Amplification of Ca2+ signaling by diacylglycerol-mediated inositol 1,4,5-trisphosphate production.

Stimulation of various cell surface receptors leads to the production of inositol 1,4,5-trisphosphate (IP3) and diacylglycerol (DAG) through phospholipase C (PLC) activation, and the IP3 and DAG in turn trigger Ca2+ release through IP3 receptors and protein kinase C activation, respectively. The amount of IP(3) produced is particularly critical to determining the spatio-temporally coordinated Ca(2+)-signaling patterns. In this paper, we report a novel signal cross-talk between DAG and the IP3-mediated Ca(2+)-signaling pathway. We found that a DAG derivative, 1-oleoyl-2-acyl-sn-glycerol (OAG), induces Ca2+ oscillation in various types of cells independently of protein kinase C activity and extracellular Ca2+. The OAG-induced Ca2+ oscillation was completely abolished by depletion of Ca2+ stores or inhibition of PLC and IP3 receptors, indicating that OAG stimulates IP3 production through PLC activation and thereby induces IP3-induced Ca2+ release. Furthermore, intracellular accumulation of endogenous DAG by a DAG-lipase inhibitor greatly increased the number of cells responding to agonist stimulation at low doses. These results suggest a novel physiological function of DAG, i.e. amplification of Ca2+ signaling by enhancing IP3 production via its positive feedback effect on PLC activity.

Stimulation of a wide variety of cells by hormones, neurotransmitters, or growth factors leads to the activation of phospholipase C (PLC) 1 and triggers inositol 1,4,5-trisphosphate (IP 3 )-mediated Ca 2ϩ signaling via activation of IP 3 receptors (IP 3 Rs) on the endoplasmic reticulum (1,2). Proper regulation of receptor-IP 3 /Ca 2ϩ signaling is very important, because the IP 3 -induced Ca 2ϩ release underlying a variety of spatio-temporal Ca 2ϩ dynamics has been shown to considerably affect various cellular functions such as smooth muscle contraction, fertilization, immune response, gene expression, synaptic plasticity, development, and so on (3,4). The molecular mechanisms responsible for the versatility of IP 3 -mediated Ca 2ϩ signaling include (i) a plasma membrane receptor-PLC system that generates the second messenger IP 3 (5-7), (ii) IP 3 Rs (8,9), (iii) Ca 2ϩ sequestration mechanisms (10,11), (iv) IP 3 phosphatases and kinases (12,13), and (v) Ca 2ϩ influx machinery (14). These molecular mechanisms interact with each other, and thus, they all affect each other. For example, intracellular Ca 2ϩ released from Ca 2ϩ stores has a positive feedback effect on PLC activity (15,16). IP 3 R activity is positively or negatively regulated by Ca 2ϩ depending on the intracellular Ca 2ϩ level (17), and Ca 2ϩ entry machinery is regulated by store depletion, conformational changes of IP 3 Rs, or direct binding of phospholipids and DAG (14). Thus, numerous regulatory mechanisms of these signaling molecules and their interactions allow cells to establish precise and complex Ca 2ϩ patterns that contribute to various physiological phenomena.
In this paper, we report a novel regulatory mechanism involved in receptor-IP 3 -Ca 2ϩ signaling, i.e. DAG-mediated positive feedback stimulation of PLC. We unexpectedly found that 1-oleoyl-2-acyl-sn-glycerol (OAG), a membrane-permeable derivative of DAG, induces Ca 2ϩ oscillation in COS-7 cells independently of PKC activity and extracellular Ca 2ϩ . The OAGinduced Ca 2ϩ oscillation was also observed in HeLa cells, CHO-K1 cells, and astrocytes, suggesting that the OAG-induced Ca 2ϩ oscillation is a general phenomenon in various types of cells. The OAG-induced Ca 2ϩ oscillation was dependent on Ca 2ϩ release through IP 3 Rs, because 2-aminoethyl diphenylborinate (2-APB) (18), an IP 3 R inhibitor, or depletion of Ca 2ϩ stores with cyclopiazonic acid (CPA) abolished the Ca 2ϩ mobilization. In addition, we found that treatment of COS-7 cells with a PLC inhibitor or expression of the IP 3 -absorbent protein "IP 3 -sponge" completely abolished the OAG-induced Ca 2ϩ oscillation, indicating that OAG stimulates IP 3 production via PLC activation. We also discovered that accumulation of endogenous DAG as a result of exposure to a DAG-lipase inhibitor increases the sensitivity of the Ca 2ϩ response in COS-7 cells to low-dose ATP stimulation. These findings suggested that DAG produced via the receptor-PLC signaling cascade leads to further PLC activation, resulting in the increased IP 3 production and amplification of receptor-Ca 2ϩ signaling. We propose that, in addition to the well known positive feedback effect of Ca 2ϩ on PLC activity, this novel signal cross-talk * This work was supported by grants from the Ministry of Education, Science, and Culture of Japan (to K. M.) and the Japan Science and Technology Agency. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. □ S The on-line version of this article (available at http://www.jbc.org) contains Supplemental Fig. 1  between DAG and PLC activity may be a crucial mechanism regulating the amount of IP 3 , which is an important factor in determining the threshold of Ca 2ϩ signaling generation and Ca 2ϩ dynamics in response to agonist stimulation, and that it may play an important role in physiological phenomena that are dependent on IP 3 -induced Ca 2ϩ release.

EXPERIMENTAL PROCEDURES
Culture-COS-7 cells, CHO-K1, and HeLa cells were cultured in Dulbecco's modified essential medium (DMEM) (Nakarai Tesque, Kyoto, Japan) supplemented with 10% fetal bovine serum, 50 units/ml penicillin, and 50 g/ml streptomycin (Nakarai). Astrocytes were prepared from the cerebral cortex of postnatal 1 day mice or postnatal 1 day Wistar rats using the standard method (19). Following trypsin treatment, cortices were dissociated by trituration and cultured in 75-mm 2 flask with DMEM containing 10% fetal bovine serum, 50 units/ml penicillin, and 50 g/ml streptomycin. The confluent cells then were treated with trypsin after overnight shaking (280 rpm) and replated on 3.5 cm of poly-L-lysine-coated glass-bottom dishes (Matsunami, Osaka, Japan) at a density of 1 ϫ 10 4 /dish. After 4 -5 days from the replating, Ca 2ϩ imaging was performed. We confirmed that the cultured cells were astrocytes by immunostaining with mouse anti-glial fibrillary acidic protein antibody (Sigma) (data not shown).
Ca 2ϩ Release from Cerebellar Microsomes-Mice cerebellar microsome fractions were prepared as described previously (17). IP 3 -induced Ca 2ϩ release from cerebellar microsomes was measured with Fura-2 and a fluorospectrometer, CAF110 (Jasco, Tokyo, Japan). The membrane fractions were suspended at a concentration of 200 g/ml protein in the buffer containing 1 g/ml oligomycin (Sigma), 2 mM MgCl 2 , 25 g/ml creatine kinase (Roche Diagnostics, Tokyo, Japan), 10 mM creatine phosphate (Sigma), and 2 M Fura-2 (Dojindo) After loading Ca 2ϩ into microsomes by activating Ca 2ϩ -ATPase using 1 mM ATP (Sigma), Fura-2 was alternately excited with 340 and 380 nm and the fluorescence changes at 510 Ϯ 10 nm were detected in response to 100 M OAG and 100 M IP 3 application. At the end of each set of experiments, maximum and minimum values of Fura-2 fluorescence were obtained in the presence of 2 mM CaCl 2 and 10 mM EGTA, respectively (21).
DAG Kinase Assay-COS-7 cells at 70% confluency in 60-mm culture plastic dishes were treated with 50 M RHC80267 or 0.1% Me 2 SO (control) for 10 min in the recording buffer containing 2 mM Ca 2ϩ and then stimulated with 0.3 M ATP for 3 min. Immediately after removal of the buffer, cells were fixed with 1 ml of methanol, scraped, and transferred to a 15-ml plastic tube. The dish was rinsed once with another 1 ml of methanol, which then was added to the previously collected ethanol, and the quantification of DAG level was performed using this solution. DAG kinase assay was performed as reported pre-viously (22,23). After separating DAG using thin-layer chromatography, the thin-layer chromatography plate was subjected to autoradiography and the optical density of radioactive spot corresponding to DAG was quantified using a custom-made software for image analysis (TI-workbench).

A DAG Derivative, OAG, Induces Ca 2ϩ Oscillation in Various
Types of Cells-To explore a novel biological function of DAG in intracellular signaling, we measured Ca 2ϩ signals following the application of a membrane-permeable DAG derivative, OAG, to COS-7 cells. Interestingly, we found that, in the nominal absence of external Ca 2ϩ , OAG application (100 M) to COS-7 cells induces Ca 2ϩ oscillation with a short latency of 1-5 min ( Fig. 1A; also see Supplemental Video 1). The frequency of Ca 2ϩ oscillation varies from cell to cell. Although the nominal Ca 2ϩ -free solution (EGTA-free) actually contains very low concentration of Ca 2ϩ (ϳ1.0 M), the Ca 2ϩ mobilization following OAG application could not have been due to Ca 2ϩ influx through cell surface Ca 2ϩ -permeable channels such as TRPC channels as previously reported (24 -26) because we also detected the OAG-induced Ca 2ϩ mobilization in the extracellular solution containing 0.5 mM EGTA (Fig. 1B). The application of a membrane-impermeable DAG derivative, SAG (100 M), failed to induce Ca 2ϩ oscillation (Fig. 1C). The effect of OAG was dose-dependent. As shown in Fig. 1D, hardly any Ca 2ϩ mobilization was detectable when 10 M OAG was applied. The OAG-induced Ca 2ϩ oscillation was first observed at 30 M OAG, and the frequency of Ca 2ϩ oscillation gradually increased as the OAG concentration was increased to 30, 100, and 300 M (Fig. 1D). The time lag between OAG application and the onset of the Ca 2ϩ response also tended to become shorter as the OAG concentration was increased. The corresponding concentrations of the vehicle, Me 2 SO, did not induce any Ca 2ϩ oscillations (data not shown). These results were unexpected, because several previous studies have reported detecting no Ca 2ϩ mobilization upon OAG application under extracellular-Ca 2ϩ -free conditions (26 -28). Because we detected similar Ca 2ϩ oscillations after exposure to OAG purchased from three different companies (see "Experimental Procedures"), the OAG-induced Ca 2ϩ response is unlikely to have been an artifact.
To determine whether OAG would induce Ca 2ϩ oscillation in other types of cells, we applied OAG to CHO-K1 cells, HeLa cells, and rat primary astrocytes and examined them for a Ca 2ϩ response. As shown in Fig. 2, in the nominal absence of external Ca 2ϩ , we observed OAG-induced Ca 2ϩ mobilization in these cells, the same as in COS-7 cells, suggesting that the OAGinduced Ca 2ϩ oscillation is a general phenomenon rather than specific to COS-7 cells.
Ca 2ϩ Release through IP 3 Rs Underlies OAG-induced Ca 2ϩ Oscillation-We next evaluated the contribution of Ca 2ϩ release from Ca 2ϩ stores to the OAG-induced Ca 2ϩ oscillation in COS-7 cells. When the Ca 2ϩ store had been depleted with a Ca 2ϩ pump inhibitor, CPA (10 M), before OAG application, OAG no longer induced Ca 2ϩ oscillation (Fig. 3A). Moreover, the application of 2-APB, an inhibitor of IP 3 Rs, completely abolished the OAG-induced Ca 2ϩ oscillation (Fig. 3B), suggesting that the OAG-induced Ca 2ϩ oscillation was caused by Ca 2ϩ release through IP 3 Rs. To determine whether OAG activates IP 3 Rs directly, we next performed an in vitro Ca 2ϩ release assay using mice cerebellar microsomes. As shown in Fig. 3C, no Ca 2ϩ release from cerebellar microsomes was detected in response to 100 M OAG or the solvent, Me 2 SO, whereas the application of IP 3 (100 M) to cerebellar microsomes induced transient Ca 2ϩ release. In addition, OAG did not enhance the amount of Ca 2ϩ release induced by IP 3 (Fig. 3C). Therefore, the OAG-induced Ca 2ϩ oscillation is likely to be mediated by IP 3 production rather than by direct activation of IP 3 Rs.
To confirm the possible involvement of IP 3 production in the OAG-induced Ca 2ϩ oscillation, we exposed the cells to a PLC inhibitor, U73122, prior to OAG application. As shown in Fig.  4, 10 M U73122 completely abolished the OAG-induced Ca 2ϩ oscillation in COS-7 cells (Fig. 4Aa), whereas the biologically inactive analog U73343 (10 M) had no inhibitory effect (Fig.  4Ab). Furthermore, when we transiently co-expressed IP 3 absorbent protein IP 3 -sponge (20) and GFP in COS-7 cells before OAG application, no Ca 2ϩ oscillation was detected in GFPpositive IP 3 -sponge-expressing COS-7 cells, whereas control cells that did not express IP 3 -sponge (GFP-negative cells) showed Ca 2ϩ oscillation (Fig. 4B). Expression of GFP alone did not affect the OAG-induced Ca 2ϩ oscillation (data not shown). Thus, these results strongly suggested that OAG and thus also DAG are capable of inducing production of IP 3 via PLC activation, which initiates Ca 2ϩ release through IP 3 Rs.
Analysis of the Molecular Mechanism of the OAG-induced Ca 2ϩ Oscillation-To gain further insight into the molecular mechanisms linking DAG to PLC activation, we analyzed the involvement of PKC activity in the OAG-induced Ca 2ϩ oscillation. As shown in Fig. 5, the OAG-induced Ca 2ϩ oscillation occurred independently of PKC activation, because acute application of PMA, a potent activator of PKC (1.0 M), to COS-7 cells did not induce any Ca 2ϩ oscillation (Fig. 5A) and pretreatment of cells with a broad serine-threonine kinase inhibitor, staurosporine (100 nM), did not abolish the Ca 2ϩ mobilization following OAG application (Fig. 5B). Next, because cellular DAG is mainly metabolized by DAG lipase, we treated COS-7 cells with the DAG-lipase inhibitor RHC80267 (25 M) before OAG application. As shown in Fig. 5C, RHC80267 did not affect the OAG-induced oscillation, indicating that it was not caused by metabolites of OAG. Interestingly, we found that a Src family tyrosine kinase inhibitor, PP2, completely abolished the OAG-induced Ca 2ϩ oscillation, whereas the inactive PP2 ana-log PP3 used as a control failed to abolish the OAG-induced Ca 2ϩ oscillation (Fig. 5D).
Endogenous DAG Modifies the Ca 2ϩ Response to ATP Stimulation in COS-7 Cells-Finally, we tried evaluating the contribution to the intracellular Ca 2ϩ mobilization in COS-7 cells of DAG endogenously produced in response to ATP stimulation. We hypothesized that if metabolization of endogenously produced DAG was inhibited by DAG-lipase inhibitor, accumulated DAG would cause further production of IP 3 , resulting in the change in Ca 2ϩ mobilization in response to agonist stimulation. The use of the DAG-lipase inhibitor, RHC80267, has been a common method to examine the effect of endogenous DAG on several physiological phenomena (24,29), and we could detect the increased endogenous DAG level by RHC80267 treatment upon ATP stimulation using DAG kinase assay (Supplemental Fig. 1). To rule out the possible effect of DAGdependent PKC activity on Ca 2ϩ mobilization, we pretreated COS-7 cells with 300 nM PMA for 36 h to down-regulate PKC activity (30) and imaged Ca 2ϩ signals evoked by application of various concentrations of ATP (0.05, 0.1, 0.3, and 0.5 M) in the presence and absence of DAG-lipase inhibitor RHC80267 (50 M). As shown in Fig. 6, the results showed that, whereas none of the control cells released Ca 2ϩ in response to 0.05 M ATP stimulation (0%), a small population of RHC80267-treated cells did release Ca 2ϩ from Ca 2ϩ stores in response to 0.05 M ATP stimulation (0.54 Ϯ 0.59%, mean Ϯ S.D.). In addition, the numbers of RHC80267-treated cells that responded to 0.1, 0. nificant difference in peak Ca 2ϩ amplitudes between the RHC80267-treated cells and the control cells. When stimulated with a higher dose of ATP (3.0 M), almost all of the control and RHC80267-treated cells released Ca 2ϩ and there were no differences between them in the percentage of cells that responded or the amplitudes of the Ca 2ϩ release (data not shown). Thus, these results suggested that a higher level of DAG endogenously produced in response to ATP stimulation could cause further IP 3 production via PLC activation, resulting in the enhanced Ca 2ϩ mobilization from Ca 2ϩ stores through IP 3 Rs in response to low-dose agonist stimulation. DISCUSSION DAG is well known as a potent PKC activator; however, other physiological actions of DAG have recently been reported including activation of RasGRP (31) and some TRPC channels (24,25) and inhibition of cyclic nucleotide-gated channels (32,33). We propose a new biological function of DAG, i.e. DAGinduced IP 3 production via PLC activation. The results of the present study demonstrated that a DAG derivative, OAG, induces Ca 2ϩ oscillation in various types of cells, and based on the following findings, we concluded that the OAG-induced Ca 2ϩ oscillation was attributable to IP 3 -mediated Ca 2ϩ release through IP 3 Rs as follows: 1) the independence of extracellular Ca 2ϩ ; 2) abolition of the oscillation by store depletion; 3) blockade by IP 3 -mediated signaling inhibitors, 2-APB and IP 3sponge; and 4) blockade by PLC inhibition. These findings indicate the existence of a positive feedback effect of DAG on PLC activity, and consistent with this finding, we have demonstrated that the DAG-mediated positive feedback signaling to PLC contributes to the amplification of agonist-induced Ca 2ϩ signaling, suggesting a physiological significance of the feedback mechanism.
The amount of IP 3 produced by OAG could be inferred from the pattern of Ca 2ϩ signaling, because the pattern is generally dependent on the concentration of the agonists applied to cells. In many types of cells, application of low concentrations of agonists, which result in the production of small amounts of IP 3 , induces a transient increase in cytosolic Ca 2ϩ lasting only a few seconds (Ca 2ϩ spike) or repetitive Ca 2ϩ spikes (Ca 2ϩ oscillation), whereas high concentrations of agonists induce sustained Ca 2ϩ elevation throughout the period of agonist application. Because at the highest concentration we tested (300 M) OAG induced only an oscillatory and not a sustained Ca 2ϩ elevation in COS-7 cells (Fig. 1D), the amount of IP 3 produced by DAG-mediated signaling must have been relatively small. This conclusion is supported by the observation that even a relatively low level of expression of IP 3 -sponge completely abolished the OAG-induced oscillation (Fig. 4B), because suppression of Ca 2ϩ mobilization by IP 3 -sponge has been shown to be dependent on both its expression level and the dose of the agonist (20). However, even the small amount of IP 3 derived from the DAG-mediated signaling would be highly effective in shaping IP 3 -mediated Ca 2ϩ signaling when agonist stimulation is weak. Consistent with this notion, we demonstrated that accumulation of intracellular DAG with an DAG-lipase inhibitor greatly increased the number of cells that responded to low doses (0.05 and 0.1 M) of ATP (Fig. 5). Therefore, we concluded that DAG-induced IP 3 production is an important molecular mechanism in the regulation of agonist-induced Ca 2ϩ signaling in addition to numerous other molecular mechanisms (e.g. modification of IP 3 R activity by several proteins and modification of signaling from surface receptors to PLC).
Grimaldi et al. (28) recently reported the finding that OAG application to glial cells induced slow Ca 2ϩ oscillation that was dependent on the extracellular Ca 2ϩ concentration. They showed that the OAG-induced Ca 2ϩ oscillation was caused by Ca 2ϩ entry through TRPC3 channels and that Ca 2ϩ release from internal stores was not involved in the Ca 2ϩ oscillation. In the presence of external Ca 2ϩ , we observed similar slow Ca 2ϩ oscillations in glial cells that were less frequent than in the absence of external Ca 2ϩ (data not shown); however, the percentage of the glial cells exhibiting low-frequency Ca 2ϩ oscillation in response to OAG in the presence of extracellular Ca 2ϩ was even smaller than reported by Grimaldi et al. (28). In addition, they did not detect the OAG-induced Ca 2ϩ mobilization observed in the present study in the absence of external Ca 2ϩ . Although we do not know the exact reasons for these discrepancies, differences among groups in cell culture protocols and conditions and thus in stages of the differentiation of the glial cells may have been responsible.
Although we found that DAG stimulates PLC activity, it was previously shown that PKC, which is activated by DAG, inhibits PLC activity via phosphorylation of PLC (34). These two effects of DAG on PLC activity seem inconsistent. Because PKC phosphorylates and inhibits only the PLC␤ isoform (34), DAG may stimulate other isoforms of PLC and enhance IP 3 -mediated Ca 2ϩ signaling. Alternatively, the above two actions of DAG may affect different phases of the Ca 2ϩ signaling induced by IP 3 -mobilizing agonists. Because PKC activation requires cytosolic Ca 2ϩ elevation, PLC activity by PKC-mediated phosphorylation may become suppressed after cytosolic Ca 2ϩ is sufficiently increased and the suppression may therefore facilitate the termination of individual Ca 2ϩ spikes and waves. By contrast, DAG together with the Ca 2ϩ released from Ca 2ϩ stores may stimulate PLC activity earlier than the PKC-mediated suppression and contribute to the formation of the rising phases of Ca 2ϩ signaling. Complex DAG actions and cytosolic Ca 2ϩ concentration changes may underlie the spatio-temporally organized Ca 2ϩ dynamics induced by IP 3 -mobilizing agonists.
The molecular mechanisms by which DAG activates PLC remain to be determined. First, what is the molecular target of DAG? Blockade of OAG-induced Ca 2ϩ oscillation by PP2 im-   recently been reported to be activated by Ras (37). In view of the fact that the activity of RasGRP, the regulator of Ras, is increased by DAG (31), PLC⑀ may be one of the candidates for the molecule responsible for the DAG-induced IP 3 production. However, since our data suggested possible involvement of Src family PTK activity in the DAG-mediated PLC activation, how Src family PTK activity is involved in the signaling is unknown. Another candidate is PLC␥, because PLC␥ can be activated by Src family PTK-mediated phosphorylation (38 -40). Because Src family PTKs are also involved in the activation of receptor-PLC signaling through G-protein tyrosine phosphorylation (41), the phosphorylation of such signaling proteins by Src family PTKs may underlie the molecular mechanism of the DAG-induced PLC activation.
In conclusion, we propose a novel biological function of DAG as an inducer of IP 3 production through PLC activation. The novel signal cross-talk between DAG and the IP 3 -mediated signal pathway amplifies agonist-induced Ca 2ϩ signaling in response to weak extracellular stimuli. Because IP 3 R-mediated Ca 2ϩ release plays an important role in various physiological functions including synaptic plasticity, gene expression, proliferation, and development, the increased IP 3 production via DAG-mediated signaling may affect numerous biological phenomena.