Localization of a Fibrillar Amyloid β-Protein Binding Domain on Its Precursor

Deposition of fibrillar amyloid-β protein (Aβ) in senile plaques and in the walls of cerebral blood vessels is a key pathological feature of Alzheimer's disease and certain related disorders. Fibrillar Aβ deposition is intimately associated with neuronal and cerebrovascular cell death both in vivo and in vitro. Similarly, accumulation of the Aβ protein precursor (AβPP) is also observed at sites of fibrillar Aβ deposition. Recently, we reported that fibrillar Aβ, but not unassembled Aβ, promotes the specific binding of AβPP through its cysteine-rich, amino-terminal region (Melchor, J. P., and Van Nostrand, W. E. (2000) J. Biol. Chem. 275, 9782–9791). In the present study we sought to determine the precise site on AβPP that facilitates its binding to fibrillar Aβ. A series of synthesized overlapping peptides spanning the cysteine-rich, amino-terminal region of AβPP were used as competitors for AβPP binding to fibrillar Aβ. A peptide spanning residues 105–119 of AβPP competitively inhibited AβPP binding to fibrillar Aβ in a solid-phase binding assay and on the surface of cultured human cerebrovascular smooth muscle cells. Alanine-scanning mutagenesis of residues 105–117 within glutathioneS-transferase (GST)-AβPP-(18–119) revealed that His110, Val112, and Ile113 are key residues that facilitate AβPP binding to fibrillar Aβ. These specific residues belong to a common β-strand within this region of AβPP. Wild-type GST-AβPP-(18–119) protected cultured human cerebrovascular smooth muscle cells from Aβ-induced toxicity whereas H110A mutant GST-AβPP-(18–119) did not. Wild-type GST-AβPP-(18–119) bound to different isoforms of fibrillar Aβ and fibrillar amylin peptides whereas H110A mutant and I113A mutant GST-AβPP-(18–119) were substantially less efficient binding to each fibrillar peptide. We conclude that His110, Val112, and Ile113, residing in a common β-strand region within AβPP-(18–119), comprise a domain that mediates the binding of AβPP to fibrillar peptides.

Fibrillar amyloid-␤ protein (A␤) 1 deposition in senile plaques in the neuropil and in the walls of cerebral blood vessels is a common pathologic feature of patients with Alzheimer's disease (AD) and certain related disorders including Down's syndrome and hereditary cerebral hemorrhage with amyloidosis of the Dutch type (1). A␤ is a 39 -43-amino acid peptide that has the propensity to self-assemble into insoluble, ␤-sheet-containing fibrils (1,2). A␤ is proteolytically derived from a large type I integral membrane precursor protein, termed the amyloid ␤-protein precursor (A␤PP), encoded by a gene located on chromosome 21 (3)(4)(5)(6). In this regard, full-length A␤PP is proteolytically cleaved by an enzyme, termed ␤-secretase, at the amino terminus of the A␤ domain. A novel aspartyl proteinase named BACE (for ␤-site A␤PP-Cleaving Enzyme) has been identified as the ␤-secretase enzyme (7)(8)(9)(10). Subsequent cleavage of the remaining amyloidogenic membrane spanning A␤PP carboxyl-terminal fragment by an enzyme termed ␥-secretase liberates the 40-or 42-amino acid residue A␤ peptide. Although the exact identity of ␥-secretase remains unclear studies suggest that the presenilin proteins may function as this enzyme or as a required cofactor for ␥-secretase function (11)(12)(13). Alternatively, full-length A␤PP can be proteolytically processed by an enzyme termed ␣-secretase through the A␤ domain. This cleavage event generates a non-amyloidogenic membrane spanning carboxyl-terminal fragment and truncated secretory forms of A␤PP␣ (sA␤PP␣) that are released into the extracellular environment (14,15).
Cerebrovascular A␤ deposition, known as cerebral amyloid angiopathy, is accompanied by smooth muscle cell degeneration suggesting a toxic effect of A␤ to these cells in vivo (16 -18). These degenerating smooth muscle cells have been implicated in the overproduction of A␤PP and A␤ in the cerebral vessel wall further suggesting the active involvement of these cells in the progression of this cerebrovascular pathology (17)(18)(19). Similar to these in vivo observations, we have reported that A␤- , the more pathogenic form of the wild-type peptide, causes severe cellular degeneration accompanied by a marked increase in the level of cell-associated A␤PP in cultured human cerebrovascular smooth muscle (HCSM) cells (20 -22). In more recent studies we demonstrated that mutations associated with familial forms of cerebral amyloid angiopathy (E22Q Dutch, E22K Italian, and D23N Iowa) markedly enhance both the fibrillogenic and cerebrovascular pathogenic properties of A␤ toward cultured HCSM cells (23)(24)(25)(26). These experiments showed that these pathogenic forms of A␤ assemble into an elaborate network of fibrils on the surfaces of HCSM cells. Furthermore, fibril assembly of pathogenic A␤ on the cell surface is required for inducing downstream pathologic responses in HCSM cells, including cell-surface accumulation of sA␤PP␣, degradation of vascular smooth muscle cell ␣-actin, and ultimately an apoptotic cell death (24 -27).
We have shown that the accumulation of sA␤PP␣ is mediated by its high-affinity binding to the A␤ fibrils that assemble on the HCSM cell surface (28). This event coincides with the induction of smooth muscle cell ␣-actin degradation and cell death. It is noteworthy that an interaction between fibrillar A␤ and A␤PP also has been implicated in neuronal cell death in vitro (29). Finally, A␤ fibril binding to Kunitz proteinase inhibitory (KPI) domain containing forms of A␤PP can enhance its proteinase inhibitory property (30). Together, these findings indicate that interactions between fibrillar A␤ binding and A␤PP may have significant physiological and pathological consequences.
In the present study, we identify the precise site in the cysteine-rich, amino-terminal region of A␤PP that facilitates its binding to fibrillar forms of A␤. Our investigations show that residues His 110 , Val 112 , and Ile 113 , all on a common ␤-strand region within A␤PP, comprise a domain on A␤PP that is involved with its binding to fibrillar A␤ peptides. This domain also mediates the binding of A␤PP to other fibrillar peptides, suggesting that this region may participate in other biologically important interactions.

EXPERIMENTAL PROCEDURES
Materials-A␤ peptides were synthesized by solid-phase Fmoc ((N-(9-fluorenyl)methoxycarbonyl)) amino acid chemistry, purified by reverse phase-HPLC, and structurally characterized as previously described (31). Amylin peptide was obtained from Bachem (San Carlos, CA). For preparation of amyloid fibrils, A␤ peptides or amylin were resuspended to a final concentration of 1.25 mM in 50 mM Tris-HCl, 150 mM NaCl, pH 7.4 and incubated at 37°C for 3 days. The ␤-sheet, fibrillar structure of each peptide was confirmed by circular dichroism spectroscopy and electron microscopy as previously described (24). For cell culture experiments, lyophilized A␤ peptide was first resuspended to a concentration of 250 M in sterile distilled water. Prior to addition to HCSM cells, the peptide was diluted to a final concentration of 25 M in serum-free culture medium. The set of overlapping 15 amino acid peptides spanning A␤PP residues 18 -119 was prepared by Multiple Peptide Systems (San Diego, CA). Wild-type sA␤PP␣-770 was purified as previously described (32) and was biotinylated according to the manufacturer's protocol using the Pierce EZ Sulfo-link Biotin (Rockford, IL). The anti-A␤PP mouse monoclonal antibody (mAb) P2-1, which specifically recognizes an epitope in the amino-terminal region of human A␤PP, was prepared as previously described (33). The anti-A␤PP mAb 22C11 was obtained from Chemicon (Temecula, CA). Secondary peroxidase-coupled sheep anti-mouse IgG and peroxidase-conjugated streptavidin were purchased from Amersham Biosciences. Supersignal Dura West chemiluminescence substrate was purchased from Pierce (Rockford, IL).
Solid-Phase Binding Assay-A␤ peptides or amylin were assembled into fibrils as described above. For most studies we used fibrillar forms of Dutch-type A␤40 since this mutant form of A␤ exhibits enhanced fibrillogenic and pathogenic properties compared with wild-type A␤. Two g of each fibrillar A␤ peptide, fibrillar amylin, or ovalbumin in 100 l of phosphate-buffered saline (PBS) were dried in a 96-well microtiter plate (Corning, Cambridge, MA) overnight at 37°C. The wells were rinsed with PBS three times and blocked with 100 l per well PBS containing 1 mg/ml bovine serum albumin (BSA) for 1 h at room temperature. After rinsing three times with PBS, known concentrations of biotinylated sA␤PP␣ in PBS containing 0.1 mg/ml BSA were incubated in triplicate (100 l per well) for 1 h at room temperature. After rinsing the wells with PBS three times, 100 l of streptavidin conjugated to horseradish peroxidase in PBS containing 0.1 mg/ml BSA (1:800) was added for 1 h at room temperature. The binding of biotinylated sA␤PP␣ was detected using the colorimetric substrate o-phenylenediamine dihydrocholride as described by the manufacturer (Invitrogen). Briefly, the substrate was diluted in buffer (0.1 M sodium citrate, pH 4.5) to a final concentration of 1 mg/ml. H 2 O 2 was added to a final concentration of 0.012% immediately before 100 l of the substrate was added to each microtiter well. The solution was developed for ϳ30 min at room temperature and quenched by the addition of 50 l of 4 N H 2 SO 4 to each well. The conversion of the colorimetric substrate was measured at a wavelength of 490 nm using a Molecular Dynamics V max kinetic plate reader (Sunnyvale, CA).
Alternatively, known concentrations of GST-A␤PP-(18 -119) proteins were added to the wells followed by the anti-A␤PP mAb 22C11 at a dilution of 1 g/ml in PBS containing 0.1 mg/ml BSA and a secondary anti-mouse IgG conjugated to horseradish peroxidase (1:1000). Bound secondary antibody was detected using the colorimetric substrate as described above.
Site-Directed Mutagenesis of GST-A␤PP-(18 -119) Fusion Protein-The pGEX-KG-human A␤PP exon 2-3 fusion construct, encoding A␤PP residues 18 -119, was prepared as previously described (28). Individual amino acids from Cys 105 through Cys 117 within exon 3 were mutagenized to Ala employing single nucleotide change approach using the QuickChange Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA). Briefly, complementary oligonucleotides used in the site-directed mutagenesis were 5Ј phosphorylated, ϳ40 bases in length with 20 matched bases on either side of the point mutation, possessing a T m Ն 78°C and containing at least 40% GC content. The 50-l PCR reaction sample included 40 ng of the pGEX-KG-human A␤PP exon 2-3 fusion construct and 125 ng of the sense and antisense oligonucleotides. The PCR program was as follows: 1 cycle of 95°C for 30 s, 20 cycles of 95°C for 30 s, 55°C for 1 min, 68°C for 20 min, and 1 cycle of 72°C for 10 min. The PCR product was then digested with DpnI for 1.5 h to remove methylated parental DNA as per the manufacturer's protocol. The digested sample was transformed into Epicurian Coli XL1-Blue supercompetent cells and grown on LB agar plates containing 100 g/ml ampicillin. Colonies were chosen and plasmid DNA was isolated and sequenced to confirm the presence of each mutation. In some cases, a second mutation was needed to change the codon to alanine. In these cases, sequential mutagenesis was done using two mutagenesis oligonucleotide sets, the second set with both mutations. The plasmids were sequenced to confirm the presence of the second mutation.
Escherichia coli BL21 cells were transformed with wild-type and mutant pGEX-KG-human A␤PP exon 2-3 fusion constructs, plated on LB agar plates containing 100 g/ml ampicillin and incubated overnight at 37°C, and colonies were picked and used to inoculate 100 ml LB media. Protein expression was induced by the addition of isopropyl ␤-D-thioglucoside to a final concentration of 1 mM for 4 h. Wild-type and mutant GST-A␤PP-(18 -119) fusion proteins were affinity-purified from the harvested cells using glutathione-Sepharose beads as previously described (28). The concentration of the eluted GST-A␤PP-(18 -119) fusion proteins was measured using the extinction coefficient of GST. The precise concentrations of wild-type and each mutant GST-A␤PP-(18 -119) protein were confirmed by titration using mAb 22C11 and compared with standard curves of known concentrations of purified sA␤PP␣. The mAb 22C11 was used for the precise titrations since its epitope on A␤PP and GST-A␤PP-(18 -119) resides upstream of residues 105-117 and was not affected by any of the alanine substitutions inserted in the mutant fusion proteins.
HCSM Cell Culture-Primary cultures of HCSM cells were established and characterized as previously described (34). Two lines of HCSM cells were used in these studies. One was derived from a 70year-old male AD patient and the other was derived from a 37-year-old female control. All HCSM cells were used between passages 4 -7 and maintained in 24-well tissue culture dishes with Dulbecco's minimum essential medium containing 10% fetal bovine serum (Gemini Bio-Products, Calabasas, CA), non-essential amino acids, and antibiotics (Invitrogen). For experiments, near-confluent cultures of HCSM cells were placed in serum-free medium containing 0.1% BSA, non-essential amino acids, and antibiotics overnight prior to treatment. Freshly solubilized Dutch-type A␤-(1-40) at a final concentration of 25 M was added to the cultures in serum-free medium and incubated at 37°C for 6 days. Cells were routinely viewed and photographed using an Olympus IX70 phase-contrast microscope. Cell viability was quantified using a fluorescent live/dead cell assay following the manufacturer's protocol (Molecular Probes, Eugene, OR). The number of live and dead cells were counted from several fields (n ϭ 4) from at least three separate wells for each experiment.

RESULTS
Peptide Mapping of the Fibrillar A␤ Binding Domain in A␤PP-(18 -119)-We showed that the amino-terminal region A␤PP-(18 -119), encoded by exons 2 and 3 of the A␤PP gene, mediates the binding of A␤PP to fibrillar forms of A␤ (28). Here we sought to identify the precise site within this region of A␤PP that is responsible for this interaction. To accomplish this, we synthesized a series of overlapping peptides of 15 amino acids starting at residue 18 of A␤PP with each subsequent peptide shifting 3 amino acids toward the carboxyl-terminal end. This approach covered the entire region of A␤PP-(18 -119) (Fig. 1).
Each of these peptides was tested for its ability to compete against biotinylated sA␤PP␣ for binding to fibrillar A␤ in a solid-phase binding assay. At a relatively high concentration of 1 mM, the peptide A␤PP-(105-119) competed for ϳ80% of the biotinylated sA␤PP␣ binding to fibrillar A␤ ( Fig. 2A, lane 3). In contrast, a 1000-fold lower concentration of GST-A␤PP-(18 -119) completely blocked biotinylated sA␤PP␣ binding to immobilized fibrillar A␤ ( Fig. 2A, lane 2). We also found that overlapping peptide A␤PP-(102-116) possessed some competing activity in the assay, although it was approximately half that of A␤PP-(105-119) (data not shown). However, none of the other 15-mer peptides showed either appreciable or consistent competing activity.
We previously showed that fibrillar A␤ assembled on the surface of HCSM cells mediates the binding of endogenously produced sA␤PP␣ on the cell surface (24,28). Consistent with the results obtained from the solid-phase binding assay, GST-A␤PP-(18 -119) completely inhibited endogenous sA␤PP␣ binding to HCSM cell surface fibrillar A␤, whereas higher concentrations of A␤PP-(105-119) could partially diminish this A␤PP binding (Fig. 2, B and C). The findings that A␤PP-(105-119) peptide was less effective than GST-A␤PP-(18 -119) and that higher concentrations of A␤PP-(105-119) peptide were needed in blocking sA␤PP␣ binding to fibrillar A␤ likely results from this region in A␤PP being highly structured containing three intrachain disulfide bonds (Fig. 1).
Alanine-scanning Mutagenesis of the Putative Fibrillar A␤ Binding Domain in A␤PP-(18 -119)-The peptide-mapping experiments described above indicated that the sequence A␤PP-(105-119) contains the fibrillar A␤ binding domain. However, this small peptide was not nearly effective as the larger, recombinantly expressed A␤PP-(18 -119) in binding to fibrillar A␤. Therefore, we performed alanine-scanning mutagenesis studies of the A␤PP-(105-119) sequence within A␤PP-(18 -119) to further identify key epitopes important in facilitating its binding to fibrillar A␤. Alanine residues were introduced from Cys 105 through Cys 117 of A␤PP-(18 -119) by site-directed mutagenesis of the A␤PP-(18 -119) cDNA. Each mutant A␤PP-(18 -119) cDNA was expressed as a GST fusion protein amd purified, and the concentrations were carefully determined by quantitative immunoblotting as described in "Experimental Procedures". Each purified mutant GST-A␤PP-(19 -118) was tested for its ability to bind immobilized fibrillar A␤ in a solidphase binding assay and compared with the level of binding observed with wild-type GST-A␤PP-(18 -119). As shown in Fig.  3 most of the alanine substitutions had modest inhibitory or enhancing effects on GST-A␤PP-(18 -119) binding to fibrillar A␤. However, alanine substitutions at residues His 110 , Val 112 , or Ile 113 showed highly diminished GST-A␤PP-(18 -119) binding to fibrillar A␤. In particular, H110A mutant GST-A␤PP-(18 -119) exhibited Ͻ10% of fibrillar A␤ binding compared with wild-type GST-A␤PP- (18 -119). The x-ray crystal structure for this amino-terminal region of A␤PP was recently reported (35). Based on a predicted structural fold of this region, it is interesting to note that residues His 110 , Val 112 , and Ile 113 all reside on the same ␤-strand of A␤PP-(18 -119) (Fig. 4). Also observed in these studies was that the C117A mutant GST-A␤PP-(18 -119) showed markedly diminished binding to fibrillar A␤ compared with wild-type GST-A␤PP-(18 -119). Residue Cys 117 participates in a disulfide bond with Cys 73 (Figs. 1 and 4), suggesting that this linkage is important in presenting a properly folded fibrillar A␤ binding domain in A␤PP-(18 -119) likely involving a ␤-strand within this region that includes specific residues His 110 , Val 112 , and Ile 113 . proteins presented in Fig. 3 were performed using fibrillar Dutch-type A␤. We determined whether key mutant GST-A␤PP-(18 -119) proteins similarly exhibited deficient binding to fibrillar wild-type A␤-(1-42) and fibrillar amylin peptide. For these studies we used purified wild-type, H110A mutant, and I113A mutant GST-A␤PP-(18 -119) proteins as shown in Fig. 6A and used in solid-phase binding experiments. As shown in Fig. 6B, H110A and I113A mutant GST-A␤PP-(18 -119) proteins displayed markedly decreased binding to fibrillar Dutch-type A␤-(1-40) compared with wild-type GST-A␤PP- (18 -119). Similarly, both mutant GST-A␤PP-(18 -119) proteins showed considerably reduced binding to fibrillar wild-type A␤-(1-42) and fibrillar amylin peptide compared with wild-type GST-A␤PP-(18 -119). As a control, none of the GST-A␤PP-(18 -119) proteins exhibited any binding to immobilized ovalbumin. These findings suggest that His 110 and Ile 113 , which reside on the same ␤-strand within this region of A␤PP, are key residues for also facilitating A␤PP-(18 -119) binding to fibrillar wildtype A␤ and fibrillar amylin peptides. The results with fibrillar amylin further suggest that the binding of A␤PP-(18 -119) through this domain is not sequence-specific but appears to involve recognition of the fibrillar structure of the peptide. DISCUSSION Deposition of fibrillar A␤ in senile plaques and in the walls of cerebral blood vessels is a key pathological feature of AD, Down's syndrome, and several related cerebral amyloid angiopathy disorders (1,2). These fibrillar A␤ deposits that occur within the brain are intimately associated with neuronal and cerebrovascular cell degeneration at their respective sites (16 -19). Similarly, fibrillar A␤ deposition is involved in neuronal cell and HCSM cell toxicity in vitro (20 -26). Fibrillar A␤ de- posits that occur both in vivo and in vitro cell culture models lead to accumulation of its precursor protein A␤PP (20 -24, 29, 40, 41). We recently reported that fibrillar A␤ mediates the pathological accumulation of A␤PP on the HCSM cell surface through interaction with a domain in the cysteine-rich, aminoterminal region of A␤PP (28). In this study the majority of the HCSM cell-accumulated A␤PP was shown to be sA␤PP␣. Although sA␤PP␣ has been postulated to be a protective molecule, the interaction between fibrillar A␤ and its precursor may have significant implications in cytopathogenic mechanisms in AD and related disorders. For example, this interaction, which results in the accumulation of A␤PP on the HCSM cell surface may contribute to the onset of cell death (28). Likewise, the recent study of Lorenzo et al. (29) has implicated an interaction between fibrillar A␤ and A␤PP in neuronal toxicity in vitro. In light of the potential importance of these findings, we sought to determine the precise site on A␤PP that facilitates its binding to fibrillar A␤.
We previously identified the amino-terminal domain of residues 18 -119 as the region on A␤PP responsible for mediating its binding to fibrillar A␤ (28). Therefore, we synthesized a set of overlapping 15-amino-acid peptides that spanned this region of A␤PP to further localize this site. In competition experiments, the peptide A␤PP-(105-119), located at the extreme carboxyl-terminal end of this amino-terminal region of A␤PP (Fig. 1), was found to compete against A␤PP for binding to both immobilized fibrillar A␤ ( Fig. 2A) and fibrillar A␤ assembled on the surface of HCSM cells (Fig. 2, B and C). However, A␤PP-(105-119) was found to be much less effective than GST-A␤PP- (18 -119) in its ability to compete for A␤PP binding. This disparity is likely caused by the highly structured nature of A␤PP-(18 -119), a region that contains three intrachain disulfide bonds (Figs. 1 and 4). Although the integrity of this structure is preserved in the recombinantly expressed GST-A␤PP- (18 -119), it is unlikely to be properly folded in the small synthetic A␤PP-(105-119) peptide. Nevertheless, these findings suggest the involvement of this focused region on A␤PP-(18 -119) in binding fibrillar A␤.
Because the peptide-competition experiments implicated the region A␤PP-(105-119) as the likely site of a fibrillar A␤ binding domain, we conducted alanine-scanning mutagenesis studies in this region to determine the key residues involved. We decided to perform this analysis in recombinantly expressed GST-A␤PP-(18 -119) fusion proteins since the wild-type GST-A␤PP-(18 -119) protein faithfully recapitulates the binding characteristics of native sA␤PP to fibrillar A␤ (28). Alanine substitutions were made for each amino acid from Cys 105 through Cys 117 of GST-A␤PP- (18 -119). These studies clearly identified His 110 , Val 112 , and Ile 113 as key residues that facilitate GST-A␤PP-(18 -119) binding to fibrillar A␤. It is noteworthy that these three particular residues reside on a predicted common ␤-strand within this region of A␤PP (Fig. 4). It has been suggested that these particular residues, along with Phe 37 , Pro 109 , Phe 111 , and Tyr 115 form a hydrophobic surface patch on A␤PP (35). Our alanine scanning mutagenesis results indicate that P109A, F111A, and Y115A had little or no effect on GST-A␤PP-(18 -119) binding to fibrillar A␤ (Fig. 3). Similarly, little effect on binding was observed with a F37A mutant GST-A␤PP-(18 -119) (data not shown). This suggests that this putative hydrophobic surface patch is not wholly involved with mediating the binding of A␤PP to fibrillar A␤. The finding that the C117A substitution substantially affects GST-A␤PP-(18 -119) binding to fibrillar A␤ further supports the notion that the disulfide bond formed between Cys 73 and Cys 117 is important for properly presenting the ␤-strand of this region containing His 110 , Val 112 , and Ile 113 as a functional fibrillar A␤ binding domain.
Treatment of cultured HCSM cells with pathogenic forms of A␤ results in a protracted period of cellular degeneration leading to apoptotic cell death (20,(23)(24)(25)(26)(27). We previously showed that GST-A␤PP-(18 -119), which contains a functionally active fibrillar A␤ binding domain, blocks cell death in HCSM cells treated with pathogenic Dutch-type A␤-(1-40) (26). GST-A␤PP-(18 -119) may act as a dominant-negative factor containing the site for fibrillar A␤ binding but lacks other downstream regions of A␤PP that mediate a cell-death response. In the present study, we show that in contrast to wild-type GST-A␤PP-(18 -119), the H110A mutant GST-A␤PP-(18 -119), which is deficient in fibrillar A␤ binding, is incapable of protecting HCSM cells from the cytotoxic effects of Dutch-type A␤-(1-40) (Fig. 5). This finding further supports the notion that an interaction between A␤PP and fibrillar A␤ contributes to the cell death response in HCSM cells.
Our earlier studies showed that biotinylated sA␤PP␣ and GST-A␤PP-(18 -119) bound to fibrils formed with either Dutchtype A␤-(1-40) or wild-type A␤-(1-42), but not unassembled forms, indicating that this interaction depends on fibrillar structures of the peptide. Similarly, in the present study we show that H110A mutant and I113A mutant GST-A␤PP-(18 -119) proteins are deficient in binding fibrils formed with either Dutch-type A␤-(1-40) or wild-type A␤-(1-42) (Fig. 6). It is noteworthy that the same pattern of binding was observed when fibrillar amylin peptide was used in the solid-phase binding assay. This finding suggests that the ␤-strand containing residues His 110 , Val 112 , and Ile 113 folds into a binding site that is not specific for the A␤ amino acid sequence but rather possesses recognition for fibrillar structures. Therefore, it is possible that this domain may mediate the binding of A␤PP to other fibrillar structures as well.
The binding of fibrillar A␤, and possibly other fibrillar proteins, to A␤PP through the domain identified here may have several potential consequences. For example, this interaction may help to explain the high levels of A␤PP that accumulate around fibrillar A␤ present in cerebrovascular and, possibly plaque, amyloid deposits (17, 40 -43). In addition, we recently showed that the binding of fibrillar A␤ to KPI-containing forms of A␤PP enhances its coagulation proteinase inhibitory properties (30). The KPI domain resides downstream from A␤PP-(105-119) starting at A␤PP residue 289. This finding suggests that fibrillar A␤ deposits may bind, localize, and stimulate the proteinase inhibitory functions of A␤PP. This activity may have implications regarding hemorrhagic stroke seen in patients with severe cerebral amyloid angiopathy. A␤PP that is produced locally in the cerebral vessel wall or released by circulating activated platelets may accumulate at sites of cerebrovascular A␤ deposition. This would result in a microenvironment high in anticoagulant activity and conducive to hemorrhaging.
More germane to the present work, several studies have reported that treatment of cultured neuronal cells with the mAb 22C11 or a polyclonal antibody (both of which recognize epitopes in the amino-terminal region of A␤PP not far from the identified fibrillar A␤ binding site) can stimulate G-protein activity and/or initiate cell death pathways in vitro (44 -46). It is thought that these responses proceed through the dimerization of A␤PP by divalent antibody binding. Also of note is the recent study of Scheuermann (47) implicating A␤PP dimerization in increased A␤ production. Although antibodies to A␤PP were used in these in vitro studies, clearly other more pathologically relevant agonists must exist in vivo to elicit these potential responses in situations such as AD. In this case the A␤ fibril, and perhaps the protofibril, may be pathological agonists that facilitate A␤PP dimerization on the cell surface stimulating cell death pathways and A␤ production. In regard to the present study, GST-A␤PP-(18 -119) may inhibit pathogenic A␤-induced HCSM cell death by interfering with A␤PP dimerization. This thought is also consistent with our finding that GST-A␤PP-(18 -119) mutants deficient in binding fibrillar A␤ are incapable of blocking pathogenic A␤-induced HCSM cell death.
In summary, the present study has identified the precise site of a fibrillar A␤ binding domain on the extracellular, aminoterminal region of A␤PP. This site involves several key amino acids located on a putative ␤-strand within this region and depends on proper disulfide bonding. The interaction of A␤PP with fibrillar A␤ mediated through this site may contribute to the pathologic accumulation of A␤PP observed at sites of fibrillar A␤ deposition that occur in vitro on the cultured HCSM cell surface and in vivo in the cerebral vessel walls of patients with severe cerebral amyloid angiopathy. This pathologic fibrillar A␤-A␤PP interaction may provide further insight into the mechanisms that lead to cerebrovascular and, possibly neuronal, cellular degeneration observed in AD and related disorders.