Combined Phosphoinositide and Ca2+ Signals Mediating Receptor Specificity toward Neuronal Ca2+ Channels*

Phosphatidylinositol 4,5-bisphosphate (PIP2) regulates Ca2+ (ICa) and M-type K+ currents in superior cervical ganglion sympathetic neurons. In those cells, M1 muscarinic and AT1 angiotensin types do not elicit Ca2+i signals and suppress both currents via depletion of PIP2, whereas the B2 bradykinin and P2Y purinergic types elicit robust IP3-mediated [Ca2+]i rises and neither deplete PIP2 nor inhibit ICa. We have suggested that this specificity arises from differential Ca2+i signals underlying receptor-specific stimulation of PIP2 synthesis by phosphatidylinositol (PI) 4-kinase. Here, we investigate which PI 4-kinase isoform underlies this signal, whether stimulation of PI 4-phosphate 5-kinase is also required, and the origin of receptor-specific Ca2+i signals. Recordings of ICa were used as a PIP2 “biosensor.” In control, stimulation of M1, but not B2 or P2Y, receptors robustly suppressed ICa. However, when PI 4-kinase IIIβ, diacylglycerol kinase, Rho, or Rho-kinase was blocked, agonists of all three receptors robustly suppressed ICa. Overexpression of exogenous M1 receptors yielded large [Ca2+]i rises by muscarinic agonist, and transfection of wild-type IRBIT decreased Ca2+i signals, whereas dominant negative IRBIT-S68A had little effect on B2 or P2Y responses but greatly increased muscarinic responses. We conclude that overlaid on microdomain organization is IRBIT, setting a “threshold” for [IP3], assisting in fidelity of receptor specificity.

Phosphoinositides comprise a diverse array of plasma membrane signaling molecules with temporal effects ranging from 1 ms to the lifetime of the organism. Most intensively studied are the polyphosphorylated phosphatidylinositols (PIs). 2 The species phosphorylated at the 4Ј-and 5Ј-positions of the inositol ring by PI 4-kinase and PI 4-phosphate (PI(4)P) 5-kinase, called PI 4,5-bisphosphate (PIP 2 ), have been implicated as a critical player in myriad cellular activities. These include cytoskeletal remodeling, vesicular/protein trafficking among intracellular membranes, transcriptional control, and regulation of ion channels and transporters. Regarding the latter role, a large and widening array of such membrane transport proteins have been shown to be regulated by plasma membrane PIP 2 , presumably due to direct binding (1,2). The action is thought to arise from either PIP 2 depletion by stimulation of phospholipase C (PLC)-coupled receptors or from alterations in the affinity of PIP 2 for the channels caused by generation of other signaling molecules (3). Using sympathetic neurons of the superior cervical ganglion (SCG) as a model, we have focused on PIP 2 -mediated regulation of voltage-gated "M-type" (KCNQ) K ϩ and "N-type" Ca 2ϩ currents. The former is so named for its depression by muscarinic stimulation, whose mechanism has been established as being due to depletion of PIP 2 (4 -9). The latter is also PIP 2 -sensitive (10 -12) as well as sensitive to arachidonic acid (13), and stimulation of PLC-linked muscarinic receptors likewise depresses I Ca in the same neurons (10,14). However, stimulation of other PLC-linked receptors (bradykinin B 2 and purinergic P2Y) in those neurons does not deplete PIP 2 and displays a pattern of M-current depression via intracellular Ca 2ϩ (Ca 2ϩ i ) signals (15,16), acting on KCNQ channels via calmodulin (CaM) (17,18), and little action on the N-type I Ca (10,14) (although many Ca V channels are modulated by CaM, the N-type channels are probably insensitive to [Ca 2ϩ ] i rises in the range (Ͻ2 M) reachable by release from stores).
This receptor specificity parallels stark receptor specificity in the induction of IP 3 -mediated [Ca 2ϩ ] i signals. Although stimulation of PLC-coupled M 1 receptors in SCG neurons produces robust PIP 2 hydrolysis and the downstream products, IP 3 and diacylglycerol (DAG) (19), little IP 3 -mediated [Ca 2ϩ ] i rises are detected, whereas stimulation of PLC-coupled bradykinin B 2 or purinergic P2Y receptors produce reliable Ca 2ϩ i signals (14, 16, 20 -23). What accounts for the pronounced receptor specificity in [Ca 2ϩ ] i signals? One hypothesis involves subcellular clustering of certain plasma membrane PLC-linked receptors into microdomains together with endoplasmic reticulum membrane IP 3 receptors. Thus, B 2 , but not M 1 , receptors have been shown to physically interact with IP 3 receptors, and the two proteins have been shown to strongly co-localize under confocal microscopy (22). Recently, however, several regulators of IP 3 receptors have been characterized that modify the efficacy of IP 3 to open its receptor (24); among those, IRBIT (IP 3 receptor-binding protein released with IP 3 ) (25,26) has seemed a likely candidate to be involved in tuning the extent of receptor-induced [Ca 2ϩ ] i rises. In this work, we perform several tests of these mechanisms.
The receptor specificity in Ca 2ϩ i signaling parallels the receptor specificity in which receptors deplete PIP 2 , whose origin is suggested to at least partly lie with Ca 2ϩ i -mediated stimulation of PIP 2 synthesis, via NCS-1 (neuronal Ca 2ϩ sensor-1) action on PI 4-kinase (9,10,14). Indeed, dramatic hormonal stimulation of PIP 2 synthesis concurrent with PLC activation is well known in the literature (e.g. stimulation of PLC-linked muscarinic receptors accelerates PIP 2 synthesis many-fold in smooth muscle and platelets (27,28), and careful measurements coupled with cellular modeling indicate that strong stimulation of PIP 2 synthesis in neuroblastoma cells and cerebellar spines in the brain is required to account for the mass of IP 3 produced (29 -32)). However, is acceleration of PI 4-kinase sufficient to maintain PIP 2 levels in the face of PLC activation, or is concurrent acceleration of PI(4)P 5-kinase required as well? In addition, among four mammalian PI 4-kinase isoforms, only PI 4-kinase III␤ has been shown to be stimulated by calcified NCS-1 (33), and this isoform has been localized within cells to the Golgi (34,35), not to the plasma membrane (PM), where PIP 2 synthesis would seem to be relevant for regulation of ion channel activity. Thus, we here also test the involvement of phosphatidic acid and the Rho monomeric GTPase, two types of signaling molecules reported to stimulate PI(4)P 5-kinase activity, either downstream of receptors or in a receptor-independent fashion (36,37). We also test whether PI 4-kinase III␤ is a critical player in stimulation of PIP 2 synthesis by B 2 and P2Y receptors in sympathetic neurons. Our work highlights the central role of intracellular Ca 2ϩ signals in conferring receptor specificity toward ion channel targets.

EXPERIMENTAL PROCEDURES
cDNA Constructs, Antibodies, and Drugs-The plasmids for wild-type and dominant negative (S68A) IRBIT, GST-tagged IRBIT(1-104), and the anti-IRBIT antibody were kindly given to us by the laboratory of Humbert De Smedt (Laboratory of Molecular and Cellular Signaling, University of Leuven, Belgium). The cDNAs for wild-type and D656A bovine PI 4-kinase III␤ and PIK93 were kindly given to us by Tamas Balla (National Institutes of Health, Bethesda, MD).
SCG Sympathetic Neuron Culture and cDNA Transfections-Sympathetic neurons were isolated from the superior cervical ganglia of 7-14-day-old rats of both genders (Sprague-Dawley) and cultured for 2-4 days. Rats were given a lethal overdose of halothane and decapitated. Neurons were dissociated using methods of Bernheim et al. (38), plated on 4 ϫ 4-mm glass coverslips (coated with poly-L-lysine) and incubated at 37°C (5% CO 2 ). Fresh culture medium containing nerve growth factor (50 ng/ml) and pertussis toxin (100 ng/ml) were added to the cells 3 h after plating. For exogenous expression of cDNA constructs, we used the PDS-1000/He biolistic particle delivery system ("gene gun," Bio-Rad), as described previously (39). Transfection efficiency was assumed to be determined by the random distribution of fired gold particles and was up to 10% of cultured neurons.
Immunostaining-Cells grown on poly-L-lysine-coated coverslips were fixed in 4% paraformaldehyde, washed twice with 100 mM sodium phosphate (PB, pH 7.4), three times with PB ϩ 150 mM NaCl (PBS), and blocked with 5% goat serum and 0.1% saponin in PBS (PBS ϩ GS). The cells were incubated for 3 h at room temperature with primary affinity-purified rabbit anti-IRBIT (40) (diluted 1:500 in PBS ϩ GS) and mouse antityrosine hydroxylase (1:5000) antibodies. In the blocking controls, the anti-IRBIT antibody was pre-adsorbed with a 10fold molar excess of affinity-purified recombinant GSTtagged IRBIT(1-104) peptide used to raise the antibody. Cells were washed six times with PBS and then incubated with goat rhodamine red-conjugated anti-rabbit and FITC-conjugated anti-mouse secondary antibodies (1:500, Jackson Immunoresearch) in PBS ϩ GS for 1 h. Cells were then washed three times with PBS, twice with PB, and three times with water. Air-dried slides were mounted on a drop of Vectashield (Vector Laboratory) and sealed with nail polish. Stained cells were viewed with a Nikon TE2000 inverted microscope, using excitation/emission filters appropriate for rhodamine red and FITC, and images were taken with a Cascade 512F enhanced-CCD camera controlled by Metamorph software running on a PC. The purification of the GST-tagged IRBIT(1-104) peptide was performed by us as previously described by others (40), except that we did not cleave the GST moiety from the peptide.
Perforated Patch Voltage Clamp Electrophysiology-Pipettes were pulled from borosilicate glass capillaries (1B150F-4, World Precision Instruments, Sarasota, FL) using a Flaming/Brown micropipette puller P-97 (Sutter Instruments) and had resistances of 1-2 megaohms when filled with internal solution and measured in standard bath solution. Membrane current was measured with pipette and membrane capacitance cancellation, sampled at 200 s, and filtered at 1 kHz by an EPC-9 amplifier and PULSE software (HEKA/Instrutech, Port Washington, NY). In all electrophysiological experiments, the perforated patch method of recording was used with amphotericin B (600 ng/ml) in the pipette (41). Amphotericin was prepared as a stock solution as 60 mg/ml in DMSO. In these experiments, the access resistance was Ͻ10 megaohms 5-10 min after seal formation. Cells were placed in a 500-l perfusion chamber through which solution flowed at 1-2 ml/min. Inflow to the chamber was by gravity from several reservoirs, selectable by activation of solenoid valves (Warner Scientific). Bath solution exchange was essentially complete by Ͻ30 s. Experiments were performed at room temperature. To evaluate the amplitude of I Ca , cells were held at Ϫ80 mV, and 20-ms depolarizing steps to 5 mV were applied every 5 s. The amplitude of I Ca was usually defined as the inward current sensitive to Cd 2ϩ (100 M). UTP, bradykinin (BK), oxotremorine methiodide (oxo-M), and Cd 2ϩ were used at concentrations of 10 M, 250 nM, 10 M, and 100 M, respectively. The external solution used to record Ca 2ϩ currents contained 150 mM NaCl, 2.5 mM KCl, 5 mM CaCl 2 , 1 mM MgCl 2 , 10 mM glucose, 10 mM HEPES, 500 nM tetrodotoxin, pH 7.4, with NaOH. The perforated patch pipette solution for voltage clamp experiments contained 150 mM CsCl, 5 mM MgCl 2 , and 10 mM HEPES. Data are presented as mean Ϯ S.E. Statistical tests were performed using a paired t test or an unpaired t test where appropriate.
Imaging-Fluorescent microscopy was performed with an inverted Nikon Eclipse TE300 microscope with an oil immersion ϫ40/1.30 numerical aperture objective. A Polychrome IV monochromator (T.I.L.L. Photonics, Martinsreid, Germany) was used as the excitation light source, and a FURA2 71000 filter cube (Chroma) was used for fura-2 imaging. SCG neurons were bath-loaded with fura-2/AM (2 M) for 30 min at 37°C in the presence of pluronic acid (0.01%). Cells were excited alternatively at 340 and 380 nm (50 -200 ms every 2 s), the fluorescence emission was collected by an IMAGO 12-bit cooled CCD camera, and images were stored/analyzed with TILLvisION 4.0 software. These fura-2 signals were not calibrated due to inherent difficulties in calibrating esterified indicator dyes (54). Because the EGFP used in our SCG transfection experiments is very poorly excited in the UV (at 340 or 380 nm), it negligibly contaminates the fura-2 emission signals. In control experiments, the combined contamination from EGFP emission and from autofluorescence of the cell was measured to alter the 340/380 fura-2 emission ratio by Ͻ0.5%.

RESULTS
To assay the extent of PIP 2 depletion upon receptor stimulation in living cells, a convenient paradigm is to use the activity of an ion channel known to be PIP 2 -sensitive as a read-out of PIP 2 levels in the membrane. Thus, we turned to our measurements of the inhibition in SCG neurons of N-type voltage-gated Ca 2ϩ channels, which have been shown to be very sensitive to PIP 2 abundance. SCG neurons were studied under perforated patch whole-cell voltage clamp and were pretreated with the G o/i blocker, pertussis toxin, to isolate G q/11mediated actions (42). Our previous work reported suppression of I Ca by muscarinic M 1 receptors, but not bradykinin B 2 or purinergic P2Y receptors, via depletion of PIP 2 molecules in the membrane, which are necessary for Ca 2ϩ channel activity (10,11,14). Thus, the Ca 2ϩ channels act as a PIP 2 biosensor, and depression of I Ca reports depletion of PIP 2 .

Compensatory, Receptor-specific Stimulation of PIP 2 Synthesis Uses PI 4-Kinase III␤-Our previous work has demonstrated receptor-specific stimulation of PI 4-kinases via Ca 2ϩ
i signals and NCS-1 to be involved in compensatory PIP 2 synthesis concurrent with PLC hydrolysis. Thus, acute treatment of cells by the broad-spectrum PI 3-and PI 4-kinase blocker, wortmannin, or transfection into the neurons of a dominant negative (DN) NCS-1 mutant that cannot bind Ca 2ϩ bestowed upon P2Y and B 2 receptors the ability to depress I Ca in SCG neurons (10,14). However, precisely which molecule might NCS-1 be targeting? Mammalian cells express at least four distinct isoforms of PI 4-kinases with varying functions and pharmacological profiles (43). Although types II␣ and II␤ are found at the PM, neither is wortmannin-sensitive (44,45), ruling out these isoforms in this process. PI 4-kinase III␣ has been localized to the PM of vertebrate cells, where it would be well placed to underlie rapid synthesis of PI(4)P, whereas type III␤ has been localized primarily to the Golgi (35,46). Thus, the literature presents us with something of a conundrum because the type III␤ isoform implicated in regulation by Ca 2ϩ /NCS-1 (33) would appear to not be in the correct subcellular location.
To test whether III␤ is actually the relevant isoform in this system, we used two approaches. The first exploited the development of a new isoform-specific PI 4-kinase blocker, PIK93, with high potency against the type III␤ but little effect on the type III␣ isoforms (35). We compared the suppression of I Ca in control neurons with those pretreated with PIK93. In Fig. 1A are plotted the amplitudes of inward I Ca , elicited by brief depolarizations to 5 mV in a control neuron, with representative current traces during different times in the experiment shown in the insets. We tested the responses of I Ca to bath application of supramaximal concentrations of UTP, BK, and oxo-M, agonists of G q/11 -coupled purinergic P2Y, bradykinin B 2 , and muscarinic M 1 receptors, respectively. As seen in our previous work, neither UTP nor BK caused significant suppressions of I Ca , although oxo-M induced a strong suppression of the current in the same cell. However, when PI 4-kinase III␤ was blocked by pretreatment of a neuron with PIK93 (0.3 M, 7-min pretreatment), all three agonists caused significant suppressions of I Ca (Fig. 1B), suggesting that PI 4-kinase III␤ activity is required to maintain PIP 2 levels during stimulation of B 2 or P2Y receptors.
Our second approach involved transfection into the neurons of the PI 4-kinase III␤ kinase-dead (KD) D656A mutant that acts as a DN (47). We again used I Ca amplitudes as a PIP 2 biosensor and compared the responses to UTP, BK, and oxo-M in neurons transfected with WT PI 4-kinase III␤ or the D656A mutant. In a SCG neuron transfected with the WT kinase, the responses to the three agonists were similar to those seen in control neurons, with only oxo-M, but not BK or UTP, inducing a robust depression of I Ca (Fig. 1C). However, in the cell transfected with the KD mutant, both UTP and BK now induced a significant depression of I Ca (Fig. 1D). These data are summarized in Fig. 1E. For control neurons, the suppressions of I Ca by UTP, BK, and oxo-M were 3.3 Ϯ 0.5, 5.5 Ϯ 0.8, and 57 Ϯ 3% (n ϭ 12), respectively. For PIK93treated neurons, UTP, BK, and oxo-M suppressed I Ca by 23 Ϯ 6% (p Ͻ 0.01), 37 Ϯ 7% (p Ͻ 0.01), and 59 Ϯ 6% (n ϭ 7), respectively. For neurons transfected with WT PI 4-kinase III␤, the suppressions of I Ca by UTP, BK, and oxo-M were 5.0 Ϯ 1.0, 3.4 Ϯ 1.3, and 60 Ϯ 5% (n ϭ 5), respectively. For neurons transfected with the KD D656A mutant, the suppressions of I Ca were 31 Ϯ 3 (p Ͻ 0.001), 38 Ϯ 4 (p Ͻ 0.001), and 56 Ϯ 5% (n ϭ 7), respectively. Taken together, these data suggest that heightened PI 4-kinase III␤ isoform activity is required to prevent PIP 2 levels from falling upon stimulation of P2Y and B 2 receptors in SCG neurons. We also quantified tonic I Ca amplitudes in these two groups of cells to see what effect expression of WT or KD PI 4-kinase III␤ might have on tonic PIP 2 levels. Indeed, those values were 60 Ϯ 5 and 37 Ϯ 6 pA/picofarad (p Ͻ 0.05), respectively (Fig. 2G), indicating that not only is PI 4-kinase III␤ necessary for receptor stimulation of PIP 2 synthesis, but it is also needed for homeostatic mainte-nance of PIP 2 abundance (although overexpression of the WT kinase did not augment tonic I Ca amplitudes).
RhoA, Rho-kinase, and DAG-kinase Are Required for PIP 2 Synthesis-Given our hypothesis of receptor stimulation of PI(4)P 5-kinase, we considered two putative signaling mechanisms found in the literature. A number of laboratories have shown PI(4)P 5-kinases to be regulated by Rho family GTPases, either directly or via its effector, Rho-kinase (48 -51). Thus, we performed two tests of the involvement of these signaling molecules in PIP 2 synthesis. To test the role of RhoA action in this process, we evaluated the effect of transfection into the neurons of the RhoA T19N mutant that acts as a DN (50). SCG neurons were transfected with EGFP either alone or together with DN RhoA, using the biolistic method. Again, we used I Ca to assay the extent of PIP 2 depletion by receptor agonists in pertussis toxin-treated SCG neurons. In the cell transfected with only EGFP, neither UTP nor BK induced a significant depression of I Ca , but oxo-M caused a robust effect ( Fig. 2A). However, in the neuron transfected with DN RhoA (Fig. 2B), all three agonists caused a significant depression of I Ca , although to a somewhat lesser degree than in control cells. Such data are summarized in Fig. 2F. In cells transfected with only EGFP, the suppressions of I Ca by UTP, BK, and oxo-M were 3.4 Ϯ 0.9, 4.2 Ϯ 1.1, and 55 Ϯ 4% (n ϭ 8), respectively. However, in the cells transfected with RhoA T19N, they were 16 Ϯ 3% (p Ͻ 0.01), 25 Ϯ 3% (p Ͻ 0.01), and 52 Ϯ 3% (n ϭ 6), respectively. In the experiment shown in Fig. 2B, the current amplitudes appear smaller than usual; indeed, that was the case in neurons transfected with DN RhoA. In cells transfected with EGFP either alone or together with RhoA T19N, the current densities were 50 Ϯ 4 pA/picofarad (n ϭ 8) and 25 Ϯ 4 pA/picofarad (n ϭ 6, p Ͻ 0.051), respectively (Fig. 2G). Thus, blockade of endogenous RhoA activity prevents compensatory PIP 2 synthesis during B 2 or P2Y receptor stimulation. We also conclude that RhoA activity is required to maintain normal PIP 2 levels because I Ca amplitudes were tonically reduced in DN RhoA-expressing cells.
The actions of RhoA are often through its effector, Rhokinase. As a probe for Rho-kinase involvement, we tested the effect of a Rho-kinase inhibitor, Y27632 (52). As before, neither BK nor UTP induced a significant suppression of I Ca in the vehicle-treated cell (Fig. 2C), whereas oxo-M had a robust effect. However, in a cell treated with Y27632 (1 M, 40-min pretreatment) (Fig. 2D), I Ca was significantly depressed by UTP, BK, and oxo-M, an effect similar to that seen by blockade of PI 4-kinase III␤. These data are summarized in Fig. 2F. In control cells, the suppressions of I Ca by UTP, BK, and oxo-M were 3.3 Ϯ 1, 5.7 Ϯ 0.8, and 56 Ϯ 3% (n ϭ 7), respectively. In cells pretreated with the Rho-kinase inhibitor, however, they were 28 Ϯ 4% (p Ͻ 0.01), 43 Ϯ 4% (p Ͻ 0.001), and 59 Ϯ 4% (n ϭ 10), respectively. Application of Y27632 itself did not lower PIP 2 levels during the 40-min pretreatment period, as indicated by the lack of reduced I Ca amplitudes before the addition of any agonist. In control cells or those treated with the drug, the current densities were 55 Ϯ 6 (n ϭ 7) and 58 Ϯ 7 (n ϭ 10), respectively (Fig. 2G). Thus, Rho-kinase activity is necessary for receptor-specific stimulation of PIP 2 synthesis, but its blockade does not result in PIP 2 depletion during the lifetime of a patch clamp experiment.
Finally, we investigated the part played in this system by phosphatidic acid (PA), which has been shown to stimulate the type I PI(4)P 5-kinase isoform (37). PA is produced downstream of DAG production by DAG-kinase, and thus PA levels would be expected to increase with activation of PLC. To probe this notion, we tested the effect of the drug DAG-kinase inhibitor II (R59949) on the suppression of I Ca by UTP, BK, and oxo-M. Indeed, when DAG-kinase was blocked by pretreatment of a neuron with DAG-kinase inhibitor II (20 M, 1-h pretreatment) (53), all three agonists caused significant suppressions of I Ca (Fig. 2E). For neurons treated with DAG-kinase inhibitor II, the suppressions by UTP, BK, and oxo-M were 25 Ϯ 6 (p Ͻ 0.01), 33 Ϯ 7 (p Ͻ 0.01), and 62 Ϯ 5% (n ϭ 8), respectively (Fig. 2F). Thus, the maintenance of PIP 2 levels in the face of PLC activity requires functional DAG-kinase activity, presumably in order for PA to participate in stimulation of PI(4)P 5-kinase. We also asked whether simultaneous blockade of PI 4-kinase III␤ and DAG-kinase would result in greater depression of I Ca by UTP or BK than blockade of either alone. However, we found neurons treated with both PIK93 and DAG-kinase inhibitor II to display responses similar to those treated with only one inhibitor. For those cells, the suppressions of I Ca were 30 Ϯ 5, 44 Ϯ 6, and 60 Ϯ 7%, respectively (Fig. 2F). To conclude this section, our working hypothesis is that stimulation of PIP 2 synthesis from PI(4)P via PI(4)P 5-kinase is controlled synergistically by RhoA, Rho-kinase, and phosphatidic acid.

Overexpression of M 1 Receptors in SCG Neurons Confers Muscarinic Agonist-induced Ca 2ϩ
i Signals-The data presented so far suggest that stimulation of G q/11 -coupled receptors, which induce Ca 2ϩ i signals, does not deplete PIP 2 due to concurrent stimulation of PIP 2 synthesis. However, for the case of M 1 receptors, which do not provoke [Ca 2ϩ ] i rises, a wealth of data indicate that PIP 2 is strongly depleted. But what mechanism underlies this receptor specificity in generation of Ca 2ϩ i signals in the first place? The "microdomain" hypothesis suggests co-localization or physical association between plasma membrane G protein-coupled receptors and endoplasmic reticulum membrane IP 3 receptors to be the key determinant (22). We thought a test of this hypothesis would be to overexpress exogenous M 1 receptors in SCG neurons, reasoning that such overexpression should override any na-tive subcellular localization of endogenous receptors. Neurons were transfected either with EGFP only or with EGFP plus M 1 receptors, and Ca 2ϩ i signals induced by P2Y, BK, and M 1 receptor stimulation were assayed using fura-2 imaging, with the dye loaded into the cells as the AM-ester. We did not attempt to calibrate these measurements, due to its inherent difficulty in such AM-ester-loaded experiments (54). Fig. 3A shows the results from a control neuron transfected with EGFP only, with the ratio of emission from 340-and 380-nm light, which reports Ca 2ϩ , plotted during the experi- ment. Shown in the inset are pseudocolor 340/380 images of the neuron at various time points. Both UTP and BK induced robust rises in [Ca 2ϩ ] i , whereas the [Ca 2ϩ ] i rises from muscarinic stimulation were insignificant, consistent with the literature (14,16,18,20,55). However, in the cell overexpressing M 1 receptors, muscarinic stimulation now induced a very significant rise in [Ca 2ϩ ] i , even larger than that produced by UTP or BK (Fig. 3B). These data are summarized in Fig. 3C. For neurons transfected with only EGFP, the increases in the 340/380 nm ratio induced by UTP, BK, and oxo-M were 0.12 Ϯ 0.01, 0.25 Ϯ 0.02, and 0.02 Ϯ 0.01 (n ϭ 20), respectively, whereas for those transfected with EGFP plus M 1 receptors, they were 0.07 Ϯ 0.01 (p Ͻ 0.05), 0.15 Ϯ 0.03 (p Ͻ 0.01), and 0.17 Ϯ 0.02 (p Ͻ 0.001) (n ϭ 24), respectively. Thus, overexpression of exogenous M 1 receptors into the neurons results in muscarinic agonists now inducing large Ca 2ϩ i signals, which we presume to be due to more widespread expression of M 1 receptors, including in microdomains containing IP 3 receptors, although we cannot exclude the possibility that the difference lies partly in enhanced production of IP 3 . We do argue, however, that stimulation of endogenous muscarinic receptors in SCG neurons does result already in very large IP 3 production, as evidenced from multiple biochemical (9,19) and PIP 2 hydrolysis probe imaging (8 -10) experiments. We note that the [Ca 2ϩ ] i rises by UTP and BK were slightly, but significantly, reduced by M 1 receptor overexpression. This result might be due to displacement of some endogenous P2Y and B 2 receptors from their preferred position in association with IP 3 receptors by the heterologously expressed M 1 receptors, but we did not investigate this further.
The Protein IRBIT Underlies a "Threshold" Mechanism for IP 3 -induced Ca 2ϩ Release-One aspect on this topic that has puzzled us is the seeming "all-or-none" nature of the Ca 2ϩ i response to receptor agonists (i.e. the Ca 2ϩ i signals emanating from M 1 (and angiotensin AT 1 (56)) receptor stimulation are not just less than from P2Y and B 2 receptors but nearly wholly absent). Thus, we wondered if another signaling molecule were involved, one that might inhibit opening of IP 3 receptors in either a receptor-dependent or receptor-independent way. IRBIT is one such candidate inhibitory molecule because it has been shown to be a competitive antagonist of IP 3 for its binding site on IP 3 receptors (25). We first asked if IRBIT is expressed in SCG neurons by immunostaining and fluorescence microscopy. Cultured neurons were fixed and immunolabeled with anti-IRBIT antibodies (kindly given to us by Herbert De Smedt (Leuven, Belgium)) and an antibody against tyrosine hydroxylase as a sympathetic neuron marker. Fig. 4 shows images of the neurons showing clear expression of IRBIT. As controls, there was no significant labeling when the anti-IRBIT antibody was preadsorbed with a 10-fold molar excess of the immunizing peptide used to raise the antibody or when the primary antibody was omitted. The interaction between IRBIT and the IP 3 receptor is regulated by phosphorylation of IRBIT at Ser 68 , such that the S68A mutant binds only weakly to the IP 3 receptor, does not compete with IP 3 , and inhibits binding of endogenous IRBIT by formation of an inactive heteromer (25). Thus, we utilized the IRBIT-S68A mutant as a DN test.
To examine the role of IRBIT in regulating Ca 2ϩ i signals in SCG neurons, we compared the [Ca 2ϩ ] i responses to stimulation of P2Y, B 2 , and M 1 receptors in neurons transfected with EGFP only, with WT IRBIT, or with IRBIT-S68A. In a cell transfected with WT IRBIT (Fig. 5A), the responses to UTP or BK were present but modest, and there was no response to oxo-M. In a cell transfected with IRBIT-S68A (Fig. 5B), the responses to UTP and BK were large, but now there was a modest but significant response to oxo-M. We quantified the data in two ways: by calculating the fraction of cells that responded to each agonist in all three groups of cells (Fig. 5C) or by quantification of the mean rise in the 340/380 nm ratio (Fig. 5D). For the case of UTP, for cells transfected with only EGFP, with WT IRBIT or with IRBIT-S68A, the fractions of cells with significant rises in [Ca 2ϩ ] i were 86, 67, and 93%, respectively. For BK, the fractions were 86, 58, and 93%, respectively, and for oxo-M, they were 15, 9, and 64%, respectively. Thus, for UTP and BK, overexpression of WT IRBIT reduced the number of responding cells, and for oxo-M, expression of IRBIT-S68A greatly increased it. We then analyzed the changes in the 340/380 nm ratio for the cells that did respond, to see if the [Ca 2ϩ ] i rises for such cells were affected by WT or S68A IRBIT. For UTP, there were no effects on the amplitude of the responses that reached the p ϭ 0.05 level of significance, which were 0.12 Ϯ 0.01, 0.08 Ϯ 0.02, and 0.16 Ϯ 0.02 (n ϭ 17, 17, and 15) for cells transfected with EGFP only, WT IRBIT, and IRBIT-S68A, respectively. For BK, there was a reduction in the 340/380 nm ratio in cells trans- Thus, IRBIT plays a role in tuning IP 3 -mediating Ca 2ϩ i signaling but probably not in a receptor-specific manner. Notably, the actions of IRBIT were seen here at saturating concentrations of agonist, although they were most manifest at subsaturating agonist concentrations in previous work in HeLa cells (24). The difference is likely to be the paucity of spatiotemporal mechanisms of specificity in tissue culture cells as opposed to primary neurons. Indeed, when one tests cloned G q/11 -coupled receptors expressed in tissue cells, they all respond with similar [Ca 2ϩ ] i rises or similar depletion of PIP 2 , etc. In SCG neurons, we think [IP 3 ] at the IP 3 receptor, even after B 2 or P2Y receptor stimulation, to be comparably modest, compared with that seen in tissue culture cells.

DISCUSSION
Our work in sympathetic neurons has highlighted the mechanistic underpinnings of receptor-specific actions toward voltage-gated Ca 2ϩ and K ϩ channels. In SCG neurons, both N-type Ca 2ϩ (10) and Kv7 (M-type) K ϩ (2, 57) channels are highly sensitive to PIP 2 abundance, and PIP 2 binding/un-binding almost certainly underlies their PIP 2 sensitivity. Another factor in the literature has been complex regulation of N-type channels by arachidonic acid (58 -60), but we do not delve into that subject here. SCG neurons express two groups of PLC-linked receptors, which we have called "mode 1" and "mode 2" (14), the former consisting of the M 1 and AT 1 types that elicit sparse Ca 2ϩ i signals, deplete PIP 2 , and suppress both M current and I Ca and the latter consisting of the B 2 and P2Y types that induce reliable Ca 2ϩ i signals, do not deplete PIP 2 or suppress I Ca , and depress M current via Ca 2ϩ /calmodulin binding (61). Thus, the I Ca in these cells makes an ideal biosensor of PIP 2 depletion by receptors, similar to the reporting by highly PIP 2 -sensitive GIRK K ϩ channels when expressed in the same neurons (9). It is important to note that such dramatic receptor specificity is only seen in native cells (23,(62)(63)(64)(65)(66)(67) and is wholly absent in reconstituted systems in which channels and receptors are heterologously expressed, in which all PLC-linked receptors elicit large Ca 2ϩ i signals and all probably work similarly (8,11,68,69).
We have suggested at least part of such receptor-specific actions to be mediated by Ca 2ϩ /NCS-1-mediated acceleration of PI 4-kinase activity. However, the apparent mismatch between the proposed PI 4-kinase isoform (III␤) and subcellular location (Golgi) provoked us to test if the III␤ isoform was actually required. Our positive result suggests that 1) either PI 4-kinase III␤ is localized to the plasma membrane in neurons, versus its localization in cell lines (35,46), or 2) PI(4)P is produced in the Golgi but is then rapidly (Ͻ1 min) transported to the membrane, where it is converted to PIP 2 by PI(4)P 5-kinase. Indeed, the latter possibility is interesting and is worth testing, given the established ability of phosphoinositide transfer proteins (70,71) or other proteins (see Ref. 72 for a review) to rapidly shuttle phosphoinositides between subcellular compartments. As recently discussed (73), it could then be the phosphoinositide transfer protein transport rate that is stimulated by the "mode 2" receptors rather than the PI 4-kinase rate itself. The recent work using voltage-sensitive PIP 2 phosphatases makes clear that PI(4)P replenishment at the PM is rate-limiting for PIP 2 synthesis (73)(74)(75). However, along with stimulation of plasma membrane PI(4)P levels by receptor stimulation, several lines of evidence indicate that acceleration of PI(4)P 5-kinase activity must also be required to maintain [PIP 2 ] upon activation of PLC. Using the Virtual Cell environment, the Loew laboratory has demonstrated the requirement for ϳ10-fold stimulation of both PI 4-and PI(4)P 5-kinases in order to fit their biochemical measurements of PI(4)P and PIP 2 levels during bradykinin stimulation of neuroblastoma cells (30) and stimulation of metabotropic glutamate receptors in cerebellar spines (32). Using their model, we find that PI(4)P must be allowed to rise to levels Ͼ30-fold over that present tonically in order to produce enough PIP 2 by mass action, without increasing the rate of PI(4)P 5-kinase activity, 3 and biochemical measurements indicate that such a massive increase in PI(4)P abundance does not occur (30,69,76,77). Likewise, Falkenburger et al. (73) examined the kinetics and dose-response relationship of muscarinic modulation of cloned M-type channels by cloned M 1 receptors heterologously expressed in tsA-201 cells, and using the Virtual Cell, showed 10-fold stimulation of both PI 4-kinase and PI(4)P 5-kinase by agonist to be required to adequately fit their data. Strong acceleration of PIP 2 synthesis by agonists is documented in platelets, in which thrombin stimulation of G q/11coupled PAR1 receptors increased PI(4)P 5-kinase activity nearly 10-fold and DAG-kinase activity 4.5-fold (78). In heterologous systems, stimulation of these same PAR1 receptors resulted in a translocation of type I PI(4)P 5-kinases from the Golgi to the PM that was Rho-dependent, representing a possible mechanism (48), although it is unclear if such translocation could occur on the time scale of a patch clamp experiment. Finally, overexpression of PI(4)P 5-kinase strongly increases the single-channel open probabilities and tonic current amplitudes of PIP 2 -sensitive Kv7 (M-type) channels (5,7,9), indicating the sensitivity of PIP 2 levels to PI(4) 5-kinase activity. Our data here conform to the thinking that stimulation of both PI 4-kinase and PI(4)P 5-kinase is required to explain our data and that of others.
In a variety of cells, Rho family GTPases have been shown to physically associate with type I PI(4)P 5-kinases (79 -81) and, together with Rho-kinase and PA, to stimulate production of PIP 2 by PI(4)P 5-kinase (for reviews, see Refs. 82 and 83). At least some of this activity seems to be constitutive, and the reduction in I Ca that we observed in DN RhoA-transfected neurons is consistent with that idea. But what could underlie receptor-mediated stimulation of PI(4)P 5-kinase activity? Clearly, PA action provides one such mechanism because PA is produced downstream of DAG production, whose levels are expected to be elevated by PLC hydrolysis of PIP 2 (see Refs. 84 and 85 for reviews). In addition, PA, via CDP-DAG, is the precursor to PI production, and blockade of PA production might eventually reduce tonic PI levels. We must also consider the subcellular localization of PI(4)P 5-kinase molecules. Indeed, as for PI 4-kinases, the subcellular localization of mammalian PI(4)P 5-kinases is controversial. Whereas several laboratories report the latter to localize to the Golgi of resting cells (48,86,87), the general literature supports the subcellular localization of PI(4)P 5-kinases to be regulated, with the regulatory molecules including Rho (via Rho-kinase), Rac and Arf GTPases, and PA. Slow (Ͻ1 h) regulation of ion channels by small GTPases, comprising trafficking steps and tuning of functional channels, has been documented (50). However, as for the discussion on the site of PI(4)P synthesis, receptor-induced translocation of PI(4)P 5-kinases to the PM as the mechanism underlying G q/11 -coupled receptor stimulation of PIP 2 synthesis suffers from a time constraint. Not only would this event have to transpire during a patch clamp measurement of current amplitudes (as we perform here), but measurements of the time course of PI(4)P phosphorylation following PIP 2 dephosphorylation by a voltage-sensitive PIP 2 phosphatase indicate a time scale of Ͻ30 s (73). Thus, careful measurements of how fast PI(4)P 5-kinases can translocate to the membrane and phosphorylate PI(4)P are required to determine if this is a plausible mechanism.
The lack of Ca 2ϩ i signals elicited by stimulation of M 1 receptors in SCG neurons has been a persistent enigma, given that such stimulation causes intense PIP 2 hydrolysis, as measured by biochemical (19) and optical (88) methods. Although the microdomain hypothesis seemed a logical explanation, observations from these neurons transfected with the PLC␦-PH probe that binds to PIP 2 and to IP 3 (89), in which the probe appears to flood the entire cytoplasm upon muscarinic agonist (8 -10), made us intuitively wonder if subcellular localization is the entire story. There seems nothing special about M 1 receptors in SCG neurons that prevents their stimulation from causing [Ca 2ϩ ] i rises because when we heterologously overexpressed them here, oxo-M produced large Ca 2ϩ i signals. We acknowledge that overexpression of M 1 receptors might result in much greater numbers of receptors expressed in the membrane of the neurons, resulting in greater or faster turn-on of G␣ q/11 and PLC for a given concentration of agonist. Quantification of the increase over endogenous levels of membrane proteins by their expression suggests that this factor might be as large as 100-fold, although most expressed proteins would probably not be in the plasma membrane (90). However, that same paper concludes that the steps of agonist binding to receptors and receptor activation of G q/11 and PLC are much faster than PLC-mediated hydrolysis of PIP 2 , and thus it is the number of PLC molecules that largely determines the speed and amount of IP 3 production, at least at the supramaximal agonist concentrations used here. Thus, we take our results as a validation of microdomain organization being a major part of the story.
The other part lies in IRBIT as an inhibitory "gatekeeper" of IP 3 receptor activation. Although [IP 3 ] may rise in all domains of the cytoplasm, it rises most near its site of production by PLC. At low or only modestly raised [IP 3 ], IRBIT remains bound to IP 3 receptors, antagonizing their activation by IP 3 . This may be the case for M 1 receptor stimulation. When [IP 3 ] rises to a certain point in the microdomain around the IP 3 receptor (between 0.3 and 1 M (26)), IRBIT is released from the IP 3 receptor, allowing unhindered binding of IP 3 , maximal opening of the IP 3 receptor, and evident rises in [Ca 2ϩ ] i . This is likely to be the case for B 2 and P2Y-receptor stimulation. Thus, IRBIT sets a threshold for [IP 3 ] for Ca 2ϩ release, conferring specificity to cellular signals that depend on co-localization of IP 3 receptor and the site of IP 3 production. A similar threshold mechanism has been proposed for CaM via Ca 2ϩ -dependent feedback inhibition of IP 3 Rs, and expression in SCG cells of DN CaM likewise increased [Ca 2ϩ ]; rises induced by oxo-M (22). We found that IRBIT affected the fraction of neurons that responded to each agonist and also affected the amplitude of the resultant [Ca 2ϩ ] i rise, although this latter effect was most evident for the case of oxo-M. Overexpression of IRBIT-S68A had little effect on the responses to UTP or BK, suggesting that [IP 3 ] at IP 3 receptors is usually high enough to overcome endogenous IRBIT (although this might not be the case at subsaturating concentrations of those agonists) but had large effects for oxo-M, for which the fraction of responding cells was much higher and the amplitude of the [Ca 2ϩ ] i rises in responding cells was larger. Finally, although there was a reduction in BK-induced [Ca 2ϩ ] i rises in neurons overexpressing WT IRBIT, the biggest effects were on the muscarinic responses. Thus, one possibility is that the physical association between IP 3 receptors and B 2 , but not M 1 , receptors (21) in some way physically or allosterically occludes the action of IRBIT on the IP 3 receptors.
Our model encompassing Ca 2ϩ i and phosphoinositide signals endowing receptor specificity is summarized in Fig. 6. Shown are two classes of receptors, the M 1 muscarinic and the B 2 /P2Y bradykinin/purinergic. The former are localized at some distance from IP 3 receptors, resulting in local [IP 3 ] being too low to overcome inhibition by bound IRBIT. Thus, the [Ca 2ϩ ] i rise observed is normally negligible. The B 2 /P2Y class is co-localized in microdomains with IP 3 receptors, as postulated previously (91). Consequently, local [IP 3 ] is high enough to overcome IRBIT action, causing release of IRBIT from IP 3 receptors, amplification of IP 3 action, and reliable [Ca 2ϩ ] i rises. The released Ca 2ϩ ions bind to and activate NCS-1, which accelerates activity of PI 4-kinase III␤, which may or may not need to reside in the PM. Because the B 2 receptors physically associate with IP 3 receptors (55), NCS-1 need not translocate to the PM during this event but is probably already preassembled in a complex with PI 4-kinase III␤. Our . Model accounting for receptor-specific phosphoinositide and Ca 2؉ signals in SCG neurons. All receptors activate PLC␤, which in turn hydrolyzes PIP 2 to IP 3 and DAG. The released Ca 2ϩ is regulated by the IP 3 R-binding protein, IRBIT, which sets a threshold for [IP 3 ] sufficient to open IP 3 Rs. Stimulation of M 1 muscarinic acetylcholine receptors (left) is ineffective in producing cytoplasmic Ca 2ϩ signals because the IP 3 produced is too far away from IP 3 Rs; thus, [IP 3 ] at the IP 3 R is too low to overcome the IRBIT threshold. Bradykinin B 2 and purinergic P2Y receptors (right) produce robust cytoplasmic Ca 2ϩ signals due to their spatial co-localization with IP 3 Rs, where [IP 3 ] is sufficiently high. Via NCS-1, bradykinin and purinergic, but not muscarinic, stimulation accelerates PI 4III␤-kinase activity. Via DAG-kinase conversion to PA, the produced DAG increases PI(4)P 5-kinase (PI4P-K) activity, in concert with Rho family proteins and Rho-kinase (R-K). PI(4)P 5-kinase activity is also increased by bradykinin and purinergic stimulation, but for clarity this is only shown for muscarinic stimulation. Acceleration of both PI 4III␤-and PI(4)P 5-kinases is required to increase PIP 2 synthesis that compensates for consumption of PIP 2 by PLC␤.
results indicate that stimulation of PI(4)P 5-kinases is also required to explain the lack of B 2 /P2Y action on I Ca . This signaling pathway requires Rho, Rho-kinase, and DAG-kinase activities because disruption of any one of these prevents compensatory PIP 2 synthesis. We depict these molecules as clustered together, associated with PI(4)P 5-kinase, but we have little information about which of them are constitutively pretethered and which associate together only upon receptor stimulation. We have no reason to think that this signal is receptor-specific. Importantly, this stimulation must not be tonic but rather receptor-stimulated. Because tonic rates of PI(4)P 5-kinase activity must be low enough to allow the PIP 2 depletion part of receptor action (30,73,92), and tonic increases in PI(4)P 5-kinase activity prevent PLC-mediated modulation of K ϩ channels (5,7,9,93), the molecules clustered around PI(4)P 5-kinases that we identify here cannot be all constitutively active, but rather one or more must be turned on by receptor agonists. Interestingly, IRBIT is only functional if phosphorylated at multiple amino-terminal serines, including Ser 68 (25), meaning that its cognate kinase could regulate IRBIT action and Ca 2ϩ i signaling in neurons. Although IRBIT was shown to be phosphorylated in vitro by casein kinase I and is known to be dephosphorylated by PP1 protein phosphatases (25), its physiological kinase in vivo is unknown; also, it is unknown whether such a kinase could be activated downstream of stimulation of G protein-coupled receptors. If so, this could underlie another mechanism of G q/11 -coupled receptor specificity.