 mutant mice in which only an absolute deficiency in MC exocytosis had an impact on anaphylaxis outcome, whereas partial defects in expression had no effects on this allergic response (20,22).
To generate Stx3 cKO mice, we obtained ES cells from EUCOMM (clone EUCE320f12). These 129/Ola ES cells contain a gene trap flanked by inversely oriented pairs of the heterotypic FRT/F3 and lox511/loxP recombinase target sequences (68). The cassette, located in intron 1 of the Stx3 gene (GRCm38; Chr19:1806110) includes the following, from 5Ј to 3Ј: a long terminal repeat, an FRT, an F3, a loxP, a lox511, a splice acceptor, ␤-gal/neomycin phosphotransferase fusion gene, a polyadenylation site, a loxP, a lox511, an FRT, and an F3. The mechanism relies on two-directional site-specific recombination systems: FLPe/FRT and Cre/loxP. The former inverts the gene trap from its sense (mutagenic) orientation to antisense, allowing expression of Stx3 (F allele); the latter reinduces the mutation by the reinversion of gene trap to its sense orientation, stopping expression of Stx3 (⌬ and Ϫ alleles). We injected the ES cells into B6 blastocysts, established a line, and backcrossed it into B6 until all 105 SNPs from a speed-congenic scanning confirmed a pure B6 background. Using FLPe-mediated recombination by crossings with B6.129S4-Gt(ROSA)26Sor tm1(FLP1)Dym /RainJ mice, we recovered Stx3 expression (Stx3 F/F ), leaving the allele susceptible to future inactivation via Cre recombinase.
To create an Stx4 cKO, we engineered a targeting vector to insert a loxP upstream of exon 1 (GRCm38; Chr7:127841278) of the mouse Stx4 gene by homologous recombination. A second loxP site, followed by the phosphoglucokinase promoter-driven neomycin resistance gene (PGK-Neo) flanked by two FRT sites, was introduced in intron 4 (GRCm38; Chr7:127842997). The herpes simplex virus thymidine kinase gene was introduced outside the homology region. The construct was electroporated into B6 ES cells, and mutant mice were produced and selected as described (20). We crossed mice with confirmed germ line transmission of the mutant Stx4 allele with B6.129S4-Gt(ROSA)26Sor tm1(FLP1)Dym /RainJ mice to remove PGK-Neo and leave exons 1-4 susceptible to Cre recombination (F allele).
We used littermates as controls in all experiments and included adult animals of both sexes. All studies were carried out using animal protocols approved by the Institutional Animal Care and Use Committee of the University of Texas M.D. Anderson Cancer Center.

MC harvesting, cultures, secretion assays, flow cytometry, and cell sorting
After euthanasia, we lavaged the peritoneal cavity as described (70). We counted cells in a Neubauer chamber and cytospins stained with Wright-Giemsa and toluidine blue (pH 0.5). The peritoneal lavage cells were washed with PBS and processed for different assays. For PCMCs, we resuspended the cells in medium supplemented with rmIL-3 (5 ng/ml) and stem cell factor (15 ng/ml; both from R&D Systems) and cultured them for 2 weeks (37°C, 5% CO 2 ) with biweekly medium exchanges. For secretion assays, 3 ϫ 10 4 PCMCs were incubated for 5 h with 100 ng/ml SPE-7 anti-DNP IgE (Sigma-Aldrich), washed, and stimulated with DNP-HSA or PMA/ionomycin to assess secretion of ␤-hexosaminidase, histamine, LTC 4 , and PGD 2 at 30 min, and of TNF␣ and IL4 at 6 h as described (20,22). For flow cytometry, we incubated peritoneal MCs and PCMCs with 200 ng of anti-mouse CD117 phycoerythrin-cyanine-7 and 200 ng of anti-mouse Fc⑀RI␣ Alexa Fluor 647 (both from eBioscience) in 100 l of PBS at room temperature for 25 min, washed them in PBS twice, and then analyzed them (LSRII; BD Biosciences), recording the number of CD117 ϩ /Fc⑀RI␣ ϩ cells and their mean fluorescence intensity. For sorting, we labeled the cells as above and collected the CD117 ϩ /Fc⑀RI␣ ϩ double-positive cells (BD FACSAria).

Histology
We harvested ears and lips and fixed them in 4% paraformaldehyde (pH 7.0) overnight at 4°C and processed for histology. Paraffin-embedded 5-m sections were labeled with 1:1000 FITC-avidin and 1:10,000 Hoechst 33342 (Molecular Probes, Life Technologies) (21). We identified MCs as Hoechst ϩ nuclei surrounded by FITC-avidin ϩ granules and identified the dermis as the area between the epidermis and the skeletal muscle or auricular cartilage and then reported the number of MCs per area of dermis (2). OCT-embedded 5-m sections were stained with toluidine blue (pH 0.5), and we counted an MC as degranulated if Ͼ50% of its metachromatic granules were visualized outside the cell (22).

EM and stereology
We activated peritoneal MCs by resuspending peritoneal lavage cells in 1 ml of 300 nM PMA and 1.2 M ionomycin in PBS. After 5 min, we stopped exocytosis and fixed the cells by adding 2 ml of cold buffered 2.5% glutaraldehyde on ice for 15 min; the fixation continued at room temperature for 2 h. Resting cells were fixed the same way. After washings in PBS, we resuspended the fixed cells in 0.1 M sodium cacodylate in PBS. The processing and imaging of the cells was done as described (2). We analyzed the images on STEPanizer (71). We used point counting to determine the Vv values of dense granules and multivesicular compound granules and used line intersections with plasma membrane and granule membranes to obtain Sv values (72). The relative electron lucency of MC granules was analyzed as described (20). In short, with a circle-cycloid stereological grid, random points were captured to generate a 0 -255 linear gray scale unique to each image, and the digital grayness of circles with an area of 0.0366 m 2 within randomly selected granules was compared against this scale. Granules with relative electron lucency of Ͻ120 were categorized as dense.

Electrophysiology
We obtained single-cell capacitance recordings as described before (20,22) with slight variations. We resuspended peritoneal cells in external solution (137 mM NaCl, 2.6 mM KCl, 10 mM HEPES, 2 mM CaCl 2 , 0.5 mM MgCl 2 , 0.4 mM MgSO 4 , 10 mM glucose, 0.44 mM KH 2 PO 4 , 0.34 mM Na 2 HPO 4 , pH 7.3, 0.310 osmol). Whole-cell recordings from individual MCs were made using 5-6-megaohm SYLGARD-coated patch pipettes. The internal solution (135 mM potassium gluconate, 10 mM HEPES, 7 mM MgCl 2 , 3 mM KOH, 0.2 mM Na 2 ATP, 0.05 mM Li 4 GTP␥S, 2.5 mM Cs 2 -EGTA, 7.5 mM Ca-EGTA, 0.1 mM Fura-2, pH 7.21, 0.302 osmol) defined intracellular [Ca 2ϩ ] and induced degranulation. Cytoplasmic [Ca 2ϩ ] was measured ratiometrically with Fura-2, and Ca 2ϩ calibration constants were determined in vitro using a multipoint calibration kit (Thermo Fisher Scientific/Molecular Probes) and analyzed in IgorPro (WaveMetrics). For recordings of membrane capacitance (C m ), membrane conductance (G m ) and series conductance (G s ), an 800-Hz sinusoidal, 30-mV peak-to-peak stimulus was applied around a holding potential of Ϫ70 mV, and the resultant signal was analyzed using the Lindau-Neher technique (73). For each 100-ms sweep, the average value was recorded, yielding a temporal resolution for C m , G m , and G s of ϳ7 Hz. Cells selected for analysis met the criteria of G m Յ 1,200 pS, G s Ն 35 nS, and steady-state intracellular [Ca 2ϩ ] ϳ400 Ϯ 100 nM. ⌬C m , rates of ⌬C m from 40 to 60% of total ⌬C m , time interval between cell access and exocytic burst, and size and number of C m steps between 1 and 15% of total ⌬C m were obtained as described (22).

Passive systemic anaphylaxis
We sensitized 15-18-week-old mice with 10 g of mouse anti-DNP IgE (SPE-7; Sigma-Aldrich) in 200 l of PBS i.p. The next day, we challenged the mice with 500 g of DNP-HSA in 200 l of PBS i.p. We monitored the basal core body temperature over time with a rectal thermometer probe (Sper Scientific), euthanized the mice at 90 min, and harvested the lips for histology.

Statistical analysis
For continuous variables, we first tested for normality with D'Agostino's K2 test. For normal data, we first compared the means of all groups by analysis of variance, and if a significant difference was found, we applied Tukey's honest significant difference test for multiple pairwise comparisons, Dunnett's test for multiple comparisons against the control group, or Student's t test for single comparisons. For nonnormal data, we first compared all groups using a Kruskal-Wallis H test and followed any significant result with Dunn's test for multiple comparisons or the Mann-Whitney U test for single comparisons. For categorical data, we used Pearson's 2 test or Fisher's exact test. We analyzed frequency distributions with the Kolmogorov-Smirnov test. Significance was set at p Ͻ 0.05.