NMR Reveals Two-Step Association of Congo Red to Amyloid β in

Publication Date (Web): November 15, 2010. Copyright ... In this study, we have examined in detail the binding of the fibril-formation inhibitor Congo...
21 downloads 3 Views 2MB Size
J. Phys. Chem. B 2010, 114, 16003–16010

16003

NMR Reveals Two-Step Association of Congo Red to Amyloid β in Low-Molecular-Weight Aggregates Marie Ø. Pedersen,†,‡,§,| Katrine Mikkelsen,†,‡,§ Manja A. Behrens,‡,§ Jan S. Pedersen,‡,§ Jan J. Enghild,†,‡,⊥ Troels Skrydstrup,†,‡,§ Anders Malmendal,*,†,‡,§ and Niels Chr. Nielsen*,†,‡,§ Center for Insoluble Protein Structures (inSPIN), Interdisciplinary Nanoscience Center (iNANO), Department of Chemistry, and Department of Molecular Biology, Aarhus UniVersity, Denmark ReceiVed: August 25, 2010; ReVised Manuscript ReceiVed: October 24, 2010

Aggregation of the Amyloid β peptide into amyloid fibrils is closely related to development of Alzheimer’s disease. Many small aromatic compounds have been found to act as inhibitors of fibril formation, and have inspired the search for new drug candidates. However, the detailed mechanisms of inhibition are largely unknown. In this study, we have examined in detail the binding of the fibril-formation inhibitor Congo Red (CR) to monomeric Aβ1-40 using a combination of 1D, 2D, saturation transfer difference, and diffusion NMR, as well as dynamic light scattering experiments. Our results show that CR binds to the fibril forming stretches of Aβ1-40 monomers, and that complex formation occurs in two steps: An initial 1:1 CR:Aβ1-40 complex is formed by a relatively strong interaction (Kd ≈ 5 µM), and a 2:1 complex is formed by binding another CR molecule in a subsequent weaker binding step (Kd ≈ 300 µM). The size of these complexes is comparable to that of Aβ1-40 alone. The existence of two different complexes might explain the contradictory reports regarding the inhibitory effects of CR on the fibril-formation process. 1. Introduction With more than 20 million patients worldwide, Alzheimer’s disease (AD) is the most common neurodegenerative disease.1 Brain tissue of AD patients shows deposition of protein plaques composed mainly of the 39-43 residue Amyloid β (Aβ) peptide, which is prone to form amyloid fibrils characterized by a crossβ-sheet structure. The propensity for aggregation into amyloid fibrils is shared with other disease-related peptides, such as R-synuclein in Parkinson’s disease and the Prion proteins in Creutzfeldt-Jacob disease.2,3 Aβ fibrils themselves have been shown to be toxic,4,5 but in the past decade, increasingly strong evidence has been established supporting that soluble Aβ oligomers are the real pathogenic species of Alzheimer’s disease.6 The toxicity of Aβ might also be linked with membrane interacting properties of the peptide, and studies have shown for instance that detergents induce β-sheet structure in Aβ1-40,7 and that Aβ1-42 destabilizes phospholipid bilayers.8 Both amyloid fibrils, and in many cases also fibril-forming peptides in solution, appear to interact with a large number of small aromatic compounds. These include various dye molecules used for amyloid detection, many of which have been found to decrease the rate of fibril formation.9,10 One prominent example of these molecules is Congo Red (CR)11 (see structure in Figure 1), which in addition to being frequently used for fibril quantification has been repeatedly reported to act as an inhibitor against aggregation of Aβ,10,12,13 and other amyloid-forming * To whom correspondence should be addressed. Address: Center for Insoluble Protein Structures (inSPIN), Interdisciplinary Nanoscience Center (iNANO) and Department of Chemistry, Aarhus University, Langelandsgade 140, DK-8000 Aarhus C, Denmark. Fax: +45 8619 6199. E-mail: ncn@ inano.dk (N.C.N.); [email protected] (A.M.). † Center for Insoluble Protein Structures (inSPIN). ‡ Interdisciplinary Nanoscience Center (iNANO). § Department of Chemistry. | Current address: Monash Institute of Pharmaceutical Sciences, Monash University, 399 Royal Parade, Parkville, VIC, 3052, Australia. ⊥ Department of Molecular Biology.

peptides.14–16 It has been observed that CR inhibits the neurotoxic effects of Aβ on hippocampal neurones from embryonic rats,17 as well as reduces the neurotoxicity of preformed fibril aggregates.18 Furthermore, CR has been proposed to be able to facilitate clearance of Aβ from the brain, by interfering with the process in which fibrillar Aβ inactivate proteases degrading soluble Aβ.19 Many analogues of CR (e.g., Curcumin, Chrysamine G, FSB) have been found to inhibit Aβ aggregation as well,14,20,21 emphasizing the importance of understanding the interaction pattern of CR in the scope of rationally designing new fibrilinhibitory compounds. The details of the underlying inhibitory mechanisms of CR are still largely unknown. One study has shown CR to inhibit aggregation by stabilizing monomers,22 and in a recent study, the authors propose that CR prevents the formation of lowweight oligomers, but rather arrests Aβ as globular structures, which upon further aggregation produce either fibrillar or globular aggregates.23 In solution, CR has been observed to self-aggregate,24,25 a property shared with other amyloid inhibiting molecules, and it has been suggested that this selfaggregation plays an important role in blocking fibril formation.26 Very recently, it was furthermore proposed that CR induces detergent-like interactions with Aβ.27 In this study, we use a combination of liquid-state NMR, diffusion NMR, and dynamic light scattering to address in more detail the interactions of monomeric Aβ1-40 in solution and CR. Our analysis of the CR:Aβ1-40 interactions shows for the first time that the formation of small complexes/oligomers occurs by a two-step mechanism involving an initial tight 1:1 CR: Aβ1-40 binding followed by a second weaker interaction. We suggest that the existence of two different complexes might explain the contradictory reports regarding the inhibitory effects of CR on the fibril-formation process.

10.1021/jp108035y  2010 American Chemical Society Published on Web 11/15/2010

16004

J. Phys. Chem. B, Vol. 114, No. 48, 2010

Figure 1. Titration of CR into Aβ1-40 followed by 1D 1H liquid-state NMR. (A) Molecular structure of CR. (B) 1H NMR spectra showing the changes in the Aβ1-40 methyl region, when CR is titrated into 70 µM Aβ1-40 in 50 mM Na-phosphate buffer at pH 7.4 containing 10% D2O. Aβ1-40 signals before addition of ligand are shown in red, and as the ligand concentration increase, spectra are colored in increasing shades of blue. Spectra are shown for ligand concentrations of 0, 50, 100, 150, 200, 400, 600, 800, and 1000 µM. (C) Amide and aromatic region of 1H NMR spectra of CR in buffer (top), and in the presence of Aβ1-40 (middle), as compared to the Aβ1-40 solution before addition of CR (bottom). The ligand concentration is 200 µM, and the Aβ1-40 concentration is 70 µM. The spectra were recorded at 400 MHz at a temperature of 37 °C.

2. Experimental Methods Sample Preparation. Aβ1-40 was synthesized by Biopeptides Co. Inc. (San Diego, CA) and used with no further purification. Samples were prepared by dissolving Aβ1-40 in 10 mM NaOH (0.6 mg/mL) and incubating at room temperature for 30 min. The incubated samples were ultracentrifuged at 4 °C for 40 min at 66 000 × g to remove high molecular aggregates. D2O and 1 M phosphate buffer (pH 7.1) were added to yield a final concentration of 70 µM Aβ in 50 mM phosphate buffer with 10% D2O, and pH was adjusted to 7.4 by addition of HCl. The concentration of Aβ was checked by amino acid analysis as described previously28 and a modified Lowry method29 from Thermo Scientific (Rockford, IL). For chemical shift and intensity reference, sodium 2,2-dimethyl-2-silapentane-5-sulfonate (DSS) was added to a concentration of 25 µM (no linebroadening of DSS was observed). 10 mM CR stock solutions were prepared fresh before each experiment in 50 mM phosphate buffer (pH 7.4) with 10% D2O and 25 µM DSS. The concentrations were checked by absorption using molar extinction coefficients of 45 000 at 498 nm. NMR Spectroscopy. All NMR experiments, except for the 1 15 H, N HSQC spectra, were carried out on a Bruker Avance

Pedersen et al. 400 MHz (9.4 T) spectrometer (Bruker Biospin, Rheinstetten, Germany) equipped with a standard 5 mm triple-resonance probe with a z-gradient. The sample temperature was 37 °C. 1H 1D spectra were recorded with the carrier centered in the resonance of water protons, using 144 scans, and suppressing the water signal by WATERGATE.30 Saturation transfer difference (STD) spectra31 were recorded using a 1:10 Aβ1-40:CR ratio. In all experiments, the frequency of the on-resonance pulse (same excitation pulse in all experiments) was applied at the resonances of the Aβ1-40 methyl groups (180-390 Hz, corresponding to 1-0.45 ppm), and the off-resonance frequency was chosen to be -16 000 Hz. The saturation time was 4 s. On- and off-resonance spectra were processed and subtracted to obtain the STD spectrum. A stimulated-echo sequence with pulsed field gradients was used for diffusion measurements.32 These pulsed-field-gradient (PFG) NMR experiments were carried out by acquiring 16 1H 1D spectra with 40 scans with stepwise increasing gradients from 0.67 to 32 G cm-1. The signal decays for protein and ligand were obtained by integrating spectral regions with none or negligible overlap: 1.1-0.4 ppm for the methyl peaks of Aβ1-40 and 8.6-8.4 ppm for CR signals. Diffusion coefficients were obtained by fitting signal decays to a Gaussian function using GOSA-fit version 3 (BIO-LOG Scientific Sofware, Toulouse, France). Dioxane was added to all samples as an internal reference for determination of the hydrodynamic (Stoke’s) radius (Rh), by using Rh,Aβ ) Rh,dioxane(Dh,dioxane/Dh,Aβ) and Rh,dioxane ) 2.12 Å, as described by Wilkins et al.33 Neither Aβ1-40 nor CR signals were observed to be perturbed by dioxane, indicating that no interactions are taking place, nor were any significant changes in diffusion coefficient observed with or without this reference compound. 1 15 H, N HSQC spectra34 were obtained on a Bruker AvanceII 800 MHz spectrometer (18.8 T) equipped with a 5 mm tripleresonance TCI cryogenic probe. The spectra were acquired at 25 °C with eight transients using 128 increments and spectral widths of 13.96 and 40 ppm in the direct and indirect dimension, respectively. Two-dimensional (2D) spectra were processed using Topspin 1.3 (Bruker Biospin, Rheinstetten, Germany), and the processed spectra were exported for further analysis in Sparky.35 Assignment of the 1H-15N HSQC spectrum was established on the basis of the assignment at 3 °C by Wahlstro¨m et al.,7 by increasing the temperature from 3 to 25 °C and recording spectra every 5 °C. Chemical shift changes (∆CS) from the individual peaks are reported as

∆CS ) √(∆CS1H)2 + (∆CS15N/6.51)2

(1)

Dynamic Light Scattering. Dynamic light scattering (DLS) was performed on an ALV system (ALV GmbH, Langen, Germany) operating at a wavelength of 633 nm, and the scattered light was detected at 90°. All measurements were carried out at 37 °C; additionally, CR experiments were performed at 25 °C. Prior to measurements, all samples were centrifuged for 30 min at 14 000 rpm in an Eppendorf centrifuge to remove dust particles. The data was analyzed in order to obtain Rh, using the regularization algorithm available in the ALV correlator 3.0 software. Data Analysis. Principal component analysis (PCA) was conducted on the chemical shifts of a set of 10 spectral features in the spectra of Aβ1-40 in the presence of increasing levels of CR. PCA was performed using the Simca-P 12.0 software (Umetrics, Umeå, Sweden).

Two-State Association between Amyloid β and Congo Red

J. Phys. Chem. B, Vol. 114, No. 48, 2010 16005

Figure 2. Titration of CR into Aβ1-40 followed by 1H,15N HSQC NMR. (A) 1H,15N HSQC NMR spectra of an Aβ1-40 monomer solution (70 µM Aβ1-40 in 50 mM Na-phosphate buffer at pH 7.4, containing 10% D2O) at increasing CR concentrations: 0 (blue), 25 (green), 100 (yellow), 200 (red), and 400 (purple) µM. (B) Line broadening measured as the ratio between peak heights for the amide signals upon addition of increasing amounts of CR (25 µM, green squares; 100 µM, yellow triangles; 400 µM, purple spheres). (C) Chemical shift changes calculated using eq 1. CR concentrations are 25 (green squares) and 100 µM (yellow triangles). The 1H,15N HSQC spectra were recorded at 800 MHz at a temperature of 25 °C.

To describe complex formation of CR and Aβ, 1D 1H NMR data was fitted to a two-step binding model as described by Chanvorachote et al.36 Each binding is represented by a dissociation constant Kdi (i ) 1,2 for the two steps):

Kdi )

[Aβi] × [CR]free [AβiCR]

(2)

where [Aβi] ) ni[Aβ], ni is the number of binding sites on the peptide, and [AβiCR] is the concentration of the formed complex. This expression can be rearranged and represented in two different ways, either in terms of the amount of free CR or the total concentration of Aβ:

[AβiCR] )

ni × [Aβ] × [CR]free Kdi + [CR]free

(3)

or in terms of the total amount of CR: [AβiCR] )

{

1 K + ni × [Aβ] + [CR]total ( 2 di

√(Kdi + ni × [Aβ] + [CR]total)2 - 4 × [CR]total × ni × [Aβ]}

(4)

Simulations were carried out using GOSA-fit version 3 (BIOLOG Scientific Sofware, Toulouse, France). 3. Results and Discussion To investigate the detailed interaction mechanism between the amyloid-formation inhibitor CR and Aβ1-40, we have used solution NMR complemented by dynamic light scattering (DLS) to monitor structural and environmental changes of both Aβ1-40 and CR molecules. Monomeric solutions of Aβ1-40 were prepared at pH 11 using NaOH as described by Petkova et al.37 The samples were ultracentrifuged, and immediately before the experiments, the pH was lowered to the physiological range (pH 7.4). 1D 1H NMR spectra showed only a small loss in signal intensity upon

lowering the pH (see Figure S1 in the Supporting Information), indicating that Aβ1-40 does not precipitate or form insoluble fibrils. A very small signal decrease after a week at 37 °C documents an insignificant rate of aggregation and/or fibril formation within the time frame of our measurements. The slow aggregation was seen repetitively and was most likely due to the lack of seeds in the sample in combination with the quiescent conditions. When a similar sample was incubated under agitation, fibrils formed in a couple of days (data not shown). Amyloid β and Congo Red NMR Signals Are Mutually Broadened. Interactions between small ligands and peptides may be monitored in different ways using liquid-state NMR. The simplest approach involves titration of ligands into a peptide solution monitored by 1D NMR spectra. This allows analysis of spectral changes of both peptide and ligand signals, typically including chemical shift changes caused by altered chemical surroundings upon noncovalent binding, and broadening of the signals due to either chemical exchange between different bound and/or unbound species or the formation of larger complexes with inherently broader lines. No NMR signals are visible for complexes larger than ∼200 kDa, but exchange with such a complex may cause line broadening of an NMR-visible state. Figure 1 illustrates how the CR and Aβ1-40 signals are affected by the CR:Aβ1-40 interaction. Figure 1B show the 1H NMR spectral changes for the methyl groups in Aβ1-40 as CR is added to a 70 µM solution of Aβ1-40, going from 0 (red) to 1 mM (blue) CR. The induced changes for the methyl region are representative for the majority of peptide 1H resonances. From the changes in the Aβ1-40 signals, it is clear that CR affects both chemical shifts and line shape. Figure 1C shows 1H NMR spectra of the aromatic and amide regions for CR in buffer (top), CR with Aβ1-40 (middle), and Aβ1-40 in buffer (bottom). In this spectral region, both the Aβ1-40 and CR signals are all perturbed to a large degree in each other’s presence. Congo Red Binds to the Fibril Forming Stretches of Amyloid β. Insight into the structural aspects of the binding of CR to Aβ1-40 monomers may be obtained from 1H-15N HSQC spectra of uniformly 15N-labeled Aβ1-40, as illustrated in Figure 2 (see Table S2 in the Supporting Information for chemical shift assignment). The HSQC spectrum (Figure 2A) reinforces our

16006

J. Phys. Chem. B, Vol. 114, No. 48, 2010

Pedersen et al.

TABLE 1: Diffusion Coefficients and Hydrodynamic Radii (Rh) diffusion coefficient · 10-10 (m2/s)a Diffusion of Aβ1-40 NaOH 2.33 ( 0.14 Aβ in buffer (0 µM CR) 2.75 ( 0.09 Aβ (20 µM CR) 2.73 ( 0.08 Aβ (100 µM CR) 2.40 ( 0.09 Aβ (500 µM CR) 2.66 ( 0.14 Aβ (1 mM CR) 2.69 ( 0.19

Rh (Å)

b,c

Diffusion of Ligandsc,d CR in DMSO 3.47 ( 0.10 CR in buffer 2.98 ( 0.16 CR in buffer (25 °C) 1.64 ( 0.21 CR + Aβ (100 µM CR) CR signals too low for determination of D CR + Aβ (500 µM CR) 2.73 ( 0.12 CR + Aβ (1 mM CR) 2.94 ( 0.13

12.9 ( 0.8 13.2 ( 0.4 12.8 ( 0.4 14.1 ( 0.5 13.3 ( 0.7 14.2 ( 1.0 6.8 ( 0.2 10.6 ( 0.6 14.2 ( 1.8 12.2 ( 0.5 12.4 ( 0.5

a If nothing else is stated, the diffusion coefficient was determined at 37 °C. b The signals from the methyl groups are used for determination of the diffusion coefficient. c Dioxane was used to determine Rh. d CR signals from PFG NMR were integrated to give the diffusion coefficient.

previous observations that CR induces changes in peak positions and extensive line broadening for Aβ1-40. Figure 2 provides residue-specific information about changes of Aβ1-40 HSQC peaks in the presence of CR. Information on signal loss due to line broadening, measured as the ratio between the peak height (I) and the initial peak height (I0) (Figure 2B), is shown along with changes in chemical shifts (Figure 2C). It is evident from both chemical shift changes and peak heights that CR influences all residues of the Aβ1-40 peptide, but the major perturbations occur for residues 14-26 and 31-37. These two regions correspond roughly to the hydrophobic parts of Aβ1-40 (residues 10-22 and 30-40) that are known to reside in the core of mature fibrils.38 Less pronounced perturbations are observed for resonances in the free C-terminus, indicating limited interactions in this part of the peptide, possibly due to repellence between the negative charges in the C-terminal carboxyl acid and the substituents on the aromatic end groups of CR. We note, however, that the chemical shift changes are too small to be a signature of significant changes in the secondary structures, such as transition from R-helical to fibril-like β-sheet structures.39 It should be noted that HSQC data was obtained at 25 °C to avoid rapid exchange of the amide protons. We observed that the CR signals decreased drastically in intensity as the temperature was decreased, and that the remaining signals were broadened. This supports the interpretation that line broadening of CR is caused by exchange processes, which are slowed down at the lower temperature. No fundamental difference in the size distribution of CR and CR aggregates at 37 °C vs 25 °C was observed by dynamic light scattering (data not shown), whereas some changes in NMR diffusion properties were observed (Table 1), indicating that the apparent size of CR has increased slightly, which also might be explained in terms of exchange processes between CR aggregates of different size. Small Oligomeric Complexes Are Formed upon Adding CR to Amyloid β. We applied saturation transfer difference (STD) NMR31,40 to probe for formation of oligomeric complexes of CR with Aβ1-40. STD NMR is based on the fact that the NOE effect is much stronger in larger molecules with large rotational correlation times as compared to smaller molecules. When a large protein is selectively saturated, the magnetization

Figure 3. Saturation transfer difference (STD) NMR experiments. Alternating normal 1D NMR spectra (1D ref) and STD spectra. From the top: reference spectra of CR and Aβ1-40 (full line) and CR in buffer (dotted line), STD spectrum of CR and Aβ1-40, reference spectrum of Aβ1-40 in buffer, and STD spectrum of Aβ1-40 in buffer. CR and Aβ1-40 concentrations were 1 mM and 70 µM, respectively. For all STD spectra, the saturation frequency was set at the maximum of the methyl signals of Aβ1-40 (0.85 ppm) and the saturation time was 4 s. The spectra were recorded at 400 MHz at a temperature of 37 °C.

will spread out to all surrounding spins through spin diffusion processes. If a ligand binds reversibly, it will be saturated too due to cross-relaxation with the protein. When this ligand is released, it will bring the transferred magnetization into the soluble ligand fraction. Subtracting spectra with and without selective saturation of the protein prior to a simple 90° excitation pulse gives the STD spectrum, which shows only the signals from ligands that have been bound to the saturated protein. This experiment is efficient if the Kd is in the µM range or larger, or for stronger binders if the formed complex is of detectable size while sufficiently large to induce STD spin diffusion effects. Figure 3 clearly illustrates that addition of a 10 times surplus of CR to the Aβ1-40 solution causes ligand signals to appear in a STD spectrum in which Aβ1-40 is present. Interestingly, we also see signals from Aβ1-40 (Figure 3, marked with asterisk), which do not appear in a sample containing Aβ1-40 only (Figure 3, bottom), showing that complexation alters the spin diffusion properties of Aβ1-40. The lower size limit for the large species in STD has been reported to be approximately 10 kDa.40 Thus, these results imply that in both cases peptide ligand complexes containing more than just one peptide and one ligand molecule are formed, which is verified by the fact that Aβ1-40 alone does not give rise to STD signals. To further characterize the CR-Aβ1-40, complexes, we used dynamic light scattering (DLS) and pulsed-field-gradient (PFG) spin-echo NMR to determine diffusion properties and hydrodynamic radii (Rh) of the constituents in the pure peptide, pure ligand, and mixed peptide-ligand samples. DLS analysis showed that the monomeric solution of Aβ1-40 is inhomogeneous, even after ultracentrifugation, but the major part of the peptide is in the monomeric state, and only a very limited fraction forms larger aggregates (Figure 4). CR in solution behaves very similarly to Aβ1-40. Again, DLS reveals

Two-State Association between Amyloid β and Congo Red

Figure 4. Particle size distributions obtained for solutions of Aβ1-40 and CR using dynamic light scattering at 37 °C. The left panels present the “unweighted” (i.e., intensity-weighted) distributions, and the right panels present the number size distributions. (A) Aβ1-40, (B) CR, and (C) Aβ1-40 and CR in 1:1 (solid line) and 1:10 (dashed line) mixtures. The samples contained 70 µM Aβ1-40 in 50 mM Na-phosphate buffer at pH 7.4.

Figure 5. Diffusion NMR experiments of Aβ1-40 and CR. Representative diffusion curves from PFG-NMR experiments: Signal intensities from 70 µM pure Aβ1-40 (squares), 70 µM Aβ1-40 with 1 mM CR (open circles), and 1 mM pure CR (crosses) are seen along with a representative fitted curve (solid line). Signal intensities from dioxane are labeled with the same marks as the Aβ1-40 signals from the corresponding sample, and a representative fitted line is shown as dashes. The spectra were recorded at 37 °C at 400 MHz.

a combination of 8-10 Å particles and a minor fraction of large aggregates, as has also been reported previously.25 By PFGNMR, CR alone is observed to have an apparent Rh value of 10.6 Å, which is larger than what would be expected on the basis of the monomeric size of 6.8 Å in dimethyl sulfoxide (DMSO) (Table 1). In regard to the extended line broadening of CR, we ascribe this larger apparent Rh value to the aggregation properties of CR.24 When CR and Aβ1-40 are mixed, the pattern of small particles accompanied by a subfraction of huge aggregates is maintained. In regard to size changes, the PFG-NMR curves extracted from Aβ1-40 signals are very similar regardless of the presence of CR (Figure 5). Around an equimolar ratio, however, we observed by DLS that the average size of the small species increased slightly as compared to both CR and Aβ1-40 alone, thus suggesting that this complex formed

J. Phys. Chem. B, Vol. 114, No. 48, 2010 16007

Figure 6. Aβ1-40-CR complex formation. (A) Quantitative analysis of Aβ1-40 1H NMR signals. Integrals of methyl signals of 70 µM Aβ1-40 as a function of ligand/protein ratio after addition of CR. The intensity has been normalized to 1 before addition of ligand. (B) Change in Aβ1-40 chemical shifts expressed as the dominant principal component from a PCA of the variations in chemical shifts for the Aβ1-40 spectra as a function of free CR concentration (symbols). The line shows a simulated binding curve of formation of the second complex (Kd2 ) 300 µM, n2 ) 1). (C) Amount of free CR estimated from the CR signal vs the total amount of CR. (D) Amount of CR bound to Aβ1-40 estimated from the difference between free and total CR vs the amount of free CR (symbols). The line shows a simulated two-step binding curve (Kd1 ) 5 µM, Kd2 ) 300 µM, n1 ) n2 ) 1).

in the first step is indeed larger than the peptide alone. As a surplus of CR is added, no significant size changes are observed for Aβ1-40. The formation of small oligomeric complexes has also been suggested in a recent study on R-synuclein interacting with the self-aggregating inhibitors CR and Lacmoid,25 and the parallels between the systems indicate that this interaction mode may be general for fibril-forming peptides or proteins. Indeed, the proposed mechanism, including exchange between different peptide-ligand complexes, supports the proposition by Feng et al.26 that it is actually the presence of small molecule aggregates executing the inhibitory action of the small molecules. Aβ1-40 and CR Form Complexes by a Two-State Mechanism. Integrals of the Aβ1-40 methyl signals in the presence of CR show that low CR concentrations give rise to major losses of intensity (Figure 6A). At higher concentrations, no further intensity losses are seen but rather a small increase in intensity. In fact, detailed investigation of the 1D 1H NMR spectra reveals that the interaction between Aβ1-40 and CR takes place in two steps. We have estimated dissociation constants for the two involved complexes from the chemical shift changes observed for Aβ1-40 for increasing amounts of CR (Figure 6B), the amount of free CR (Figure 6C), the amount of bound CR (Figure 6D), and from new Aβ1-40 signals emerging as a signature of the first CR-Aβ1-40 complex (Figure 7). The concentration of free CR in the presence of Aβ1-40 has been estimated from the intensities of the CR signals when titrated into Aβ1-40 and into pure buffer (Figure 6C). This value is subtracted from the total concentration to obtain the concentration of CR bound to Aβ1-40 (Figure 6D). It should be noted that, when CR is titrated into an aqueous solution, the signals are very broad (Figure 1C), most likely due to exchange

16008

J. Phys. Chem. B, Vol. 114, No. 48, 2010

Figure 7. New spectral components that appear upon CR addition. (A) Aβ1-40 signals observed prior to addition of ligand are shown in red, and as the ligand concentration increases, spectra are colored in increasing shades of blue. Spectra are shown for ligand concentrations of 0, 25, 50, 75, 100, 200, 400, 600, 800, and 1000 µM. (B) Aβ1-40 signal at ligand concentrations of 0 (red), 400 (purple), and 1000 (blue) µM. (C) Intensity of the new spectral component at 3.35 ppm, curves simulate infinitely strong binding of n1 ) 0.5 (red), 1 (green), and 2 (blue). (D) Intensity of the new spectral component at 3.35 ppm fitted to eq 4 with n1 ) 1 and Kd1 values of 0.1 (red), 1 (green), 10 (blue), and 100 (orange) µM. Triangles and squares represent signal intensities from two independent titration experiments. The Aβ1-40 concentration is 70 µM in 50 mM phosphate buffer, pH 7.4.

processes caused by the presence of CR aggregates, as has been reported by Skowronek et al.24 However, since the signal intensity is proportional to the CR concentration (see Figure S3 in the Supporting Information), it appears that the relative populations of the different CR species are unaltered within this concentration regime. An analysis like this has not been made before, as previous experiments used to study CR interactions with Aβ1-40,27 as well as R-Synuclein,25 have been carried out at 25 °C, where severe line broadening renders the CR signals practically invisible. Up to equimolar concentrations, both Aβ1-40 and CR signals are broadened (Figure 6A) by the formation of a CR-Aβ1-40 complex. However, this line broadening might be attributed either to conformational exchange or transient interactions between complexes on the intermediate chemical shift time scale, and/or to formation of large-sized complexes with longer rotational correlation times. While most Aβ1-40 signals decrease, a few new signals appear, e.g., the signal at 3.35 ppm in Figure 7A, which possibly arise from side chain interactions with CR. This suggests that a new complex with NMR visible size is formed between Aβ and CR. Judging from the DLS and diffusion NMR measurements, the decreased intensities are caused mainly by conformational exchange within this complex or transient interactions between the complexes rather than a drastic size increase. A quantitative measure on the formation of this CR-Aβ1-40 complex may be obtained by plotting the intensity of the new signal as a function of the ligand concentration (Figure 7C and D). The binding stoichiometry is assessed by plotting theoretical curves describing infinitely strong CR:Aβ1-40 binding reactions of different stoichiometry (Figure 7C). From this analysis, we

Pedersen et al. deduce that the complex has a stoichiometry of approximately one CR molecule binding per Aβ1-40 peptide (n1 ) 1). The strength of the interaction is determined by comparison with theoretical curves describing binding with Kd1 values of 0.1, 1, 10, and 100 µM (Figure 7D) using eq 4. The binding curve agrees well with a Kd1 value between 1 and 10 µM for a 1:1 stoichiometry, corresponding to a relatively strong binding. It should be noted that it was possible to make reasonable fits using n1 values between 0.7 and 1, all resulting in Kd1 values in the same range. A binding stoichiometry of 1:1 has earlier been reported for the binding of CR to R-synuclein,25 as well as binding of the CR analogue Curcumin to an R-helical prion protein intermediate,41 suggesting that formation of roughly equimolar complexes is a typical feature of this class of inhibitors. As the CR concentration is further increased, no more intensity losses are observed and the new peaks saturate. However, the amount of bound CR (Figure 6D), as well as the chemical shift changes of Aβ1-40 (extracted by PCA analysis;42 Figure 6B), do not saturate. There is thus a second binding reaction in which the gradual changes in chemical shifts correspond to a shift in the equilibrium between two different CR-Aβ1-40 complexes in fast-to-intermediate exchange. In comparison to the complex first formed, this state has less conformational exchange within the complex, as judged by the increase in intensity for the Aβ1-40 signals (Figure 6A). The formation of the second complex is confirmed by the appearance of new signals, e.g., at 0.2 ppm at the end of the titration, as illustrated in Figure 7B. Interestingly, this signal is close to the resonance at 0 ppm observed for a 100 kDa oligomer of Aβ1-40,43 which might indicate similar peptide structures in this second 2:1 CR:Aβ1-40 complex and the pure Aβ1-40 oligomer. A more general cause of the upfield change in methyl resonances could most likely be induced ring current effects from the aromatic groups on the bound CR. However, according to the PFG NMR measurements, the size of the 2:1 complex is not significantly larger than that of the Aβ1-40 monomer. The increase of Aβ1-40 signal seen at high CR concentrations indicates stabilization of the structure in this complex and/or a lower tendency to interact with other complexes. To obtain an estimate of Kd2, we have simulated binding curves describing the amount of bound CR using a two-state binding reaction assuming that Kd1 ) 5 µM and n1 ) 1. Kd2 has been fitted at different stoichiometries (see S4 in the Supporting Information). The best fits are achieved with n2 of approximately 1-1.25, which strongly indicates a 1:1 stoichiometry also for this interaction. For n2 ) 1, Kd2 was fitted to be 320 ( 120 µM. Further refinement was obtained by exploiting the chemical shift (CS) changes, which reflect the formation of the second complex. When applying eq 3 to the CS changes using n2 ) 1 and Kd2 ) 320 µM, a very good fit is obtained (see Figure 6B), supporting the data above. Conclusively, the estimated Kd values of the two complexes have been used to simulate the concentration of bound CR (CRbound) using the concentration of free CR (CRfree). As seen in Figure 6D, Kd1 and Kd2 values of 5 and 300 µM, respectively, and n1 and n2 of 1 reproduced the binding curve in a satisfactory manner. This second binding mode has not been reported before, neither for Aβ1-40 nor in the analysis of CR interaction with other amyloid-forming peptides. On the contrary, there are reports that the action of CR on Aβ1-40 has no concentration dependency.23 Only by in-depth analysis of simple 1D 1H NMR spectra was it possible to extract a model describing the system, thus emphasizing the benefits of applying NMR in combination

Two-State Association between Amyloid β and Congo Red

J. Phys. Chem. B, Vol. 114, No. 48, 2010 16009

Figure 8. Model of the Aβ1-40-CR interaction based on NMR and DLS data. (1) Upon addition of low to equimolar concentrations of CR as compared to Aβ1-40, small complexes are formed at a 1:1 stoichiometry with a Kd1 value between 1 and 10 µM. (2) As more CR is added, a second small complex with 1:2 Aβ1-40:CR stoichiometry is formed. (3) Aβ1-40 alone slowly forms aggregates/amyloid fibrils. (4) CR alone is found as a small oligomer, most likely in exchange with larger aggregates.

with PCA to obtain additional, otherwise inaccessible knowledge of the states of the system, as recently shown by Malmendal et al.44 Even though most studies prove CR to be a fibril-formation inhibitor,12,14 or an inhibitor against the formation of certain oligomers without any pronounced effect against fibril formation,10 other studies prove CR to have an aggregation promoting effect.27,45 That, in combination with the lack of specificity of CR,46 makes elucidating the detailed mechanism of its inhibitory action highly nontrivial. In general, CR is known to prolong the lag phase of fibril formation rather than completely inhibiting fibril formation,13 which can be effectuated if the formation of the small oligomers observed here represents an off-pathway in the fibril-formation process. However, Lendel et al.27 recently presented results showing a fibril-promoting effect upon incubation of Aβ1-40 with equimolar amounts of CR ascribed to a detergent-like mechanism. The exchange-broadened lines observed in our study of the initially formed complex support the formation of a complex that transiently interacts with other species. Such interactions could potentially lead to a fibrillationpromoting effect, as also seen when Aβ is incubated with anionic detergent micelles.7 It should be noted that, in the work presented by Lendel et al.,27 the authors did not observe the stepwise formation of two distinct complexes, which is only noticeable upon the detailed analysis of 1D 1H spectra as presented here. Upon formation of the second 2:1 complex, we see decreasing exchange broadening, suggesting a more inert complex. A sample containing a 10-fold excess of CR also proved to be very stable, and even after long time storage (>7 days), no signs of aggregation were observed, as judged by invariant NMR signals. These observations lead us to suggest that the first complex could be a promoter rather than an inhibitor of fibril formation, whereas the second complex might be the actual inhibitor of fibril formation. It should be noted, however, that both complexes might play important roles in reducing the neurotoxicity, caused by protofibrillar species, as seen in ViVo of Aβ1-40.18 4. Conclusions To gain knowledge about the interaction mechanisms between the fibril-formation inhibitor CR and Aβ1-40, we have used a combination of 1D and 2D NMR spectra with DLS and diffusion NMR. On the basis of our data, it is evident that two distinct CR-Aβ1-40 complexes are formed as CR is titrated into Aβ1-40. As outlined in Figure 8, the first complex is formed with a 1:1 stoichiometry and a Kd1 value between 1 and 10 µM (step 1), while the second 2:1 complex is formed by a weaker interaction

with Kd2 ∼ 300 µM (step 2). It should be noted that we observed the major fraction of Aβ1-40 in solution to be monomeric, but a smaller fraction of larger aggregates seemed to be present at all times, and a slow aggregation of the monomers was seen over time (step 3). For the first time, we have shown that CR binds to Aβ1-40 by a two-state mechanism, and we suggest that the two resulting complexes have different effects upon fibril formation, possibly with equimolar concentrations of CR promoting aggregation, while a large surplus act inhibitory. Acknowledgment. This research has been supported by the Danish National Research Foundation. We acknowledge use of the 800 MHz NMR instrument at the Danish Center for NMR Spectroscopy of Biological Macromolecules at the Carlsberg Laboratory, Copenhagen, Denmark. Supporting Information Available: Further experimental documentation in the form of spectra, tables containing chemical shift assignments, calibration curves, and simulation of binding steps. This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) Goedert, M.; Spillantini, M. G. Science 2006, 314, 777. (2) Tycko, R. Methods Enzymol. 2006, 413, 103. (3) Selkoe, D. J. Nature 2003, 426, 900. (4) Burgevin, M. C.; Daniel, N.; Passat, M.; Messence, K.; Bertrand, P.; Doble, A.; Blanchard, J. C. J. Neurochem. 1994, 63, S74. (5) Eraser, P. E.; Levesque, L.; McLachlan, D. R. J. Neurochem. 1994, 62, 1227. (6) Walsh, D. M.; Selkoe, D. J. J. Neurochem. 2007, 101, 1172. (7) Wahlstro¨m, A.; Hugonin, L.; Peralvarez-Marin, A.; Jarvet, J.; Gra¨slund, A. FEBS J. 2008, 275, 5117. (8) Lau, T. L.; Ambroggio, E. E.; Tew, D. J.; Cappai, R.; Masters, C. L.; Fidelio, G. D.; Barnham, K. J.; Separovic, F. J. Mol. Biol. 2006, 356, 759. (9) Reinke, A. A.; Gestwicki, J. E. Chem. Biol. Drug Des. 2007, 70, 206. (10) Necula, M.; Kayed, R.; Milton, S.; Glabe, C. G. J. Biol. Chem. 2007, 282, 10311. (11) Frid, P.; Anisimov, S. V.; Popovic, N. Brain Res. ReV. 2007, 53, 135. (12) Lorenzo, A.; Yankner, B. A. Neurobiol. Alzheimer’s Dis. 1996, 777, 89. (13) Bartolini, M.; Bertucci, C.; Bolognesi, M. L.; Cavalli, A.; Melchiorre, C.; Andrisano, V. ChemBioChem 2007, 8, 2152. (14) Masuda, M.; Suzuki, N.; Taniguchi, S.; Oikawa, T.; Nonaka, T.; Iwatsubo, T.; Hisanaga, S.; Goedert, M.; Hasegawa, M. Biochemistry 2006, 45, 6085. (15) Caughey, B.; Ernst, D.; Race, R. E. J. Virol. 1993, 67, 6270.

16010

J. Phys. Chem. B, Vol. 114, No. 48, 2010

(16) Heiser, V.; Scherzinger, E.; Boeddrich, A.; Nordhoff, E.; Lurz, R.; Schugardt, N.; Lehrach, H.; Wanker, E. E. Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 6739. (17) Burgevin, M. C.; Passat, M.; Daniel, N.; Capet, M.; Doble, A. NeuroReport 1994, 5, 2429. (18) Lorenzo, A.; Yankner, B. A. Proc. Natl. Acad. Sci. U.S.A. 1994, 91, 12243. (19) Chander, H.; Chauhan, A.; Chauhan, V. J. Alzheimer’s Dis. 2007, 12, 261. (20) Yang, F. S.; Lim, G. P.; Begum, A. N.; Ubeda, O. J.; Simmons, M. R.; Ambegaokar, S. S.; Chen, P. P.; Kayed, R.; Glabe, C. G.; Frautschy, S. A.; Cole, G. M. J. Biol. Chem. 2005, 280, 5892. (21) Ishii, K.; Klunk, W. E.; Arawaka, S.; Debnath, M. L.; Furiya, Y.; Sahara, N.; Shoji, S.; Tamaoka, A.; Pettegrew, J. W.; Mori, H. Neurosci. Lett. 2002, 333, 5. (22) Podlisny, M. B.; Walsh, D. M.; Amarante, P.; Ostaszewski, B. L.; Stimson, E. R.; Maggio, J. E.; Teplow, D. B.; Selkoe, D. J. Biochemistry 1998, 37, 3602. (23) Bose, P. P.; Chatterjee, U.; Xie, L.; Johansson, J.; Go¨thelid, E.; Arvidsson, P. I. ACS Chem. Neurosci. 2010, 1, 315. (24) Skowronek, M.; Stopa, B.; Konieczny, L.; Rybarska, J.; Piekarska, B.; Szneler, E.; Bakalarski, G.; Roterman, I. Biopolymers 1998, 46, 267. (25) Lendel, C.; Bertoncini, C. W.; Cremades, N.; Waudby, C. A.; Vendruscolo, M.; Dobson, C. M.; Schenk, D.; Christodoulou, J.; Toth, G. Biochemistry 2009, 48, 8322. (26) Feng, B. Y.; Toyama, B. H.; Wille, H.; Colby, D. W.; Collins, S. R.; May, B. C. H.; Prusiner, S. B.; Weissman, J.; Shoichet, B. K. Nat. Chem. Biol. 2008, 4, 197. (27) Lendel, C.; Bolognesi, B.; Wahlstro¨m, A.; Dobson, C. M.; Gra¨slund, A. Biochemistry 2010, 49, 1358. (28) Ozols, J. Methods Enzymol. 1990, 182, 587.

Pedersen et al. (29) Lowry, O. H.; Rosebrouhg, N. J.; Farr, A. L.; Randall, R. J. J. Biol. Chem. 1951, 193, 265. (30) Liu, M. L.; Mao, X. A.; Ye, C. H.; Huang, H.; Nicholson, J. K.; Lindon, J. C. J. Magn. Reson. 1998, 132, 125. (31) Mayer, M.; Meyer, B. Angew. Chem., Int. Ed. 1999, 38, 1784. (32) Stejskal, E. O.; Tanner, J. E. J. Chem. Phys. 1965, 42, 288. (33) Wilkins, D. K.; Grimshaw, S. B.; Receveur, V.; Dobson, C. M.; Jones, J. A.; Smith, L. J. Biochemistry 1999, 38, 16424. (34) Bodenhausen, G.; Ruben, D. J. Chem. Phys. Lett. 1980, 69, 185. (35) Goddard, T. D.; Kneller, D. G. SPARKY 3; University of California: San Francisco, CA, 2008. (36) Chanvorachote, B.; Nimmannit, U.; Muangsiri, W.; Kirsch, L. J. Fluoresc. 2009, 19, 747. (37) Petkova, A. T.; Leapman, R. D.; Guo, Z. H.; Yau, W. M.; Mattson, M. P.; Tycko, R. Science 2005, 307, 262. (38) Petkova, A. T.; Ishii, Y.; Balbach, J. J.; Antzutkin, O. N.; Leapman, R. D.; Delaglio, F.; Tycko, R. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 16742. (39) Sibley, A. B.; Cosman, M.; Krishnan, V. V. Biophys. J. 2003, 84, 1223. (40) Meyer, B.; Peters, T. Angew. Chem., Int. Ed. 2003, 42, 864. (41) Hafner-Bratkovic, I.; Gaspersic, J.; Smid, L. M.; Bresjanac, M.; Jerala, R. J. Neurochem. 2008, 104, 1553. (42) Spearman, C. Am. J. Psych. 1904, 15, 201. (43) Narayanan, S.; Reif, B. Biochemistry 2005, 44, 1444. (44) Malmendal, A.; Underhaug, J.; Otzen, D. E.; Nielsen, N. C. PLoS One 2010, 5, e10262. (45) Kim, Y. S.; Randolph, T. W.; Manning, M. C.; Stevens, F. J.; Carpenter, J. F. J. Biol. Chem. 2003, 278, 10842. (46) Khurana, R.; Uversky, V. N.; Nielsen, L.; Fink, A. L. J. Biol. Chem. 2001, 276, 22715.

JP108035Y