Sustained entry of Ca2+ is required to activate Ca2+-calmodulin-dependent phosphodiesterase 1A.

Regulation of adenylyl cyclases (ACs) by Ca2+ requires capacitative Ca2+ entry (CCE) (Cooper, D. M. F. (2003) Biochem. J. 375, 517-529), but whether Ca2+-sensitive phosphodiesterases (PDEs) are similarly discriminating has never been addressed. In the present study, a variety of conditions were devised to manipulate [Ca2+]i so that we could ask whether PDE1 selectively responds to different modes of elevating [Ca2+]i, viz. Ca2+ released from intracellular stores and various modes of Ca2+ entry. In 1321N1 human astrocytoma cells, the endogenous PDE1 (identified as PDE1A by reverse transcriptase-PCR) was largely insensitive to Ca2+ released from carbachol-sensitive stores but was robustly stimulated by a similar rise in [Ca2+]i due to carbachol-induced Ca2+ influx. Gd3+, which effectively blocked thapsigargin-induced CCE and its effect on PDE1A, also inhibited the activation of PDE1A by carbachol-induced Ca2+ entry. However, non-selective ionomycin-mediated Ca2+ entry also activated PDE1A, so that, unlike Ca2+-sensitive ACs, PDE1A cannot discriminate between the different sources of Ca2+ entry. Fractionation of the cells revealed that the Ca2+-calmodulin-stimulated PDE activity was not present at the plasma membrane but was associated with the cytosol and the organellar compartments of the cell. Therefore, the apparent disparity between PDE1A and ACs is likely to be the consequence of their differential subcellular localization. Nevertheless, in a physiological context, where artificial modes of elevating [Ca2+]i are not available, as with ACs, a dependence on CCE would be evident, and it would be the duration of this influx of Ca2+ that would determine how long PDE1A was activated.

A rise in [Ca 2ϩ ] i 1 leads to an inhibition of cAMP accumulation in a variety of cell types (1)(2)(3)(4)(5)(6)(7)(8). In some cases, the inhibition of cAMP accumulation is exerted on Ca 2ϩ -inhibitable adenylyl cyclases (ACs) (1-3, 5-7), whereas in others the effect may be mediated by Ca 2ϩ -calmodulin-dependent phosphodiesterases (PDE1) (4,8). ACs are extremely discriminating in terms of the source of the Ca 2ϩ to which they respond (9). In non-excitable cells, Ca 2ϩ -sensitive ACs respond only to CCE (9), whereas other modes of elevating [Ca 2ϩ ] i , including release from intracellular stores (6,7,10) and ionophore- (6,7,10) or arachidonic acid-mediated Ca 2ϩ entry (11), are ineffective. This dependence, along with other evidence, suggests that Ca 2ϩ -sensitive ACs and CCE channels must be functionally co-localized and that cellular strategies are in place to ensure their association (12,13). By contrast, although type I PDEs (PDE1) are known to be markedly stimulated by Ca 2ϩ acting via calmodulin in vitro, little if anything is known about the mode of [Ca 2ϩ ] i rise to which they will respond in the intact cell.
In the human astrocytoma cell line 1321N1, the activation of receptors that stimulate the formation of inositol 1,4,5trisphosphate (InsP 3 ) substantially inhibits cAMP accumulation (8,14,15). Complete reversal of the inhibition by PDE1specific inhibitors is consonant with the agonist-evoked rise in [Ca 2ϩ ] i -activating PDE1 and hence increasing the rate of cAMP hydrolysis (16). Indeed, these early studies established that PDE1 activity is markedly increased when Ca 2ϩ is introduced into the extracellular medium (14,15). Therefore, PDE1 does seem to be regulated by [Ca 2ϩ ] i , but the source of the Ca 2ϩ to which PDE1 responds and whether PDE1 is as discriminating as Ca 2ϩ -sensitive ACs has never been addressed.
In the present study, we first characterized the various modes by which [Ca 2ϩ ] i could be elevated in 1321N1 cells. We then established a variety of conditions to manipulate [Ca 2ϩ ] i so that we could ask whether PDE1 discriminates between Ca 2ϩ signaling pathways, viz. Ca 2ϩ released from intracellular stores, and the different modes of Ca 2ϩ entry. We also established by RT-PCR that PDE1A was the only PDE1 isoform expressed in these cells. Furthermore, we compared the effect of Ca 2ϩ entry following stimulation with the muscarinic agonist carbachol (CCh) or non-selective entry mediated by the ionophore, ionomycin, and the triggering of CCE by the sarco-(endo)plasmic reticulum Ca 2ϩ -ATPase inhibitor, thapsigargin, on isoproterenol-evoked cAMP accumulation. Our findings establish that Ca 2ϩ entry is the major stimulus for PDE1A in 1321N1 cells, but unlike Ca 2ϩ -sensitive ACs, PDE1A does not discriminate between different modes of Ca 2ϩ entry. We wondered whether the lack of selectivity for the source of Ca 2ϩ entry was related to the subcellular distribution of PDE1A. Following fractionation of the cells, we found that Ca 2ϩ -calmodulin-stimulated PDE activity was detected in the cytosol and the non-plasma membrane organellar compartments of the cell. Therefore, it would appear that the more diffuse subcellular organization of PDE1A renders it susceptible to non-selective Ca 2ϩ entry. Nevertheless, under physiological conditions, CCE is the dominant Ca 2ϩ entry pathway in these cells and the dependence of PDE1A activation on CCE would be evident.
Measurement of Intracellular cAMP Accumulation-Intracellular cAMP levels were measured as previously described by Evans et al. (17) with some modifications. 1321N1 cells were incubated in Dulbecco's modified Eagle's medium (2 h, 37°C) supplemented with [2-3 H]adenine (1 Ci/ml) to label the ATP pool. The cells were washed and incubated (30°C, 10 min) with KBS. After further washing, cells were incubated in either nominally Ca 2ϩ -free KBS or Ca 2ϩ -free KBS (nominally Ca 2ϩfree KBS supplemented with 1 mM EGTA) with test reagents. The assays were terminated by aspirating the medium and replacing it with ice-cold 10% trichloroacetic acid. Unlabeled cAMP (100 l, 10 mM), ATP (10 l, 65 mM), and [␣-32 P]ATP (ϳ3,000 cpm) were added to measure the recovery of cAMP and ATP. After centrifugation (16,089 ϫ g, 6 min), the [␣-32 P]ATP and [2, H]cAMP content of the supernatant was determined by anion exchange chromatography using the standard Dowex/ alumina protocol (18).
Preparation of Membranes from 1321N1 Cells for Adenylyl Cyclase Assay-1321N1 cells that were grown for 5 days in 75-cm 2 flasks were detached with PBS containing 0.03% w/v EDTA and centrifuged at 195 ϫ g for 5 min. The supernatant was removed, and the pellet was resuspended in 2 ml of hypotonic buffer (HB buffer: 2 mM MgCl 2 , 1 mM EDTA, 50 mM Tris-HCl, 1 mM 4-(2-aminoethyl)benzensulfonyl fluoride, 1 mM benzamidine, 1 g of DNase I, pH 7.4), and the cells were lysed by repeatedly passing the cell suspension through a 21-gauge needle. After centrifugation (195 ϫ g, 5 min) and further dissociation, the lysate was centrifuged at 17,257 ϫ g (15 min, 4°C). The supernatant was removed, and the pellet (crude membrane fraction) was resuspended in 250 -500 l of assay buffer (assay buffer: 40 mM Tris-HCl, 800 M EGTA, 0.25% w/v BSA, pH 7.4) and stored in liquid nitrogen until required.
Measurement of Adenylyl Cyclase Activity-Adenylyl cyclase activity was measured as previously described (2). Briefly, 1321N1 membranes (4 g of protein) were incubated ( (see under "Determination of Free Ca 2ϩ Concentrations" below). Reactions were terminated by the addition of 100 l ice-cold stopper buffer (43.9 mM ATP, 3.19 mM cAMP, and 1% w/v SDS). [2, H]cAMP (ϳ3000 cpm) was added as a recovery marker, and the amount of [ 32 P]cAMP that was formed was quantified as previously described (18).
Cell Fractionation-1321N1 cells were grown for 5 days in 150-mm diameter dishes and then detached with PBS containing 0.03% w/v EDTA and centrifuged at 195 ϫ g for 5 min. The pellet was resuspended in 1 ml of assay buffer (50 mM Tris-HCl, pH 7.4) supplemented with 1 mM phenylmethylsulfonyl fluoride, 1 mM benzamidine, 2 mM 4-(2-aminoethyl)benzensulfonyl fluoride, 1 mM EDTA, 130 M bestatin, 14 M E-64, 1 M leupeptin, 0.3 M aprotinin, and 1 g of DNase I. The cells were then lysed by repeatedly passing the cell suspension through a 21-gauge needle. Unbroken cells and nuclei were removed from the lysate by low speed centrifugation (195 ϫ g, 5 min, 4°C). The supernatant was removed and centrifuged (17,257 ϫ g, 15 min, 4°C) to pellet the crude membrane fraction (containing plasma membranes) (19), and the resulting supernatant ("crude cytosolic" fraction) was either used immediately or fractionated further before use. To separate the cytosol from the non-plasma membrane organellar components of the cell, the "crude cytosolic" fraction was centrifuged at high speed (105,000 ϫ g, 60 min, 4°C) (20). The final supernatant was designated as the cytosolic fraction and the pellet as the organellar fraction (21). All final fractions were suspended in an equal total volume of 50 mM Tris-HCl supplemented with 0.2% w/v BSA and 4 mM dithiothreitol (pH 7.4) and used immediately. The protein content was determined by using the bicinchoninic acid method (22).
Measurement of Cyclic Nucleotide Phosphodiesterase Activity-Phosphodiesterase activity was measured as described by Shahid  [Adenine-U-14 C]cAMP (ϳ3000 dpm) was added to each sample to measure the recovery of [2, H]cAMP from alumina columns (23). The amount of [2, H]cAMP that was hydrolyzed was quantified by expressing the corresponding [2, H]cAMP counts following recovery from the columns as a function of the total [2, H]cAMP added to each assay tube. Blank values were obtained by using previously boiled preparations. There was no significant difference between the decrease in the amount of cAMP detected with boiled enzymes and when no protein was added. The amount of cAMP that was hydrolyzed was expressed as either picomoles of cAMP/ min/mg of protein or picomoles of cAMP/min/l of fraction.
Determination of Free Ca 2ϩ Concentrations-Free Ca 2ϩ concentrations were calculated as previously described (24). Briefly, this involved an iterative computing program that solved equations that described the complexes formed within a mixture composed of the assay buffer components (see under "Measurement of Adenylyl Cyclase Activity" and "Measurement of Phosphodiesterase Activity" above). Final assay mixture concentrations of free Ca 2ϩ (in the presence of 200 M EGTA) are shown.
Reverse Transcriptase-PCR-Total RNA was extracted from 1.5 ϫ 10 6 1321N1 cells using the SV total RNA isolation kit (Promega). RNA (1 g) was reverse-transcribed by Superscript II enzyme with 0.5 g of oligo(dT) 16 . The reaction mixture was incubated at 42°C for 50 min, followed by a further incubation at 70°C for 15 min. PCR was performed using 10 ng of cDNA, 10 pmol of each of the two specific primers, 1.5 mM MgCl 2 , 0.2 mM dNTPs, and 2.5 units of TaqDNA polymerase. The specific primers were CCATTTTCCCCACTTTGTGATCG, an Nterminal primer common to all PDE1 isoforms; TCCACCTCTCTTTGT-TCTGC, a C-terminal primer specific for PDE1A; CGTCAATGGA-CATCTGGTTGG, a C-terminal primer specific for PDE1B; and TTCTCCTCTTTGGGTACCTTGGC, a C-terminal primer specific for PDE1C. The amplification profile consisted of heating the mixture at 94°C for 30 s, annealing at 55°C for 30 s, and extension at 72°C for 30 s, for 30 cycles (an initial heating to 94°C for 3 min was performed). To ensure the fidelity of mRNA extraction and reverse transcription, the cDNA was subjected to PCR amplification with oligonucleotide primers specific for the constitutively expressed gene ␤-actin.
Statistical Analysis-Results are shown as the means Ϯ S.E. of the mean (means Ϯ S.E.) of at least three individual experiments. Statistical significance was assessed by using paired Student's t test, where p Յ 0.05 was considered significant.

CCh Mediates a Ca 2ϩ -dependent Decrease in Isoproterenolevoked cAMP Accumulation-Stimulation of endogenous
␤-adrenoceptors in 1321N1 cells with isoproterenol evoked a concentration-dependent increase in cAMP accumulation (EC 50 ϭ 8.7 Ϯ 0.1 nM, n ϭ 3) (Fig. 1A, inset). The time course for cAMP accumulation after stimulation with a maximal concentration of isoproterenol (10 M) in the presence and absence of CCh (1 mM) was examined (Fig. 1A). In Ca 2ϩ -free conditions, isoproterenol stimulated a time-dependent increase in cAMP accumulation that reached a maximum level after 10 min. A similar profile was observed in cells that were co-stimulated with isoproterenol and CCh in Ca 2ϩ -free conditions or isoproterenol alone in the presence of 3 mM Ca 2ϩ . However, co-stimulation with isoproterenol (10 M) and CCh (1 mM) in Ca 2ϩ -containing medium produced two notable effects. First, there was a significant decrease in the amount of cAMP detected after each time interval compared with that evoked by isoproterenol alone, and second, the time required to achieve a maximal cAMP response was reduced from 10 to 2 min (Fig. 1A).
To confirm that the effect of CCh on isoproterenol-evoked cAMP accumulation was indeed Ca 2ϩ -dependent, cells were depleted of Ca 2ϩ by pre-treatment with ionomycin in the absence of extracellular Ca 2ϩ . This has the effect of causing extrusion of Ca 2ϩ from intracellular stores out of the cell. Under such circumstances, it was possible to isolate any potential action of CCh on isoproterenol-evoked cAMP accumulation that was independent of Ca 2ϩ mobilization from Ca 2ϩ -dependent effects. Because the removal of both extracellular and intracellular sources of Ca 2ϩ eliminated the inhibitory effect of CCh on isoproterenol-evoked cAMP accumulation (Fig. 1B), it is possible to conclude that the observed inhibition of cAMP accumulation by CCh is a consequence of a rise in [Ca 2ϩ ] i . The relative effects of what was assumed to be Ca 2ϩ release from InsP 3 -sensitive stores and Ca 2ϩ entry on isoproterenolevoked cAMP accumulation were compared as a function of the total inhibition evoked by CCh, i.e. that attributable to the increase in [Ca 2ϩ ] i due to release plus entry in the presence of extracellular Ca 2ϩ . Of the total inhibition of the cAMP response by CCh, 22 Ϯ 6% (n ϭ 25) (viz. in the absence of extracellular Ca 2ϩ ) was dependent on Ca 2ϩ released from intracellular stores and 78 Ϯ 6% (n ϭ 25) on Ca 2ϩ influx.
The extrusion of cAMP from the cell may provide a theoretical means (in addition to phosphodiesterases) whereby an intracellular cAMP signal may be rapidly diminished (25). Although the identity of the cAMP transporters remains unclear (25), we considered the possibility that the rise in [Ca 2ϩ ] i evoked by CCh could enhance the extrusion of cAMP from the cell. It turned out that, in the presence of 3 mM Ca 2ϩ , the amount of cAMP detected in the extracellular medium following exposure to isoproterenol (10 M) and CCh (1 mM) was 11.88 Ϯ 3.27% (n ϭ 4) of the total cAMP level compared with 7.96 Ϯ 1.10% (n ϭ 4) with isoproterenol alone. Obviously, changes in this small fraction of the total cAMP cannot account for the present effects of CCh on isoproterenol-evoked cAMP accumulation.
Characterization of Adenylyl Cyclase in 1321N1 Cells-The regulation of adenylyl cyclases (ACs) by elevated [Ca 2ϩ ] i provides a means whereby Ca 2ϩ can modulate the cAMP pathway at the earliest opportunity. Of the 10 known isoforms of AC, five are either stimulated or inhibited by physiological increases in [Ca 2ϩ ] i (9). AC1, AC3, and AC8 are activated by Ca 2ϩ in a calmodulin-dependent manner, whereas AC5 and AC6 are Ca 2ϩ -inhibitable, but calmodulin-independent (9). Therefore, a possible explanation for the inhibitory effect of CCh on isoproterenol-evoked cAMP accumulation (Fig. 1A) could be the inhibition of AC activity by Ca 2ϩ . To address this issue, the sensitivity of ACs to Ca 2ϩ was measured using membranes prepared from 1321N1 cells. AC activity in 1321N1 membranes was inhibited by Ca 2ϩ in a concentration-dependent manner (IC 50 ϭ 273 Ϯ 25 nM, n ϭ 3), which was unaffected by calmodulin (IC 50 ϭ 279 Ϯ 80 nM, n ϭ 3) (Fig. 2). Thus 1321N1 cell membranes express either AC5 or AC6 adenylyl cyclase. However, inhibition of adenylyl cyclase cannot play any role in the inhibition of cAMP accumulation observed in the present studies, because the inhibition is eliminated in the presence of PDE inhibitors (see below); a situation that does not arise when direct inhibition of adenylyl cyclase occurs in intact cells (2,7).
The Activation of PDE1A by Ca 2ϩ Entry Underlies the Mechanism for the Inhibition of Isoproterenol-evoked cAMP Accumulation-Of the 11 known members of the PDE family (26), only PDE1 (27) and PDE4 (28) are directly regulated by a rise in [Ca 2ϩ ] i . However, the mechanisms whereby Ca 2ϩ regulates these enzymes are different: PDE1 is directly activated following the binding of Ca 2ϩ -calmodulin (29), whereas intracellular Ca 2ϩ facilitates the association of PDE4A1 with the Golgi membrane (28) without affecting its activity (30). Indirect effects of Ca 2ϩ (e.g. protein kinase C) on other Ca 2ϩ -insensitive PDEs may also account for the enhanced cAMP hydrolysis in the presence of CCh.
PDE2 is the only PDE directly activated by cGMP (26) and is the most likely candidate for the third PDE identified in 1321N1 cells (16). Although protein kinase C phosphorylates and activates PDE2 (36), in fact, stimulation of protein kinase C enhances cAMP accumulation in 1321N1 cells independently of PDE (37). Therefore, cGMP-stimulated PDE plays little or no role in controlling intracellular cAMP signals in these cells. We conclude that the inhibitory effect of CCh on cAMP accumulation is a direct effect of intracellular Ca 2ϩ on PDE1, and any indirect effects of the rise in [Ca 2ϩ ] i on other PDEs is highly unlikely. This issue is tackled more directly in the cell fractionation studies described later (see Fig. 8).
To identify the subtype of PDE1 that was expressed in 1321N1 cells, RT-PCR was performed using total RNA and isoform-specific primers for the PDE1A, PDE1B, and PDE1C genes and a primer that was common to all PDE1s. Of the three PDE1 isoforms (26), only PDE1A mRNA was detected in 1321N1 cells (Fig. 4). Therefore, the observed inhibitory effect of Ca 2ϩ entry on isoproterenol-evoked cAMP accumulation is mediated by PDE1A.
PDE1A Is Activated by Capacitative Ca 2ϩ Entry-The experiments described hitherto have inferred that Ca 2ϩ entry plays a role in activating PDE1 in 1321N1 cells, but without meas- uring [Ca 2ϩ ] i , it is not possible to address the mechanism with any precision. When [Ca 2ϩ ] i is monitored it is possible to manipulate and isolate various modes of [Ca 2ϩ ] i rise and determine their significance for the regulation of PDE1. Consequently, in the following series of experiments we established conditions where known modes of Ca 2ϩ rises were triggered, and we then applied those conditions to measurements of cAMP accumulation.
In excitable cells, a major mode of Ca 2ϩ entry is through voltage-gated Ca 2ϩ channels (42). To evaluate whether voltagegated Ca 2ϩ channels might represent an additional route for Ca 2ϩ -mediated regulation of PDE1A in 1321N1 cells, we looked for Ca 2ϩ influx in the presence of depolarizing concentrations of KCl (20 -60 mM). Not even the highest concentration of KCl could trigger Ca 2ϩ entry (data not shown), which concurs with the reported absence of voltage-gated Ca 2ϩ entry channels in 1321N1 cells (43,44). Therefore, this issue cannot be resolved with this model system.
The effect of CCE on isoproterenol-evoked cAMP accumulation was examined under analogous experimental conditions as those used for the Ca 2ϩ measurements described above (Fig.  5D). Incubating the cells with Tg (1 M, 15 min) alone did not significantly affect isoproterenol-evoked cAMP accumulation in 1321N1 cells (Fig. 5D). However, the addition of 3 mM Ca 2ϩ after 15-min incubation with Tg significantly reduced the isoproterenol-evoked cAMP response (51 Ϯ 5%, n ϭ 4), an effect that was abolished by preincubating cells with low concentrations of IBMX (50 M, 10 min), MMX (50 M, 10 min), or Gd 3ϩ (10 M, 5 min) (Fig. 5D). These results show that Ca 2ϩ influx through CCE channels can enhance the hydrolysis of cAMP by activating PDE1A.
Characterization of CCh-induced Ca 2ϩ Entry and Its Effect on PDE1A-Recently, the physiological role of CCE has been brought into question by some reports suggesting that arachidonic acid-dependent "non-capacitative Ca 2ϩ entry" (NCCE) is a dominant Ca 2ϩ entry pathway that is preferentially activated by physiological concentrations of agonists (41,45,46). The activation of muscarinic M 3 receptors leads to the formation of InsP 3 and diacylglycerol via phospholipase C. Diacylglycerol may be further metabolized to produce arachidonic acid. Even though arachidonic acid-dependent NCCE has only been demonstrated in a few cell types (41,46), it seemed reasonable to entertain the possibility that CCh might activate both NCCE and CCE in 1321N1 cells.
In nominally Ca 2ϩ -free medium, CCh (1 mM) evoked a transient increase in [Ca 2ϩ ] i (1143 Ϯ 152 nM, n ϭ 5), which rapidly returned to its basal level (t1 ⁄2 ϭ 24 Ϯ 2s, n ϭ 5) (Fig. 6A). The subsequent addition of Ca 2ϩ to the medium, triggered a second Ca 2ϩ signal that represented Ca 2ϩ entry. The peak increase in [Ca 2ϩ ] i that followed Ca 2ϩ entry was 712 Ϯ 109 nM (n ϭ 5). The Ca 2ϩ signal attributable to Ca 2ϩ entry did not return to the basal level, but achieved a steady state at an elevated [Ca 2ϩ ] i . The rate at which the Ca 2ϩ signal reached the steady-state was significantly longer for the Ca 2ϩ entry phase than that of the release phase (half-time, t1 ⁄2 ϭ 93 Ϯ 29s, n ϭ 5).
The effects of the CCE blocker, Gd 3ϩ , on Ca 2ϩ entry induced by 1 mM CCh is shown in Fig. 6A. Preincubating cells (5 min) with 10 M Gd 3ϩ inhibited CCh-induced Ca 2ϩ entry by 86 Ϯ 2% (n ϭ 5) (cf., Fig. 5B). Therefore, the inhibition of isoproterenolevoked cAMP accumulation by CCh shown in Fig. 2 can be interpreted to be the consequence of CCh-induced CCE activat-ing PDE1A. This, however, does not exclude the possibility that, at lower concentrations of CCh, NCCE could regulate the activity of PDE1A. 2 To address this issue, we examined the nature of the Ca 2ϩ influx that was promoted by low concentrations of CCh (10 -30 M) in the presence of 100 M 2-APB. A short incubation (5 min) with 100 M 2-APB reduced the Ca 2ϩ entry that followed the application of 10 -30 M CCh (data not shown). This blocking effect of 2-APB on Ca 2ϩ entry triggered by low concentrations of CCh indicates that NCCE does not occur in 1321N1 cells.
A Generalized Rise in [Ca 2ϩ ] i Can Activate PDE1A-To determine whether a similarly intimate relationship existed between PDE1A and CCE in 1321N1 cells, as between Ca 2ϩsensitive ACs and CCE, we compared the effects of Tg-induced CCE with a generalized ionophore-mediated Ca 2ϩ entry on PDE1 activity. At submicromolar concentrations, ionomycin 2 It is considered that, in sources where both occur, NCCE is triggered by low concentrations of agonist (or arachidonate) and is not blocked by 2-APB, whereas high concentrations of agonist trigger CCE, which is blocked by 2-APB (56). selectively permeabilizes intracellular membranes (47) facilitating the flux of Ca 2ϩ from organelles into the cytosol to trigger CCE. However, at higher concentrations, ionomycin also permeabilizes the plasma membrane to divalent cations (48) to cause a generalized increase in [Ca 2ϩ ] i that is independent of Ca 2ϩ entry channels and which cannot be blocked by CCE channel blockers.
Ionomycin (10 M, 15 min) evoked a dramatic increase in [Ca 2ϩ ] i that corresponded to the emptying of intracellular Ca 2ϩ stores (Fig. 7A). The subsequent addition of Ca 2ϩ to the extracellular medium caused a concentration-dependent increase in [Ca 2ϩ ] i . The response to 30 M Ca 2ϩ was similar in magnitude to that evoked by 3 mM Ca 2ϩ in Tg pre-treated cells (576 Ϯ 89 nM, n ϭ 9 and 527 Ϯ 86 nM, n ϭ 10, respectively) (Fig. 7A). This ionomycin-mediated Ca 2ϩ entry was unaffected by 2-APB (100 M) (Fig. 7B) at a concentration that inhibited CCE (Fig. 4C) confirming that it was a non-selective entry mechanism.
Under identical conditions to those used in the Ca 2ϩ measurements, ionomycin alone did not significantly affect isoproterenol-evoked cAMP accumulation (Fig. 7C). However, the addition of 30 M Ca 2ϩ to the extracellular medium produced a substantial inhibition of isoproterenol-evoked cAMP accumulation in ionomycin-pre-treated cells (41 Ϯ 8%, n ϭ 4). The effect was completely reversed by IBMX (50 M) and MMX (50 M) (Fig. 7C). This result establishes that not only CCE, but also a nonspecific, generalized rise in [Ca 2ϩ ] i can activate PDE1 in 1321N1 cells.
PDE1A Is Distributed within the Cytosol and Non-plasma Membrane Organellar Compartments of 1321N1 Cells-The functional co-localization of Ca 2ϩ -sensitive ACs with CCE channels is reinforced by their subcellular distribution (12,13). The disparity between the selectivity of ACs and the nonselectivity of PDE1 for particular sources of Ca 2ϩ entry pathways might therefore be explained by the spatial distribution of the enzymes within the cell. To examine this possibility, we initially measured Ca 2ϩ -calmodulin-dependent PDE activity in crude membrane and "crude cytosolic" fractions prepared from 1321N1 cells (Fig. 8). In the presence of calmodulin, Ca 2ϩ evoked a concentration-dependent increase in PDE activity in the cytosolic fraction (Fig. 8A), which was completely inhibited by the PDE1-specific inhibitor MMX (50 M) (Fig. 8B). By contrast, Ca 2ϩ -calmodulin-dependent PDE activity was not detected in the membrane fraction (Fig. 8C).
We considered the possibility that the crude cytosol described above could include non-plasma membrane organelles with which PDE1A would be associated. Therefore, we fractionated the crude cytosolic fraction into cytosol and organellar components and measured the Ca 2ϩ -calmodulin-dependent PDE activity. AC activity was also measured to show the subcellular distribution of PDE1A relative to plasma membraneassociated ACs. Ca 2ϩ -calmodulin-dependent PDE activity oc-curred in the cytosol and, to a slightly lesser extent, in the organellar components of the cells (Fig. 9A, panels i and ii), but was absent from the plasma membrane (Fig. 9A, panel iii). In all cases, the Ca 2ϩ -calmodulin-dependent increase in PDE activity was abolished by the PDE1-specific inhibitor MMX (50 M). In contrast, all AC activity was localized to the crude plasma membrane fraction (Fig. 9B) (19).
The experiments described above examined the subcellular distribution of PDE1A at resting [Ca 2ϩ ] i . Because the effect of elevated [Ca 2ϩ ] i on the subcellular localization of PDE1 has never been explored, we entertained the further possibility that the intracellular targeting of PDE1A may be a function of the rise in [Ca 2ϩ ] i that follows agonist stimulation in a manner analogous to PDE4A1 (28). 1321N1 cells were treated with ionomycin (10 M) and Ca 2ϩ (30 M) as shown in Fig. 7 prior to lysis, and the PDE activity in the different cellular fractions was measured (data not shown). Increasing [Ca 2ϩ ] i prior to lysis did not affect the subcellular distribution of PDE1A compared with resting [Ca 2ϩ ] i . Therefore, PDE1A is likely to remain cytosolic or associated with organellar membranes in 1321N1 cells, and it seems reasonable to conclude that it is this diffuse organization that renders the enzyme susceptible to non-selective Ca 2ϩ entry. DISCUSSION The regulation of ACs by Ca 2ϩ has been extensively characterized (9). In the intact cell, they show a remarkable dependence on CCE: neither release from intracellular stores (6, 7, 10) nor ionophore-(6, 7, 10) or arachidonic acid-mediated Ca 2ϩ entry (11) affect AC activity. As a consequence, Ca 2ϩ -sensitive ACs have been postulated to lie close to CCE channels in the plasma membrane. In contrast, very little is known about the regulation of Ca 2ϩ -sensitive PDEs by Ca 2ϩ . Although earlier work has established that activation of Ca 2ϩ -calmodulin-dependent PDE1 requires extracellular Ca 2ϩ (15,16) the nature of the Ca 2ϩ signal involved has never been explored. In the present study, we first characterized the various modes of [Ca 2ϩ ] i elevation in the human astrocytoma cell line, 1321N1. By using selective PDE inhibitors, we established that 1321N1 cells express endogenous PDE1, and through RT-PCR identified PDE1A as the only PDE1 isoform expressed in these cells. We then manipulated [Ca 2ϩ ] i by various means to ask (i) whether Ca 2ϩ release from intracellular stores regulates PDE1 activity, (ii) whether CCE activates PDE1 and, (iii) whether any other mode of elevating [Ca 2ϩ ] i could regulate PDE1. Finally, we explored the subcellular distribution of PDE1A as a potential determinant in its selectivity toward Ca 2ϩ from a particular source.
In 1321N1 cells, CCh evokes an increase in [Ca 2ϩ ] i through the formation of InsP 3 and the subsequent release of Ca 2ϩ from InsP 3 -sensitive stores (49). Because removal of extracellular Ca 2ϩ almost eliminates the inhibitory effects of InsP 3 -coupled agonists on cAMP accumulation in 1321N1 cells, we can infer that release of Ca 2ϩ from intracellular stores plays little role in the regulation of PDE1A. The importance of Ca 2ϩ entry versus Ca 2ϩ release for the regulation of PDE1A is emphasized when the substantial, but relatively ineffective rise in [Ca 2ϩ ] i due to Ca 2ϩ release is compared with the robust effect of a similar rise in [Ca 2ϩ ] i due to Ca 2ϩ influx (see Fig. 6).
In non-excitable cells, CCE plays a critical role in determining the amplitude of sustained elevations in [Ca 2ϩ ] i and replenishing depleted intracellular Ca 2ϩ stores. Although the identity of store-operated Ca 2ϩ channels remains unclear, prime candidates for the role are mammalian homologues of the Drosophila transient receptor potential (trp) protein (50). The localization of these trp proteins and Ca 2ϩ -sensitive ACs in cholesterol-rich domains of the plasma membrane (51) may provide the means whereby Ca 2ϩ -sensitive ACs are exclusively regulated by CCE. Our findings show that CCE triggered independently of receptor activation (i.e. Tg) causes a profound increase in PDE1A activity (Fig. 5). Although CCE clearly stimulates PDE1A, the use of Tg to induce CCE may overlook modes of Ca 2ϩ entry that are stimulated by agonists, in addition to CCE. For example, when CCh increases InsP 3 formation in 1321N1 cells (49), it may be assumed that the depletion of CCh-sensitive Ca 2ϩ stores will trigger CCE. However, CCh may also activate NCCE through a diacylglycerol/arachidonic acid-driven pathway. It turned out that known CCE blockers abolished CChinduced Ca 2ϩ entry and its effect on PDE1 (Fig. 6). Therefore, we can conclude that CCE is the primary route of Ca 2ϩ entry and the major stimulus for PDE1A activation in 1321N1 cells.
The dependence of PDE1 activity on Ca 2ϩ influx might reflect either the functional co-localization of the enzyme with Ca 2ϩ entry channels, localization of the enzyme near the plasma membrane, or simply a requirement for a sustained elevation of [Ca 2ϩ ] i , irrespective of its origin. To distinguish between these possibilities, we devised conditions where the effect on PDE1A activity of non-selective Ca 2ϩ influx could be compared with that of CCE. Unlike CCE, ionophore-mediated Ca 2ϩ entry is not restricted to discrete regions of the plasma membrane, but allows a nonspecific influx of Ca 2ϩ into the cell (6). The experimental conditions developed permitted ionomycin to cause a substantial increase in [Ca 2ϩ ] i in the presence of a low concentration of Ca 2ϩ (30 M) that evoked no CCE in Tg-treated cells (Figs. 5 and 7). The failure of CCE blockers to inhibit ionomycin-mediated Ca 2ϩ entry established that iono- mycin stimulated nonspecific Ca 2ϩ entry in 1321N1 cells. This ionomycin-mediated Ca 2ϩ entry also activated PDE1A, and so we conclude that PDE1A requires sustained Ca 2ϩ entry for activation, but unlike Ca 2ϩ -sensitive ACs, it does not discriminate between the different routes of Ca 2ϩ entry. However, the possibility should not be discounted that other PDE1 isoforms (viz. PDE1B or PDE1C) might be more selective for the source of Ca 2ϩ to which they will respond. Future studies involving heterologous expression of the various isoforms might be insightful in this regard.
The disparity between the regulation of Ca 2ϩ -sensitive ACs and PDE1A by Ca 2ϩ is most likely to reflect the subcellular placement of these enzymes. A major contributing factor for the regulation of Ca 2ϩ -sensitive ACs by CCE is their compartmentalization in cholesterol-rich domains of the plasma membrane (12,13). Indeed, other proteins that are specifically regulated by CCE are also targeted to this domain, as are trp proteins (51,52). Unlike ACs, which possess twelve transmembrane-spanning domains and are always associated with the plasma membrane, only some of the known PDEs have the potential to be associated with the plasma membrane, and this is dependent on lipid modifications to amino acids in their N-terminal regions (20,26). The subcellular distribution of PDEs is largely determined by the formation of signaling complexes with scaffolding proteins (53). Although extremely little is known about the intracellular targeting of PDE1, at resting [Ca 2ϩ ] i the enzyme is likely to be distributed within the cytosol (54) or associated with cytoskeletal components (55) (Fig. 9).
Although PDE1A does not discriminate between specific and nonspecific routes of Ca 2ϩ entry it is highly selective for Ca 2ϩ entering the cell versus Ca 2ϩ released from intracellular stores. The dependence of PDE1A on Ca 2ϩ entering the cell may be another manifestation of its subcellular distribution. Selective targeting of PDE1A to the sub-plasmalemmal space (possibly through association with cytoskeletal, accessory proteins, or organellar structures) could allow regulation of the enzyme by Ca 2ϩ entry, while potentially shielding it from Ca 2ϩ released from intracellular stores.
In conclusion, the present study has significantly refined our understanding of how PDE1 is regulated by [Ca 2ϩ ] i . It is clear that Ca 2ϩ released from intracellular stores plays little or no role in the regulation of PDE1A compared with robust activation by Ca 2ϩ influx. However, PDE1A is unable to discriminate between the different sources of Ca 2ϩ entry owing to its subcellular distribution. Nevertheless, it is worth noting that, in a physiological context, where artificial modes of elevating [Ca 2ϩ ] i are not available, as with ACs, a dependence of PDE1A on CCE would be manifest, and it would be the duration of the influx of Ca 2ϩ that would determine for how long PDE1A was activated.