Cell Surface Binding and Internalization of Aβ Modulated by Degree of Aggregation

The amyloid peptides, Aβ40 and Aβ42, are generated through endoproteolytic cleavage of the amyloid precursor protein. Here we have developed a model to investigate the interaction of living cells with various forms of aggregated Aβ40/42. After incubation at endosomal pH 6, we observed a variety of Aβ conformations after 3 (Aβ3), 24 (Aβ24), and 90 hours (Aβ90). Both Aβ4224 and Aβ4024 were observed to rapidly bind and internalize into differentiated PC12 cells, leading to accumulation in the lysosome. In contrast, Aβ40/4290 were both found to only weakly associate with cells, but were observed as the most aggregated using dynamic light scattering and thioflavin-T. Internalization of Aβ40/4224 was inhibited with treatment of monodansylcadaverine, an endocytosis inhibitor. These studies indicate that the ability of Aβ40/42 to bind and internalize into living cells increases with degree of aggregation until it reaches a maximum beyond which its ability to interact with cells diminishes drastically.


Introduction
Alzheimer's disease (AD) is a progressive neurological disorder resulting from the deposition of Alzheimer β-Amyloid peptide (Aβ) as senile plaques, the appearance of neurofibrillary tangles, and selective neuronal loss. The most abundant forms of Aβ are 40 and 42 amino acid residues long and referred to as Aβ40 and Aβ42, respectively [1].
Here we present a method for generating a mixed population of Aβ conformations using model endocytic conditions. Using this method, we demonstrate that when Aβ is exposed to endosomal conditions for an extended period of time, the ability of the peptide to bind and internalize into living rat adrenal pheochromocytoma (PC12) cells increases with time until it reaches a maximum beyond which its ability to interact with PC12 cells diminishes drastically.

Peptide Synthesis and Purification.
Aβ40 and Aβ42 were synthesized and purified as described previously [22]. Before cleavage from the resin, the fluorophore, Nα-(9-Fluorenylmethoxycarbonyl)-Nε-tetramethylrhodamine-(5-carbonyl)-L-lysine (Molecular Probes, Eugene, OR) (abbreviated TMR), was coupled to the N-terminus via a glycine linker. The crude peptides were purified by HPLC using a Superdex Tricorn 10/300 GL Peptide column (Amersham Biosciences, Piscataway, NJ) with 30 mM NH 4 OH running buffer. To maintain stock peptide solutions free from fibril seeds, solutions were stored at pH 10 and 4 • C immediately after chromatographic separation of monomeric peptides. Aβ preparations were never lyophilized, as this process may allow for seeds to form. These solution conditions have been previously shown to maintain the monomeric state [16,23,24]. Peptide purity and identity was confirmed using both MALDI mass spectrometry and amino acid analysis. Concentrations of stock peptide solutions were determined using amino acid analysis and confirmed by either tyrosine absorbance (275 nm, ε = 1390 cm −1 M −1 ) or TMR absorbance for labelled peptides (550 nm, ε = 92000 cm −1 M −1 ). At least three separate synthesized lots of Aβ were used in this study and each displayed identical cell association rates when compared for quality assurance. Oligomeric samples were prepared by diluting stock Aβ samples to 30 μM with 30 mM NH 4 OH and reducing to pH 6 with 0.2 M HCl and incubating for zero (Aβ 0 ), 3 (Aβ 3 ), 24 (Aβ 24 ), or 90 hours (Aβ 90 ) in the dark at 20 • C.

Dynamic Light Scattering (DLS).
Hydrodynamic radius (Rh) measurements were made at 20 • C with a DynaPro DLS instrument (Protein Solutions Inc., Piscataway, NJ). Peptide samples (30 μM) were reduced to pH 6 using 0.2 mM HCl, centrifuged at 12000 ×g for 3 minutes and then rapidly added to a 1 cm path length cuvette and left in the instrument. DLS data was collected at various time points over 90 hours. Particle translational diffusion coefficients were calculated from decay curves of autocorrelation of light scattering data and converted to hydrodynamic radius (Rh) with the Stokes-Einstein equation. Histograms of intensity versus Rh were calculated using Dynamics data analysis software (Protein Solutions Inc., Piscataway, NJ).

Filter Assay.
Aβ samples (30 μM) were either not filtered or spin-filtered for 30 minutes at 14,000 ×g using 10, 30, 100 kDa (Amicon ultra cellulose KMWO), or 0.1 μm (Amicon PVDF) spin filters. Absorbance at 550 nm was collected on a Molecular Devices SpectraMax M5 (Molecular Devices Corp., Sunnyvale, CA) and graphs were created normalizing the absorbance signal from each filtered sample to the corresponding unfiltered sample.

Thioflavin-T Assay.
Fluorescence measurements were obtained using 200 μL of 30 μM Aβ samples, within a 96well plate, after addition of 5-fold molar excess of thioflavin-T and incubation at room temperature for 30 minutes.
Emission at 485 nm was collected using 440 nm excitation on a Molecular Devices SpectraMax M5 (Molecular Devices Corp., Sunnyvale, CA).

Cell
Culture. PC12 cells were maintained in DMEM/F12 containing 10% fetal bovine serum (HyClone, Logan, UT) with 100 units/mL penicillin and 100 μg/mL of streptomycin. To induce differentiation of PC12 cells and for cell imaging, they were plated at 2.2 × 10 4 cells/cm 2 in Lab-tech chambered cover glass chambers and suspended in phenol red free DMEM/F12 containing N2 supplement and 10 ng/mL NGF. Cells were differentiated for 72 hours before media was replaced and peptide treatments (final concentration 1.5 μM) were performed. Cells were maintained at 37 • C in a humidified incubator with 5% carbon dioxide.
2.6. Analytical Ultracentrifugation. TMR-labelled Aβ samples (6 μM) were prepared in phenol red free DMEM/F12 media containing N2 supplement and 10 ng/mL NGF. Either fresh media or cultured supernatants, obtained from cell culture after 72 hours, were used. To avoid interference from cell culture components, molecular weights of TMR-labelled Aβ were obtained by selective monitoring of TMR absorbance at 550 nm. Sedimentation experiments were performed at 20 • C on a Beckman XLI analytical ultracentrifuge using an AN50-Ti rotor. Molecular weights were calculated using Beckman XLI data analysis software in which absorbance versus radial position data were fitted to the sedimentation equilibrium equation using nonlinear least-squares fitting.

Confocal Microscopy.
Three-dimensional stacks of fluorescence micrographs were taken at 20 • C with a confocal laser-scanning system consisting of an LSM 510 Zeiss META NLO confocal microscope with a C-APO 40X water immersion objective (numerical aperture 1.2) and HeNe laser with a 543 nm laser line. The displayed images were captured using Zeiss LSM Image version 4 and prepared using ImageJ version 1.37v and represent a single crosssection through the cells.

Inhibition of Endocytosis.
Differentiated PC12 cells were treated with 50 μM monodansylcadaverine (MDC) for 20 minutes at 37 • C with 5% carbon dioxide. After inhibitor treatment, Aβ 24 (final concentration 1.5 μM) was added to the media containing MDC and imaged after 4 hours at 37 • C. Aβ 24 (final concentration 1.5 μM) was also added to differentiated PC12 cells media for 4 hours at either 37 • C or 4 • C prior to imaging.
2.9. Intracellular Localization. Differentiated PC12 cells were first treated with Aβ 24 (final concentration 1.5 μM) for 6 hours at 37 • C followed by treatment with 40 nM Mito-Tracker Deep Red and 50 nM LysoTracker Green DND-26 for 20 minutes. After treatment, the media was exchanged with fresh phenol red free DMEM/F12 media containing N2 supplement and 10 ng/mL NGF with 1.5 μM DAPI for nuclear staining. Cells were imaged using Argon laser with 488 nm laser line for LysoTracker, HeNe laser with 633 nm laser line for MitoTracker, HeNe Laser with 543 nm laser for TMR-labelled Aβ 24 , and a tunable Chameleon laser at 730 nm for two-photon excitation of DAPI. 2D histograms and correlation coefficients were determined using Image J version 1.42q with colocalisation threshold plugin [25].
2.10. Toxicity. Differentiated PC12 cell media was replaced with media containing 0.6 to 20 μM Aβ 0 , Aβ 24 , Aβ 90 , or 10 μM Melitin as a positive control for 48 hours. Cell survival was quantified using the Sulforhodamine B assay [26], and absorbance was measured at 560 nm using Molecular Devices SpectraMax M5 (Molecular Devices Corp., Sunnyvale, CA). LC50 values were determined as the concentration of Aβ required to kill fifty percent of the cells from an absorbance versus Aβ concentration plot.

Aβ Aggregation Is Mediated through Cell Interaction.
To study the interaction between Aβ and live cells, we synthesized and fluorescently labelled Aβ40/42. The synthesized Aβ was maintained in solution from purification to storage and was never lyophilized, as these solution conditions are known to significantly reduce the formation of Aβ aggregation seeds [16,23,24]. We covalently attached tetramethylrhodamine (TMR) to the N-terminus of Aβ via a flexible glycine linker to generate TMR-Aβ. The N-terminus of Aβ is highly accessible even in the fibril state [21,27,28] and attaching a fluorescent label to this site has been shown to neither alter its amyloidogenic properties [16,29,30], nor its solubility behaviour [31,32].
We have previously shown that treating cultured cells with 1.5 μM monomeric TMR-Aβ42 leads to the formation of visible aggregates on the surface of PC12 cells within one hour of treatment [29]. We initiated our current study by investigating whether Aβ aggregation could occur in cell culture media alone. Using analytical ultracentrifugation, we measured the molecular weights of Aβ40 and Aβ42 present in phenol red free cell culture medium, both freshly prepared and conditioned media taken from differentiated PC12 cells 3 days postdifferentiation. Following the addition of 6 μM Aβ to each medium and subsequent 24-hour incubation at room temperature, both media preparations were centrifuged to equilibrium at room temperature in an analytical ultracentrifuge in order to determine the molecular weight of Aβ conformations. The molecular weights of Aβ40 and Aβ42 from both media preparations were measured to be approximately 4103 Da and 4425 Da, respectively. These values both correspond to the expected monomeric molecular weights of Aβ, falling within the 95% confidence intervals of 3650-4570 Da for Aβ40 and 3650-4770 Da for Aβ42. Thus, the aggregation of Aβ seen by confocal microscopy apparently occurs only after interaction with the differentiated PC12 cells and not with cultured supernatants. It should be noted that the concentration previously used to treat cells (1.5 μM) and the concentration used for ultracentrifugation (6 μM) are considerably lower than the reported 20 to 50 μM range required for in vitro aggregation [33][34][35].

Aβ Oligomerizes at Endosomal pH.
We have previously shown that Aβ40 and Aβ42 aggregate significantly at pH 6 [29]. Since Aβ40 and Aβ42 are generated through endoproteolytic cleavage [7][8][9], and the pH of the endosome and recycled cellular vesicles is equivalent to pH 6 [12], we characterized the Aβ conformations formed under endosome conditions. Aβ (30 μM) was reduced to pH 6, and hydrodynamic radius calculations were collected after zero (Aβ 0 ), 3 (Aβ 3 ), 24 (Aβ 24 ), and 90 (Aβ 90 ) hours using dynamic light scattering ( Figure 1). The average hydrodynamic radius of Aβ40 was found to increase from 2.0 nm at time zero (Aβ40 0 ) to 216 nm after 90 hours at pH 6 (Aβ40 90 ). A more striking increase was found with Aβ42, beginning with 1.7 nm at time zero (Aβ42 0 ) to 451 nm after 90 hours at pH 6 (Aβ42 90 ). For each of these samples, the development of increasingly higher ordered aggregates was observed over time and the samples that were treated for 90 hours contained particles over 1000 nm in radii.
To further investigate the relative levels of peptide aggregation at endocytic pH, we filtered the Aβ40 and Aβ42 samples through various molecular weight cutoff (MWCO) spin filters ( Figure 2). Approximately 67% of Aβ40 3 , 60% of Aβ40 24 , and 55% of Aβ40 90 were recovered through the 10 kDa MWCO spin filter and approximately 80% were recovered through the 100 kDa MWCO filter, except for Aβ40 90 with only 60% recovered. In contrast, only 55% of Aβ42 3 , Aβ42 24 , and Aβ42 90 were recovered through the 10 kDa filter and approximately 60% were recovered through the 100 nm filter. These results indicate that the majority of peptide conformations present under these conditions were able to pass through a 10 kDa molecular weight filter, but that just over 40% could not be recovered through the 100 nm filter for the Aβ40 90 , Aβ42 3 , Aβ42 24 , and Aβ42 90 samples.
We also used thioflavin-T to assess the time-dependence of the extent of Amyloid fibril formation at endocytic pH. Thioflavin-T is a dye known to shift its fluorescence from 430 nm to 490 nm upon binding specifically to the cross-β-structure of Amyloids but not to monomeric or small oligomeric complexes [36,37]. We observed enhanced thioflavin-T fluorescence at all time points ( Figure 3); however, Thioflavin-T bound most strongly to Aβ40 90 , Aβ42 24 , and Aβ42 90 . The high thioflavin-T binding to Aβ 90 samples suggests that these late stage Aβ conformations are the most aggregated.

Endocytic Aβ Undergoes Rapid Cellular Interaction.
Using confocal microscopy, cell surface association and internalization of peptides can be monitored. We have previously shown that monomeric Aβ42 associates with cells more rapidly than Aβ40, with significant staining observable after six hours of treatment [29]. To determine whether the aggregation state of Aβ affects cell association, we exposed each of the Aβ40 and Aβ42 samples to differentiated PC12 cells and monitored the kinetics of association by confocal microscopy. Upon treating cells with Aβ40 24 , the cell surface association was observed within one hour of treatment ( Figure 4). Moreover, Aβ40 24 was observed to significantly internalize into these cells after only 6 hours, whereas significant internalization of Aβ40 0 was only visualized around    24 hours (Figure 4). Similarly, Aβ42 24 internalized into differentiated PC12 cells after only one hour of treatment, whereas Aβ42 0 treatment only became observable at 24 hours ( Figure 4). Interestingly, when differentiated PC12 cells were treated with late-stage Aβ40 90 or Aβ42 90 , very few cells underwent internalization or even exhibited cell surface interaction with these aggregated peptide forms (Figure 4). To illustrate the contrast between these treatments, we collected the three-dimensional image slices through a cell from the base to the apex for the 6-hour treatment with Aβ42 0 , Aβ42 24 , and Aβ42 90 ( Figure 5). Aβ42 0 was only observed around the periphery of the cell, whereas some Aβ42 24 was located inside the cell. By contrast, Aβ42 90 treatment seemed to localize to extracellular regions and did not produce the punctate pattern as observed with the Aβ42 0 sample.
To quantify the frequency of cells that exhibited peptide internalization, we randomly selected five fields of view from at least three separate 6-hour treatment experiments and plotted the percentage of cells having internalized Aβ ( Figure 6). Approximately 25% of cells internalized Aβ40/42 3 , whereas more than 90% of cells internalized Aβ40/42 24 . In contrast, very few cells were found to internalize Aβ40/42 90 . Since late Aβ40/42 90 was found to have a large number of aggregates with hydrodynamic radius over 1000 nm (Figure 1) and were found to significantly bind thioflavin-T (Figure 3), then these aggregates may favour self-association over cell association.

Internalization of Aβ Is Mediated through Cellular Import
Mechanisms. To determine if the internalization of endocytic Aβ is mediated through cellular processes, such as receptormediated endocytosis or through direct peptide-mediated processes like membrane pore formation, we monitored the effects of temperature and monodansylcadaverine (MDC) on internalization. MDC is a known inhibitor of receptormediated clathrin-dependent endocytosis [38,39]. At physiological temperature (37 • C), after a 4-hour treatment of cells with Aβ40 24 and Aβ42 24 , internalization was observed (Figure 7). At 4 • C membrane vesicle formation is inhibited, preventing endocytosis of extracellular and cell surface components [40]. When Aβ40 24 or Aβ42 24 association with differentiated PC12 cells was monitored at 4 • C, none of the cells were found to have internalized these peptides  ( Figure 7). Similarly, when differentiated PC12 cells were treated with MDC, neither Aβ40 24 nor Aβ42 24 were observed within the cells. These observations indicated that Aβ 24 was internalized through a cell-directed import mechanism, rather than an independent penetration route through the cell membrane. nucleus staining (Figure 8). The staining pattern of each cellular organelle marker in the same image plane as the Aβ40 24 and Aβ42 24 treatment was determined ( Figure 8) and quantified (Figure 9). Only the signal from the lysosome marker costained with both internalized Aβ40 24 and Aβ42 24 . The red signal from the peptides costained well with green signal from the lysosomes, resulting in the yellow signal in the merged image. When quantified, 71% of the lysosome signal intensity colocalised with the Aβ40 channel and 80% with the Aβ42 channel ( Figure 9). Whereas only 9% of the mitrochondria signal intensity colocalised with Aβ40 channel and less than 1% with the Aβ42 channel ( Figure 9). These findings suggest that the internalized vesicles containing Aβ40/42 24 are directed to the lysosome. The intensity of both Aβ40 24 and Aβ42 24 fluorescence signals was found to increase in the lysosomes over time, which may reflect the accumulation of Aβ40/42 24 in the lysosome.

Internalized Aβ Is
To determine the relative toxicity of these Aβ conformations, a cytotoxicity assay was performed where differentiated PC12 cells were treated with various peptide concentrations for 48 hours (Figure 10). We calculated the lethal concentration at which 50% of cells were killed (LC50), for each of the Aβ conformations. We found that Aβ40 24 was moderately more toxic than Aβ40 0 , with a lower LC50 value ( Figure 10). Surprisingly, Aβ40 90 was also found to have a lower LC50 value compared to Aβ40 0 . Aβ42 24 was also found to be moderately more toxic than Aβ42 0 , whereas Aβ42 90 had a much higher LC50 value compared to Aβ42 0 ( Figure 10).

Discussion
We have developed a method to produce a collection of Aβ conformations that differ in their extent of aggregation and investigated the interaction between these states of Aβ40 and Aβ42 and differentiated PC12 cells. Others have described the methods to isolate individual soluble oligomeric forms of Aβ, using various chemical reagents and protocols [41][42][43][44]. In addition, purified oligomeric Aβ molecules from cell culture [45,46] or transgenic mice [47] have also been monitored. These purified sources of oligomeric Aβ offer great potential in understanding the progression of Aβ aggregation; however, they cannot be directly visualized over the time course of their effects on cells during maturation from earlier to later conformations. Since it is well established that Aβ monomers can oligomerize under the physiological pH of the endosome [13][14][15], and Aβ is formed through endoproteolytic cleavage, we have utilized this condition to determine which conformations are present and how these mixed conformations interact with differentiated PC12 cells. It is known that the extent of oligomerization/fibrillation is very dependent on experimental conditions. Necula et al. [48] have indicated that Aβ can be induced to form fibrils via dilution from 100 mM NaOH to neutral pH in the presence of 10 mM HEPES/100 mM NaCl buffer, while dilution into phosphate buffered saline results in oligomer formation. In our experiments, the peptide was diluted from 30 mM ammonium hydroxide (pH 10) to pH 6 with final condition of 1 mM ammonium chloride as the only additional chemical. We do note that the normal human physiological concentration of ammonium chloride in blood and cerebrospinal fluid is approximately 20 to 50 μM [49,50], and that hyperammonemia has been linking to Alzheimer type II astrocytosis [51,52].
When extracellular monomeric Aβ associates to the surface of cells, we speculate there are three possible outcomes: (1) it can act as a stable template to allow further Aβ aggregation; (2) the peptide can penetrate through the cell membrane depositing in the cytoplasm; or (3) the peptide may be internalized into the cell within endocytic vesicles, which would result in a reduction in the surrounding pH. In the third case, the endocytic vesicles containing Aβ can Mitochondria Mitochondria Mitochondria Mitochondria theoretically be recycled back to the cell surface or directed to the lysosome where a further reduction in pH would occur. We have previously visualized oligomeric Aβ on the surface of neuronal cell lines [29], and here sought to determine first whether this aggregation occurred prior to cell surface deposition in the cell culture mediaor are aggregation and cell surface binding concomitant and linked processes. Using ultracentrifugation analysis of Aβ preparations either in freshly prepared cell culture media or in conditioned media removed from cultured cells after 3 days, we did not observe any conformation larger than the monomer.
These results indicate that components in our cell media or secreted factors from cultured cells are not responsible for the observed Aβ aggregation present on the surface of neuronal cells. After adding monomeric Aβ to cells, we noted a maturation time for the visual appearance of Aβ on International Journal of Alzheimer's Disease the cell surface and in the cell interior ( Figure 4; [29]). We postulate that the peptides on the cell surface might undergo a series of pH reductions through endocytic recycling, before becoming visible punctae on and inside the cell. Since pH 6 is approximately endocytic pH, we characterized the kinetic effects of this condition on peptide conformations. Exposing Aβ40 for up to 24 hours at pH 6, we observed an increase in the average hydrodynamic radius from 2.0 nm (Aβ40 0 ) to 183 nm (Aβ40 24 ) (Figure 1). In addition, approximately 60% of these peptides were recovered through a 10 kDa MWCO filter and 80% were recovered through a 100 kDa MWCO filter (Figure 2), and only minor thioflavin-T binding was found ( Figure 3). Following exposure of up to 24 hours at pH 6, Aβ42 had an increase in the average hydrodynamic radius from 1.7 nm (Aβ42 0 ) to 246 nm (Aβ42 24 ) (Figure 1), approximately 55% were recovered through 10 kDa MWCO filter and only 60% were recovered through 100 nm filter ( Figure 2) and more significant thioflavin-T binding was measured ( Figure 3). Both Aβ40 and Aβ42 treated for 90 hours at pH 6 were observed to have large average hydrodynamic radius particles, with 40% of these samples being withheld by a 100 nm filter and both displayed significant thioflavin-T staining. Also, these early-and latestage Aβ samples were observed to have an ensemble of Aβ conformations (Figure 1). The interactions of these early-and late-stage Aβ conformations were monitored with differentiated PC12 cells. Whereas late stage Aβ40 90 and Aβ42 90 exhibited very few interactions with cells (Figures 4, 5) and internalized at a very low frequency (Figure 6), early stage Aβ40 24 and Aβ42 24 interacted with cells rapidly (Figure 4) and internalized into the majority of the cells present ( Figure 6). Interestingly, Aβ40 24 showed a more rapid cellular association and internalization compared to Aβ40 0 (Figure 4). Similarly, Aβ42 24 was found to internalize more rapidly than Aβ42 0 , indicating that a specific conformation may play a role in cell binding and internalization. Furthermore, our model of endocytic Aβ toxicity ( Figure 10) is in agreement with previous findings utilizing chemically-produced oligomers of Aβ which showed that Aβ fibrils were less toxic than soluble oligomers [41,53]. Our results with Aβ 90 are also in agreement with studies that have indicated that regions with large Amyloid plaques do not correlate directly with regions of significant neuronal loss [54,55].
Many studies have reported that certain oligopeptides are freely imported into cells. For example, the peptide sequences corresponding to Tat (48)(49)(50)(51)(52)(53)(54)(55)(56)(57)(58)(59)(60), penetratin, and oligoarginine are known to internalize into live cells. Cellular import of Tat (48)(49)(50)(51)(52)(53)(54)(55)(56)(57)(58)(59)(60) and penetratin was shown to be temperature dependent, indicating the possible role of endocytosis for internalization [56,57]. In contrast, oligoarginine containing C-terminal tryptophan was shown to follow a nonendocytic pathway, independent of energy requirements and temperature [57]. Another type of internalization process has been studied in the bacterium Clostridium septicum, whose alpha toxin contains functional domains responsible for oligomerization and cellular pore formation. Using specific domains that bind cell surface receptors, alpha toxin monomers oligomerize to form pores in human cells and thus impose direct entry [58]. By exposing Aβ to endosomal conditions, we were able to observe internalization of Aβ40/42 24 , which was inhibited at 4 • C or with the endocytosis inhibitor MDC (Figure 7). Throughout the course of our confocal experiments, we did not observe any disruption of the cell membrane, which would have been expected if membrane channels or pores were being formed by Aβ. Thus, the structural conformation of Aβ that was observed to internalize into cells seems to follow a cellmediated import mechanism.
Our results with the various Aβ conformations correlate directly with previous reports that indicate the deposition of Aβ in the lysosome [15,59,60] and with the previous study that internalized Aβ can persist undegraded for days [61]. From our findings taken together with previous studies, we present the following model for the interaction of Aβ with neuronal cells ( Figure 11). As monomeric Aβ binds the cell surface, this event could lead to a conformational change allowing for the further catalytic aggregation of Aβ. Recently, it has been shown that oligomeric Aβ binds to GM1 ganglioside and alters physical properties of the plasma membrane which stimulates the amyloidogenic processing pathway of APP [62,63]. Aβ that is bound to the cell surface may become internalized and recycled, allowing for acidification and further aggregation, which in turn may stimulate additional Aβ generation [62]. Once the surface Aβ reach some specific structural form, they may be internalized and directed to the lysosome for potential degradation. However, the reduced pH of the lysosome may instead cause further aggregation and persistence of Aβ, resulting in lysosomal overload and cell death. The recently reported high turnover rate of monomeric Aβ [64,65] and the formation of large aggregate pools of Aβ may indicate that the human body uses these mechanisms to sequester and remove the toxic Aβ oligomers from neuronal cells.
With age, the turnover rate for monomeric Aβ may slow down, which could then result in the initiation of our proposed mechanism and may contribute to the late onset of Alzheimer's disease. It is also possible that endocytosis of Aβ and lysosomal targeting are mechanisms by which the cell clears Aβ aggregates from the cell surface. Also, certain large aggregates of Aβ may just be too large for the cell to internalize, and they may represent the precursor for extracellular Amyloid formation.

Conclusion
Our study shows that using a simple method to generate various Aβ conformations, the rates of cellular interaction and targeting can be followed with live cell cultures. Using this model, we found that early endocytic conformations, rather than highly aggregated late forms, serve as the major contributors to rapid cell internalization. The mechanism of internalization likely involves a cell surface receptormediated process instead of peptide-mediated direct entry, resulting in accumulation in the lysosome. This method allows for conformation-specific therapeutics and conditions to be screened with live cells, circumventing the need to purify specific Aβ conformations.