The inhibition of cellular toxicity of amyloid- b by dissociated transthyretin

The protective effect of transthyretin (TTR) on cellular toxicity of b -amyloid (A b ) has been previously reported. TTR is a tetrameric carrier of thyroxine in blood and cerebrospinal fluid, the pathogenic aggregation of which causes systemic amyloidosis. However, studies have documented a protective effect of TTR against cellular toxicity of pathogenic A b , a protein associ-ated with Alzheimer ’ s disease. TTR binds A b , alters its aggregation, and inhibits its toxicity both in vitro and in vivo . In this study, we investigate whether the amyloidogenic ability of TTR and its antiamyloid inhibitory effect are associated. Using protein aggregation and cytotoxicity assays, we found that the dissociation of the TTR tetramer, required for its amyloid pathogenesis, is also necessary to prevent cellular toxicity from A b oligomers. These findings suggest that the A b -binding site of TTR may be hidden in its tetrameric form. Aided by computational docking and peptide screening, we identified a TTR segment that is capable of altering A b aggregation and toxicity, mimicking TTR cellular protection. EM, immune detection analysis, and assessment of aggregation and cytotoxicity revealed that the TTR segment inhibits A b inhibiting A b 42 cytotoxicity and amyloid seeding. We also found that TTR-S binds to soluble and fibrillar species, causing a structural rearrangement that leads to an increase of protease sensitivity. Finally, we found that TTR-S inhibits amyloid seeding catalyzed by A b fibrils extracted from the brain of an AD patient. Our results encourage further evaluation of the inhibitory effect of TTR-derived peptides by additional experimenta-tion. Additional experiments may include independent cell death assays to confirm our MTT-based cytotoxicity assays and evaluation of the species that result from the inhibition of seeded polymerization by ex vivo fibrils using techniques such as EM imaging, CD, and protease digestion analysis. The pres-ent study represents an expansion of the current knowledge on the mechanism of protection of TTR over A b 42 cellular toxicity and opens a potential therapeutic avenue.

The physiological importance of transthyretin in Alzheimer's disease was first reported by Schwarzman et al. in 1994 (1). One of the hallmarks of Alzheimer's disease (AD) is the formation of brain plaques composed of b-amyloid peptide (Ab). A 42-residue-long b-amyloid peptide (Ab42) is the predominant variant in neuritic plaques of AD patients, with higher amyloidogenicity and cellular toxicity in vitro (2,3). Many studies have shown that transthyretin (TTR) binds to Ab, alters its aggregation, and inhibits its toxicity both in vitro and in vivo (4)(5)(6)(7)(8). In vitro, TTR co-aggregates with Ab oligomers into large nontoxic assemblies, thereby inhibiting cellular toxicity (9,10). In vivo, TTR sequesters Ab and facilitates its clearance in the brain (1). More recently, the Buxbaum laboratory (6) showed that overexpression of WT human TTR suppressed disease progression in the APP23 transgenic AD mouse model. They also showed that silencing the endogenous TTR gene in AD transgenic mice accelerated Ab42 deposition (5).
What makes the pair of TTR and Ab42 particularly interesting is the amyloid nature of the two elements. In health, TTR functions as a transporter of retinol and thyroxine in blood, cerebrospinal fluid, and the eye and is secreted by the liver, choroid plexus, and retinal epithelium, respectively. However, dissociation of tetrameric TTR leads to amyloid fibril formation and systemic TTR amyloid deposition in patients of transthyretin amyloidosis (11)(12)(13). Whether the amyloidogenicity of TTR is linked to its interaction to Ab42 is still under debate. Previous work showed that different aggregation propensities of TTR result in distinct interaction capabilities: fewer amyloidogenic variants had increased affinity for Ab (7). However, this did not impact on the levels of inhibition of Ab aggregation, and cytotoxicity protection was not assessed. Here we attempt to fill the experimental gap by studying the protective effect of TTR variants at different aggregated states over Ab42 cellular toxicity. In addition, we show the identification and characterization of a segment of TTR that binds Ab42 triggering the formation of nonamyloid amorphous aggregates that are more sensitive to protease digestion, therefore mimicking TTR inhibitory effect over Ab42.

Results
We first evaluated the protection against Ab42 cytotoxicity by nine TTR variants with distinct aggregation propensities in several aggregated states ( Table 1). The amyloidogenic behavior of three representative TTR variants is shown in Fig. 1 (A  and B). For this assay, recombinant transthyretin was incubated at 37°C, pH 4.3; protein aggregation was followed by immunodot blot of the insoluble fractions collected by centrifugation at several time points (Fig. 1A); and EM was performed after 4 days of incubation (Fig. 1B). We found that M-TTR, a monomeric-engineered form of transthyretin that is soluble at physiological pH (14), shows notable aggregation after 1 day of incubation at pH 4.3 (Fig. 1, A-C). NSTTR is an artificially mutated variant of TTR carrying the double mutation N99R/S100R that, although remaining tetrameric in solution (Fig. 1C), shows a slower aggregation pattern than M-TTR (Fig. 1, A and B). T119M is a very stable mutant that did not show any sign of aggregation in 4 days (Fig. 1, A- in a significant delay of the onset of hereditary neuropathic TTR amyloidosis in patients who carry both ttr-V30M and ttr-T119M genes (15). We then incubated Ab42 with the samples obtained from Fig. 1A for 16 h and evaluated its cytotoxicity by following MTT reduction. Three observations were made. The first observation is that once aggregated, TTR does not prevent Ab42 toxicity, for any of the variants (Fig. 1D). In addition, the variant NSTTR, which aggregates at a slow pace, resulted to be cytoprotective longer than M-TTR (Fig. 1D). The third observation is that T119M, which does not dissociate and aggregate, does not prevent toxicity either (Fig. 1D). We observed the same phenomena when the other six TTR variants were evaluated (Fig. S1). Although variants S112I, M13R/ L17R, or A108R/L110R are mainly monomeric or dimeric in solution, the cytotoxic protection at day 0 is not greater than NSTTR or L55P that are mainly tetrameric in solution ( Fig.  1C and Fig. S1B). This finding suggests that the capacity to protect from Ab42 toxicity does not correlate with the initial oligomeric state of soluble variants ( Fig. 1C and Fig.  S1B) but rather with the dissociation state at the moment of the co-incubation with Ab42 (Fig. 1, A and B, and Fig. S1C). A control experiment shows that the TTR variants used in this assay were not cytotoxic in the absence of Ab42, with the exception of M-TTR, which resulted in a 20% reduction of cell viability (Fig. S1D). Overall, our results indicate that dissociation of TTR is required for the inhibition of Ab cytotoxicity. In addition, the data suggest that a more stable dissociated TTR exerts more Ab42 inhibition than a more amyloidogenic variant. These findings led us to wonder whether a segment of TTR that is only exposed when dissociated but not in the aggregate might be similarly capable of altering Ab toxicity.
Computational modeling of the interaction of a fibrillar segment of Ab42 and the TTR monomer shows a tight packing of the thyroxine-binding pocket and the fibrillar structure (Fig. 2). Others have shown that the amyloidogenic segment KLVFFA is protected when TTR is bound to Ab (16). In previous studies, the Eisenberg laboratory determined the structures of three fibrillar polymorphs of KLVFFA by X-ray microcrystallography (17). We used these three polymorphs, as well as the Figure 1. The inhibition of Ab42 cytotoxicity by TTR depends on the aggregation propensity of each TTR variant. A, aggregation assay of three TTR variants followed by immunodot blot of insoluble fractions collected prior to incubation or after 1, 2, and 4 days of incubation at 37°C, as labeled. All samples were treated equally and onto the same membrane. Cropping was applied for ethetical reasons. B, electron micrographs of aggregated TTR variants after 4 days of incubation. Scale bar, 200 nm. C, size-exclusion chromatography of soluble TTR variants at pH 7.4. D, cytotoxicity assay of Ab42 in the presence of TTR variants at different stages of aggregation, followed by MTT reduction. A 5-fold molar excess of soluble TTR (50 mM, considering monomeric concentration) and aggregated samples collected after 1, 2 and 4 days of incubation were added to soluble 10 mM Ab42 and incubated overnight. The samples were added to HeLa cells, and MTT reduction was measured after 24 h. Buffer-treated cells were considered 100% viability and used for normalization (n = 3). All replicates are shown. Error bars, S.D. **, p 0.005; ***, p 0.0005. N.D., significance not detected. These results suggest that the dissociation of the tetrameric TTR structure precedes Ab42 cytotoxicity protection, whereas aggregated TTR does not exert any effect.  (Fig. 2). For every fibrillar form, the segment TTR(105-117) was the longest interacting TTR segment. The segment TTR(105-117) is only exposed in the monomeric or dimeric form of TTR. We wondered whether this segment might mimic TTR protective effects when isolated.
The peptide TTR(105-117) was found to be highly amyloidogenic. We first analyzed the amyloidogenicity of TTR sequence by ZipperDB, which measures the propensity of every six-residue segment to form amyloid fibrils (19,20) (Fig. S2A). TTR predictions show that the region that contains this segment is highly prone to form amyloid structures. In fact, we found the segment TTR(105-117) and the shorter peptide TTR(106-117) to be amyloidogenic in solution (Fig. S2B), which may explain the lack of inhibitory effect in previous studies (21). In contrast to the amorphous aggregates generated by recombinant full-length TTR, these peptides form amyloid-like fibrils (Fig. S2B). We then explored several sequence modifications to increase solubility, by eliminating the first tyrosine or adding a charged tag to the N-terminal end. The sequences and names of all the analyzed peptides are listed in Table 2. We found that TTR(105-117) amyloidogenicity was fully hindered by the addition of a polyarginine tag (Fig. S2B).
The soluble tagged derivatives of the segment TTR(105-117) inhibited Ab42 fibril formation ( Fig. 3 and Figs. S3 and S4). We first evaluated the inhibitory effect of TTR-derived peptides over Ab42 fibril formation in thioflavin T (ThT) assays ( Fig. 3A and Fig. S3). Of all the peptides that we analyzed, the best Ab42 inhibitory effect was obtained with TTR-S (sequence YTIAALLSPYSYSRRRRR), which contains a four-arginine tag followed by TTR(105-117) (Fig. S3). After an incubation of 2 days, we analyzed the samples by EM and found that the addition of TTR-S promoted the formation of amorphous aggregates ( Fig. 3B) that were not birefringent when stained with Congo red (Fig. 3C). Remarkably, we found that TTR-S also promoted the formation of amorphous species when incubated with preformed Ab42 fibrils (Fig. 3B). These aggregates were also found to be thioflavin T-negative (Fig. 3A), and nonbirefringent when stained with Congo red (Fig. 3C). The structural characterization of Ab42 amorphous aggregates by CD showed that the addition of TTR-S to preformed fibrils resulted in a significant structural shift from b to helical secondary motifs (Fig.  3D). Additionally, we found that this structural modification results in cytotoxicity protection (Fig. 4). HeLa, PC12, and SH-5YSY cells were subjected to Ab42 in the absence and presence of TTR-S at different molar ratios, and cell metabolic activity was followed by MTT reduction (Fig. 4, A-D). We found that the addition of TTR-S results in a significant reduction of Ab42 cellular toxicity. This protective effect may be explained by the inhibition of the formation of A11-positive oligomers as a result of the incubation with TTR-S (Fig. 4D). It is worth noting that the larger aggregates did not resolve in the SDS-PAGE gel.
Next we found that the binding of TTR-S to soluble and fibrillar Ab42 alters their protease resistance. Others have shown that TTR increases Ab42 clearance in vivo (1). We reasoned that the large assemblies that form upon binding to TTR might be more sensitive to proteolytic activity than amyloid fibrils, thereby facilitating clearance. This hypothesis would also explain the protective effect found in vivo (6). We explored this hypothesis by analyzing proteolytic sensitivity of TTR-Sderived amorphous aggregates (Fig. 5). We incubated soluble and fibrillar Ab42 with TTR-S overnight and collected the insoluble fractions by centrifugation. The immunodot blot of insoluble fractions showed that amorphous aggregates that result from the incubation of TTR-S with both soluble and fibrillar Ab42 were easily digested by proteinase K after 1 h (Fig. 5, A  and B). In contrast, Ab42 fibrils showed a significantly higher resistance to proteinase K digestion. We included soluble Ab42 in the assay as a control of proteolytic activity.
Finally, we evaluated TTR-S ability to hinder amyloid seeding caused by fibrils extracted from the brain of an AD patient (Fig. 5C). The extraction of Ab fibrils was performed by several cycles of homogenization and ultracentrifugation as described by Tycko (22). Amyloid seeding was measured by thioflavin T fluorescence. As expected, we found that the addition of sonicated patient-derived fibrils to soluble Ab42 resulted in the acceleration of fibril formation. In contrast, incubation with TTR-S resulted in inhibition of Ab42 fibril formation even in the presence of patient-derived Ab seeds.

Discussion
Our results indicate that the dissociation of the TTR tetramer is required to prevent cytotoxicity from Ab oligomers. These results are consistent with the model proposed by Buxbaum and co-workers (16), in which dissociated TTR  monomers efficiently bind Ab oligomers, which are thought to be responsible for cellular Ab-associated toxicity (23). They also found that tetrameric TTR binds to soluble monomeric Ab, thereby suppressing aggregation in vitro (16). Others have also found an association between genetic stabilization of transthyretin and decreased risk of cerebrovascular disease (24). Consistently, an additional study found that the stabilization of the tetrameric form of TTR promotes Ab clearance in a mouse model of Alzheimer's disease (25). In our study, we did not find any effect of tetrameric TTR on Ab42 cytotoxicity. For instance, the tetrameric nonamyloidogenic T119M variant did not inhibit Ab42 cytotoxicity (Fig. 1). Consistently, the nonamyloidogenic variant L55P/ T119M did not inhibit cytotoxicity, whereas the amyloidogenic L55P variant did (Fig. S1). Taken together, other studies and our results suggest that tetrameric TTR may be protective in vivo perhaps not by recruiting toxic oligomers, but by sequestering and promoting clearance of Ab monomers. In contrast, dissociated TTR seems to inhibit Ab cytotoxic-ity by binding to toxic oligomeric species triggering the formation of large nontoxic assemblies. Using computational tools, we generated a TTR-derived peptide that inhibits Ab42 fibril formation and cellular toxicity (Fig. 3). This TTR-derived peptide, called TTR-S, contains the highly amyloidogenic segment TTR(105-117) followed by a polyarginine tag (Fig. S3). Our results are consistent with previous NMR studies showing that the interaction of TTR residues Lys 15 , Leu 17 , Ile 107 , Ala 108 , Ala 109 , Leu 110 , Ala 120 , and Val 121 with Ab contributes to the inhibition of Ab fibril formation in vitro (16). In addition, others have shown that the segment TTR(106-117), one residue shorter, was capable of binding Ab when immobilized on a membrane (8). However, the same peptide was not effective inhibiting Ab aggregation in solution (21), unless the peptide was made cyclic, therefore not amyloidogenic (26), probably because of the high amyloidogeneicity of this segment (Fig. S3). In our study, we solubilized TTR (105-117) by adding a charged tag, which confers higher solubility and also hinders self-aggregation (Fig. S3). The addition of a polyarginine tag was sufficient to convert a highly amyloidogenic peptide into an inhibitor of Ab42 cellular toxicity even at substoichiometric concentrations (Fig. 4A).
TTR-S inhibits Ab42 oligomer formation while promoting the formation of nontoxic, nonamyloid, amorphous aggregates. Previous studies of the inhibition of Ab cytotoxicity by TTR have revealed that TTR co-aggregates with Ab oligomers into large nontoxic assemblies, thereby inhibiting cellular toxicity in vitro (9). TTR-S mimics this effect and promotes the formation of large aggregates (Fig. 3B) that do not bind thioflavin T (Fig.  3A), do not show birefringence upon staining with Congo red (Fig. 3C), and display a non-b-secondary structure when subjected to CD (Fig. 3D). These amorphous species share some resemblance with the large unstructured aggregates found after the addition of EGCG to intrinsically disordered Ab and other amyloid proteins (27). Similar to EGCG-derived Ab aggregates, TTR-S-derived aggregates display a unique secondary structure that results in reduced thioflavin T fluorescence and a lack of birefringence upon binding to Congo red (Fig. 3, A and C).
Ab42 amorphous aggregates generated upon binding to TTR-S are more sensitive to protease digestion (Figs. 4 and 5). As discussed above, TTR inhibits Ab oligomer toxicity by coaggregation into large nontoxic assemblies (9). However, those studies did not assess protease sensitivity of Ab species. We observe that TTR-S-derived aggregates, from both soluble and fibrillar Ab, are more sensitive to proteinase K than preformed Ab fibrils. We speculate that TTR may promote Ab clearance in vivo by a similar mechanism; this is, the formation of large assemblies that are more prone to digestion and clearance.
Finally, we evaluated the inhibitory effect of TTR-S on Ab amyloid seeding. As shown previously, we found that the addition of ex vivo fibril extracts to soluble Ab accelerates fibril formation (28,29). In contrast, we found that TTR-S inhibits amyloid seeding catalyzed by Ab extracted from the brain tissue of an AD patient (Fig. 5C).
In summary, our results suggest that the dissociation of the TTR tetramer into monomers is required to prevent cytotoxicity from Ab oligomers. In addition, we found that a segment derived from TTR, TTR-S, exerts antiamyloid activity, thereby Figure 5. Evaluation of the inhibitory effect of TTR-S by protease digestion and amyloid seeding. A, anti-Ab42 immunodot blot of Ab42 aggregates after proteinase K digestion. Soluble Ab42 and preformed fibrils were incubated with TTR-S. Half of the sample was subjected to centrifugation, and soluble (S) and insoluble (I) fractions were collected. Increasing concentrations of proteinase K were added to the other half of the sample and incubated at 37°C for 1 h. The samples were analyzed by dot blot using E610 specific anti-Ab42 antibody. As negative control, we included a sample with TTR-S alone. B, the signal was quantified using ImageJ and plotted as shown. This assay shows that Ab42 amorphous aggregates generated after the addition of TTR-S are degraded by proteinase K more readily than Ab42 fibrils. C, amyloid seeding assay followed by ThT signal. Soluble Ab42 was incubated with AD ex vivo seeds in the presence or absence of TTR-S. Samples with only TTR-S, with only AD seeds, or with buffer were included as negative control. Note that the addition of TTR-S results in inhibition of Ab42 seeding (n = 3). All replicates are shown.
inhibiting Ab42 cytotoxicity and amyloid seeding. We also found that TTR-S binds to soluble and fibrillar species, causing a structural rearrangement that leads to an increase of protease sensitivity. Finally, we found that TTR-S inhibits amyloid seeding catalyzed by Ab fibrils extracted from the brain of an AD patient. Our results encourage further evaluation of the inhibitory effect of TTR-derived peptides by additional experimentation. Additional experiments may include independent cell death assays to confirm our MTT-based cytotoxicity assays and evaluation of the species that result from the inhibition of seeded polymerization by ex vivo fibrils using techniques such as EM imaging, CD, and protease digestion analysis. The present study represents an expansion of the current knowledge on the mechanism of protection of TTR over Ab42 cellular toxicity and opens a potential therapeutic avenue.

Patients and tissue material
Post-mortem brain tissue from the occipital lobe of an 83year-old female patient of Alzheimer's disease was obtained from Dr. Vinters at the Pathology Department of the University of California, Los Angeles. The patient was previously evaluated for the presence of amyloid plaques by immunohistochemistry and pathologically diagnosed for Alzheimer's disease. The University of California, Los Angeles Office of the Human Research Protection Program granted exemption from internal review board review because the specimen was anonymized.

Recombinant protein purification
TTR mutants were cloned and purified as previously described (18). Briefly, exponentially growing Escherichia coli Rosetta TM (DE3)pLysS competent cells (Millipore) were treated with 1 mM of isopropyl b-D-thiogalactopyranoside for 3 h. Mutant TTR were purified by nickel-affinity chromatography using HisTrap columns (GE Healthcare), followed by gel-filtration chromatography using a HiLoad 16/60 Superdex 75 column (GE Healthcare) running on an AKTA FPLC system. b-Amyloid peptide (Ab42) was overexpressed through E. coli recombinant expression system and was purified as reported previously (30). The fusion construct for Ab42 expression contains an N-terminal His tag, followed by 19 repeats of Asn-Ala-Asn-Pro, TEV protease site, and the human Ab42 sequence. Briefly, the fusion construct was expressed into inclusion bodies in E. coli BL21(DE3) cells. 8 M urea was used to solubilize the inclusion bodies. Fusion proteins were purified through HisTrap HP columns, followed by reversed-phase HPLC. After TEV cleavage, Ab42 peptide was purified from the cleavage solution by reversed-phase HPLC followed by lyophilization. To disrupt preformed aggregation, lyophilized Ab42 was resuspended in 100% hexafluoroisopropanol, which was finally removed by evaporation.
Size-exclusion chromatography with multiangle light scatter and refractometer detection Components of 0.5-mg protein samples were separated by a silica-based size-exclusion column (Toso Biosep G3000SWXL; 5 mm beads, 7.8 3 300 mm with guard column). The elution buffer contained 0.025 M NaH 2 PO 4 , 0.1 M Na 2 SO 4 , 1 mM NaN 3 , which was weighed on an analytical scale, titrated together to pH 6.5 with NaOH, and diluted to final volume in a 2-liter volumetric flask. The flow rate was 0.4 ml/min. Samples were 0.1mm filtered (Millipore UltraFree MC centrifugal filters). Each injection was 40-50 ml. The peaks were detected by UV absorbance at 280 nm (Waters 2487), light scatter, and index of refraction (Wyatt Technologies MiniDAWN and OPTILAB DSP). Molecular weights were calculated with Astra software, from the light scatter and refractometer signals, transferring the dn/dc and normalization parameters determined from the monomer peak of BSA.

TTR aggregation assay
TTR aggregation assays were performed as previously described (31). Briefly, 1 mg/ml TTR sample in 10 mM sodium acetate, pH 4.3, 100 mM KCl, and 10 mM EDTA was incubated at 37°C during a maximum of 7 days. Transmission EM micrographs were taken after 7 days of incubation. Insoluble fractions were sampled and used to follow TTR aggregation by immunodot blot.

Transmission EM
Transmission EM was performed to visualize TTR mutant aggregation and the fibrillation of Ab42 in the presence of TTR mutants or TTR-derived inhibitors. 5 ml of solution was spotted onto freshly glow-discharged carbon-coated EM grids (Ted Pella, Redding, CA, USA). The grids were rinsed three times with 5 ml of distilled water after 3 min of incubation, followed by staining with 2% uranyl acetate for 2 min. A T12 Quick CryoEM electron microscope at an accelerating voltage of 120 kV was used to examine the specimens. The images were recorded digitally by a Gatan 2kX2k CCD camera.

MTT-based cell assay
We performed MTT-based cell viability assay to assess the cytotoxicity of Ab42 with or without the addition of TTR mutants or TTR-derived peptide inhibitors. A CellTiter 96 aqueous nonradioactive cell proliferation assay kit (MTT) (catalog no. G4100, Promega, Madison, WI, USA) was used. Prior to toxicity test, HeLa, PC-12, and SH-5YSY cells were plated at 10,000, 15,000, and 10,000 cells/well, respectively, in 96-well plates (catalog no. 3596, Costar, Washington, D.C., USA). The cells were cultured in 96-well plates for 20 h at 37°C in 5% CO 2 . For Ab42 and inhibitors samples preparation, purified Ab42 was dissolved in PBS at the final concentration of 10 mM, followed by the addition of TTR mutants or TTR-derived inhibitors at indicated concentrations. The mixtures were filtered with a 0.2-mm filter and further incubated for 16 h at 37°C without shaking for fiber formation. To start the MTT assay, 10 ml of preincubated mixture was added to each well containing 90 ml medium. After 24 h of incubation at 37°C in 5% CO 2 , 15 ml of dye solution (catalog no. G4102, Promega) was added into each well. After incubation for 4 h at 37°C, 100 ml of solubilization solution/stop mix (catalog no. G4101, Promega) was added to each well. After 12 h of incubation at room temperature, the absorbance was measured at 570 nm with background absorbance recorded at 700 nm. Three replicates were measured for each of the samples. The MTT cell viability assay measured the percentage of survival cell upon the treatment of the mixture of Ab42 and inhibitors. The cell viability (%) after treatment with Ab42 with and without TTR-derived peptide inhibitors was calculated by normalizing the cell survival rate using the PBS buffer-treated cells as 100% viability and 2% SDS-treated cells as 0% viability.

Computational docking of fibrillar KLVFFA and monomeric TTR
Protein-protein docking between monomeric TTR and Ab42 fibrillar segments was performed using the ClusPro server (32). Monomeric TTR was generated by removing chain B of the asymmetric unit of PDB code 4TLT (18). Three fibrillar polymorphs of the Ab42 segment KLVFFA were analyzed: PDB codes 2Y2A, 2Y21, and 3OW9 (17). To increase binding surface, only one sheet of each polymorph was included in the modeling.

Thioflavin T fibrillation assay
Purified Ab42 was dissolved in 10 mM NaOH at the concentration of 300 mM. Ab42 was diluted into PBS buffer at the final concentration of 30 mM and was mixed with 30 mM ThT and different concentrations of TTR mutants or TTR-derived peptide inhibitors. The reaction mixture was split into four replicates and placed in a 394-well plate (black with flat optic bottom). The ThT fluorescence signal was measured every 5 min using the Varioskan plate reader (Thermo Scientific, Inc.) or FLUOstar Omega plate reader (BMG Labtech) with excitation and emission wavelengths of 444 and 484 nm, respectively, at 37°C.

Western and immunodot blot
The aggregation of His-tagged TTR mutants was followed by immunodot blot analysis as described by Saelices et al. (18) using SuperSignal ® West HisProbe TM kit following the manufacturer's instructions (Life Technologies). Briefly, 100 ml of samples was spun at 13,000 rpm for 30 min, and the pellet was resuspended in the same volume of fresh buffer and spun again. The final pellet was resuspended in 6 M guanidine chloride and dotted onto nitrocellulose membranes (0.2 mM, Bio-Rad). We used a concentration of the HisProbe antibody of 1:10,000. The aggregation of Ab42 was followed by Western and immunodot blotting analysis. For the Western blots, 7.5 ml of 6 mM samples were separated by SDS-PAGE (NuPAGE 4-12% Bis-Tris gel, Life Technologies) or native gels (NativePAGE 4-16% Bis-Tris gel, Life Technologies) and transferred onto a nitrocellulose membrane (iBlot ® 2 NC mini stacks, Life Technologies) by iBlot ® 2 membrane transfer system (Life Technologies). For the immunodot blot analysis, 2-15-ml of samples were dotted onto a nitrocellulose membrane (Bio-Rad). We used concentrations of 6E10 antibody of 1:2000 (Biolegend®) (33), OC antibody of 1:25,000, and A11 antibody of 1:500 (Millipore and Life Technologies, respectively) (34). The horseradish peroxidaseconjugated anti-mouse IgG antibody (Sigma-Aldrich) was used as secondary antibody at a concentration of 1:2000, and anti-rabbit IgG antibody (Thermo Scientific) was used to detect OC and A11 at a concentration of 1:1000. The membrane was finally developed using the SuperSignal ® West Pico chemiluminescent substrate kit (Life Technologies), following the manufacturer's instructions.

CD spectroscopy
Secondary structures of Ab42 samples were analyzed by CD spectroscopy. The samples (200 ml) were placed into a 1-mmpath length quartz cell (Hëllma Analytics). A Jasco J-810 UVvisible spectropolarimeter was employed. The spectra were obtained in a wavelength range of 190-250 nm, with a time response of 2 s, a scan speed of 50 nm/min, and a step resolution of 0.1 nm. Each spectrum was the average of five accumulations. All of the samples were assayed at a concentration of 50 mg/ml in 15 mM phosphate buffer (pH 7.4). The spectra were recorded at 25°C. The results are expressed as the mean residue molar ellipticity, ½u ¼ u 3 100 3 M C 3 l 3 n where u is the ellipticity in degrees, l is the optical path in cm, C is the concentration in mg/ml, M is the molecular mass, and n is in the number of residues in of the sample molecule. The ellipticity of the samples of Ab after addition of TTR-S was normalized by using the ellipticity of TTR-S as a blank. The percentage of the various structural conformations was calculated by using the software package CDPro, which includes three programs to determine the secondary structure fractions: CONTIN, SELCON, and CDSSTR (35). The results are displayed as the mean values and standard deviations of the three methods.

Proteinase K assay
Soluble Ab42 was prepared in PBS and filtered (0.22 nm). Ab42 fibrils were obtained after 1-2 weeks of incubation in PBS at room temperature. Three-fold molar excess of TTR-S was added to soluble or fibrillar Ab42 and incubated for 16 h. Soluble and insoluble fractions were extracted by centrifugation of half of the sample prior to incubation. PK digestion was performed in a reaction volume of 100 ml containing 10 mM of Ab42 samples and varying concentrations of PK (from 0.2 to 50 mg/ml) in 0.1 M Tris-HCl buffer, pH 7.5, for 1 h at 37°C. The reactions were quenched with 1 mM phenylmethylsulfonyl fluoride, and the samples were analyzed dot blot using E610 specific anti-Ab42 antibody. As negative control, we included a sample with TTR-S alone. The signal was quantified using ImageJ.

Statistical analysis
Statistical analysis of cell assays was performed with Prism 7 for Mac (GraphPad Software) using an unpaired t test. All samples were included in the analysis. All quantitative experiments are presented as independent replicates or as means 6 S.D. of at least three independent experiments.

Data availability
The data sets and images used and/or analyzed during the current study are available from the corresponding author on reasonable request. All remaining data are contained within the article.