Binding of Apolipoprotein E Inhibits the Oligomer Growth of Amyloid-β Peptide in Solution as Determined by Fluorescence Cross-correlation Spectroscopy*

Background: ApoE is the most significant risk factor for Alzheimer disease, with known effects on Aβ deposition in the brain. Results: ApoE binds to aggregating Aβ peptides and maintains a faster diffusion rate for the Aβ peptide over time. Conclusion: Binding of apoE to Aβ slows the oligomerization of Aβ. Significance: FCCS measurements quantify isoform-dependent differences in apoE binding to Aβ in solution. One of the primary neuropathological hallmarks of Alzheimer disease is the presence of extracellular amyloid plaques resulting from the aggregation of amyloid-β (Aβ) peptides. The intrinsic disorder of the Aβ peptide drives self-association and progressive reordering of the conformation in solution, and this dynamic distribution of Aβ complicates biophysical studies. This property poses a challenge for understanding the interaction of Aβ with apolipoprotein E (apoE). ApoE plays a pivotal role in the aggregation and clearance of Aβ peptides in the brain, and the ϵ4 allele of APOE is the most significant known genetic modulator of Alzheimer risk. Understanding the interaction between apoE and Aβ will provide insight into the mechanism by which different apoE isoforms determine Alzheimer disease risk. Here we applied alternating laser excitation fluorescence cross-correlation spectroscopy to observe the single molecule interaction of Aβ with apoE in the hydrated state. The diffusion time of freely diffusing Aβ in the absence of apoE shows significant self-aggregation, whereas in the presence of apoE, binding of the protein results in a more stable complex. These results show that apoE slows down the oligomerization of Aβ in solution and provide direct insight into the process by which apoE influences the deposition and clearance of Aβ peptides in the brain. Furthermore, by developing an approach to remove signals arising from very large Aβ aggregates, we show that real-time single particle observations provide access to information regarding the fraction of apoE bound and the stoichiometry of apoE and Aβ in the complex.

Alzheimer disease (AD) 4 is a neurodegenerative disorder of aging that affects the cognitive ability of the brain. AD is characterized by two histopathological features of the brain: insoluble extracellular plaques composed of amyloid-␤ (A␤) peptides and intracellular neurofibrillary tangles formed from hyperphosphorylated Tau, a microtubule-associated protein.
Although the primary cause and progression of AD are still not well understood, they are thought to be linked to the aggregation of A␤ peptides. The A␤ peptides are generated as cleavage fragments by the action of ␥ and ␤ secretases on the amyloid precursor protein, a constitutively expressed transmembrane protein. Due to their inherently disordered and "sticky" nature, the resulting A␤ peptides easily aggregate into oligomers, then fibrils, and finally, mature plaques in the brain.
To date, the ⑀4 allele of the apolipoprotein E (APOE) gene is the strongest known risk factor for the late onset form of AD (1)(2)(3)(4). The apoE protein is involved in lipid transport throughout the body and is the principal lipid transport protein in the central nervous system. There are three apoE isoforms: E2, E3, and E4, and studies have demonstrated (1)(2)(3)(4) increased risk of AD and earlier age of onset in individuals carrying the ⑀4 allele. Although the ⑀4 allele is linked to both the sporadic and the familial late onset forms of AD, the mechanism of this association remains unknown. However, studies have revealed the presence of apoE in the amyloid fibrils and plaques of Alzheimer brains, which strongly suggests that apoE plays a critical role in the pathogenesis of AD through its interaction with aggregating A␤ peptides (5)(6)(7)(8). It has also been established that apoE plays an important role in the homeostasis of A␤ in the brain, through its influence on both the deposition and the clearance of the peptide (9 -13). Emerging in vivo techniques in humans and other animal models further strengthen our understanding of the distributions of and relationships between A␤ and apoE in the brain (14,15). However, the interaction of A␤ with apoE is still poorly understood, with conflicting evidence with respect to differences in isoform interaction with A␤ (11). In addition, many previous studies were also designed with particular attention on isoform influence on amyloid burden rather than A␤ toxicity, and therefore focused on apoE associations with fibril/plaque species rather than the oligomeric forms of A␤ that are now recognized as the pathogenic species. As oligomeric A␤ represents a dynamic intermediate along the fibrillization pathway, it is very difficult to investigate this interaction and determine the affinity of apoE with A␤ directly in solution, particularly at the single molecule level. We therefore require insights into the oligomeric state of A␤, its binding with apoE, and the distribution of these species across the system to understand how apoE influences A␤ deposition and clearance in the brain.
Fluorescence correlation spectroscopy (FCS) is a statistical technique to detect chemical reactions and determine translational and rotational diffusion coefficients of molecules and complexes (16 -18). It is based on monitoring intensity fluctuations emitted from fluorescent molecules diffusing through a tightly focused laser excitation volume (ϳ1 fl). By subjecting these fluctuations to an autocorrelation analysis, G (2) () ϭ (͗I(t)I(t ϩ )͘)/(͗I͘ 2 ), the molecular diffusion time, sample concentration, and photophysical properties can be extracted. With precise knowledge of the diffusion time, D and beam waist, , of the excitation laser spot, the diffusion coefficient, D ϭ ( 2 )/(4 D ) can be determined, which is proportional to the Einstein-Stokes hydrodynamic radius R H ϭ (K B T)/(6n 0 D).
If two differently labeled species are in the sample, their colocalization can be monitored using fluorescence cross-correlation spectroscopy (FCCS), originally developed by Schwille et al. (19). FCCS has been applied to study binding events (20) and enzyme kinetics such as oligonucleotide cleavage (21) and protease cleavage (22), and to monitor calcium activity in cells containing calmodulin (23). The addition of alternating laser excitation (ALEX) eliminates spectral cross-talk between fluorophores and reduces the possibility of false positives because the different fluorophores are not excited simultaneously and their signals can be temporally separated. ALEX was first developed by Kapanidis et al. (24) for fluorescence resonance energy transfer (FRET) measurements to determine the stoichiometry between biomolecules and later extended to FCCS to eliminate cross-talk between two fluorescent proteins in cells (25), to monitor single molecule interactions (26,27), and also to antibody-based protein detection (28).
We applied ALEX-FCCS to investigate the interaction of apoE with A␤ in the hydrated state. The ability of this technique to report on the distribution of A␤ species, along with the binding of A␤ to other proteins, provides a powerful tool for studying the interaction of the peptide with apoE in the oligomeric state. To probe the molecular basis of the role of apoE in the development of Alzheimer disease, the E3 and E4 isoforms were selected as representative examples for this study. Because the fluorescent labeling required for this study takes advantage of thiol binding chemistry at a cysteine residue, it was necessary to avoid binding to the native cysteine residue found at position 112 in apoE3. Thus, we utilized the apoE3-like (apoE3L) pro-tein, in which a serine is substituted for the cysteine at position 112. A thiol-reactive fluorescent label was then introduced to the C-terminal domain of apoE3L or apoE4 by replacing Trp-264 with a cysteine residue. It has been shown that the cysteine substitution and subsequent modification of the W264C mutation of apoE with the thiol-specific label do not alter its predicted distribution among plasma lipoproteins, and circular dichroism analysis of the labeled protein is indistinguishable from the wild-type apoE (29 -31).

EXPERIMENTAL PROCEDURES
Materials-Hexafluoro-2-propanol was purchased from Sigma-Aldrich. Dimethyl sulfoxide (DMSO) was purchased from Fisher Scientific. Alexa Fluor 488 C 5 -maleimide was obtained from Invitrogen Molecular Probes, and Atto 647N NHS ester was obtained from Fluka Analytical/Sigma-Aldrich.
Preparation of Amyloid-␤-Amyloid-␤  peptide was purchased from Bachem (catalogue number H-1194, Torrance, CA). The peptide was dissolved in hexafluoro-2-propanol and incubated at room temperature with gentle rocking for 48 -72 h. SpeedVac or evaporation was then used to remove the hexafluoro-2-propanol, resulting in a monomeric A␤ pellet. To direct preferential labeling of the N-terminal amine group of A␤, a 0.1-mg aliquot of peptide was dissolved in 10 l of DMSO and reacted at pH 7.0 with 3 l of Atto 647N NHS ester label (10 mM stock in DMSO) and 500 l of phosphate-buffered saline (PBS; 137 mM NaCl, 2.7 mM KCl, 10.1 mM Na 2 HPO 4 , 1.8 mM KH 2 PO 4 , pH 7.0). The mixture was incubated for 1 h at room temperature, after which it was washed six times with fresh PBS. After the final PBS wash was removed, hexafluoro-2-propanol was added to the labeled peptide and allowed to evaporate. The resulting pellet was stored at Ϫ20°C until use. Immediately before the experiment, the pellet was warmed to room temperature and dissolved in fresh DMSO to achieve a stock solution of 1 mM A␤. To generate oligomers, the A␤ solution was then diluted into PBS buffer to a final concentration of 10 M. The 10 M solution was allowed to incubate at room temperature for 0 -3 h to produce oligomers. As demonstrated previously (29,32,33), these oligomeric preparations are A11-positive, prefibrillar oligomers (34), with a 10 M solution producing particles of ϳ10 nm by atomic force microscopy imaging.
Cloning, Purification, and Labeling of Apolipoprotein E-Human apoE4 contains no endogenous Cys, so site-specific Alexa Fluor 488 incorporation was achieved by substituting the native tryptophan at position 264 with a cysteine and reacting the purified protein with Alexa Fluor 488 C 5 -maleimide. To specifically target the fluorophore label to an -SH group in the C-terminal region of the apoE3 protein, a cysteine-free version of apoE3 was first generated by substituting the native Cys residue at position 112 with a Ser as described previously (29). This apoE3L gene was then used as a template for introducing a cysteine substitution at position 264 by PCR mutagenesis. Multiple lines of evidence indicate that this apoE3-like protein serves as a reasonable mimic of apoE3. These include structural studies (reduced domain interaction (29)), LPS binding properties (35), and formation of an SDS-resistant complex with A␤ unique to native apoE3 (36). The gene encoding human apoE3L-W264C or apoE4-W264C was then cloned, expressed, and purified (29), with an addition of a single pass through a His-bind Ni(II) chelating column prior to the size exclusion. Labeling of apoE was accomplished by incubating the sample with 200 M Alexa Fluor 488 C 5 -maleimide for 1 h at room temperature in the presence of 100 M tris(2-carboxyethyl)phosphine to maintain reduced disulfides. Excess dye was removed by running the sample through a Bio-Spin 6 column (Bio-Rad). The labeled apoE was stored at 4°C and diluted into PBS buffer (pH 7.4) to obtain the desired concentration immediately before the experiments.
Instrumentation-We conducted our experiments using a MicroTime 200 confocal fluorescence spectroscopy system (PicoQuant GmbH, Berlin, Germany) equipped with two pulsed diode lasers (470-and 640-nm wavelengths, ϳ80-ps pulse width) operating at a repetition rate of 20 MHz. The 640-nm laser pulse was delayed by 25 ns with respect to the 470-nm laser to produce alternating laser excitation (Fig. 1A). The lasers were coupled into a polarization-preserving single mode optical fiber, recollimated, and then focused to a diffraction-limited spot of ϳ250-nm diameter by an Olympus 1.45 NA 100ϫ oil objective to a height of 5 m above a glass coverslip surface. The average power of each laser was 50 microwatts at the sample. The fluorescence emission was split by a dichroic mirror (600DCXR, Chroma Technology Corp., Bellows Falls, VT), spectrally filtered with emission bandpass filters (HQ520/40 m and HQ680/75 m, Chroma Technology Corp.), and detected by two avalanche photodiode detectors (SPCM-AQR-14, PerkinElmer Life Sciences). The signals were processed by a time-correlated single photon counting board (PicoHarp 300, PicoQuant, Westfield, MA), operating in timetagged time-resolved mode. The time-tagged time-resolved mode of the data acquisition records the photon arrival time from the last excitation pulse (micro-time) with 50-ps relative time resolution and the photon arrival time from the start of the experiment (macro-time) with 100-ns absolute time resolution. Time-correlated single photon counting board of separate detection channels allows for the temporal analysis of all detected photons. In particular, it enables the determination of which excitation laser (470 or 640 nm) leads to the detection of a photon. Auto-and cross-correlations were calculated and fitted using the SymPhoTime software package (PicoQuant GmbH).
ApoE3L was labeled with a single Alexa Fluor 488 fluorophore, which exhibits an emission peak at ϳ519 nm after excitation with the 470-nm laser, and detected at avalanche photodiode 2, the "green" channel. Similarly, A␤ was labeled with a single Atto 647 fluorophore, which exhibits an emission peak at ϳ668 nm after excitation with the 640-nm laser, and detected by avalanche photodiode 1, the "red" channel. The red channel detects both free and bound A␤, and the green channel detects free and bound apoE3L. Because time-correlated single photon counting board electronics assign time tags to all detected photons, only photons that arrive at the two detectors simultaneously are analyzed. Cross-correlations were formed from photons detected in the green channel while the 470-nm laser was on and from photons detected in the red channel while the 640-nm excitation laser was on. In this way, leakage of photons from Alexa Fluor 488 into the red channel and direct excitation of the Atto 647 by the 470-nm excitation laser were excluded from the analysis, eliminating sources of spurious cross-correlation signals. Therefore, ALEX-FCCS allows us to resolve signals only from the truly bound species. Time traces of both A␤ and apoE3L are shown in Fig. 1B.
The cross-correlation signal from freely diffusing fluorescent molecules illuminated by two excitation lasers is where the term denotes the temporal decay of the cross-correlation function by the bound molecule with diffusion time D,XY in the effective superimposed observation volume (V eff ). C X and C Y are the concentrations of free X and Y molecules, and C XY is the concentration of bound molecules. At lag time ϭ 0, Equation 1 can be rewritten as (21) G x and G y are the autocorrelations of channels x and y. In autocorrelation analysis, the number of molecules N in the excitation volume is inversely proportional to the amplitude of the autocorrelation function G(0), whereas in cross-correlation analysis, the number of bound molecules N XY is proportional to G XY (0) in the volume. By analyzing the auto-and cross-correlation amplitudes, the number of bound molecules can be determined.
Aggregate Removal Algorithm-One problem encountered in taking accurate FCS measurements of A␤ is the presence of extremely large aggregates resulting in huge fluorescent bursts. These aggregates are most likely from A␤ that has formed large oligomers and are not necessarily a true representation of the average particle size in our sample. The large aggregates may also result from the tendency of A␤ to stick to glass surfaces, such as those used for the experimental measurements. Regardless of the cause, the large fluorescent bursts detected can skew the results of our analysis. As an example, Fig. 1C shows a large A␤ aggregate with a burst size of almost 10 times the average signal. To eliminate these aggregates from our data, we implemented a custom algorithm that cuts a portion of the intensity time trace when photon burst counts larger than five times the average signal are observed. The remaining portion of the time trace is then stitched back into the original time trace for photon correlation analysis.

RESULTS AND DISCUSSION
Kinetics and Stoichiometry of the Binding Reaction-Although it is not as toxic as the A␤  species, A␤  induces a similar, albeit attenuated, pathology in neurons. We limited our initial study to A␤  because it demonstrates more predictable behaviors, including lower surface affinity, and has a slower aggregation rate in solution. To establish the binding interactions between A␤ and apoE3L, we used a sample solution consisting of 10 M A␤ and 10 M apoE3L. Because micromolar A␤ in solution undergoes a process of oligomerization (33,37,38), a series of time measurements was performed at time 0, 15 min, 30 min, 1 h, 3 h, and 4 h after introducing A␤ into solution with and without apoE3L protein. For each FCCS measurement, a small volume of this sample was diluted to less than 1 nM in PBS to ensure that the excitation volume contained at most one molecule per laser pulse, which gives a high signal to noise ratio. The data were recorded for 5 min at each time interval. To obtain an accurate statistical error distribution, the whole time series experiment was repeated five times with five independent A␤-apoE3L samples.
The measurement taken immediately following the dilution (time 0) reveals very little correlated signal from the two probes, reflecting initially weak binding between A␤ and apoE3L. This is evident in Fig. 2A by the flat black cross-correlation curve. As the reaction is monitored over time, the amplitude of the crosscorrelation slowly increases from a value of G(0) ϭ 0 during the initial measurement to G(0) ϳ 0.59 after 4 h, clearly indicating binding between A␤ and apoE3L. Additional measurements were taken over a course of 48 h, but no significant change in the correlation amplitude was observed, indicating that equilibrium was established.
Following the kinetic assessment of the binding reaction between A␤ and apoE3L over time, we determined the fraction of A␤ and apoE3L that bind to each other to form the complex with both molecules added at a concentration of 10 M. At time 4 h, the autocorrelation was fitted to values of G A␤(0) ϳ 3.7 and G apoE3L(0) ϳ 5.2, which are inversely proportional to the number of A␤ and apoE3L molecules in the excitation volume, or N A␤ ϳ 0.27 and N apoE3L ϳ 0.19. The cross-correlated value is G A␤/apoE3L ϳ 0.59 (Fig. 2B). Solving Equation 3 with these values yields N A␤/apoE3L ϳ 0.03, the number of fully bound particles detected in the excitation volume. This implies that approximately N fraction bound , A␤ ϭ (N A␤/apoE3L )/(N A␤ ϩ N A␤/apoE3L ) ϳ 10.0 Ϯ 3% of the total A␤ concentration and N fraction bound,apoE3L ϭ (N A␤/apoE3L )/(N apoE3L ϩ N A␤/apoE3L ) ϳ 13.6 Ϯ 3% of the total apoE3L concentration form a binarycomplex species. It should be noted that although A␤ and apoE have a single fluorophore attached, the diffusion bound complex has variations in its number of apoE and A␤ proteins, particularly the latter from self-aggregation. Due to the way the cross-correlation is calculated, any such inhomogeneity can result in a lower apparent fraction bound as the free A␤ available for binding becomes much lower than the free apoE. It should also be noted that the amount of oligomeric A␤ in the system is significantly lower than the total peptide added. Thus, the values for apparent dissociation constants (below) should be considered as upper limits.
Diffusion Rate of the Binary Complex-Next, we compared the diffusion time of the unbound A␤ and apoE3L with the bound species (Fig. 2C) by analyzing the normalized correlation data. In principle, if every A␤ molecule binds to every apoE3L molecule, then the autocorrelation curves for the two channels would be identical. However, because the red channel measures both free and bound A␤ (Fig. 2C, red) and the green channel contains data for the mixture of free and bound apoE3L (Fig.  2C, green), at equilibrium we expect the two autocorrelation curves to be similar but not identical. Analysis of the autocorrelation signal at 4 h provides average diffusion times of 110 s for apoE3L and 100 s for A␤. This corresponds to a hydrodynamic radius of ϳ1.6 nm for both apoE3L and A␤. By using cross-correlation spectroscopy to analyze the signals from both channels arriving within a very short time interval, signals from the free proteins can be separated from that of the bound complex. The cross-correlated signal, representing a complex of A␤-apoE3L, had a diffusion time of ϳ2 ms (Fig. 2C, black) with an average hydrodynamic radius of 27 nm, which suggests that the complex in solution forms from the cooperative association of more than one apoE3L and A␤ oligomer. The formation of large, multimeric complexes of apoE and A␤ is consistent with our previous observations of apoE structure upon A␤ binding (29).
Comparison of ApoE Isoforms-As described earlier, the relevance of the A␤-apoE interaction was first recognized due to the increased AD risk for individuals carrying the ⑀4 isoform allele of apoE. We therefore investigated the binding of A␤ as a function of apoE concentration for both the E3L and the E4 proteins by FCCS (Fig. 3). In these experiments, we fixed the concentration of A␤ at 10 M while adjusting the concentration of apoE3L or apoE4 to 5, 10, 20, and 40 M. The correlation spectroscopy data were acquired at uniform incubation times (2 h) for all mixtures. The fraction of A␤ bound to either apoE3L or apoE4 was calculated from the correlation data as described in the previous section. The results indicate that the E3L protein has a substantially higher affinity for oligomeric A␤ when compared with apoE4. The slightly cooperative concentration dependence is consistent with the notion that apoE3L binding to A␤ results in a complex that involves more than one apoE molecule.
Previously, we used surface plasmon resonance and EPR spectroscopy of site-directed spin labels to explore the affinity of these two principal isoforms with oligomeric A␤ (29). Although surface plasmon resonance utilizes an immobilized substrate, both species are free in solution with measurements  by FCCS and EPR. The A␤-apoE association by EPR can be observed on the basis of (i) changes in the rotational diffusion (both local and global) and (ii) dipolar coupling via increased self-association of the apoE C-terminal domain in the presence of oligomeric A␤. However, FCCS has clear advantages over EPR for detecting the A␤-apoE complex as both rotational and translational diffusion can be measured and also well separated as these two effects occur at very different timescales. Furthermore, the sampling of single particles in FCCS provides superior resolution regarding the heterogeneity of the species in solution.
We then compared the hydrodynamic radius of the bound particle to free A␤ incubated in the absence of apoE3L and apoE4 (Fig. 4). A␤ is known to form soluble oligomers that are conformationally and pathologically distinct (34,37). Under the conditions employed here, A␤ oligomers assemble into relatively disordered peptides defined as prefibrillar oligomers (37). Monomeric A␤ has a hydrodynamic radius of ϳ0.7 nm as measured by FCS. Rapid aggregation of 10 M A␤ over the course of the first 2 h resulted in many large particles approaching 100-nm hydrodynamic radius, which then dissociated afterward to an average size of 60 nm at around 4 h. This is consistent with previous measurements of A␤ aggregation (37,38). Although the basis for this partial disassembly is unclear, it may be related to a reorganization of the prefibrillar oligomer, including the elimination of antiparallel interactions (38).
The change in hydrodynamic radius is markedly different when an equal amount of apoE is present (Fig. 4). When mixed with the apoE3L protein, the average bound complex particle size slowly approaches 27 nm with no observable dissociation into smaller particles. This suggests that apoE3L, when bound to A␤, forms a more stable complex and interferes with the ability of A␤ to form larger oligomers. Inclusion of apoE4 also results in smaller hydrodynamic radii over time, although to a lesser extent than is achieved by apoE3L. As shown in Fig. 4, incubation of A␤ ϩ apoE4 results in an average particle size of 46 nm at 4 h, a size midway between A␤ ϩ apoE3L and A␤ alone. The basis of this distinction is not clear; however, the lower A␤ affinity for apoE4 may relate to a diminished ability to arrest A␤ oligomerization.
Although the interaction of apoE and A␤ as detected by FCCS does not reflect submicromolar affinity, apoE is known to form strong, SDS-resistant complexes with A␤ under conditions similar to those employed here (36,39). As mentioned above, we estimate apoE affinity for A␤ based on the total peptide in the system, although the concentration of oligomeric species is considerably lower. The dissociation constants calculated here represent relative upper limits and provide a useful mechanism to compare the affinity for A␤ between apoE isoforms. Equilibrium binding measurements are further hindered by the increased aggregation encountered with employing higher levels of A␤ and/or longer incubation times. Due to the inherent self-assembly of A␤, FCCS measurements of A␤ binding are most useful for determining differences in initial rates, rather than steady states. Nevertheless, the reduced A␤ binding and polymerization rate seen with apoE4 is significant in a physiological context. Low affinity interactions have significant physiological relevance as competition among relatively weak interactions can have a profound consequence in age-dependent diseases. For example, the balance between low affinity metal ions has been proposed to influence the degree of A␤ toxicity over a period of decades (40). In addition, elevated local concentrations of apoE draw physiological connections to our findings. In response to injury, neurons rapidly release large amounts of apoE (41), where its accumulation in the extracellular matrix (42) results in effective concentrations exceeding the levels applied here. Future FCCS work will also consider the interaction of lipid-bound apoE isoforms and A␤ using FCCS as previous measurements have shown a higher affinity for oligomeric A␤ when the protein is assembled in HDL-like lipoprotein particles (29). Such measurements will demand additional analysis as A␤ has the potential to interact with the lipid phase alone in such particles. However, previous experience (43,44) in using FCS and photon antibunching to determine lipoprotein particle size and protein stoichiometry will be helpful in this pursuit.
Consequences of ApoE Association-We have shown that the presence of the apoE3L protein retards the progression of A␤ monomers into oligomers. If such an effect occurs in vivo, it may correlate with a packaging of A␤ to facilitate its clearance from the brain (45). We and others have previously postulated that reduced binding of apoE4 to oligomeric A␤ may correlate with a loss of apoE protection with regard to A␤ clearance (29,(45)(46)(47)(48). The ability of ALEX-FCCS to provide a quantitative measure of apoE binding to A␤ and also A␤ assembly will be helpful in not only elucidating the molecular mechanism of A␤ pathogenicity and protection, but also in identifying factors that influence these processes.
Conclusion-The ⑀4 allele of the APOE gene represents the most significant genetic risk factor for AD (49 -51). The differential ability of apoE isoforms to interact with and clear A␤ is likely key to the mechanism of the isoform influence on AD (12,FIGURE 4. Bar chart of the hydrodynamic radii at different reaction times as measured by FCCS. At time 0, A␤ (Abeta) has a hydrodynamic radius of 0.7 nm, which increases to over 60 nm after a 4-h reaction (blue). After a 4-h reaction in the presence of apoE3L or apoE4, the bound complex has a size of 28 or 46 nm, respectively. Note that at time 0, there is no binding between the two molecules. and therefore, the red and blue bars have been omitted. Error bars represent S.D. 29). An accumulating body of evidence demonstrates just how vital a role apoE plays in the aggregation and clearance of A␤ peptides in the brain (9,45,52). This dynamic process presents an intriguing point of intervention for rational therapies designed to prevent and/or delay the progression of AD pathology, but to approach this question, it is first necessary to understand the precise interactions of apoE with A␤ as they relate to the deposition and clearance of A␤ peptides. Although other methods can detect both A␤ binding and oligomerization (33,37), they are limited in their ability to describe the size and stoichiometry distribution of species in the system. We have shown using ALEX-FCCS that apoE inhibits the oligomerization of A␤ in the hydrated state. We have also demonstrated the ability of this method to report on the size and composition of biological complexes in solution, therefore providing a powerful tool for unraveling the molecular interaction of A␤ with apoE in Alzheimer disease. Furthermore, because the entropically driven growth of A␤ oligomers is an indicator for a high energy (and likely pathogenic) state of the peptide, ALEX-FCCS provides a quantitative approach for the real-time, in-solution evaluation of small molecules that can modulate apoE interaction and/or the stability of A␤.