P2Y2 Nucleotide Receptors Enhance α-Secretase-dependent Amyloid Precursor Protein Processing*

The amyloid precursor protein (APP) is proteolytically processed by β- and γ-secretases to release amyloid β, the main component in senile plaques found in the brains of patients with Alzheimer disease. Alternatively, APP can be cleaved within the amyloid β domain by α-secretase releasing the non-amyloidogenic product sAPPα, which has been shown to have neuroprotective properties. Several G protein-coupled receptors are known to activate α-secretase-dependent processing of APP; however, the role of G protein-coupled nucleotide receptors in APP processing has not been investigated. Here it is demonstrated that activation of the G protein-coupled P2Y2 receptor (P2Y2R) subtype expressed in human 1321N1 astrocytoma cells enhanced the release of sAPPα in a time- and dose-dependent manner. P2Y2 R-mediated sAPPα release was dependent on extracellular calcium but was not affected by 1,2-bis(2-aminophenoxy)ethane-N,N,N,-trimethylammonium salt, an intracellular calcium chelator, indicating that P2Y2 R-stimulated intracellular calcium mobilization was not involved. Inhibition of protein kinase C (PKC) with GF109203 or by PKC down-regulation with phorbol ester pre-treatment had no effect on UTP-stimulated sAPPα release, indicating a PKC-independent mechanism. U0126, an inhibitor of the mitogen-activated protein kinase pathway, partially inhibited sAPPα release by UTP, whereas inhibitors of Src-dependent epidermal growth factor receptor transactivation by P2Y2 Rs had no effect. The metalloprotease inhibitors phenanthroline and TAPI-2 and the furin inhibitor decanoyl-Arg-Val-Lys-Arg-chloromethylketone also diminished UTP-induced sAPPα release. Furthermore, small interfering RNA silencing of an endogenous adamalysin, ADAM10 or ADAM17/TACE, partially suppressed P2Y2R-activated sAPPα release, whereas treatment of cells with both ADAM10 and ADAM17/TACE small interfering RNAs completely abolished UTP-activated sAPPα release. These results may contribute to an understanding of the non-amyloidogenic processing of APP.

P2 nucleotide receptors modulate a wide range of physiological responses following their activation by extracellular nucle-otides (1,2). The G protein-coupled P2Y 2 receptor (P2Y 2 R) 1 subtype is fully activated by equivalent concentrations of ATP or UTP (3)(4)(5) and is up-regulated in salivary gland models of stress and disease (6 -8) as well as in blood vessels after balloon angioplasty and in collared carotid arteries, where it induces intimal hyperplasia and inflammation by increasing smooth muscle cell proliferation and leukocyte infiltration (9,10). Moreover, nucleotides are released from damaged cells of all tissues and from excited neurons, aggregating platelets, and contracting smooth muscle under physiological conditions (2,11).
The diversity of cellular responses mediated by P2Y 2 Rs is due in part to unique structural features that enable these receptors to stimulate a variety of signal transduction pathways. In addition to the classical stimulation of G␣ q -dependent phospholipase C (12,13), the P2Y 2 R contains two Src homology 3 binding sites (PXXP motifs) in its intracellular C terminus that interact directly with Src to transactivate epidermal growth factor (EGF), platelet-derived growth factor, and vascular epidermal growth factor receptor 2 receptors (14,15). The P2Y 2 R also contains an Arg-Gly-Asp sequence in the first extracellular loop, which has been shown to interact with ␣ v ␤ 3 /␤ 5 integrins (16) that regulate cell chemokinesis and chemotaxis (17)(18)(19), responses associated with cell-mediated inflammation (20). Activation of the P2Y 2 R on endothelial cells and astrocytes also up-regulates the expression of cell surface adhesion molecules that play a role in monocyte-mediated inflammation (21) and reactive astrogliosis, 2 respectively. The present study describes a novel function for the P2Y 2 R in enhancing the ␣-secretase-dependent cleavage of the amyloid precursor protein (APP) to generate a neuroprotective peptide rather than neurodegenerative A␤, a finding that has relevance to Alzheimer disease.
APP is a transmembrane glycoprotein that is present in a variety of tissues but predominantly in the brain (22). APP contains an extracellular N terminus and a short C-terminal region that lies in the cytoplasm. Within APP, a single membrane-spanning region of 39 -42 amino acids represents the amyloidogenic A␤ peptide, the major component of plaques found in Alzheimer patients (23,24). Proteolytic cleavage of APP in vivo can occur at the amino terminus of the A␤ domain (␤-cleavage), within the A␤ domain (␣-cleavage), and at the C terminus of the A␤ domain (␥-cleavage); for review see Mills and Reiner (25). Cleavage of the APP protein by both ␤-secretase and ␥-secretase gives rise to the amyloidogenic A␤ fragment. APP cleavage by ␣-secretase generates sAPP␣, a soluble, non-amyloidogenic N-terminal fragment (ϳ100 -140 kDa) that is released into the extracellular medium, whereas the membrane-retained fragment undergoes further cleavage and endocytotic processing (26 -28). The sAPP␣ fragment has been shown to have both neurotrophic (29) and neuroprotective (30 -33) activities, suggesting that the presence of sAPP␣ in the external milieu may promote neuronal viability.
Stimulation of several G protein-coupled receptors (GPCRs) has been reported to induce sAPP␣ release through the activation of protein kinase C (PKC)-dependent and -independent pathways (34 -36). Initial studies conducted in HEK293 cells showed that overexpression of the human M1 and M3 muscarinic receptors stimulated sAPP␣ secretion (37). Subsequently, G protein-coupled glutamate, serotonin (5-HT), thrombin, and bradykinin receptors have been shown to regulate sAPP␣ release (38 -43). Furthermore, receptor-mediated activation of sAPP␣ release is associated with a reduction in A␤ formation (44 -46). Although the molecular identities of the secretase(s) stimulated by GPCRs have not been fully elucidated, both constitutive and receptor-regulated sAPP␣ release have been linked to two members of the ADAM (a disintegrin and metalloprotease) family, namely ADAM 10, which is the Kuz enzyme (47), and ADAM17/TACE (tumor necrosis factor-␣ converting enzyme), which is the protease responsible for releasing tumor necrosis factor-␣ (TNF-␣) from the plasma membrane (48). In the present study we demonstrate that activation of a P2Y 2 nucleotide receptor expressed in the human 1321N1 astrocytoma cells stimulates ADAM10-mediated APP processing and ADAM17/TACE-mediated APP processing, leading to the release of sAPP␣. Because nucleotide release from damaged cells in the central nervous system would be expected to activate P2Y 2 Rs, these findings suggest a novel neuroprotective role for P2Y 2 R-mediated APP processing in neurodegenerative disorders, including Alzheimer disease.
Cell Culture-Human 1321N1 astrocytoma cells stably transfected with human P2Y 2 R cDNA (5) were cultured in Dulbecco's modified Eagle's medium supplemented with 5% (v/v) fetal bovine serum, 100 units/ml penicillin, 100 g/ml streptomycin, and 500 g/ml G418 and maintained at 37°C in a humidified atmosphere of 5% CO 2 and 95% air. Cells transfected with a pLXSN expression vector served as a negative control.
sAPP␣ Release-Cells were plated on 12-well culture dishes and grown until ϳ80 -90% confluence. Cells were washed twice with a modified Krebs-HEPES buffer (15 mM HEPES, pH 7.4, 120 mM NaCl, 4 mM KCl, 1.2 mM MgSO 4 , 1.2 mM KH 2 PO 4 , 1 mM CaCl 2, and 10 mM D-glucose) and then incubated on a rocking platform in a humidified incubator at 37°C for specified times with 200 l of buffer containing compounds as indicated in the Fig. 4 legend. When indicated, cells were pretreated with inhibitors for 30 min prior to the addition of nucleotides. At the end of the incubation period, the medium was collected, centrifuged at 12,000 ϫ g for 1 min to remove cellular debris, and the protein concentration of the supernatant was determined by the Lowry method (49). 80 g of supernatant protein was diluted 1:4 with 5ϫ Laemmli buffer, boiled for 3 min, subjected to electrophoresis on 7.5% SDS-polyacrylamide mini-gels, and transferred to nitrocellulose mem-branes. Membranes were blocked for 1 h with 5% (w/v) nonfat dry milk in TBST (Tris-buffered saline containing 0.1% (v/v) Tween 20) and immunoblotted overnight at 4°C in 3% (w/v) bovine serum albumin with 0.02% (w/v) sodium azide in TBST with a 6E10 monoclonal antibody (1:1000 dilution; Senetek, Maryland Heights, MO) that recognizes residues 1-17 of the A␤ domain of sAPP␣ (50). Membranes were washed three times during a 45-min period with TBST and incubated with peroxidase-linked goat anti-mouse IgG antibody (1:1000 dilution; Santa Cruz Biotechnology, Santa Cruz, CA) at room temperature for 1 h. After three more washes with TBST, the membrane was subjected to chemiluminescence, and the protein bands detected on x-ray film were quantified using a computer-driven scanner and Quantity One software (Bio-Rad). The amount of sAPP␣ generated was expressed as a percentage of untreated controls. All experiments were performed in duplicate and repeated at least three times. Analysis of variance and unpaired Student's t test was used to determine statistical significance (p Ͻ 0.05).
Measurement of the Intracellular Calcium Concentration-The intracellular free calcium concentration ([Ca 2ϩ ] i ) was measured by dual excitation spectrofluorometric analysis of cell suspensions loaded with fura-2 (6) and BAPTA as described previously (51). Cells were assayed in modified Krebs-HEPES buffer containing 0.1% (w/v) bovine serum albumin. The intracellular calcium concentration was calculated by the method of Grynkiewicz et al. (52).
siRNA Targeting of ADAM10 and ADAM17/TACE Genes-Transfection of cells with siRNA duplexes (Integrated DNA Technologies, Coralville, IA) to inhibit endogenous ADAM10 or ADAM17/TACE expression was performed using Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions. Briefly, 0.7 g of each of the three ADAM10 or ADAM17/TACE siRNA duplexes was incubated with cells in serum-free medium for 6 h and then replaced with normal growth media. Sequences of the siRNA duplexes were AUUCGUAGGUUGAA-AUGUCdTdT, UUCCAUUUCCACAAAUAGGdTdT, and AGCCAUUA-CAUAUUCCUUCdTdT (ADAM10) and AGUUUGCUUGGCACACCU-UdTdT, AGUAAGGCCCAGGAGUGUdTdT, and CAUAGAGCCACUU-UGGAGAdTdT (ADAM17/TACE). Cells were assayed for sAPP␣ release 48 h after siRNA transfection, and silencing of the targeted protein expression was confirmed by Western analysis of cell lysates as described above for the detection of sAPP␣ by using goat anti-human ADAM17/TACE antibody (1:1000 dilution; Santa Cruz Biotechnology) or goat anti-human ADAM10 antibody (1:1000 dilution; Sigma). Blots were stripped and reprobed with goat anti-human actin antibody (1: 1000 dilution; Cytoskeleton, Denver, CO) to verify the equivalence of protein loading per lane.

P2Y 2 R Activation Stimulates the Release of sAPP␣-UTP
(100 M) caused a time-dependent increase in sAPP␣ (ϳ120 kDa) release from human 1321N1 astrocytoma cells stably transfected with P2Y 2 R cDNA (1321N1-P2Y 2 cells) that was significantly greater than basal sAPP␣ release (Fig. 1, A and  B). UTP also caused a dose-dependent increase in sAPP␣ from 1321N1-P2Y 2 cells with an EC 50 value of 2.5 M (Fig. 2), which is characteristic of P2Y 2 R activation (5,6,53,54). In contrast, UTP-induced sAPP␣ release was undetectable in untransfected cells or in cells transfected with the expression vector only (data not shown).
UTP-induced sAPP␣ Release Is Independent of PKC Activation-The PKC activator phorbol myristate acetate (PMA; 1 M) also caused a time-dependent release of sAPP␣, similar to that caused by UTP (Fig. 1A). Because P2Y 2 R activation causes the phospholipase C-dependent stimulation of PKC (1), we determined whether P2Y 2 R-mediated sAPP␣ release is dependent on activation of PKC. Therefore, 1321N1-P2Y 2 cells were pre-treated for 30 min with GF109203 (10 M), an inhibitor of PKC, followed by incubation with UTP (100 M) or PMA (1 M) for an additional 2 h. GF109203 completely inhibited PMA-activated release of sAPP␣ but had no effect on UTPinduced release (Fig. 3A). Furthermore, PKC down-regulation by overnight treatment with PMA (1 M) abolished PMA-stimulated sAPP␣ release but had no effect on the UTP-induced sAPP␣ release (Fig. 3B). Thus, UTP-induced sAPP␣ release is independent of P2Y 2 R-mediated activation of PKC.
UTP-induced sAPP␣ Release Is Dependent on Extracellular Calcium but Not P2Y 2 R-mediated Intracellular Calcium Mobilization-Activation of G protein-coupled P2Y 2 Rs with UTP in 1321N1-P2Y 2 cells is known to cause increases in [Ca 2ϩ ] i through the G␣ q -dependent stimulation of phospholipase C, the generation of inositol-1,4,5 triphosphate, and the release of calcium from inositol-1,4,5 triphosphate-sensitive calcium stores (12,13). In addition, P2Y 2 Rs can promote calcium entry through the stimulation of plasma membrane calcium channels (5,13,54,55). Therefore, we examined the role of calcium signaling pathways in UTP-induced sAPP␣ release. In the absence of extracellular calcium, UTP-stimulated sAPP␣ release was completely inhibited (Fig. 4A). In contrast, PMA-stimulated sAPP␣ release was unaffected by the absence of extracellular calcium, suggesting the existence of both calcium entrydependent and -independent pathways for stimulating sAPP␣ release. Unexpectedly, introduction into 1321N1-P2Y 2 cells of the intracellular calcium chelator BAPTA, which suppresses P2Y 2 R-mediated increases in [Ca 2ϩ ] i (Fig. 4B), had no effect on sAPP␣ release induced by UTP (Fig. 4A). These data suggest that sAPP␣ release due to P2Y 2 R activation is independent of increases in cytosolic calcium. Consistent with this conclusion, neither ionomycin, a calcium ionophore, nor thapsigargin, an inhibitor of the plasma membrane Ca 2ϩ -ATPase, caused an increase in sAPP␣ release (data not shown).
Inhibition of MAPK (ERK1/2) Partially Inhibits UTP-stimulated sAPP␣ Release-Previous studies in our laboratory have shown that UTP activates ERK1/2 phosphorylation in 1321N1-P2Y 2 cells, in part via the Src-dependent transactivation of the EGF and platelet-derived growth factor receptors (14), consistent with Src-dependence of ERK1/2 phosphorylation by other GPCRs (56,57). Therefore, we investigated the role of the MAPK/EGF receptor (EGFR) pathway in UTP-stimulated sAPP␣ release. U0126, an inhibitor of MAPK/ERK kinase (MEK) whose substrates are ERK1/2, partially inhibited (ϳ40%) sAPP␣ release induced by UTP but not by PMA (Fig. 5). Incubation of cells with the Src inhibitor pyrazole pyrimidinetype 2 (10 M) or the EGFR kinase inhibitor AG1478 (10 M) had no effect on the UTP-stimulated or PMA-stimulated sAPP␣ release (data not shown). These results suggest that UTPstimulated sAPP␣ release in 1321N1-P2Y 2 cells is partially dependent on phosphorylation of ERK1/2 but independent of Src and EGFR activation.
Inhibitors of Metalloproteases and Furin, the Proprotein Convertase, Decrease UTP-induced sAPP␣ Release-Disintegrin metalloproteases, including ADAM10 and ADAM17/TACE, catalyze the shedding of the ectodomain of APP and other transmembrane proteins (58). We determined that P2Y 2 R-mediated sAPP␣ release was inhibited by phenanthroline, a broad range metalloprotease inhibitor, and TAPI-2, a selective, competitive inhibitor of ADAM17/TACE (Fig. 6). PMA-stimulated sAPP␣ release was also inhibited by phenanthroline and TAPI-2 (Fig. 6A). TAPI-2 inhibited UTP-stimulated sAPP␣ release with an estimated IC 50 of 1.8 M (Fig. 6B), comparable with a value reported previously for the inhibition of ADAM17/ TACE (59,60). These results confirm that APP processing due to P2Y 2 R activation requires metalloprotease activity and that ADAM17/TACE is likely involved. However, micromolar concentrations of TAPI-2 have been reported to inhibit other members of the adamalysin family (61), which may contribute to P2Y 2 R-mediated sAPP␣ release.
ADAM10 and ADAM17/TACE require proteolytic processing to become active (62). The proprotein convertase furin has been linked to the proteolytic cleavage of adamalysins (63,64), and, therefore, we tested the ability of the furin inhibitor, CMK, to suppress UTP-induced sAPP␣ release. As shown in Fig. 7, both basal and UTP-stimulated sAPP␣ release were completely inhibited by CMK, suggesting that both constitutive and receptor-mediated sAPP␣ release via metalloproteases in 1321N1-P2Y 2 cells are dependent on furin or a furin-like protease.
ADAM10 and ADAM17/TACE Are Involved in UTP-stimulated sAPP␣ Release-To unambiguously determine whether ADAM10 and/or ADAM17/TACE were the metalloproteases responsible for P2Y 2 R-mediated sAPP␣ release, 1321N1-P2Y 2 cells were transfected with ADAM10 and/or ADAM17/TACE siRNAs to inhibit expression of the endogenous adamalysins. Inhibition of either ADAM10 or ADAM17/TACE expression with the corresponding siRNAs partially inhibited basal, UTPstimulated, and PMA-stimulated sAPP␣ release (Fig. 8A), whereas the simultaneous inhibition of both ADAM10 and ADAM17/TACE expression virtually prevented sAPP␣ release induced by either UTP or PMA (Fig. 8A). The extent of suppression of ADAM10 and ADAM17/TACE protein expression by the siRNAs is shown in Fig. 8B. Based on these data, we conclude that ADAM10 and ADAM17/TACE are the primary metalloproteases responsible for P2Y 2 R-mediated APP processing in 1321N1-P2Y 2 cells. DISCUSSION In this report, we demonstrate that activation of the G protein-coupled P2Y 2 nucleotide receptor subtype stimulates the ␣-secretase-dependent processing of APP. Utilizing the human 1321N1 astrocytoma cell line that lacks endogenous P2 receptor expression, we demonstrate that stable transfection with P2Y 2 R cDNA promoted UTP-induced release of sAPP␣ in a time-and dose-dependent manner, similar to sAPP␣ release mediated by muscarinic, glutamate, bradykinin, and serotonergic receptors (37)(38)(39)(40)(41)(42).
G protein-coupled receptors have been shown to stimulate ␣-secretase-dependent APP processing via PKC-dependent and PKC-independent pathways (34 -36, 65). Nonetheless, stimulation of either pathway can lead to activation of the same proteolytic enzyme (65). Here, we show that P2Y 2 R-mediated sAPP␣ release occurred independently of PKC activation be- cause the PKC inhibitor GF109203 (Fig. 3A) or PKC downregulation (Fig. 3B) had no effect on UTP-induced APP processing, although direct activation of PKC with the phorbol ester PMA increased sAPP␣ release and was inhibited by GF109203 or PKC down-regulation (Fig. 3), in agreement with previous reports (66,67). Although P2Y 2 Rs in 1321N1-P2Y 2 cells are known to activate PKC (1) these data indicate that activation of PKC is not necessary for P2Y 2 R-mediated increases in APP processing, suggesting the involvement of other cellular signaling pathways coupled to the P2Y 2 R in UTP-induced sAPP␣ release.
It is well known that activation of the P2Y 2 R causes an increase in [Ca 2ϩ ] i through both the PLC-dependent release of calcium from intracellular stores and the entry of extracellular calcium through ion-activated calcium channels (1,12). Several physiological responses resulting from P2Y 2 R activation are dependent on the rise of [Ca 2ϩ ] i (e.g. ion secretion, fluid secretion, and vascular cell adhesion molecule-1 expression) (21,54,68); however, our results indicate that P2Y 2 R-mediated increases in [Ca 2ϩ ] i by either calcium released from intracellular stores or calcium influx were not required for sAPP␣ release, because chelation of intracellular calcium with BAPTA had no effect on UTP-induced sAPP␣ release (Fig. 4A). Furthermore, incubation with ionomycin or thapsigargin, agents that increase [Ca 2ϩ ] i, did not promote sAPP␣ release (data not shown), confirming that APP processing in 1321N1-P2Y 2 cells is not a result of an increase of cytosolic calcium. However, the absence of extracellular calcium does inhibit UTP-induced but not PMA-induced sAPP␣ release (Fig. 4A), suggesting the presence of an extracellular calcium-dependent protein that regulates P2Y 2 R-mediated ␣-secretase activity. Because PKC-regulated ␣-secretase is found to be localized to the trans-Golgi network and not at the cell surface (69), it is possible that the PKC-independent ␣-secretase expressed in 1321N1-P2Y 2 cells is localized to the cell surface where its activity could be modulated by the extracellular calcium concentration. Nonetheless, the role of extracellular calcium in P2Y 2 R-mediated sAPP␣ release requires further investigation.
Our data indicate that UTP-stimulated sAPP␣ release in 1321N1-P2Y 2 cells is partially dependent on the activation of ERK1/2 (Fig. 5). Activation of GPCRs, including the P2Y 2 R, has been shown to stimulate ERK1/2 activity either directly through G protein signaling or by transactivation of EGFR via P2Y 2 R interaction with Src (14,70). Because EGFR activation has been reported to promote sAPP␣ release (71), we investigated whether Src-dependent transactivation of the EGFR was involved in UTP-induced sAPP␣ release. However, inhibition of EGFR tyrosine phosphorylation by AG1478 or inhibition of Src by the pyrazole pyrimidine-type 2 inhibitor did not affect UTPinduced sAPP␣ release (data not shown), suggesting that Srcmediated EGFR transactivation is not involved. Similar to the P2Y 2 R, sAPP␣ release due to activation of M1 (72) or M3 muscarinic receptors (66,73) occurs primarily by MAPK-and PKC-independent pathways. These data suggest that P2Y 2 and muscarinic receptor-mediated stimulation of sAPP␣ release are coupled to an as yet unidentified signaling pathway, which highlights the mechanistic diversity of GPCR-mediated APP processing.
In recent years, several studies have identified and characterized secretases responsible for APP processing (58, 74 -77). These secretases belong to the furin-dependent metalloprotease/disintegrin family of proteins referred to as ADAMs, proteolytic enzymes that catalyze ectodomain shedding of proteins to release an active peptide into the extracellular milieu. The TNF-␣-converting enzyme TACE/ADAM17 was originally described as the main protease responsible for TNF-␣ release from membranes (48,78) and was proposed to be the ␣-secretase responsible for APP processing (79). Recently, ADAM9 and ADAM10, two metalloprotease family members among the 30 cloned, have also been proposed to be ␣-secretases responsible for processing APP (77). Our data demonstrate that inhibition of the expression of ADAM10 or ADAM17/TACE with siRNAs results in a decrease in both UTP-stimulated sAPP␣ release and PMA-stimulated sAPP␣ release, whereas co-suppression of ADAM10 and ADAM17/TACE completely prevents UTP-or PMA-stimulated sAPP␣ release (Fig. 8A), implicating ADAM10 and ADAM17/TACE as the primary ␣-secretases involved. Furthermore, the inhibition of sAPP␣ release by the furin inhibitor CMK (Fig. 7) supports the conclusion that furin-dependent metalloproteases/disintegrins are responsible for ␣-secretasedependent APP processing in 1321N1-P2Y 2 cells.
In conclusion, this study provides the first evidence for the regulation of APP processing by any P2 nucleotide receptor subtype. Our results indicate that P2Y 2 R-mediated APP processing is regulated by extracellular calcium, but not by increases in [Ca 2ϩ ] i , and is partially dependent on ERK1/2, but not PKC. Furthermore, we have demonstrated that P2Y 2 Rmediated sAPP␣ release in 1321N1-P2Y 2 cells is largely due to the activities of the furin-dependent secretases ADAM10 and ADAM17/TACE. Proteolytic processing of APP by ␣-secretase occurs within the sequence of A␤, thus precluding the formation of the amyloidogenic and neurodegenerative A␤ fragment (44 -46). Additionally, the ␣-secretase product sAPP␣ has both neuroprotective and mitogenic effects (29 -33). Therefore, activation of P2Y 2 Rs in the brain may promote neuronal viability.