Coagulation factor XIa cleaves the RHDS sequence and abolishes the cell adhesive properties of the amyloid beta-protein.

Amyloid β-protein (Aβ) is the major constituent of senile plaques and cerebrovascular amyloid deposits in Alzheimer's disease and is proteolytically derived from its transmembrane parent protein the amyloid β-protein precursor (AβPP). Although the physiological role(s) of secreted AβPPs are not fully understood, several potential functions have been described including the regulation of hemostatic enzymes factors XIa and IXa and a role in cell adhesion. In the present study, we investigated the proteolytic processing of AβPP by factor XIa (FXIa). Incubation of the human glioblastoma cell line U138 stably transfected to overexpress the 695 isoform of AβPP with FXIa (2.5-5 nM) resulted in proteolytic cleavage of secreted AβPP. Higher concentrations of FXIa (>25 nM) resulted in loss in cell adherence. Coincubation of FXIa with purified, recombinant Kunitz protease inhibitor domain of AβPP blocked both the proteolytic processing of AβPP and the loss of cell adhesion. The RHDS cell adhesion site of AβPP resides within residues 5-8 of the Aβ domain. Incubation of synthetic Aβ1-40 peptide with increasing concentrations of FXIa resulted in cleavage of Aβ between Arg5 and His6 within the cell adhesion domain of the peptide. FXIa-digested Aβ1-40 or AβPP695 lost their abilities to serve as cell adhesion substrates consistent with cleavage through this cell adhesion site. Together, these results suggest a new potential biological function for FXIa in the modulation of cell adhesion. In addition, we have shown that FXIa can proteolytically alter Aβ and therefore possibly modify its physiological and perhaps pathological properties.

Amyloid ␤-protein (A␤) is the major constituent of senile plaques and cerebrovascular amyloid deposits in Alzheimer's disease and is proteolytically derived from its transmembrane parent protein the amyloid ␤-protein precursor (A␤PP). Although the physiological role(s) of secreted A␤PPs are not fully understood, several potential functions have been described including the regulation of hemostatic enzymes factors XIa and IXa and a role in cell adhesion. In the present study, we investigated the proteolytic processing of A␤PP by factor XIa Deposition of the amyloid ␤-protein (A␤) 1 in senile plaques in the neuropil and in the walls of cerebral blood vessels is a pathologic feature of Alzheimer's disease. A␤ is a 39 -42-amino acid protein that is proteolytically derived from its transmembrane parent protein, amyloid ␤-protein precursor (A␤PP) (1)(2)(3)(4). A␤PP is a multidomain protein that can be translated from alternatively spliced transcripts from a single gene located on chromosome 21 (5)(6)(7)(8)(9)(10)(11). The major mRNA species encode proteins of 695, 751, and 770 amino acids. The latter two isoforms contain a 56-amino acid domain that is homologous to Kunitztype serine protease inhibitors (KPI) (9 -11). These isoforms are identical to the cell secreted inhibitor identified as protease nexin-2 (PN-2) (12,13). Secretory cleavage of A␤PP occurs within the A␤ domain, and therefore, processing through this pathway precludes A␤ formation (14,15).
Although the physiological roles of secreted A␤PP are not fully understood, several potential functions have been ascribed to it. For example, several laboratories have provided evidence that A␤PP can mediate both cell-cell and cell-surface adhesion in neural and non-neural cells (16 -20). Recently, A␤PP has been shown to be involved in regulating intracellular calcium levels in neurons, thus providing protection to these cells (21,22). Both the cell adhesion and neuroprotective activities have been localized to the carboxyl-terminal region of the secreted A␤PP (17,21,22). In addition, growth-promoting activities of A␤PP in non-neural and neural cells have been reported (23,24). The region responsible for this autocrine activity has been identified as the sequence RERMS located in the middle portion of A␤PP (24).
Secreted forms of A␤PP that contain the KPI domain have been shown to inhibit several different serine proteases, which include trypsin, chymotrypsin, and coagulation Factors XIa and IXa (12,(25)(26)(27)(28)(29). Through inhibition of these latter two proteases, it has been suggested that the KPI-containing isoforms of A␤PP may play a role in regulating hemostasis by acting as an anticoagulant (30). Factor XIa (FXIa) participates in the middle phase of the intrinsic pathway of blood coagulation by converting the zymogen Factor IX to the active serine protease Factor IXa. Factor IXa then converts Factor X to Factor Xa, the next serine protease to participate in the coagulation cascade ultimately leading to fibrin clot formation (31).
In the present study, we investigated the proteolytic processing of A␤PP by FXIa. We report here that FXIa can cleave the RHDS sequence in the A␤ domain, altering the cell adhesive properties of both A␤ and secreted A␤PP. These data suggest a new potential biological function for FXIa in the modulation of cell adhesion. These findings also indicate that the KPI domain of A␤PP can regulate the activity of a protease that can process the A␤ domain, thus altering its potential physiological and pathological properties.

EXPERIMENTAL PROCEDURES
Materials-Purified human FXIa was purchased from Enzyme Research Laboratories. Purified ␣-thrombin (3554 units/mg) was purchased from Calbiochem. Dulbecco's modified Eagle's medium, fetal bovine serum, and Geneticin were obtained from Life Technologies, Inc. A␤  was synthesized and structurally characterized as described previously (32). The recombinant KPI domain of A␤PP was expressed and purified as described (33). Purified secreted A␤PP 695 was prepared in a baculovirus system (34) and was kindly provided by Drs. L. Gregori and D. Goldgaber of the State University of New York at Stony Brook. * This work was supported by Grant HL49566 and Research Career Development Award HL03229 (to W. E. V. N.) from the National Institutes of Health and Grant-in-Aid 94006240 from the American Heart Association. 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.
‡ Supported by National Institute on Aging National Research Service Award AG00096-12 from the University of California, Irvine.
The anti-PN-2/A␤PP monoclonal antibody (mAb) P2-1 was prepared as previously reported (12). mAb 6E10 to A␤ was a generous gift from Dr. K. S. Kim. Hybond nitrocellulose membranes, peroxidase-coupled secondary antibodies, and enhanced chemiluminescence reagents were purchased from Amersham Corp.
Cells-The human glioblastoma cell line U138 stably transfected to overexpress the 695-amino acid isoform of A␤PP (U138/695) was generously provided by Dr. M. Murphy of the Salk Institute Biotechnology/ Industrial Associates, Inc. These cells were routinely cultured in Dulbecco's modified Eagle's medium containing 2 mM glutamine, 1 mM sodium pyruvate, 100 units/ml penicillin, 100 g/ml streptomycin sulfate, nonessential amino acids, 10% fetal bovine serum, and 20 g/ml Geneticin.
Effects of Factor XIa on Proteolytic Processing of A␤PP-U138/695 cells were grown to near confluency in 12-well tissue culture plates in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum and 20 g/ml Geneticin as described above, rinsed one time with 1 ml of serum-free Dulbecco's modified Eagle's medium containing 0.1% bovine serum albumin, and then incubated in 1 ml of the same medium for 3 h. Then the medium was removed, replaced with 0.5 ml of serum-free medium alone or serum-free medium containing various concentrations of FXIa, and incubated for 18 h at 37°C. In parallel experiments, cultures were treated with either 50 nM FXIa, 100 nM KPI, 50 nM FXIa, and 100 nM KPI or 100 nM thrombin for 18 h at 37°C. The medium from control and treated cultures was collected, and the cells were solubilized as described previously (35). All samples were stored at Ϫ30°C until immunoblotting analysis.
Proteolytic Processing of A␤ by FXIa-Synthetic A␤ 1-40 (25 M) was incubated with increasing concentrations of FXIa for 24 h at 37°C in serum-free medium. In a similar manner, A␤ 1-40 was also treated with either 50 nM FXIa, 100 nM KPI, or 50 nM FXIa and 100 nM KPI. Aliquots of the digested peptide were analyzed by immunoblotting as described below.
Immunoblotting-For analysis of secreted and cellular A␤PP proteins, aliquots of culture medium samples or cell lysate samples were separated by electrophoresis on nonreducing SDS 10% polyacrylamide gels (36). For analysis of A␤ 1-40 peptides, samples were electrophoresed on Tris/Tricine/SDS 10 -20% gradient polyacrylamide gels. The proteins were then transferred onto Hybond membranes and probed using mAb P2-1 (5 g/ml) or mAb 6E10 ascites (1:3000) for detection of A␤PP or A␤, respectively. Bound mouse monoclonal antibody was detected using a peroxidase-coupled sheep anti-mouse IgG (1:1000). Immunoreactivity was detected by enhanced chemiluminescence followed by exposure to Kodak X-Omat film.
Amino Acid Sequence Analysis-Synthetic A␤ 1-40 (10 g) was incubated with 100 nM FXIa at 37°C for 24 h. An ϳ3.4-kDa FXIa cleaved A␤ 1-40 fragment was gel-purified by electrophoresis on a Tris/Tricine/ SDS 10 -20% polyacrylamide gel. The peptide was transferred to a polyvinylidene difluoride membrane and subjected to automated sequential Edman degradation analyses using on a 475A protein sequencer.
Cell Attachment Assay-35-mm plastic dishes were coated by evaporation overnight at 37°C with the appropriate substrate diluted in 0.1 ϫ phosphate-buffered saline. The substrates used were 2 g A␤ 1-40 ; 400 ng FXIa; 65 ng KPI; 2 g A␤ 1-40 and 65 ng KPI; 400 ng FXIa and 65 ng KPI; 2 g A␤ 1-40 digested with 400 ng FXIa for 48 h at 37°C; 2 g A␤ 1-40 digested with 400 ng FXIa then incubated with 65 ng KPI for 2 h at 37°C; 10 g of purified A␤PP 695 ; 10 g of A␤PP 695 digested with 200 ng of FXIa then incubated with 65 ng KPI for 2 h at 37°C; and 10 g of A␤PP 695 digested with 200 ng of thrombin. Exponentially dividing U138/695 cells were collected and suspended in serum-free medium. A 2-ml suspension containing 3.4 ϫ 10 5 cells was pipetted into the coated dishes and incubated for 1 h at 37°C. Nonadherent cells were removed by two gentle washes with serum-free culture medium. Adherent cells were then photographed using an Olympus phase contrast microscope. Cells were counted from 3-5 fields for each condition tested.

FXIa Induces Proteolytic Processing of Secreted A␤PP in
Cultured U138/695 Cells-We examined the effects of FXIa on A␤PP processing in cultured U138/695 cells that have previously been shown to overproduce the A␤PP 695 isoform by 80fold over untransfected glioblastoma cells (35). Because A␤PP 695 lacks the KPI domain, the effects of FXIa could be tested without intrinsic inhibition by the PN-2 form of A␤PP. Proteolytic processing of secreted A␤PP was observed at 2.5 nM FXIa as evidenced by the appearance of a truncated secreted A␤PP of ϳ85 kDa in the culture media (Fig. 1A). At 50 nM FXIa, Ͼ90% of the secreted A␤PP was cleaved. With higher concentrations of FXIa, additional cleavage sites were utilized as suggested by the appearance of additional smaller A␤PP fragments. Cellular A␤PP was not affected (data not shown).
It was noted that the ϳ85-kDa truncated secreted A␤PP was very similar in size to that we and others have shown to be generated by thrombin proteolysis (35,37,38). Therefore, we compared the effects of 50 nM FXIa and 100 nM thrombin on A␤PP proteolysis in U138/695 cells (Fig. 1A). The amino acid sequence at the thrombin cleavage site in A␤PP was determined to be EPR with cleavage occurring on the carboxyl side of the arginine residue (37). This is the same sequence as the synthetic chromogenic substrate used to assay FXIa activity in vitro (39). Based on sequence similarity and the size of the truncated secreted A␤PP protein, the initial cleavage site utilized by FXIa is most likely the same as the thrombin cleavage site located at Arg 510 -Ile 511 of the A␤PP 695 sequence. FXIa cleavage of secreted A␤PP 695 could be inhibited when FXIa was preincubated with 100 nM KPI (Fig. 1B).
Factor XIa Alters Cell Adhesion of U138/695 Cells-In the course of studying the effects of increasing concentrations of FXIa on the proteolytic processing of A␤PP 695 , we observed that the cells lost adherence after 18 h when treated with 5-25 nM FXIa. Many of the cells that remained were rounded and no longer showed their characteristic elongated morphology. After an 18-h incubation with 50 nM FXIa, less than 20% of the cells remained adhered to the culture dish (Fig. 2B). In parallel experiments, similar concentrations of thrombin (Fig. 2E) and coagulation Factors IXa and Xa (data not shown) had no effect on cell adhesion. In addition, these effects of FXIa treatment could be inhibited by the addition of 100 nM KPI to the FXIa prior to incubation with the cells (Fig. 2D). Similar but less robust losses in cell adhesion were observed in the untransfected parent glioblastoma cell line, a neuroblastoma cell line, and cultured cerebrovascular smooth muscle cells (data not shown).
FXIa Cleaves the Cell Adhesion RHDS Sequence of A␤-A cell adhesion domain has been localized to the RHDS sequence of the A␤ domain located at the carboxyl terminus of secreted A␤PP (18). To determine if disruption of this domain by FXIa was responsible for the loss of adherence observed in the FXIa treated U138/695 cells, we examined the effect of FXIa on a synthetic A␤ 1-40 peptide. With increasing FXIa concentrations, we detected a decrease in A␤ immunoreactivity using mAb 6E10 (Fig. 3A). At 50 nM FXIa, Ͼ80% of A␤ immunoreactivity was lost. In parallel experiments, neither thrombin nor KPItreated FXIa had any effect on A␤ immunoreactivity (Fig. 3B).
To further identify the site of FXIa cleavage within A␤ 1-40 , 10 g of the peptide was digested with 100 nM FXIa for 48 h. The digested peptide was analyzed on a Tris/Tricine/SDS 10 -20% polyacrylamide gel. The Coomassie-stained gel revealed a truncated ϳ3.4-kDa A␤ fragment (Fig. 4). In parallel experiments, the truncated A␤ peptide was transferred to a polyvinylidene difluoride membrane and subjected to amino-terminal sequence analysis. The resulting sequence derived from five cycles of sequential Edman degradation is shown in Table I. The amino-terminal sequence of the truncated A␤ peptide identified the FXIa cleavage between Arg 5 -His 6 within the RHDS cell adhesion domain.
FXIa Cleavage of the RHDS Sequence Abolishes the Cell Adhesion Properties of A␤ and Secreted A␤PP-To determine if FXIa cleavage of the RHDS sequence disrupts the ability of A␤ to promote cell adhesion, we compared untreated and FXIadigested A␤ 1-40 as substrates in cell-surface adhesion. The U138/695 cells showed enhanced binding to the plastic culture dish treated with A␤ 1-40 by nearly 20-fold over cells plated on 0.1 ϫ phosphate-buffered saline-treated plastic dishes (Fig.  5A). However, when FXIa-digested A␤ 1-40 was used as a substrate, cell adherence was abolished (Fig. 5A). In control experiments, the numbers of cells adhered to FXIa, KPI, or FXIa and KPI closely resembled the numbers of cells on phosphate-buffered saline-treated plastic (Fig. 5A). Similar to A␤ 1-40 , cells adhered to purified A␤PP 695 but not FXIa-digested A␤PP 695 (Fig. 5B). It is noteworthy that cells adhered to thrombindigested A␤PP 695 (Fig. 5B), consistent with a lack of A␤ cleavage by thrombin (Fig. 3B). No significant differences in numbers of adhered cells using fibronectin or FXIa-treated fibronectin were observed in parallel experiments (data not shown).

DISCUSSION
The present studies show that FXIa, a target protease for inhibition by the KPI domain of A␤PP, proteolytically cleaves secreted A␤PP 695 from U138/695 cells. Initial cleavage at low FXIa concentrations (ϳ2.5 nM) occurred at the site previously reported for thrombin cleavage at Arg 510 -Ile 511 of A␤PP 695 . It is noteworthy that the amino acid sequence at the thrombin cleavage site in A␤PP is Glu 508 -Pro 509 -Arg 510 , the same sequence as the synthetic chromogenic substrate used to assay  FXIa activity in vitro (39). Additional cleavage sites were utilized at higher concentrations of FXIa (Fig. 1A). Several potential functions have been ascribed to secreted A␤PP 695 , including cell adhesion, growth supportive activity, and neuroprotection (16 -24). Proteolytic processing of A␤PP 695 by FXIa may disrupt one or more of these properties.
Reduced cell adhesion to the substratum was observed in FXIa-treated U138/695 cells. This may have resulted from disruption of the cell adhesion domain by FXIa. A cell adhesion site in A␤PP has been localized within the amino-terminal region of the A␤ domain, which is present in secreted forms of the protein (18). To test this hypothesis, synthetic A␤ 1-40 was digested with increasing concentrations of FXIa. The loss of A␤ immunoreactivity observed in Fig. 3 was consistent with disruption of the mAb 6E10 epitope, which has been mapped within the amino-terminal region of A␤ (40). Accompanying the loss of A␤ immunoreactivity, we observed the appearance of a truncated A␤ peptide when treated with FXIa (Fig. 4). Aminoterminal sequence analysis of the first five amino acids of this truncated peptide yielded the A␤ sequence HDSGY (Table I), consistent with cleavage through the cell adhesion domain RHDS. It is noteworthy that the RHDS cleavage site in the A␤ domain shows homology to the factor IX activation site for FXIa (32). Because cleavage through the RHDS sequence may disrupt the cell adhesive properties of the A␤ domain, we compared the abilities of A␤ 1-40 and FXIa-cleaved A␤ 1-40 to serve as cell adhesion substrates. A␤ 1-40 promoted adhesion of U138/ 695 cells to the substratum by nearly 20-fold over buffertreated substratum (Fig. 5). However, FXIa-cleaved A␤ 1-40 lost its ability to serve as cell adhesive substrate. Similarly, FXIa diminished the ability of secreted A␤PP 695 to promote cell adhesion. These results suggest that FXIa could disrupt adhesion of cells that may use this region of the A␤ domain as an extracellular matrix protein.
Together, these findings have identified a new potential function for FXIa in the modulation of cell adhesion. This suggests that in addition to a function in coagulation, FXIa may participate in other roles of wound repair and tissue remodeling at sites of vascular injury. It is noteworthy that platelets activated by physiological agonists release large amounts of PN-2/A␤PP (27,41). Therefore, PN-2/A␤PP released by platelets at sites of vascular damage may participate in both the regulation of coagulation and cell adhesion. Through its intimate interactions with PN-2/A␤PP, FXIa may also be involved in both of these processes. The present data also show that the KPI domain of PN-2/A␤PP can regulate the activity of a target protease that possesses the ability to modulate another potential biological function on a distal region of the protein. Moreover, these findings demonstrate that the KPI domain of PN-2/A␤PP can regulate the activity of a protease that can cleave the A␤ peptide. Proteolytic processing of this nature could alter possible physiologic and perhaps pathologic properties of A␤.