Role of species-specific primary structure differences in Aβ42 assembly and neurotoxicity

Abstract

A variety of species express the amyloid β-protein (Aβ). Those species expressing Aβ with primary structure identical to that expressed in humans have been found to develop amyloid deposits and Alzheimer’s disease-like neuropathology. In contrast, the Aβ sequence in mice and rats contains three amino acid substitutions, Arg5Gly, His13Arg, and Tyr10Phe, which apparently prevent the development of AD-like neuropathology. Interestingly, the brush-tailed rat, Octodon degus, expresses Aβ containing only one of these substitutions, His13Arg, and does develop AD-like pathology. We investigate here the biophysical and biological properties of Aβ peptides from humans, mice (Mus musculus), and rats (Octodon degus). We find that each peptide displays statistical coil→β-sheet secondary structure transitions; transitory formation of hydrophobic surfaces; oligomerization; formation of annuli, protofibrils, and fibrils; and an inverse correlation between rate of aggregation and aggregate size (faster aggregation produced smaller aggregates). The rank order of assembly rate was mouse > rat > Aβ42. The rank order of neurotoxicity of assemblies formed by each peptide immediately after preparation was Aβ42 > mouse ≈ rat. These data do not support long-standing hypotheses that the primary factor controlling development of AD-like neuropathology in rodents is Aβ sequence. Instead, the data support a hypothesis that assembly quaternary structure and organismal responses to toxic peptide assemblies mediate neuropathogenetic effects. The implication of this hypothesis is that a valid understanding of disease causation within a given system (organism, tissue, etc.) requires the co-evaluation of both biophysical and cell biological properties of that system.

Graphical abstract

Introduction

Alzheimer’s disease (AD) is the most common form of late-life dementia (1). An important working hypothesis of disease causation is aberrant folding and assembly of Aβ42 (2). This assembly leads to the production of extracellular amyloid plaques by the amyloid β-protein (Aβ) and of intracellular neurofibrillary tangles by tau. Such histopathologic findings are pathognomonic for AD and accompany progressive declines in cognitive ability and executive function (1, 3–5). AD pathology also has been observed in dogs, polar bears, rabbits, cows, sheep, pigs, guinea pigs, orangutans, and rhesus monkeys, but until recently, it has not been found in mice or rats (6–8). In 2005, Inestrosa et al. (9, 10) reported that the brush-tailed rat, Octodon degus (Od; a rodent indigenous to Chile), normally expresses neuronal APP695 and displays both intra- and extra-cellular deposits of Aβ, intracellular accumulation of phosphorylated tau, strong astrocytic responses, and pyramidal neurons rich in acetylcholinesterase. Od naturally develop these neuropathological signs of AD between 12–36 months of age. This pathology has been correlated with decreases in spatial and object recognition memory, postsynaptic function, and synaptic plasticity (11, 12). Interestingly, the brains from another rodent, the naked mole rat (the longest lived rodent, with a life span ≈32 years) showed levels of Aβ42 similar to a 3X-Tg-AD mouse model of AD (13).

Examination of the primary structure of Aβ42 in Od and in the naked mole rat reveals 100% identity (13, 14), and this amino acid sequence is identical to that of human Aβ42, except for a His13Arg substitution (Fig. 1). In contrast, mouse (Mus musculus; Mm) Aβ42 differs from human Aβ42 in having three amino acid substitutions, Arg5Gly, His13Arg, and Tyr10Phe (6, 15–17). Mutations in the amyloid β-protein precursor (APP) that result in amino acid substitutions at other sites in Aβ cause familial AD and cerebral amyloid angiopathy (CAA) (1). It is particularly interesting that an identical amino acid substitution in humans, His→Arg (the English mutation), but at His6 rather than His13, may be associated with AD (18). Whether this mutation is causative or simply a polymorphism remains unclear, but incorporation of this substitution into Aβ has been found to substantially increase β-sheet formation, fibril seeding ability, and formation of toxic oligomers (19). It thus is reasonable to hypothesize that the presence of the two additional amino acid substitutions in Mm Aβ42 explains why wild type mice do not show AD-like pathology.

Figure 1. Primary structures of human, Mm, and Od Aβ42.

One-letter amino acid code is used to present the sequences of each Aβ42 peptide. Underlined, italicized letters indicate amino acid differences or substitutions among the peptides.

It has been suggested that species-specific differences in Aβ42 primary structure are key factors controlling the development of AD-like pathology (15, 20). However, some of the work in this area is contradictory (16, 17, 21). Fraser et al. (22) and Hilbich et al. (16) argue that differences in primary structure do not account for the lack of amyloid deposits in the brains of aged rats and mice, and have no effect on the morphology and organization of fibrils. In contrast, Ötvös et al. (15) and Dyrks et al. (21) suggest that subtle interspecies differences in amino acid residues may account for the inability of the rodent peptide to form amyloid fibrils in situ. The reductionist approach of arguing that amino acid sequence differences in Aβ are the sole explanation for species-specific development, or lack of development, of AD-like pathology, ignores the fact that mice aren’t human. Other explanations for the species-specific differences in neuropathology may exist. Here, we investigate how the natural differences in APP gene sequence among humans, Mm, and Od affect the biophysical, biochemical, and biological properties of the respective Aβ peptides.

Results and Discussion

Ion mobility-mass spectrometry (IM-MS) and Aβ42 monomer structure

To probe the effects of primary structure difference (Fig. 1) on the peptide structure and oligomer distributions of all Aβ42 alloforms, we performed an ion mobility coupled mass spectrometry analysis. Ion mobility is capable of separating species that have the same mass-to-charge (m/z) ratio but different conformations or oligomer orders¹ (23). It has been successfully applied to the study of Aβ assembly and its inhibition by small molecules (24–33). We first recorded the mass spectra of all three Aβ42 alloforms immediately after preparation (Fig. 2a–c). All alloforms displayed three peaks, corresponding to the z/n = −4, −3, and −5/2 charge states, where z is the charge and n is oligomer number. It is notable that the intensity of the z/n = −5/2 peak of Mm is relatively high compared to those of Aβ42 or Od. These results suggest that Mm oligomerizes more readily than do Aβ42 and Od. Interestingly, this suggestion is consistent with the observation that Mm clogged the nano-electrospray tips more easily during the experiment than did either Aβ42 or Od.

Figure 2. Ion mobility-mass spectrometry and Aβ42 monomer structure.

Panels A–C: Mass spectra of all Aβ42 alloforms. The charge state z/n is noted for each species where z is the charge and n is the oligomer number. Panels D–F: The ATDs of z/n = −3 for all Aβ42 alloforms. M₁, M₂ and M₃ represent three monomer conformers with different cross-sections.

To evaluate peptide monomer structures, arrival time distributions (ATDs) were determined for the z/n = −3 species (Fig. 2d–f). In previous studies of the ATD of the z/n = −3 Aβ42 species(24, 25), two major features, with arrival times at ~620 (M₁) and 670 (M₂) μs, were observed. M₁ and M₂ were assigned as a compact solvent free-like conformer and an extended solution-like conformer, respectively. Here, the ATD of the z/n = −3 Od also showed two features, which by analogy with the prior samples and analysis of collision cross-sections (σ; Table S2), were assigned as M₁ and M₂. Interestingly, the ATD of the z/n = −3 Mm displayed three features, assigned as M₁, M₂, and M₃. M₃ was not observed with the Aβ42 or Od peptides. This indicates that Mm produces an additional monomer conformer with a more extended structure. As shown in Table S2, the collision cross-sections of M₁ and M₂ monomers for each peptide were very similar, suggesting that all three Aβ42 alloforms have similar monomer conformations (M₁ and M₂). Mm forms one additional extended conformer (M₃), which may be of relevance to its aggregation kinetics.

Ion mobility spectrometry and Aβ42 oligomer distributions

To understand the early oligomer distributions of the Aβ42 species, we recorded ATDs of the z/n = −5/2 peaks (Fig. 3). The ATD of Aβ42 shows four features, with arrival times at ≈710, 680, 620 and 540 μs. These features were assigned as dimer, tetramer, hexamer, and dodecamer based on their collision cross-sections (Table S2; see also Bernstein et al. (26) for a detailed discussion of Aβ42 oligomer assignments). The ATD of Od also shows these four features. However, the ATD of Mm shows only three features, corresponding to dimer, tetramer, and hexamer. Finally, the ATD of the z/n = −2 Mm was recorded and showed predominant features, corresponding to dimer and trimer (Fig. 3d), indicating that Mm formed trimer, something not observed for Aβ42 or Od under the same experimental conditions. The lack of dodecamer formation by Mm is interesting, as it may correlate with the fact that Mm does not naturally display AD-like neuropathology, in contrast to mice expressing the human form of Aβ (34), which express substantial levels of Aβ^★56, a dodecamer linked to memory deficits.

Figure 3. Ion mobility spectrometry and Aβ42 oligomer distributions.

Panels A–C: The ATDs of the z/n = −5/2 peaks for all Aβ42 alloforms. Panel D: The ATD of the z/n = −2 peak for mouse Aβ42. The oligomer number (n) is noted for each feature.

Oligomerization of Aβ42 by PICUP

We performed Photochemical Cross-linking of Unmodified Proteins (PICUP) and SDS-PAGE (Fig. 4), an approach that “freezes” the monomer ⇌ oligomer equilibrium, thus allowing quantitative determination of Aβ42 oligomer size distributions (35). Aβ42 displayed a typical PICUP wild type Aβ42 oligomer distribution (see (36)) comprising a series of oligomers with a prominent node of band intensity at pentamer-hexamer. Mm showed a relatively faint dimer, but intense trimer and tetramer bands. With the exception of the faint dimer band, this oligomerization pattern is identical to that of non-cross-linked Aβ42 (Fig. S1)—which we would expect due to the fact that the Tyr10Phe amino acid substitution in the Mm peptide eliminates the Tyr residue that is highly reactive in the PICUP chemistry. The intense trimer and tetramer bands also were present in the non-cross-linked control Mm sample (Fig. S1), again consistent with a lack of cross-linking of Mm.² The Od PICUP results, on the other hand, are highly similar to that of Aβ42. The His13Arg substitution thus did not appear to alter peptide oligomerization, as measured by PICUP. These results are consistent with the ion mobility results in that Mm forms smaller oligomers than do Aβ42 or Od.

Figure 4. Oligomer distributions determined by PICUP.

PICUP, followed by SDS-PAGE and silver staining, was done to determine the oligomer size distributions of the peptides. ^aOligomer order arrows point to positions of bands from the Aβ42 lane. Some variation in electrophoretic mobility is observed among the other peptides.

Monitoring Aβ assembly by quasielastic light scattering spectroscopy (QLS)

QLS is a method for the non-invasive monitoring of the diffusion coefficients (D) of particles in solution (37). We used QLS, and the Stokes-Einstein equation (see Methods), to derive R_H distributions of each peptide after 0.5 h and 24 h of incubation (Fig. 5). Aβ42 initially produced a distribution composed primarily of small particles of R_H ≈ 8–10 nm (darkened area). Particles of R_H ~ 100–1000 nm also were observed, but because scattering intensity is proportional to the square of molecular weight, these peaks must have been produced by relatively few particles. After 24 h, the contributions to the scattered light intensity of larger particles had increased substantially, consistent with a process of Aβ assembly. However, oligomers of R_H ≈ 8–10 nm remained³. Such oligomers have been reported previously (38). In contrast, the other peptides did not form oligomers of this size. After 24 h of incubation, the distribution of Mm changed little, showing only a slightly increased average R_H of its predominant peak. Deconvolution of R_H distributions depends on the specific parameterizations used, especially that for data smoothing (37). Although deconvolution yielded two peaks for Mm, it is possible that only one heterodisperse population of scatterers was present. For this reason, the average R_H was calculated across the entire distribution. Od showed substantial increases in R_H over 24 h, yielding three nodes of R_H intensity, at ≈90 nm, ≈500 nm, and ≈4000 nm. The average R_H was 1740 nm. Interestingly, the time-dependence of the distributions of R_H showed that those of Aβ42 and Od primarily comprised relatively small scatterers initially ( $\overline{R_H}\approx 10nm$ and $\overline{R_H}\approx 62nm$, respectively) but that after 24 h distribution range expanded to larger sizes with maximal R_H > 1000 nm. Mm was unique in initially displaying larger scatterers $(\overline{R_H}\approx 233nm)$ whose size remained relatively constant over 24 h.

Figure 5. Quasielastic light scattering spectroscopy (QLS).

Intensity (counts/sec) is plotted versus hydrodynamic radius (R_H in nm). The data are from samples monitored ≈. The number in the top right corner of the panel represents the average R_H for the shaded regions of the distributions. Buffer spectra produce no scattering and thus are not shown.

It should be noted that all samples were filtered through a 20 nm porosity filter immediately after preparation. During the first 30 min of monitoring, for all peptides except Aβ42, we observed a rapid increase in scatterer size, reflecting a rapid assembly process. The particle size distributions then were relatively stable while overall scattering intensity continued to rise due to increasing contribution to scattering from large particles. We calculated the rate of growth in the scattering intensity during the first hour of incubation (Table 1). Rates for Aβ42 and Mm were moderate, consistent with relative constancy of their particle size distributions. In contrast, Od displayed a much higher rate, as would be expected from the relatively large scatterer size observed after incubation for 24h (Fig. 5).

Table 1.

Rate of change in scattering intensity. The rate of change in scattering intensity, dI/dt, determined by fitting I_t to a linear function spanning the first hour of incubation. I₀ was determined from the derived line I_t = dI/dt × t + I₀ by substitution of t=0. Both dI/dt and I₀ depend on the aperture in which light is collected, which in turn is a function of the instrument used. To determine the instrument-independent rate of change in relative intensity, we report the normalized quantity (dI/dt)/I₀. We note that these quantities are precise, but likely not absolutely accurate, because the intensity versus time dependency over this time interval is significantly non-linear.

Peptide	(dI/dt)/I0
Aβ42	0.05
Mm	0.09
Od	1.43

Taken together, the data on the time-dependent evolution of R_H and scattering intensity suggest that Mm rapidly forms small aggregates that are relatively stable, whereas both Aβ and Od display a lag time before frank fibril assembly is observed.

Thioflavin T binding

ThT binding was performed to monitor the assembly of β-sheet-rich structures (39–41) (Fig. 6). Aβ42 displayed a classical ThT fluorescence curve starting at a low fluorescence level (~25 FU), remaining constant for ≈1 h, and then monotonically increasing during the next 5 h. In contrast, Mm fluorescence tripled within the first hour. Od displayed a somewhat higher initial level of ThT fluorescence than did Aβ42 but only a gradual increase during the initial 6 h. We continued monitoring for 21 d, during which time all three peptides displayed increasing rates of increase of ThT fluorescence (dFU/dt), until a plateau level was reached. The final fluorescence level of Aβ42 was substantially lower than that of Mm (≈425 versus ≈800 FU) and somewhat lower than that of Od (≈425 versus ≈500 FU). It was interesting that the maximal value of dFU/dt for each peptide was similar, but this phase of assembly occurred substantially earlier for Od (7h) than for the other peptides (5d). This observation is consistent with the substantially higher rate of increase in scattering intensity (dI/dt)/I₀; Table 1) of Od compared to Aβ42 and Mm observed using QLS. A similar concordance of ThT fluorescence and QLS data was observed with respect to R_H, for which Mm quickly displayed substantial fluorescence, and the highest R_H (Fig. 5), of the three peptides. The fluorescence of Mm remained relatively constant for the first day, before increasing substantially over the next three weeks. The ThT fluorescence of Aβ42, as with its R_H, was the lowest among the three peptides after the first day of incubation.

Figure 6. Aβ assembly kinetics.

ThT fluorescence was used to monitor the kinetics of β-sheet formation during assembly reactions. Peptides (20 µM) were incubated with 40 µM ThT at pH 7.5 and 37°C with shaking. Data are present on a semi-log plot to allow visualization of data in a single figure for the entire time range (t=0 d, 0.04 d (1h), up to 21 d).

These data are consistent with a reaction coordinate model (Fig. S2) in which Mm has a relatively low activation energy (E_a¹) for formation of relatively stable (ΔG_Ol) oligomers. This produces the rapid kinetics for oligomer formation and the relatively long lifetimes of the oligomers thus formed. However, thermodynamically, the free energy of fibril formation, ΔG_ff, is similar to those of Aβ and Od (Table S1; Fig. S2), thus eventually Mm does form fibrils.

Bis-ANS Fluorescence

We monitored exposure of hydrophobic surfaces during peptide assembly using Bis-ANS fluorescence (Fig. 7). Bis-ANS is a hydrophobic molecular probe that is essentially non-fluorescent in water but fluorescent in nonpolar or hydrophobic environments. This property makes it a sensitive indicator of protein conformation (e.g., molten globules, exposed hydrophobic surfaces, or native folds) (42–46).

Figure 7. Bis-ANS fluorescence.

Aliquots from Aβ assembly reactions were removed at regular intervals, mixed with Bis-ANS, and then monitored for fluorescence (as arbitrary fluorescence units (FU)).

Aβ42 showed a rapid but modest increase in fluorescence during the first 4 h of incubation (to ≈200 FU), after which a slow monotonic decrease in fluorescence was observed. Mm displayed an initial level of fluorescence approximately twice that of Aβ42 that eventually reached ≈950 FU, the highest among these peptides. Fluorescence then decreased monotonically with time. Od displayed initial fluorescence levels identical, within experimental error, to that of Mm. The peak level of Od fluorescence was ~600 FU, which occurred at ≈24 h, later than for Aβ42 or Mm. Experiments also were done with SYPRO Orange (data not shown), a dye with properties similar to that of Bis-ANS. The rates of fluorescence increase and the plateau levels of fluorescence produced by SYPRO Orange exhibited the same rank order as with Bis-ANS, namely Mm > Od > Aβ42.

The rapid increase in fluorescence in the Mm sample suggests early exposure of hydrophobic surfaces. The formation of such surfaces has been postulated to occur as the intrinsically disordered Aβ monomer begins folding (47, 48). This phenomenon would produce rapid aggregation, as seen by ThT (Fig. 6). The different initial fluorescence levels observed between Od and Aβ42 also is consistent with the ThT fluorescence data, which showed higher initial levels of β-sheet in Od compared to Aβ42.

Secondary structure dynamics

We used circular dichroism spectroscopy (CD) to determine the time-dependence of the distribution of secondary structures for each peptide (Fig. S3) (49, 50). Aβ42 displayed statistical coil structure initially, but then exhibited a progressive increase in β-sheet content that was clear within hours and produced a classical β-sheet spectrum at 24 h. The spectra appeared to display an isodichroic point at ≈208 nm. This isodichroic point was observed with all three peptides, suggesting the existence of a two-state transition (likely statistical coil→α-helix) (51). Mm exhibited some β-sheet content immediately after preparation. Progressive increases in β-sheet were observed throughout the 24 h monitoring period, as with Aβ42. However, the level of β-sheet was higher than in Aβ42, as indicated by the relative increase in absolute values of the molar ellipticities at ≈195 and ≈215 nm. Od had the least statistical coil content of any of the peptides when its CD spectrum was first acquired. Its initial β-sheet content was similar to that of Mm, and β-sheet content displayed a progressive increase with a rate similar that of Aβ42.

To visualize the relative rates of change in β-sheet content, we plotted [Θ]₂₁₅ versus time (Fig. 6). Mm showed the highest initial β-sheet content (the most negative [Θ]₂₁₅), the fastest increase in the level of this secondary structure element (largest |d[Θ]₂₁₅/dt|), and the highest overall β-sheet content at the end of the experiment (the most negative [Θ]₂₁₅). Od also displayed relatively high initial β-sheet content, but the evolution of β-sheet occurred at a slower rate. Aβ42 had the lowest initial β-sheet level, but this level eventually reached that of Od. The high initial β-sheet in the Mm peptide is consistent with its rapid oligomerization. The monotonic increase in β-sheet for all three peptides suggests a progressive assembly phenomenon that was confirmed in subsequent EM experiments (see below).

Assembly Morphology

We used electron microscopy to determine the morphologies of assemblies present at the initiation of peptide incubation (day 0) and after the assembly process was complete (day 7) (Fig. 9). Small structures (“globules”) were observed initially in the Aβ42 sample. These globules ranged in diameter from 11–23 nm (white arrows, Fig. 9), often displaying pore-like central cavities filled with uranyl acetate stain. These annuli had outside diameters ranging from ≈13–25 nm and inside diameters ranging from ≈4–11 nm. Mm, in contrast, formed aggregates comprising short beaded chains of different diameters (black arrows) and irregular structures (open arrows). These irregular assemblies ranged in diameter from 11–21 nm and had lengths from 16–75 nm. Od displayed filamentous aggregates (open arrows), as well as occasional long filaments to which smaller assemblies appeared to be associated (curved arrow). Small globules (black arrows) were interspersed among these other structures. The globules range in diameter from 11–54 nm. The diameters of the long fibrils ranged from ≈7–19 nm.

Figure 9. Assembly morphology.

EM was used to determine assembly morphology. Aliquots of assembly reactions were removed at days 0 and 7. Determination of geometric parameters (lengths, diameters) was done using Image J software. Scale bars are 200 nm.

Considering the heterogeneity of assembly morphologies observed immediately after sample preparation, it was interesting that by 7 days all samples formed long straight or curved fibrils. Aβ42 predominately formed fibrils of quite different lengths (56–303 nm) and with diameters of ≈7–12 nm (Table 2). Long fibrils had a diameter of 9–21 nm. The Mm sample contained numerous fibrils that were thinner than those of Aβ42 (Table 2), but also abundant smaller curved “protofibril-like” structures (black and white arrows) (52). Od formed long and short fibrils interspersed with numerous globules. Bifilar structures with helical twist were observed (thick black arrow) along with thinner fibrils, often with helical twists (thin black arrows). A number of annuli also were observed (white arrows). The results of electron microscopic examination of assembly morphology correlated with the results from the prior experiments. For example, the initial rank order of average assembly size was Mm > Od > Aβ42, and as would be predicted from the QLS, ThT, and CD experiments, large (fibrillar) assemblies were observed after 7 d. Mm continued to behave somewhat uniquely in that the distribution of assemblies at day 7 included abundant protofibrillar species (not seen with the other peptides) and narrower fibrils. The continued presence of the small protofibrils is consistent with the relatively constant distribution of scatter sizes seen during the first 24 h by QLS.

Table 2.

Thermodynamic properties of fibril formation. peptides were incubated at a final concentration of 20 μM at 37°C for 21 days without shaking. C_r is the molar concentration of the peptide in the supernate after ultracentrifugation (436,000 × g for 1 h) after cessation of fibril growth. ΔG⁰ =−RT ln (1/C_r). p-values are relative to Aβ42. The results are from four independent experiments.

Peptide	Cr (μM)	p-value	ΔG0 (kcal/mole)	p-value
Aβ42	1.10		−8.45 (0.12)
Mm	0.45	0.065	−9.20 (0.65)	0.123
Od	4.04	0.016	−7.64 (0.14)	<0.001**

Neurotoxic activities of Aβ assemblies

MTT assays were performed to evaluate the effects of freshly prepared peptides on MTT metabolism (53). The assay was performed both on PC12 cells (Fig. 10A) and rat primary cortical neurons (Fig. 10B). Aβ42 caused an ≈23% decrease in MTT metabolism in PC12 cells, whereas Mm and Od were non-toxic. The differences in toxicity between Aβ42 and each of the other peptides was highly significant (p<0.001). The data from primary cortical neurons were similar to those from PC12 cells. Quantitatively, the levels of toxicity of all peptides assayed with primary cortical neurons was greater than with PC12 cells. LDH assays were performed to measure cell death. The results of these assays were consistent with those of the MTT assays (Fig. 10C, D). Aβ42 was most toxic while Mm and Od were least toxic.

Figure 10. Peptide neurotoxicity.

To determine the effects of the different peptides on cellular metabolism, MTT assays were performed using (A) differentiated PC12 cells or (B) rat primary cortical neurons. Freshly prepared peptides were added to the cultures for 24h at 37°C, after which MTT was added and incubated for 4h at 37°C, stop solution was added to the cultures, the cultures were incubated overnight, and then formazan optical absorbance was measured at 570–630 nm (OD_570–630). Data are representative of that obtained in three independent experiments (6 wells per data point). The data are normalized to the media control group and expressed as mean ± the standard error of the mean (SEM). Statistical significance between samples is indicated by asterisks (*** and ** indicate p values < 0.001 and <0.01 respectively). NS is “not significant.” To determine cell death, LDH activity was measured in the media from (C) differentiated PC12 cells or (D) rat primary cortical neurons after 48 h of incubation of the cells with the different peptides. Data are representative of that obtained in three independent experiments (6 wells per data point). The data are normalized to the media control group and expressed as mean ± the standard error of the mean (SEM). Statistical significance between samples is indicated by asterisks (*** and ** indicate p values < 0.001 and <0.01 respectively). NS is “not significant.”

To measure the level of apoptosis induced by freshly prepared Aβ, SHSY-5Y cells were stained for phosphatidylserine using Annexin-V conjugated FITC (54, 55). Propidium iodide (PI) staining was done concurrently to reveal cell nuclei. Negative control samples displayed only red nuclei, but no cell surface Annexin-V fluorescence. Positive control samples incubated with 1 μM staurosporine, showed complete disruption of membrane integrity (data not shown). All peptides produced green fluorescence, indicating binding of Annexin-V (Fig. 11). Quantitative analysis of the number of Annexin-V-positive (fluorescent) cells per 100 μm² field (Fig. 11B, bar graph) revealed an average of 4.7 for Aβ42 and 4.2 for Od, a difference that was not statistically significant. Mm showed significantly lower numbers (3.3; p<0.01) than did Aβ42. The difference in fluorescence between Od and Mm was not significant.

Figure 11. Monitoring apoptosis.

(A) Annexin-V staining of SH-SY5Y cells treated with each of the peptides was performed to estimate apoptosis levels. The SH-SY5Y cells were grown on cover slips and treated with freshly prepared Aβ42 peptides for 24h, after which they were stained with Annexin V-FITC. Images were recorded using confocal laser scanning microscopy. (B) Bar graph: Quantitation of Annexin-V staining. The number of apoptotic cells was counted in a 100 μm² area using Image-J software. A minimum of 30 microscopic fields was used in each of two independent experiments. Significance between groups is indicated by asterisks (*** and ** indicate p values of <0.001 and <0.01, respectively). No significant difference existed between Mm and Od, although it is possible that Od trended towards greater apoptotic activity.

Taken together, the results among these assays were consistent and showed a rank order of toxicity of Aβ42 >> Mm ≥ Od. An identical rank order was observed in our Annexin-V fluorescence apoptosis assay. This rank order was the opposite of the rank order observed in our biophysical studies. Considered from the perspective of assembly rate (which is inversely related to assembly size) versus toxicity, our data suggest that assembly populations comprising smaller assemblies are relatively more toxic than are those containing larger assemblies. This result agrees with the hypothesis that the most important neurotoxin in AD are the small, presumably oligomeric, assemblies (56, 57, 58). However, our data extend this hypothesis by providing evidence, through the use of peptides from different species (and of different primary structure), that assembly state, rather than primary structure per se, is likely to contribute most significantly to peptide neurotoxicity.

It is important to note that we do not argue that primary structure is irrelevant with respect to determining peptide neurotoxic activity (59), as some have with respect to the existence of “generic” amyloid structures (60–63). Instead, we suggest that it is the combination of at least three factors that control peptide neurotoxicity: (1) primary structure; (2) assembly structure; and (3) cellular responses. Primary structure determines the intrinsic propensity of a peptide sequence to fold into an energetically determined distribution of tertiary structures, according to Anfinsen (64). It is the basis of Factor #2, which encompasses the vast and complex folding landscape of Aβ and other amyloid proteins (62) that includes assemblies such as oligomers (irregular, globular, annular, worm-like), protofibrils, and fibrils. Factor #3 encompasses how distinct cell types (e.g., neurons, glia, microglia, or non-neuronal) respond to extracellular or intracellular (including cytoplasmic and intra-organellar or intramembranous (plasma, endosomal, lysosomal, mitochondrial, nuclear)) Aβ assemblies. This response includes components of the unfolded protein response, chaperones, and lysosomal activities. It also includes anabolic features of Aβ metabolism, which help to determine steady state concentrations of Aβ in different anatomic sites). Factors #1 and #2 are intrinsic to the Aβ peptide per se. Factor #3 is specific to each type of organism and falls under the rubric of specific organismal responses to neurotoxins, such as Aβ. Organismal responses determine both the assembly space of Aβ aggregates as well as their biological half lives. An excellent example of “organismal control” comes from recent studies of superoxide dismutase 1 (SOD1) aggregation in transgenic mouse models of amyotrophic lateral sclerosis. Bergh et al. (65) report that SOD1 aggregates formed in the brains and spinal cords of these animals differ in structure from those produced in vitro—a clear example of how organismal factors mediate the intrinsic assembly propensities of amyloid proteins and thus may affect their neurotoxic activities.

Conclusions

Taken together, these data do not support long-standing hypotheses that the primary factor controlling development of AD-like neuropathology in rodents is Aβ sequence. If it were, then we would not expect, a priori, to observe the folding and assembly of rodent peptides into intrinsically toxic oligomeric, protofibrillar, and fibrillar structures. Instead, our data support the hypothesis that the factors of assembly quaternary structure and organismal response control development of neuropathology. The implication of this hypothesis is that a valid understanding of disease causation within a given system (organism, tissue, etc.) requires the co-evaluation of both biophysical and physiologic properties of that system. One obvious property that might contribute to amyloid formation is life span. AD amyloidosis in humans is a neuropathologic phenomenon that is age-dependent. Rats of the species O. degus, like the naked mole rat, live substantially longer than do animals of the species M. musculus (≈6–8 in captivity versus ≈2 years) and thus they may be more likely to develop amyloid and experience its attendant effects. Another interesting possibility, although one beyond the scope of study at this time, is that O. degus neurons are more susceptible to Aβ-induced toxicity.

Materials and Methods

Chemicals and Reagents

All chemicals were purchased from Sigma Chemical Co. (Saint Louis, MO) and were of the highest purity available. Water was de-ionized and filtered using a Milli-Q system (Millipore Corp., Bedford, MA). Xpress™ silver-staining kits were from Invitrogen (Carlsbad, CA). Buffers were prepared with sterile, autoclaved water containing 0.002% (w/v) sodium azide. SYPRO Orange dye was purchased from Invitrogen Corp. Annexin-V staining kits were purchased from BioVision, Inc, Milpitas, CA.

Peptide synthesis and preparation

Aβ42 peptides were synthesized using 9-fluorenylmethoxycarbonyl (Fmoc) chemistry and purified by reverse phase high performance liquid chromatography (RP-HPLC), essentially as described (52). The identity and purity (usually >97%) of the peptides were confirmed by amino acid analysis, mass spectrometry, and reverse phase high performance liquid chromatography (RP-HPLC). Peptides were solvated with 1,1,1,3,3,3-hexafluoroisopropanol (HFIP) as described (66). Briefly, 100–300 μg of peptide lyophilizate were dissolved in 200 μl of 100% HFIP in a 1.5 ml V-bottom polypropylene microcentrifuge tube (Eppendorf, Fisher Scientific). Each tube was covered with a Kimwipe® (Fisher Scientific) tissue and placed in a chemical fume hood overnight to allow evaporation of the HFIP. Tubes then were placed in a SpeedVac rotary evaporator (Savant SPD121P, ThermoScientific) for 2 h to ensure complete HFIP removal. The resulting peptide films were dissolved in 10% (v/v) 60 mM NaOH, 45% (v/v) Milli Q water, followed by 45% (v/v) 20 mM sodium phosphate, pH 7.5, containing 0.002% (w/v) sodium azide (“phosphate buffer”). The solutions then were thoroughly mixed and sonicated for 1 min in a Branson 1200 bath sonicator (Branson Ultrasonics Corp., Danbury, CT). The resulting peptide solution either was used immediately or stored at −20°C for future use. When used immediately, the sample was placed on ice and the peptide concentrations were estimated by UV absorbance (ɛ₂₈₀=1280 cm⁻¹M⁻¹ for human and Od; and ɛ₂₅₇=200 cm⁻¹M⁻¹ for Mm). The final concentration of Aβ42, as necessary for each experiment, was achieved by dilution with phosphate buffer. Equimolar amounts of Aβ42 were used in each experiment and a minimum of three independent experiments were performed.

Ion mobility-mass spectrometry (IM-MS)

Samples were prepared at a final peptide concentration of 10 μM in 10 mM ammonium acetate, pH 7.4. The samples were analyzed on a home-built ion mobility spectrometry-mass spectrometer (67). Briefly, for ion-mobility measurements, ions are generated continuously by a nano-ESI source, focused and stored in the ion funnel. A pulse of ions is injected into a temperature-controlled drift cell filled with helium gas (3–5 torr), in which the ions move under the influence of a weak electric field. The injection energy can be varied from ~20–150 eV, but it is usually kept as low as possible to minimize thermal heating of the ions during the injection process. After exiting the drift cell, the ions are further analyzed with a quadrupole mass filter, detected by the conversion dynode and channel electron multiplier, and recorded as a function of their arrival time distribution (ATD). The ATD can be related to the time the ions spend in the drift cell, which is directly related to the ion mobility and collision cross-section (σ) of the ion (68). The measured σ provides information about the three-dimensional configurations of the ion. The width of the ATD can be compared to the width calculated for a single ion structure (68), which gives information on the structural distribution favored in the ATD. If a feature in the experimental ATD is broader than the calculated one, then the feature may represent a family of structures, rather than a single structure.

Oligomerization of Aβ42 by PICUP

Aβ oligomerization was monitored using Photo-Induced Cross-linking of Unmodified Proteins (PICUP), essentially as described (35, 69). Briefly, peptide samples were prepared, as described in Peptide synthesis and preparation, to produce final peptide concentrations 25–35 μM. Eighteen μl of sample were periodically subjected to the PICUP reaction (69). Briefly, 1μl of 2 mM Tris (2,2′-bipyridyl) dichlororuthenium (II) hexahydrate (Ru(bpy)) was added to a 0.2 ml thin-walled PCR tube (Eppendorf AG, Hamburg, Germany) containing the peptide, followed by addition of 1μl of 40 mM ammonium persulfate (APS) in phosphate buffer. The tube then was irradiated for 1 s with incandescent light using a high intensity illuminator (150 W, Dolan-Jenner Industries Inc, Model 170-D). The reaction was quenched immediately with 1 μl of 1M DTT in water and then the sample was vortexed and placed on ice. To determine the oligomer size distribution, an equal volume of 2⨉ Tris-Tricine SDS sample buffer (Invitrogen, Carlsbad, CA) was added to each sample. The samples then were boiled in a 100°C water bath for 10 min and electrophoresed on a 10–20% T, 1 mm thick, Tris-Tricine SDS gel (Invitrogen, Carlsbad, CA). Silver staining was done using X-press™ silver staining.

Quasielastic light scattering spectroscopy (QLS)

Mm, Od, and Aβ42 were dissolved at a concentration of 0.5 mg/ml in phosphate buffer, briefly vortexed, sonicated for 20 s, and filtered using a 20 nm Anotop filter (Whatman, Maidstone, England). Samples were monitored at ≈22°C for 7–10 days. Measurements were done using a custom optical setup comprising a 40 mW He-Ne laser (λ=633nm) (Coherent, Santa Clara, CA) and PD2000DLS detector/correlator unit (Precision Detectors, Bellingham, MA). Light scattering was measured at a 90° angle. The intensity correlation function and the diffusion coefficient (D) frequency distribution were determined using Precision Deconvolve software (Precision Detectors, Bellingham, MA). Hydrodynamic radius R_H was calculated from D according to the Stokes-Einstein equation R_H = k_BT/6πηD; where k_B is Boltzmann’s constant, T is Kelvin, and η is the solvent viscosity (70).

To determine the instrument-independent rate of change in relative intensity, we calculate the normalized quantity (dI/dt)/I₀ (units of h⁻¹). The rate of change in scattering intensity, dI/dt, was determined by fitting to a linear function spanning the first hour of incubation. I₀ was determined from the derived line I_t=(dI/dt)t + I₀ by substitution of t=0. Both dI/dt and I₀ depend on the aperture in which light is collected, which in turn is a function of the instrument used. We note that these quantities are precise, but likely not absolutely accurate, because the intensity versus time dependency over this time interval is significantly non-linear.

Thioflavin T (ThT) binding

HFIP-treated Aβ42 peptide films were prepared in phosphate buffer at a nominal concentration of 1 mg/ml on ice and added to 96-well optical-bottom microtiter plates (Thermo Fisher Scientific, Rochester NY), followed by 1.6 µl of ThT at a concentration of 5 mM in phosphate buffer. Phosphate buffer was added to the wells to produce a final volume of 200 µl and final Aβ and ThT concentrations of 20 and 40 µM, respectively. The solutions were mixed gently by pipetting solution into and out of the pipette tip. The plates were sealed using an adhesive plate sealer and incubated at 37°C with gentle shaking (160 rpm) (Innova 4080 incubator shaker, New Brunswick Scientific, NJ). Fluorescence was determined using λ_ex=450 nm and λ_em=482 nm. Readings were taken immediately (0 h), every hour for 6 h, and subsequently at 24 h intervals. Buffer alone, instead of peptide, constituted the ThT blank. A minimum of 5 replicates of each sample was measured. The mean of the blank readings were subtracted from the mean of the sample readings at each time point and the corrected values, along with mean and SD, were plotted using KaleidaGraph (v 4.1, Synergy Software, Reading, PA). Statistical analyses (t-test and Mann-Whitney Rank test) were performed using SigmaStat (Jandel Scientific, San Jose, CA).

Bis-ANS fluorescence

Bis-ANS (1,1′-Bis(4-anilino, 5-naphthalene sulfonic acid)) fluorescence was monitored periodically in 20 µM Aβ42 samples incubating at 37°C with shaking (160 rpm). Fifty μl of the sample (5 μM final concentration in the cuvette) was removed at regular intervals and added to 150 μl of 133.3 μM Bis-ANS solution in a 1.5 ml semi-micro disposable fluorescence cuvette (Brand, Germany). The cuvette was placed in the dark at room temperature (RT; 22.5°C) for 5 min without shaking and then fluorescence (λ_ex=400 nm and λ_em=495 nm) was determined using a Hitachi model F-4500 Fluorescence Spectrophotometer (Hitachi America, NJ). The excitation and emission slit widths were both 5 nm. Each sample was read in triplicate, averaged, and corrected using a buffer blank incubated for the same amount of time. The samples were read at 2 h intervals for the first 12 h and subsequently at 24 h intervals. The data were plotted in KaleidaGraph (v 4.1, Synergy Software, Reading, PA). Statistical analyses on the data (t-test and Mann-Whitney Rank test) were performed using SigmaStat (Jandel Scientific, San Jose, CA).

Circular Dichroism Spectroscopy

Mm, Od, and Aβ42 peptides were prepared from HFIP dried films at final concentrations of 20 µM. For the mixture Aβ42 samples, the final concentration of each peptide in the mixture was 10 µM. The peptides were incubated in 1 mm path length cuvettes, without shaking, at 37ºC. Spectra were acquired using a JASCO J-810 spectropolarimeter (Tokyo, Japan) every 15 min for the first 3 h and subsequently every 30–60 min. CD measurement parameters were: wavelength range, 190–260 nm; data pitch, 0.2 nm; continuous scan mode; scan speed, 100 nm/min; response time, 1 s; band width, 2 nm; scan number per sample, 7. The raw spectra were smoothed using the means movement smoothing parameters within the data acquisition software (Spectra Manager 2). The data were subsequently plotted in KaleidaGraph (v 4.1, Synergy Software, Reading, PA).

Electron microscopy (EM)

Formvar 400 mesh grids were glow discharged on a Med010 mini-deposition system EM glow discharge attachment model BU007284-T (Balzers Union Ltd, Hudson, NH) containing a cylindrical discharge compartment and an adjacent discharge control and timer unit. Aβ42 was incubated at 37°C with continuous shaking (160 rpm). Eight μl aliquots were mixed thoroughly and then they were applied to the grid at days 0 and 7. The grid was covered and incubated for 20 min at RT. Liquid was wicked off carefully using a filter paper wick by gently touching the tip of the filter paper to the edge of the grid. Five μl of 2.5% (v/v) glutaraldehyde in water were added to the grid, which then was incubated for 3 min in the dark. The glutaraldehyde solution was wicked off and replaced with 5 μl of 1% (w/v) uranyl acetate in water, which then was incubated for 3 minutes in the dark. The grids then were wicked off and air-dried. A JEOL 1200 EX (JEOL Ltd., Tokyo, Japan; 40–120 kV) transmission electron microscope was used to visualize the samples (71).

Neuronal Cell Cultures

Primary cortical or hippocampal neurons were prepared as described previously (72). Briefly, pregnant E18 rats were euthanized with CO₂ and the pups were removed immediately. Brains were dissected in chilled Leibovitz’s L-15 medium, pH 7.5, (ATCC, Manassas, VA) in the presence of 1 μg/ml penicillin/streptomycin (Invitrogen, Carlsbad, CA). The tissue was incubated with 0.25% (w/v) trypsin in phosphate-buffered saline, pH 7.5 (PBS), containing 0.02% (w/v) EDTA for 30 min and then mechanically dissociated in a small volume of Leibovitz’s L-15 medium using a fire-polished Pasteur pipette. The neurons were suspended in Dulbecco’s modified Eagle’s medium, pH 7.5, (DMEM; ATCC) containing 10% (v/v) heat-inactivated fetal bovine serum (FBS) (ATCC) and penicillin/streptomycin (1 μg/ml) and plated on poly-D-lysine (0.1 mg/ml, Sigma)-coated 96-well COSTAR plates (Corning, Lowell, MA) at a density of 3 × 10⁵ cells/ml. Twenty-four hours after the cells had been plated, the medium was replaced with fresh medium supplemented with 5 µM cytosine β-d-arabinofuranoside (Sigma Chemical Co, St Louis, MO) to inhibit the proliferation of glial cells. The cultures were maintained for 6 d before being treated with Aβ42. PC-12 cells were cultured and differentiated with 50 ng/ml nerve growth factor (NGF) 24 h prior to treatment with peptides, as described previously (38). SHSY-5Y cells were grown in DMEM/F12K (1:1) media, pH 7.5, containing 10% (v/v) heat-inactivated FBS and penicillin/streptomycin (1μg/ml), 2 mM glutamine, and 1.5 g/L sodium bicarbonate. The cultures were maintained at 37°C in a humidified atmosphere of 5% (v/v) CO₂.

MTT (3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) Assay

Cell metabolism was evaluated using the MTT assay, as described previously (73). Briefly, rat primary cortical neurons or NGF-differentiated PC12 cells were treated with 10 µM freshly prepared Aβ42 for 24 h at 37°C. Following treatment, 15 μl of MTT, prepared by dissolution in 10% (v/v) Triton-X 100 in 2-propanol containing 0.1N HCl, according to the manufacturer’s instructions (Promega, Madison, WI), was added to each well and the samples were incubated for 4 h at 37°C. “Stop solution” (Promega, Madison, WI) then was added and the samples were incubated overnight at RT. OD_570–630 was measured using a Synergy plate reader (Bio-TEK Instruments, Winooski, VT). A minimum of three independent experiments (six wells per data point) were performed. Data were normalized to the medium control group and expressed as the mean ± the standard error of the mean (SEM).

LDH (Lactate Dehydrogenase) Assay

Rat primary cortical neurons and NGF-differentiated PC12 cells were incubated with 10 µM Aβ42 peptide for 48 h at 37°C in Eagle’s minimal essential medium (MEM, Earle’s salts, supplied glutamine-free) supplemented with 5% (v/v) heat-inactivated horse serum, 5% (v/v) fetal bovine serum, 2 mM glutamine, and 25 mM glucose. Cell death was assayed by measuring released LDH activity, as described previously (71). Data from a minimum of three independent experiments (six wells per data point) were normalized to media control and expressed as mean ± SEM.

Apoptosis Assay

Apoptosis was estimated using Annexin V-FITC fluorescence. Briefly, SH-SY5Y neuroblastoma cells (ATCC® CRL-2266) were grown on cover slips coated with poly D-lysine (0.1 mg/ml) in DMEM/F12K (1:1) medium containing 10% (v/v) heat-inactivated fetal bovine serum (FBS), penicillin/streptomycin (1 μg/ml), 2 mM glutamine, and 1.5 g/L sodium bicarbonate. The cells were maintained at 37°C in a humidified atmosphere of 5% (v/v) CO₂ in air until a cell density of 1–5 × 10⁵ cells per well was obtained, at which point the cells were exposed to peptide (10 μM final concentration) for 24 h. The peptides were prepared by dissolution of HFIP films in cell culture medium. Staurosporine (1 µM final concentration) was added to cells as a positive control for apoptosis. The negative control was cells treated with medium containing no Aβ42. After the incubation period, the coverslip was washed in PBS and fixed with 4% (v/v) paraformaldehyde in PBS. The cells were stained with an Annexin V-FITC apoptosis detection kit (Bio Vision Incorporate, Milpitas, CA) according to the manufacturer’s instructions. Images were recorded using a confocal laser scanning microscope (Zeiss, Model: LSM 700; λ_ex=488 nm and λ_em=535 nm). The numbers of apoptotic cells were counted in a 100 μm² area using Image-J software (Image-J, NIH). A minimum of 30 microscopic fields was used for morphometric analysis from two independent experiments.

Figure 8. Time-dependence of [Θ]₂₁₅.

Molar ellipticity at 215 nm is plotted versus time as a measure of time-dependent changes in β-sheet secondary structure.

Table 3.

Dimensions of assemblies observed by EM. Peptides were incubated at 37°C with shaking. Aliquots of assemblies were removed at day 0 and day 7. Following incubation, different classes of assemblies were observed, including globules, short fibrils, and long fibrils. If present, the numbers represent the size range, in units of nm, of each assembly type. Assembly lengths are reported in nm within parentheses.

Peptide	Day 0			Day 7
	globules	short fibrils	long fibrils	globules	short fibrils	long fibrils
Aβ42	11–23	–	–	–	7–12 (56–303)	9–21
Mm	11–21 (16–75)	–	–	–	–	4–19
Od	11–54	–	7–19	7–17	7–17 (37–182)	7–17

ABBREVIATIONS

Aβ42

Amyloid β−protein (1–42)

AD

Alzheimer’s disease

IM-MS

Ion mobility-mass spectrometry

Mm

Mus musculus Aβ42

PICUP

Photo-Induced Cross-linking of Unmodified Proteins

Ru (Bpy)

Tris (2,2′-bipyridyl) dichloro ruthenium (II) hexahydrate

Od

Octodon degus Aβ42