Controlling (cid:1) -Amyloid Oligomerization by the Use of Naphthalene Sulfonates TRAPPING LOW MOLECULAR WEIGHT OLIGOMERIC SPECIES *

Aggregation of proteins and peptides has been shown to be responsible for several diseases known as amyloidoses, which include Alzheimer disease (AD), prion diseases, among several oth-ers. AD is a neurodegenerative disorder caused primarily by the aggregation of (cid:1) -amyloid peptide (A (cid:1) ). Here we describe the stabi-lizationofsmalloligomersofA (cid:1) bytheuseofsulfonatedhydropho-bic molecules such as AMNS (1-amino-5-naphthalene sulfonate); 1,8-ANS (1-anilinonaphthalene-8-sulfonate) and bis-ANS (4,4 (cid:1) -di-anilino-1,1 (cid:1) -binaphthyl-5,5 (cid:1) -disulfonate). The experiments were performed with either A (cid:1) -1–42 or with A (cid:1) -13–23, a shorter version

The aggregation of proteins and peptides in vivo plays a fundamental role in the onset of several human pathologies known as amyloid diseases (1,2). Despite the lack of significant sequence homology among the proteins involved in these diseases, amyloid fibrils seem to share a common structural motif, namely the cross-␤-pleated sheet (3)(4)(5)(6). The mechanisms responsible for fibril formation have been extensively investigated but they are still poorly understood, which makes it difficult to develop new drugs against these diseases and even to select suitable targets (4). Recent studies indicate that early aggregates and fibrils have a great contribution of hydrophobic interactions, whereas late amyloid fibrils are more stabilized by hydrogen bonds (7,8).
Alzheimer disease (AD) 3 is one of the most common forms of senile dementia and is believed to be caused by aggregation of ␤-amyloid peptide (A␤). A␤ is a peptide of 39 -43 amino acid that is derived from a larger, type I transmembrane protein called amyloid precursor protein (APP). During APP processing by proteases called ␤and ␥-secretases, this precursor protein generates A␤, which includes a segment of the APP protein that is embedded in the plasma membrane (9). Thus, A␤ is highly apolar and undergoes aggregation rapidly when dissolved in an aqueous environment, especially the larger peptide with 42-43 amino acids. A␤-1-40 and A␤-1-42 are the most common peptides found in amyloid plaques (10). A␤-1-42 undergoes aggregation in vitro within a few minutes (11), while the A␤-1-40 needs several hours or even days for its complete aggregation (12).
Due to the low solubility of A␤ peptides in water, all the structures obtained thus far in liquid media deposited in the Protein Data Bank (National Institutes of Health, Bethesda, MD) were obtained by NMR in apolar solvents (Me 2 SO, trifluorethanol) (13,14) or in a membranemimicking environment such as SDS (15)(16)(17). The description of experimental conditions in which this peptide remains soluble could be crucial for a better understanding of the mechanisms involved in its fibrillogenesis (4). In addition, greater solubility would allow the design of drugs that could trap this soluble conformation, impeding its aggregation and opening new strategies for treatment.
It has been proposed that A␤ aggregation occurs in several steps, including the formation of soluble, low molecular weight (LMW) oligomers (18 -20). The LMW species that are present prior to fibril formation include dimers and tetramers of A␤, which can also be extracted from neuritic and vascular amyloids (21). Stable dimers have been purified from fibril suspensions of the synthetic peptide A␤-1-40 (18,20,22). Also it has been shown that there is a population of soluble oligomers in AD brains but not in normal brains (23) or in cell cultures (24,25).
Much evidence has shown that A␤ has to be in a fibrillar arrangement to be neurotoxic (26,27). However, dimers produced by the synthetic A␤-1-42 have also been shown to be highly neurotoxic in vitro (21) but only in the presence of microglia cells. On the other hand, several other reports have shown that LMW species of A␤ are bioactive either by interacting with synapses (28, 29), which could explain the loss of memory in AD, either by disrupting cognitive function (30) or by deregulating calcium homeostasis (31), which ultimately leads to cell death. These observations led the researchers to investigate further the role of these LMW species in the aggregation pathway of A␤ and in the development of AD (32,33). Several attempts have been made to block fibril formation (34,35) or even to promote fibril solubilization (36,37). However, it is very important to know how and when these LMW species are formed and what is their participation in AD pathogenesis. Nakagami et al. (38) have shown that is possible to inhibit the toxic effect of A␤ to HeLa cells without affecting its fibrillogenesis.
Tjernberg et al. (39) mapped the regions of A␤ that are crucial for fibrillogenesis. They showed that a pentapeptide (KLVFF) comprising the A␤ sequence between residues 16 -20 blocked fibril formation and that this region in A␤-1-40 is crucial for interpeptide interactions involved in fibril formation. Thereafter, Li et al. (40) showed that the residues 17-21 were essential to amyloid core formation. Norstedt and co-workers (39) showed that the region comprising residues 14 -23 of A␤ was still capable of forming amyloid fibrils, and deletions or substitutions in this region resulted in the loss of fibril formation capability (41). For these reasons, besides A␤-1-42, we selected for our studies A␤-13-23, a derivative of A␤ that is still able to form fibrillar aggregates in solution (Fig. 1A). This region includes the hydrophobic core ( 17 LVFF 20 ) flanked by two charged residues, 16 K and 22 ED 23 .
Since hydrophobic and ionic interactions play an important role in A␤ fibril formation, several hydrophobic and/or charged molecules have been used successfully to block A␤ aggregation (35,37,42). Recently, we observed that the aggregation of a peptide derived from the murine prion protein was significantly attenuated by the hydrophobic anionic disulfonate fluorescent probe 4,4Ј-dianilino-1,1Ј-binaphthyl-5,5Ј-disulfonate (bis-ANS) in the micromolar range (43). The dye binds in a hydrophobic pocket, which appears to be the binding site for nucleic acids (44).
In the present study, we investigate the effects of bis-ANS and other compounds of the same family, namely ANS (1-anilino-naphthalene-8-sulfonate) and AMNS (1-amino-5-naphthalene sulfonate; Fig. 1B) on the aggregation of two A␤ peptides, A␤-1-42 and A␤-13-23. We report that this family of compounds was able to abolish the aggregation of these peptides in aqueous solution almost completely, keeping them soluble. By size exclusion chromatography we were able to trap small oligomers that accumulate when aggregation was performed in the presence of bis-ANS. These small oligomers were toxic to RAW (mouse monocyte-macrophage) cell cultures when assayed by MTT (3-(4,5dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) reduction. These results allowed us to perform TOCSY experiments with A␤-13-23 in the presence of AMNS, an effective inhibitor of the aggregation of this peptide. However, the peptide presented an enormous flexibility under these conditions, which prevented us from solving its structure. On the other hand, we were able to solve the structure of A␤-13-23 in the presence of SDS, where it forms an ␣-helix extending from Lys 16 through Phe 20 . The chemical shifts observed in the presence of AMNS and SDS were almost superimposable, suggesting similar structures. We postulate that ANS-related compounds inhibit fibril formation through hydrophobic and electrostatic interactions established between these compounds (rings and sulfonate groups, respectively) and the amino acid residues of the peptide (apolar and charged, respectively).

EXPERIMENTAL PROCEDURES
Material-Synthetic A␤-1-42 was obtained from Dr. Luis Juliano (Universidade Federal do Estado de São Paulo, São Paulo, Brazil). The fragment A␤-13-23 was purchased from Genemed Synthesis (San Francisco, CA). They were lyophilized and stored at Ϫ80°C. D 2 O and SDS-D 25 were purchased from Cambridge Isotope Laboratories (Woburn, MA). Both peptides were subsequently purified to greater than 98% homogeneity by reverse-phase HPLC. The purity of the peptides was verified by electron spray mass spectrometry. All ANS-related compounds were from Molecular Probes (Eugene, OR). They were dissolved in deionized, filtered water and their concentrations measured as described in the catalogue from Molecular Probes. All other reagents were of the highest analytical grade available.
Preparation of Peptide Stock Solutions-For aggregation assays, a stock solution of A␤ was prepared as follow: 1 mg of each peptide was weighted and dissolved with 100 l of 1,1,1,3,3,3-hexafluoro-2-propanol to disassemble preformed aggregates. The samples were then dried in a SpeedVac system. Then, 1 ml of 100% Me 2 SO was added to each sample, and these stock solutions were stored at Ϫ20°C until use. The stock concentrations of the peptides were ϳ250 M for A␤-1-42 and 1000 M for A␤-13-23.
In Vitro Aggregation Measurements-Kinetics of aggregation were monitored in an ISS K2 spectrofluorometer (Champaign, IL) by following the increase in light scattering (LS) using a 1.0 cm quartz cuvette lightening the samples at 320 nm and collecting scattered light at 320 nm. Aggregation of A␤-1-42 and A␤-13-23 was accomplished by mixing an aliquot of the peptides from the Me 2 SO stock solution into aggregation buffer (50 mM sodium phosphate in 100 mM NaCl, pH 7.4) at 37°C. Light scattering by the buffer was used as baseline. Negative controls for aggregation were performed by diluting the peptides in 100% Me 2 SO. The residual Me 2 SO concentration present in the experiments (10% for A␤-1-42 and 2.5% for A␤-13-23) did not interfere with the aggregation assays (data not shown). Aggregation assays in the presence of the ANS derivatives ( Fig. 1) were performed by diluting the peptides into a solution already containing the ANS derivatives. The extent of inhibition caused by ANS derivatives was based on the maximum aggregation attained in the buffer without the probe. Cell Cultures and MTT Assay-RAW cells (mouse monocyte-macrophage, American Type Culture Collection) were routinely cultured in RPMI medium (Invitrogen) supplemented with 10% (v/v) bovine calf serum and 3 mM glutamine, 100 units ml Ϫ1 penicillin and 10 g ml Ϫ1 streptomycin, in a 5% CO 2 humidified atmosphere at 37°C. Cells were used after five passages and were plated at a density of 10,000 cells per well in 100 l of fresh medium on 96-well plates. After 24 h, RPMI medium was exchanged for 100 l of RPMI without phenol red and supplemented with 10% bovine calf serum and antibiotics. Cell-mediated reduction of MTT was used to monitor A␤ toxicity as described by Bucciantini et al. (45). Aggregates of A␤-1-42 and A␤-13-23 were prepared as described above and added to cell cultures to a final concentration of 25 M. For controls, the buffer with or without the probes was added. Reduced MTT was determined by measuring absorbance at 540 nm after 24 h using an automatic microplate reader.
Thioflavin T (ThT) and Congo Red (CR) Binding Assays-CR and ThT binding assays were performed after fibril formation to ensure that the observed increase in light scattering was indeed due to the presence of amyloid fibrils. The amount of CR (15 M) bound was evaluated by measuring the ratio between the absorbance values at 480 and at 540 nm, according to Lai et al. (46). The binding of ThT (10 M) was monitored by the increase in ThT fluorescence at 482 nm, setting excitation at 440 nm.
Size Exclusion Chromatography-Size exclusion chromatography was performed on an HPLC system (Shimadzu) using prepacked TSK G3000sw xl (0.78 ϫ 30 cm) and Synchropak GPC 60 (0.5 ϫ 20 cm) columns. The system was equilibrated with aggregation buffer at 1.0 ml/min. Sample elution was monitored by absorption at 280 nm and by fluorescence emission at 315 nm (excitation at 275 nm, data not shown). The column and buffer were maintained at room temperature. TSK G300sw xl column resolves globular proteins ranging in molecular mass from 5,000 to 150,000 daltons, while Synchropak GPC 60 resolves proteins from 250 to 28,000 daltons.
SDS-PAGE Electrophoresis-The aggregates obtained after 1 h under aggregating conditions (100 l) were dried, and 15 l of SDS-PAGE sample buffer was added. The samples were loaded onto a 12% SDSpolyacrylamide gel (47) without boiling. Low molecular weight markers purchased from Sigma were used as molecular weight standards. The gels were stained with Comassie Brilliant Blue.
NMR Experiments and Structural Calculations-The A␤-13-23 stock was prepared as described above. A 3 mM sample was diluted in 50 mM phosphate buffer, 100 mM NaCl, pH 7.4, in the presence of either D 25 -SDS (35 mM) or AMNS (20 mM). The pH was previously verified and 10% D 2 O was then added for lock. The resulting solutions were immediately centrifuged for 15 min at 13,000 ϫ g, and no pellet was observed. 1 H NMR data were acquired in a Bruker DRX-600 MHz spectrometer employing pulsed field gradient probes. The experiments were performed at 310 or 298 K, when SDS-D 25 or AMNS were present, respectively. NOESY and TOCSY spectra were acquired with 2048 ϫ 512 complex data points and phase-sensitive mode using States-TPPI (time proportional phase incrementation) for quadrature detection in the t 1 dimension. Water signal was suppressed using the watergate sequence (48). Mixing times for NOESY and TOCSY were 150 and 70 ms, respectively. Spectra were processed using NMRPipe (49) and analyzed using NMRView (50). Structure calculations were performed using NOE as distance restraints and 3 J HNH␣ couplings constants as dihedral restraints in peptide. Fifty-two intraresidue and 22 interresidue NOEs and 6 dihedral restraints from 3J HNH␣ coupling constants were used. One-hundred structures were calculated with simulated annealing protocols using CNS program (51). Twenty structures with the low-est energies were selected for root mean square deviation calculation and shown in Fig. 7B.

In Vitro Inhibition of Fibril Formation by Anilinonaphthalene
Compounds-Aggregation of ␤-1-42 and ␤-13-23 in vitro is already well documented (11,39). We performed aggregation kinetics with both peptides in stagnant solutions under optimal conditions (pH 7.4, 37°C) by following the increment in the LS that takes place when the peptides are diluted out from a pure Me 2 SO solution (Fig. 2).
As seen in Fig. 2, the increase in the LS for both peptides is very rapid and pronounced, taking less than 5 min for completion. The aggregation of A␤-1-42 is slightly faster and considerably more extensive than the aggregation of A␤-13-23 (TABLE ONE), although both peptides present a similar aggregation profile (Fig. 2). To confirm that the increase in LS represents the formation of amyloid fibrils, thioflavin T and Congo Red binding were also evaluated and the data are presented in To gain a preliminary idea of the mass of the species that coexist with the amyloid fibrils or are rescued from fibrils by the action of SDS, we carried out SDS-PAGE with the fibrils of A␤-1-42 (inset of Fig. 2, left) and A␤-13-23 (inset of Fig. 2, right) after 1 h in aggregating buffer. As seen, the fibrils of A␤-1-42 solubilized in SDS show the presence of species with ϳ4 and 16 -20 kDa, which are compatible with the mass of monomers and tetramers. These species were either rescued from the fibrils by the action of SDS or were present in equilibrium with them. The gel obtained with the fibrils of A␤-13-23 treated with SDS also presented low molecular weight species, although the bands are very faint. Since the fibrils of ␤-13-23 are SDS-resistant (data not shown), the presence of these low molecular weight species (range ϳ10 -16 kDa) suggests they are probably in equilibrium with the fibrils before SDS addition.
Since the light-scattering increase reflects the formation of amyloid fibrils with both two peptides, we were able to examine the effect of small, hydrophobic molecules belonging to the family of ANS com- pounds. The use of light scattering measurements to evaluate the extent of fibril formation is essential, since the ANS compounds fluoresce in the same range as thioflavin T.
To verify the effect of bis-ANS on the aggregation of A␤-1-42 and A␤-13-23, experiments similar to those in Fig. 2 were performed in the presence of increasing concentrations of this probe in the aggregating solution. The final LS values attained after 20 min were compared with the LS of the control sample (absence of bis-ANS, 100% aggregation), and the results (Fig. 3) show that bis-ANS inhibits the aggregation of both peptides in a dose-dependent fashion. In the presence of 200 M bis-ANS (peptide:bis-ANS molar ratio ϭ 1:8), aggregation of A␤-1-42 and A␤-13-23 is inhibited by 80 and 95%. However, as seen, aggregation of A␤-13-23 is more sensitive to the probe with a half-maximal effect attained in 15 M bis-ANS compared with 62 M for the longer peptide.
To evaluate the size of the species formed during aggregation of A␤ in the presence of bis-ANS, size exclusion chromatography was performed. Fig. 4 shows the chromatograms of A␤-1-42 (A) and A␤-13-23 (B) after aggregation in the absence (dashed line) or in the presence (solid line) of 200 M bis-ANS. After 1 h in aggregating buffer, the samples of A␤-1-42 and A␤-13-23 were injected directly into a G3000sw xl or GPC 60 column, respectively, without previous centrifugation. In both cases there are high molecular weight species present in the control samples, which elute in the void volume of the columns; these species are absent from the samples incubated in the presence of bis-ANS. This indicates that bis-ANS is able to prevent the formation of high molecular weight species in both A␤ samples. Furthermore, the main population present in the samples aggregated in the presence of bis-ANS consists of LMW species that elute close to or after the total volume of the columns. Although their precise weights cannot be inferred from these data, the presence of these LMW species during the aggregation of A␤ is consistent with the data from other groups (19,20).
To better visualize the LMW species of A␤-1-42 stabilized by bis-ANS, the samples were briefly centrifuged to remove the large oligomers before injection into the G3000sw xl (C) or GPC 60 (D) column. As seen in C, the high molecular weight species present in the aggregation of A␤-1-42 that elute at ϳ4 min (A) disappeared with the centrifugation, leaving a more homogeneous population containing LMW. In the sample of A␤-1-42 aggregated in the presence of 200 M bis-ANS, the population is more heterogeneous, comprising species with different weights, but all of them smaller then the resolution of the column (C). This indicates that bis-ANS is stabilizing small oligomeric forms of A␤ (LMW species), which are present in the aggregation pathway of A␤. This can be confirmed in D, which shows larger amounts of these LMW species in the sample aggregated in the presence of bis-ANS. Altogether, these data suggest that bis-ANS is able to prevent fibril formation by stabilizing a large range of LMW species of A␤.
It has been proposed that the LMW species formed during aggregation of several amyloidogenic proteins are in fact the toxic species (27-31, 33, 45). To evaluate whether bis-ANS is trapping such species, MTT assays were performed to evaluate RAW cell viability (Fig. 5) (52). Thus, we incubated 25 M of A␤-1-42 with different concentrations of bis-ANS (50 and 100 M) during 20 min, to stabilize the LMW species. These suspensions were added to RAW cells and MTT reduction was evaluated 24 h later (Fig. 5). The species trapped in the presence of 100 M bis-ANS was more toxic (bar 6; *, p Ͻ 0.01) to these cells than the fibrils of A␤ produced in the absence of any addition (bar 4) or in the presence of 50 M bis-ANS (bar 5), a concentration that inhibited aggregation by less than 50% (Fig. 3). These results are consistent with a role for the LMW species, which are more abundant in the presence of bis-ANS. We note that these concentrations of bis-ANS are not toxic to this cell line (Fig. 5, bars 2 and 3), although higher concentrations of bis-ANS (Ͼ200 M) exhibited a toxic effect (data not shown).
Since both peptides responded similarly to the inhibitory effect of bis-ANS, we chose A␤-13-23 as a model for further experiments performed with other bis-ANS related compounds (ANS and AMNS). As seen in Fig. 6, ANS (triangles) also inhibited the aggregation of ␤-13-23 but less effectively than bis-ANS (Fig. 3). Strikingly, AMNS (Fig. 6, circles) was more potent than ANS and 100 -150 M AMNS abolished completely the aggregation of this peptide. The half-maximal effects were obtained at 150 and 30 M for ANS and AMNS, respectively. At these concentrations, the drug:peptide molar ratio was 6:1 (ANS) and 1.2:1 (AMNS).
Comparing the Structure of ␤-13-23 in the Presence of SDS or in the Presence of AMNS by NMR-Several laboratories have shown that A␤ remains soluble in SDS micelles and the region comprising the hydrophobic residues 17-21 tends to assume an ␣-helical structure (15,53,54). In the case of A␤-13-23, CD measurements were performed in the presence of SDS as shown in Fig. 7A. The spectrum presented two minima at 208 and 222 nm, a profile that is characteristic of ␣-helix structure. Although A␤-13-23 also remains soluble in the presence of AMNS, the CD spectrum of A␤-13-23 in this case (data not shown) was

Sample Increase in LS at 320 nm (LS/LS 0 )
ThT binding (spectral area of ThT bound/ spectral area of free ThT)

CR binding (mmol CR/L fibrils)
A␤-1-42 ( too noisy to provide useful data due to the large absorption of AMNS in the UV region. Since A␤-13-23 remains soluble and ordered in the presence of SDS, we performed 1 H NMR experiments at pH 7.4 and 37°C to determine its three-dimensional structure. Although we already knew that this peptide adopts an ␣-helical structure in the presence of SDS (15)(16)(17), we aimed to determine which region of the peptide is involved in this helix. For this purpose, TOCSY and NOESY experiments were performed as described under "Experimental Procedures." The three-dimensional structure of A␤-13-23 resolved in the presence of SDS micelles is presented in Fig. 7B, which shows the 20 lowest energy structures calcu-lated using NOEs and 3 J HNH␣ coupling constants as restraints. As shown, there is a short ␣-helix comprising the residues from Lys 16 to Phe 20 .
1 H NMR (TOCSY and NOESY) spectra of A␤ 13-23 in the presence of AMNS (molar ratio 4:1) were also collected. The TOCSY spectrum in the presence of AMNS presented chemical shift dispersion very similar to that observed in the presence of SDS. Fig. 8 shows the TOCSY spectra in the amide region of 4 mM A␤-13-23 in the presence of 35 mM SDS (black) or in the presence of 20 mM AMNS (red). All spin systems are present in both samples and could be seen by changing the spectra threshold. This result suggests that the soluble form of A␤-13-23 sta- bilized by AMNS may be similar to the helicoidal structure observed in presence of SDS. A␤-13-23 NOESY spectrum in the presence of AMNS displayed fewer NOE peaks when compared with the same spectrum performed in the presence of SDS, which suggests that soluble A␤-13-23 in AMNS must be dynamically more flexible than the peptide in SDS micelles. Due to the limited number of NOE restraints on the NOESY spectrum of A␤ 13-23 in the presence of AMNS, we were not able to perform structural calculations. We are working now to establish other conditions that will make it possible to determine the three-dimensional structure of A␤-13-23 in aqueous solution.

DISCUSSION
The neurotoxic effects observed in AD have been linked to fibrils present in the brain parenchyma of the patients ("amyloid cascade" hypothesis, by 55). Several laboratories have been trying to characterize the species present in the very first steps of A␤ aggregation. From these studies, oligomeric forms of A␤ formed prior to fibril formation have been identified as neurotoxic (28 -30, 38). These soluble oligomers are common to most amyloids and may be responsible for the amyloidogenic diseases. Recently, a polyclonal antibody that recognizes soluble oligomeric intermediates of A␤ was also able to recognize soluble oligomers from several other types of amyloidogenic proteins and peptides, indicating that all of them have a common architecture (56). This antibody abolished the in vitro neurotoxic effect of these oligomers, reinforcing the idea that they are indeed the pathogenic agents in these diseases.
Several subfragments of A␤ were tested in an effort to dissect the minimum core necessary to keep fibril formation. The region 16 -20 has been shown to be essential for fibrillogenesis of A␤ peptides (11,39,57). Molecular modeling performed with this region using A␤- 14 -23 showed that oligomers could be formed by an antiparallel ␤-sheet kept by favorable hydrophobic interactions and stabilized by salt bridges between charged residues (39). Besides, A␤-1-40 and A␤-12-28 had their three-dimensional structures determined in SDS micelles at acidic pH and presented the region 15-24 structured as an ␣-helix. This structure seems to be stabilized by charged groups present in the ␤-amyloid primary sequence (15,53). Thus, this same sequence of amino acids could be organized in ␤-sheets or as ␣-helix, depending on the environ-   Studies performed by Montserret et al. (58) using a basic and an acidic peptide whose primary sequence was derived from the sequence of human platelet factor 4 (PF-4) showed that only the basic peptide (QAPAYKKAAKKLAES) in the presence of SDS could assume an ␣-helical structure at neutral pH (58). This helix was stabilized by strong, electrostatic interactions established between the anionic groups of SDS and the cationic groups of the lysines. This region seems to be necessary to initiate the folding of this basic peptide. From these results, a model was proposed to explain the effects of SDS as a ␣-helix inductor. This model proposes that, although the folding of the peptide is mostly driven by hydrophobic effects, electrostatic interactions (salt bridges between lysine residues and SDS sulfate groups) might play a significant role in the formation and stabilization of the ␣-helical structure (58).
Recently, LeVine (59) has shown that bis-ANS interacts strongly with the soluble A␤-1-40 in acidic buffer solutions but interacts poorly with the formed fibrils. In this study, however, the effects of this probe on the aggregation of A␤ were not evaluated. Interestingly, bis-ANS was also able to inhibit the aggregation of a peptide derived from the hamster prion protein (43). In the present study, we were able to show that ANS-related compounds abolish almost completely the aggregation of A␤-1-42 and -13-23 in a dose-dependent manner. These compounds seem to interact with and stabilize soluble species present in the aggre-gation pathway that leads to fibril formation. These species were observed in size exclusion chromatography experiments and proved to be more toxic than the aggregates generated in the absence of ANSrelated compounds. One possibility would be the formation of bis-ANS micelles with monomers inserted in them, which would elute in a size exclusion chromatography experiment at the same position as the oligomers described here. This possibility was discarded because a solution of 200 M bis-ANS elutes in a size exclusion column at the same position as 2 M (data not shown). Besides, bis-ANS has no aliphatic tail to allow micelle formation.
What could be the molecular basis for the inhibition of aggregation? The ANS family of compounds might interfere with electrostatic and hydrophobic interactions that are crucial for fibril formation and stabilization. This effect could be achieved due to the dual nature of the ANS-derived molecules, which have hydrophobic rings and charged groups. Thus, it is possible that, similarly to SDS, bis-ANS, ANS, and AMNS might interact with specific, crucial regions of the A␤ peptide in the very beginning of the nucleation step, shielding them from interaction with other peptide molecules. As a result, the oligomerization reaction would be halted. This mechanism of inhibition has been proposed for other peptides and proteins solubilized in SDS micelles (58,60).
It has been shown that the binding of bis-ANS to tubulin monomers blocks their polymerization induced by microtubule-associated proteins (61). In addition, Teschke et al. (62) showed that bis-ANS was able to inhibit the assembly of coat proteins from bacteriophage P22 into procapsids. These results suggest that bis-ANS can interfere with intersubunit contacts regardless of whether these are related to the assembly of a microtubule, a viral particle, or in our case, a fibril.
Another relevant observation is the fact that highly sulfated glycosaminoglycans have been found to be associated with A␤ fibrils in vivo (63). In vitro, glycosaminoglycans and CR compete for the same binding sites on A␤ fibrils (64). CR is a specific dye for amyloid fibrils sharing similarities with bis-ANS such as the presence of sulfonate groups. CR was able to inhibit aggregation of A␤ into fibrils (27,65) decreasing concomitantly its neurotoxicity (66). Li et al. (40) proposed a model to explain CR effects on the fibrillogenesis of A␤. These authors postulated that Lys 16 could interact electrostatically with the sulfonate groups of CR, in addition to any hydrophobic contacts that could be established between uncharged regions of the two molecules.
Bis-ANS and CR share structural similarities in size and number of aromatic rings as well as in chemical properties such as hydrophobicity and the presence of charged groups (sulfonates). Thus, it is possible that both compounds might act similarly in inhibiting the aggregation of A␤. However, while the LMW species trapped in the presence of bis-ANS are toxic leading to an increase in cell death, the one trapped in the presence of CR are not. We do not have any explanation for this observation, but it is possible that the LMW species trapped in the presence of these compounds are different in terms of size and consequently in toxicity. Recently, Gazit et al. (67) showed that phenol red, an aromatic compound with a sulfonate radical, could inhibit with high efficiency fibril formation of the human islet amyloid polypeptide. The authors propose that heteroaromatic interactions exerts a crucial role in the inhibitory mechanism. A similar mechanism might be underlying the effect of bis-ANS describe here, especially because the A␤-13-23 contains two adjacent phenilalanines.
The search for molecules that can interfere with A␤ aggregation could lead to the understanding of the driving forces behind inter-and intrapeptide interactions as well as the structural motifs responsible for them. These molecules could also be used as lead compounds for developing drugs against this devastating disease.