The Pathogenic A2V Mutant Exhibits Distinct Aggregation Kinetics, Metal Site Structure, and Metal Exchange of the Cu2+- Aβ Complex
Abstract: A prominent current hypothesis is that impaired metal ion homeostasis may contribute to Alzheimer’s Disease (AD). We elucidate the interaction of Cu2+ with wild-type (WT) Aβ1-40 and the genetic variants A2T and A2V which display increasing pathogenicity as A2T
Aggregation kinetics of all three variants (WT, A2T, and A2V) was elucidated by Thioflavin T (ThT) fluorescence spectroscopy (Figure 1, S1-S5, and Tables S1-S4). In this experiment series we used edta for a dual purpose: 1) In the absence of added Cu2+ to bind trace amounts of metal ions, and 2) to control the concentration of free Cu2+: With 9 µM edta and 0, 10, and 20 µM Cu2+ in the three series (Figure 1A-C), thus effectively having 0, 1, and 11 µM (or 0, 0.05 and 0.55 eq. with respect to Aβ) free Cu2+. In itself edta does not appear to affect the aggregation kinetics (Figure S6). In the absence of Cu2+ only minor differences in the lag time are observed for the three variants, and qualitatively, the expected concentration dependence of the lag time and final ThT fluorescence is observed (Figure 1A, S1 and S5). Day to day and batch to batch variations are presented in Figures S7-S8, indicating that these observations are reproducible. Similar results have previously been obtained[30-33], although with some controversy concerning the order of the lag times for the three variants. We show that there is a batch to batch variation of this order, see Figure S9. Upon addition of Cu2+ the lag phase is extended for all three variants in agreement with literature[34–36] (Figure 1, A-C). Most interestingly, the A2V variant is particularly susceptible to the presence of Cu2+, and exhibits a much extended lag phase as compared to the WT and the A2T variant. This may be pathogenically relevant, as the oligomers formed prior to fibrillation are believed to be the pathogenic species[37-39]. Moreover, Cu2+ addition gives rise to a rapid low- intensity ThT response, most prominent for the A2V variant (Figure 1F). CD spectroscopic and AFM data, vide infra, imply that this early weak ThT response reflects formation of large amorphous aggregates with no well-defined secondary structure, or a spectrum of species with differing secondary structure. Even very low free Cu2+ concentration (1 µM free Cu2+ / 20 µM A, Figure 1B) significantly increases tlag, and does so particularly for the A2V variant. Thus, with as little as 0.05 eq. Cu2+, all the peptides are affected by the presence of Cu2+. Accordingly, a simple model based on the high affinity of Aβ for Cu2+,[40] where Cu2+ binds to 5% of the peptides and leaves the remaining 95% unaffected, is not in agreement with the data.
This conclusion is supported by the paramagnetic relaxation enhancement (PRE) recorded in 1H-NMR, where all peptides are affected by the presence of 0.1 eq. Cu2+ within a very short time span, vide infra. These findings agree with the sub-stoichiometric early Cu2+-induced non-fibrillar aggregation observed for the wild- type Aβ by Pedersen et al.[41] Higher free Cu2+ concentrations, yet still at sub-stoichiometric conditions (11 µM free Cu2+ / 20 µM A, Figure 1C) further delays the fibrillation. All the observed trends of the ThT experiments for the A2T, WT, and A2V variants of Aβ are qualitatively maintained at low ionic strength (Figure S3), and when no shaking is applied (Figure S4), i.e. these conclusions are robust against such changes in experimental conditions.AFM data (Figure 2 A-C, Figure S10 A-C) recorded for samples that were dropcasted on mica sheets and dried for 4 hours, display no significant indication of aggregation as compared to the background (Figure S13). This observation corroborates the ThT fluorescence data, indicating no significant aggregation at early times in the absence of Cu2+, and is further supported by the CD spectroscopic data (Figure 3A), indicating no significant presence of ordered secondary structure, vide infra. After two weeks fibrils appear for both the WT and A2T variant, while the A2V variant has assembled into aggregates with no clear presence of fibrils (Figure 2G-I, Figure S11 A-C). After three weeks (Figure 2 M-O, Figure S12 A-C), the WT and A2T variant remain fibrillary (Figure 2M and 2N), and while the A2V variant exhibits clear fibril-like structures, the morphology and length differ significantly from the other two variants (Figure 2O, Figure S12C). The fibrils are considerably longer, the height distribution significantly shifted towards larger values, and there appear to be “beads” deposited along the fibrils, possibly reflecting amorphous aggregates associating with the fibrils.
Addition of Cu2+, controlling the free Cu2+ concentration using edta in the same manner as in the ThT fluorescence, vide supra, gives rise to formation of large aggregates for all three variants (Figure 2 D-F, J-L, P-R) already after 4 hours, and overall these large aggregates persist for the duration of the experiment series. The results are in agreement with literature data for the WT.[42] literature,[30,42] and although there are minor changes upon addition of Cu2+, no obvious formation of ordered secondary structure occurs at short time scales (Figure 3C).Figure 3B and 3D display CD spectra recorded for the end point samples of the ThT time trace (Figure S3, Table S3) without and with Cu2+ present, respectively. In the absence of Cu2+ the spectra are very similar for WT and A2T, while the A2V variant gives a distinct signal. This is intriguing since all three samples have reached essentially the same final level of ThT response (Figure S3C). The difference cannot originate simply from differences in the concentration of Aβ because the absorption spectra exhibit only minor differences (Figure S14). This indicates that the secondary structure composition of fibrils of A2V differs from that of the other two variants.The CD spectra are typical for antiparallel β-sheet structure. However, these features are clearly reduced for the A2V variant, possibly due to signal cancellation from contributions by less ordered species (Table S5), in agreement with the distinct morphological properties as probed by AFM. Thus, the CD and AFM data indicate that the A2V variant forms more complex aggregates, encompassing both fibrillar structures and components richer in helix or coil structure, even when the fibrillation has converged towards the final ThT fluorescence. In the presence of Cu2+, interestingly, the A2V variant still displays no well-defined secondary structure after 90 hours in the Cu2+- induced prolonged lag phase (Figure 3D). This implies that the A2V peptides are trapped in Cu2+-induced aggregates with no well-defined or a spectrum of secondary structures prior to fibrillation. The WT and A2T variant have fibrillized (Figure S3D) and display very similar CD spectra indicating well-defined secondary structure (Figure 3D), similar to spectra observed for the WT after eight days by Kirkitadze et al., which were interpreted as indicating increased α-helix character.[43] Based on deconvolution of the CD data, the change induced by Cu2+ corresponds to a ~10% lower content of antiparallel -sheet and an increase in both helix and random coil (~6-7%) content (Table S5).
Paramagnetic relaxation enhancement (R1,PRE) for the aromatic region of the 1H-NMR spectrum for all three Aβ1-16 variants is shown in Figure 4A (for 1H-NMR spectra see Figure S15). The observed PRE of the His6, His13, and His14 residues is higher than for the others, reflecting that they are involved in copper coordination in the WT,[23,28] and that this is also the case for the two genetic variants. The PRE for WT and A2T are similar, while for A2V it is significantly higher. The experiments are carried out with only 0.1 eq. Cu2+ relative to the peptide, i.e. the rapid relaxation of protons of all the peptides indicates that Cu2+ is rapidly exchanged between peptides. To test if this might occur via free Cu2+ and be due to weaker Cu2+ binding of the A2V variant, potentiometric and fluorometric determination of the Kd for metal ion dissociation was conducted, see Figures S16-S18, Table S6. The results demonstrate only minor (less than a factor of ~3) differences in the Kd values between the A2T, WT, and A2V variants, and therefore the affinity of the peptides for Cu2+ is too high for the exchange to occur via free Cu2+.[44] Thus, the PRE experiments reflect Cu2+ exchange occurring via transient formation of metal ion bridged dimers or higher oligomers, and this process is apparently significantly faster for the A2V variant than for the WT and A2T variant residues in A1-16. 300 μM A, 30 μM CuCl2, 60 μM glycine, 150 mM of NaCl,~10% D2O and pH 7.4 at 25 C, no buffer. B) EPR spectra obtained at 77 K for 300 M A in the presence of 240 M CuCl2, 10 mM hepes and 110 mM NaCl at pH 7.5, 8.0, and 8.5.
EPR spectra recorded for all three Aβ1-16 variants (Figure 4B) reveal the coexistence of two Cu2+ coordination geometries. The spin Hamiltonian parameters for component I are g = 2.258, g = 2.05, A = 17710-4 cm-1 and for component II g = 2.230, g = 2.05, A = 15710-4 cm-1 in agreement with literature.[29,45] Despite the fact that the mutations (A2T and A2V) occur near the metal site binding region of the peptide, the parameters of the two sites do not change between variants, but the population differs and depends on pH. The pKa values for the transition from component I to component II is ~7.4 (A2T), ~7.9 (WT), and ~8.4 (A2V) (Figure S19). Importantly, this feature reflects the trend in clinical pathogenicity (A2T