The beta- and gamma-isoforms of type I PIP5K regulate distinct stages of Ca2+ signaling in mast cells.

Crosslinking of IgE receptors by antigen initiates Ca2+ mobilization in mast cells by activating phospholipase-C gamma-mediated hydrolysis of phosphatidylinositol-4,5-bisphosphate [PtdIns(4,5)P2]. The resulting inositol 1,4,5-trisphosphate-mediated Ca2+ release from the endoplasmic reticulum (ER) activates store-operated Ca2+ entry, which is necessary for exocytotic release of inflammatory mediators. To investigate roles for PtdIns(4,5)P2-synthesizing isozymes of the type I phosphatidylinositol 4-phosphate 5-kinase family (PIP5K-I) in mast cell signaling, we compared the ectopic expression of wild-type and catalytically inactive PIP5K-I beta in RBL-2H3 mast cells. Surprisingly, both antigen and thapsigargin-stimulated Ca2+ influx were reduced by overexpression of active PIP5K-I beta, whereas antigen-stimulated Ca2+ release from ER stores was unaffected. Consistent with these results, Ca2+ entry stimulated by antigen or thapsigargin was enhanced by expression of a plasma-membrane-associated inositol polyphosphate 5'-phosphatase, whereas antigen-stimulated Ca2+ release from stores was reduced. To investigate the role of PIP5K-I gamma in antigen-stimulated Ca2+ mobilization, we used bone-marrow-derived mast cells from PIP5K-I gamma(-/-) mice. Antigen-stimulated Ca2+ release from ER stores was substantially reduced in the absence of PIP5K-I gamma, but thapsigargin-mediated Ca2+ entry was unaffected. In summary, PIP5K-I gamma positively regulates antigen-stimulated Ca2+ release from ER stores, whereas PIP5K-I beta negatively regulates store-operated Ca2+ entry, suggesting that these different PIP5K-I isoforms synthesize functionally distinct pools of PtdIns(4,5)P2 at the plasma membrane.


Introduction
Regulation of cytoplasmic Ca 2+ concentration is a hallmark of eukaryotic cells and is crucial for processes ranging from fertilization to smooth muscle contraction (Takahashi et al., 1999). In mast cells, the antigen-stimulated increase in cytoplasmic Ca 2+ mediated by FcεRI receptors for IgE is necessary for the release of allergic mediators such as histamine and de novo synthesis and secretion of cytokines (Vig et al., 2008). In response to antigen stimulation, tyrosine-kinase-mediated activation of phospholipase Cγ (PLCγ) results in hydrolysis of the membrane glycerophospholipid, phosphatidylinositol 4,5-bisphosphate [PtdIns(4,5)P 2 ] to produce inositol 1,4,5-trisphosphate (IP 3 ), which binds to IP 3 receptors to activate Ca 2+ efflux from the ER. Depletion of Ca 2+ from the ER lumen activates store-operated Ca 2+ entry (SOCE), which is initiated by oligomerization and redistribution of the luminal ER Ca 2+ sensor STIM1 (Liou et al., 2007;Luik et al., 2008). Interaction of STIM1 with the tetraspan Ca 2+ channel subunit Orai1/CRACM1 at the plasma membrane initiates SOCE (Feske, 2007;Lewis, 2007).
One factor that regulates Ca 2+ release from the ER is the availability of PtdIns(4,5)P 2 , which is the substrate for PLCγ. The type I family of phosphatidylinositol 4-phosphate 5-kinases (PIP5K-I) synthesize PtdIns(4,5)P 2 by phosphorylation of its substrate, PtdIns(4)P. This family comprises three isoforms: α, β and γ, including three splice variants of the γ isoform (γ87, γ90 and γ93) (Kanaho et al., 2007). Nomenclature for the murine and human isoforms of PIP5K-Iα and -Iβ were previously inconsistent. In this manuscript, the human nomenclature is used as per recently adopted convention. In HeLa cells, Wang et al. (Wang et al., 2004) identified PtdIns(4,5)P 2 synthesized by PIP5K-Iγ87 as the source for PLCβmediated IP 3 production. The kinase isoform responsible for synthesis of the pool of PtdIns(4,5)P 2 that is hydrolyzed by PLCγ in mast cells has not been identified. Knockout of PIP5K-Iβ in mice was shown to result in enhanced antigen-stimulated Ca 2+ response in bone marrow-derived mast cells (BMMCs) (Sasaki et al., 2005), indicating that this β isoform does not initiate Ca 2+ mobilization by FcεRI.
In the present study, we investigated roles for different PIP5K-I isoforms in mast cell signaling, and we found that the β and γ isoforms regulate distinct stages of the mast cell Ca 2+ response. By comparing ectopic expression of wild-type (wt) PIP5K-Iβ and a catalytically inactive mutant in RBL-2H3 mast cells, we find that PtdIns(4,5)P 2 produced by PIP5K-Iβ does not contribute to IP 3mediated Ca 2+ release from the ER, but, rather, it negatively regulates SOCE. We further demonstrate that PIP5K-Iγ contributes positively to antigen-stimulated Ca 2+ release from the ER in BMMCs, most likely by synthesizing the pool of PtdIns(4,5)P 2 that is hydrolyzed by PLCγ to produce IP 3 . These results indicate that PIP5K-Iβ and -Iγ synthesize functionally and possibly spatially distinct pools of PtdIns(4,5)P 2 in the mast cell plasma membrane.

Crosslinking of IgE receptors by antigen initiates Ca 2+ mobilization in mast cells by activating phospholipase-Cγmediated hydrolysis of phosphatidylinositol-4,5-bisphosphate
[PtdIns(4,5)P 2 ]. The resulting inositol 1,4,5-trisphosphatemediated Ca 2+ release from the endoplasmic reticulum (ER) activates store-operated Ca 2+ entry, which is necessary for exocytotic release of inflammatory mediators. To investigate roles for PtdIns(4,5)P 2 -synthesizing isozymes of the type I phosphatidylinositol 4-phosphate 5-kinase family (PIP5K-I) in mast cell signaling, we compared the ectopic expression of wildtype and catalytically inactive PIP5K-Iβ in RBL-2H3 mast cells. Surprisingly, both antigen and thapsigargin-stimulated Ca 2+ influx were reduced by overexpression of active PIP5K-Iβ, whereas antigen-stimulated Ca 2+ release from ER stores was unaffected. Consistent with these results, Ca 2+ entry stimulated by antigen or thapsigargin was enhanced by expression of a plasma-membrane-associated inositol polyphosphate 5Ј-phosphatase, whereas antigen-stimulated Ca 2+ release from stores was reduced. To investigate the role of PIP5K-Iγ in antigen-stimulated Ca 2+ mobilization, we used bone-marrowderived mast cells from PIP5K-Iγ -/mice. Antigen-stimulated Ca 2+ release from ER stores was substantially reduced in the absence of PIP5K-Iγ, but thapsigargin-mediated Ca 2+ entry was unaffected. In summary, PIP5K-Iγ positively regulates antigenstimulated Ca 2+ release from ER stores, whereas PIP5K-Iβ negatively regulates store-operated Ca 2+ entry, suggesting that these different PIP5K-I isoforms synthesize functionally distinct pools of PtdIns(4,5)P 2 at the plasma membrane.

Results
Ectopically expressed PIP5K-Iβ synthesizes a plasma membrane pool of PtdIns(4,5)P 2 in RBL-2H3 cells To investigate the role of PtdIns(4,5)P 2 synthesis by PIP5K-Iβ in IgE receptor signaling in RBL-2H3 mast cells, we stably expressed murine HA-tagged wt PIP5K-Iβ and a catalytically inactive (D203A) mutant of this kinase (Tolias et al., 2000). Three wt clones (IβWT) and two mutant clones (IβMut) were characterized and their phenotypes compared with untransfected RBL-2H3 cells (2H3). Fig.  1A shows representative images for EGFP expression from the pIRES2-EGFP vector and kinase localization in individual IβWT and IβMut clones. When labeled with an anti-HA mAb that detects the kinase, both wt and mutant PIP5K-Iβ showed relatively low expression levels with some localization to the plasma membrane and in cytoplasmic structures. The 2H3 cells did not show EGFP fluorescence or significant labeling by the anti-HA mAb. Expression was confirmed by western blotting: wt and mutant PIP5K-Iβ were immunoprecipitated from individual IβWT and IβMut clones, respectively, and detected using anti-HA antibodies, with 2H3 cells as a negative control (Fig. 1B).
To evaluate the effects of wt PIP5K-Iβ and mutant PIP5K-Iβ on PtdIns(4,5)P 2 levels at the plasma membrane, the PtdIns(4,5)P 2specific PLCδ-PH domain (PLCδ-PH-dsRed) was used to quantitatively monitor plasma membrane PtdIns(4,5)P 2 using realtime confocal microscopy and a plasma-membrane-masking algorithm as described in the Materials and Methods. Similar strong plasma membrane localization of the expressed PH domain was observed in 2H3, IβWT and IβMut cells, with no significant differences detected in the ratio of plasma membrane to cytoplasmic PLCδ-PH-dsRed in these cells under resting conditions ( Fig. 2A, insert, and data not shown). Stimulation of 2H3 cells with multivalent antigen did not cause a significant change in average PtdIns(4,5)P 2 levels at the plasma membrane after monitoring for 300 seconds with PLCδ-PH-dsRed, despite a substantial IP 3dependent Ca 2+ response under these conditions ( Fig. 2A; see below). Although antigen-stimulated PtdIns(4,5)P 2 hydrolysis is expected to reduce the plasma membrane PtdIns(4,5)P 2 levels, it is apparently compensated by stimulated PtdIns(4,5)P 2 synthesis on this timescale (Apgar, 1995). Also shown in Fig. 2A, insignificant differences in the antigen-stimulated changes in PtdIns(4,5)P 2 levels were observed for the IβWT and IβMut cells compared with 2H3 cells.
Similarly to antigen stimulation, no significant changes in plasma membrane PtdIns(4,5)P 2 levels were detected following stimulation with the Ca 2+ ionophore A23187 in 2H3 cells (Fig. 2B). However, A23187 stimulated an increase in the plasma membrane recruitment of the PH domain in IβWT cells (Fig. 2B), suggesting a net increase in stimulated PtdIns(4,5)P 2 synthesis at the plasma membrane by wt PIP5K-Iβ. By contrast, IβMut cells exhibited an ionophorestimulated decrease in plasma membrane PtdIns(4,5)P 2 levels ( Fig.  2B), suggesting dominant-negative suppression of the endogenous kinase activity by this catalytically inactive mutant, coupled with PtdIns(4,5)P 2 hydrolysis. It is not clear why a similar dominantnegative effect of this mutant PIP5K-Iβ is not observed in antigenstimulated cells, but it is possible that antigen stimulates additional pathways of PtdIns(4,5)P 2 synthesis that might compensate for this effect (see below).
We also examined the capacity of these cells to ruffle in response to antigen stimulation (Pfeiffer et al., 1985). IβWT and IβMut cells showed similar F-actin concentration at the periphery of unstimulated cells (Fig. 3A, left panels). Following stimulation by antigen, IβMut cells exhibited greatly suppressed ruffling at the dorsal cell surface compared with either 2H3 or IβWT cells. Results from multiple experiments are summarized in Fig. 3B. They show that stimulated ruffling in IβMut cells was suppressed by ~80% compared with 2H3 or IβWT cells, which showed similarly strong ruffling responses. These results are consistent with the characteristics of the PIP5K-Iβ -/-BMMCs, which showed reduced F-actin content before and after antigen stimulation (Sasaki et al., 2005). They further support the conclusion above that mutant PIP5K-Iβ can act as a dominant-negative suppressor of endogenous PIP5K-Iβ activity in stimulated cells.

PIP5K-Iβ negatively regulates store-operated Ca 2+ influx
To investigate the role of PIP5K-Iβ in antigen-stimulated Ca 2+ mobilization in RBL cells, we measured Ca 2+ levels in suspended 2H3, IβWT and IβMut cells using the Ca 2+ indicator indo-1 and steady state fluorimetry. As shown in Fig. 4A, antigen stimulated a biphasic Ca 2+ response that was somewhat larger and faster for the IβMut cells than for the 2H3 cells. By contrast, the IβWT cells exhibited a rapid initial response with a clearly attenuated plateau phase compared with the 2H3 cells. When several experiments were averaged, the IβWT cells exhibited a time-integrated response that was about 50% of the 2H3 cell response, whereas the IβMut cells exhibited a slightly larger integrated response than the parental 2H3 cells (Fig. 4D). These results are consistent with those reported for antigen-stimulated BMMCs from PIP5K-Iβ -/mice, in which the absence of PIP5K-Iβ resulted in a ~20% larger sustained phase of the Ca 2+ response compared with that for wt BMMCs (Sasaki et al., 2005). To determine whether PIP5K-Iβ contributes to the pool of PtdIns(4,5)P 2 that is hydrolyzed by antigen-stimulated PLCγ to produce IP 3 , we examined the Ca 2+ response to antigen in the absence of extracellular Ca 2+ . Under these conditions, the transient response observed was completely dependent on IP 3 -mediated Ca 2+ release from ER stores . As shown in Fig. 4B, 2H3, IβWT and IβMut cells all showed similar, transient responses under these conditions, and these integrated responses from several experiments were not significantly different from each other (Fig.  4D). These results suggest that PIP5K-Iβ is not the primary enzyme responsible for the pool of PtdIns(4,5)P 2 that is hydrolyzed by PLCγ in these cells.

Journal of Cell Science 122 (14)
To investigate directly the role of PIP5K-Iβ in regulating Ca 2+ influx in RBL mast cells, we bypassed FcεRI-mediated Ca 2+ release from stores and activated store-operated Ca 2+ influx using thapsigargin to inhibit the sarcoplasmic/endoplasmic reticulum Ca 2+ -ATPase (SERCA). As shown in Fig. 4C, low concentrations of thapsigargin caused a rapid increase in cytoplasmic Ca 2+ levels in these cells, followed by a sustained phase that is due to store operated Ca 2+ influx (Ali et al., 1994) (and data not shown). IβWT cells showed an integrated response to thapsigargin that was ~45% less than that for 2H3 cells, whereas the response to thapsigargin in the IβMut cells was not significantly different from that for 2H3 cells (Fig. 4D). These results indicate that the negative regulatory effect of overexpression of wt PIP5K-Iβ observed upon antigenstimulated Ca 2+ mobilization is due primarily to negative regulation of store-operated Ca 2+ influx.
A plasma-membrane-associated yeast inositol polyphosphate 5Ј-phosphatase alters stimulated Ca 2+ responses To investigate further the roles of PtdIns(4,5)P 2 at the plasma membrane in FcεRI-mediated Ca 2+ responses, we stably expressed an ECFP-tagged yeast inositol polyphosphate 5Ј-phosphatase, Inp51 (Inp cells) (Stolz et al., 1998) that dephosphorylates PtdIns(4,5)P 2 to generate PtdIns(4)P. As shown in supplementary material Fig. S1A, Inp51 exhibits strong plasma membrane association, with some cytoplasmic localization. These Inp51-expressing cells did not show significant differences in the ratio of PLCδ-PH-dsRed at the plasma  membrane to that in the cytoplasm when this PtdIns(4,5)P 2 reporter was co-expressed, indicating that expression of the Inp51 phosphatase did not detectably alter the steady state levels of PtdIns(4,5)P 2 at the plasma membrane (data not shown). However, using this PtdIns(4,5)P 2 probe, we observed a time-dependent reduction in PtdIns(4,5)P 2 at the plasma membrane in thapsigargin-stimulated Inp cells compared with levels in stimulated 2H3 cells (supplementary material Fig. S1B), suggesting that Inp51 is hydrolyzing newly synthesized PtdIns(4,5)P 2 .
To investigate whether a reduction in stimulated PtdIns(4,5)P 2 levels by Inp51 alters Ca 2+ responses to FcεRI crosslinking, we compared the antigen-stimulated Ca 2+ response in Inp cells with that in 2H3 cells. As shown in Fig. 5A, the antigen-stimulated Ca 2+ response is enhanced by ~20% in Inp cells. This enhancement is somewhat greater for the sustained phase of thapsigargin-stimulated cells (~30%; Fig. 5C), indicating a principal effect on SOCE. By contrast, antigen-stimulated Ca 2+ release from stores, monitored in the absence of extracellular Ca 2+ , was reduced by ~40% in Inp cells  (Fig. 5B). These results are summarized for integrated responses from multiple experiments in Fig. 5D. They support the findings with PIP5K-Iβ overexpression indicating a negative regulatory effect of PtdIns(4,5)P 2 at the plasma membrane on store-operated Ca 2+ influx (+Ca 2+ ) (Fig. 4A,C,D). Furthermore, they also implicate PtdIns(4,5)P 2 at the plasma membrane in antigen-stimulated Ca 2+ release from stores, most likely as the source of IP 3 generated by activated PLCγ (-Ca 2+ ) (Fig. 5B,D).

PIP5K-Iγ contributes to antigen-mediated Ca 2+ release from stores
The lack of effect of mutant or wt PIP5K-Iβ on antigen-stimulated Ca 2+ release from stores in RBL mast cells (-Ca 2+ ) (Fig. 4B,D) suggests that this isoform does not contribute to the pool of PtdIns(4,5)P 2 that is hydrolyzed by PLCγ in response to FcεRI activation in these cells. Consistent with this, antigen-stimulated IP 3 production was not reduced in PIP5K-Iβ -/-BMMCs (Sasaki et al., 2005). A previous study demonstrated that PIP5K-Iγ (but not PIP5K-Iβ) is important for the pool of PtdIns(4,5)P 2 that is hydrolyzed by PLCβ in response to G-protein-coupled receptor activation in HeLa cells (Wang et al., 2004). We attempted to investigate the role of PIP5K-Iγ in RBL mast cells by siRNA knockdown and overexpression strategies, but we were unsuccessful in modulating the expression level of this isoform sufficiently. To evaluate the role of PIP5K-Iγ in FcεRI-mediated Ca 2+ mobilization, we examined this response in BMMCs from wt and PIP5K-Iγ -/mice (Di Paolo et al., 2004). BMMCs were differentiated from newborn wt and PIP5K-Iγ -/mice using IL3 and SCF, and similar expression of FcεRI on these cells was confirmed using Alexa Fluor 488-IgE (supplementary material Fig. S2A).
As shown in Fig. 6A, both antigen-stimulated Ca 2+ release from stores and antigen-stimulated Ca 2+ influx are substantially reduced in the absence of PIP5K-Iγ. As summarized for a number of experiments in Fig. 6B, the integrated Ca 2+ response to antigen in the presence of extracellular Ca 2+ was reduced by ~40% in the absence of PIP5K-Iγ, and the response to antigen in the absence of extracellular Ca 2+ was reduced by ~50%. No difference in the magnitude of thapsigargin-mediated SOCE was observed for BMMCs in the presence and absence of PIP5K-Iγ (supplementary material Fig. S3). These results indicate that PIP5K-Iγ contributes to FcεRI-mediated Ca 2+ mobilization, most likely by synthesizing the pool of PtdIns(4,5)P 2 that is hydrolyzed by PLCγ to produce IP 3 . Thapsigargin-mediated SOCE does not depend on IP 3 production or on PIP5K-Iγ. Because both PIP5K-Iγ87 and PIP5K-Iγ90 splice variants and the catalytically inactive mutants of these enzymes strongly localize to the plasma membrane of RBL mast cells (supplementary material Fig. S2B), the γ isoform of PIP5K-I is likely to synthesize PtdIns(4,5)P 2 at the plasma membrane, similarly to the β isoform of PIP5K-I as described above. These results point to functionally distinct pools of PtdIns(4,5)P 2 that arise from these different isoforms of PIP5K.

Discussion
Phosphoinositides, including PtdIns(4,5)P 2 , have been shown to have multiple roles in cell signaling and trafficking (Di Paolo and De Camilli, 2006), and organelle-specific synthesis and localization of different phosphoinositides are involved in these processes (Behnia and Munro, 2005;Balla and Balla, 2006). However, the regulation of these organelle-specific pools is poorly understood, and possible roles for separate pools of phosphoinositides within the same organelle have not been widely explored. Our studies complement and extend others in the literature to yield new insights into the participation of PIP5Ks in mast cell signaling mediated by IgE receptors. Our investigation using overexpression and genetic deletion provides evidence that PIP5K-Iβ and PIP5K-Iγ synthesize functionally distinct pools of PtdIns(4,5)P 2 : reduction of FcεRImediated Ca 2+ release from ER stores in mast cells from PIP5K-Iγ -/mice indicates that PtdIns(4,5)P 2 synthesized by PIP5K-Iγ contributes to this process. Overexpression of wt PIP5K-Iβ or its catalytically inactive mutant does not affect this initial step of Ca 2+ mobilization, but rather wt PIP5K-Iβ negatively regulates SOCE in RBL mast cells, and knockout of this gene enhances Ca 2+ responses in BMMCs (Sasaki et al., 2005), indicating consistent results for these two different mast cell types. Expression of a yeast inositol polyphosphate 5Ј-phosphatase, Inp51, in RBL cells both reduces Ca 2+ release from stores and enhances store-operated Ca 2+ influx, consistent with separate roles for PtdIns(4,5)P 2 in each of these processes.

PtdIns(4,5)P 2 generated by PIP5K-Iγ functions as PLC substrate
Yin and colleagues (Wang et al., 2004) used siRNA-mediated knockdown of PIP5K isoforms to reveal participation of PIP5K-Iγ87 in G-protein-coupled receptor-mediated Ca 2+ signaling in HeLa cells. They showed that this isoform is responsible for stimulated IP 3 production by PLCβ, whereas PIP5K-Iα or PIP5K-Iβ did not contribute, suggesting functional compartmentalization of PIP5K-I isoforms in these cells. Consistent with these observations, knockout of PIP5K-Iβ did not reduce antigen-stimulated IP 3 production in mast cells, despite reduced levels of PtdIns(4,5)P 2 (Sasaki et al., 2005). Our results are in agreement with these data, indicating that PIP5K-Iγ, but not PIP5K-Iβ, synthesizes a pool of PtdIns(4,5)P 2 that is hydrolyzed to IP 3 by PLCγ to cause Ca 2+ release from the ER in mast cells. It is notable that PIP5K-Iγ binds to PLCγ1, and this is regulated by phosphorylation of Tyr634 in PIP5K-Iγ (Sun et al., 2007).
A recent study in platelets indicates that PIP5K-Iα contributes to the pool of PtdIns(4,5)P 2 that is hydrolyzed to IP 3 in these cells . It is possible that this isoform also contributes to antigen-stimulated IP 3 production in mast cells. Interestingly, although PIP5K-Iγ contributes to the pool of PtdIns(4,5)P 2 used for the initiation of SOCE by antigen, it is not necessary for antigenstimulated degranulation in mouse BMMCs. As shown in supplementary material Fig. S4, antigen-stimulated degranulation in PIP5K-Iγ -/mast cells is actually enhanced compared with wt BMMCs, and this response is less sensitive to inhibition by the protein kinase C (PKC) inhibitor, bisindolylmaleimide (BIM) in the PIP5K-Iγ -/cells. Furthermore, the degranulation response to the Ca 2+ ionophore, A23187, is insensitive to BIM and not dependent on PIP5K-Iγ. These results indicate that PIP5K-Iγ negatively regulates a step in FcεRI signaling that contributes to stimulated exocytosis in a PKC-independent manner. These results are consistent with previous observations indicating a negative downstream role for PtdIns(4,5)P 2 in degranulation by rat peritoneal mast cells (Hammond et al., 2006). Further studies will be necessary to characterize the nature and mechanism of this negative regulatory pathway.

Mechanisms of regulation of Ca 2+ influx by PtdIns(4,5)P 2
The reduction in SOCE that we observe upon overexpression of wt PIP5K-Iβ, but not its catalytically inactive mutant, reveals a negative regulatory role for this source of PtdIns(4,5)P 2 . Our results are consistent with previous studies on PIP5K-Iβ -/mice (Sasaki et al., 2005), but the specific role in antigen-stimulated Ca 2+ mobilization played by this enzyme was not defined previously. Sasaki and colleagues demonstrated that PIP5K-Iβ participates in F-actin polymerization and its negative regulation of mast cell degranulation (Sasaki et al., 2005). The capacity of mutant PIP5K-Iβ to inhibit antigen-stimulated ruffling in RBL mast cells (Fig. 3) is consistent with these findings. They also suggest that the enhancement of FcεRI-mediated signaling observed in PIP5K-Iβ -/-BMMCs might be due to enhancement in the interaction of FcεRI with lipid rafts (Sasaki et al., 2005). By contrast, we observe negative regulation of SOCE by overexpression of wt PIP5K-Iβ in RBL cells following stimulation by both antigen and thapsigargin, indicating that this effect is downstream of Ca 2+ store depletion, and thus independent of effects of PIP5K-Iβ on the initial steps in FcεRI signaling. It will be of interest to determine whether thapsigarginstimulated SOCE is enhanced in PIP5K-Iβ -/-BMMCs. The similar enhancements we observe for Ca 2+ entry stimulated by both antigen and thapsigargin as a result of expression of Inp51 in RBL cells (Fig. 5), are consistent with a direct effect of PtdIns(4,5)P 2 on SOCE.
The mechanism by which PIP5K-Iβ negatively regulates SOCE is not yet understood. Although mutant PIP5K-Iβ suppresses stimulated ruffling in RBL cells (Fig. 3), and PIP5K-Iβ knockout reduces F-actin levels in BMMCs (Sasaki et al., 2005), enhanced actin polymerization does not appear to mediate negative regulation of SOCE by this enzyme, because inhibition of actin polymerization by cytochalasin D does not alter the effects of overexpression of PIP5K-Iβ on SOCE (data not shown). Recent evidence from our laboratory indicates that stimulated interactions between STIM1 and Orai1 depend on electrostatic interactions involving negatively charged residues at the C-terminus of Orai1 (Calloway et al., 2009). Increased levels of PtdIns(4,5)P 2 at the plasma membrane might negatively regulate the interaction of this sequence with positively charged residues at the C-terminus of STIM1 (Li et al., 2007;Yuan et al., 2009). It is also possible that PtdIns(4,5)P 2 produced by PIP5K-Iβ negatively regulates gating of the Orai1 Ca 2+ channel, in a similar manner to that of PtdIns(4,5)P 2 on TRPV1 channels (Prescott and Julius, 2003). Alternatively, the negative regulatory effect could result from modulation of a channel other than Orai1. For example, Ma and co-workers recently described evidence that the Ca 2+ channel protein TRPC5 contributes to SOCE in RBL cells (Ma et al., 2008). The Ca 2+ -activated cation channel, TRPM4, negatively regulates SOCE in BMMCs (Vennekens et al., 2007) and is positively regulated by PtdIns(4,5)P 2 (Nilius et al., 2006). Further investigations will be necessary to determine the molecular mechanism of regulation of store-operated Ca 2+ entry by PIP5K-Iβ-produced PtdIns(4,5)P 2 in mast cells.
Plasma membrane targeting of PIP5K-I isoforms and the spatio-temporal regulation of PtdIns(4,5)P 2 synthesis Previous studies suggested specific roles for the different PIP5K isoforms. Our results are consistent and further reveal functionally distinct pools of PtdIns(4,5)P 2 at the plasma membrane. Both PIP5K-Iγ (supplementary material Fig. S2B) and PIP5K-Iβ (Fig.  1A) localize to the plasma membrane, and each synthesizes a pool of PtdIns(4,5)P 2 that has a distinctive role there: PIP5K-Iγ contributes to PtdIns(4,5)P 2 that is hydrolyzed by PLCγ to mediate Ca 2+ release from ER stores in BMMCs, and PIP5K-Iβ synthesizes PtdIns(4,5)P 2 that negatively regulates SOCE in both mast cell types. PIP5K-Iβ modulates stimulated changes in PtdIns(4,5)P 2 at the plasma membrane (Fig. 2B) and regulates stimulated changes in plasma membrane ruffling (Fig. 3). Furthermore, plasmamembrane-associated 5Ј-phosphatase, Inp51, inhibits antigenstimulated Ca 2+ release from ER stores while enhancing SOCE, indicating that the processes are differentially regulated by PtdIns(4,5)P 2 at the plasma membrane. These results, taken together, provide strong evidence for two functionally distinct pools of PtdIns(4,5)P 2 at the plasma membrane synthesized by PIP5K-Iβ and γ isoforms.
The physical basis for functional distinction of these independently synthesized pools of PtdIns(4,5)P 2 is not yet clear. Roles for several independent signaling microdomains have been proposed to contribute to the diversity of cellular responses (Simons and Toomre, 2000). Previous evidence for restricted diffusion of PtdIns(4,5)P 2 at the plasma membrane (Cho et al., 2005), together with indications for localized synthesis of PtdIns(4,5)P 2 (Hilgemann, 2007), indicate possible mechanisms for heterogeneity in PtdIns(4,5)P 2 concentration. Pike and Miller (Pike and Miller, 1998) described evidence for preferential hydrolysis of lipid raftassociated PtdIns(4,5)P 2 by receptor-activated PLCγ. Recently, Rodgers and colleagues (Johnson et al., 2008) provided biochemical evidence in T cells for segregation of functionally distinguishable pools of PtdIns(4,5)P 2 into lipid raft and non-raft domains using targeted 5Ј-phosphatases. Together, these studies suggest that compositional heterogeneity in the plasma membrane based on lipidphase-like separation can contribute to functionally distinct pools PIP5Ks regulate Ca 2+ signaling of PtdIns(4,5)P 2 that differentially regulate cell signaling. We hypothesize that PIP5K-Iβ is targeted to ordered lipid rafts, whereas PIP5K-Iγ is excluded from these rafts, thereby allowing the two pools of PtdIns(4,5)P 2 to be spatially and functionally segregated at the plasma membrane. Interactions of these PIP5K-I isoforms with proteins that target them to different plasma membrane microdomains might also participate in maintaining the heterogeneous distribution that regulates their roles in cell signaling.
Cell culture and generation of stable cell lines RBL-2H3 cells were cultured as monolayers in minimum essential medium (MEM; Invitrogen Corp., Carlsbad, CA) with 20% fetal bovine serum (Atlanta Biologicals, GA) and gentamicin (Invitrogen). Murine HA-tagged wt and catalytically inactive mutant (D203A) PIP5K-Iβ constructs were cloned into a pIRES2-EGFP vector (Invitrogen) using XhoI-SacII restriction sites. Cells were transfected using Lipofectamine 2000 in OptiMEM reduced serum medium (Invitrogen). Cells were incubated with the DNA-lipofectamine complexes for 1 hour, then 0.1 μM phorbol 1,2-dibutyrate (Sigma) was added to the cells for 4 hours to enhance fluid phase pinocytosis and DNA uptake in OptiMEM. Transfected cells were selected in MEM containing 380 μg/ml neomycin (G418). Cells positive for PIP5K-Iβ were identified initially based on EGFP expression. Three wt clones (IβWT) and two mutant clones (IβMut) were produced and characterized.
PIP5K-Iγ -/mice were described previously (Di Paolo et al., 2004). Stem cells were extracted by flushing the bone marrow from legs of newborn mice. Cells were maintained in Dulbecco's MEM with 10% fetal bovine serum, and 20 ng/ml each of IL-3 and stem cell factor (Peprotech, NJ). After about 2 weeks in culture, differentiation into mast cells was confirmed by labeling FcεRI with Alexa Flour 488-IgE. The use of stem cell factor was discontinued at this point. The cells were maintained in suspension culture for up to 6 weeks.

Immunocytochemistry
Cells were plated at a subconfluent density of 0.5ϫ10 6 cells/ml in 35 mm MatTek dishes (MatTek Corporation, Ashland, MA) and cultured overnight. Cells were fixed the next day using 3.7% formaldehyde, permeabilized with 0.1% Triton X-100, and labeled for 1 hour with appropriate antibodies in PBS with 10 mg/ml BSA. Images were collected using a Leica TCS SP2 laser scanning confocal system (Leica Microsystems, Exton, PA) with a 63ϫ 0.9 NA, HCX APO L U-V-I water-immersion objective. For detecting HA-tagged wt and mutant PIP5K-Iβ, anti-HA mAb (1:100; Covance Research Products, Berkeley, CA) was used as the primary antibody, followed by Alexa Fluor 568-anti mouse IgG 1 as the secondary antibody (1:100; Invitrogen). FITC-phalloidin (1:200; Invitrogen) was used to label membrane ruffles.

Immunoprecipitation
HA-tagged PIP5K-Iβ was immunoprecipitated using polyclonal anti-HA (Covance). Briefly, cells were resuspended in buffered saline solution (BSS: 20 mM HEPES, 135 mM NaCl, 1.8 mM CaCl 2 , 2 mM MgCl 2 , 5.6 mM glucose, 1 mg/ml BSA, pH 7.4) at a concentration of 6.2ϫ10 6 cells/ml. 500 μL of 3ϫ solubilization buffer (15 mM N-ethylmaleimide, 3 mM Na 3 VO 4 , 15 mM Na 2 P 2 O 7 , 10 mM NaF, 6 mM iodoacetate, 15 mM EDTA, 0.3% Triton X-100, 150 mM NaCl, and 150 mM Tris) was added to 1 ml cells for 5 minutes at 4°C. Supernatant (1.3 ml) from the cell lysate following centrifugation at 10,000 g. for 10 minutes at 4°C, was incubated with 10 μg anti-HA and 25 μl Protein A beads (Amersham, Piscataway, NJ) for 2 hours at 4°C on a rotator. Beads were washed with 1ϫ solubilization buffer three times by pelleting and resuspension in 500 μl. Washed beads were boiled in 40 μl of 2ϫ sample buffer (Invitrogen). For polyacrylamide gel electrophoresis, 40 μl sample was loaded per lane of a 12% Tris-glycine gel (Invitrogen) and electrophoresed for 1 hour at 35 mA, followed by electrotransfer to Immobilon P (Millipore, Bedford, MA). Blots were sequentially incubated with 1 mg/ml anti-HA mAb (Covance), 0. Measurement of PtdIns(4,5)P 2 at the plasma membrane PLCδ-PH-dsRed was transiently transfected into the cells using either lipofectamine as described above or electroporation, and used as a marker for PtdIns(4,5)P 2 at the plasma membrane (Varnai and Balla, 2006). For electroporation, cells were harvested and resuspended at 5ϫ10 6 /ml in an electroporation buffer (137 mM NaCl, 2.7 mM KCl, 1 mM MgCl 2 , 1 mg/ml glucose, 20 mM HEPES, pH 7.4). Cells (500 μl) at 4°C were premixed with 6 μg DNA and pulsed once (exponential decay, 280 V, 90 μF) in a Bio-Rad Gene Pulser (Bio-Rad, Hercules, CA). Cells were immediately resuspended in medium and plated at 0.5ϫ10 6 cells/ml in MatTek dishes, then cultured overnight. PtdIns(4,5)P 2 levels were measured using real time confocal microscopy as previously described (Das et al., 2008). Briefly, the plasma membrane of IgE-sensitized cells was labeled with 2 μg/ml Cy5-cholera toxin B for 4 minutes before imaging. Images were collected every 3 seconds for a total of 3 minutes in the absence of stimulation using a Leica confocal TCS SP2 microscope. DNP-BSA (0.8 μg/ml), A23187 (0.5 μM), or thapsigargin (0.25 μM) was added, and images were collected every 3 seconds for a total of 5 minutes after stimulation. ImageJ (http://rsb.info.nih.gov/ij/) and Matlab (The MathWorks) were used for quantitative image analysis as described (Das et al., 2008). Briefly, a region of interest enclosing the selected cell was chosen in ImageJ. Matlab was used to generate a plasma membrane (PM) mask by thesholding the cholera toxin B image. The PLCδ-PH-dsRed fluorescence was calculated as the mean intensity of all pixels in the PM mask for each image time point.

Ca 2+ measurements
Cytoplasmic Ca 2+ levels were measured using an SLM 8100C steady-state fluorimeter (SLM Instruments, Urbana, IL). Cells in suspension were loaded with the Ca 2+ indicator, indo-1 (excitation 330 nm, emission 400 nm; Invitrogen), and sensitized with anti-DNP IgE in BSS containing 0.5 mM sulfinpyrazone (Sigma). Cells were stimulated with 0.2 μg/ml DNP-BSA or 0.25 μM thapsigargin, and their Ca 2+ responses were monitored for ~500 seconds. The cells were lysed by the addition of 0.1% Triton X-100 to obtain the maximal value of indo-1 fluorescence, and then quenched by the addition of 6 mM EDTA to obtain background indo-1 fluorescence levels.
Representative Ca 2+ responses were plotted as the change in fluorescence intensity of indo-1 with time, where the fluorescence intensity was expressed as the ratio of the actual fluorescence value to the difference between the value following addition of Triton X-100 and that for Triton X-100 + EDTA. Under the conditions of these measurements, indo-1 fluorescence is largely proportional to Ca 2+ concentration. Ca 2+ responses in the presence of extracellular Ca 2+ were integrated for 300-500 seconds. For these cells, integrated Ca 2+ responses exhibit a broad dose-response to multivalent antigen that is near maximal between 0.1 and 1.0 μg/ml DNP-BSA. Responses were quantified as a ratio of the integrated area of the Ca 2+ response for the sample to that of untransfected 2H3 cells measured in the same experiment.

Ruffling assay
Cells were plated at a subconfluent density of 0.5ϫ10 6 cells/ml in 35 mm MatTek dishes and sensitized with anti-DNP IgE overnight. Cells were stimulated with 0.8 μg/mL DNP-BSA for 15 minutes at 37°C, fixed using 3.7% formaldehyde, permeabilized with 0.1% Triton X-100, and labeled with FITC-conjugated phalloidin (Invitrogen) to visualize ruffles at the dorsal cell surface. Positive transfectants were scored based on F-actin labeling as being ruffled or not ruffled (see Results).