Control of Calcium Signal Propagation to the Mitochondria by Inositol 1,4,5-Trisphosphate-binding Proteins*

Cytosolic Ca2+ ([Ca2+]c) signals triggered by many agonists are established through the inositol 1,4,5-trisphosphate (IP3) messenger pathway. This pathway is believed to use Ca2+-dependent local interactions among IP3 receptors (IP3R) and other Ca2+ channels leading to coordinated Ca2+ release from the endoplasmic reticulum throughout the cell and coupling Ca2+ entry and mitochondrial Ca2+ uptake to Ca2+ release. To evaluate the role of IP3 in the local control mechanisms that support the propagation of [Ca2+]c waves, store-operated Ca2+ entry, and mitochondrial Ca2+ uptake, we used two IP3-binding proteins (IP3BP): 1) the PH domain of the phospholipase C-like protein, p130 (p130PH); and 2) the ligand-binding domain of the human type-I IP3R (IP3R224–605). As expected, p130PH-GFP and GFP-IP3R224–605 behave as effective mobile cytosolic IP3 buffers. In COS-7 cells, the expression of IP3BPs had no effect on store-operated Ca2+ entry. However, the IP3-linked [Ca2+]c signal appeared as a regenerative wave and IP3BPs slowed down the wave propagation. Most importantly, IP3BPs largely inhibited the mitochondrial [Ca2+] signal and decreased the relationship between the [Ca2+]c and mitochondrial [Ca2+] signals, indicating disconnection of the mitochondria from the [Ca2+]c signal. These data suggest that IP3 elevations are important to regulate the local interactions among IP3Rs during propagation of [Ca2+]c waves and that the IP3-dependent synchronization of Ca2+ release events is crucial for the coupling between Ca2+ release and mitochondrial Ca2+ uptake.

Inositol 1,4,5-trisphosphate (IP 3 ) 1 -induced Ca 2ϩ liberation from intracellular stores results in a [Ca 2ϩ ] c signal that con-trols a wide spectrum of cell functions, including energy metabolism, gene transcription, and cell proliferation. Appropriate exposure of the effectors to Ca 2ϩ throughout the cell is supported by several mechanisms that include the propagation of Ca 2ϩ release throughout the cell without attenuation, the recruitment of Ca 2ϩ entry, and efficient delivery of the [Ca 2ϩ ] c signal into organelles such as the nucleus and mitochondria (1).
Spreading of the [Ca 2ϩ ] c signal is facilitated by regenerative mechanisms of Ca 2ϩ mobilization, which may derive from interactions between adjacent Ca 2ϩ release sites. Non-metabolizable IP 3 analogs have been found to evoke [Ca 2ϩ ] c oscillations, providing support to the idea that at a constant [IP 3 ] Ca 2ϩ released through IP3Rs is sufficient to trigger regenerative Ca 2ϩ release via binding to and activating IP3Rs and ryanodine receptors (RyR) (2). Based on IP 3 microinjection and uncaging studies, the regenerative Ca 2ϩ release during [Ca 2ϩ ] c waves was also claimed to be largely independent of [IP 3 ] (3,4). Along this line, IP3R-mediated [Ca 2ϩ ] c waves were observed in the absence of any stimulated IP 3 formation (5). However, released Ca 2ϩ may also promote phospholipase C (PLC) activation and, in turn, increase IP 3 , providing a potential amplification mechanism for IP3R-mediated Ca 2ϩ release waves (6 -8). Ca 2ϩ may activate the PLC coupled to the agonist receptor to stimulate cleavage of phosphatidylinositol 4,5-bisphosphate (PIP 2 ) in the plasma membrane. The localization of phospholipase C enzyme isoforms (9,10) and PIP 2 (11) in intracellular membranes has also been documented. IP 3 formed at the plasma membrane may rapidly diffuse throughout the cytoplasm (12), but the presence of PIP 2 and PLC in the vicinity of the IP3R could also provide for a local IP 3 feedback. Thus, the role of IP 3 fluctuations in Ca 2ϩ wave propagation requires further investigation.
Recharging of the ER Ca 2ϩ stores during the agonist-induced [Ca 2ϩ ] c signal involves Ca 2ϩ entry mediated by so-called storeoperated Ca 2ϩ -entry mechanisms that might be mediated by the canonical transient receptor potential channel family (13). One of the proposed mechanisms for activation of Ca 2ϩ entry during Ca 2ϩ release is a conformational coupling between IP3Rs and store-operated Ca 2ϩ channels (14). Although storeoperated Ca 2ϩ entry is activated by agents that directly target the ER Ca 2ϩ store (15,16), IP 3 binding to the IP3R has been claimed to have a role in activation of Ca 2ϩ entry (17,18). Nonetheless, the question of whether an increase in [IP 3 ] is required for the channel activation remains elusive. IP 3 -induced [Ca 2ϩ ] spikes are also delivered to the mitochondria to control the activity of several enzymes that participate in ATP production as well as that of other proteins compartmentalized to the matrix space (19 -21). Mitochondrial Ca 2ϩ uptake sites display low affinity toward Ca 2ϩ and appear to respond mostly to the large local [Ca 2ϩ ] c transients that occur in the vicinity of the activated IP3Rs and RyRs (21)(22)(23)(24), but a rapid mode of Ca 2ϩ uptake at relatively low [Ca 2ϩ ] has also been documented (25). A local interaction between RyRs and mitochondrial Ca 2ϩ uptake sites may secure that even fundamental Ca 2ϩ release events (Ca 2ϩ sparks) induce a [Ca 2ϩ ] m increase in the neighboring mitochondrion (Ca 2ϩ mark) (26). However, a substantially larger [Ca 2ϩ ] m rise occurs during a global [Ca 2ϩ ] c signal, suggesting that numerous Ca 2ϩ release events cooperate with each other to establish a [Ca 2ϩ ] m signal in a mitochondrion. At the ER-mitochondrial interface, great quantities of IP3Rs have been visualized and enrichment in ER Ca 2ϩ pumps and mitochondrial Ca 2ϩ uniporters is probable (reviewed in Refs. 27 and 28). Morphology of the ER-mitochondrial associations and concentration of the Ca 2ϩ transporters at the interface may play a role in controlling coordinated activation of the individual Ca 2ϩ release events that give rise to the IP 3 -linked [Ca 2ϩ ] m spikes. Thus, the role of IP 3 in coordination of Ca 2ϩ release events is of great interest in a variety of Ca 2ϩ signaling mechanisms. We reasoned that mobile IP 3 buffers should be useful to test the role of IP 3 in [Ca 2ϩ ] c wave propagation, activation of Ca 2ϩ entry, and in IP3R-dependent activation of mitochondrial Ca 2ϩ uptake.
An intensely investigated IP 3 -binding module is the IP 3binding domain of the IP3R. The N-terminal cytoplasmic region of the human type-I IP3R (residues 1-604) binds IP 3 with comparable affinity to the full-length IP3R (29). Removal of residues 1-223 (suppressor region) further increases the affinity for IP 3 (30). Expression of IP 3 R 224 -605 has been shown to attenuate the ATP-induced [Ca 2ϩ ] c signal in human embryonic kidney 293 and in COS-7 cells (31,32). Another structurally unrelated IP 3 -binding module is the PH domain of certain proteins. Some PH domains are known to bind IP 3 and/or PIP 2 , for example, the PH domains of PLC␦1 and the PLC-like 130-kDa protein (p130). p130 was isolated from rat brain as an IP3BP (33) and has been shown to be important in the signaling by GABA A receptors (34). p130 shares 38.2% sequence homology to PLC-␦1 but lacks catalytic activity. The PH domain of p130 (residues 95-233) has also been shown to inhibit the [Ca 2ϩ ] c signal evoked by IP 3 -linked agonists (31,35).
Our main objective was to evaluate the possible role of fluctuations of IP 3 in the spatio-temporal organization of the calcium signal, utilizing IP 3 R 224 -605 and p130PH. Our results indicate that these two proteins are freely distributed in the cytosol, inhibit Ca 2ϩ release induced by suboptimal IP 3 but do not suppress Ca 2ϩ release evoked by sensitization of the IP3R to IP 3 , and fail to attenuate non-IP3R-mediated Ca 2ϩ release. Thus, IP 3 R 224 -605 and p130PH act as mobile cytosolic IP 3 buffers. IP 3 R 224 -605 and p130PH were used next to explore the effects of IP 3 buffering on the propagation of IP 3 (1,4,5)IP 3 -binding domain (residues 224 -605) of the human type-I (1,4,5)IP 3 receptor with cyan, green, or yellow fluorescent protein have been described previously (31). In addition, the PH domain of p130 and the R134L mutant were also inserted into a plasmid encoding a monomeric red fluorescent protein (RFP-p130PH and RFP-p130PH-R134L) (36). Ratiometric pericam targeted to the mitochondria (pericam-mt) has also been published previously (37). Bacterial expression of the GFP-fused protein domains (p130PH-GFP, GFP-IP 3 R 224 -605 , and GFP-PLC␦1PH R40L) was carried out as described previously (31).
Cell Culture-COS-7 cells (obtained from ATCC) were cultured in Dulbecco's modified essential medium supplemented with 10% fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin in humidified air (5% CO 2 ) at 37°C. RBL-2H3 cells were cultured as described previously (22). For imaging experiments, cells were plated onto poly-D-lysine-treated glass coverslips at a density of 20,000 -25,000/cm 2 and were grown for 3-4 days. For cell suspension studies, cells were cultured for 4 -6 days in 75-cm 2 flasks.
Transfection of Cells-Cells plated onto poly-D-lysine-coated coverslips were transfected with plasmid DNA (1.5 g/ml) for 6 -12 h using Lipofectamine (Invitrogen) and Opti-MEM I medium (Invitrogen) according to the manufacturer's instructions. Cells were observed 24 -36 h after transfection. Measurement of [ 3 H]IP 3 -COS-7 cells (5 ϫ 10 4 cells/ml) were cultured on 12-well culture dishes for 1 day and transfected with either the RFP-p130PH or RFP-p130PH-R134L construct. After 24 h, cells were labeled in 0.75-ml inositol-free Dulbecco's modified essential medium containing 0.1% BSA, 2.5% fetal bovine serum, and myo-[ 3 H]inositol (20 Ci/ml, Amersham Biosciences) for 24 h. After two washes, cells were preincubated for 30 min at 37°C before stimulation with 50 M ATP for the indicated times. Incubations were terminated by the addition of ice-cold perchloric acid (5% final). Inositol phosphates were extracted and separated on high pressure liquid chromatography as described previously (38) Coverslips were mounted at the thermostated stage (35°C) of an Olympus IX70 inverted microscope fitted with a ϫ40 (UApo, NA 1.35) oil immersion objective. For simultaneous measurements of fura2, pericam, and RFP fluorescence, a Leica IRB2 inverted microscope equipped with a motorized turret under computer control and fitted with the above described objective was used. Fluorescence images were collected using a cooled CCD camera (PXL, Photometrics or, for fast-imaging, a frame-transfer device, Pluto, PixelVision) (22,31,39). The following multi-parameter imaging protocols were applied: fura2 and GFP (fura2, 340-and 380-nm excitations or 360 nm for Mn 2ϩ quenching; enhanced GFP, 490-nm excitation, 510-nm longpass dichroic mirror, and a 520-nm longpass emission filter); CFP and YFP (CFP, 435-nm excitation, 480/40-nm emission; FRET, 435-nm excitation, 535/30-nm emission); and fura2, pericam, and RFP (fura2, 340-and 380-nm excitation; pericam, 414-and 495-nm excitation, 510-nm longpass dichroic mirror, and a 520-nm longpass emission filter; RFP, 545-nm excitation, 570-nm longpass dichroic mirror, and a 603/75-nm emission filter; CCD imaging system). In most of the experiments, a set of images was obtained every 3 s, whereas in recording of the waves 15 frames were obtained in each second, and in the measurements that involved a switch of the filter cube in each cycle, a set of images was obtained every 6 s.
For evaluation of [Ca 2ϩ ] c , fura2 fluorescence was calculated for the total area of individual cells and the background fluorescence obtained over cell-free regions of each image was subtracted prior to calculation of the fluorescence ratios. In the cells expressing high amounts of fluorescent proteins, detectable GFP/pericam fluorescence was obtained at the excitation wavelengths used for fura2 (40). The GFP-related fluorescence was not affected by changes in [Ca 2ϩ ] c . Although a comparison of the fura2 fluorescence between fluorescent protein-expressing cells and non-transfected cells was used to visualize the kinetics of the signal in each image sequence, to avoid underestimation of the fluorescence ratios reflecting the changes in [Ca 2ϩ ] c , GFP-N versus p130PH-GFP or GFP-IP 3 R 224 -605 and RFP-N versus RFP-p130PH were also analyzed in all of the comparisons. For evaluation of [Ca 2ϩ ] m , the pericam-mt signal was masked. Recordings obtained from all of the transfected cells on the field were averaged for comparison in each experiment. Experiments were carried out with at least four different cell preparations, and 20 -60 cells were monitored in each experiment. Significance of differences from the relevant controls was calculated by Student's t test.
Confocal Imaging of Fluorescence Recovery after Photobleaching (FRAP) Measurements in COS-7 Cells-COS-7 cells were transfected with GFP-N, GFP-IP 3 R 224 -605 , p130PH-GFP, mito-YFP, or PLC␦ 1 PH-GFP. Experiments were performed in 0.25% BSA at 35°C. A Bio-Rad Radiance 2100 confocal laser-scanning system coupled to an Olympus IX70 inverted microscope with a ϫ40 (UApo, NA 1.35) oil immersion objective was used to record image series. The 488-nm line of a krypton/ argon laser was used to excite the GFP. For the FRAP measurements, images were taken continuously at 512 ϫ 512 resolution at a high digital zoom (0.1 m/pixel) with a frame rate of 0.3 Hz. Two 64 m 2 (80 ϫ 80 pixels) regions were selected for bleaching. Two images were recorded pre-bleach followed by five bleaching images with the selected regions illuminated at 16.7 (GFP-N) and 25 times (GFP-IP 3 R 224 -605 , p130PH-GFP, and mito-YFP) the normal scanning intensity. The cell was scanned continuously for 5.5 min after bleaching.
Measurement of [Ca 2ϩ c ] in Suspension of Permeabilized COS-7 Cells-Cells harvested using 0.25% trypsin were washed in an ice-cold Ca 2ϩ -free extracellular buffer containing 120 mM NaCl, 5 mM KCl, 1 mM KH 2 PO 4 , 0.2 mM MgCl 2 , 0.1 mM EGTA, and 20 mM HEPES-NaOH, pH 7.4. Equal aliquots of cells (10 -12 ϫ10 6 cells) were resuspended and permeabilized with 40 g/ml digitonin in 1.4 ml of an intracellular medium (ICM) composed of 120 mM KCl, 10 mM NaCl, 1 mM KH 2 PO 4 , and 20 mM HEPES-Tris, pH 7.2, supplemented with 1 g/ml each of antipain, leupeptin, and pepstatin A. ICM was passed through a Chelex column prior to the addition of protease inhibitors to lower the ambient [Ca 2ϩ ]. All of the measurements were carried out in the presence of 2 mM MgATP and 2 mM succinate. Cytosolic [Ca 2ϩ ] was measured using fura2/FA (0.5 M, excitation 340 and 380 nm; emission, 500 nm) added to the incubation medium. For GFP, 490-nm excitation and 535-nm emission were used. Fluorescence was monitored in a multi-wavelength-excitation dual wavelength-emission fluorimeter (Delta RAM, PTI). Incubations were carried out at 35°C in a stirred thermostated cuvette. Calibration of fura2 signal was carried out at the end of each measurement by adding 1.5 mM CaCl 2 and subsequently 10 mM EGTA-Tris, pH 8.5. [Ca 2ϩ ] c was calculated using a K d of 224 nM.

Fluorometric Measurements of [Ca 2ϩ ] c and [Ca 2ϩ ] m in Suspensions of
Permeabilized RBL-2H3 Cells-Measurements were carried out as described before (22,41). fura2FF-loaded cells (5.5 ϫ 10 6 cells/ml) were permeabilized in ICM supplemented with 25-35 g/ml digitonin for 5 min at 35°C followed by washout of the released cytosolic fura2FF (125 ϫ g for 4 min). Cell permeabilization was evaluated by trypan blue exclusion, and after a 5-min incubation, Ͼ95% cells were trypan bluepositive. Compartmentalized fura2FF has been shown to occur in the mitochondria of RBL-2H3 cells. Permeabilized cells were resuspended in ICM supplemented with succinate 2 mM and 0.25 M rhod2/FA and maintained in a stirred thermostated cuvette at 35°C. Rhod2/FA was added to monitor [Ca 2ϩ ] in the intracellular medium that exchanges readily with the cytosolic space. Fluorescence was monitored in a multiwavelength-excitation dual wavelength-emission fluorimeter using 340-and 380-nm excitation and 500-nm emission for fura2FF and 540-nm excitation and 580-nm emission for rhod2. In every experiment, five data triplets were obtained per second. Calibration of the Ca 2ϩ signals was carried out at the end of each measurement as described previously (41).

Cellular Localization of IP 3 -and Inositol
Lipid-binding Domains Fused to GFP-We first visualized by confocal microscopy the subcellular localization of the IP 3 -binding domains, IP 3 R 224 -605 and p130PH, in COS-7 cells, which were expressed as GFP fusion proteins. GFP-N was used as a control for cytosolic distribution (Fig. 1A, i), whereas PLC␦ 1 PH-GFP that binds to PIP 2 through its PH domain was used as a reference for plasma membrane localization (Fig. 1A, iv). As indicated by the GFP fluorescence, GFP-IP 3 R 224 -605 and p130PH-GFP were found to be present in the cytoplasm (Fig. 1A, ii and iii). The cells were stimulated next with ATP that induces PLC-medi-ated cleavage of PIP 2 to enhance IP 3 formation. In ATP-stimulated COS-7 cells, no change was observed in the distribution of GFP-N, GFP-IP 3 R 224 -605 , and p130PH-GFP ( Fig. 1A, v, vi, and vii). The ATP-induced hydrolysis of PIP 2 was confirmed by the partial translocation of PLC␦ 1 PH-GFP from the membrane to the cytosol, which appeared as a fluorescence decrease at the plasma membrane and an increase in the cytosol (Fig. 1A, viii).
FRAP studies were used to assess the mobility of the IP3BPs in the cytosol (Fig. 1B). When a large area of a p130PH-GFPexpressing cell was illuminated, an essentially homogeneous decrease in fluorescence appeared throughout the cell within 3s (Fig. 1B, i versus v), suggesting fast cytosolic distribution of the fluorescent protein. Furthermore, when small regions of the GFP-IP 3 R 224 -605 -and p130PH-GFP-expressing cells were photobleached, the post-bleach fluorescence in these regions was not different from the pre-bleach signal, indicating that the fluorescence was recovered completely within 3s (Fig. 1B, ii versus vi and iii versus vii). As a positive control, GFP targeted to the mitochondrial matrix, mito-GFP, which shows slower redistribution than the cytosolic GFP (42), was also photobleached. In this case, the loss of fluorescence in the region of bleaching was apparent in the post-bleach image (Fig. 1B, iv versus viii; 29 Ϯ 3%, n ϭ 12). Although the rate of image acquisition was too low to determine the half-recovery time for GFP-IP 3 R 224 -605 and p130PH-GFP, it is likely to be in the subsecond range, similarly to the half-recovery time measured for a freely diffusible cytosolic PH domain-GFP construct (PLC␦1PH(R40L)-GFP, 0.2s) (43). Collectively, the above data suggest that the GFP-IP 3 R 224 -605 and p130PH-GFP are highly mobile cytosolic proteins.
Suppression of IP 3 -induced Ca 2ϩ Release by p130PH-GFP and GFP-IP 3 R 224 -605 in Permeabilized Cells-To evaluate the extent to which the GFP-fused IP 3 -binding domains can buffer IP 3 in the cytoplasm, we analyzed the effect of recombinant p130PH-GFP on IP 3 -induced Ca 2ϩ release in permeabilized COS-7 cells (Fig. 2A). The change of [Ca 2ϩ ] in the cytosolic buffer ([Ca 2ϩ ] c ) was monitored by fura2. First, a Ca 2ϩ pulse (750 nM CaCl 2 ) was added to the permeabilized cells. The added Ca 2ϩ -induced [Ca 2ϩ ] c rise and subsequent rapid decay were not affected by the presence of p130PH-GFP (250 nM), indicating that cytoplasmic Ca 2ϩ buffering and the ER Ca 2ϩ uptake were not influenced by p130PH-GFP ( Fig. 2A, red versus black line). After a steady state [Ca 2ϩ ] c was attained, submaximal IP 3 was added in three steps, 100 nM each, followed by the addition of a supramaximal dose (7.5 M). The three additions of 100 nM IP 3 triggered a dose-dependent increase in [Ca 2ϩ ] c , and the supramaximal IP 3 evoked an additional large [Ca 2ϩ ] c increase ( Fig.  2A). When recombinant p130PH-GFP was present, the [Ca 2ϩ ] c rise elicited by the suboptimal IP 3 was largely attenuated but the [Ca 2ϩ ] c change in response to maximum IP 3 dose was not affected ( Fig. 2A, red versus black line). On average, Ͼ50% inhibition of the [Ca 2ϩ ] c signal evoked by 200 nM IP 3 was obtained in the presence of p130PH-GFP (250 nM), whereas no inhibition was caused by the GFP fusion protein of PLC␦ 1 PH R40L, a PLCd 1 PH mutant that does not bind IP 3 (Fig. 2B, left) (31). The dose response relation also showed highly effective inhibition of the IP 3 -induced [Ca 2ϩ ] c signal by p130PH-GFP (Fig. 2B, right). Similar to p130PH-GFP, GFP-IP 3 R 224 -605 (50 nM) also inhibited the [Ca 2ϩ ] c signal induced by 100 nM IP 3 (40% inhibition) and failed to affect the [Ca 2ϩ ] c response elicited by 7.5 M IP 3 (not shown, n ϭ 2). These results suggest that p130PH-GFP and GFP-IP 3 R 224 -605 were able to compete with the IP3R for IP 3 until IP 3 was available for saturation of both the IP3R and the recombinant IP3BPs. As compared with IP 3 , Tg (2 M), an inhibitor of the ER Ca 2ϩ pumps, induced a slow but progressive [Ca 2ϩ ] c rise that was mediated by Ca 2ϩ leak from the ER (Fig. 2C). p130PH-GFP did not affect the Tg-induced [Ca 2ϩ ] c signal, suggesting that the ER Ca 2ϩ leak was not controlled by p130PH-GFP (Fig. 2C). Thus, IP 3 -binding domains attenuated selectively the IP3R-mediated Ca 2ϩ release from the ER.

Effect of IP 3 -binding Domains on the [Ca 2ϩ ] c Signal Evoked by IP 3 -linked Agonists in
Intact Cells-To further analyze the effect of p130PH and IP 3 R 224 -605 on Ca 2ϩ signaling, agonists activating PLC␤ or PLC␥ were added to intact COS-7 cells expressing either p130PH-GFP or GFP-IP 3 R 224 -605 . Microscopic imaging of GFP fluorescence (490-nm excitation) simultaneously with fura2 fluorescence (340-and 380-nm excitation) allowed us to evaluate the agonist-induced [Ca 2ϩ ] c signal both in transfected and non-transfected (control) cells (Fig. 3). ATP (50 M) that activates PLC␤ through P2-purinergic receptors elicited a robust and rapid global [Ca 2ϩ ] c elevation in control cells (Fig. 3, A and B, 50-versus 66-s image). In contrast, the GFP-positive cells (Fig. 3, A and B, GFP image) did not exhibit a [Ca 2ϩ ] c rise or displayed only a delayed response (Fig. 3 Ca 2ϩ accumulation and release by permeabilized COS-7 cells were monitored in a fluorometer using the ratiometric Ca 2ϩ indicator, fura2/FA. A, cells were first subjected to a 750 nM CaCl 2 pulse that optimized the Ca 2ϩ -loading state of the ER and then exposed to sequential 100 nM  Fig. 3B graph, respectively). To determine the effect of the IP 3 -binding domains on ATP-induced Ca 2ϩ mobilization, stimulation of the cells was also carried out in Ca 2ϩ -free bathing medium (Fig.  3C). In the absence of extracellular Ca 2ϩ , ATP induced a rapidly rising and decaying [Ca 2ϩ ] c spike in control cells and a delayed and relatively small [Ca 2ϩ ] c increase in GFP-IP 3 R 224 -605expressing cells (Fig. 3C). This result further supports that the ATP-stimulated Ca 2ϩ mobilization was effectively inhibited in the presence of the IP 3 -binding domain of the IP3R. EGF (100 ng/ml) that promotes PLC␥-mediated IP 3 formation also elicited a large [Ca 2ϩ ] c signal in control cells, which was delayed as compared with the ATP-induced [Ca 2ϩ ] c signal (Fig. 3, D and E, note the different time scales in panels A-C and D and E). Nevertheless, the expression of GFP-IP 3 R 224 -605 (Fig. 3D) or p130PH-GFP (Fig. 3E) resulted in further extension of the lag time and smaller magnitude of the EGF-stimulated [Ca 2ϩ ] c rise.
To quantitate the effect of expression of cytosolic IP 3 -binding domains on the generation of the [Ca 2ϩ ] c signal, we counted the cells that failed to respond to the agonist and, in the responsive cells, measured the lag time of the [Ca 2ϩ ] c signal (t lag ), the fura2 ratio before stimulation (R 0 ), the change in R fura2 during the [Ca 2ϩ ] c signal (⌬R), and the mean GFP fluorescence (F GFP ) ( Table I). In addition to the cells that did not express GFP-IP 3 R 224 -605 or p130PH-GFP, cells transfected with GFP-N were also used as a control. Both the fraction of cells that did not respond to agonist and the lag time of the [Ca 2ϩ ] c signal were similar in GFP-N cells to that in non-transfected cells (Table I). Notably, R 0 and ⌬R for any of the agonists were smaller in GFP-N cells than in non-transfected cells (Table I). This result is likely to reflect a bleed through the bright F GFP to the records of the fura2 fluorescence (in particular to F 380 ) in the GFP-N cells (Table I). Thus, the cross-talk between GFP and fura2 fluorescence could also contribute to the small ⌬R obtained in GFP-IP 3 R 224 -605 or p130PH-GFP-expressing cells. However, both F GFP and ⌬R were larger in GFP-N cells than in GFP-IP 3 R 224 -605 or p130PH-GFP cells (  (Table I). When compared with non-transfected cells, the GFP-IP 3 R 224 -605 -or p130PH-GFP-expressing cells showed an apparent attenuation of both the initial and the sustained phases of the ATP-and EGF-induced [Ca 2ϩ ] c signal (Fig. 3, A, B, D, and E). However, the difference in the sustained phase is not a reflection of the inhibition of Ca 2ϩ entry because no difference in the sustained phase was observed when the GFP-IP 3 R 224 -605 or p130PH-GFP-expressing cells were compared with GFP-expressing cells (see below, Fig. 6, A  and B). Taken together, these data show that the expression of IP 3 -binding domains suppress [Ca 2ϩ ] c signaling that is dependent on IP 3 formation and IP3R-mediated Ca 2ϩ mobilization. By contrast, no effect appeared on [Ca 2ϩ ] c elevation that was dependent on non-IP3R-mediated discharge of the ER.
After showing that the IP3R-mediated Ca 2ϩ mobilization is effectively inhibited by the recombinant IP3BPs (Fig. 2), their effect on the IP 3 formation was tested (Fig. 4). Because only a small fraction of the transfected cells expressed GFP-IP 3 R 224 -605 (Յ20% efficacy) and no single-cell IP 3 assay was available, we determined the basal and agonist-stimulated total [IP 3 ] in cells transfected with p130PH-GFP (ϳ50% efficacy) or its mutant form unable to bind IP 3 . The basal and ATP-stimulated [ 3 H]IP 3 levels were ϳ20 and 40% higher, respectively, in the p130PH-GFP-transfected cells (Fig. 4). After correction for transfection efficiency, the p130PH-GFPexpressing cells showed ϳ40 and 80% increase in the basal and ATP-stimulated [ 3 H]IP 3 levels, respectively. This result suggests that the buffering effect of p130PH-GFP brought about a compensatory increase in total IP 3 , enabling the cells to keep the basal-free IP 3 levels close to normal and to show an [IP 3 ] signal during stimulation by an agonist. Although the total IP 3 formation was increased, the IP3BPs would have suppressed the free [IP 3 ] rise to attenuate the IP 3 -linked [Ca 2ϩ ] c signal.
Based on the above results, the cytosolic IP3BPs may serve as mobile cytosolic IP 3 -buffers, providing a novel means to study the role of [IP 3 ] elevation in calcium signaling phenomena that have been proposed to depend on regenerative activation of the IP3Rs (e.g. [Ca 2ϩ ] c wave propagation) or to depend on local interactions between IP3R and other Ca 2ϩ channels (e.g. activation of store depletion-induced Ca 2ϩ entry and Ca 2ϩ signal delivery to the mitochondria).

Effect of Cytosolic IP3BPs on [Ca 2ϩ ] c Wave Propagation-To
track the spatio-temporal pattern of the ATP-induced [Ca 2ϩ ] c signal, we carried out fast recording of fura2 fluorescence (F 380 , 15 frames/s) in control and GFP-IP 3 R 224 -605 -expressing COS-7 cells (Fig. 5). To visualize the sites of the [Ca 2ϩ ] c rise, F 380 nm images were differentiated through time (⌬F, purple overlay on gray-scale images). In each of the four control cells surrounding the GFP-IP 3 R 224 -605 -expressing cell (in the center), the [Ca 2ϩ ] c rise started in one region and spread throughout the whole cells in 0.6 s or less (Fig. 5, time series of ⌬F/F380 images). In cells expressing GFP-IP 3 R 224 -605 , the onset of the [Ca 2ϩ ] c rise was delayed as described above, and although the [Ca 2ϩ ] c signal still appeared as a [Ca 2ϩ ] c wave, its propagation was much slower than that in control cells (Fig. 5, lower two rows of  images). Direction of the [Ca 2ϩ ] c wave is marked by arrows for two control cells and a GFP-IP 3 R 224 -605 -expressing cell (upper row, right), and for three regions selected along the wave propagation, the time course of the [Ca 2ϩ ] c rise (decrease in F 380 fluorescence) is shown (Fig. 5, right; control, upper and middle;  Interestingly, a wave of IP 3 production has been recently claimed to underlie the fertilization Ca 2ϩ wave in Xenopus oocytes (45). Because PIP 2 may exert an inhibition on the IP3R Ca 2ϩ channel (46), cleavage of PIP 2 in the vicinity of the IP3Rs would not only provide IP 3 for channel activation but it could also relieve an inhibition elicited by PIP 2 . The effect of IP3BPs may also involve attenuation of the effect of IP 3 formed close to the IP3Rs.
Effect of Cytosolic IP3BPs on Store-operated Ca 2ϩ Entry-Comparison of the [Ca 2ϩ ] c signal of the GFP-IP 3 R 224 -605 -or p130PH-GFP-expressing cells with that of cells expressing GFP alone showed that the IP3BP-dependent inhibition of the [Ca 2ϩ ] c signal was confined to the initial phase (2-5 min) of the stimulation by both ATP and EGF (Fig. 6, A and B). To separate Ca 2ϩ entry from Ca 2ϩ mobilization, the cells were placed next into a Ca 2ϩ -free medium stimulated with ATP and Ca 2ϩ was added back at 5 min of the stimulation (Fig. 6C). Also, the IP3BP was fused to monomeric RFP to minimize the bleed through to the fura2 fluorescence and Tg was added together with ATP to result in similar Ca 2ϩ store depletion in each condition. The non-transfected and the RFP-alone transfected cells showed a robust ATP ϩ Tg-induced [Ca 2ϩ ] c spike and, in response to the Ca 2ϩ back addition, a rapid and massive elevation of [Ca 2ϩ ] c (Fig. 6C, black and red traces). The RFP-p130PH-expressing cells exhibited a slow and relatively small ATP ϩ Tg-induced [Ca 2ϩ ] c spike, but the effect of the Ca 2ϩ back addition was as prompt and large as it was in the non-trans- fected or RFP-expressing cells (Fig. 6C, green trace). Based on these results, the IP3BPs do not appear to affect the Ca 2ϩ entry evoked by the IP 3 -linked agonists.
Store-operated Ca 2ϩ entry can be activated by ER Ca 2ϩ depletion both in the presence and absence of a rise in total cellular [IP 3 ]. Recent studies have suggested that a portion of the IP 3 -sensitive Ca 2ϩ store is closely associated with the plasma membrane and interact with the store-operated Ca 2ϩ channels in an IP 3 -dependent manner to stimulate Ca 2ϩ entry (conformational coupling, reviewed in Ref. 14). IP3BPs could have failed to inhibit the Ca 2ϩ influx in the ATP-stimulated cells if the agonist-induced IP 3 signal was supramaximal for the Ca 2ϩ entry. To determine whether any fluctuations in IP 3 are involved in the control of the coupling between Ca 2ϩ release and entry, we investigated the Ca 2ϩ entry in GFP-IP 3 R 224 -605expressing cells using two fura2-imaging approaches. In the first protocol, cells were incubated in the absence of extracellular Ca 2ϩ , the ER Ca 2ϩ store was discharged using Tg, and then Ca 2ϩ was added back to initiate Ca 2ϩ entry (Fig. 7, A and  B). The Tg-induced Ca 2ϩ release appeared as a gradually rising and decaying [Ca 2ϩ ] c signal, whereas Ca 2ϩ entry gave rise to a rapid and massive [Ca 2ϩ ] c signal (Fig. 7, A and B). The time course of the [Ca 2ϩ ] c signal resulting from intracellular Ca 2ϩ mobilization and Ca 2ϩ entry was not changed by the expression of GFP-N or GFP-IP 3 R 224 -605 in the cells (Fig. 7, A and B, thick  lines). As a measure of the Ca 2ϩ entry, we calculated the lag time of the Ca 2ϩ addition-induced Ca 2ϩ signal (Fig. 7C). There was no significant difference among control (6.99 Ϯ 0.28 s), GFP-N (6.86 Ϯ 0.32 s), and GFP-IP 3 R 224 -605 (7.13 Ϯ 0.33 s). In the second approach, we used Mn 2ϩ as a surrogate for Ca 2ϩ to characterize unidirectional ion flux through store depletionactivated Ca 2ϩ channels (Fig. 7, D and E). The rate of Mn 2ϩ transport through this pathway was shown by the quench of fura2 fluorescence measured at 360 nm, a Ca 2ϩ -independent excitation wavelength. Specifically, cells were pretreated with Tg for 300 s to deplete the ER Ca 2ϩ store and to activate Ca 2ϩ entry and then MnCl 2 was added (Fig. 7, D and E). The time course of Mn 2ϩ quench was not affected by the presence of GFP-IP 3 R 224 -605 (Fig. 7, E versus D). These results show that buffering of IP 3 does not interfere with the activation of Ca 2ϩ entry during depletion of the ER Ca 2ϩ store. Notably, we tested Ca 2ϩ entry only after complete store depletion had been reached; therefore, the results do not exclude the possibility that the kinetic of the activation of Ca 2ϩ entry is modulated by IP 3 (reviewed in Ref. 47).
Effect of Cytosolic IP3BPs on the Propagation of [Ca 2ϩ ] c Signals into the Mitochondria-In a wide variety of cell types, the IP 3 -induced Ca 2ϩ mobilization is effectively propagated to the mitochondria (for a summary see Table I in Ref. 48). Mitochondria form local interactions with subdomains of the ER (24), and at the sites of interaction, mitochondrial Ca 2ϩ uptake is tightly coupled to IP3R-mediated Ca 2ϩ release reminiscent of the organization of synaptic transmission (22). However, the functional organization of the local communication remains elusive. Using the IP3BPs, we tested the IP 3 requirements for the ER-mitochondrial Ca 2ϩ coupling. We first used a permeabilized cell model that allowed fluorometric measurement of [Ca 2ϩ ] m simultaneously with [Ca 2ϩ ] c during IP 3 -induced Ca 2ϩ mobilization (22,49). As we noted for the [Ca 2ϩ ] c signal (Fig. 2), the [Ca 2ϩ ] m elevations evoked by the addition of a submaximal IP 3 were also suppressed in the presence of p130PH-GFP (125 nM) (Fig. 8, upper left)  contribute to the relatively large inhibition of the IP 3 -induced [Ca 2ϩ ] m rise by p130PH-GFP. By lowering the dose of IP 3 , we found that 125 nM IP 3 evoked a [Ca 2ϩ ] c signal identical to that of 200 nM IP 3 in the presence of 125 nM p130PH-GFP (Fig. 8). Furthermore, 125 nM IP 3 elicited a [Ca 2ϩ ] m rise that was comparable to the effect of 200 nM IP 3 in the presence of 125 nM p130PH-GFP (Fig. 8). Based on these data, IP3BPs suppress the Ca 2ϩ release mediated by the IP3Rs that provide Ca 2ϩ for mitochondrial Ca 2ϩ uptake. Furthermore, suppression of Ca 2ϩ release results in attenuation of the [Ca 2ϩ ] m signal more than the [Ca 2ϩ ] c rise. An important clue to this point is that a second-order relationship exists between [Ca 2ϩ ] c and the mitochondrial Ca 2ϩ uptake (for review see Ref. 50).
To evaluate in intact cells the calcium signal propagation to the mitochondria, we carried out measurements of [Ca 2ϩ ] m with ratiometric pericam-mt (37). Because the fluorescence spectrum of pericam-mt is similar to the spectrum of enhanced GFP, p130PH was introduced to the cells fused to RFP. Simultaneous recording of F RFP , F pericam-mt , and F fura2 was achieved by combination of a filter wheel with a motorized turret, which permitted the selection of the appropriate excitation filters and filter cubes containing the dichroic reflectors and emission filters. Although this design allowed us to minimize the crosstalk between fluorophores, F pericam-mt was still detectable at 380-nm excitation, one of the excitation wavelengths used for fura2. Because pericam-mt was always excluded from the nucleus, the nuclear area was selected in each cell to calculate [Ca 2ϩ ] c . As shown in Fig. 9, most of the cells expressing either RFP-N or RFP-p130PH also expressed pericam-mt (upper two rows of images) and the subcellular distribution of pericam-mt was similar to the distribution of enhanced GFP targeted to the mitochondria (Fig. 1B, iv). Furthermore, during stimulation with ATP, the cells expressing RFP alone displayed a substantial change in pericam-mt fluorescence, and the change in fluorescence followed distribution of the mitochondria. In contrast, the RFP-p130PH-expressing cells showed little or no change in pericam-mt fluorescence (Fig. 8, third row of images). To quantitate [Ca 2ϩ ] c and [Ca 2ϩ ] m , for every cell, the fura2 and pericam ratios were obtained and then the cell population average was calculated (Fig. 9, graphs) Because individual mitochondria appear to utilize discrete paths of local communication to access the Ca 2ϩ released from the ER, subpopulations of mitochondria could be less sensitive to the increase in IP 3 -buffering capacity than others. We acquired images at higher spatial resolution to evaluate the ATPinduced [Ca 2ϩ ] m signal in several small groups of mitochondria in individual RFP-p130PH-expressing cells (Fig. 10). The large [Ca 2ϩ ] c spike was not associated with a considerable [Ca 2ϩ ] m increase in any of the mitochondria (Fig. 10, red traces).
To  (Fig. 11). The distribution of the points representing the RFP-p130PH cells suggests that these cells show a relatively small [Ca 2ϩ ] m rise at any given [Ca 2ϩ ] c signal (Fig. 11,  upper panel). As an example, the mean [Ca 2ϩ ] c and [Ca 2ϩ ] m responses were calculated for the RFP and RFP-p130PH cells that displayed 50 -150% increase in the fura2 ratio. The mean [Ca 2ϩ ] c signal was similar in both the control and the IP 3 buffer-expressing cells, but the [Ca 2ϩ ] m rise was reduced by Ͼ80% in the RFP-p130PH cells (Fig. 11, lower panel)  ship between [Ca 2ϩ ] c and mitochondrial Ca 2ϩ uptake/[Ca 2ϩ ] m , which has been documented for both isolated mitochondria (for review see Ref. 50) and permeabilized cells (22,23). The Ca 2ϩ dependence of the [Ca 2ϩ ] m is apparently unaffected by IP3BPs in the permeabilized cells, because in this model, both the [Ca 2ϩ ] c and [Ca 2ϩ ] m signals evoked by a given dose of IP 3 in the presence of an IP3BP could be reproduced by the addition of a lower dose of IP 3 alone (Fig. 8). However, in the intact cells, the mitochondria were effectively dissociated from the global [Ca 2ϩ ] c signal by the IP3BPs (Figs. 9 and 10) and depression of the [Ca 2ϩ ] c dependence of the [Ca 2ϩ ] m signal was also shown (Fig. 11). The key to this point seems to be that the Ca 2ϩ uptake of the mitochondria depends on a local perimitochondrial [Ca 2ϩ ] c rise mediated by the adjacent IP3Rs (21,22,24). We have shown that activation of a cluster of Ca 2ϩ release sites is sufficient to evoke a [Ca 2ϩ ] m signal ("Ca 2ϩ mark") (26) but that optimal activation of the mitochondrial Ca 2ϩ uptake sites requires coordinated activation of the IP3Rs that deliver Ca 2ϩ to the mitochondria (22). Such coordinated activation of the IP3Rs is expected to be largely affected in the presence of an IP 3  ] m signal was more sensitive to IP3BPs in intact cells than in permeabilized cells, because IP 3 was added as a bolus to the permeabilized cells, whereas it was produced by PLCs in intact cells, enabling the IP3BPs to decrease the rate of [IP 3 ] rise.

CONCLUSIONS
The most important observation of this study is that the propagation of the [Ca 2ϩ ] c signal to the mitochondria is strikingly sensitive to the IP 3 -buffering power of the cytoplasm. Thus, the kinetics and spatial distribution of the [IP 3 ] rise appear to control the recruitment of the mitochondrial Ca 2ϩ buffering and ATP production to the [Ca 2ϩ ] c signal. Our work also evaluated for, the first time, the effect of IP 3 buffers on Ca 2ϩ wave propagation and on store-operated Ca 2ϩ entry and has provided evidence that the rapid increase in [IP 3 ] is important for the regenerative mechanism of the [Ca 2ϩ ] c wave but that rapid fluctuations in [IP 3 ] may not be required for the optimal activation of Ca 2ϩ entry by store depletion.
The IP 3 signal has been envisioned to originate from the G qor tyrosine kinase-stimulated activation of PLC and hydrolysis of PIP 2 at the plasma membrane and to utilize rapid diffusion of IP 3 throughout the cells. However, recent studies have indicated that Ca 2ϩ -induced IP 3 formation may also occur at intra- The data are representative of experiments repeated six times. cellular membranes (9 -11). Furthermore, IP 3 -binding proteins are expressed in several cell types and may also contribute to the shaping of the [IP 3 ] signal (e.g. p130) (29 -33). Thus, the [IP 3 ] signal may display a complex spatio-temporal pattern. Overexpression of IP 3 -binding proteins provided us with a tool to change the temporal pattern of the [IP 3 ] rise and to interfere with spatially confined fluctuations in [IP 3 ]. Our data indicate that the kinetics of the [IP 3 ] signal affects both the initiation and the propagation of the [Ca 2ϩ ] c signal. Along this line, regional differences in IP 3 production and buffering may converge with regional differences in the Ca 2ϩ feedback on the IP3R/RyR to generate regional differences in [Ca 2ϩ ] c wave velocity, which have been observed in several models. If an increase in IP 3 buffering leads to slower propagation of the [Ca 2ϩ ] c wave but at each location the size of the [Ca 2ϩ ] c rise is relatively stable, the activation of cytoplasmic Ca 2ϩ effectors may remain unaffected. Also, IP 3 buffers fail to modulate the store depletion-induced Ca 2ϩ entry, providing evidence that neither a global nor a local increase in IP 3 production is involved in the coupling between massive ER Ca 2ϩ depletion and store-operated Ca 2ϩ entry. By contrast, the presence of IP 3 buffers results in a dramatic inhibition of mitochondrial Ca 2ϩ accumulation during the agonist-induced [Ca 2ϩ ] c signal. The large effect of the IP 3 buffer on mitochondrial Ca 2ϩ uptake could be partly explained by its non-linear dependence on extramitochondrial [Ca 2ϩ ]. However, we demonstrate that the relationship between the [Ca 2ϩ ] c and [Ca 2ϩ ] m signals is also suppressed when the IP 3 buffering is enhanced. A clue to this problem is that coordinated activation of the IP3Rs seems to be important for mitochondrial Ca 2ϩ uptake (22) and that IP 3 buffers may influence the rate of global [IP 3 ] increase or attenuate a local [IP 3 ] signal to compromise synchronized opening of the IP3Rs.
Taken together, the present results show IP3BPs as a valuable tool in the study of the mechanisms of local calcium signaling and indicate that a dynamic local interplay between Ca 2ϩ and IP 3 is of great significance for the propagation of the calcium signal throughout the cytosol and into the mitochondria. Furthermore, based on their relatively large effect on the [Ca 2ϩ ] m signal, IP3BPs may provide a tool to selectively attenuate the mitochondrial effects of the [Ca 2ϩ ] c signal and can be useful to determine the mitochondrial contribution in a range of signaling cascades.