Real-Time Imaging and Quantification of Amyloid-β Peptide Aggregates by Novel Quantum-Dot Nanoprobes

Background Protein aggregation plays a major role in the pathogenesis of neurodegenerative disorders, such as Alzheimer's disease. However, direct real-time imaging of protein aggregation, including oligomerization and fibrillization, has never been achieved. Here we demonstrate the preparation of fluorescent semiconductor nanocrystal (quantum dot; QD)-labeled amyloid-β peptide (QDAβ) and its advanced applications. Methodology/Principal Findings The QDAβ construct retained Aβ oligomer-forming ability, and the sizes of these oligomers could be estimated from the relative fluorescence intensities of the imaged spots. Both QDAβ coaggregation with intact Aβ42 and insertion into fibrils were detected by fluorescence microscopy. The coaggregation process was observed by real-time 3D imaging using slit-scanning confocal microscopy, which showed a typical sigmoid curve with 1.5 h in the lag-time and 12 h until saturation. Inhibition of coaggregation using an anti-Aβ antibody can be observed as 3D images on a microscopic scale. Microglia ingested monomeric QDAβ more significantly than oligomeric QDAβ, and the ingested QDAβ was mainly accumulated in the lysosome. Conclusions/Significance These data demonstrate that QDAβ is a novel nanoprobe for studying Aβ oligomerization and fibrillization in multiple modalities and may be applicable for high-throughput drug screening systems.


Introduction
Neurodegenerative disorders such as Alzheimer's disease (AD), Parkinson's disease, Huntington's disease, and prion diseases are characterized by misfolded protein aggregates, termed amyloids, which are usually high in b-sheet content [1]. However, the exact mechanism of amyloid aggregation and its links to multiple disease pathogeneses are not fully understood. Amyloid-b peptide (Ab) is the major component of senile plaques and is a hallmark of AD [2]. An early hypothesis stated that the accumulation of fibrillar Ab deposits in senile plaques was neurotoxic [3]. In contrast, recent studies have identified the smaller soluble Ab oligomer as potentially more neurotoxic than amyloid fibrils [4,5,6]. Meanwhile, Ab peptide has been observed in various cellular localities, including lysosomes, aggresomes, mitochondria, dendritic spines, and within neurons, microglia, astrocytes and the extra-cellular space [7,8,9,10,11], but the exact cellular origin of Ab aggregation is not known. To understand the mechanism of Ab misfolding and locate the origin of Ab assemblage, we have developed a real-time imaging tool for monitoring Ab aggregation.
Fluorescent semiconductor nanocrystals (quantum dots; QD) have evolved over the past decade as highly useful fluorescence probes in biological staining and diagnostics [12,13]. QD properties include long-term photostability, chemical and physical stability, nano-scale size, and multicolor fluorescence emission with single excitation [14]. These features are extremely useful for long-term, single-molecule imaging in vitro and in vivo [15,16]. In fact, a single QD can be observed and tracked using basic widefield fluorescence microscopy [17], confocal microscopy [12], total internal reflection microscopy [18], and two-photon fluorescent microscopy [19]. For these reasons, QD could be an excellent tool for real-time monitoring of Ab aggregation and localization. Nevertheless, there have been no reports of successful preparation and characterization of QD-crosslinked Ab peptide, possibly due to the difficulty of covalently coupling the QD to the peptide without also reducing the ability of Ab to aggregate. Recently, Ji et al. [20] imaged Ab42 and Ab40 fibrils linked with QD, although the labeling was performed by non-specific ionic interaction between the fibrils and the QD. Therefore, the method is not applicable for tissue culture or in vivo studies. While fluoresceinlabeled Ab peptides have also been used in amyloid aggregation studies [21,22], this application is limited to short-term live imaging studies (less than 1 second) and is not appropriate for small oligomer imaging as fluorescein is not suitable for single molecule imaging nor live imaging due to poor signal levels and quenching [23]. In addition, standard amyloid plaque staining by thioflavin or Congo red is not suitable due to poor binding between the fluorescent dyes and b-sheet structures of Ab oligomers. Although potential cytotoxicity is a concern for longterm QD applications in cells [24], masking the core surface cadmium atom with a polyethylene glycol (PEG) coating greatly reduced the cytotoxicity [25]. Here, we have successfully generated a PEG-QD-crosslinked Ab peptide, which has enabled us to quantitatively examine, for the first time, Ab fibril and oligomer formations in vitro and in an intact cell system.

Generation of QDAb probe
Our first step was to examine whether Ab42 or Ab40 is more suitable as a QD probe. Both can be major components of amyloid plaques [2]. Aggregation of Ab42 has been shown to be more rapid than Ab40 [26]. Indeed, we confirmed that Ab42, without SDS, begins to aggregate within minutes of preparation-already forming oligomers or protofibrils during the one hour labeling process (Fig.  S1a). It formed trimeric and tetrameric species within 0.3 h from the start of incubation. On the other hand, Ab40, without SDS, did not form oligomers after 5 days ( Fig. S1c and S1d). Consequently, to provide a reasonable timeframe over which to study aggregation, we employed Ab40 for QD-labeling in this study.
Since one QD-PEG-NH 2 has ,80 to 100 amine groups on the surface, CL saturates almost all PEG amino groups under these conditions. Next, 100, 20, and 0 mM Cys-Ab40 (CAb) was added to QD-PEG-CL, yielding the binding ratios (Ab/QD) of 6, 1, and 0, for QDAb(6), QDAb(1), and QDAb(0) (Ab unconjugated control) conjugates, respectively (Fig. 1b). The initial concentration of Cys-Ab40 could control the Ab/QD ratio. A study to determine the rate of nonspecific binding between Cys-Ab40 and noncrosslinked QD-PEG-NH 2 showed an Ab/QD binding ratio of 0.4, suggesting that approximately 7% of Cys-Ab40 was bound to the QD surface via non-specific binding (Fig. 1b). The yield of QD particles was approximately 50% using this preparative method.

Oligomer formation of QDAb
Many recent studies have implicated soluble Ab oligomers as a potential toxic species in AD pathology [4,5,6]. Since formation of the toxic, b-sheet rich, Ab oligomer can be enhanced by certain concentrations of SDS [5,6], we examined whether QDAb forms oligomers in the presence or absence of SDS. Prior to conducting this experiment, we needed to confirm that 1 mM SDS enhances oligomerization of unlabeled Ab40 and Ab42 [5] by measuring the kinetics of Ab40 and Ab42 oligomerization with and without SDS (Fig. S1). SDS promoted oligomer and fibril formations of both Ab42 and Ab40 at 1 mM concentration with especially enhanced Ab40 dimer formation (Fig. S1d). We then applied these conditions for monitoring oligomerization of QDAb.
Oligomerization of QDAb was imaged according to the method in Figure S2a. Incubation of QDAb(6) in water for 3 weeks on ice does not alter its fluorescent image ( Fig. 2a top right), suggesting that QDAb(6) can be stored in water on ice without aggregation. In contrast, brighter and larger spots were observed by incubation of QDAb(6) at 37uC with and without 1 mM SDS (Fig. 2a bottom micrographs), suggesting its oligomerization.
To examine the formation of oligomers by QDAb, we measured the relative fluorescence (RF) and the number of fluorescence spots using the ''analyze particle'' tool of ImageJ (NIH) (Fig. S2b). In this analysis, the average RF of unconjugated QD-PEG-NH 2 was expressed as 1 RF unit (RF1). Since fluorescence intensity is generally proportional to the number of fluorescence molecules, it is likely that the summed RF values indicate the total number of QDAb molecules in each RF class. Therefore, we tallied the RF values for each RF class as total QDAb spot intensity (RF#1 to $5, Fig S2c). The data established that the distribution profile, as determined by the total intensity, of incubated-QDAb(6) in water for 3 weeks on ice was similar to that of the negative control QD-PEG-NH 2 ( Table S1a). The results of incubation in the presence of SDS revealed that the percentage of QDAb(6) molecules in the RF#1 class were reduced from 76.2% to 29.1% after 24 hrs incubation at 37uC (Fig. 2b and Table S1b), suggesting that majority of QDAb(6) particles formed oligomers in this condition. Although the oligomer formation was also observed with QDAb(6) samples in the absence of SDS, the total value of RF2-RF$5 (39.8%, Table S1c) was much less than in the presence of SDS (70.9%, Table S1b) (Fig. 2b). This enhancement of Ab aggregation by SDS is consistent with the results obtained using unconjugated Ab40 peptides (Fig. S1 b and d).
We also examined the effects of the Ab/QD labeling ratio, in conjunction with SDS, on QDAb oligomerization. In the presence of SDS, the frequency of spots belonging to the RF#1 class significantly decreased as the Ab/QD ratio increased (QDAb(0).QDAb (1).QDAb(6)) ( Table S1b). Accordingly, the number of spots in RF3, RF4, RF$5 classes significantly increased in the order of QDAb(0),QDAb(1),QDAb(6) ( Table S1b). These data demonstrated that the Ab/QD binding ratio is correlated with oligomer formation. To confirm whether the bright, large spots were QDAb oligomers, the incubated samples were examined by atomic force microscopy (AFM). The results revealed that several types of QD clustering were observed in QDAb(6) but not in QD-PEG-NH 2 samples (Fig. 2c). AFM imaging also revealed two types of trimers: one type entails a tandem repeat of three QDAbs (Fig. 2e, top), and another contains a triangular complex of three QDAbs (Fig. 2e, bottom). The distribution of oligomers that were measured from the AFM data (Table S2b) was similar to the fluorescence spot data (Table S2a) of the small oligomers (monomer, dimer, and trimer), suggesting that these small oligomer sizes can be estimated from RF values of fluorescence spots in fluorescence micrographs. We have also detected QDAb oligomer formation as dimer, trimer, or tetramer by electron microscopy (Fig. 2f).

Amyloid fibril formation with QDAb
Although we observed oligomerization solely between QDAb particles, we were unable to observe amyloid fibril formation by QDAb alone. This was expected because the size of the QD is significantly larger than that of Ab. In fact, recent structural work on Ab fibrils have revealed a non-registered parallel b-sheet structure stacking approximately 4 peptide molecules in 1 nm fibril length [27]. This implies that fibril formation is inhibited due to steric hindrance by the QDs. Therefore, unconjugated Ab has to be incorporated with QDAb to effectively image Ab fibril formation. In this study, a QDAb:Ab42 ratio of 1:1000 (0.1%) or 1:10000 (0.01%) was examined for Ab aggregation (Fig. 3). When 0.1% QDAb(6) was mixed with 50 mM Ab42 peptide, bright aggregates were observed (Fig. 3a). The aggregates were stained with a monofluoro bis-styrylbenzene (FSB) derivative [28] (Fig.  S3), demonstrating that these aggregates contain a b-sheet structure. Aggregates were also observed when 0.1% QDAb(1) was mixed with 50 mM Ab42. However, the mean fluorescence intensity of these aggregates was approximately 32% of that of QDAb(6) (Fig. 3b), indicating that the insertion efficiency of QDAb(1) into Ab fibrils is lower than that of QDAb (6). Although Ab aggregates could also be visualized by incubation with 0.1% QDAb(0), the fluorescence intensity was only approximately 14% of that of QDAb(6) (Fig. 3b). This is probably due to non-specific binding between QD-PEG-NH 2 and Ab fibrils, as observed during the preparation of QDAb (Fig. 1b) and in a previous report [20]. Individual Ab filaments can be observed by high-power magnification ( Fig. 3c left). Electron microscopy imaging revealed periodical insertion of QDAb(6) into Ab fibrils (Fig. 3c right). When 0.01% QDAb(6) was incubated with Ab42, the periodicity of single QD molecules was directly observed by fluorescent microscopy (Fig. 3d). The average interval length of the periodicity was 1.961.0 mm, which is close to the estimated value (2.5 mm) based on the NMR fibril structure [27]. These data suggest that QDAb(6) was incorporated into Ab fibrils with a similar efficiency as unconjugated Ab42.
When various concentrations of Ab42 containing 0.1% QDAb(6) were incubated, time-and dose-dependent aggregation was observed (Fig. S4). No fibrils were observed in the sample of 6.3 mM Ab42, suggesting that critical concentration of Ab42 in fibril formation is between 6.3 and 13 mM. In addition, the coaggregation process with unconjugated Ab42 was temperature-dependent, as we did not observe aggregation 1 day after incubation on ice (Fig. S5). This enables us to examine the aggregation process by live imaging system in a time-controlled manner.

4D imaging of Ab aggregation in vitro
Since we succeeded in direct imaging of Ab aggregation under a regular wide-field fluorescent microscope, we next improved the image quality by conducting time-dependent 3D imaging (4D imaging) of Ab aggregation using automated Z-stack image acquisition of a Swept-field confocal microscope ( Fig. 4a and Movie S1). When 0.1% QDAb(6)-containing 100 mM Ab42 was incubated at 37uC, small aggregates were observed on the glass bottom of the well within 1-2 h incubation time. The Ab fibrils then ''grew upwards'' as Ab that aggregated in solution precipitated on top of the fibrils (Movie S1 and S2). The time-course of the Ab aggregation showed a typical sigmoidal curve [29] which consisted of the characteristic time lag, growth, and steady state phases (Fig. 4b). The lag time was approximately 1.5 h, and the aggregation reached a plateau around 12 h. Aggregation of Ab can also be monitored by turbidity at 400 nm [26] and fluorescence measurement of thioflavin T (ThT) binding [5]. Turbidity measurements showed that the aggregation of 20 mM Ab42 reached a plateau around 10-20 h in phosphate buffer (pH 7.4) [26] which is consistent with our 4D imaging in this study (Fig. 4A). In contrast, ThT binding of 10-35 mM Ab42 reached a plateau around 1-2 h in phosphate buffer (pH 7.5 or pH 7.4) with approximately several minutes of lag time [5]. Our data suggest that the ThT binding assay monitors development of b-structure in both protofibrils and fibrils, and that the turbidity assay monitors the amount of fibrils but not protofibrils, presumably because the protofibrils (,5 nm) [30] are too small to induce turbidity at 400 nm. These data imply the existence of two rate-limits in Ab aggregation: b-structure formation, which can be detected as lag time of fluorescence of ThT, and protofibrillization. Total time of b-structure formation and protofibrillization is displayed as the lag time of QDAb 4D imaging. Since the critical concentration of Ab42 aggregation is between 6.3 and 13 mM (Fig.  S5), as described above, approximately 90% of Ab formed aggregates in a steady state phase. On the basis of 4D imaging data, we estimated that the density of stacked-Ab aggregates on the glass bottom in a steady state was 32+/26 mg/ml. Next, we examined whether inhibition of Ab aggregation could be observed using this technique (Fig. 4c). When 0.6 mM (0.1 mg/ ml) anti-b-tubulin (bTb) mouse monoclonal control antibody was mixed with 13 mM of 0.1% QDAb(6) containing Ab42, the fibril formation was unaffected. In contrast, when 0.6 mM anti-Ab mouse monoclonal antibodies (6E10 and 4G8, specifically recognizing Ab1-16 and Ab18-22 epitopes, respectively) were incubated, fibril formation was significantly inhibited (Fig. 4c). The effects of inhibition differed depending on antibodies: 6E10 blocked fibril elongation but not small aggregate formation, whereas 4G8 completely blocked Ab aggregation. Since the 4G8 epitope corresponds to the region that forms a b-structure [31], the binding of 4G8 to the region may directly block the aggregation, whereas 6E10 may affect higher-order Ab aggregation. 3D reconstruction of the Swept-field confocal microscope images demonstrated clear differences in the depth and size of Ab aggregation in the presence of different antibodies ( Fig. S6 and Movies S3-S5).
Fortunately, this image acquisition does not require the fixation or immobilization procedures necessary for AFM and electron microscopic observation. Furthermore, the fibril formation can be observed at a microscopic scale with the use of a simple bioincubator system placed on the microscope stage. Thus, this technology can be applied to micro-scale screening of inhibitory drugs for Ab aggregation (Fig. 4d).

Live imaging of Ab in cells
Microglia have been extensively shown to phagocytose Ab [8,9,32]. Our recent study revealed that the uptake efficiency of the oligomeric Ab was significantly lower (0.2-0.5%) than that of the monomeric form (1-10%) [9]. Therefore, we imaged Ab phagocytosis by microglia using monomeric or oligomeric QDAb (Fig. 5). When monomeric QDAb(6) was incubated with primary cultured mouse microglia for 24 hr, QDAb(6) uptake and accumulation was observed. In contrast, the uptake and accumulation of oligomeric QDAb(6) was significantly less, supporting our recent finding [9]. When monomeric QDAb (6) or QDAb(1) was added to microglia (Fig. 5b), the number of cells containing phagocytosed material increased in a time-dependent manner ( Fig. 5c and Fig. S7). In addition, the Ab/QD ratio also affected the uptake rate as the amount of ingested QDAb(6) was much higher than that of QDAb(1). In contrast, uptake of QDAb(0) was hardly observed under these conditions ( Fig. 5b and  c). Although high-power magnification imaging revealed that QDAb(0) was also ingested by microglia, the ingestion amount was significantly lower than that of QDAb(6) and QDAb(1) (Fig. S8), indicating the ingestion and accumulation are due to Ab peptides on the QD surface. There was no obvious cytotoxicity by ingestion of QD-probes, consistent with the report of the PEG-coated QD [25]. To determine the localization of the accumulated-QDAb in . Samples were incubated in microcentrifuge tubes at 37uC for 1 day, spread between glass slides and cover slips, and observed by regular microscopy using a 100x objective lens with QD filter set (left micrograph) and electron microscopy (right three micrographs). (d) 0.01% QDAb-containing Ab42 (final concentration 50 mM) were incubated in microcentrifuge tubes at 37uC for 1 day, spread between glass slides and cover slips, and observed by regular microscopy using a 100x objective lens with QD filter (left micrograph). The right three micrographs are magnified and brightened micrographs. The top right micrograph is the boxed area in the left micrograph. doi:10.1371/journal.pone.0008492.g003 microglia, we observed the uptake of monomeric QDAb(6) using Lysotracker or Mitotracker. The results showed QDAb(6) partially colocalized with lysosomes (Fig. 5d) as our group and others have reported [8,9]. On the other hand, colocalization of QDAb(6) and Mitotracker, although observed in a recent report [10], was less than that of Lysotracker (Fig. 5e). These results indicate that the majority of accumulated Ab colocalized with lysosomes, but not with mitochondria. The colocalization of QDAb(6) and Lysotracker was reconstructed in 3D-images ( Fig. S9 and Movie S6).

Discussion
In this study, we developed a method for QD-labeling of Ab, which can then be utilized to monitor Ab aggregation for real-time imaging both in vitro and in cells. We believe that this technology can be applied to a wide variety of amyloidogenic peptides and proteins.
Our study found that QDAb forms oligomers and that small oligomer sizes can be estimated from fluorescent microscope imaging. In addition, we could also observe monomeric QDAb uptake by microglia (Fig. 5), suggesting the application of QDAb for the functional analysis of Ab oligomers. Although the data in vitro and in cells showed that the properties of QDAb oligomers are similar to those of untagged Ab oligomers, it is still not known whether the structures of QDAb oligomers and native Ab oligomers are quite the same. Indeed, QDAb failed to form fibrils by itself assumedly due to steric hindrance by the QDs, suggesting a possibility that this also precludes high molecular oligomer formation (such as nonamer or dodecamer). However, QDAb is a useful nanoprobe, if used as a small fraction of the unmodified Abeta, for monitoring the aggregation under the microscope.
In this study, we also successfully observed quantitative 4D live imaging of Ab aggregation (Fig. 4 a and b, Movie S1). Moreover, the inhibition of the Ab aggregation by anti-Ab antibody could be observed in 3D reconstructed imaging. This method could visualize a detailed configuration of Ab aggregates at a microscopic scale, suggesting an application for advanced micro drug screening systems that can distinguish different inhibition mechanisms of Ab aggregation at different stages.
In this study, we successfully observed different ingestion manners between monomeric and oligomeric QDAb by microglia (Fig. 5). The lysosomal accumulation of oligomeric QDAb was poorer than that of the monomeric form, suggesting that it is  difficult for microglia to phagocytize oligomerc Ab. These data imply that cytotoxic Ab oligomers [4,5] are less prone to degradation by microglia in the brain.
Since QD can be detected by multi-photon fluorescence microscopy [19], this technology could be applied to monitor localization and aggregation of Ab in brain. Recently, it was reported that transferrin (Tf)-conjugated quantum rods transmigrated across an in vitro blood-brain barrier model via receptormediated transport [33]. If QDAb can be coated with Tf and the nanoprobe retains the transmigration capability, it may become a powerful tool for in vivo live imaging of Ab aggregation in the brain. Further development of QDAb nanoprobes on the basis of this outcome promises to yield useful information in the analysis of beta-amyloidosis-a hallmark of AD.
Preparation of QDAb 12.5 ml of 8 mM QD-PEG-NH 2 (100 pmol) was put in microcentrifuge tubes and centrifuged at 10,0006g for 1 min at 4uC to eliminate any aggregates. The supernatants were transferred into centrifugal filter units (Microcon YM-100, Millipore), and the remaining volume of the unit was filled with 450 ml of PBS. After centrifugation at 2,8006g for 15 min at 4uC, the unit was refilled with 450 ml of PBS and centrifuged again at 4uC until the volume was reduced to 5 ml. The QD-PEG-NH 2 solutions were adjusted to 9 ml with PBS, supplemented with 1 ml of 10 mM sulfo-EMCS (final concentrations: 10 mM QD-PEG-NH 2 , 1000 mM sulfo-EMCS, and PBS), and incubated for 1 h at 22uC. To quench unreacted sulfo-EMCS, the reacted samples (QD-PEG-CL) were supplemented with 1 ml of 100 mM K-glutamate (pH 7.4) and incubated for 10 min at 22uC. The buffer was changed by micro spin desalting columns (Zeba Micro Spin Desalting Column, Pierce) that were equilibrated with 5 mM EDTA in PBS (pH 6.8) (PBSE), and the volumes were adjusted to 9 ml with PBSE. Meanwhile, dried Cys-Ab40 aliquots were dissolved at a concentration of 1000 or 200 mM in dimethyl sulfoxide (DMSO). The QD-PEG-CL solutions were mixed with various concentrations of Cys-Ab40 (final concentration 100, 10, or 0 mM), and incubated for 1 h at 22uC. To quench the unreacted maleimide group of EMCS, 1 ml of 100 mM 2-mercaptoethanol was added and incubated for 10 min at 22uC. The buffer was changed to pure water using micro spin desalting columns. Concentrations of QD in QDAb were determined at the absorbance of 504 nm according to the instruction manual (Invitrogen). Concentrations of Ab40 in QDAb were measured using Human b Amyloid 1-40 ELISA KIT (Biosource).

Kinetic analysis of Ab40 and Ab42 aggregations by SDS-PAGE
Dried Ab40 and Ab42 aliquots were dissolved at a concentration of 1 mM in DMSO. These Ab solutions were diluted at 50 mM in PBS with or without 1 mM SDS, and 5 ml of aliquots were incubated for various time periods at 37uC. The samples were mixed with sample buffer (final concentration: 50 mM Tris-HCl (pH 6.8), 2% SDS, 0.1% bromophenol blue, and 10% glycerol) and immediately electrophoresed using 16.5% Tris-Tricine gels [34] or 16% Tris-Glycine gels [35]. These gels were stained with Coomassie brilliant blue.

Imaging of QDAb oligomers
QDAb samples were adjusted to a concentration of 3.0 mM in PBS with or without 1 mM SDS, and 5 ml aliquots were incubated for various time periods at 37uC. The oligomer samples were observed by wide-field fluorescence microscopy, atomic force microscopy (AFM), and electron microscopy. Details of sample preparation and analysis in fluorescence microscopy observations are provided in Supplementary Figure 2.

Imaging of fibril formation using QDAb
Dried Ab42 aliquots were dissolved at various concentrations (100, 50, 25, 13, and 6.3 mM) in PBS and mixed with 0.1% or 0.01% QDAb (6). To remove any aggregates, the mixtures were centrifuged at 10,0006g for 1 min at 4uC. The mixtures (50 ml) were put in microcentrifuge tubes or 96-well glass bottom plates (MatTek). The samples in microcentrifuge tubes were incubated for various time periods at 37uC in an air incubator, and observed by wide-field fluorescence microscopy or electron microscopy. The samples in 96well glass bottom plates were incubated at 37uC in 5% CO 2 in a culture chamber (LiveCell TM , Pathology Devices), and directly observed by wide-field fluorescence microscopy or swept-field laserscanning confocal microscopy [QDAb: Excitation: 488 nm at 15% power; Emission filter: Chroma Quad Filter (#C68208) for FITC].

Primary culture of microglia
Microglia were prepared according to the previous report [36]. Microglia were prepared from wild type mouse day 0 newborn pups as described [37], and cultured in Dulbecco's modified eagle medium supplemented with heat-inactivated 10% fetal bovine serum, heat-inactivated 5% horse serum, and 50 mg/ml penicillin/ streptomycin (all from Invitrogen). Microglia released in the tissue culture media by shaking were collected at 14 days after the plating. After confirmation of their purity to be more than 90% by immunocytochemistry (CD11b for microglia, GFAP staining for contaminating astrocytes, and Hoechst 33342 for nuclear staining), cells were used for experiments. The primary cultures were cultured at 37uC in 5% CO 2 .

Preparations of monomeric and oligomeric QDAb for in vivo imaging
To prepare oligomeric QDAb samples, QDAb was adjusted to a concentration of 3.0 mM in PBS with 1 mM SDS, and then incubated for 1 day at 37uC. Monomeric QDAb samples were identical but were not incubated.

Uptake of QDAb by microglia and imaging with Lysotracker and Mitotracker
Mouse microglia were seeded at a density of 50,000 cells/well in 96-well glass bottom plates and preincubated for 10 days. The cells were incubated with 50 nM monomeric or oligomeric QDAb for 1 day, and then supplemented with 50 nM Lysotracker (Invitrogen) or 100 nM Mitotracker (Invitrogen). After an additional incubation for 30 min, the cells were fixed with 4% paraformaldehyde (PFA) for 15 min at 22uC and washed with PBS three times. Vectashield (Vector Laboratories) was added to the wells, and the cells were observed by wide-field fluorescence microscopy or swept-field confocal microscopy. [Lyso/Mitotracker: Excitation: 568 nm at 25% power; Emission filter: Chroma Quad Filter (#C68208) for Texas Red. QDAb: Excitation: 488 nm at 20% power; Emission filter: Chroma Quad Filter (#C68208) for FITC].

Atomic force microscopy
Reaction mixtures were deposited on a 1-(3-Aminopropyl) silatrane-(APS-) modified mica [38,39] glued to the glass slide. AFM images were taken in air, height, amplitude and phase mode using MFP-3D Asylum Research Instrument (Santa Barbara, CA). Regular silicon probes (TESP) with spring constant 40 N/m and resonance frequencies 270-320 kHz were used. Image processing and the cross-section measurements were performed using Femtoscan (Advanced Technologies Center, Moscow, Russia).

Electron microscopy
Reaction mixtures were spread on carbon-coated grids, negatively-stained with 2% uranium acetate pH 7.0, and examined under an electron microscope (H-7500, Hitachi) with an acceleration voltage of 75 kV as described [40]. Images were processed using the FFT bandpass filter (ImageJ 1.40 g, NIH).

Swept-field laser-scanning confocal microscopy
Co-aggregation of QDAb-Ab42 and microglial cells in 96-well glass bottom plates was observed using a swept-field laser-scanning confocal microscopy system (TE-2000U, Nikon). QD was excited by an argon laser, and Lysotracker and Mitotracker were excited by an argon/krypton laser.

Statistics
Data were analyzed by analysis of variances, followed by oneway ANOVA (Newman-Keuls multiple comparison tests) using statistics software (Prism 4.0, GraphPad Software inc.).

Supporting Information
Table S1 Distribution of QDAb molecules belonging to each RF class as determined by the total intensity of QDAb. (a) Distribution of total fluorescent intensity (%) of unconjugated QD-PEG-NH2 and QDAb (6). QD-PEG-NH2 in 50 mM borate was diluted with PBS (final 1-10 nM) and then analyzed immediately. QDAb(6) samples (3.0 mM) in water, PBS, and PBS containing 1 mM SDS were incubated for 3 weeks at 0uC, for 6 weeks at 4uC, and for 3 weeks at 37uC, respectively. The samples were diluted with PBS (final 1-10 nM) and then analyzed immediately. (b and c) 3.0 mM QDAb(0), QDAb(1), and QDAb (6) were incubated in PBS with (b) or without (c) 1 mM SDS for 1 day at 37uC. The samples were diluted with PBS (final 1-10 nM) and then analyzed immediately. The data show averages of 10 fields (86686 mm). Incubation of QDAb(6) in PBS for 6 weeks at 4uC led to a significant increase in the total value of the RF2-RF$5 classes (23.8% to 70.3%) and a decrease in the RF#1 class (76.2% to 29.7%), suggesting that QDAb(6) forms oligomers in PBS at 4uC. In contrast, QDAb(6) incubated in water for 3 weeks on ice was similar to that of the negative control QD-PEG-NH2, suggesting that QDAb (6) can be stored in water on ice but not in PBS in the refrigerator. Distribution of QDAb molecules belonging to each RF class as determined by the total intensity of QDAb. (a) Distribution of total fluorescent intensity (%) of unconjugated QD-PEG-NH2 and QDAb (6). QD-PEG-NH2 in 50 mM borate was diluted with PBS (final 1-10 nM) and then analyzed immediately. QDAb(6) samples (3.0 mM) in water, PBS, and PBS containing 1 mM SDS were incubated for 3 weeks at 0uC, for 6 weeks at 4uC, and for 3 weeks at 37uC, respectively. The samples were diluted with PBS (final 1-10 nM) and then analyzed immediately. (b and c) 3.0 mM QDAb(0), QDAb(1), and QDAb (6) were incubated in PBS with (b) or without (c) 1 mM SDS for 1 day at 37uC. The samples were diluted with PBS (final 1-10 nM) and then analyzed immediately. The data show averages of 10 fields (86686 mm). Incubation of QDAb (6) in PBS for 6 weeks at 4uC led to a significant increase in the total value of the RF2-RF$5 classes (23.8% to 70.3%) and a decrease in the RF#1 class (76.2% to 29.7%), suggesting that QDAb(6) forms oligomers in PBS at 4uC. In contrast, QDAb(6) incubated in water for 3 weeks on ice was similar to that of the negative control QD-PEG-NH2, suggesting that QDAb (6) can be stored in water on ice but not in PBS in the refrigerator. Although longer incubation (3 weeks) showed a slight promotion of Ab aggregation in the presence of 1 mM SDS (a, far right), the distribution profile was similar to the 1 day incubated sample (b, far right). These results revealed that oligomer formation of QDAb(6) nearly saturates after 24 hrs, and that approximately 30% of QDAb(6) remains as monomers under these conditions. Found at: doi:10.1371/journal.pone.0008492.s001 (0.05 MB DOC)

Table S2
Comparison of QDAb comets as determined by fluorescence microscopy and AFM imaging. (a) Frequency of spot number belonging to each RF class from fluorescence microscope observations. The data table shows differences before (1) and after incubation (2). The data of RF#1 (parenthetic data) alone were estimated according to the following calculation method because the RF#1 value of (2) -(1) was not correct. RF#1 value of (2) -(1) calculated by 100 -(RF2+RF3+RF4+RF$5). (b) Frequency of multimerization from AFM observations. The data represent averages of 9 fields (160061600 nm). This comparison shows that the frequency of small oligomers (1-mer, 2-mer, and 3-mer) is similar to the frequency of RF values, suggesting that small oligomer sizes can be estimated from fluorescence intensities. Found at: doi:10.1371/journal.pone.0008492.s002 (0.04 MB DOC) Figure S1 Kinetics of Ab42 and Ab40 aggregations. 50 mM Ab42 peptide (a and b) and 50 mM Ab40 peptide (c and d) were incubated in PBS with or without 1 mM SDS for various time periods at 37uC. After the incubation, these samples were electrophoresed using 16.5% Tris-Tricine [1] (a and c) and 16% Tris-Glycine gels [2] (b and d). Aggregation of Ab42 was more rapid than Ab40 in PBS both with and without SDS.  [3]. An aliquot (2 ml) of oligomer sample solution, which was diluted to 1-10 nM, was spread between the glass slide and the coverslip. The coverslip was taken off, dried, and placed on a wide-field fluorescence microscope. The gray images (2040 pixel61536 pixel: 175 mm6132 mm) were obtained using a 100x objective lens with a QD filter set. A micrograph represented an average of 5 frames (each exposure time was 0.2 s). (b) Measurement of relative fluorescence. The micrographs were analyzed using ImageJ software (NIH). In this analysis, we used a 100061000 pixel area in the central region of the micrographs because of aberration at the periphery. The micrographs were thresholded under the same conditions and then were analyzed using the ''analyze particles'' program of ImageJ. Relative fluorescence (RF) was defined as the product of the area size (pixel) and mean fluorescence intensity. The average RF of unlabeled QD-PEG-NH2 was expressed as 1 RF unit (RF1). In this study, #1.5, 1.5-2.5, 2.5-3.5, 3.5-4.5, $4.5 of RF were indicated as RF#1, RF2, RF3, RF4, and RF$5, respectively. Each analysis averaged 10 micrographs (one micrograph contained several hundred particles). (c) Frequency of spot number and total fluorescent intensity. Spot number (Table S2) and total fluorescent intensity (Figure 2b and Table S1) reflect the number of oligomers and the number of QDs belonging to each RF class. The aggregates were observed by wide-field fluorescence microscopy using a 20x objective lens with FITC (QD) or Blue (Blue) filter sets. Since the FSB derivative binds to the b-sheet structure of Ab fibrils [4], it is likely that these aggregates are typical Ab fibrils containing b-sheet structure. were incubated at 37uC in 96 well glass bottom plates. The samples were observed at 0, 4, and 21 h from the start of incubation by wide-field fluorescence microscopy using a 20x objective lens with FITC filter set. No aggregates were observed in all 0 h samples. Although dose-and time-dependent aggregation were observed in the 50, 25, and 13 mM samples, aggregates were not observed in the 6.3 mM sample, suggesting that the critical concentration for Ab42 aggregation was between 6.3-13 mM under these conditions. Found at: doi:10.1371/journal.pone.0008492.s006 (7.12 MB TIF) Figure S5 Temperature-dependent Ab aggregation. 0.1% QDAb(6)-containing Ab42 (final concentration 50 mM) was incubated for 1 day at 37uC (left), for 1 day on ice (middle), and for 1 day at 37uC after 1 day on ice (right), and observed by widefield fluorescence microscopy using a 100x objective lens with FITC filter set. The results showed that Ab aggregates were not formed after 1 day at 0uC incubation (middle). The sample on ice formed aggregates by additional incubation (right), suggesting that the 0.1% QDAb (6) (6) and QDAb(0). Primary mouse microglia were incubated with monomeric QDAb (6) or QDAb(0) (final concentration 50 nM) for 24 h, followed by fixation with 4% PFA, and observed by wide-field fluorescence microscopy using a 100x oil objective lens (TE-300, Nikon Instruments) and QD filter set (green). Found at: doi:10.1371/journal.pone.0008492.s010 (5.75 MB TIF) Figure S9 Co-localization of ingested QDAb6) and Lysotracker in microglia. Primary mouse microglia were incubated with 50 nM monomeric QDAb6) for 24 h, followed by incubation with 50 mM Lysotracker for an additional 30 min. The cells were fixed with 4% PFA, and observed by Swept-field laser-scanning confocal microscopy using 488 nm excitation (QD, green) and 568 nm excitation (Lysotracker, red). Far right panel is the 3D reconstruction image in the same field. The movie of this 3D image is in Movie S5. Movie S1 4D imaging of Ab aggregation. 0.1% QDAb(6)containing Ab42 (final concentration 100 mM) was incubated in PBS for 24 h at 37uC in 96 well glass bottom plate. The sample was observed every 30 min by swept-field confocal microscopy using 488 nm excitation laser (20%) and 60x objective lens. The bird's-eye movie is played at a speed of 4 h/s. Movie S2 4D imaging of Ab aggregation. 0.1% QDAb(6)containing Ab42 (final concentration 100 mM) was incubated in PBS for 24 h at 37uC in 96 well glass bottom plate. The sample was observed every 30 min by swept-field confocal microscopy using a 488 nm excitation laser (20%) and 60x objective lens. The movie in angled bird's-eye view is played at a speed of 4 h/s. Found at: doi:10.1371/journal.pone.0008492.s013 (5.68 MB MOV) Movie S3 3D imaging of Ab aggregates in PBS. 0.1% QDAb(6)containing Ab42 (final concentration 50 mM) was incubated in PBS for 1 day at 37uC in a 96 well glass bottom plate. The sample was observed by swept-field confocal microscopy using a 488 nm excitation laser (75%) and 100x objective lens. Found at: doi:10.1371/journal.pone.0008492.s014 (5.53 MB MOV) Movie S4 3D imaging of Ab aggregates in PBS with anti-bTubulin (control) antibody. 0.1% QDAb(6)-containing Ab42 (final concentration 50 mM) was incubated in PBS containing anti-bTubulin antibody for 1 day at 37uC in a 96 well glass bottom plate. The sample was observed by swept-field confocal microscopy using a 488 nm excitation laser (75%) and 100x objective lens.