Effects of Aβ-derived peptide fragments on fibrillogenesis of Aβ

Amyloid β (Aβ) peptide aggregation plays a central role in Alzheimer’s disease (AD) etiology. AD drug candidates have included small molecules or peptides directed towards inhibition of Aβ fibrillogenesis. Although some Aβ-derived peptide fragments suppress Aβ fibril growth, comprehensive analysis of inhibitory potencies of peptide fragments along the whole Aβ sequence has not been reported. The aim of this work is (a) to identify the region(s) of Aβ with highest propensities for aggregation and (b) to use those fragments to inhibit Aβ fibrillogenesis. Structural and aggregation properties of the parent Aβ1–42 peptide and seven overlapping peptide fragments have been studied, i.e. Aβ1–10 (P1), Aβ6–15 (P2), Aβ11–20 (P3), Aβ16–25 (P4), Aβ21–30 (P5), Aβ26–36 (P6), and Aβ31–42 (P7). Structural transitions of the peptides in aqueous buffer have been monitored by circular dichroism and Fourier transform infrared spectroscopy. Aggregation and fibrillogenesis were analyzed by light scattering and thioflavin-T fluorescence. The mode of peptide-peptide interactions was characterized by fluorescence resonance energy transfer. Three peptide fragments, P3, P6, and P7, exhibited exceptionally high propensity for β-sheet formation and aggregation. Remarkably, only P3 and P6 exerted strong inhibitory effect on the aggregation of Aβ1–42, whereas P7 and P2 displayed moderate inhibitory potency. It is proposed that P3 and P6 intercalate between Aβ1–42 molecules and thereby inhibit Aβ1–42 aggregation. These findings may facilitate therapeutic strategies of inhibition of Aβ fibrillogenesis by Aβ-derived peptides.

. Aβ 1-42 amino acid sequence in single-letter format, with stretches corresponding to peptides P1 through P7 shown by horizontal bars of different colors above or below the sequence, along with respective theoretical isoelectric points (according to ExPASy ProtParam function, https:// web. expasy. org/ protp aram). Horizontal blue arrows below the sequence show the segments corresponding to β-sheet structures in Aβ  or Aβ 1-42 fibrils determined experimentally, with Protein Data Bank entries shown on the right-hand side. Segments left blank are described as loop or turn regions. In the lower panel, one structure for Aβ  and three structures for Aβ  in fibrillar form are shown (fibril axis and the direction of interstrand H-bonding are perpendicular to the plane of the image). Backbone regions in β-sheet and turn/loop conformations are colored blue and rose, respectively.  [71][72][73] . The aforementioned mutated peptide, KLVFWAK, also interacted in a non-homologous manner with the C-terminus of Aβ aggregates 67 .
These findings indicate that Aβ aggregation can be effectively modulated by Aβ-derived peptides. However, a complete understanding of the differential effects of Aβ segments along the entire Aβ 1-42 sequence has not been achieved. This work aimed at systematic studies on 10 to 12 amino acid residue long overlapping fragments of Aβ   (Fig. 1) to understand their intrinsic β-sheet formation and fibrillogenesis properties and their effects on same properties of the parent peptide. Out of seven peptide fragments, three stretches, corresponding to sequence 11-20, 26-36, and 31-42, have been identified that demonstrated strong β-sheet and aggregation propensity. The former two sequences were able to inhibit Aβ 1-42 fibrillogenesis by up to 80%, whereas the C-terminal dodecapeptide, as well as the fragment 6-15, displayed moderate (20-30%) inhibitory potency. Further spectroscopic studies shed light on the mode of peptide-peptide interactions. Our data thus provide significant new information on intrinsic biophysical properties of Aβ 1-42 segments and their effects on the behavior of the full-length peptide, which may be useful in designing peptide-based AD therapies.

Results
Choice of peptide fragments. Seven overlapping fragments of Aβ  have been studied, i.e. sequences 1-10, 6-15, 11-20, 16-25, 21-30, 26-36, and 31-42, named P1, P2, P3, P4, P5, P6, and P7, respectively. All peptides have been N-terminally acetylated and C-terminally amidated to eliminate the charge effects of terminal amine and carboxyl groups. We chose to study overlapping peptides to avoid missing the effects of sequences at boundaries of consecutive segments. The locations of these seven fragments along the Aβ 1-42 sequence and their respective theoretical isoelectric point (pI) values are presented in Fig. 1.
Secondary structure from FTIR spectroscopy. The lyophilized peptides were dissolved in hexafluoroisopropanol (HFIP) to disperse possible pre-existing aggregates 74 . HFIP solution of the Aβ 1-42 peptide was spread over a CaF 2 disk, air-dried and desiccated for 15 min, followed by measurements of Fourier transform infrared (FTIR) spectra. The dry peptide acquired α-helical structure manifested by a sharp amide I peak at 1658 cm −1 ( Figure S1 of Supplementary Information). Before cleavage from APP, most of the Aβ sequence corresponds to the transmembrane α-helix embedded in neuronal membranes 75 . Hence, this result is consistent with the intrinsic α-helical propensity of the Aβ 1-42 sequence in dehydrated state. On the other hand, determining the structure of the dehydrated peptide is important for analysis of structural transitions once the peptide is exposed to an aqueous medium.
Next, HFIP solutions of Aβ 1-42 and the seven fragments were separately dried in glass vials and aqueous (D 2 O) buffer was added to reach concentrations of 50 μM for Aβ 1-42 and 100 μM for the fragments. The samples were constantly stirred with a small magnetic stir bar, and FTIR spectra of 70 μL aliquots were measured periodically over 100 h to examine structural changes of the peptides. The amide I spectrum of Aβ 1-42 initially showed two prominent components at 1672 and 1647 cm −1 , suggesting mostly β-turn and unordered or loop conformations [76][77][78] (Fig. 2a). Upon incubation in buffer, the 1647 cm −1 peak was gradually replaced by a component at 1627 cm −1 , indication formation of intermolecular β-sheet structure 76,78 . This is consistent with the fibrillar structure of Aβ 1-42 comprising β-sheet and β-turn/loop conformations [32][33][34] (Fig. 1) and indicates that formation of turn/loop structures precedes aggregation and fibrillogenesis through intermolecular H-bonding.
The fragments exhibited distinct structural transitions in aqueous buffer. All fragments except for P7 initially displayed significant fraction of β-turn structure in addition to irregular structure, indicated by amide I components in 1672-1670 cm −1 and 1640 cm −1 regions, respectively (Fig. 2). Among these 6 peptides, only P3 and P6 showed β-sheet component at 1628-1623 cm −1 (Fig. 2d, g). In case of P3, both β-turn and β-sheet structures grew at the expense of irregular structure (Fig. 2d), whereas in P6 the β-turn became the dominant structure upon prolonged incubation in buffer (Fig. 2g). P7 exhibited a unique behavior; it rapidly formed β-sheet structure with a distinctive peak around 1626 cm −1 , which gradually shifted to higher wavenumbers possibly reflecting transition to turn and/or α-helix structures. Thus, peptides P3, P6, and P7 display β-sheet formation propensity with distinct dynamics of structural transitions whereas the other four peptides tend to adopt turn/loop conformations. Secondary structure from CD spectroscopy. The structural transitions of the peptides in aqueous medium were further analyzed by a complementary method, circular dichroism (CD). In the same time, the question was addressed as to how the structural transitions of the peptides are related to fibrillogenesis. This was accomplished by near-simultaneous measurements of CD spectra, light scattering, and thioflavin-T (ThT) fluorescence of same peptide samples in aqueous buffer for over 100 h. CD spectra were measured for peptides dissolved in HFIP, in dry state, and in aqueous buffer. The spectrum of the full-length Aβ 1-42 peptide in HFIP displayed a deep minimum at 204 nm and a shoulder around 220 nm ( Figure S2a of Supplementary Information), indicating the presence of type I and II β-turns, β-strand, and possibly α-helix conformations 79,80 . CD spectra of seven peptide fragments exhibited two negative ellipticities as well. The stronger component occurred in the 205-208 nm region (P1 and P2) or 195-200 nm region (P3 through P7) ( Figure S2b-h of Supplementary Information), assigned to β-turn and unordered structures, respectively [81][82][83] . The weaker component was located between 220 and 230 nm, which can be assigned to α-helix, β-turn and β-strand conformations.
The solvent was removed by desiccation and CD spectra of dry peptides were measured. These spectra showed significant structural changes upon drying. Aβ 1-42 adopted mostly α-helical structure, evidenced by a negative component at 223 nm and weaker one at 208 nm ( Figure S2a of Supplementary Information), which are generated by n-π* and π-π* electronic transitions 80,83 . Similar spectral features were displayed by P1, P3, and P6. Weaker intensity of the π-π* transition can be attributed to a more flexible or distorted α-helix 80 . In spectra of peptides P4 and P5, the π-π* transition was stronger, suggesting formation of more stable α-helix. Dry peptides www.nature.com/scientificreports/ P2 and P7 showed just one component around 220-223 nm, which is most likely to represent type-I β-turn and β-strand structures 79,81 .
Following structural characterization of the peptides dissolved in HFIP and in dry state, aqueous buffer was added and structural transitions were followed by recording consecutive CD spectra. Spectra of Aβ 1-42 initially showed a shallow negative band around 217 nm, which upon incubation in buffer for up to 48 h became a welldefined band accompanied with a positive component of nearly equal intensity around 200 nm, clearly indicating formation of β-sheet structure (Fig. 3a). Structural transitions in peptide fragments occurred at slower rate, therefore they were examined for longer time. Peptides P1, P2, P4, and P5 retained strong negative ellipticity around 200 nm and weaker components between 225 and 235 nm (217 nm for P2), implying these peptides assume mostly unordered conformation with fractions of β-turn/β-strand structures (Fig. 3b, c, e, f). In contrast, peptides P3, P6, and P7 readily (in less than 1 h) adopted defined secondary structures ( Fig. 3d, g, h). Spectra of these three peptides display a negative n-π* band around 220-225 nm and strong positive ellipticity in the 200 nm region. These spectral features can be assigned to twisted β-sheet structure, as the intense π-π* transition is diagnostic for twisted β-strands, and the n-π* band can be significantly red-shifted depending on the degree of strand twist [80][81][82]84,85 . Note that CD spectra become excessively noisy below 200 nm due to the relatively long optical path-length (4 mm) of the cuvette, which was used for simultaneous light scattering and fluorescence measurements. However, this does not invalidate above interpretations of CD spectra in terms of peptides' structural features.
Some differences have been observed between peptide structures deduced from FTIR and CD measurements, such as different relative contents of β-turn and irregular structures. These differences can be attributed to nonidentical sample preparation and handling protocols, as Aβ structure is exquisitely sensitive to sample preparation conditions, including the method of sample stirring 24,32,[86][87][88] . In addition, both N-terminal acetylation and C-terminal amidation introduce additional amide bonds that generate FTIR signal in the unordered region, i.e. between 1650 and 1638 cm −1 . Nonetheless, both FTIR and CD data unveil important common structural trends of these peptides, i.e., P3, P6, and P7 stand out by exhibiting strong β-sheet propensity, like the parent peptide, while peptides P1, P2, P4, and P5 tend to stay in β-turn or unordered conformations.

Aggregation and fibrillogenesis.
ThT fluorescence and static light scattering at 90 degrees were used to examine the kinetics of aggregation and fibrillogenesis of the peptides in aqueous buffer. Of note, these measurements were conducted in parallel with CD on same samples, which allowed analysis of structural changes, peptide aggregation and fibril formation in real time, excluding sample-to-sample errors. Increase in ThT fluorescence around 480 nm when excited at 435 nm was interpreted as fibril formation with an understanding that non-fibrillar structures could also bind ThT and produce fluorescence in that region. Aβ 1-42 underwent fibrillogenesis without delay, as indicated by increase in ThT fluorescence around 480 nm (Fig. 4a, b). Saturation of ThT fluorescence was reached by around 1 day of incubation in buffer with constant stirring on a rotary mixer. Light scattering at 550 nm increased with steeper initial rate and approached saturation by 10-20 h of incubation  In samples with peptide fragments P1, P2, P4, and P5, very weak ThT fluorescence was detected ( Fig. 5a, b, d, e). Peptides P3 and P6 generated higher level of ThT fluorescence, and peptide 7 produced maximum ThT fluorescence ( Fig. 5c, f, g), unraveling a trend of fibrillogenesis propensity P7 > P6 ≈ P3 >> P4 ≈ P5 ≈ P2 ≈ P1 ≈ 0 (Fig. 5h). Statistical t-test analysis of all data of Fig. 5 between 24 and 104 h of incubation in buffer produced nearly 98% credibility for significant difference between fibrillogenesis propensities of P7 and P3 (p = 0.0214 ± 0.0084) and more than 96% credibility between P7 and P6 (p = 0.0385 ± 0.0101). P3 and P6 were not significantly different (p = 0.593 ± 0.241), and the difference between P3 and P1, P2, P4, P5 was 96.9-98.7% significant.
Concurrent light scattering measurements indicated relatively weak signal for peptides P1, P2, P4, and P5 and about threefold higher signal for peptides P3, P6, and P7 ( Figure S3 of Supplementary Information). For the first group of peptides, t-test analysis produced p values between 0.172 ± 0.085 (P2/P5) and 0.747 ± 0.232 (P1/P4), and for the second group of peptides, p values varied between 0.366 ± 0.270 (P3/P7) and 0.685 ± 0.159 (P3/P6), indicating little statistical difference within each group. The difference between these two groups was significant (> 96%), as indicated by p values for P2/P3 of 0.0367 ± 0.0122. Negligible ThT fluorescence but appreciable light scattering of peptides P1, P2, P4, and P5 apparently reflects formation of non-fibrillar aggregates by these fragments. Higher level of ThT fluorescence by P7 compared to P3 and P6 but similar light scattering (cf. Fig. 5h and S3h) suggests a higher ThT binding capacity of P7, plausibly due to its exceptional hydrophobicity and/or a larger number of amino acid residues. Overall, these data provide evidence that peptides P3, P6, and P7 possess significant aggregation/fibril formation capability as compared to the other four Aβ fragments. In conjunction with structural data, it appears that the aggregation/fibrillogenesis property of these peptides correlates with their β-sheet formation ability.
Effects of Aβ fragments on Aβ fibrillogenesis. If some of peptide fragments demonstrate strong fibrillogenesis propensity, they might be able to intercalate into the assemblies of the parent peptide by binding to respective stretches and thereby interfere with its aggregation. To test this hypothesis, fibrillogenesis of Aβ  was measured in the absence and presence of all seven peptide fragments at two-fold molar excess. In the presence of peptides P1, P2, P4, P5, and P7, ThT fluorescence of Aβ 1-42 samples gradually increased and levelled off by 48 h of incubation with little effect of these fragments (Fig. 6a, b, d, e, g, h). In contrast, P3 and P6 strongly reduced the level of ThT fluorescence (Fig. 6c, f, h). Quantitative evaluation of inhibition of Aβ 1-42 fibril formation by peptide fragments was performed by calculating the inhibition percentages (IP) for each fragment, as described in "Materials and methods". P3 and P6 strongly inhibited Aβ 1-42 fibrillization by 80.5 ± 9.0% and 77.7 ± 5.8%, respectively, with insignificant statistical difference between them (p = 0.673) (Fig. 7). Peptides P2 and P7 exhibited moderate inhibitory potencies (IP = 18.8 ± 2.1% and 28.7 ± 3.8%, respectively), with considerable statistical difference (p = 0.0163), and the other three peptides (P1, P4, and P5) were not potent inhibitors The difference between these three groups, presented by bars of distinct colors in Fig. 7, was statistically significant (p < 0.01).
Mode of peptide interactions from Phe-to-Tyr energy transfer. The next question we explored was whether the intrinsic fluorescence of peptide fragments can be utilized to gain insight into the mode of their interactions. Aβ 1-42 is devoid of tryptophan, but has one tyrosine (Tyr) at position 10 and three phenylalanines (Phe) at positions 4, 19, 20. Phe to Tyr resonance energy transfer (RET) has been described earlier for a tryptophan-less protein, with Förster distance of R 0 = 13.5 Å 89,90 , which is much shorter than that of many other donor-acceptor pairs, typically in the 20 Å to 60 Å range 90 . This provides a unique opportunity to examine the interactions between peptides containing Phe and Tyr by RET.
As both Tyr and Phe can be excited in a wide wavelength range 90 , an optimal wavelength needs to be identified where Phe absorbs maximally and Tyr absorbs minimally. Our measurements showed that this wavelength range was around 220 nm. Figure S4 of Supplementary Information shows how RET works for the P3/P1 donor/acceptor pair. Upon excitation at 220 nm, Phe and Tyr of peptides P3 and P1 emit around 287 and 310 nm, respectively ( Figure S4a,b). In equimolar mixture of these peptides at same total peptide concentration, the intensity of Phe emission gradually decreases, indicating energy transfer to Tyr. Negative change of Phe fluorescence (DF Phe < 0)  As seen from Fig. 1, only peptide fragments P1, P2, P3, and P4 contain Phe and/or Tyr fluorophores, hence these four peptides have been analyzed in RET experiments. Figure 8 shows RET data for five peptide pairs that satisfy this criterion, i.e. possibility of Phe-to-Tyr RET. These data indicate slightly negative DF Phe in P1/P2 and P1/P4 systems, implying little RET in these peptide samples (Fig. 8a, c, f, h). Significant RET is seen in combinations of P3 with P1 and P2; in both cases negative DF Phe develops soon after mixing the peptides, reaches maximum values between 10 and 30 h, and returns to zero level between 50 and 100 h of incubation in buffer (Fig. 8b, d, g, i). In P2/P4 combination, there is relatively moderate RET (DF Phe < 0) for the whole time period of the experiment, i.e. ~ 200 h (Fig. 8e, j), indicating stable hetero-aggregation between peptides P2 and P4. Values of DF Tyr are always positive, consistent with RET; their variations can be attributed to quenching effects of water and/or carbonyl groups of peptides during the aggregation process, as described earlier 71 .
These data indicate that P3 not only readily self-aggregates, as seen from ThT fluorescence and light scattering data (Figs. 5h and S3h), but also effectively interacts with non-homologous stretches of Aβ, such as P1 and P2 segments, pointing to the "sticky" nature of P3. In the P1/P3 system, the peptides are likely to interact in parallel in-register manner, positioning Tyr 10 against Phe 20 ( Figure S5a of the Supplementary Information). Strong RET between P2 and P3 suggests parallel out-of-register pairing, with Phe 19 facing Tyr 10 ( Figure S5n of the Supplementary Information). These interactions apparently start falling apart at the end of first day of incubation (Fig. 8g, i), as P3 predominantly forms homologous aggregates, as described below. Data of Fig. 8e, j suggest that P2 and P4 form persistent aggregates, presumably of parallel out-of-register sense allowing juxtaposition of Phe 19 and Tyr 10 ( Figure S5c of the Supplementary Information). It is interesting to note that in these proposed models of peptide interactions, ionic bonds between side chains of Asp 7 and Lys 16 play an important role by stabilizing hetero-aggregation in both P2/P3 and P2/P4 systems ( Figure S5b,c of the Supplementary Information).
To gain insight into the mode of interaction of P3 with Aβ 1-42 , RET experiments have been conducted with this peptide system. Fluorescence spectra were measured for 35 μM Aβ 1-42 , 70 μM P3, and 35 μM Aβ 1-42 + 70 μM P3 over a time period of 48 h, with excitation at 220 nm ( Figure S6 of the Supplementary Information). First, we checked if RET effect was involved in self-aggregation of Aβ 1-42 by subtracting the spectrum of Aβ 1-42 measured at 1 h from spectra measured at 2 h and beyond. These difference spectra showed time-dependent changes in Tyr fluorescence, reflecting at least two effects, i.e., suspension of the dry peptide into the aqueous medium (decrease of fluorescence because of quenching by water) and sequestration from the aqueous medium through fibrillogenesis (increase of fluorescence), but a combination of negative ΔF Phe and positive ΔF Tyr has not been detected ( Figure S6d of the Supplementary Information). This means absence of RET between Aβ 1-42 molecules, consistent with formation of parallel, in-register β-sheet aggregates. A similar procedure for the Aβ 1-42 + P3 system, as well as subtraction of spectra of both Aβ 1-42 and P3 from those of the combined sample at respective times, yielded RET effect at 32-48 h of co-incubation, i.e. DF Phe < 0 and DF Tyr > 0 ( Figure S6e,f of the Supplementary  Information). This outcome implies that during interaction of P3 with Aβ 1-42 , Phe 19 and Phe 20 of at least a fraction of P3 come to close proximity to Tyr 10 of Aβ 1-42 at 32-48 h. However, the inhibitory effect of P3 on Aβ 1-42 fibrillogenesis starts at earlier times (Fig. 6h), indicting that P3 interacts more effectively with other regions of Aβ  , which need to be identified in further studies.

Discussion
This work reports an analysis of structural and aggregation propensities of Aβ 1-42 fragments and their effects on Aβ 1-42 fibrillogenesis. The results facilitate a better understanding of segment-specific properties of Aβ  and identify peptide fragments capable of inhibiting its aggregation. We chose 10-to 12-residue overlapping fragments because fragments shorter than 10 residues might be unable to exhibit robust secondary structure formation and fibrillogenesis abilities because of insufficient length. Although short Aβ fragments containing 6-8 residues have been shown to adopt parallel or antiparallel β-sheets in crystal form 91,92 , the hydrophobic C-terminal sequences of Aβ 1-42 containing eight amino acid residues, as well as an 11-residue stretch (Aβ 32-42 ), did not aggregate into β-sheet fibrils in aqueous buffer even at concentrations exceeding 100 mM 36 . On the other hand, use of overlapping sequences was important for characterization of segments at boundaries of consecutive stretches; properties of the boundary between P(i) and P(i + 2) were reflected in P(i + 1), which covers the C-and N-termini of flanking sequences. Peptides P1 and P2 represent the N-terminus of Aβ, which has been consistently found in unordered conformation in the context of Aβ 1-40 or Aβ 1-42 24,31,32,34,86,93 . The Aβ 1-8 fragment was found in extended coil confirmation in complex with a monoclonal antibody 94 . Despite its proposed role in stabilizing Aβ fibrils 95 , above data and our findings indicate that the N-terminus of Aβ is intrinsically unordered and unable to undergo fibrillogenesis. Consistent with this, RET data indicate that P1 and P2 do not form hetero-aggregates as well (Fig. 8a, f). This feature is apparently related with the amino acid composition of these segments; the paucity of nonpolar side chains and the excess negative charge (pI 4.54 and 5.72 for P1 and P2, respectively) may prevent aggregation due to electrostatic repulsion and lack of hydrophobic interactions. P1 is also a poor inhibitor of Aβ 1-42 aggregation, whereas P2 demonstrates moderate inhibitory activity (Fig. 7), which probably stems from its C-terminal part, as P3 is highly aggregation prone and a potent inhibitor of Aβ 1-42 fibrillogenesis.
The P3 segment corresponds to one of the two major β-strands of Aβ 1-40 fibrillar structure Aβ 1-42 24,31,86 (Fig. 1). The whole P3 sequence or at least its central part are in β-strand conformation in fibrils formed by of Aβ 1-42 as well [32][33][34] (Fig. 1). P3 has one glutamic acid (Glu) and one lysine (Lys) residues and nearly zero net charge (pI = 7.01). These features would allow the peptide to form β-strand-like non-H-bonded out of register aggregates in antiparallel sense with oppositely charged side chains facing each other, as shown in Figure S7 of the Supplementary Information. This arrangement will juxtapose one Glu-Lys pair and one histidine pair between strands so the structure will be stabilized by Glu-Lys salt bridges and π-π stacking interactions between histidines 96,97 . As a fraction of imidazole side chains will be protonated due to the close-to-neutral pK = 6.3 ± 0.1 98 , His-His attraction will be further strengthened by nucleophile-electrophile interactions. This structure may grow laterally beyond tetramer until decreasing internal energy is compensated by decreasing entropy at a minimum of free energy. The fibril will grow by means of parallel β-sheet formation stabilized by H-bonding perpendicular to the plane of side chains. The nonpolar overhangs may form turn or parallel β-sheet structure stabilized by hydrophobic interactions and π-π stacking interactions between C-terminal phenylalanines 96 . These conjectures are consistent with earlier observations of dimer formation by Aβ 10-23 peptide that grow into β-sheet fibrillar structure 99 . FTIR data (Fig. 2d) show a low frequency (1623 cm −1 ) β-sheet component without a high-frequency counterpart diagnostic for antiparallel β-sheet, consistent with parallel β-sheet formation. Figures 5, 6, and 7 show that P3 forms fibrils and substantially blocks the fibrillogenesis of Aβ 1-42 . Inhibition of Aβ 1-42 fibrillogenesis presumably occurs through intercalation of P3 aggregates between Aβ 1-42 molecules and/or binding of P3 to the homologous or other segments of Aβ 1-42 . Heterologous binding is supported by the ability of P3 to effectively interact with P1 and P2 ( Figures S5 and 8b, d, g, i). Consistent with our findings, it was reported earlier that Aβ segment 11-17 (EVHHQKL), i.e. 0.7 part of P3, but not other Aβ-derived short peptides, effectively blocked Aβ 1-40 -induced Ca 2+ influx into cultured cells possibly by inhibiting membrane pore structure formation 58 .
In conclusion, this work thoroughly characterizes the intrinsic aggregation and structural properties of overlapping peptide segments throughout the Aβ 1-42 sequence and the effects of these fragments on Aβ 1-42 fibrillogenesis. Although previous studies have utilized a similar approach and have focused on certain parts of Aβ 1-42 , such as the C-terminal hydrophobic stretch, systematic dissection and analysis of the whole Aβ 1-42 sequence has not been reported. Or data identify Aβ fragments that are capable of self-aggregation and β-sheet structure formation and substantially inhibit fibrillogenesis of Aβ 1-42 , presumably by intervention into forming aggregates or prevention of efficient intermolecular contacts between Aβ 1-42 molecules. However, there is no universal correlation between self-aggregation/β-sheet propensities of the fragments and suppression of Aβ 1-42 fibril formation. For example, P7, the most hydrophobic segment that undergoes efficient fibrilization (Fig. 5h) and β-sheet formation ( Fig. 2h and 3h) is a relatively week inhibitor of Aβ 1-42 fibrillization of (Figs. 6h and 7). This leads to a conclusion that peptide fragments with high homo-aggregation propensity prefer to self-associate instead of forming hetero-aggregates with the full-length peptide. These findings, specifically the potent inhibitory effects of P3 and P6 peptides on Aβ 1-42 aggregation, may help develop new DNA-, mRNA-or peptide-based anti-AD therapeutic strategies.

Materials and methods
Materials. The synthetic Aβ 1-42 peptide was from Innovagen (Lund, Sweden) and was > 98% pure. Synthetic peptide fragments, acetylated at N-terminus and amidated at C-terminus, were purchased from Peptide 2.0 Inc. (Chantilly, VA, USA) and were 98-99% pure, as verified by high performance liquid chromatography and mass-spectrometry. HFIP, ThT, salts, buffers, and most of other reagents were purchased from Sigma-Aldrich Experimental procedures. Circular dichroism and fluorescence. Circular dichroism (CD) and fluorescence spectra were measured using a J-810 spectropolarimeter equipped with a fluorescence attachment and a temperature controller (Jasco, Tokyo, Japan) at 25 °C. Lyophilized peptides were dissolved in HFIP at a desired concentration and CD spectra were measured using a 4 mm × 4 mm rectangular quartz cuvette. Then, the solvent was removed by desiccation and CD spectra of the dry peptide stuck on the walls of the cuvette were measured. This was followed by addition of aqueous buffer (25 mM NaCl, 25 mM Na,K-phosphate, 20 μM ThT, pH 7.2) to achieve a target peptide concentration. The cuvette was capped with a Teflon stopper, sealed with parafilm, and rotated on a Fisherbrand™ Multi-Purpose Tube Rotator for several days. CD, ThT or intrinsic peptide fluorescence, and light scattering at 90 degrees were measured periodically on same sample, at pre-determined time pints, to follow the time dependence of structural changes and aggregation/fibrillogenesis of the peptides, using the multifunctional J-810 instrument. Measurements of these parameters on same samples in real time excluded sample to sample errors. For each peptide system, time-dependent spectroscopic measurements were repeated in three or four independent experiments.
ThT fluorescence was averaged between 475 and 485 nm, when excited at 435 nm, and plotted against time of incubation in buffer. For fluorescence RET experiments, incident light at 220 nm was used for predominant excitation of phenylalanine as compared to tyrosine, and spectra were measured between 200 and 400 nm. For www.nature.com/scientificreports/ each donor/acceptor peptide pair, the fluorescence of individual peptides was measured at 100 μM concentration, then spectra were recorded for the two peptides mixed at 50 μM each to maintain the same total peptide concentration. To determine the presence of RET, spectra of each peptide were subtracted from the spectrum of the combination multiplied a factor of 2. Reduction of the donor (Phe) but not acceptor (Tyr) fluorescence in the combination indicated RET effect. For light scattering, incident light at 550 nm was used to avoid generation of fluorescence signal and scattered light intensity at 90 degrees was measured between 530 and 570 nm. The signal intensity was averaged between 550 and 555 nm and plotted as a function of time.
The degree of inhibition of fibrillization of Aβ 1-42 by peptide fragments was evaluated based on ThT fluorescence data. Total saturating ThT fluorescence between 32 and 48 h in the presence of both Aβ 1-42 and a given fragment was considered as a sum of fluorescence due to each component. Then, fluorescence owing to Aβ 1-42 fibrils in the presence of a fragment, F*, was determined by subtracting the signal measured for the fragment alone from total fluorescence. Fluorescence signal measured for pure fragments was corrected for concentration before subtraction, i.e. multiplied by 0.7, assuming a linear relationship between fluorescence and peptide concentration. Inhibition percentage of Aβ 1-42 fibrillogenesis was calculated as IP = (1 − F*/F 0 ) × 100%, where F 0 is ThT fluorescence of Aβ 1-42 in the absence of fragments.
FTIR spectroscopy. HFIP solution of the peptide was placed in small glass vials and the solvent was removed by desiccation. D 2 O-baseb buffer containing 25 mM NaCl and 25 mM Na,K-phosphate, adjusted to pD 7.2, was added to the dry peptide to achieve 50 μM concentration for the full-length peptide Aβ 1-42 and 100 μM for the fragments. Peptide samples were continuously stirred by a small (3 mm in diameter) magnetic stir bar for > 100 h. At defined time points, a 70-μL peptide sample was placed between two CaF 2 disks (32 mm in diameter, 3 mm thick), separated by a 50 μm thick Teflon spacer, and FTIR spectra were measured at 2 cm −1 nominal resolution, using a Vector-22 FTIR spectrometer (Bruker Optics, Billerica, MA, USA) equipped with a liquid nitrogen-cooled Hg-Cd-Te detector. Transmittance spectra of the blank buffer were measured and used as reference to calculate the absorbance spectra. For measurements of FTIR spectra of dry peptide sample, HFIP solution of the peptide was spread on a CaF 2 disk, dried for 15 min in a desiccator, followed by measurement of the transmittance spectrum. A transmittance spectrum of the clean CaF 2 disk was used as reference to calculate the absorbance spectrum. The instrument was constantly purged with dry air to avoid contribution of H 2 O vibrational/rotational features in the amide I region. Absorbance spectra of atmospheric H 2 O were measured separately and subtracted from spectra of peptides when such contribution was detected.