Phosphorylation Interferes with Maturation of Amyloid-β Fibrillar Structure in the N Terminus*

Neurodegeneration is characterized by the ubiquitous presence of modifications in protein deposits. Despite their potential significance in the initiation and progression of neurodegenerative diseases, the effects of posttranslational modifications on the molecular properties of protein aggregates are largely unknown. Here, we study the Alzheimer disease-related amyloid-β (Aβ) peptide and investigate how phosphorylation at serine 8 affects the structure of Aβ aggregates. Serine 8 is shown to be located in a region of high conformational flexibility in monomeric Aβ, which upon phosphorylation undergoes changes in local conformational dynamics. Using hydrogen-deuterium exchange NMR and fluorescence quenching techniques, we demonstrate that Aβ phosphorylation at serine 8 causes structural changes in the N-terminal region of Aβ aggregates in favor of less compact conformations. Structural changes induced by serine 8 phosphorylation can provide a mechanistic link between phosphorylation and other biological events that involve the N-terminal region of Aβ aggregates. Our data therefore support an important role of posttranslational modifications in the structural polymorphism of amyloid aggregates and their modulatory effect on neurodegeneration.

Aggregation of the amyloid-␤ (A␤) 3 peptide into oligomers, protofibrils, and amyloid fibrils is the pathological hallmark of Alzheimer disease (AD) (1). Neurotoxic protein aggregation is often characterized by high levels of posttranslational modifications (2,3). A particularly abundant modification in amyloid deposits is phosphorylation (4). For example, the Tau protein found in intracellular neurofibrillary tangles of AD brains is hyperphosphorylated, and in Lewy bodies in Parkinson disease brains, a large fraction of ␣-synuclein is phosphorylated at serine 129 (4). In the case of A␤, it has been shown that phosphorylation at Ser 26 is particularly abundant in intraneuronal depos-its at very early stages of AD (5), whereas phosphorylation at Ser 8 is detected in the later stages of the disease (6).
Several aspects of protein activity are influenced by phosphorylation of neurodegeneration-related proteins (4). It has been shown that phosphorylation of the Parkinson disease-related ␣-synuclein alters its interactions with other proteins (7) and membrane lipids (8) and plays a role in regulating its subcellular localization (9) and degradation (10 -12). In the case of A␤, phosphorylation at Ser 8 attenuates its proteolytic degradation by certain proteases (13). It has also been suggested that the Ser 8 phosphorylation level defines the biochemical stage of A␤ aggregate maturation, which is associated with the conversion from preclinical to symptomatic AD (6,14). Despite its mechanistic importance, little is known on the effect of phosphorylation on the molecular properties of protein aggregates, especially their structure.
Progress over the past decade has provided substantial insight into the structure of A␤ fibrils (15,16). In most models, A␤ molecules adopt a U-shaped ␤-strand-turn-␤-strand fold. Despite the fact that the N-terminal region of A␤ is mainly unstructured in these fibrillar models, several lines of evidence point to an important role of this region in pathogenic aggregation. In particular, it has been shown that toxic A␤ assembly is enhanced by mutations in this region (A2V, H6R, or D7N) (17,18) as well as N-terminal truncation with or without pyroglutamate formation (19,20). Furthermore, antibodies against the N-terminal region inhibit A␤ amyloid formation (21), and the conformational propensity of the N-terminal region controls the partitioning between A␤ oligomers and protofibrils (22). In addition, phosphorylation at Ser 8 promotes toxic A␤ aggregation both in vitro and in vivo (23) and increases the stability of A␤ aggregates against pressure-and SDS-induced dissociation into monomers (24).
Because of their putative role in the initiation and spreading of neurodegenerative diseases and their potential as therapeutic targets, there is growing interest in elucidating the effects of posttranslational modifications on the molecular properties of protein aggregates (4). In the present study, we used hydrogendeuterium exchange coupled with solution-state NMR techniques and demonstrated that phosphorylation of Ser 8 favors more exposed conformations in the N-terminal region of A␤ in insoluble aggregates. Our data support an important role of posttranslational modifications in the structural polymorphism of amyloid protein deposits and suggest a modulatory role in the course and severity of neurodegenerative diseases.

Results
Monomeric A␤ Remains Disordered after Ser 8 Phosphorylation-First, we investigated the structural features of non-aggregated A␤ by CD and NMR. The far-UV CD spectra of both non-phosphorylated A␤1-40 (npA␤) and Ser 8 -phosphorylated A␤1-40 (pS8A␤) were indicative of largely disordered conformations as manifested in prominent negative minima at ϳ199 nm and weak shoulders at ϳ218 nm (Fig. 1). The onedimensional proton NMR spectra of the two peptides showed poor resonance dispersion in the amide and methyl regions, further supporting their lack of ordered structure (Fig. 2, A and  B). Minor differences between the two spectra were observed in the amide and aromatic regions, including the expected downfield signal that originated from the amide proton of phosphoserine (at ϳ9.25 ppm) and two other peaks from the backbone amide proton of Tyr 10 and the side chain ⑀1 proton of His 6 , both in the vicinity of the phosphorylated residue ( Fig. 2A). In pulsed field gradient NMR experiments, the two peptides showed identical decay curves of NMR signal intensity, indicating that they had the same apparent diffusion coefficient of 8.27 ϫ 10 Ϫ7 cm 2 ⅐s Ϫ1 at 5°C and thus the same assembly state (Fig. 2C). A calculated hydrodynamic radius of ϳ1.6 nm demonstrated that the NMR-visible states of the two peptides were mainly monomeric.
Serine 8 Undergoes Extensive Conformational Dynamics in Monomeric A␤-NMR chemical shifts are sensitive probes of local conformational dynamics in polypeptide chains. The random coil index (25) calculated from backbone chemical shifts (HN, HA, N, CA, and CO) of npA␤ reveals that serine 8 is located in a region of high conformational flexibility in the N-terminal region of A␤ that extends down to residue Glu 11 (Fig. 3). Interestingly, Ser 26 , the second potential site of A␤ phosphorylation, occurs in a region of high mobility too, although the mobility level of these two regions is quantitatively different. It has been shown that phosphorylation of Ser 26 results in a significant loss of conformational plasticity in A␤ that is otherwise essential for the progression of A␤ aggregation toward fibrils (26). It is therefore important to see how Ser 8 phosphorylation may influence the local conformational dynamics in A␤.
Phosphorylation at Ser 8 Alters Local Conformational Dynamics of A␤-The HN and HA chemical shifts obtained by two-dimensional homonuclear experiments showed that the backbone conformation of npA␤ and pS8A␤ differs mainly around the phosphorylation site. The largest chemical shift  A and B, one-dimensional 1 H NMR spectra of npA␤ (black) and pS8A␤ (gray) in the amide/aromatic (A) and aliphatic/methyl (B) regions, respectively. The low chemical shift dispersion of amide and methyl proton resonances is in agreement with the unstructured nature of the two peptides. Marked (*) peaks originate from the amide proton of phosphorylated Ser 8 , the amide proton of Tyr 10 , and the ⑀1H of His 6 . C, NMR signal decay observed for npA␤ (black) and pS8A␤ (gray) at 5°C in pulsed field gradient NMR experiments. To estimate the hydrodynamic radii, the signal decay of dioxane was also recorded. The data show that npA␤ and pS8A␤ have identical diffusion coefficients in agreement with a similar assembly state.
deviations were observed at Ser 8 and Asp 7 followed by residues 4 -16 (Fig. 4A). The amide proton of His 6 was not detectable in npA␤, probably due to severe exchange broadening, but with pS8A␤, a strong His 6 HA-HN correlation peak was observed. The signs of HA shift changes were opposite for residues preceding and following Ser 8 : Phe 4 , Asp 7 , and Ser 8 rose in HA shift, whereas residues 9 -12 fell (Fig. 4A, inset). This suggests a differential influence of phosphorylation on these residues: the conformation of the N-terminal residues gets more extended, whereas succeeding residues tend to become less extended. The latter is in line with enhanced HN(i)-HN(i ϩ 1) NOE peaks of residues Gly 9 -Glu 11 in pS8A␤ (Fig. 4, B and C). The side chain resonances of Asp 7 and Tyr 10 provided further support for the differential impact of Ser 8 phosphorylation on the structural dynamics downstream and upstream of the phosphorylation site. The two HB resonances of Asp 7 , unresolved in npA␤, became well separated following phosphorylation, whereas the two well resolved HB resonances of Tyr 10 in npA␤ lost their dispersion upon phosphorylation (Fig. 5). Notably, a significant increase in the intensity of sequential NOE peaks between the amide protons of residues Asp 23 -Gly 25 and Gly 29 -Ala 30 was also observed, suggesting that a long range conformational change may occur after Ser 8 phosphorylation. Together, our data show that monomeric A␤ preserves its disordered structure in the Ser 8 -phosphorylated state with structural changes mainly induced around the site of phosphorylation.
Phosphorylation at Ser 8 Decreases Compaction of A␤ Aggregates in the N-terminal Region-Next, the effect of Ser 8 phosphorylation on the structure of aggregated A␤ was investigated. First, a control experiment confirmed that pS8A␤ possessed a higher rate and amount of thioflavin T-reactive aggregation than npA␤, in line with previous reports (data not shown) (23). Then we used hydrogen-deuterium exchange coupled to solution-state NMR to provide high resolution information on the backbone structure of A␤ aggregates dependent on Ser 8 phosphorylation. After 7 days of aggregation (37°C with gentle stirring), which led to formation of amyloid fibrillar aggregates, the  after calibration with respect to the NOE peak between Gln 15 side chain protons. Higher intensities of sequential HN-HN peaks for Gly 9 -Glu 11 support a phosphorylation-induced decrease of extended conformation for residues C-terminal to the phosphorylation site. Error bars represent S.D. and were estimated on the basis of the signal-to-noise ratio in the spectra. JULY 29, 2016 • VOLUME 291 • NUMBER 31 insoluble A␤ aggregates were spun down by ultracentrifugation and then resuspended in pure D 2 O in a way that unprotected protons could be exchanged with deuterons. Then, after a second run of ultracentrifugation and following dissociation of the pellet with 5% dichloroacetic acid, 95% DMSO, we obtained two-dimensional 1 H, 1 H TOCSY spectra and evaluated the intensity of cross-peaks between backbone amide protons and non-exchangeable aliphatic protons. For residues 15-40, the ratio of cross-peak intensity between pS8A␤ and npA␤ was 0.54 Ϯ 0.03 (Fig. 6A). Because a similar ratio of ϳ0.55 was observed for the intensity of non-exchangeable methyl resonances, the lower cross-peak intensity of residues 15-40 is attributed to a lower incorporation of pS8A␤ into insoluble aggregates in agreement with the higher propensity of pS8A␤ to form soluble oligomers (23). However, the intensity ratio dropped even further in the N-terminal residues: the smallest ratio was observed for Asp 7 and Ser 8 at ϳ0.14 around the phosphorylation site, and the average ratio for flanking residues Ala 2 -Arg 5 and Tyr 10 -Glu 11 was 0.28 Ϯ 0.05 (Fig. 6A).

Phosphorylation and Structure of Amyloid-␤ Aggregates
To investigate whether the lower peak intensity of N-terminal residues in pS8A␤ had its origin in a higher intrinsic exchange rate around the phosphoserine, we evaluated the collective water-amide proton exchange rates of monomeric npA␤ and pS8A␤ through a one-dimensional CLEANEX experiments at 15°C (Fig. 6B). Water-amide proton exchange  . The N-terminal region is more solvent-exposed in insoluble pS8A␤ aggregates than in non-phosphorylated aggregates. A, intensity ratio between the TOCSY cross-peaks of pS8A␤ and npA␤ in hydrogen-deuterium (H-D) exchange NMR experiments. For residues Gln 15 -Val 40 , the observed ratio of ϳ0.54 (dashed line) is due to a lower amount of insoluble aggregates formed by pS8A␤ when compared with npA␤. The prominent drop observed in the intensity ratio of the N-terminal residues indicates a higher solvent exposure of this region in pS8A␤ aggregates. B, in one-dimensional CLEANEX NMR experiments, monomeric pS8A␤ shows a smaller collective water-amide proton exchange rate than npA␤, excluding the possibility that the intensity profile of the hydrogen-deuterium exchange NMR data (shown in A) arises because of a phosphorylation-induced increase in the intrinsic exchange rates. Error bars represent S.D. and were estimated on the basis of the signal-to-noise ratio of the spectra. rate constants of 26.0 Ϯ 2.5 and 20.5 Ϯ 2.9 s Ϫ1 were obtained for npA␤ and pS8A␤, respectively. Because the intrinsic exchange rate in pS8A␤ is not larger than in npA␤, the possibility that the enhanced efficiency of hydrogen-deuterium exchange in pS8A␤ arises because of its higher intrinsic exchange rate is excluded. Our data thus demonstrate that the peptide backbone in the N-terminal region is more solventexposed in insoluble pS8A␤ aggregates than in aggregates of npA␤.
To further investigate the solvent exposure of the N-terminal regionofA␤aggregates,weperformeddynamicquenchingmeasurements in which the fluorescence emission of Tyr 10 was probed in the presence of the neutral quencher acrylamide. When the lifetime of fluorescence excited states does not significantly vary, the quenching constant (the Stern-Volmer constant (K SV )), which is obtained through the dependence of fluorescence intensities on quencher concentration, reflects how much the fluorophore is exposed to solvent. For monomeric npA␤ and pS8A␤, similar quenching constants of 6.30 Ϯ 0.26 and 6.18 Ϯ 0.17 M Ϫ1 were obtained. In contrast, K SV was significantly higher in pS8A␤ (3.98 Ϯ 0.08 M Ϫ1 ) than in npA␤ aggregates (3.30 Ϯ 0.06 M Ϫ1 ) (Fig. 7). The higher K SV of pS8A␤ aggregates provides further support for the lower compaction of the N-terminal region of phosphorylated A␤ aggregates.
Phosphorylated A␤ Fibrils Exhibit Distinct Sedimentation Behavior-Phosphorylation at Ser 8 enhances the oligomeric and fibrillar aggregation of A␤ (23). The fibrils generated by npA␤ and pS8A␤ are similar in morphology, but they show significant differences in the stability against SDS-and pressure-induced dissociation (24). To further investigate the effects of Ser 8 phosphorylation on the properties of A␤ aggregates, 5-day-aggregated A␤ samples were briefly centrifuged (16,000 ϫ g for 30 min), and the pellets and supernatants were examined separately. NMR measurements of the dissociated pellets showed that the concentration of pS8A␤ in the pelleted aggregates was smaller than in the pellet obtained from the npA␤ sample (Fig. 8) in agreement with higher CD intensities of pS8A␤ in the supernatant (Fig. 9). EM examination further demonstrated that the pS8A␤ supernatant but not the npA␤ supernatant was rich in fibrils (Fig. 10). In addition, the fibrils observed in the supernatant of pS8A␤ exhibited detailed features on their surface that were not present in the few fibrils that were found in the npA␤ supernatant (Fig. 10). The data suggest that Ser 8 phosphorylation alters the surface properties of A␤ fibrils in a manner that reduces their higher order assembly and increases the number of dispersed A␤ fibrils.

Discussion
A␤ aggregation proceeds through several steps in which A␤ forms different conformational and assembly states. Although a U-shaped strand-turn-strand conformation of A␤ is a structural motif commonly preserved along the aggregation pathway (27,28), a high level of conformational rearrangement underlying the interconversion of A␤ aggregates is essential for progression of A␤ aggregation (29). One striking example of the role of conformational plasticity in A␤ aggregation is observed in the turn region around Ser 26 where rigidification of this region by Ser 26 phosphorylation is connected to the prevention of fibrillar A␤ aggregates (26). Similarly, maturation of A␤ fibrillar aggregates in later stages of aggregation involves extension of the ␤-sheet structure toward the N terminus (30). Moreover, the N-terminal region of A␤ fibrils manifests significant structural polymorphism with various degrees of solvent protection due to variations in experimental conditions (31). Our data demonstrate that the conformational dynamics of Ser 8 and its adjacent residues are significantly affected by phosphorylation. In agreement with the crucial importance of local flexibility for structural remodeling of A␤ aggregates, our hydrogen-  JULY 29, 2016 • VOLUME 291 • NUMBER 31 deuterium exchange and fluorescence data demonstrate that A␤ phosphorylation at Ser 8 influences the structure of the N-terminal region of A␤ aggregates, leading to a lower degree of compaction in pS8A␤ aggregates.

Phosphorylation and Structure of Amyloid-␤ Aggregates
Phosphorylation at Ser 8 promotes oligomerization and fibrillation of A␤ (23). In addition, Ser 8 phosphorylation enhances the stability of A␤ aggregates against high pressure-and SDSinduced monomer dissociation (24). Because of a higher compressibility of pS8A␤ aggregates without monomer release, pS8A␤ aggregates undergo a smaller volume decrease upon pressure-assisted monomer dissociation (24). The data reported in the current study demonstrate that phosphorylation at Ser 8 not only influences the stability but also the structure of A␤ aggregates. Although an increase in the stability of pS8A␤ aggregates against pressure-induced dissociation, despite the higher solvent accessibility of pS8A␤ aggregates, seems at first sight counterintuitive, it should be noted that the pressure stability and solvent accessibility methods monitor two different aspects of A␤ aggregate structure. For example, in the npA␤ aggregates, the N-terminal region of A␤ is rather protected from the solvent as evidenced by hydrogen-deuterium exchange and fluorescence quenching data (Figs. 6 and 7). At the same time, this does not exclude the presence of waterexcluded cavities in this region in the structure of npA␤ aggregates, and the lower pressure stability of npA␤ aggregates (24) suggests that indeed this might be the case. In line with this model, the solvent-exposed N-terminal region of pS8A␤ aggregates would contain fewer water-excluded cavities, resulting in a higher resistance of pS8A␤ aggregates against pressure-induced monomer dissociation. In addition, we showed that the volume change upon monomer release from pS8A␤ aggregates is around 3 times smaller than that of npA␤ aggregates (8 versus 25 ml/mol), suggesting that the pS8A␤ aggregates have higher compressibility without undergoing monomer release. According to molecular dynamics simulations (24), the increased number of intramolecular electrostatic interactions in pS8A␤ aggregates and their susceptibility to pressure-induced disruption contribute to the higher compressibility of pS8A␤ aggregates without A␤ monomer release. The combined data show that phosphorylation at Ser 8 modulates both the stability and structure of A␤ aggregates.
Despite the largely disordered structure of A␤ in aqueous solution, several short segments of the A␤ sequence exhibit non-random conformational preferences (32). Of special interest in this study is the segment Ser 8 -Val 12 , which adopts a transient turnlike structure (32,33) and thereby partially brings the FIGURE 8. A smaller amount of pS8A␤ than npA␤ is incorporated into insoluble A␤ aggregates as indicated by one-dimensional proton NMR spectra of npA␤ (black) and pS8A␤ (gray). After 5 days of aggregation, the A␤ samples were centrifuged (16,000 ϫ g for 30 min), and the pellets were dissociated into monomers by 5% dichloroacetic acid, 95% DMSO and examined by NMR. The signal intensity ratio in the methyl region (between 1 and 0.5 ppm) was ϳ0.8 for pS8A␤:npA␤. FIGURE 9. The aggregated pS8A␤ contains higher amounts of soluble A␤ assemblies than npA␤. After 5 days of aggregation, CD spectra of the whole aggregated samples (solid lines) and supernatant fractions (dotted lines; 16,000 ϫ g for 30 min) were recorded. Stronger CD intensities were observed in the supernatant of aggregated pS8A␤ than in that of npA␤. mdeg, millidegrees. N-terminal region of A␤ close to the functionally important central hydrophobic cluster of A␤ (residues Leu 17 -Phe 20 ) (34,35). In addition, residues Phe 20 -Val 24 tend to form a helical turn (32), which is in line with the helical propensity of this segment observed in SDS micelle environments and non-aqueous helix-inducing solvents (36 -38). Helical propensity of residues His 13 -Asp 23 has been suggested to be crucial for the early aggregation of A␤ and its interaction with membranes (39). Local perturbation of A␤ conformation by Ser 8 phosphorylation, as demonstrated by our data, can therefore be associated with subtle long range conformational alterations, further influencing its aggregation behavior. The observation that the fine modulation of A␤ conformation by Ser 8 phosphorylation influences its aggregation is of potential wider interest, especially because A␤ aggregation in vivo usually occurs in various environments such as cellular membrane interfaces where interaction with membrane lipids influences A␤ conformation (40,41).
The N-terminal region of A␤ is host for a wide range of events such as phosphorylation (23), metal binding (28,42), tyrosine nitration (43), and truncation (19,20). These events may modulate the role of A␤ in oxidative damage (44) and induction of inflammation (45) and alter the efficiency of prionlike self-propagation (24) and their interaction with biological targets such as cellular lipid membranes (46). Based on the observation that phosphorylation of Ser 8 increases the N-terminal exposure of A␤ aggregates, we hypothesize that phosphorylation might contribute to a mechanism of cross-talk between different N-terminal modifications of A␤ aggregates and thus influence the biochemical maturation of A␤ aggregates (14) and progressive course of AD. This hypothesis remains to be tested by additional experiments probing different N-terminal modifications dependent on Ser 8 phosphorylation.
In conclusion, phosphorylation of A␤ at Ser 8 changes the structure of the N-terminal region of A␤ aggregates in favor of more solvent-exposed conformations. Our data provide experimental support at the molecular level for the potential role of posttranslational modifications in structural polymorphism of amyloids.

Experimental Procedures
Materials-Synthetic npA␤ and pS8A␤ were obtained from Peptide Specialty Laboratory (Germany); both had a purity of more than 95% and were used without further purification. The 15 N, 13 C-labeled A␤1-40 (non-phosphorylated) was purchased from AlexoTech (Sweden); it had a purity above 95% and was used without further purification. For monomerization, the A␤ powders were dissolved in 20 mM NaOH at 2 mg/ml peptide concentration, and after 1 min of sonication and 30 min of shaking in the cold room (4°C), they were split into 50-l aliquots, flash frozen in liquid nitrogen, and stored at Ϫ80°C until use.
Transmission Electron Microscopy-For EM examination, aggregated samples of npA␤ and pS8A␤ were deposited onto carbon-coated copper mesh grids and negatively stained with 2% (w/v) uranyl acetate. Excess stain was washed away, and the sample grids were allowed to air-dry. Subsequently, the samples were examined using a Philips CM 120 BioTwin transmission electron microscope (Philips Inc., Eindhoven, The Netherlands).
Far-UV CD Spectroscopy-After a two-step solubilization, first in 1,1,1,3,3,3-hexafluoro-2-propanol and then in 100 mM NaOH, synthetic npA␤ and pS8A␤ peptide variants were dissolved in 20 mM sodium phosphate (pH adjusted to 7.40) at a concentration of 0.3 mg/ml. CD spectra were recorded on a Chirascan spectropolarimeter using a cuvette with 1-mm path length at 10°C. The CD spectra were recorded from 260 to 190 nm at 0.5-nm intervals with an acquisition time per data point of 8 s. Temperature control with an accuracy of Ϯ0.5°C was achieved with a heating/cooling accessory using a Peltier element.
Dynamic Fluorescence Quenching Experiment-Tyr 10 fluorescence emission spectra of monomeric and aggregated npA␤ and pS8A␤ samples (50 M in 50 mM sodium phosphate, 50 mM NaCl, pH 7.4) were measured in the presence of specified concentrations of the neutral quencher acrylamide. The aggregated A␤ samples were incubated in aggregation-prone conditions (37°C with gentle stirring) for 72 h. The excitation wavelength was 279 nm, and emission spectra were recorded between 292 and 350 nm. The maximum emission intensity obtained through interpolation was used for Stern-Volmer analysis. K SV was obtained from the following equation.
in which I 0 is the maximal emission intensity obtained in the absence of quencher and [Q] is the molar concentration of the quencher.
Two-dimensional Homonuclear NMR Experiments-NMR samples contained 0.4 mg/ml A␤ in 20 mM sodium phosphate buffer, pH 7.2, and measurements were performed at 5°C. Chemical shift referencing at this temperature was made with respect to external 4,4-dimethyl-4-silapentane-1-sulfonic acid (0.0 ppm). All NMR spectra were processed and analyzed with NMRPipe (47) and Sparky (50). Two-dimensional 1 H, 1 H TOCSY and NOESY spectra were acquired on a Bruker (Germany) Avance 800-MHz spectrometer equipped with a cryogenic probe. The time domain data contained 2,048 and 600 complex data points in t2 and t1, respectively. MLEV17 was used for mixing in the TOCSY experiment with a total duration of 60 ms. The mixing time in the NOESY experiment was 200 ms. Water signals were suppressed through a WATERGATE element. 1 H resonance assignments were made on the basis of the two-dimensional 1 H, 1 H TOCSY and NOESY spectra. The combined HN and HA chemical shift perturbation (CSP) induced by Ser 8 phosphorylation was evaluated using the following equation.
Pulsed Field Gradient NMR-Pulsed field gradient NMR experiments were performed on a Bruker AMX 600-MHz spectrometer equipped with a triple resonance cryogenic probe. The sample contained dioxane as an internal hydrodynamic radius standard and viscosity probe. Sixteen one-dimensional 1 H spectra were collected as a function of gradient amplitude using the stimulated echo sequence with bipolar gradient pulses with gradient strengths increasing from 0.674 to 32.030 G/cm (after correction for the sine shape of the gradient pulses) in a linear manner. The gradient distance (⌬) was 200 ms, and the total Phosphorylation and Structure of Amyloid-␤ Aggregates gradient length (␦) was 4 ms. Peaks in the aliphatic region of the 1 H spectra were picked, and the apparent translational diffusion coefficients were calculated after fitting the intensity versus gradient strength data to a corresponding diffusion equation.
Two-dimensional and Three-dimensional Heteronuclear NMR Experiments-NMR samples containing 0.4 mg/ml 15 N, 13 C-labeled A␤ (20 mM sodium phosphate, pH 7.2) were measured at 5°C on a 700-MHz Bruker spectrometer with cryogenic probe. Chemical shift referencing was made with respect to an internal 4,4-dimethyl-4-silapentane-1-sulfonic acid standard (0.0 ppm). The 1 H, 15 N heteronuclear single quantum correlation, 1 H, 13 C heteronuclear single quantum correlation, HNCA, and HNCO experiments were used for backbone assignment. The residue-specific random coil index and corresponding squared order parameter (S 2 ) values were calculated through the random coil index web server.
Hydrogen-Deuterium Exchange NMR Experiments-npA␤ and pS8A␤ samples (50 M) in PBS were incubated in aggregation-prone conditions (37°C with gentle stirring) for 7 days. Subsequently, aggregated A␤ samples were ultracentrifuged (100,000 ϫ g at 25°C) for 4 h, and the pellet was dried using strips of filter paper, resuspended in D 2 O with gentle vortexing, and incubated at room temperature for 1 h to allow for forward exchange from D 2 O to exchangeable amide protons. After D 2 O treatment, insoluble A␤ aggregates were again collected by ultracentrifugation (100,000 ϫ g at 25°C for 4 h) and carefully dried to remove residual D 2 O. The collected A␤ aggregates were then dissolved in 250 l of 5% dichloroacetic acid, 95% DMSO mixture and immediately transferred to an NMR tube to start data acquisition on an 800-MHz Bruker NMR spectrometer. One-dimensional 1 H, two-dimensional 1 H, 1 H TOCSY, and NOESY spectra were obtained. Mixing times of 60 ms for TOCSY and 200 and 300 ms for NOESY experiments were used. Peak assignments were made through a standard homonuclear sequential assignment strategy and further checked with the assignments reported previously (48). The relative extent of hydrogen-deuterium exchange in npA␤ and pS8A␤ aggregates was evaluated through comparing the TOCSY cross-peak intensities between the two peptides.
Water-Amide Proton Exchange Rates-Intrinsic water-amide proton rates in monomeric npA␤ and pS8A␤ samples (75 M; buffered with 20 mM sodium phosphate, pH 7.4) were measured through one-dimensional CLEANEX-PM experiments on a 700-MHz Bruker spectrometer at 15°C. In this experiment, a selective water excitation pulse was followed by a mixing time ( m ) of durations 5, 10, 15, 25, 50, 100, 200, and 500 ms during which chemical exchange between water and amide protons takes place. The global intensity of amide protons as a function of mixing time was fitted to the following equation.
where V 0 is the intensity in a control experiment, k is the normalized rate constant related to the forward exchange rate constant between water and amide protons, and R 1A and R 1B are apparent longitudinal relaxation rates for protein and water (49).
Author Contributions-N. R.-G. designed and conducted the NMR, fluorescence, and CD experiments; analyzed the results; and wrote the paper. S. K. contributed to CD experiments. J. W. conceived the project. M. Z. conceived the project and wrote the paper.