InsP3R-associated cGMP Kinase Substrate Determines Inositol 1,4,5-Trisphosphate Receptor Susceptibility to Phosphoregulation by Cyclic Nucleotide-dependent Kinases*

Ca2+ release through inositol 1,4,5-trisphosphate receptors (InsP3R) can be modulated by numerous factors, including input from other signal transduction cascades. These events shape the spatio-temporal characteristics of the Ca2+ signal and provide fidelity essential for the appropriate activation of effectors. In this study, we investigate the regulation of Ca2+ release via InsP3R following activation of cyclic nucleotide-dependent kinases in the presence and absence of expression of a binding partner InsP3R-associated cGMP kinase substrate (IRAG). cGMP-dependent kinase (PKG) phosphorylation of only the S2+ InsP3R-1 subtype resulted in enhanced Ca2+ release in the absence of IRAG expression. In contrast, IRAG bound to each InsP3R subtype, and phosphorylation of IRAG by PKG attenuated Ca2+ release through all InsP3R subtypes. Surprisingly, simply the expression of IRAG attenuated phosphorylation and inhibited the enhanced Ca2+ release through InsP3R-1 following cAMP-dependent protein kinase (PKA) activation. In contrast, IRAG expression did not influence the PKA-enhanced activity of the InsP3R-2. Phosphorylation of IRAG resulted in reduced Ca2+ release through all InsP3R subtypes during concurrent activation of PKA and PKG, indicating that IRAG modulation is dominant under these conditions. These studies yield mechanistic insight into how cells with various complements of proteins integrate and prioritize signals from ubiquitous signaling pathways.

Cells express an array of cell surface receptors that couple neurotransmitters, hormones, and growth factors to cellular responses. In vivo, cells are seldom exposed to single modulating agents, and thus initiation of multiple signal transduction pathways concurrently is the norm. As a result of interaction between individual signal transduction cascades, one pathway can markedly influence the activity of another; the overall cellular response will therefore be determined by the integra-tion and prioritization of these multiple inputs. Cell surface receptors coupled to a release of intracellular Ca 2ϩ are expressed in all mammalian cells, and this pathway is a particularly rich source of potential interaction between distinct signal transduction systems (1,2). The Ca 2ϩ rise can initiate further signaling cascades, for instance by influencing the generation and metabolism of other second messengers, including cAMP and cGMP (3,4). Importantly, the precise kinetic and spatial properties of the Ca 2ϩ signal are pivotal to the appropriate stimulation of effectors, and thus regulatory input modulating the activity of the Ca 2ϩ handling machinery itself is central to the spatio-temporal "shaping" of the Ca 2ϩ signal (5).
A primary locus for modifying the characteristics of the intracellular Ca 2ϩ signal is through regulating the activity of the InsP 3 R 2 family of Ca 2ϩ release channels. InsP 3 R are encoded by three genes, leading to the expression of three distinct proteins (InsP 3 R-1, InsP 3 R-2, and InsP 3 R-3) (6 -8). Additional diversity at the protein level is generated from numerous splice variants of the InsP 3 R-1 and InsP 3 R-2 and the formation of heterotetrameric channel proteins (9). Ca 2ϩ release is allosterically regulated by a diverse array of modulatory events allowing input from other intracellular factors or events. These include the levels of intracellular Ca 2ϩ , ATP levels, phosphorylation events, and binding of protein partners (2, 9 -11). Outside of the conserved NH 2 -terminal InsP 3 binding pocket and the COOH-terminal channel domain, the primary sequence of the individual proteins is quite divergent, allowing for potential InsP 3 R subtype-specific regulation of Ca 2ϩ release. This regulation, along with the particular complement of InsP 3 R expressed, is thought to make a significant contribution to defining the particular Ca 2ϩ signals observed in individual cell types.
A relatively well studied mode of regulation of InsP 3 R activity occurs following the phosphorylation of the receptor by cAMP-dependent protein kinase (PKA) (11), a primary interaction or point of "cross-talk" between cascades that increase Ca 2ϩ or cAMP. PKA has been shown to phosphorylate defined serine residues on each isoform of InsP 3 R (12)(13)(14) and clearly enhances the single channel activity of at least the InsP 3 R-2 (12) and InsP 3 R-1 (15,16). In contrast, the Ca 2ϩ release activity of InsP 3 R-3 is apparently unaffected by PKA phosphorylation (17). The increased InsP 3 R activity has been proposed to be physiologically important for processes as diverse as neuronal plasticity and fluid secretion from salivary epithelia (15,18).
cGMP-dependent kinases (PKG) phosphorylate similar consensus sequences on substrates as PKA (RRX(S/T) or RXX(S/T), where R is basic) and thus would be expected to have functional effects similar to those of PKA. PKG can be activated following ligand binding of receptors with intrinsic guanylate cyclase activity or as a consequence of the action of nitric oxide on soluble guanylate cyclases (19). An elevation in cAMP can also lead to PKG activation either directly or indirectly by increasing cGMP levels through substrate competition at the level of shared phosphodiesterases. When the functional effects on InsP 3 R-1 were studied directly, PKG phosphorylation of identical sites regulated by PKA resulted, as expected, in a marked enhancement of the neuronal S2ϩ InsP 3 R-1 activity (20). The alternatively spliced S2Ϫ InsP 3 R-1 variant, the major InsP 3 R-1 splice variant expressed outside the neural system, resulting in 40 amino acids excised between phosphorylation sites, was not subject to regulation by PKG (20). These data indicate that all PKA consensus sequences are not necessarily PKG substrates. No data are available regarding direct effects of PKG on InsP 3 R-2 and InsP 3 R-3.
Elevations in cGMP have, however, been predominantly linked to an attenuation of Ca 2ϩ signaling (21)(22)(23)(24); this may reflect either PKG phosphorylation of InsP 3 R-2/-3 or PKGdependent phosphorylation of other substrates. For example, PKG can also influence the activity of InsP 3 R by phosphorylation of a binding partner termed InsP 3 R-associated PKG substrate (IRAG) (25,26). This protein constitutively binds to both InsP 3 R-1 and PKG1␤, and the tight association between proteins allows for efficient targeted phosphoregulation. In smooth muscle and platelets, IRAG is phosphorylated on serine 696 and leads to decreased InsP 3 -induced Ca 2ϩ release (27). Although it is not known if IRAG interacts with all subtypes of InsP 3 R or if its expression is ubiquitous, this mechanism may reconcile earlier observations of decreased Ca 2ϩ release following PKG and PKA activation in various tissues (22,24,28).
The goal of the present study was to further investigate the regulation of Ca 2ϩ release by PKA and PKG. By using an expression system that is functionally null for both InsP 3 R and IRAG, we define the particular InsP 3 R that are subject to direct regulation by PKG and indirect regulation by interaction through IRAG. Furthermore, because PKG and PKA are commonly activated concurrently, we define the conditions and the molecular mechanism by which a particular mode of regulation is specified and is dominant at the cellular level.
Generation and Transfection of Expression Constructs-A vector containing the full-length cDNA for IRAG was obtained from the RIKEN cDNA collection. The open reading frame was cloned by PCR into pCI-Neo-EGFP. To create the deletion mutant (⌬ residues 460 -506) of IRAG (IRAG⌬E12), we used the QuikChange Lightning kit (Stratagene). Correct incorporation of mutations was confirmed by DNA sequencing. COS-7 cells were cultured in Dulbecco's modified Eagle's medium with 10% fetal bovine serum, penicillin, and streptomycin at 37°C. Transient transfections were performed using Lipofectamine 2000 (Invitrogen) according to the manufacturer's directions. Cells were harvested 1 day after transfection. cDNA was transfected into DT40 cells stably expressing rat InsP 3 R-1, mouse InsP 3 R-2, and rat InsP 3 R-3 cells (for details of generation of stable cell lines, see Ref. 29) by an electroporation-based protocol using an Amaxa Nucleofector system using 5 g of each cDNA (Amaxa, Cologne, Germany) following the manufacturer's instructions (kit T, program B23).
Immunoprecipitation and Immunoblotting-Transfected COS-7 cells were incubated in lysis buffer (50 mM Tris, 150 mM NaCl, 10 mM EDTA, 1% Triton X-100, pH 7.4, adjusted with NaOH), and the preparation was centrifuged at 10,000 ϫ g for 5 min at 4°C. The resulting supernatant was incubated with 30 l of protein G-agarose (Santa Cruz Biotechnology, Inc.) for 1 h, at 4°C, to control nonspecific binding to the protein G-agarose. The clarified supernatant was incubated with 2 g of ␣-GFP monoclonal antibody (Roche Applied Science) or 2.5 g of ␣-FLAG monoclonal antibody (Sigma) for 1 h. The mixture was supplemented with 50 l of protein G-agarose and incubated for another 1 h at 4°C. The agarose was washed three times with the lysis buffer, and then immune complex-associated proteins were resolved by 5% SDS-PAGE and electrophoretically transferred to nitrocellulose membranes. The membranes were incubated with primary antibody and then with secondary antibody. The bands were visu-alized by enhanced chemiluminescence (PerkinElmer Life Sciences).
Digital Imaging of [Ca 2ϩ ] i in DT40 Cells-Imaging was performed as described previously (32). Briefly, DT40 cells were loaded with 2 M fura-2/AM at room temperature for 10 min. Fura-2-loaded cells were allowed to adhere to a glass coverslip forming the bottom of a perfusion chamber. Cells were perfused in HEPES-buffered physiological saline containing 137 mM NaCl, 0.56 mM MgCl 2 , 4.7 mM KCl, 1 mM Na 2 HPO 4 , 10 mM HEPES, 5.5 mM glucose, and 1.26 mM CaCl 2 at pH 7.4. Imaging was performed using an inverted Olympus IX71 microscope using a ϫ40 oil immersion objective lens (UApo/ 340; numerical aperture 1.35). Fura-2-loaded cells were excited alternately with light at 340 and 380 nm by using a monochrometer-based illumination system (TILL Photonics), and the emission at 510 nm was captured by a digital frame transfer CCD camera controlled by the Vision suite of software. In experiments where GFP-tagged IRAG (IRAG(GFP)) or IRAG⌬E12 (IRAG⌬E12(GFP)) was transiently expressed, GFP fluorescence was detected by excitation at 488 nm and monitoring the emission at Ͼ500 nm and was used to select transfected cells. In other experiments, cDNA encoding HcRed was included to select transfected cells. HcRed fluorescence was detected by excitation at 560 nm and observing the emission at Ͼ600 nm.
Statistical Analysis-Data are presented as mean Ϯ S.E. Data were subjected to one-way analysis of variance. Statistical significance is indicated where p Ͻ 0.05.

IRAG Physically Interacts with All InsP 3 R Subtypes-
Schlossman and colleagues (26,27,33) have reported that IRAG physically interacts with and regulates the activity of InsP 3 R-1 both in vivo and in vitro. Because the binding determinants of IRAG in InsP 3 R-1 are not known, the likelihood of an interaction with other InsP 3 R subtypes is difficult to predict. In order to establish if IRAG can potentially modulate other InsP 3 R family members and splice variants, experiments were first performed to ascertain whether IRAG physically interacts with all subtypes of InsP 3 R and is capable of forming a tertiary complex with PKG1␤. COS-7 cells were chosen for these experiments because of the ability to achieve high levels of heterologous protein expression together with low endogenous InsP 3 R expression (12, 31, 34). Following transfection with InsP 3 R subtypes, GFP-tagged IRAG and PKG1␤ immune complexes were isolated from cell lysates by incubation with ␣-GFP antisera as detailed under "Experimental Procedures." Similar experiments were performed in cells transfected with a GFP-IRAG construct (IRAG⌬E12(GFP)), which lacks the coiled-coil domain of IRAG necessary for interaction with InsP 3 R-1 (27,33). In Fig. 1A, a representative blot is presented, which shows that immunoprecipitation with ␣-GFP antibody captures both S2ϩ and S2Ϫ InsP 3 R-1 and PKG1␤ when full-length IRAG(GFP) is co-expressed (lanes 5 and 7, respectively). In contrast, in cells expressing IRAG⌬E12(GFP), PKG1␤ but not InsP 3 R-1 was recovered from the lysates following IP (lanes 6 and 8). Similar experiments were performed to determine if InsP 3 R-2 and InsP 3 R-3 interact in an analogous fashion with IRAG and PKG1␤. The lower panels of Fig. 1 demonstrate that immunoprecipitation of full-length IRAG(GFP) robustly captures InsP 3 R-2 ( Fig. 1B) or InsP 3 R-3 ( Fig. 1C) and PKG1␤, whereas immunoprecipitation of IRAG⌬E12(GFP) does not. Immune complexes were captured by immunoprecipitation (IP) from cell lysates with ␣-GFP monoclonal antibody. In A, lanes 1-4 show the input, representing 5% of the lysate prior to immunoprecipitation. Immunoprecipitated samples were subjected to Western blot analysis (IB) for detection of S2ϩ/S2Ϫ InsP 3 R-1, PKG1␤, and IRAG(GFP) or IRAG⌬E12(GFP) (lanes 5-8).
Both splice variants of InsP 3 R1 were co-immunoprecipitated only in the presence of full-length IRAG (lanes 5 and 7). In contrast, PKG1␤ was co-immunoprecipitated with antibody independent of IRAG type. In B, immunoprecipitated samples were subjected to Western blot analysis for detection of InsP 3 R-2, PKG1␤, and IRAG(GFP) or IRAG⌬E12(GFP). Lanes 1 and 2 show input representing 5% of the sample. InsP 3 R-2 was co-immunoprecipitated with antibody only in the presence of full-length IRAG (lane 3). In contrast, PKG1␤ was co-immunoprecipitated with antibody independent of IRAG type. In C, immunoprecipitated samples were subjected to Western blot analysis for detection of InsP 3 R-3, PKG1␤, and IRAG(GFP) or IRAG⌬E12(GFP). Lanes 1 and 2 show input representing 5% of the sample. InsP 3 R-3 was coimmunoprecipitated with antibody only in the presence of wild type IRAG (lane 3). In contrast, PKG1␤ was co-immunoprecipitated with antibody independent of IRAG type. Results presented are representative of at least three independent similar experiments.

PKG1␤ Phosphorylation of IRAG Modulates InsP 3 -induced
Ca 2ϩ Release via InsP 3 R-1-Next, experiments were performed to determine if the interaction of IRAG/PKG1␤ with individual InsP 3 R subtypes modulates Ca 2ϩ release. By virtue of targeted deletion of both copies of the three chicken InsP 3 R genes, the DT40-3KO pre-B lymphocyte cell line is a unique experimental platform to monitor the function of defined populations of mammalian InsP 3 R in an unambiguously null background (35). Our previous studies have generated stable cell lines expressing individual mammalian InsP 3 R splice variants and subtypes (29,32,36). To monitor the effects of PKG activation on Ca 2ϩ release, these lines were transfected with muscarinic M3 receptor (M3R), together with IRAG constructs and PKG1␤. Ca 2ϩ release was monitored in fura-2loaded cells following stimulation of M3R with low concentrations of CCh. This paradigm has been shown to be a convenient and relatively faithful reflection of Ca 2ϩ release, given that little desensitization of the response is observed over multiple exposures to agonist, and the initial peak height is largely independent of Ca 2ϩ influx (16,31). Initially, experiments were performed with DT40-3KO cells stably expressing the S2Ϫ peripheral splice variant of InsP 3 R-1 and transiently expressing M3R, IRAG(GFP), and PKG1␤. Ca 2ϩ release was initiated by brief exposure to a low concentration of CCh, and following agonist washout, the cells were incubated for 5 min with the cell-permeable and phosphodiesterase-resistant cGMP analog PET-cGMP and subsequently restimulated with CCh. As shown in Fig. 2A (pooled data in Fig.  2C), activation of PKG resulted in a marked inhibition of the CCh-induced Ca 2ϩ signal, which was partially reversible upon removal of the cGMP analog. The inhibition of Ca 2ϩ release was dependent on both the expression of PKG1␤ and IRAG because failing to express either protein abrogated the response (Fig. 2C). The current experiments are consistent with our earlier data, which suggested that S2Ϫ InsP 3 R-1 are not a direct substrate for PKG (20,31). Furthermore, the inhibition was dependent on the interaction between IRAG and S2-InsP 3 R-1 because no attenuation of the CCh-induced Ca 2ϩ signal following PET-cGMP exposure was observed in cells expressing IRAG⌬E12(GFP) (Fig. 2B). These data indicate the importance of the InsP 3 R-1⅐IRAG⅐PKG1␤ complex and that DT40-3KO cells are functionally null for IRAG and PKG1␤ in the absence of ectopic expression.
Similar experiments were performed in cells stably expressing the S2ϩ neuronal InsP 3 R-1 splice variant. PET-cGMP treatment of cells transiently expressing M3R, IRAG(GFP), and PKG1␤ resulted in a marked inhibition of the CCh-induced Ca 2ϩ rise (Fig. 3, A and pooled data in D). Again, the attenuation of Ca 2ϩ release was dependent on the expression of PKG1␤ and binding of IRAG to the InsP 3 R-1 (Fig. 3, B and  D). In contrast, in the absence of IRAG, PET-cGMP incubation led to a striking enhancement of the CCh-induced Ca 2ϩ signal (Fig. 3, C and pooled data in E), presumably as a result of the fact that S2ϩ InsP 3 R-1 is a direct substrate for PKG and phosphorylation of Ser 1755 of S2ϩ InsP 3 R results in a marked increase in the open probability of the receptor (16). A small but statistically significant increase in Ca 2ϩ release was also observed in cells expressing IRAG⌬E12(GFP) and PKG1␤ (Fig. 3, B and D). This observation is also consistent with direct phosphorylation of S2ϩ InsP 3 R-1; however, the finding that this effect is smaller than observed in the absence of IRAG may reflect the fact that IRAG⌬E12(GFP) retains PKG1␤ binding (27) and thus may sequester a fraction of the kinase from other substrates. The data in Fig. 3A also suggest that the inhibitory effect of phosphorylation of IRAG is dominant over any positive effect of direct phosphorylation of S2ϩ InsP 3 R-1.
PKG1␤ Phosphorylation of IRAG Modulates InsP 3 -induced Ca 2ϩ Release via InsP 3 R-2 and -3-Next, experiments addressed whether binding of IRAG to InsP 3 R-2 and InsP 3 R-3 is translated into modulation of Ca 2ϩ release through these particular receptors. Cells stably expressing mouse InsP 3 R-2 were transfected with M3R, PKG1␤, and IRAG(GFP). Exposure of these cells to PET-cGMP resulted in a considerable attenuation of the CCh-induced Ca 2ϩ signal, which again was fully reversible upon removal of the cGMP analog (Fig. 4, A and pooled data in D). This effect was dependent on PKG activity because the extent of inhibition was significantly reduced by concurrent exposure to the PKG inhibitor Rp-8-Br-PET-cGMP (Fig. 4, B and D). In a similar fashion to InsP 3 R-1, the reduction of the Ca 2ϩ signal was dependent on the expression and binding of IRAG (Fig. 4, C and D). Further, in cells expressing PKG1␤ in the absence of IRAG, no effect of PKG activation was observed. These data indicate that, despite InsP 3 R-2 being a substrate for PKA, InsP 3 R-2 is unlikely to be a direct substrate for PKG. Experiments were also performed to gauge the impact of IRAG modulation of Ca 2ϩ signaling during Ca 2ϩ oscillations, considered a more physiological mode of signaling. Ca 2ϩ oscillations initiated by continued exposure to low [CCh] in cells expressing M3R, IRAG(GFP), and PKG1␤ were rapidly inhibited by activation of PKG (supplemental Fig. 1A). This inhibition was attenuated by prior incubation with PKG antagonist or expression of IRAG⌬E12(GFP) (supplemental Fig. 1, B and C). Oscillations initiated by activation of the B cell receptor, and thus the endogenous signaling pathway, were similarly inhibited by PKG activation in cells expressing full-length but not truncated IRAG (supplemental Fig. 1, D and E).
In cells stably expressing InsP 3 R-3 and transiently expressing M3R, PKG1␤, and IRAG(GFP), activation of PKG resulted in an attenuation of CCh-induced Ca 2ϩ release (Fig. 5, A and pooled data in C). In comparison with the effects observed on InsP 3 R-1-and InsP 3 R-2-induced release, the degree of inhibition was modest but statistically significant in cells expressing InsP 3 R-3 (compare Fig. 5C with Figs. 2C, 3D, and 4D). Although not investigated further, significant differences in the expression level of IRAG(GFP) and PKG1␤ were not noted in InsP 3 R-3-expressing cells, and thus a possibility exists that IRAG may not interact as strongly with InsP 3 R-3 when compared with other subtypes. Nevertheless, the attenuation of Ca 2ϩ release was similarly dependent on IRAG and PKG expression and the formation of a tertiary complex with InsP 3 R-3 (Fig. 5, B and pooled data in C). Similar to cells expressing InsP 3 R-2, no PKG effects were observed that were independent of IRAG, and thus direct PKG phosphorylation is unlikely to occur or, if it does occur, is unlikely to be functionally relevant for modulating Ca 2ϩ release through InsP 3 R-3. In summary, IRAG binding and its subsequent phosphorylation by PKG reduces Ca 2ϩ release through all InsP 3 R subtypes. In addition, direct phosphorylation by PKG does not occur or does not have functional implications for the InsP 3 R-2 or InsP 3 R-3. An enhancement of Ca 2ϩ release is, however, unmasked in the absence of IRAG in cells expressing the S2ϩ "neuronal" InsP 3 R-1.
IRAG Expression Inhibits Direct Modulation of InsP 3 R-1 by PKA-In many cell types, signaling through the cAMP and cGMP pathways coexists. Given that direct phosphorylation by PKA and PKG on InsP 3 R-1 and the indirect effects of IRAG modulation by PKG are functionally opposite, we next carried out experiments to determine under what conditions the individual pathways are dominant. Fig. 6A shows a typical experimental trace, which illustrates the effects of exposure to cBIMPS, a PKA-specific analog of cAMP. As shown previously, activation of PKA results in an ϳ2-3-fold increase in the peak CCh-induced Ca 2ϩ release in cells stably expressing S2Ϫ InsP 3 R-1 (Fig. 6, A and pooled data in E). Surprisingly, when identical experiments were performed in S2-InsP 3 R-1 cells expressing IRAG(GFP) and PKG1␤, Ca 2ϩ release was unaffected by cBIMPS treatment (Fig. 6, B and E). The binding of IRAG to S2Ϫ InsP 3 R was necessary for rendering the cells refractory to the effects of PKA activation because the anticipated marked enhancement of CCh-induced Ca 2ϩ release was observed in cells expressing IRAG⌬E12(GFP) and treated with cBIMPS (Fig. 6C). In cells initially incubated with cBIMPS, subsequent activation of PKG with PET-cGMP resulted in the expected inhibition of CCh-stimulated Ca 2ϩ release (Fig. 6D). Identical results were obtained in cells expressing the neuronal S2ϩ InsP 3 R-1 (i.e. no effect of PKA activation was observed in cells expressing full-length IRAG-(GFP), but the expected enhancement of Ca 2ϩ release was readily observed either in cells not expressing IRAG or transfected with IRAG⌬E12(GFP)) (supplemental Fig. 2). The lack of effect of PKA activation on S2Ϫ InsP 3 R-1 in cells expressing IRAG, illustrated in Fig. 6, B and D, has multiple implications. First, it suggests that InsP 3 R-1 under these conditions is not functionally altered by PKA activation; second, it indicates that, despite Ser 696 in IRAG being present in a canonical PKA consensus phosphorylation motif, it is not efficiently phosphorylated by PKA. PKA activation, therefore cannot substitute for the effects of PKG1␤ anchored to IRAG. These data are entirely consistent with experiments in vascular smooth muscle, which show that although 8-bromo-cyclic GMP reduced Ca 2ϩ release, cBIMPS neither enhanced nor inhibited Ca 2ϩ release in vascular smooth muscle cells (33).
IRAG Expression Attenuates PKA Phosphorylation of InsP 3 R-1-The lack of functional effect of PKA phosphorylation on InsP 3 R-1 in cells expressing IRAG could potentially occur because IRAG binding to InsP 3 R-1 either physically inhibits the phosphorylation of the receptor or, alternatively, hinders the coupling of the phosphorylation event to the enhanced gating of the channel. The former idea was tested by directly monitoring the phosphorylation status of InsP 3 R-1 with an antibody that recognizes phosphorylated Ser 1755 in COS-7 cells expressing IRAG. IRAG(GFP) or IRAG⌬E12(GFP) expression appeared to decrease the level of expression of InsP 3 R-1, and thus the -fold change in phos-phorylation in each treatment group was evaluated. Exposure of cells transfected with InsP 3 R-1 to a PKA-activating mixture of forskolin and isobutylmethylxanthine resulted in robust phosphorylation of S2Ϫ InsP 3 R-1 (Fig. 7A, compare lanes 1  and 2, and pooled data) and S2ϩ InsP 3 R-1 (Fig. 7B, lanes 1 and 2, and pooled data). The extent of phosphorylation was markedly reduced in cells expressing IRAG (Fig. 7, A (lanes 3  and 4) and B for S2Ϫ and S2ϩ InsP 3 R-1, respectively, and pooled data). The increase in phosphorylation was at least partially restored in cells expressing IRAG⌬E12(GFP) (Fig. 7,  6) and B for S2Ϫ and S2ϩ InsP 3 R-1, respectively, and pooled data). These data are consistent with the idea that IRAG binding to InsP 3 R-1 reduces the PKA-dependent direct phosphorylation of the receptor and subsequent potentiation of Ca 2ϩ release but does not rule out a contribution by other mechanisms.
IRAG Does Not Impact PKA Regulation of InsP 3 R-2 Unless Phosphorylated by PKG-Next, we addressed whether IRAG expression impacts PKA modulation of InsP 3 R-2 in a similar fashion to InsP 3 R-1. A typical example of the effect of PKA activation on InsP 3 R-2 is shown in Fig. 8A, in which a threshold response to M3R stimulation is significantly enhanced following incubation with cBIMPS (12). In contrast to cells expressing InsP 3 R-1, transfection with IRAG did not alter the effect of PKA activation on Ca 2ϩ release via InsP 3 R-2 (Fig. 8, B and pooled data in E). These data indicate that the binding of IRAG to InsP 3 R-2 does not itself alter receptor phosphorylation and may reflect the fact that the phosphorylation sites in the individual InsP 3 R subtypes are distinct and physically distant in the receptor's linear sequence (12). This idea was confirmed experimentally by monitoring the phosphorylation of InsP 3 R-2 in COS-7 cells transiently expressing InsP 3 R-2 and IRAG constructs. Using an antibody that recognizes phosphorylated Ser 937 in InsP 3 R-2, robust receptor phosphorylation following PKA activation was detected in cells expressing IRAG(GFP) and IRAG⌬E12(GFP) but not in cells expressing a mutant InsP 3 R-2 in which the phosphorylation site at serine 937 was mutated to alanine (S937A) (supplemental Fig. 3).
Finally, experiments were performed to determine if the indirect modulation of InsP 3 R-2 activity following PKG phosphorylation of IRAG could overcome the direct modulation of InsP 3 R-2 activity following Ser 937 phosphorylation. The potentiating action of PKA activation could be maintained during repetitive exposure to cBIMPS and subsequent challenge with CCh (Fig. 9A). In contrast, when PKA and PKG were activated concurrently, a marked inhibition of the PKA-enhanced Ca 2ϩ signal was observed (Fig. 9B), indicating that the phosphorylation of IRAG exerts a dominant effect over direct InsP 3 R-2 phosphorylation.

DISCUSSION
Regulation of InsP 3 R activity following activation of PKG and PKA has been extensively studied (for reviews, see Refs. 11 and 37). It is clear that PKA can directly phosphorylate all InsP 3 R subtypes, whereas PKG has been shown to phosphorylate the InsP 3 R-1. PKA activation has generally been reported to increase InsP 3 R activity (15, 16, 20, 31, 38 -40). In contrast, increasing PKG activity has typically been reported to decrease InsP 3 R-dependent Ca 2ϩ release (22,24,41), a prominent exception being in hepatocytes, where PKG activation results in phosphorylation of InsP 3 R-1 at a shared PKA site and raising cGMP enhances agonist-stimulated Ca 2ϩ signals (42). In the present study, we have used the DT-40 InsP 3 R null expression system to define the functional effects of PKG activation, both directly and indirectly through IRAG on each InsP 3 R subtype in unambiguous isolation. In addition, these studies also provide insight into the interplay between the PKA and PKG signaling modules and the molecular mechanisms that dictate which pathway dominates regulation of InsP 3 R activity when the pathways are activated together.
In the absence of IRAG expression, raising cGMP levels had no effect on Ca 2ϩ release in cells expressing S2Ϫ InsP 3 R-1, InsP 3 R-2, or InsP 3 R-3. Because phosphorylation of InsP 3 R-2 by PKA markedly enhances Ca 2ϩ release (12), these data would suggest that serine 937 in InsP 3 R-2 is not a substrate for PKG. In addition, it is unlikely that any additional, functionally important PKG sites are present in these receptors, leaving the neuronal form of InsP 3 R-1 as the primary splice variant with potential to be directly regulated by PKG. Of note, a significant proportion of neuronal InsP 3 R-1 are phosphorylated at Ser 1755 in an activity-dependent and region-specific manner (30), and it is possible that this results at least partially from PKG activity. Ser 1755 is the major phosphorylated residue in the brain (43), and analysis of nonphosphorylatable (Ser 3 Ala) or phosphomimetic (Ser 3 Glu) mutations at these sites in the context of S2ϩ InsP 3 R clearly indicates that only phosphorylation of Ser 1755 has functional consequences (20,31). Paradoxically, PKG has, however, been shown to be predominantly, if not exclusively, phosphorylated at Ser 1589 in S2ϩ InsP 3 R-1 (44,45). Thus, it remains to be established if direct phosphorylation of Ser 1755 in S2ϩ InsP 3 R by PKG occurs and is thus physiologically relevant.
This study demonstrates that IRAG binds to all InsP 3 R subtypes and can form the basis of a tertiary complex with PKG1␤. Phosphorylation of IRAG by PKG1␤ leads to inhibitory modulation of Ca 2ϩ release through each InsP 3 R subtype and would be predicted to influence Ca 2ϩ release in cells where the other components of the complex are expressed. Only a limited amount of information is available regarding the expression of IRAG outside of platelets and tissues/ organs with a smooth muscle component (46). In these systems, InsP 3 R-1 is the predominant subtype expressed, although in platelets, InsP 3 R-2 is present (47). Smooth muscle of various origins also expresses variable amounts of both InsP 3 R-2 and InsP 3 R-3. Notably, the complement and relative amounts of each isoform have been reported to change in proliferative states of smooth muscle (48). IRAG is expressed in various regions of the brain (46) and in osteoclasts (49), and in these cells, if IRAG and PKG1␤ FIGURE 6. PKA activation fails to enhance Ca 2؉ release through S2-InsP 3 R-1 in the presence of IRAG/PKG1␤. DT40-3KO cells stably expressing S2Ϫ InsP 3 R-1 were transfected with cDNA encoding M3R and IRAG(GFP) in B and D or IRAG⌬E12(GFP) in C, together with PKG1␤ in B and C. In A, treatment with 30 M cBIMPS resulted in a significantly enhanced CChinduced Ca 2ϩ release in the absence of IRAG and PKG1␤. In B, no effect of cBIMPS treatment was observed in cells expressing full-length IRAG. In C, cBIMPS incubation resulted in a markedly enhanced CCh-induced Ca 2ϩ release in cells expressing IRAG⌬E12(GFP). In D, cBIMPS treatment did not alter CCh-induced Ca 2ϩ release; however, subsequent incubation with the cGMP analog PET-cGMP resulted in significant attenuation of Ca 2ϩ release. E, pooled data. Data in the open bars represent the -fold change of the second response compared with the first response in the absence of treatment. The filled bars show the normalized -fold increase of the second peak over the first peak for the indicated experimental condition. Columns represent mean Ϯ S.E. (error bars). *, p Ͻ 0.001; ****, p Ͻ 0.01; NS, not statistically significant. The number of cells in each condition is indicated in parentheses. are coexpressed, PKG regulation of Ca 2ϩ release would be predicted to occur.
In smooth muscle cells, elevating either cAMP or cGMP results in muscle relaxation. The importance of PKG regulation of IRAG and its interaction with InsP 3 R in this process is clear. For example, deletion of IRAG results in defective regulation by NO of smooth muscle tone (50). In addition, highlighting the importance of "targeting," an identical phenotype was reported when exon 12, encoding the InsP 3 R interaction domain of IRAG was deleted (25,33). In contrast, the effects of elevating cAMP in smooth muscle are generally explained either by direct or indirect activation of PKG (51,52) and phosphorylation of the contractile machinery (53). The effects of cAMP are therefore inconsistent with any role of direct phosphorylation of InsP 3 R by PKA. Our studies may provide a mechanism to reconcile why IRAG regulation of InsP 3 R dominates any modulation by direct receptor phosphorylation. These data indicate that the simple expression of IRAG appears to attenuate PKA phosphorylation of InsP 3 R-1 and negate any functional effects on Ca 2ϩ release. Because attenuation of direct PKA regulation requires IRAG binding to InsP 3 R-1, it is reasonable to suggest that the interaction of the pro-teins hinders PKA access to the phosphorylation sites at Ser 1589 and Ser 1755 . Further studies defining the sites of interaction on each InsP 3 R for IRAG are necessary to confirm this idea. Our findings also show that signaling via InsP 3 R-2 is potentially more versatile because, in the presence of IRAG expression, both PKA and PKG modulation can occur and exert opposing effects on Ca 2ϩ release. Nevertheless, even in the continued presence of a PKA activator, regulation by cGMP through IRAG predominates and can overcome the enhanced activity initiated by direct receptor phosphorylation.
In summary, the present study adds to our understanding of the cross-talk between signaling pathways that interact at the level of regulation of Ca 2ϩ release. Specifically, we have defined the effects on Ca 2ϩ release of PKG and PKA activation through each InsP 3 R subtype in the presence or absence of IRAG expression. This work effectively establishes a set of basic "rules" that can be used to predict the effect of PKA and PKG activation, alone and concurrently, on InsP 3 -induced Ca 2ϩ release, in particular cells expressing various complements of InsP 3 R subtypes and IRAG/PKG1␤. Further testing of these predictions will involve defining in detail the expression profile of IRAG and FIGURE 7. IRAG expression reduces PKA-induced phosphorylation of S2؊/S2؉ InsP 3 R-1. COS-7 cells were transfected with cDNA encoding S2Ϫ InsP 3 R-1 in A or S2ϩ InsP 3 R-1 in B, together with IRAG(GFP) or IRAG⌬E12(GFP) and PKG1␤. Analysis of phospho-InsP 3 R band densities for three experiments is shown in the right-hand panels for each splice variant; although the expression level of InsP 3 R was reduced by the transfection of IRAG, the amount of InsP 3 R when analyzed in each treatment group was not significantly different, and thus the -fold change in each group is shown. Cells were treated with 10 M forskolin and 200 M isobutylmethylxanthine for 10 min at room temperature prior to cell lysis. Protein was subjected to SDS-PAGE and then Western blotting for the indicated proteins. Phosphorylation of S2Ϫ InsP 3 R-1 in A and S2ϩ InsP 3 R-1 in B was diminished in the presence of full-length IRAG (compare lanes 1 and 2 with lanes 3 and 4). In contrast, the extent of phosphorylation was largely unaltered by expression of IRAG⌬E12(GFP) (lanes 5 and 6). Error bars, S.E, p Ͻ 0.05.