The Type III Inositol 1,4,5-Trisphosphate Receptor Preferentially Transmits Apoptotic Ca2+ Signals into Mitochondria*

There are three isoforms of the inositol 1,4,5- trisphosphate receptor (InsP3R), each of which has a distinct effect on Ca2+ signaling. However, it is not known whether each isoform similarly plays a distinct role in the activation of Ca2+-mediated events. To investigate this question, we examined the effects of each InsP3R isoform on transmission of Ca2+ signals to mitochondria and induction of apoptosis. Each isoform was selectively silenced using isoform-specific small interfering RNA in Chinese hamster ovary cells, which express all three InsP3R isoforms. ATP-induced cytosolic Ca2+ signaling patterns were altered, regardless of which isoform was silenced, but in a different fashion depending on the isoform. ATP also induced Ca2+ signals in mitochondria, which were inhibited more effectively by silencing the type III InsP3R than by silencing either the type I or type II isoform. The type III isoform also co-localized most strongly with mitochondria. When apoptosis was induced by activation of either the extrinsic or intrinsic apoptotic pathway, induction was reduced most effectively by silencing the type III InsP3R. These findings provide evidence that the type III isoform of the InsP3R plays a special role in induction of apoptosis by preferentially transmitting Ca2+ signals into mitochondria.

Real Time Polymerase Chain Reaction-Real time PCR analysis was performed using an Applied Biosystems 7700 sequence detector (Foster City, CA) as described (22). Sets of primers were chosen for each InsP 3 R isoform and for glyceraldehyde-3-phosphate dehydrogenase to give PCR products less than 100 base pairs in length. Dual labeled fluorogenic probes complementary to a sequence within each PCR product were added to the PCRs. Primers and probes were custom synthesized by Applied Biosystems. Reverse transcription was performed on 5 g of CHO total RNA with SuperScript II Ribonuclease H2 reverse transcriptase from Invitrogen according to the manufacturer's protocol. Complementary DNA was amplified in a 50-l volume containing 25 l of 2ϫ TaqMan Universal PCR Master Mix (Applied Biosystems), 100 nmol/liter probe (Applied Biosystems), and 300 nmol/liter each primer. After a denaturing step at 95°C for 10 min, 40 cycles were performed at 95°C for 15 s and then 60°C for 1 min. Mathematical analysis of the results was performed as recommended by the manufacturer.
Reverse Transcription-PCR-Reverse transcription-PCR was performed to determine whether CHO cells express the ryanodine receptor. Total RNA from CHO cells was isolated using Trizol reagent. First strand cDNA was synthesized using the primer oligo-(dT) 16 and Moloney murine leukemia virus reverse transcriptase. RNA samples were then subjected to DNase-and RNase-free treatment to extract any possible genomic DNA. A negative control was carried out in which RNA but no reverse transcriptase was added (RNA control). Degenerate primers were designed to amplify a 530-base pair product from a portion of the 3Ј region common to all three ryanodine receptor isoforms (23). PCR amplification then was performed in a PTC-100 automated thermocycler (MJ Research, Watertown, MA) using 2 l of the first strand cDNA reaction, 200 nM each primer, 200 M dNTPs, 2.5 mM FIGURE 1. CHO cells express all three InsP 3 R isoforms. Real time quantitative PCR was used to measure the relative mRNA distribution for each InsP 3 R isoform. The type II isoform is most abundant in CHO cells, present in an InsP 3 R/glyceraldehyde-3-phosphate dehydrogenase ratio of 2.43. Isoforms I and III are present in ratios of 1.29 and 1.04, respectively. FIGURE 2. Silencing of specific InsP 3 R isoforms. CHO (A-C) or HEK293 (D) cells were transfected with 40 nM siRNA for InsP 3 R-I, InsP 3 R-II, InsP 3 R-III, or scrambled InsP 3 R-III, followed by a 48-h incubation. Immunoblot of whole-cell protein from CHO cells demonstrates that siRNA for InsP 3 R-I (A), InsP 3 R-II (B), and InsP 3 R-III (C) specifically knocks down expression of each InsP 3 R isoform. Expression of ␤-actin serves as a loading control. Densitometric analysis confirms reduction of InsP 3 R-I expression by 99 Ϯ 0.3% (A), InsP 3 R-II by 70 Ϯ 0.6% (B), and InsP 3 R-III by 97 Ϯ 0.6% (C), compared with control. Each result is representative of four individual experiments. D, immunoblot of whole-cell protein from HEK293 cells demonstrates that siRNA for InsP 3 R-I, InsP 3 R-II, and InsP 3 R-III specifically knocks down expression of each InsP 3 R isoform in this cell type as well. Each result is representative of three individual experiments. MgCl 2 , and 2.5 units of AmpliTaq DNA polymerase for a total volume of 100 l. The PCR samples were subjected to hot start (2 min at 94°C), followed by 30 cycles at 94°C, 1 min at 50°C, and 1 min at 72°C. The reaction was followed by a final extension at 72°C for 10 min, and the PCR product was electrophoretically size-fractionated in polyacrylamide gel.
Preparation of siRNA-Potential target sites within the rat InsP 3 R genes were selected and then searched with NCBI BlatN to confirm specificity for each InsP 3 R isoform. The siRNAs for the type I, II, and III InsP 3 R and an siRNA containing the same nucleotides for type III InsP 3 R but in a scrambled sequence were prepared by a transcriptionbased method using the Silencer kit according to the manufacturer's instruction. The sense and antisense oligonucleotides of siRNAs were, respectively, as follows: type I, 5Ј-AAAGTTGTAGCTGCTGGT-GCTCCTGTCCTC-3Ј and 5Ј-AAAGCACCAGCAGCTACAACTC-CTGTCCTC-3Ј; type II, 5Ј-AACAGCCTAATCAAGATCTCCCCT-GTCTC-3Ј and 5Ј-AAGGAGATCTTGATTAGGCTGCCTGTCTC-3Ј; type III, 5Ј-ATGGTGCTGGCAAACTTGTTTCCTGTCCTC-3Ј and 5Ј-AAACAAGTTTGCCAGCACCATCCTGTCCTC-3Ј; scrambled Type III, 5Ј-AACAGCTACAAAGCTTCTGCACCTGTCTC-3Ј and 5Ј-AATGCAGAAGCTTTGTAGCTGCCTGTCTC-3Ј.
Transfection of pNTCP-GFP and siRNA-CHO cells were maintained in culture at 37°C in an atmosphere of 5% CO 2 for 24 h prior to transfections. For studies of bile acid-induced apoptosis, cells were transfected using Effectene with 0.5 g of pNTCP-GFP, a construct encoding a green fluorescent protein (GFP)-tagged bile acid transporter (24). For siRNA studies, the cells were washed and supplied with 1 ml of fresh tissue culture medium, and then 1 g of siRNA and 3 l of transfection reagent were mixed with 100 l of tissue culture medium. The mixture was incubated for 15 min at room temperature for complex formation, and then 900 l of tissue culture medium was added to the mixture, and this solution was placed dropwise onto the cells, resulting in a final siRNA concentration of 40 nM. The cells were incubated at 37°C in an atmosphere of 5% CO 2 for 48 h prior to use.
Immunoblotting-Standard methods were used for immunoblots (25,26). Briefly, cells grown in 35-mm dishes were washed three times with phosphate-buffered saline (PBS) and solubilized in 80 l of Non-idet P40 containing a protease inhibitor mixture (Roche Applied Science). The protein concentration was determined spectrophotometrically, and 40 g of protein was separated by electrophoresis in a 6% polyacrylamide gel and then transferred to an Immobilon polyvinylidene membrane. The membrane was blocked with 5% skim milk in PBST (PBS plus 0.1% Tween 20) for 60 min and then incubated with primary antibody. The primary antibodies used were as follows: rabbit anti-InsP 3 R-I polyclonal antibody (1:1000), rabbit anti-InsP 3 R-II polyclonal antibody (1:50), mouse anti-InsP 3 R-III monoclonal antibody (1:500), and rabbit anti-actin monoclonal antibody (1:400). This incubation was carried out for 2 h at room temperature. After three washes with PBST, the membrane was incubated with peroxidase-conjugated secondary antibody (1:5000) for 1 h at room temperature. Bands were revealed by enhanced chemiluminescence (ECL plus; Amersham Biosciences). The film was scanned with a GS-700 imaging densitometer (Bio-Rad), and then quantitative analysis was performed using Multi-Analyst software (Bio-Rad).
Immunofluorescence-Confocal immunofluorescence was performed as described previously (25,26). Cells were fixed with 4% paraformaldehyde in PBS for 10 min and then washed three times in PBS. The cells were incubated for 1 h in blocking solution (PBS containing 1% bovine serum albumin, 0.5% Triton, 5% goat serum), and then incubated for 2 h at room temperature in PBS with 1% bovine serum albumin containing one of the following primary antibodies: anti-InsP 3 R-I polyclonal antibody (diluted 1:100), anti-InsP 3 R-II polyclonal antibody (diluted 1:25), or anti-InsP 3 R-III monoclonal antibody (diluted 1:100). The cells were washed three times in PBS and incubated for 1 h in PBS with 1% bovine serum albumin containing secondary antibody conjugated to Alexa-488 (diluted 1:500). Negative controls were stained with secondary antibody alone. Cells were washed six times, and coverslips were mounted with an antifade reagent. Immunofluorescence images were obtained with a Zeiss LSM 510 confocal microscope (Thornwood, NY) using a ϫ63 water immersion objective. The 488-nm line of an argon laser was used to excite Alexa-488 and emission was collected between 505 and 530 nm. To ensure specificity of staining, images were obtained using confocal machine settings at which no Alexa-488 fluorescence was detectable in negative control specimens labeled with the secondary antibody alone. InsP 3 R immunofluorescence was determined, keeping the pinhole and detector gain setting identical while analyzing wild type CHO cells and CHO cells transfected with InsP 3 R siRNA. For mitochondrial co-localization studies, CHO cells were incubated with 500 nM Mitotracker Red for 20 min at room temperature and then processed and examined as described above, using excitation at 543 nm with observation at 560 nm to detect Mitotracker Red. Co-localization of the mitochondrial label and InsP 3 R staining was quantified using Image-J software (available on the World Wide Web at rsb.info.nih.gov/ij/).
Detection of Cytosolic and Mitochondrial Ca 2ϩ Signals-Cytosolic and mitochondrial Ca 2ϩ were monitored in individual CHO cells by time lapse confocal microscopy as described previously (25,26). CHO cells were cultured on glass coverslips, incubated for 30 min in the presence of 6 M Fluo-4/AM to monitor cytosolic Ca 2ϩ or 6 M rhod-2/AM to monitor mitochondrial Ca 2ϩ and then transferred to a perfusion chamber on the stage of a Bio-Rad MRC-1024 confocal microscope. Cells were in a HEPES-buffered solution during experiments and were observed using a ϫ63, 1.4 numerical aperture objective lens. The 488-nm line of a krypton/argon laser was used to excite the dye, and emission signals between 505 and 550 nm were collected. Cells were stimulated with 2 M ATP, and images were acquired at a rate of 2-10 FIGURE 3. Subcellular localization of InsP 3 R isoforms in CHO cells. Confocal immunofluorescence with isoform-specific antibodies demonstrates that each isoform is distributed heterogeneously in the cytosol, with variable nuclear staining (scale bar, 10 m). Nonspecific staining by secondary antibody was not seen (not shown). Expression of each isoform is specifically silenced by its respective siRNA. Results are representative of five independent experiments. frames/s. Neither autofluorescence nor background signals were detectable at the machine settings used.
Induction of Apoptosis-Apoptosis was induced in two separate ways. To induce apoptosis via the intrinsic pathway, CHO cells were treated with 0.5 M staurosporine at 37°C in 5% CO 2 for 12 h (27). To induce apoptosis via the extrinsic pathway, CHO cells first were transiently transfected with the GFP-tagged bile acid transporter, Na ϩ -taurocholate co-transport polypeptide (NTCP), and then were treated with 50 nM of the hydrophilic bile acid GCDCA at 37°C in 5% CO 2 for 2 h (28). In either case, apoptosis was quantified by loading the cells with 1 M of the nuclear binding dye DAPI (28), and then chromatin condensation and nuclear fragmentation were assessed by confocal microscopy.
Statistical Analysis-All experiments were performed in at least triplicate, and results are expressed as mean Ϯ S.E. Statistical analyses were performed using PRISM statistical software (GraphPad; San Diego, CA). Groups of data were compared using one-way analysis of variance. A value of p Ͻ 0.05 was considered to indicate a statistically significant difference.

Expression and Knockdown of InsP 3 R Isoforms-CHO cells were used
for these studies, because this epithelial cell line expresses all three InsP 3 R isoforms (4). Real time PCR (Fig. 1) confirmed previous studies at the protein level indicating that CHO cells express all three isoforms (4) and suggests that the InsP 3 R-II is most highly expressed, followed by the type I and then the type III isoform (Fig. 1). PCR additionally showed that CHO cells do not express the ryanodine receptor (data not shown), the other intracellular Ca 2ϩ release channel that is present in some epithelial cells (29,30). Together, these data show that InsP 3 R is the principal intracellular Ca 2ϩ release channel in CHO cells. The data also confirm that all three isoforms of InsP 3 R are expressed in CHO cells and show that the type II InsP 3 R is the predominant isoform at the RNA level. Next, siRNA constructs were used to silence each isoform of the InsP 3 R. Immunoblot analysis using isoform-specific antibodies demonstrated that the siRNA constructs were able to specifically and selectively knock down InsP 3 R-I ( Fig. 2A), InsP 3 R-II (Fig. 2B), or InsP 3 R-III (Fig. 2C), respectively. In contrast, scrambled siRNA for InsP 3 R-III had To the right is the summary of the different patterns of Ca 2ϩ signaling when cells were stimulated with ATP. An oscillatory pattern of Ca 2ϩ signaling predominates in wild type cells. Compared with these controls, an increase in the fraction of cells responding with a plateau pattern of Ca 2ϩ signaling is observed when cells were transfected with siRNA for InsP 3 R-I or InsP 3 R-III. On the other hand, an increase in the number of cells responding with a transient pattern of Ca 2ϩ signaling is observed when cells were transfected with siRNA for InsP 3 R-II. Results are representative of those seen in three separate experiments (n ϭ 70 cells) for control CHO cells, three separate experiments (n ϭ 70 cells) for cells transfected with siRNA for InsP 3 R-I, three separate experiments (n ϭ 70 cells) for cells transfected with siRNA for InsP 3 R-II, and three separate experiments (n ϭ 70 cells) for cells transfected with siRNA for InsP 3 R-III (*, p Ͻ 0.001 relative to control cells). no effect on InsP 3 R gene expression (Fig. 2C). Densitometric analysis confirmed reduction of InsP 3 R-I expression by 99% ( Fig. 2A), InsP 3 R-II by 70% (Fig. 2B), and InsP 3 R-III by 97% (Fig. 2C), each compared with nontransfected control. The effect of each siRNA was specific for the respective InsP 3 R isoform, since the amount of other InsP 3 R isoforms or ␤-actin was unaffected (Fig. 2, A-C). To further evaluate the siRNA constructs, they were used to knock down each InsP 3 R isoform in HEK293 cells, which also express all three receptor isoforms (Fig. 2D). Densitometric analysis of isoform-specific immunoblots showed reduction of InsP 3 R-I expression by 88%, InsP 3 R-II by 81%, and InsP 3 R-III by 70% (Fig. 2D), each compared with nontransfected control. Efficacy of each siRNA was slightly lower in HEK293 cells than in CHO cells, which may reflect the fact that the siRNAs were designed for rodent InsP 3 R sequences, whereas HEK293 cells are of human origin. As in CHO cells, however, the effect of each siRNA was specific for the respective InsP 3 R isoform. The effects of each siRNA construct were then confirmed by examining the subcellular localization of InsP 3 Rs in CHO cells using confocal immunofluorescence (Fig. 3). CHO cells were labeled with the same InsP 3 R-I, II, or III antibodies used for immunoblots. Both InsP 3 R-I and InsP 3 R-II were distributed heterogeneously throughout the cell (Fig. 3, left and middle). On the other hand, InsP 3 R-III was distributed mainly in the cytosol (Fig. 3, right). Immunofluorescence also showed that each siRNA specifically knocked down InsP 3 R-I, -II, or -III (Fig. 3), consistent with immunoblot results. Together, these findings demonstrate that InsP 3 R isoforms are distributed in heterogeneous patterns in CHO cells and that the siRNA constructs used here are highly efficient and specific for silencing each InsP 3 R isoform.
Each InsP 3 R Isoform Affects Ca 2ϩ Signaling Patterns in CHO Cells-Although each of the three InsP 3 R isoforms acts as an InsP 3 -gated Ca 2ϩ channel, the isoforms are not uniformly sensitive to InsP 3 (6). Isoformspecific differences in tissue expression and subcellular distribution further suggests that the various InsP 3 Rs serve distinct roles in Ca i 2ϩ signaling. Since there is no selective antagonist for each InsP 3 R isoform, we used the isoform-specific siRNAs to test this. Ca i 2ϩ signaling was induced by extracellular stimulation of the cells with ATP (2 M), since this nucleotide binds to P2Y nucleotide receptors expressed on the plasma membrane of CHO cells, and since activation of these receptors initiates the InsP 3 -induced Ca 2ϩ signaling cascade (31). Ca i 2ϩ signaling was monitored in individual CHO cells loaded with the Ca 2ϩ -sensitive fluorescent dye fluo-4/AM using time lapse confocal microscopy ( (2.5 Ϯ 2.5%). Together, these results suggest that InsP 3 preferentially induces Ca i 2ϩ oscillations in CHO cells but that a sustained increase in Ca i 2ϩ is the next most common pattern when InsP 3 R-II is expressed, whereas a transient Ca i 2ϩ increase is more common if InsP 3 R-II expression is suppressed.

Each InsP 3 R Isoform Has a Differential Effect on Mitochondrial Ca i 2ϩ
Signaling-Uptake of Ca i 2ϩ by mitochondria alters Ca i 2ϩ signaling patterns (32, 33), in part by modulating the feedback effect of Ca 2ϩ on the InsP 3 R, and by defining the threshold for InsP 3 to trigger Ca i 2ϩ signals in different subcellular regions (34). Because silencing each InsP 3 R isoform led to changes in Ca i 2ϩ signaling patterns, we examined the subcellular location of each isoform relative to the distribution of mitochondria. A close proximity between mitochondria and certain subpopulations of InsP 3 Rs and regions of the endoplasmic reticulum has already been reported (35). We used confocal immunofluorescence to determine whether mitochondria preferentially co-localize with specific InsP 3 R isoforms. Some co-localization was seen between mitochondria and each of the three InsP 3 R isoforms (Fig. 5A). In order to determine the extent of co-localization for each isoform, ImageJ software was used to quantify the fraction of pixels within the cytosol in which there was labeling for both the specific InsP 3 R isoform and mitochondria (Fig. 5B). Mitochondria co-localized most extensively with the type III InsP 3 R (28.9 Ϯ 2.8% of pixels), followed by the type II InsP 3 R, which co-localized with mitochondria in only 9.6 Ϯ 1.7% of pixels. The type I InsP 3 R co-localized the least with mitochondria, in only 3.6 Ϯ 0.7% of pixels (p Ͻ 0.001 by analysis of variance). Since mitochondria can take up a significant fraction of the Ca 2ϩ released by nearby InsP 3 Rs, we also investigated free mitochondrial Ca 2ϩ (Ca m 2ϩ ) signals induced by stimulation of CHO cells with ATP. In this system, the cationic Ca 2ϩ dye

. Co-localization of InsP 3 R isoforms and mitochondria in CHO cells. A, cells were double-labeled with isoform-specific InsP 3 R antibodies (green) and
Mitotracker Red (red) to identify mitochondria. The subcellular distribution of each InsP 3 R isoform, along with mitochondria, was determined by confocal immunofluorescence. B, areas of co-localization appear yellow in the merged image (best appreciated in the magnified region of cytosol at the right). The fraction of the pixels within the cytosol that are yellow was quantified for each InsP 3 R isoform (bottom) and revealed that co-localization with mitochondria follows a rank order of InsP 3 R-III Ͼ InsP 3 R-II Ͼ InsP 3 R-I (p Ͻ 0.001 among groups by analysis of variance). Results are mean Ϯ S.E. of measurements made in 12 cells for each isoform.
rhod-2 co-localized nearly completely with the mitochondrial dye Mit-oFluo Green (Fig. 6A), indicating that rhod-2 was specifically loaded into mitochondria. Ca m 2ϩ signals, as measured by rhod-2 fluorescence, were quantified in both control and siRNA-treated cells stimulated with 2 M ATP (Fig. 6B). Rhod-2 fluorescence increased by 1230 Ϯ 83% in control cells stimulated with ATP (n ϭ 62 cells). Similarly, rhod-2 fluorescence increased by 1173 Ϯ 102% in cells pretreated with InsP 3 R-I siRNA (n ϭ 36; p Ͼ 0.05 relative to nontransfected controls) and by 1104 Ϯ 95% in cells pretreated with InsP 3 R-II siRNA (n ϭ 43; p Ͼ 0.05 relative to controls). In contrast, silencing InsP 3 R-III led to a fluorescence increase of only 838 Ϯ 55% (n ϭ 62; p Ͻ 0.01 relative to controls). This represents only 68% of the increase that was observed in nontransfected controls (Fig. 6C). To examine the generality of these findings, mitochondrial Ca 2ϩ signals also were observed in HEK293 cells treated with each type of siRNA and then stimulated with 2 M ATP. As in CHO cells, ATP-induced increases in Ca m 2ϩ were reduced by 41% relative to controls in HEK293 cells pretreated with InsP 3 R-III siRNA but were not reduced in cells pretreated with siRNA for InsP 3 R-I or InsP 3 R-II (p Ͻ 0.001; Fig. 6D). Thus, knockdown of InsP 3 R-III but not InsP 3 R-I or II significantly reduces mitochondrial Ca 2ϩ signals, and the findings suggest that this is because InsP 3 R-III is the isoform in closest proximity to mitochondria in these cell types.
Differential Effects of InsP 3 R Isoforms on Apoptosis-The distinct effect of each InsP 3 R isoform on Ca m 2ϩ signaling suggests that each isoform may have distinct effects on Ca m 2ϩ -mediated events as well. To investigate this, we examined the effects of each InsP 3 R isoform on apoptosis. Apoptosis was induced by activation of either the extrinsic or intrinsic apoptotic pathway, since the two pathways converge at the level of Ca m 2ϩ signaling. Previous studies suggest that either InsP 3 R-I or InsP 3 R-III can play a role in apoptosis (15)(16)(17), but little is known about either the role of InsP 3 R-II or the relative effects of each InsP 3 R isoform on apoptosis. Apoptosis was triggered here by two separate effector mechanisms, one that involves the extrinsic or death receptor pathway and the other that includes the intrinsic or mitochondria-mediated pathway. The extrinsic pathway can be activated by uptake of the prototypical toxic bile acid GCDCA, which induces apoptosis by activation of the Fas receptor independent of the presence of Fas ligand (28). GCDCA was used to induce apoptosis in CHO cells in which individual InsP 3 R isoforms were knocked down. CHO cells were first transfected with a GFP-tagged form of the NTCP (24) to allow uptake of GCDCA into the cells. Cells with visible expression of NTCP-GFP were treated for 2 h with GCDCA (50 M), and then apoptosis was assessed by examining for nuclear changes with DAPI staining (28). No chromatin condensation or nuclear fragmentation was observed in nontransfected CHO cells (Fig. 7A, top), consistent with the fact that such cells are not able to take up GCDCA. Similarly, no nuclear changes were observed in CHO cells expressing NTCP-GFP but not incubated with GCDCA (Fig.  7A, middle). In contrast, nuclear fragmentation was observed in 73.9 Ϯ 3.9% of cells expressing NTCP-GFP and treated with GCDCA (n ϭ 54; Fig. 7A, bottom). The incidence of apoptotic nuclear changes was decreased in cells lacking any one of the three InsP 3 R isoforms (Fig. 7B). Only 62.0 Ϯ 0.5% of NTCP-GFP-transfected cells lacking InsP 3 R-I were apoptotic (n ϭ 45, p Ͻ 0.05 relative to control; Fig. 7C). When InsP 3 R-II was silenced, the frequency of apoptotic cells was reduced further to 47.1 Ϯ 0.4% (n ϭ 53, p Ͻ 0.001 relative to control; Fig. 7C). Finally, knockdown of InsP 3 R-III led to the greatest reduction in GCDCA-induced apoptosis, to only 30.6 Ϯ 3.4% of cells expressing NTCP-GFP (n ϭ 83, p Ͻ 0.001 relative to control; Fig. 7C). These results suggest that all three InsP 3 R isoforms participate in induction of apoptosis induced by the extrinsic pathway in CHO cells, but InsP 3 R-III plays the greatest role, followed by InsP 3 R-II and then InsP 3 R-I.
Finally, InsP 3 R-III silencing conferred similar protection against apoptosis when activation of the intrinsic pathway was examined by staurosporine (Fig. 8). The percentage of cells that were apoptotic decreased to 22.6 Ϯ 1.5 and 22.6 Ϯ 0.9% when InsP 3 R-I (n ϭ 620; p Ͼ 0.05 relative to control group) and InsP 3 R-II (n ϭ 560; p Ͼ 0.05 relative to control group) were knocked down, respectively. However, only 15.3 Ϯ 0.3% of cells were apoptotic after silencing InsP 3 R-III (n ϭ 680; p Ͻ 0.001 relative to controls). Together, these results show that InsP 3 R-III plays a preferential role in mediating apoptosis in CHO cells, regardless of whether apoptosis is induced via the extrinsic or intrinsic pathway and although quantitative PCR suggests that InsP 3 R-III is the least heavily expressed isoform in these cells. This finding is consistent with the observations that InsP 3 R-III is the isoform that co-localizes most strongly with mitochondria and transmits Ca 2ϩ signals most effectively into mitochondria in these cells.

DISCUSSION
The versatility of Ca 2ϩ as a second messenger is in part due to the complex temporal pattern of cytosolic Ca 2ϩ signals that can occur. For example, the amplitude of Ca 2ϩ elevations can regulate differentiation in memory B cells (36), whereas the frequency of Ca 2ϩ oscillations differentially regulates expression of inflammatory cytokines in T cells (37,38). The ability to produce patterns such as Ca 2ϩ oscillations results in part from unique features of the InsP 3 R (39). However, the three isoforms of the InsP 3 R each have differences in biophysical properties (7,40), so that they may have distinct effects on Ca 2ϩ signaling as well. Indeed, Ca 2ϩ signaling patterns differ among DT40 cells engineered to express only one of each of the three InsP 3 Rs (10). In addition, selective loss of either InsP 3 R-I or InsP 3 R-III has distinct effects on Ca 2ϩ signals, as examined in A7R5 vascular smooth muscle cells (41) and in HeLa and COS-7 cells (9). In the current study, isoform-specific siRNA revealed that InsP 3 R-I and -III also have specific effects in CHO cells. In addition, the availability of specific siRNA for InsP 3 R-II made it possible to evaluate the relative role of this isoform in Ca 2ϩ signaling. Nontransfected cells responded to nucleotide stimulation predominantly with Ca i 2ϩ oscillations, but silencing either InsP 3 R-I or InsP 3 R-III led to an increase in the fraction of cells with a plateau Ca 2ϩ signaling pattern. In contrast, when InsP 3 R-II was knocked down, there was an increase in the number of cells that responded to ATP with a transient pattern of Ca i 2ϩ increase. How might this be explained? Although each of the three InsP 3 R isoforms acts as an InsP 3 -gated Ca 2ϩ channel, the isoforms are not uniformly sensitive to InsP 3 . The relative order of affinity is type II Ͼ type I Ͼ type III (6). Since PCR further suggested that InsP 3 R-II is the most abundant isoform in CHO cells, this isoform thus may be the one pref-FIGURE 7. Effects of InsP 3 R isoforms on bile acid-induced apoptosis. Wild type or NTCP-GFP-transfected CHO cells were treated with the toxic bile acid GCDCA (50 nM). The cells were stained with DAPI and examined using confocal microscopy. Apoptosis was evaluated in GFP-expressing cells by assessing nuclear changes. A, top panels show nontransfected cells with normal nuclei. The middle panels show a cell expressing NTCP-GFP that has not been exposed to GCDCA and has a nonapoptotic nucleus as well. The bottom panels show an NTCP-GFP-transfected cell treated with bile acid. Note chromatin condensation and nuclear fragmentation, indicative of apoptosis. B, cells transfected with NTCP-GFP that had each isoform of InsP 3 R knocked down. The top panels show an NTCP-GFP cell that was also transfected with siRNA for InsP 3 R-I. Nuclear fragmentation was observed in the majority of these cells. The middle panels show an NTCP-GFP cell that was co-transfected with siRNA for InsP 3 R-II. Nuclear fragmentation was observed in more than half of such cells. The bottom panels show an NTCP-GFP cell that was co-transfected with siRNA for InsP 3 R-III. No chromatin condensation was observed in the majority of these cells. C, summary of the effect of silencing each InsP 3 R isoform on bile acid-induced apoptosis. Silencing InsP 3 R-I, InsP 3 R-II, and InsP 3 R-III decreased the number of apoptotic cells to 62.0 Ϯ 0.5% (p Ͻ 0.05, n ϭ 45), 47.1 Ϯ 0.4% (p Ͻ 0.001, n ϭ 53), and 30.6 Ϯ 3.4% (p Ͻ 0.001, n ϭ 83), respectively. These results represent the mean Ϯ S.E. of five separate experiments. erentially activated in these cells. Thus, an absence of this isoform may result more often in transient Ca 2ϩ signals, because this may be a weaker form of response.
A number of proteins modulate apoptosis in part through the InsP 3 R or through related Ca 2ϩ release mechanisms. Apoptosis generally occurs through one of two pathways. The extrinsic pathway is initiated by activation of death receptors, whereas the intrinsic pathway acts on mitochondria to induce formation of the permeability transition pore (43,44). However, the death receptor pathway may act through mitochondria as well by inducing Bid-mediated mitochondrial permeabilization (45). Mitochondrial permeabilization, therefore, is important in each apoptotic pathway. We found that both apoptotic pathways were dependent on InsP 3 R expression and were most sensitive to expression of the type III isoform in particular. This observation provides further evidence for a final common mechanistic link between the two apoptotic pathways. Apoptosis is inhibited by members of the Bcl-2 family of proteins, and their common mechanism of action involves inhibition of transmission of Ca 2ϩ signals from InsP 3 R to mitochondria. Bcl-2 acts in part by decreasing the size of ER Ca 2ϩ stores (46), whereas Bcl-x L acts in part by inhibiting expression of the InsP 3 R (47). In addition, cytochrome c that has leaked from mitochondria via the permeability transition pore binds to the InsP 3 R, which facilitates release of toxic amounts of Ca 2ϩ from the ER (18). This is thought to result in a positive feedback loop that causes further Ca 2ϩ overload of mitochondria and then further leakage of cytochrome c (18). Inhibition of the interaction between cytochrome c and the InsP 3 R inhibits development of apoptosis by blocking this positive feedback loop (19). How does this relate to isoform-specific effects of InsP 3 Rs on apoptosis? Under certain circumstances, the type III InsP 3 R lacks the feedback inhibition by high Ca 2ϩ concentrations that is exhibited by the type I isoform (7). The type III isoform therefore may preferentially initiate the positive feedback cycle for Ca 2ϩ signaling that is necessary to establish permeability transition pore formation and the associated release of cytochrome c. This may explain our observation that the type III InsP 3 R preferentially mediates apoptosis in CHO cells, although it is not the predominant isoform in these cells.
Mitochondria play an integral role in Ca 2ϩ signaling pathways and patterns. A subset of mitochondria are in close proximity to InsP 3 Rs (48), and mitochondrial and cytosolic Ca 2ϩ signals are interrelated (33,49). Mitochondria play a key role in mediating most forms of apoptosis, and transmission of Ca 2ϩ signals from InsP 3 Rs to mitochondria is a critical step in this process (50). Increases in mitochondrial Ca 2ϩ can directly induce formation of the permeability transition pore (32). This in turn permits leakage of cytochrome c from mitochondria, which is associated with progression to apoptosis (51). Both the type I (52) and type III (15) InsP 3 R have been shown to induce apoptosis. Our findings corroborate this and show that the type II isoform can induce apoptosis as well. The current work furthermore shows that each isoform has a different propensity to induce apoptosis. This may be due in part to their differential distribution relative to mitochondria. Subpopulations of InsP 3 Rs and mitochondria can be clustered in the cytosol, and this is associated with local, subcellular differences in Ca 2ϩ signaling patterns (53,54). The current findings raise the possibility that the type III InsP 3 R has a particular propensity to form these subcellular clusters with mitochondria, resulting in the formation of associated signaling microdomains. InsP 3 R isoforms localize to distinct regions of the cytosol in a number of cell types (12,13,20), but little is known about the factors that regulate subcellular targeting of each isoform (9,42). Our findings raise the possibility that each InsP 3 R isoform has a distinct propensity to target to regions near mitochondria, which in turn suggests an additional and previously unsuspected level of complexity in subcellular Ca 2ϩ signaling.