Phosphorylation of the β-Amyloid Precursor Protein at the Cell Surface by Ectocasein Kinases 1 and 2*

The β-amyloid precursor protein (βAPP) is one of the rare proteins known to be phosphorylated within its ectodomain. We have shown previously that βAPP can be phosphorylated within secretory vesicles and at the cell surface (Walter, J., Capell, A., Hung, A. Y., Langen, H., Schnölzer, M., Thinakaran, G., Sisodia, S. S., Selkoe, D. J., and Haass, C. (1997) J. Biol. Chem.272, 1896–1903). We have now specifically characterized the phosphorylation of cell surface-located βAPP and identified two ectoprotein kinases that phosphorylate βAPP at the outer face of the plasma membrane. By using selective protein kinase inhibitors and by investigating the usage of ATP and GTP as cosubstrates, we demonstrate that membrane-bound βAPP as well as secreted forms of βAPP can be phosphorylated by casein kinase (CK) 1- and CK2-like ectoprotein kinases. The ectodomain of βAPP was also phosphorylated by purified CK1 and CK2 in vitro, but not by protein kinases A and C. Phosphorylation of βAPP by ectoprotein kinases and by purified CK1 and CK2 occurred within an acidic domain in the N-terminal half of the protein. Heparin strongly inhibited the phosphorylation of cell-surface βAPP by ecto-CK1 and ecto-CK2, indicating a regulatory role of this extracellular matrix component in βAPP phosphorylation.

The ␤-amyloid precursor protein (␤APP) is one of the rare proteins known to be phosphorylated within its ectodomain. We have shown previously that ␤APP can be phosphorylated within secretory vesicles and at the cell surface ( We have now specifically characterized the phosphorylation of cell surface-located ␤APP and identified two ectoprotein kinases that phosphorylate ␤APP at the outer face of the plasma membrane. By using selective protein kinase inhibitors and by investigating the usage of ATP and GTP as cosubstrates, we demonstrate that membrane-bound ␤APP as well as secreted forms of ␤APP can be phosphorylated by casein kinase (CK) 1and CK2-like ectoprotein kinases. The ectodomain of ␤APP was also phosphorylated by purified CK1 and CK2 in vitro, but not by protein kinases A and C. Phosphorylation of ␤APP by ectoprotein kinases and by purified CK1 and CK2 occurred within an acidic domain in the N-terminal half of the protein. Heparin strongly inhibited the phosphorylation of cell-surface ␤APP by ecto-CK1 and ecto-CK2, indicating a regulatory role of this extracellular matrix component in ␤APP phosphorylation.
Alzheimer's disease is the most common form of dementia and is pathologically characterized by the invariant accumulation of senile plaques and neurofibrillary tangles in brains of Alzheimer's patients (1,2). A major constituent of senile plaques is the amyloid ␤-peptide (A␤) 1 (3,4), which derives from the larger ␤-amyloid precursor protein (␤APP) by proteolytic processing (5). ␤APP is a type I membrane protein with a large N-terminal ectodomain, a single transmembrane domain, and a C-terminal cytoplasmic tail (6). Two major proteolytic processing pathways for ␤APP have been identified. ␤APP can be cleaved by ␣-secretase within the A␤ domain, resulting in the release of a truncated soluble form of ␤APP (APP S -␣) (7)(8)(9). Alternatively, ␤APP can be cleaved by ␤-secretase at the N terminus of the A␤ domain, which results in the generation of APP S -␤ and a membrane-bound C-terminal fragment bearing the complete A␤ domain (10 -12). Subsequent cleavage of this fragment at the C-terminal end of the A␤ domain results in the release of A␤ (13)(14)(15). Cell biological studies revealed that ␣-secretory processing can occur at the cell surface (9) and in Golgi-derived vesicles during the transport of ␤APP to the plasma membrane (16,17). One cellular mechanism of A␤ production involves re-internalization of full-length ␤APP from the cell surface into endosomes, where the ␤-secretase cleavage appears to occur (10,18). Some of the resulting C-terminal fragments recycle to the cell surface, where they can be cleaved by ␥-secretase (18,19).
Mutations in the ␤APP gene (20) and in the two presenilin genes (21) have been shown to be associated with early onset familial Alzheimer's disease. Most familial Alzheimer's disease-associated mutations analyzed lead to an enhanced production of ␣␤- , indicating a critical role of this elongated form in amyloidogenesis (1,20,21).
The structure of ␤APP resembles a cell-surface receptor (6). However, the physiological functions of ␤APP are not well understood. Work in Drosophila melanogaster suggests a role of ␤APP in cell-cell or cell-matrix interactions (22). Indeed, direct binding of ␤APP to the extracellular matrix component heparin was demonstrated (23,24). Soluble forms of ␤APP promote neurite outgrowth and cell adhesion and are neuroprotective (25,26). In addition, splice variants of ␤APP containing the Kunitz protease inhibitor domain have been shown to be inhibitors of blood coagulation factors IXa and XIa (27,28). Since ␤APP is also released from activated platelets, ␤APP might play a role in homeostasis (29,30).
␤APP is post-translationally modified by NЈ-and OЈ-glycosylation, sulfation, and phosphorylation (7). Besides phosphorylation in its cytoplasmic tail (31,32), ␤APP is predominantly phosphorylated within its ectodomain (33,34). Ectodomain phosphorylation of ␤APP was mapped to serine residues located within an acidic region in the N-terminal half of the protein (34,35). Recently, we found that phosphorylation of the ␤APP ectodomain can occur during transport to the cell surface within secretory vesicles (35). In addition, ␤APP can also be phosphorylated at the cell surface by ectoprotein kinase activities (35).
Ectoprotein kinases act on the outer side of the plasma membrane and use extracellular nucleoside triphosphates as cosubstrates, which can be released by intact cells upon certain stimuli (36,37). Ectoprotein kinases phosphorylate membranebound proteins as well as soluble extracellular proteins. Phosphorylation of cell-surface proteins has been implicated in the regulation of certain cellular functions, including long-term potentiation in neurons (38) and synaptogenesis (39). However, the relevant protein substrates that are phosphorylated by ectoprotein kinases in these processes remain to be identified. Several ectoprotein kinase activities have been characterized, including cAMP-dependent protein kinase-like (40) and casein kinase (CK)-like (41)(42)(43) enzymes. Evidence of protein kinase C (44,45) and tyrosine kinase (46) activities at the cell surface has also been demonstrated.
By analyzing the cell-surface phosphorylation of ␤APP, we have now identified CK1-and CK2-type ectoprotein kinases (ecto-CK1 and ecto-CK2), which phosphorylate membranebound as well as soluble ␤APP. Phosphorylation of ␤APP is significantly inhibited by heparin, indicating a regulatory role of extracellular matrix components in ␤APP phosphorylation.

MATERIALS AND METHODS
Cell Culture-Human embryonic kidney (HEK) 293 cells were cultured in Dulbecco's modified Eagle's medium with Glutamax (Life Technologies, Inc.) supplemented with 10% fetal calf serum (Life Technologies, Inc. ). The cell lines stably overexpressing ␤APP-(1-695) and a truncated form of ␤APP ending at the ␣-secretase cleavage site have been described previously (34,35). Cell lines were maintained in medium containing 200 g/ml G418 (BIOMOL Research Labs Inc.).
Immunoprecipitation and Antibodies-Immunoprecipitations were carried out as described previously (48). Antibodies 6687 and 5313 were raised against fusion proteins of maltose-binding protein and the C terminus of ␤APP (amino acids 652-695) or of maltose-binding protein and the N-terminal domain of ␤APP (amino acids 444 -592).
Phosphorylation of Cell-surface Proteins-Cell-surface phosphorylation was carried out as described (47). Subconfluent monolayer cell cultures grown on Dulbecco's modified Eagle's medium containing 10% fetal calf serum were washed twice with prewarmed (37°C) isotonic phosphorylation buffer (30 mM Tris (pH 7.3), 70 mM sodium chloride, 5 mM magnesium acetate, 0.5 mM EDTA, and 5 mM KH 2 PO 4 /K 2 HPO 4 ; 290 Ϯ 10 mosM) and incubated for 10 min at 37°C in the same buffer. Phosphorylation was started by the addition of 1.5 M [␥-32 P]ATP or [␥-32 P]GTP (Amersham Pharmacia Biotech) and was allowed to proceed for 20 min at 37°C. To characterize the protein kinases that phosphorylate ␤APP, reactions were carried out in the presence or absence of the following selective kinase inhibitors: 20 M 5,6-dichloro-1-␤-D-ribofuranosylbenzimidazole (DRB; BIOMOL Research Labs Inc.), an inhibitor of CK2; 1 M CKI-7 (Medac), an inhibitor of CK1; 1 M hymenialdisine (HD; gift from L. Meijer), an inhibitor of CK1; 1 M H-89 (Calbiochem), an inhibitor of cAMP-dependent kinase; 10 M roscovitine (Calbiochem), an inhibitor of cyclin-dependent kinases; 5 M KN-62 (Calbiochem), an inhibitor of Ca 2ϩ /calmodulin-dependent kinase II; and 1 M GF109203X (Calbiochem), an inhibitor of protein kinase C. Heparin (Sigma) was added to the cell supernatants at the concentrations indicated. The phosphorylation reactions were terminated by removing cell supernatants, followed by two immediate washes of the cells with ice-cold phosphorylation buffer containing 2 mM unlabeled ATP. Subsequently, cells were lysed in the presence of 2 mM ATP for 7 min on ice as described (48). Lysates and cell supernatants were clarified by centrifugation (10 min at 14,000 ϫ g). Membrane-bound and secreted ␤APPs were isolated by immunoprecipitation as described above and separated by SDS-polyacrylamide gel electrophoresis (PAGE). Radiolabeled proteins were detected by autoradiography or by phosphoimaging.
Cell viability during phosphorylation assays was evaluated by the criteria of Kü bler et al. (47).
In Vitro Phosphorylation Assays-Recombinant rat CK1␦ (New England Biolabs Inc.), the recombinant ␣-subunit of human CK2 (gift from Dr. W. Pyerin), and the catalytic subunit of protein kinase A purified from bovine heart (gift from Dr. V. Kinzel) were used for in vitro phosphorylation assays in buffer containing 20 mM Tris (pH 7.5), 5 mM magnesium acetate, and 5 mM dithiothreitol. Protein kinase C purified from rat brain (BIOMOL Research Labs Inc.) was assayed in a similar buffer supplemented with 1 M phorbol 12,13-dibutyrate, 0.5 mM calcium chloride, and 100 g/ml phosphatidylserine under mixed micellar conditions (49). Fusion proteins of MBP and ␤APP (see above) and soluble ␤APP secreted from HEK 293 cells were used as substrates. Soluble ␤APP was immunoprecipitated with antibody 1736, which specifically recognizes APP S -␣ (17). Immunoprecipitates were washed three times in 1 ml of the respective assay buffer. In vitro phosphorylation assays were carried out in a final volume of 30 l for 10 min at 32°C. To control the kinase activities, parallel phosphorylation reactions were carried out using phosvitin (1 mg/ml; Sigma) or histone (0.5 mg/ml; Sigma) as protein substrate. Reactions were stopped by the addition of SDS sample buffer.
Phosphoamino Acid Analysis-Phosphoamino acid analysis was carried out by one-dimensional high voltage electrophoresis according to Jelinek and Weber (50). Radiolabeled proteins electrotransferred onto polyvinylidene difluoride (PVDF) membrane (Immobilon, Millipore Corp.) were hydrolyzed in 6 M HCl for 90 min at 110°C. Subsequently, supernatants were dried in a SpeedVac concentrator, and pellets were dissolved in 8 l of pH 2.5 buffer (5.9% glacial acetic acid, 0.8% formic acid, 0.3% pyridine, and 0.3 mM EDTA) and spotted onto 20 ϫ 20-cm cellulose TLC plates (Merck) together with unlabeled phosphoamino acids (1 g each of Ser(P), Thr(P), and Tyr(P); Sigma). High voltage electrophoresis was carried out for 45 min at 20 mA. Radioactive phosphoamino acids were localized by autoradiography and identified by comparison with comigrating phosphoamino acids after ninhydrin staining.
Phosphopeptide Mapping-Proteins were separated by SDS-PAGE and transferred onto PVDF membrane. One-dimensional phosphopeptide mapping was carried out as described previously (35).

RESULTS
␤APP Is Phosphorylated by CK1 and CK2 in Vitro-To identify protein kinases that are capable of phosphorylating ␤APP within its ectodomain, we first investigated phosphorylation of soluble ␤APP in vitro. Soluble ␤APP was isolated by immunoprecipitation from the conditioned medium of HEK 293 cells expressing recombinant ␤APP, which corresponds to ␣-secretase-processed ␤APP (APP S -␣) (34). Purified APP S -␣ was incubated with [␥-32 P]ATP in the presence or absence of several protein kinases that have been demonstrated to be also present at the cell surface (40 -44). After the phosphorylation reactions, proteins were separated by SDS-PAGE, and radiolabeled ␤APP was visualized by autoradiography. Incubation with CK1 and CK2 resulted in the strong phosphorylation of ␤APP, whereas protein kinases A and C were not effective ( Fig. 1, left panel). Phosphorylation reactions carried out under the same conditions using the cognate protein substrates phosvitin for CK1 and CK2 and histone for protein kinases A and C demonstrated that the respective kinases were enzymatically active (data not shown). When [␥-32 P]GTP was used as cosubstrate, ␤APP was phosphorylated exclusively by CK2 (Fig. 1, right panel) due the unique feature of CK2 to use GTP and ATP equally well as cosubstrate (51). Together, these data demonstrate that APP S -␣ is a protein substrate for CK1 and CK2 in vitro.
Cell-surface Phosphorylation of ␤APP by Ectocasein Kinases-To characterize the phosphorylation of ␤APP at the surface of cultured cells directly, [␥-32 P]ATP was added to the supernatants of HEK 293 cells overexpressing ␤APP-(1-695) (10). Since [␥-32 P]ATP does not penetrate the cell membrane, phosphorylation of cell-surface proteins by ectoprotein kinases can be selectively detected (47). After the phosphorylation reaction, ␤APP was immunoprecipitated from cell lysates with antibody 6687, directed against the cytoplasmic tail of ␤APP; and precipitates were separated by SDS-PAGE. Two prominent bands at 105 and 120 kDa corresponding to immature and mature (fully NЈ-and OЈ-glycosylated) forms of ␤APP, respectively, were detected by Coomassie staining (Fig. 2a). The identity of both bands was verified with antibody 5313 by Western blotting (data not shown; see also Fig. 3d). Phosphoimaging revealed that only the mature form of ␤APP was phosphorylated by ectoprotein kinase activity (Fig. 2a).
To characterize the protein kinases involved in the in vivo phosphorylation of cell-surface ␤APP, we first analyzed the usage of ATP and GTP as cosubstrates. Incubation of HEK 293 cells stably overexpressing ␤APP with both [␥-32 P]ATP and [␥-32 P]GTP resulted in the phosphorylation of ␤APP (Fig. 2b). The ability to use GTP as cosubstrate indicates the involvement of the ectoprotein kinase CK2 (ecto-CK2) in the phosphorylation of the ␤APP ectodomain.
To determine if ectoprotein kinases other than ecto-CK2 also phosphorylate cell surface-located ␤APP, we added 1 mM unlabeled ATP or GTP to the phosphorylation reactions together with [␥-32 P]ATP or [␥-32 P]GTP to suppress the incorporation of 32 P by cosubstrate competition. Labeling of ␤APP with 1.5 M [␥-32 P]ATP as cosubstrate could be almost completely suppressed by the addition of excess amounts (1 mM) of unlabeled ATP, whereas the addition of 1 mM unlabeled GTP only partly inhibited 32 P incorporation from [␥-32 P]ATP (Fig. 2b). Phosphorylation of ␤APP using [␥-32 P]GTP as cosubstrate could be completely suppressed by both unlabeled ATP and GTP (Fig.  2b). As shown by Coomassie staining of the gels, the addition of 1 mM ATP or GTP did not affect the expression of ␤APP (Fig.  2b, lower panels). Similar results were obtained by analyzing soluble ␤APP, which had been secreted from cells into the supernatant during the incubation period (data not shown; see also Fig. 4). Taken together, these data demonstrate that the phosphorylation of cell-surface ␤APP involves at least two distinct ectoprotein kinases, one of which is ecto-CK2, capable of using GTP as cosubstrate. Phosphoamino acid analysis of ␤APP from cells labeled with [␥-32 P]ATP or [␥-32 P]GTP revealed that ␤APP was phosphorylated predominantly on serine residues (Fig. 2c).
To prove the involvement of ecto-CK2 and to identify other protein kinases that potentially phosphorylate ␤APP at the cell surface, we tested the effects of several selective protein kinase inhibitors. First, cells were incubated with [␥-32 P]ATP or [␥-32 P]GTP in the presence or absence of DRB, which is a selective inhibitor of CK2 (52). Phosphorylation of cell-surface ␤APP by ectoprotein kinase using [␥-32 P]ATP or [␥-32 P]GTP as cosubstrate was significantly decreased by DRB (Fig. 3a). These results provide further evidence for the involvement of ecto-CK2 in the phosphorylation of ␤APP. We next carried out cell-surface phosphorylation assays in the presence or absence of HD, a recently characterized selective inhibitor of CK1-type protein kinases (53). The addition of HD to the cell supernatant resulted in a slightly decreased phosphorylation of ␤APP when [␥-32 P]ATP was used as cosubstrate (Fig. 3b, left panels). To selectively detect ectoprotein kinase activities other than CK2, phosphorylation assays with [␥-32 P]ATP were carried out in the presence of 1 mM unlabeled GTP, which was sufficient to completely saturate ecto-CK2 activity (Fig. 2b). Under these conditions, HD almost fully suppressed the phosphorylation of ␤APP (Fig. 3b, right panels). Since HD did not decrease the phosphorylation of ␤APP when [␥-32 P]GTP was used as cosub- Precipitates were separated by SDS-PAGE. Gels were Coomassie-stained, and radiolabeled proteins were detected by phosphoimaging. Mature and immature forms of ␤APP are indicated by arrows. Note that only the mature form of ␤APP was phosphorylated. b, cell-surface phosphorylation reactions were carried out using [␥-32 P]ATP or [␥-32 P]GTP as cosubstrate in the presence or absence of 1 mM unlabeled ATP or GTP as indicated. Phosphorylation of ␤APP was analyzed as described for a. To visualize ␤APP, gels were Coomassie-stained (lower panels). Note that ␤APP was phosphorylated by ectoprotein kinases using ATP or GTP as cosubstrate. c, shown are the results of the phosphoamino acid analysis of ␤APP that was phosphorylated by intact cells using [␥-32 P]ATP or [␥-32 P]GTP as cosubstrate. After separation of immunoprecipitated ␤APP by SDS-PAGE, proteins were electrotransferred onto a PVDF membrane, and radiolabeled ␤APP was subjected to one-dimensional phosphoamino acid analysis after limited acidic hydrolysis (see "Materials and Methods"). The migration of phosphoamino acid standards is indicated by arrowheads. Ectoprotein kinases using ATP and GTP as cosubstrates phosphorylated the ectodomain of ␤APP predominantly on serine residues. Ctr., control. strate (data not shown), this inhibitor did not interfere with ecto-CK2 activity. Similar results were obtained using CKI-7 (54), another selective inhibitor of CK1 (data not shown). These results indicate that besides ecto-CK2, ecto-CK1 is also involved in the phosphorylation of ␤APP. The involvement of ecto-CKs in the cell-surface phosphorylation of ␤APP is also supported by the finding that the addition of phosvitin, an acidic protein substrate of both CK1 and CK2, significantly suppressed the phosphorylation of ␤APP (Fig. 3c). Since the cell-surface phosphorylation of ␤APP can be almost completely suppressed by the inhibition of ecto-CK1 and ecto-CK2, these enzymes appear to be the major kinases in the ectodomain phosphorylation of ␤APP. To prove this further, we tested the effect of other selective protein kinase inhibitors on the cellsurface phosphorylation of ␤APP. The compounds H-89 (an inhibitor of protein kinase A) (55), roscovitin (an inhibitor of cyclin-dependent kinases) (56), KN-62 (an inhibitor of Ca 2ϩ / calmodulin-dependent kinase II) (57), and GF109203X (an inhibitor of protein kinase C) (58) were added during the incubation of cells with [␥-32 P]ATP. To suppress the activity of CK2, incubations were carried out in the presence 1 mM unlabeled GTP. Under these conditions, the kinase inhibitors H-89, KN-62, roscovitine, and GF109203X did not have a significant effect on the phosphorylation of cell-surface ␤APP (Fig. 3d), whereas the addition of HD almost completely inhibited the phosphorylation of ␤APP (see also Fig. 3b). The selective CK2 inhibitor DRB did not lead to a further decrease in ␤APP phosphorylation, demonstrating that CK2 activity was saturated by 1 mM unlabeled GTP. Taken together, these data demonstrate that cell surface-located ␤APP is predominantly phosphorylated by ecto-CK1 and ecto-CK2. However, these results do not rule out that other kinases could also phosphorylate ␤APP to some extent.
Phosphorylation of Cell-surface ␤APP within an Acidic Domain-Previously, we identified serines 198 and 206 as in vivo phosphorylation sites of secreted ␤APP. Phosphorylation on Ser 198 and Ser 206 was shown to occur in post-Golgi secretory compartments (35). To determine whether cell surface-located ␤APP is also phosphorylated on these serine residues, we incubated HEK 293 cells stably expressing wild-type ␤APP or ␤APP carrying serine-to-alanine substitutions at positions 198 and 206 (S198A/S206A) with [ 32 P]orthophosphate or with [␥-32 P]ATP. ␤APP was then immunoprecipitated from cell lysates or from the conditioned media, and phosphorylation was analyzed by autoradiography. Labeling cells that express wildtype ␤APP with [ 32 P]orthophosphate resulted in the detection of phosphorylated soluble APP S -␣ in the conditioned medium (Fig. 4a, left panels) and phosphorylated full-length ␤APP in cell lysates (middle panels). Phosphorylation of ␤APP S198A/ S206A was strongly reduced under these conditions, indicating that Ser 198 and Ser 206 are the major sites phosphorylated in the secretory pathway. In contrast, both wild-type ␤APP and ␤APP S198A/S206A were phosphorylated when cells were incubated with [␥-32 P]ATP, demonstrating that ecto-CKs use additional sites in the ␤APP ectodomain. Serines 198 and 206 are located within an acidic domain that contains two additional serine residues (Ser 193 and Ser 221 ) representing potential phosphorylation sites for CKs (see Fig. 5c). We therefore analyzed the phosphorylation of wild-type ␤APP and ␤APP S198A/S206A by purified CK1 and CK2. Soluble wild-type ␤APP or ␤APP S198A/S206A was immunoprecipitated from the conditioned medium of HEK 293 cells stably expressing the respective derivatives of ␤APP and incubated with CK1 or CK2 in the presence of [␥-32 P]ATP. Interestingly, both kinases efficiently phosphorylated ␤APP S198A/S206A (Fig. 4b). To analyze if CK1 and CK2 phosphorylate ␤APP within this acidic domain, one-dimensional phosphopeptide mapping was carried out. Wild-type ␤APP and ␤APP S198A/S206A were phosphorylated in vitro by CK1 or CK2 and then digested with trypsin. Peptides were separated by SDS-PAGE on a Tris/Tricine gel and analyzed by autoradiography. Phosphorylation of wildtype ␤APP and ␤APP S198A/S206A by both CK1 and CK2 occurred in a characteristic low molecular weight tryptic peptide (Fig. 4b), which was identified previously by mass spectrometry to represent amino acids 181-224 (Ref. 35; see also Fig. 5c). These data demonstrate that CK1 and CK2 phosphorylate the acidic domain of ␤APP in vitro.
To determine if this acidic domain is indeed the major site of phosphorylation, we deleted the sequence corresponding to the tryptic peptide of ␤APP (amino acids 181-224) (Fig. 5c). MBP-␤APP and MBP-␤APP⌬181-224 fusion proteins were incubated with purified CK1 and CK2 as well as with both kinases. MBP-␤APP was readily phosphorylated by CK1, CK2, and the CK1/ CK2 mixture. In contrast, the phosphorylation of MBP-␤APP⌬181-224 was significantly reduced (Fig. 5a). However, some residual phosphorylation occurred also in MBP-␤APP⌬181-224, indicating that CK1 and CK2 can also phosphorylate additional sites of ␤APP to some extent. Similar results To suppress ecto-CK2 activity, all reactions were carried out in the presence of 1 mM unlabeled GTP. Radiolabeled ␤APP was detected by phosphoimaging (upper panels), and total ␤APP was detected by immunoblotting with antibody 5313 (lower panels). Phosphorylated endogenous ␤APP is indicated (*). Ctr., control.
were obtained when the respective fusion proteins were incubated with HEK 293 cells in the presence of [␥-32 P]ATP to allow phosphorylation by ectoprotein kinases (Fig. 5b). Taken together, these data demonstrate that the phosphorylation of ␤APP by ecto-CKs and by purified CKs predominantly occurs within the acidic domain between amino acids 181 and 224 (Fig. 5c).
Phosphorylation of Secreted ␤APP by Ectocasein Kinases-To address the question of whether ecto-CKs can also phosphorylate proteolytically processed soluble ␤APP, we incubated untransfected HEK 293 cells with conditioned phosphorylation buffer derived from cells stably expressing recombinant APP S -␣ (34). Phosphorylation reactions were carried out using [␥-32 P]ATP or [␥-32 P]GTP as cosubstrate. APP S -␣ was readily phosphorylated by intact cells upon incubation with [␥-32 P]ATP or [␥-32 P]GTP (Fig. 6, lanes 3 and 4). As shown for membranebound ␤APP (Fig. 3c), the phosphorylation of soluble ␤APP was also significantly suppressed when phosvitin (2 mg/ml) was added to the cell supernatants during the incubation (data not shown). Incubation of cells with phosphorylation buffer not containing APP S -␣ carried out as a control did not result in the detection of phosphorylated APP S -␣ (Fig. 6, lane 2), demonstrating that the phosphorylated APP S -␣ (see lanes 3 and 4) did not originate from endogenously expressed ␤APP. Incubation of phosphorylation buffer containing APP S -␣ in the absence of cells resulted in a minor phosphorylation of ␤APP (Fig. 6, lane  1), indicating that soluble ␤APP is predominantly phosphorylated by membrane-bound ectoprotein kinases. These results demonstrate that soluble forms of ␤APP resulting from ␣-secretory processing can be phosphorylated by ecto-CKs, similar to membrane-bound full-length ␤APP.
Cell-surface Phosphorylation of ␤APP Is Inhibited by Heparin-Because ectoprotein kinases are exposed to the external side of the plasma membrane, we tested whether the phosphorylation of ␤APP is regulated by the extracellular matrix constituent heparin, which is known to inhibit the activities of CK1 and CK2 (51). HEK 293 cells stably expressing ␤APP were incubated with [␥-32 P]ATP in the presence or absence of heparin. Heparin strongly inhibited the phosphorylation of cellsurface ␤APP (Fig. 7). Similar results were obtained when the phosphorylation of soluble ␤APP, which was secreted by the cells during the labeling period, was analyzed (data not shown). These results demonstrate that the phosphorylation of ␤APP can be modulated at the cell surface by heparin in a concentrationdependent manner.

DISCUSSION
Phosphorylation of proteins is a common mechanism to regulate their functions, and protein kinases might represent pharmacological targets for the therapeutic intervention of diseases. In this study, we aimed to identify protein kinases that phosphorylate the ectodomain of ␤APP. A cell-based phosphorylation assay was used to selectively detect kinase activities at the outer face of the plasma membrane. The specificity of this FIG. 5. ␤APP is predominantly phosphorylated within the acidic domain. a, MBP-␤APP and MBP-␤APP⌬181-224 (MBP-APP⌬) were incubated with purified CK1 or CK2 or a mixture of both kinases in the presence of [␥-32 P]ATP. Reaction mixtures were separated by SDS-PAGE and transferred to PVDF membrane. Phosphorylated fusion proteins were detected by autoradiography (upper panels). Overexposed autoradiograms revealed that CK1 also phosphorylated MBP-␤APP⌬181-224 (data not shown). As a loading control, proteins were detected by immunoblotting with antibody 5313. b, MBP-␤APP and MBP-␤APP⌬181-224 were incubated with HEK 293 cells in the presence of 1 M [␥-32 P]ATP for 20 min. Fusion proteins were immunoprecipitated with antibody 5313 and processed as described for a. Note that deletion of the acidic domain (amino acids 181-224) strongly decreased the phosphorylation of ␤APP by both purified CKs and intact cells. c, shown is the amino acid sequence representing a tryptic peptide of ␤APP (amino acids 181-224) that was deleted in MBP-␤APP⌬181-224. Serine residues representing potential phosphorylation sites of CKs are shown in boldface.

FIG. 4.
Mapping of in vivo phosphorylation sites of ␤APP. a, HEK 293 cells stably expressing wild-type ␤APP (wt) or ␤APP S198A/S206A were labeled with [ 32 P]orthophosphate and with [␥-32 P]ATP, respectively. ␤APP was immunoprecipitated from the conditioned medium with antibody 5313 or from cell lysates with antibody 6687, separated by SDS-PAGE, and transferred to PVDF membrane. Radiolabeled proteins were detected by autoradiography (upper panels). Subsequently, ␤APP was visualized by immunoblotting with antibody 5313 (lower panels). Note that mutation of serines 198 and 206 to alanine (S198A/S206A) resulted in a strong decrease in phosphate incorporation when cells were labeled with [ 32 P]orthophosphate. In contrast, S198A/S206A had only little effect when cells were labeled with [␥-32 P]ATP. b, shown are the results from the in vitro phosphorylation of wild-type ␤APP and ␤APP S198A/S206A by CK1 and CK2. Wild-type ␤APP and ␤APP S198A/S206A were immunoprecipitated from the conditioned medium and incubated with purified CK1 and CK2. Phosphorylation was detected by autoradiography (upper panels). Radiolabeled bands were subjected to a tryptic digest, and the resulting peptides were separated by SDS-PAGE. CK1 and CK2 phosphorylated ␤APP within a characteristic tryptic peptide (arrow) (35). The bands at ϳ46, 19, and 15 kDa (labeled by asterisks) most likely represent partially digested fragments of phosphorylated ␤APP. sup, supernatant. assay critically depends on the intactness of the cellular plasma membrane. This is demonstrated by the selective phosphorylation of mature ␤APP (Fig. 2a). In cells with disturbed membrane integrity and nonspecific uptake of [␥-32 P]ATP, one would expect phosphorylation of both mature and immature ␤APPs. Therefore, our results demonstrate that the ␤APP ectodomain is specifically phosphorylated at the plasma membrane by ectoprotein kinases.
By using various protein kinase inhibitors in a cell culture phosphorylation assay, we found that the phosphorylation of ␤APP is selectively suppressed by the CK1 inhibitors HD (53) and CKI-7 (54) and by the CK2-specific inhibitor DRB (52). In addition, ␤APP is phosphorylated by a protein kinase using GTP as cosubstrate. Usage of GTP as cosubstrate is a specific feature of CK2 (51). These results therefore indicate that cell surface-located ␤APP can be phosphorylated by ecto-CK1 and ecto-CK2. Our data obtained by the cell-surface phosphorylation of ␤APP are consistent with the results of the in vitro phosphorylation experiments with several purified protein kinases, which demonstrated that the ␤APP ectodomain was readily phosphorylated by purified CK1 and CK2, whereas protein kinases A and C were not effective in phosphorylating ␤APP. A search for potential kinase recognition motifs in ␤APP revealed the presence of several sites for CK1 ((D/E)XX(S/T)) and CK2 ((S/T)XX(D/E)) (59) within the ␤APP ectodomain (data not shown). Previously, we identified serines 198 and 206 as in vivo phosphorylation sites of ␤APP, which are phosphorylated on the passage of ␤APP through the secretory pathway.
These serine residues are located within recognition motifs of CK1 and CK2 (35), and mutation of these residues to alanine significantly inhibited the phosphorylation of ␤APP during secretion (60). However, the S198A/S206A double mutation can still be phosphorylated at the cell surface by ecto-CKs. Therefore, our results indicate that ectoprotein kinases can phosphorylate cell surface-located ␤APP at additional sites as compared with the phosphorylation of ␤APP at Ser 198 and Ser 206 during secretion. We mapped the phosphorylation of ␤APP by ecto-CKs to an acidic domain within amino acids 181-224. Besides Ser 198 and Ser 206 , this domain contains two additional serine residues flanked by acidic amino acids. Indeed, our in vitro data demonstrate that CK1 and CK2 are capable of phosphorylating ␤APP within this domain. These results indicate that ectodomain phosphorylation of ␤APP is a regulated consecutive process. Although the phosphorylation of ␤APP during its transport to the cell surface occurs predominantly at serines 198 and 206, ecto-CK1 and ecto-CK2 can phosphorylate cell surface-located and soluble ␤APPs at additional sites.
Expression of ectoprotein kinases at the cell surface was shown in several cell types, including tumor cells (42,43,47), endothelial cells (61,62), neutrophils (63), platelets (64), and neuronal cells (65,66). Interestingly, both CK1 and CK2 have been shown to be released from the cell surface upon the addition of the cognate protein substrates casein and phosvitin (41,42,63) and appear to be constituents of a functional kinase complex (43). To our knowledge, ␤APP is the first protein substrate identified to be phosphorylated by both ecto-CK1 and ecto-CK2. We found that both kinases phosphorylate membrane-bound as well as soluble ␤APP, which is generated by ␣-secretory cleavage (Fig. 8). Therefore, ␤APP function might be regulated by phosphorylation even after its secretion from cells.
Notably, we found that the extracellular matrix component heparin efficiently inhibits the phosphorylation of ␤APP in a concentration-dependent manner. This inhibition is likely to be due to the direct inhibition of ecto-CK1 and ecto-CK2 because heparin inhibits both CKs (51). However, we cannot exclude that binding of heparin to ␤APP itself (23, 24) could affect its phosphorylation. It was previously reported that heparin enhances the phosphorylation of the microtubule-associated protein tau (71,72), most likely by inducing a conformational change in tau that facilitates access of certain kinases (73). In addition, it was shown that heparin promotes the aggregation of tau into filaments (72,74). Heparin therefore exhibits distinct effects on the phosphorylation of two proteins that are associated with the two major pathological lesions of Alzheimer's disease. The inhibition of ␤APP phosphorylation by heparin might represent a regulatory mechanism in vivo because heparin and ecto-CKs could interact in the extracellular environment. The effect of heparin on the biological functions of ␤APP such as its neurite outgrowth-promoting activity (23,75) might therefore depend not only on its direct binding to ␤APP, but also on the inhibition of ␤APP phosphorylation.