Neurotoxicity and Physicochemical Properties of A (cid:1) Mutant Peptides from Cerebral Amyloid Angiopathy IMPLICATION FOR THE PATHOGENESIS OF CEREBRAL AMYLOID ANGIOPATHY AND ALZHEIMER’S DISEASE*

Cerebral amyloid angiopathy (CAA) due to (cid:1) -amyloid (A (cid:1) ) is one of the specific pathological features of familial Alzheimer’s disease. A (cid:1) mainly consisting of 40- and 42-mer peptides (A (cid:1) 40 and A (cid:1) 42) exhibits neurotoxicity and aggregative abilities. All of the variants of A (cid:1) 40 and A (cid:1) 42 found in CAA were synthesized in a highly pure form and examined for neurotoxicity in PC12 cells and aggregative ability. All of the A (cid:1) 40 mutants at positions 22 and 23 showed stronger neurotoxicity than wild-type A (cid:1) 40. Similar tendency was observed for A (cid:1) 42 mutants at positions 22 and 23 whose neurotoxicity was 50–200 times stronger than that of the corresponding A (cid:1) 40 mutants, suggesting that these A (cid:1) 42 mutants chromatography; MALDI-TOF-MS, matrix-assisted laser desorption/ionization time-of-flight mass spectrometry; PEG, polyethylene glycol; MTT, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphe-nyltetrazolium bromide; Th-T, thioflavin-T; PS, polystyrene support.

Alzheimer's disease (AD) 1 is neuropathologically characterized by the progressive deposition of amyloid in the brain parenchyma and cortical blood vessels (1). This deposition mainly consists of 40-and 42-mer ␤-amyloid peptides (A␤40 and A␤42) generated from amyloid precursor protein by two proteases, ␤and ␥-secretases (2,3). Cerebral amyloid angiopathy (CAA) in familial Alzheimer's disease is linked to missense mutations inside the A␤-coding region in the amyloid precursor protein. The mutations of A␤ sequence are concentrated at positions 21-23 and are called Flemish (A21G) (4), Arctic (E22G) (5,6), Dutch (E22Q) (7), Italian (E22K) (8), and Iowa (D23N) (9) mutations. These A␤ mutant peptides may play a pathological role in the CAAs because wild-type A␤ peptides induce neuronal death in vitro (10). Neurotoxicity and formation of amyloid fibrils of some CAA-related A␤40 mutants have been independently reported by several groups (11)(12)(13)(14)(15). However, there are no reports on the neurotoxicity and aggregation of the CAA-related A␤42 mutants with the exception of Dutch mutation (E22Q) (11), the investigation of which is essential to reveal the mechanism of CAA because wild-type A␤42 shows considerably stronger neurotoxicity and aggregative ability than wild-type A␤40 (11). Moreover, it is indispensable to simultaneously compare neurotoxicity and aggregative ability of all of the CAA-related A␤40 and A␤42 mutants in the same conditions such as pH, peptide concentration, reaction buffer, and temperature.
It is difficult to synthesize A␤42 with 14 hydrophobic and/or bulky amino acid residues at the C terminus in a highly pure form, because it easily aggregates even under weakly acidic and neutral conditions (16). We recently established a highly efficient method for synthesizing long peptides over 50 amino acid residues with a continuous flow-type peptide synthesizer (Pioneer TM ) using N-[(dimethylamino)-1H-1,2,3-triazolo [4,5b]pyridin-1-ylmethylene]-N-methylmethanaminium hexafluorophosphate N-oxide (HATU) (17) as an effective coupling reagent for Fmoc chemistry (18 -22). This enabled us to obtain the CAA-related A␤42 mutants with high purity in combination with the purification under the alkaline condition (23). Our continuous research on the CAA-related A␤ peptides led to the synthesis of all of the CAA-related A␤40 and A␤42 mutants at positions 21-23 ( Fig. 1). Because Lys-Asp (Italian), Gln-Asp (Dutch), Gly-Asp (Arctic), and Glu-Asn (Iowa) sequences at positions 22 and 23 of CAA-related A␤ mutants are frequently found in a two-residue ␤-turn (24), we synthesized several A␤40 and A␤42 derivatives at position 22 substituted by the amino acid residues, which influence the formation of the tworesidue ␤-turn as follows: 1) proline and serine residues in the first position as ␤-turn inducers and 2) valine and leucine residues as ␤-turn breakers (24). This paper describes a comprehensive study that shows the neurotoxicity in PC12 cells, aggregative ability, and secondary structure of a series of A␤ derivatives at positions 21-23 including all of the CAA-related mutants and discusses the contribution of the CAA-related mutation to the implication of the pathogenesis of CAA and AD and to the secondary structure at positions 22 and 23 of A␤ peptides.
Synthesis of A␤ Derivatives-Each A␤ derivative ( Fig. 1) was synthesized in a stepwise fashion on 0.1 mmol of preloaded Fmoc-Val-PEG-PS (for A␤40 derivatives) or Fmoc-Ala-PEG-PS (for A␤42 derivatives) resin by Pioneer using the Fmoc method as reported previously (18 -22). The coupling reaction was carried out using Fmoc amino acid (0.4 mmol), HATU (0.4 mmol), and DIPEA (0.8 mmol) in DMF for 30 min. After each coupling reaction, the N terminus Fmoc group was deblocked with 20% piperidine in DMF.
After completion of the chain elongation, each peptide-resin washed with DMF and CH 2 Cl 2 was treated with a mixture containing trifluoroacetic acid, m-cresol, ethanedithiol, and thioanisole for final deprotection and cleavage from the resin. After 2 h of shaking at room temperature, the crude peptide precipitated by diethyl ether was purified by HPLC under alkaline conditions as reported previously (19,22,23). Lyophilization gave a corresponding pure A␤ peptide, the purity of which was confirmed by HPLC (Ͼ98%).
Cell Culture of PC12 Cells-Rat pheochromocytoma PC12 cells were obtained from Riken Cell Bank and were cultured as reported previously (23). For experimental purposes, near-confluent cultures of the cells were plated at ϳ10 4 cells/100 l/well fresh culture medium in 96-well tissue culture plate coated with collagen and incubated at 37°C under 5% CO 2 overnight before experiments.
MTT Assay Using PC12 Cells-Reduction of MTT by mitochondrial reductase was carried out by the protocols based on a previous report (25) with slight modifications. Each solution of A␤ derivatives (0.1% NH 4 OH) sterilized by the filter (0.22 m) was diluted with 0.1% NH 4 OH at concentrations ranging from 0.12 to 120 M. 10 l of the resultant solution and 10 l of 50 mM sodium phosphate, pH 7.4, containing 100 mM NaCl were, respectively, added to the above-mentioned 100-l cell culture, which was incubated at 37°C under 5% CO 2 for 48 h. After removal of 30 l from the medium, 10 l of 5 mg/ml MTT in the phosphate buffer solution described above was added to the cell culture, which was incubated at 37°C under 5% CO 2 for 4 h. After evacuation of the culture medium, the cell lysis buffer (100 l/well, 10% SDS, 0.01 M NH 4 Cl) was subsequently added to the cells and the cell lysate was incubated overnight in the dark at room temperature. The colorimetric determination of MTT was made at 600 nm. The absorbance obtained by the addition of vehicle was taken as 100%.
Sedimentation Assay for Fibril Formation-Each A␤ derivative was dissolved in 0.02% NH 4 OH at 250 M. After a 10-fold dilution by 50 mM sodium phosphate containing 100 mM NaCl at pH 7.4, the resultant peptide solution (25 M) was incubated at 37°C for 4, 8, 16, 24, or 48 h. After centrifugation at 15,000 rpm in an Eppendorf microcentrifuge at 4°C for 10 min, 25 l of the supernatant was then analyzed by HPLC as reported previously (23). The area of the absorption at 220 nm was integrated and expressed as a percentage of the control.
Th-T Fluorescence Assay-Each A␤ derivative was dissolved in 0.02% NH 4 OH at 250 M. The peptide solution (25 M) diluted with the phosphate buffer solution described above was incubated at 37°C for 4, 8, 16, 24, or 48 h. Ten microliters of each A␤ solution was added to 1 ml of 5 M Th-T in 50 mM Gly-NaOH, pH 8.5. Fluorescence intensity was measured at 450-nm excitation and 482-nm emission as reported previously (26).
Transmission Electron Micrographs of Negatively Stained Preparations of A␤42 Fibrils-The fibril formation of the A␤42 derivatives was detected by electron microscope. Each A␤42 derivative (25 M) was incubated in 50 mM phosphate buffer, pH 7.4, containing 100 mM NaCl for 48 h at 37°C. After centrifugation, the supernatant was removed from pellets. Aggregates were then suspended in water by gentle vortex mixing. These suspensions were applied to a 400-mesh collodion-coated copper grid (Nissin EM, Tokyo, Japan) and allowed to dry in air before being negatively stained for 2 min with 2% uranyl acetate. Fibrils were examined with the JEOL JEM-2000EX electron microscope.
ATR-FTIR Measurements to Estimate the Secondary Structure of A␤42 Derivatives-Each A␤42 derivative was dissolved in 50 mM sodium phosphate buffer, pH 7.4, containing 100 mM NaCl at 250 M. The A␤42 solution was lyophilized or incubated at 37°C for 48 h. The 48-h incubated sample was centrifuged at 15,000 rpm in an Eppendorf microcentrifuge at 4°C for 10 min, and the supernatant was removed. The pellet was suspended in distilled water, centrifuged, and lyophilized to give a white powder, which was loaded on an ATR cell in the FTIR spectrometer (Jasco 480plus). At the same time, each A␤42 derivative without incubation was also subjected to the ATR-FTIR measurement. 50 scans were accumulated to improve the signal/noise ratio, and the spectra were recorded at a resolution of 4.0 cm Ϫ1 . The program IR-SSE (Jasco) was used to estimate the secondary structure of each derivative by principal component regression.

Synthesis of A␤ Derivatives-
The CAA-related A␤40 and A␤42 mutants at positions 21-23 were synthesized to examine their relation to the pathogenesis of these CAAs (Fig. 1). Several A␤ derivatives in which the Glu-22 residue of wild-type A␤ was substituted by amino acid residues (proline, serine, valine, and leucine) that effectively influence the formation of the two-residue ␤-turn (24) were also prepared to investigate the secondary structure of A␤ derivatives at positions 21-23 ( Fig.  1). These peptides were synthesized with continuous flow-type peptide synthesizer (Pioneer) using HATU as an effective activator for Fmoc chemistry. After final deprotection and cleavage from the resin followed by purification using HPLC in the alkaline condition (CH 3 CN, 0.1% NH 4 OH), each A␤ derivative was successfully taken in a highly pure form as reported previously (19,22,23). Total yields of these derivatives were consequently between 2 and 24%, indicating that average coupling yield of each condensation step was 95-97%. Their molecular weights were confirmed by MALDI-TOF-MS (Table I), and their purity was determined by HPLC analysis (Ͼ98%).
Neurotoxicity Studies Using PC12 Cells-Neurotoxicity of the A␤ derivatives was estimated by the MTT assay (25). Effect of the A␤ derivatives on the mitochondrial reductase, which converts MTT to colored formazan, was examined by exposing the cells to the peptides and measuring the cellular proliferation and viability at various time points. We used rat pheochromocytoma PC12 cells, which are particularly sensitive to A␤ peptides (25). The IC 50 value for inhibition of formazan formation by each A␤ peptide was calculated using the probit procedure (27). The concentration of the A␤ derivatives to inhibit the formazan formation ranged from 10 Ϫ8 to 10 Ϫ5 M as shown in Table II.
Wild-type A␤40 did not inhibit the formazan formation even at 10 Ϫ5 M, whereas wild-type A␤42 considerably inhibited it at 10 Ϫ6 M. All of the CAA-related A␤40 mutants at position 22 inhibited significantly the formazan formation at 10 Ϫ5 M. On the other hand, A21G-A␤40 (Flemish) did not show any neurotoxic effect even at 10 Ϫ5 M. The CAA-related A␤42 mutants at position 22 (Dutch, Italian, and Arctic) also showed approximately 10-fold the neurotoxicity of wild-type A␤42. A concentration of 50% inhibition of the formazan formation was approximately 10 Ϫ7 M. D23N-A␤42 (Iowa) mutant at position 23 showed 2-3-fold more potent cytotoxicity than wild-type A␤42. A21G-A␤42 (Flemish) showed a slightly weaker yet still significant neurotoxic effect on the PC12 cells compared with wildtype A␤42 (Table II).
Aggregation Studies-The aggregative ability of A␤ can be estimated by sedimentation assay using HPLC after centrifugation of the A␤ peptide solution. A␤ aggregates are also detectable using Th-T, which binds specifically to the aggregates to show intense fluorescence (28). One drawback of the Th-T assay is that it is not always quantitative because Th-T fluorescence can vary depending on the structure and morphology of the fibrils. The sedimentation assay is more suitable for quantification of the aggregates.
Almost of all soluble wild-type A␤40 was detected even after 24-h incubation by the sedimentation assay ( Fig. 2A), indicating that wild-type A␤40 hardly aggregated by 24-h incubation. This result is in good agreement with that of the Th-T assay (Fig. 2B). The fluorescent intensity of A␤40 was almost equal to that of the control. On the other hand, soluble wild-type A␤42 could not be detected after 24-h incubation by the sedimentation assay (Fig. 3A), indicating that wild-type A␤42 aggregated completely after 24-h incubation. This result correlated very well with that of the Th-T assay where the fluorescent intensity already reached plateau after 16-h incubation (Fig. 6B). These results indicate that the aggregative ability of A␤42 is far more potent than that of A␤40, and their aggregative ability correlates very well with their neurotoxicity.
Among all of the CAA-related A␤40 mutants at position 22, E22Q-A␤40 (Dutch), E22K-A␤40 (Italian), and E22G-A␤40 (Arctic) aggregated faster than wild-type A␤40 as shown in Fig.  2A and the aggregative rate of the A␤40 mutants at position 22 was larger than that of the A␤40 mutant at position 23. Reflecting the significant difference in the aggregative ability between A␤40 and A␤42, the aggregative ability of all of the A␤40 mutants at position 22 failed to exceed that of wild-type A␤42. E22G-A␤40 showed a potent Th-T fluorescence while the Th-T intensity of E22Q-and E22K-A␤40 did not reflect their high aggregative rate in the sedimentation assay (Fig. 2B). The aggregative potency of A21G-A␤40 (Flemish) was lowest among the A␤40 mutants ( Fig. 2A), and the Th-T fluorescence of this peptide was also weaker than that of wild-type A␤40 (Fig. 2B). The aggregative ability of all of the A␤40 mutants estimated by the sedimentation assay correlated generally with their neurotoxicity (Table II). Aggregative ability of the CAA-related A␤42 mutants is shown in Fig. 3. E22Q-A␤42 (Dutch) and E22K-A␤42 (Italian) aggregated faster than wild-type A␤42 in the sedimentation assay (Fig. 3A). Although the aggregative potency of E22G-A␤42 (Arctic) and D23N-A␤42 (Iowa) was slightly lower than that of wild-type A␤42, the aggregative rate of E22G-A␤42 after 4-h incubation was almost similar to that of wild-type A␤42 and the aggregated amount of D23N-A␤42 after 24-h incubation was nearly comparable with that of wild-type A␤42 (Fig. 3A). Th-T fluorescence intensity of all of the A␤42 mutants at positions 22 and 23 after 4-h incubation was almost similar, and these results did not reflect exactly those of the sedimentation assay (Fig. 3B). The aggregative ability of A␤42 mutants at positions 22 and 23 examined by sedimentation assay correlated generally with their neurotoxicity (Table II). However, A21G-A␤42 (Flemish) with a substantial neurotoxicity (Table II) aggregated hardly after 24-h incubation (Fig. 3A) and showed no Th-T fluorescence after 4-h incubation (Fig. 3B). Only this peptide did not follow the general pattern that the aggregative ability of the peptide correlates with its neurotoxicity.
Transmission Electron Micrographs of Negatively Stained Preparations of A␤42 Fibrils-Fibril formation of the A␤42 mutants was evaluated by transmission electron microscopy after a 48-h incubation at 37°C (Fig. 4). Typical fibrils were formed in all of the CAA-related A␤42 mutants with the exception of A21G-A␤42 (Flemish), which did not aggregate at all as described above. The morphology of these fibrils resembled each other well. However, the length of the D23N-A␤42 (Iowa) fibril was slightly shorter than that of the others.
ATR-FTIR Measurements of the A␤42 Derivatives-A␤ aggregates are generally considered to contain a significant amount of the ␤-sheet structure (29). To estimate the secondary structure of the A␤ mutants, ATR-FTIR measurement was carried out on the CAA-related A␤42 peptides. Each A␤42 mutant lyophilized immediately after dissolving in 50 mM sodium phosphate buffer, pH 7.4, containing 100 mM NaCl or after a 48-h incubation in the same sodium phosphate buffer at 37°C was subjected to the ATR-FTIR analysis (Fig. 5). The amide I profiles of the freshly prepared A␤42 mutants at positions 22 and 23 exhibited a major absorbance between 1620 and 1640 cm Ϫ1 (Fig. 5A) corresponding to the ␤-sheet structure (30). The secondary structures calculated from the amide I absorption in these spectra using the program IR-SSE (Jasco) are summarized in Table III. All of the CAA-related A␤42 mutants at positions 22 and 23 had slightly higher ␤-sheet content (49 -56%) than wild-type A␤42 (48%) when freshly prepared. The ␤-sheet content of E22K-A␤42 (56%) with potent aggregative ability was highest, whereas that of A21G-A␤42 (44%) with quite low aggregative potency was significantly lower than that of the other A␤ mutants and 5% ␣-helix was detected instead for ␤-sheet.
In all of the fibrils of the A␤ mutants, intensity of the absorption between 1620 and 1640 cm Ϫ1 increased significantly (Fig. 5B), indicating that the ␤-sheet content increased by fibril formation. The ␤-sheet content of the fibrils of the A␤ mutants reflected very well the aggregative potency estimated by the sedimentation assay. The ␤-sheet content of the E22K-A␤42 fibrils was especially high, whereas that of the E22G-A␤42 and D23N-A␤42 was slightly lower than that of wild-type A␤42.
Neurotoxicity, Aggregation, Secondary Structure, and Transmission Electron Microscopy of Several A␤ Derivatives at Position 22-Because Lys-Asp (Italian), Gln-Asp (Dutch), Gly-Asp (Arctic), and Glu-Asn (Iowa) sequences are often found in the two-residue ␤-turn (24), high aggregative ability of these CAArelated A␤40 and A␤42 mutants suggests that ␤-turn formation at positions 22 and 23 might correlate with the aggregative ability of A␤ peptides. To confirm this speculation, several A␤40 and A␤42 derivatives at position 22, substituted by amino acid residues that induce or suppress the formation of two-residue ␤-turn, were synthesized (Fig. 1). Proline and serine residues at the first position are known as ␤-turn inducers, whereas valine and leucine residues are known as ␤-turn breakers (24). E22P-A␤42 and E22P-A␤40 showed stronger neurotoxicity with potent aggregative ability than the corresponding wild-type A␤42 and A␤40, respectively (Table II and Fig. 6). The neurotoxicity and aggregative ability of E22P-A␤42 were considerably stronger than those of E22P-A␤40 and were comparable with those of E22Q-A␤42 (Dutch) and E22K-A␤42 (Italian) (Table II and Figs. 3 and 6).
ATR-FTIR spectra of freshly prepared E22P-A␤42 indicated a major absorbance between 1620 and 1640 cm Ϫ1 along with a little absorbance between 1640 and 1700 cm Ϫ1 because of turn and random structure (Fig. 5A). Calculation of the spectra using the IR-SSE program showed higher ␤-sheet content (53%) than wild-type A␤42 (48%), correlating with its higher neurotoxicity and aggregative ability. E22P-A␤42 also formed typical fibrils quite similar to those of wild-type A␤42 as shown in Fig. 4. High ␤-sheet content (66%) of E22P-A␤42 fibrils was suggested by the calculation of the spectrum (Fig. 5B), indicating that the turn formation at positions 22 and 23 increased significantly ␤-sheet content of these A␤ fibrils. On the other hand, the neurotoxicity and aggregative ability of E22V-A␤42 were quite low. Because a valine residue is scarcely found in the two-residue ␤-turn, turn formation at positions 22 and 23 seems to be critical for the neurotoxicity and aggregative ability of A␤ peptides. Unfortunately, we failed to measure the neurotoxicity and aggregative ability of E22S-A␤42 and E22L-A␤42 because of their poor solubility in the sodium phosphate buffer.

DISCUSSION
Neurotoxicity of A␤ is significantly implicated in the pathogenesis of neuronal degeneration in AD. Despite many previous studies on A␤ peptide-derived neurotoxicity, the precise mechanism has not yet been clarified. Because it is generally accepted that wild-type A␤ shows neurotoxicity in vitro mainly through aggregation, inhibition of the aggregative process might be a promising therapeutic strategy. To reach this goal, it is indispensable to reveal the pathological conformation of A␤ peptides that leads to neurotoxicity and aggregation. We focused on the inside mutations of A␤ in CAA because these mutations may be related to the neurotoxicity of the variant A␤ peptides. According to the method recently established for long peptide synthesis (18 -23), all of the variant forms of A␤40 and A␤42 were synthesized in a highly pure form.
The CAA mutations at positions 22 and 23 of A␤ peptides enhanced their neurotoxicity, suggesting that these A␤ mutants are involved in the pathogenesis of the CAAs. Especially, the neurotoxicity of the A␤42 mutants at positions 22 and 23 were considerably higher than that of the corresponding A␤40 mutants, which never exceeded that of the wild-type A␤42. These results indicate that the A␤42 mutants are crucial determinants in the pathogenesis of these CAAs. The neurotoxicity of E22Q-A␤42 and E22K-A␤42 was especially potent (Table II), supporting that Dutch and Italian mutations cause hereditary cerebral hemorrhage with amyloidosis. The Dutch and Italian patients develop cerebral hemorrhage, whereas parenchymal amyloid deposits are rare and neurofibrillary tangles are constantly absent. This appearance of the Dutch and Italian patients clearly differs from that of the Arctic and Iowa patients (12) and coincides well with the remarkable aggregative ability of the corresponding A␤ mutants.
It has recently been reported that E22Q-A␤40 rather than E22Q-A␤42 plays a significant role in Dutch-type CAA because E22Q-A␤42 did not show any cytotoxic effects (11). However, we could clearly demonstrate the most potent cytotoxicity of E22Q-A␤42 among the A␤42 mutants. In our assay system, wild-type A␤42 aggregated far more rapidly than wild-type A␤40, differing slightly from the data of other groups (11,(31)(32)(33). Establishment of a practical and reliable synthetic method of pure A␤42 peptides (19,22,23) would be one of the major reasons for these discrepancies. Our synthetic method enabled precise estimation of the neurotoxicity and aggregative ability of all of the CAA-related A␤ mutants simultaneously in the same conditions for the first time.
It is quite noteworthy that A21G-A␤42 (Flemish) showed a substantial albeit weak neurotoxicity in PC12 cells, although the aggregative ability of A21G-A␤42 was significantly lower than that of wild-type A␤42. A21G-A␤42 is the only exception in which the neurotoxicity fails to correlate with the aggregation. It has recently been reported that the neurotoxicity of A␤ peptides might be the result of the oligomeric species, not the fibrils (34,35). If the oligomeric A21G-A␤42 peptides are attributable to the neurotoxicity, aggregation to form fibrils would not be a requisite condition for the Flemish-type CAA.
Tsubuki et al. (36) speculate the cause of AD as the catabolism rather than the anabolism of A␤ since the catabolic dysregulation of A␤ well accounts for the neuronal degeneration. Recently, they reported (36) that A21G-A␤42 as well as other CAA-related A␤ mutants at position 22 is hardly decomposed by the degradation enzyme of A␤ peptides. This catabolic characterization of A21G-A␤42 might be an alternative cause of Flemish-type CAA. The CAAs at positions 22 and 23 other than Flemish-type may also be explained by this degradation scheme. The lack of the cytotoxicity of A21G-A␤40 is reasonable because the cytotoxicity of wild-type A␤40 is quite low in our assay system.
There was a good correlation between the neurotoxicity in PC12 cells and the aggregative ability determined by the sedimentation assay of the CAA-related A␤40 and A␤42 mutants at positions 22 and 23 ( Figs. 2A and 3A and Table II). This finding suggests that these A␤ mutants showed neurotoxicity through the aggregative process. Because Th-T recognizes a specific structure of amyloid fibrils, aggregative ability estimated by sedimentation would not always correlate with the Th-T fluorescence. We think that the aggregative ability itself can be more precisely estimated by the sedimentation assay than by the Th-T fluorescence assay. The discrepancies of the results between the sedimentation and Th-T assay would not imply that most of the aggregates generated by A␤ mutants do FIG. 5. A, ATR-FTIR spectra of freshly prepared A␤42 derivatives. B, ATR-FTIR spectra of fibrils of the A␤42 derivatives after 48-h incubation at 37°C. 50 scans were accumulated to improve signal/noise ratio, and the spectra were recorded at a resolution of 4.0 cm Ϫ1 . not correspond to amyloid fibrils since the FTIR spectra and morphology of these fibrils were similar to those of the wildtype A␤42 fibrils (Figs. 4 and 5).
The electron microscope measurements clearly showed fibrillar materials for all of the CAA-related A␤42 mutants at positions 22 and 23 (Fig. 4). Recently, Miravalle et al. (12) reported that the aggregates of E22K-A␤40 (Italian) showed less fibrillar organization and a rather amorphous organization regardless of its strong neurotoxicity. Their report correlates well with lower Th-T fluorescence of E22K-A␤40 compared with wild-type A␤40 (Fig. 2B). Extremely rapid aggregative rate of E22K-A␤42 peptide should result in amorphous rather than fibrillar organization. However, we could not detect the intrinsic difference in morphology between E22K-A␤42 and wild-type A␤42 (Fig. 4). This inconsistency might be because of the accidental failure of detection of the amorphous.
ATR-FTIR analysis of freshly dissolved A␤42 derivatives indicated that ␤-sheet content of all of the CAA-related A␤42 mutants at positions 22 and 23 (49 -56%) was slightly higher than that of the wild-type A␤42 (48%), supporting the previous hypothesis that ␤-sheet structure is closely related to the neurotoxicity and aggregative ability of A␤ peptides. The FTIR analysis of these mutant fibrils showed that aggregation of these A␤42 peptides increased significantly their ␤-sheet content. Because the ATR-FTIR profiles of the fibrils of the CAArelated A␤42 mutants are almost similar to that of wild-type A␤42, these fibrils would adopt a similar tertiary structure as exemplified by the electron microscopic analysis. Although the shape of the fibrils derived from D23N-A␤42 was a little bit different from the other A␤42 mutants, the intrinsic difference was not detected by the ATR-FTIR measurement.
This work suggests that pathogenesis of the CAAs is related to the alteration of the amino acid residues at position 22 of the A␤ peptides. Miravalle et al. (12) and Melchor et al. (13) proposed that a change or loss of charge at position 22 enhanced the pathogenic properties of the A␤40 peptides. However, our results using A␤42 derivatives at position 22 contradict their hypothesis because E22V-A␤42 without charge at position 22 showed very weak neurotoxicity in PC12 cells and aggregative ability. Moreover, the neurotoxicity and aggregative ability of E22A-A␤42 were almost similar to those of E22D-A␤42 and a little bit weaker than those of the wild-type A␤42 as reported previously (23). These results led us to speculate that the steric factor of the side chain at position 22 would change the conformation of A␤ peptides to affect their neurotoxicity and aggregative ability.
It is noteworthy that the fibrils as well as freshly prepared A␤42 mutant peptides contain a turn structure in addition to the ␤-sheet structure (Table III). Because Lys-Asp (Italian), Gln-Asp (Dutch), Gly-Asp (Arctic), and Glu-Asn (Iowa) sequences are frequently found in the two-residue ␤-turn (24), A␤42 might form the two-residue ␤-turn at positions 22 and 23, which eventually leads to intramolecular and/or intermolecular ␤-sheet structure that causes the neurotoxicity and aggregative ability. Several A␤ derivatives at position 22 were prepared to confirm this speculation. The neurotoxicity and aggregative ability of E22P-A␤42, E22P-A␤40, and E22V-A␤42 strongly supported this hypothesis because prolines most frequently occur in the first position of the two-residue ␤-turn as a Pro-X corner (24,37) and bulky hydrophobic residues like valine and leucine are rarely found in the two-residue ␤-turn (24). D23N-A␤ peptides are only examples of the mutation at position 23. It is interesting that Asn residue most frequently occurs at the second position of the two-residue ␤-turn (24). The D23N mutation does not contradict the ␤-turn formation at positions 22 and 23. Moreover, our recent results on the proline replacement at positions 19 -21 and 24 -26 of A␤42 suggest that these amino acid residues are involved in the ␤-sheet formation (38). This is a strong circumstantial evidence that ␤-turn at positions 22 and 23 of A␤42 enhances the neurotoxicity and aggregative ability. On the basis of these considerations, we propose the pathological conformation of A␤42 peptides at positions 21-24 as shown in Fig. 7, A and B.
After the completion of this study, a structural model for   Fresh  A␤42  0  48  25  27  A21G-A␤42  5  44  24  27  E22Q-A␤42  0  49  24  27  E22K-A␤42  0  56  21  23  E22G-A␤42  0  49  24  27  D23N-A␤42  0  50  24  26  E22P-A␤42  0  53  22  25  Fibrils  A␤42  0  58  20  22  E22Q-A␤42  0  58  20  22  E22K-A␤42  0  69  15  16  E22G-A␤42  0  56  21  23  D23N-A␤42  0  54  22  24  E22P-A␤42  0  66  16  18 ␤-amyloid fibrils derived from A␤40 has been proposed using solid state NMR spectroscopy ( Fig. 7C and Ref. 39). The model shows that the residues 25-28 contain a bend of the backbone that brings the two ␤-sheets in contact through side chain-side chain interactions. Turn formation at positions 26 and 27 is reasonable because Ser-Asn sequence is often found in the two-residue ␤-turn (24). Although our present results support the presence of the bend structure, the positions are slightly different from the Tycko's model (39). To examine the bend structure at positions 25-28, G25P-A␤42, S26P-A␤42, N27P-A␤42, and K28P-A␤42 peptides were synthesized and examined for their aggregative ability. However, these peptides hardly aggregated after 24 h of incubation, unlike wild-type A␤42 estimated by the sedimentation assay, and were far less neurotoxic than wild-type A␤42 (data not shown), suggesting that positions 25-28 of A␤42 are involved in the ␤-sheet formation rather than the hairpin formation because prolines are rarely present in the ␤-sheets (24,37). Because A␤42 is highly neurotoxic than A␤40, the pathological conformation of A␤42 might be different from that of the A␤40. It is also possible that the pathological conformation of the CAA-related A␤ peptides, especially E22K-A␤40 (Italian), is different from that of wildtype A␤40. In the Tycko's model (Fig. 7C) 7B). This conformational change, which increases the intermo-lecular parallel ␤-sheet region, would enhance the aggregative ability of A␤ peptides. In summary, the present results using highly pure A␤ mutants indicated that A␤42 mutants at positions 22 and 23 play a more crucial role in the pathogenesis of the CAAs than the corresponding A␤40 mutants. The secondary structure analysis of these A␤ derivatives and the systematic proline replacement of A␤42 led to the new hypothesis that ␤-turn at positions 22 and 23 is the key secondary structure related to neurotoxicity and aggregative ability of A␤ peptides. This hypothesis could also explain why many CAA mutations are concentrated at positions 22 and 23 of A␤ peptides. The large difference in neurotoxicity and aggregative ability of A␤42 and A␤40 peptides also implies the importance of the C-terminal two residues. The present results unambiguously indicate that the residues at positions 22 and 23 and 41 and 42 should be taken into account to clarify the precise aggregation mechanism of A␤ peptides and to develop new medicinal leads, which inhibit the aggregation of A␤. Systematic proline replacement of A␤ peptides, one of the promising approaches to this goal, along with the solid state NMR study of A␤42 is under investigation. Acknowledgment-We thank Mr. Naoto Kanbayashi at Jasco Engineering for the ATR-FTIR measurement.
Note Added in Proof-Approximately 50 -100 g of each peptide sample was spread evenly on the surface of a horizontal single reflection diamond ATR plate. The data in Fig. 5 were collected using JASCO FT/IR-480 plus (Jasco) with a DLATGS detector in the following conditions: resolution, 4.0 cm Ϫ1 ; scanning speed, 2 mm/s; apodization, cosine; and one level of zero filling. The conspicuous artifacts of the spectra would be ascribable to water in a sample, although all of the samples were lyophilized sufficiently. In estimation of the secondary structure of each A␤42 derivative, the peak of the water vapor was subtracted. The difference between previously published spectra of wild-type A␤42 (for example, Janek, K., Rothemund, S., Gast, K., Beyermann, M., Zipper, J., Fabian, H., Bienert, M., and Krause, E. (2001) Biochemistry 40, 5457-5463) and that in this paper might be because of various measurement conditions (solvent, concentration, pH, and so on) and/or A␤42 employed. We used highly pure A␤42 prepared by high pressure liquid chromatography under the basic conditions. Commercially available A␤42 is usually purified under the acidic conditions where the peak of A␤42 is too broad to be purified strictly. The program IR-SSE (Jasco) for estimation of peptide secondary structure was made according to the previously reported method (Sarver, R. W., Jr., and Krueger, W. C. (1991) Anal. Biochem. 194, 89 -100). In this paper, the IR data matrices were constructed from the normalized FTIR spectra from 1700 to 1600 cm Ϫ1 of 17 proteins and the secondary structure matrices (␣-helix, ␤-sheet, turn, and random) were constructed from the x-ray data of the 17 proteins. There is a possibility that the calculated ␤-sheet contents, especially turn and random coil, are not always true because it has never been demonstrated that estimation of the secondary structure of A␤ peptides can be obtained from the amide I spectra. However, we have another circumstantial evidence of increase of the ␤-sheet structure in several A␤42 mutants by CD spectra with the exception of E22K-A␤42, which aggregates very rapidly (23). The mean Ϯ S.D. of the estimation in Table III was within 2% when the same sample was measured several times. When wild-type A␤42 was prepared on another day by the same procedure, the mean Ϯ S.D. was 1.5%. In ATR-FTIR measurements, there is an optical distortion due to dispersion that would affect absolute values of percentage of the secondary structures. This paper discusses only the relative increase of the ␤-sheet contents compared with wild-type A␤42.