Amyloid-β Interactions with Chondroitin Sulfate-derived Monosaccharides and Disaccharides

In Alzheimer's disease, the major pathological features are diffuse and senile plaques that are primarily composed of the amyloid-β (Aβ) peptide. It has been proposed that proteoglycans and glycosaminoglycans (GAG) facilitate amyloid fibril formation and/or stabilize the plaque aggregates. To develop effective therapeutics based on Aβ-GAG interactions, understanding the Aβ binding motif on the GAG chain is imperative. Using electron microscopy, fluorescence spectroscopy, and competitive inhibition ELISAs, we have evaluated the ability of chondroitin sulfate-derived monosaccharides and disaccharides to induce the structural changes in Aβ that are associated with GAG interactions. Our results demonstrate that the disaccharides GalNAc-4-sulfate(4S), ΔUA-GalNAc-6-sulfate(6S), and ΔUA-GalNAc-4,6-sulfate(4S,6S), the iduronic acid-2-sulfate analogues, and the monosaccharidesd-GalNAc-4S, d-GalNAc-6S, andd-GalNAc-4S,6S, but not d-GalNAc,d-GlcNAc, or ΔUA-GalNAc, induce the fibrillar features of Aβ-GAG interactions. The binding affinities of all chondroitin sulfate-derived saccharides mimic those of the intact GAG chains. The sulfated monosaccharides and disaccharides compete with the intact chondroitin sulfate and heparin GAGs for Aβ binding, as illustrated by competitive inhibition ELISAs. Therefore, the development of therapeutics based on the model of Aβ-chondroitin sulfate binding may lead to effective inhibitors of the GAG-induced amyloid formation that is observed in vitro.

Alzheimer's disease is characterized neuropathologically by amyloid deposits, neurofibrillary tangles, and selective neuronal loss. The major component of the amyloid deposits is a 39 -43-residue peptide, amyloid-␤ (A␤). 1 A␤ fibrillogenesis in vitro is a nucleation-dependent process consisting of a slow lag phase for nucleation followed by faster propagation of fibrils (1)(2)(3). However, in vivo fibrillogenesis is likely a complex pathway involving many factors that modulate the aggregation of A␤. Two mechanisms have been proposed for the nucleation of A␤ fibrils. The first involves the self-assembly of A␤ monomers, which undergo a conformational change to become the fibril nucleus. The second involves an alternative pathway of heterogenous nucleation, which results from outgrowth of fibrils from non-A␤ seeds (1).
Many proteins are associated with amyloid plaques; their presence may result in heterogeneous nucleation of A␤ (4 -11). Although heparan sulfate proteoglycans have been extensively correlated with plaque formation, in Alzheimer's disease at least four types of proteoglycans are associated with amyloid plaques (12)(13)(14)(15)(16)(17). A␤-proteoglycan interactions are mediated predominantly through A␤-glycosaminoglycan (GAG) binding with GAGs acting as a scaffold for the assembly of the fibrils. The scaffold may function by enhancing the structural features that favor a ␤-sheet conformation thereby increasing the number of nucleation seeds, as demonstrated by a virtually instantaneous structural transition in A␤ upon addition of GAGs (18). In the later stages of the amyloid pathway, GAGs also act by enhancing lateral aggregation of small fibrils to confer insolubility and protection from proteolysis (18 -21). A structureactivity relationship for A␤-GAG interactions is slowly emerging based on the affinities of various GAGs. In vitro studies have shown that the chondroitin sulfates are more effective at both nucleation and lateral aggregation of A␤ fibrils than the heparin GAGs (18). Chondroitin sulfates are sulfated on a single face of the polymer and may represent an ideal distribution of charge for A␤ interactions. Therefore, these GAGs were used as the prototype to determine A␤ binding and potentially to develop compounds that could compete with all identified proteoglycans associated with plaques.
The GAG binding site on the A␤ peptide has been investigated using amino acid substitution of the A␤1-28 peptide and as a function of aggregation state (22,23). These studies have demonstrated that although electrostatic interactions through basic amino acids contribute to GAG binding, nonionic interactions, such as hydrogen bonding and van der Waals packing, play a role in GAG-induced A␤ folding and aggregation (22). Furthermore, GAG-A␤ interactions are more sensitive to the conformation and aggregation state of A␤ rather than the primary sequence (22,23). Together these results suggested that inhibition of A␤-GAG interactions through targeting of the GAG binding site on A␤ may not provide a viable therapeutic. Alternatively, the A␤-GAG interaction may be mediated by a unique binding site on the GAG backbone that could serve as a target for inhibition of amyloid formation. This therapeutic strategy is supported by in vitro studies, which demonstrated * This work was supported by grants from the Ontario Mental Health Foundation (to J. M. and P. E. F.), the Scottish Rite Charitable Foundation (to P. E. F.), Neurochem Inc. (to P. E. F.), University of Toronto Dean's Fund (to J. M.), and The Banting Foundation (to J. M.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
¶ that polysulfated compounds could inhibit binding of heparan sulfate to A␤ (24). Also using an in vivo model of splenic amyloidosis, small sulfonated or sulfated molecules have been shown to be active inhibitors of amyloid deposition (25). Alternatively, small GAG-derived saccharides may have the alternate effect of enhancing A␤ precipitation into nontoxic plaques and thereby decreasing the presence of toxic A␤ species. To develop more specific therapeutics directed toward A␤ fibrillogenesis, we determined the minimum GAG unit necessary for A␤ binding, fibrillogenesis, and lateral aggregation. Here, we examine the interaction of chondroitin sulfate-derived monosaccharides and disaccharides with A␤40 and A␤42. Fluorescence spectroscopy and electron microscopy demonstrate a potent effect of both monosaccharides and disaccharides on the formation and structure of A␤ fibrils. In addition, competitive inhibition ELISAs demonstrate that the binding of both monosaccharides and disaccharides to A␤ inhibits interaction with the polymeric chondroitin sulfate and heparan sulfate GAGs.

MATERIALS AND METHODS
A␤ Peptides-A␤40 and A␤42 were synthesized by solid phase Fmoc (N-(9-fluorenyl)methoxycarbonyl) chemistry by the Hospital for Sick Children's Biotechnology Center (Toronto, ON). Peptides were isolated by reverse phase high pressure liquid chromatography on a C18 bondapak column, and purity was determined using mass spectrometry and amino acid analyses. Peptides were initially dissolved in 0.5 ml of 100% trifluoroacetic acid (Aldrich), diluted in distilled H 2 O, and immediately lyophilized (26). Peptides were then dissolved in 40% trifluoroethanol (Aldrich) in H 2 O and stored at Ϫ20°C until use. Alternatively, the lyophilized peptides were dissolved in distilled H 2 O at 2.5 mM concentration and used immediately. Tyrosine Fluorescence Assay-Tyrosine emission spectra from 290 to 340 nm were collected (excitation wavelength 281 nm, 0.5 s/nm, band pass ϭ 4 nm). A cuvette with a 1-cm path length was used. For the centrifugation studies, 1 M A␤40 or A␤42 was incubated in the presence or absence of chondroitin sulfate subunits at a 1:1 ratio for 24 h. Samples were centrifuged for 30 min at 15,600 ϫ g to sediment aggregates and fibrils as described previously (27,28). The relative amount of tyrosine in the supernatant was then determined. The fluorescence of the noncentrifuged sample was used as a measure of the total tyrosine fluorescence.
Electron Microscopy-For negative staining, carbon-coated pioloform grids were floated on aqueous solutions of peptides (100 g/ml). After grids were blotted and air-dried, the samples were stained with 1% (w/v) phosphotungstic acid, pH 7.0. The peptide assemblies were observed in a Hitachi H-7000 operated with an accelerating voltage of 75 kV (28).
Amyloid Staining-Thioflavin T fluorescence assay of A␤ in the presence and absence of GAGs (29,30) and GAG-derived disaccharides and monosaccharides was used to evaluate the similarity between A␤-GAG fibrils and classical amyloid fibers. Samples were incubated at a 1:1 ratio by weight with a final A␤ concentration of 200 M for 3 days. Samples were vortexed and 40-l aliquots were added to 960 l of 10 M Thioflavin T in phosphate-buffered saline, pH 6.0. Steady state fluorescence was measured at 20°C using a Photon Technology International QM-1 fluorescence spectrophotometer equipped with excitation intensity correction and magnetic stirrer. Thioflavin T emission spectra from 475 to 495 nm were collected (excitation wavelength 437 nm, 0.5 s/nm, band pass ϭ 4 nm). A cuvette with a 1-cm path length was used.
Competitive Inhibition ELISAs-Nunc Immunosorp plates were coated with 100 l of GAGs (5 g/ml) and incubated overnight, 4°C. Simultaneously, A␤40 and A␤42 were incubated with chondroitin sul-fate-derived monosaccharides and disaccharides at a 1:1 or 1:10 ratio by weight. The plates were rinsed twice with water and blocked with 100 l of 1% bovine serum albumin in phosphate buffered saline. After incubation for 1 h at room temperature, the plates were washed 3 times with 0.05% Tween 20/phosphate-buffered saline and twice with phosphate-buffered saline. A␤ was then added to the plates and incubated for 2 h at room temperature with shaking. Plates were washed as above before the addition of monoclonal antibody against A␤, 6F/3D (Dako), 6E10, or 4G8 (Senetek, Carpinteria, CA). The reaction with 50 l of horseradish peroxidase-conjugated goat anti-mouse IgG 1:2000 was performed at room temperature for 1 h. Color development was achieved with 100 l of 2,2Јazino-di(3-ehtyl-benzthiazoline-6-sulfonic acid) in 0.1 M acetate buffer, pH 4.2. The absorbance was monitored at 415 nm on a Bio-Rad Benchmark microtiter plate reader.

Induction and Morphology of A␤ Fibril in the Presence of
Chondroitin Sulfate-derived Disaccharides-We have previously shown that the chondroitin sulfate GAGs are the most efficient at inducing a ␤-structural transition in A␤ and subsequent fibrillogenesis (18). Therefore, we have used the chondroitin sulfate-derived monosaccharides and disaccharides ( Fig. 1) to elucidate the minimum sugar moiety necessary for A␤ binding and fibrillogenesis. Chondroitin sulfate subunits are derived by enzymatic cleavage of the GAG chain by chondroitinases ABC, AC-I, -B, or -C. The saccharides used in this study represent repeat disaccharides present in chondroitin-4sulfate, chondroitin-6-sulfate, and dermatan sulfate, which retain the charge distribution present in the intact GAG (Fig. 1). The monosaccharides are generated by removal of uronic acid from and desulfation of the disaccharides.
To investigate the effect of chondroitin sulfate-derived saccharides on A␤ nucleation, the intrinsic tyrosine fluorescence of A␤40 and A␤42 was used to monitor the amount of soluble peptide after incubation in the presence and absence of the disaccharides. After 24 h of incubation, soluble A␤ was separated from aggregated and fibrillar peptide by centrifugation (23,24). Different chondroitin sulfate-derived saccharides had variable effects on the amount of pelletable aggregates detected with greater effects seen for A␤42 ( Fig. 2) over A␤40 (data not shown). At 24 h, the disaccharides ⌬UA-GalNAc-4S, ⌬UA-Gal-NAc-6S, and ⌬UA-GalNAc-4S,6S significantly increased the amount of aggregated A␤40/42 with ⌬UA-GalNAc-4S,6S being the most effective (Fig. 2C). It is interesting to note that all disaccharides and in particular ⌬UA-GalNAc-4S,6S induced the same extent of fluorescence loss as the intact GAG, dermatan sulfate (Fig. 2D). These results demonstrate that the presence of chondroitin sulfate-derived disaccharides can be corre- lated with an increased amount of aggregated A␤.
The characteristics of A␤40 and A␤42 fibrils in the presence and absence of chondroitin sulfate-derived saccharides were examined by electron microscopy. Previous investigations indicated that GAG promoted morphological changes in the fibrous structures formed by A␤40 and A␤42 (18). Unseeded samples of both A␤40 and A␤42 were incubated in the presence of chondroitin sulfate-derived disaccharides, intact chondroitin sulfate GAGs, and alone for up to 96 h. Negative stain electron microscopy demonstrated that A␤42 fibrils were 50 -70 Å in diameter with an average length of 750 Å (Fig. 3A). These were indistinguishable from those of A␤42 in the presence of the desulfated ⌬UA-GalNAc (Fig. 3B) or ⌬UA-GlcNAc (data not shown). The monosulfated disaccharides, ⌬UA-GalNAc-6S and ⌬UA-GalNAc-4S, induced fibrils of similar size but with increased lateral aggregation as compared with control (Fig. 3C). The bundles of fibers were similar to those seen in the presence of polymeric chondroitin sulfate GAGs (18). In the presence of ⌬UA-GalNAc-4S,6S, A␤42 formed many fibers displaying extensive lateral aggregation as illustrated by the heavily stained clusters of fibrils (Fig. 3D). These results demonstrate that the sulfated disaccharides derived from chondroitin sulfate are representative of the intact GAG chains in terms of activity. The extent of lateral aggregation induced by the disaccharides reflects that of the intact GAGs, with 4-sulfate Ͻ 6-sulfate Ͻ 4/6-sulfate.
Another major component of polymeric GAGs are the 2-sulfated iduronic acid disaccharides, which are present predominantly in heparan sulfate GAGs. Certain repeating disaccharide motifs containing sulfated iduronic acid are also present as part of the chondroitin and dermatan sulfate backbones. Many studies have demonstrated that charge distribution across small molecules represents the limiting factor for A␤ binding (28,31). Therefore, the introduction of a sulfate group onto the iduronic acid may present a more favorable or deleterious surface for A␤ binding. To investigate these possibilities, we examined the effect of 2-sulfated iduronic acid-containing disaccharides on fibril assembly and morphology. Negative stain electron microscopy demonstrated that the extra sulfate on the ⌬UA-2S-GalNAc had no effect on the A␤42 fibers formed in comparison to A␤42 alone or in the presence of ⌬UA-GalNAc (data not shown). In the presence of ⌬UA-2S-GalNAc-4S and ⌬UA-2S-GalNAc-6S, many short fibers could be detected displaying limited aggregation. (Fig. 3E). Incubation of A␤42 with ⌬UA-2S-GalNAc-4S,6S results in the development of thick aggregated fibers displaying a helical twist (Fig. 3F). Close examination of the fibers reveal a length of 300 -500 Å with a diameter of 100 Å, which correspond to mature amyloid fibers. These results demonstrate that the sulfate group on the second position of the iduronic acid does not effect the morphology of A␤ fibrils formed in the presence of sulfated GalNAc disaccharides. Furthermore, these data suggest that using a disaccharide with a sulfated iduronic acid does not contribute to the minimal unit necessary for A␤ binding and enhanced fibrillogenesis.
GAGs also laterally aggregate preformed A␤ fibrils into large masses characteristic of insoluble plaques. To evaluate this phenomenon as induced by chondroitin sulfate-derived disaccharides, we incubated the saccharides with preformed A␤40 fibrils and examined the morphology by negative stain electron microscopy. Preincubated A␤40 forms long fibers that are nonaggregated (Fig. 4A). In contrast, dermatan sulfate induced a consistent lateral aggregation of A␤40 fibrils with apparent helical twisting (Fig. 4B). The fiber bundles had a cumulative diameter of up to 200 Å, but the average length of the fibers remained unaffected. In the presence of chondroitin sulfatederived disaccharides, A␤40 lateral aggregation was similar to that of an extended GAG polymer, with the disulfated ⌬UA-GalNAC-4S,6S being the most effective (Fig. 4C). The fibers were aligned in large bundles rather than a haphazard distribution across the grid. These data indicate that the binding of chondroitin sulfate-derived disaccharides to A␤ fibrils is sufficient to stabilize the macromolecular structure of pre-existing fibers by lateral aggregation and represent the smallest unit responsible for interfiber stabilization. Similar to polymeric GAGs, all data taken together demonstrate that the extent of sulfation on the disaccharide backbone defines the extent to which the disaccharide interacts with and stabilizes A␤. Furthermore, these small GAG-derived saccharides demonstrate the potential use of these molecules to decrease the soluble pool of A␤ in situ by enhancing the precipitation of nontoxic fibers.
Effect of Chondroitin Sulfate-derived Monosaccharides on A␤ Fibrillogenesis-The effect of chondroitin sulfate-derived monosaccharides on the induction of A␤42 fibrillogenesis was investigated. To investigate the effect of chondroitin sulfate-derived saccharides on A␤ nucleation, the intrinsic tyrosine fluorescence of A␤40 and A␤42 was used to monitor the amount of soluble peptide after incubation in the presence and absence of the monosaccharides. At 24 h, no significant difference could be detected in the amount of A␤40/42 pelleted in the presence and absence of the monosaccharides D-GalNAc, D-GalNAc-4S, and D-GalNAc-6S, whereas a slight increase in the amount of A␤42 pelleted in the presence of D-GalNAc-4S,6S could be detected (Fig. 2B).
The morphology of A␤ fibrils in the presence of chondroitin sulfate-derived monosaccharides was investigated using negative stain electron microscopy. Similar to A␤42 alone (Fig. 5A), the presence of nonsulfated D-GalNAc and D-GlcNAc induced fibers similar to mature amyloid fibers (Fig. 5B). In the presence of D-GalNAc-4S or D-GalNAc-6S (Fig. 5C), the fibers were less abundant and of varying lengths but exhibited some aggregation as demonstrated by the uneven, heavily stained distribution of fibers across the grid. Alternatively, in the presence of D-GalNAc-4S,6S and D-galacturonic acid, A␤42 formed many protofibrils characterized by short flexible fibrils (Fig. 5D). This suggests that binding of the sulfated monosaccharide D-Gal-NAc-4S,6S or D-galacturonic acid to A␤ enhances the nucleation stage of fibrillogenesis resulting in the formation of many protofibrils. The ability of D-galacturonic acid but not D-GalNAc to nucleate A␤ may not be surprising because of the differences in charge distribution across the sugar backbone. D-Galacturonic acid is derived by removal of the N-acetylamine group from position 4 of the sugar backbone, and the resultant charge distribution may represent a preferential binding motif. These data further suggest that monosaccharides can inhibit the formation of mature amyloid fibers by blocking the self-association of protofibrils. Further evidence to support this hypothesis is derived from the lack of lateral aggregation of both A␤40 and A␤42 preformed fibrils by all monosaccharides (data not shown).
Thioflavin T specifically stains amyloid deposits in vivo and has been shown to bind both A␤ fibers and aggregates in vitro (30). We investigated the binding of Thioflavin T to chondroitin sulfate-derived saccharide-A␤ complexes to further characterize the nature of these fibers. Thioflavin T fluorescence intensity increased for both A␤40 and A␤42 in the presence of intact chondroitin sulfate GAGs (Fig. 6). In the presence of D-GalNAc, the Thioflavin T fluorescence was indistinguishable from A␤42 alone, which is consistent with our electron microscopy data. Thioflavin T fluorescence increased in the presence of D-Gal-NAc-6S but to a lesser extent than chondroitin-6-sulfate. When incubated with D-galacturonic acid or D-GalNAc-4S,6S, A␤42 demonstrated a morphology similar to protofibrils and exhibited a Thioflavin T fluorescence greater than A␤42 alone. These results demonstrate that the protofibrils induced by chondroitin sulfate-derived monosaccharides have characteristics similar to typical amyloid fibers. In summary, our data demonstrate that chondroitin sulfate-derived monosaccharides  bind A␤ and induce fibrillogenesis without lateral aggregation.
Characterization of the A␤-Saccharide Binding Site-To compare the specificity of chondroitin sulfate-derived saccharide binding, we used competitive inhibition ELISAs to determine whether the chondroitin sulfate-derived saccharides could compete with intact GAG chains for A␤40 and A␤42 binding (Table I). Concentration-dependent studies were used to determine both the specificity of competition and the relative binding strengths of each component. The chondroitin sulfatederived monosaccharides and disaccharides were preincubated with A␤40 and A␤42 before incubation with chondroitin-4sulfate, chondroitin-6-sulfate, or dermatan sulfate. The amount of A␤ bound to the intact GAG chain on the microtiter plate was determined using the anti-A␤ antibodies 6E10, 4G8, or 6F/3D. All antibodies demonstrated similar concentrationdependent inhibition profiles indicating that the detection antibody was not a determining factor. D-GalNAc and ⌬UA-Gal-NAc were unable to compete with the chondroitin sulfate GAGs for A␤ binding, which is in agreement with our electron microscopy data in which A␤40/42 fibrils were indistinguishable in the presence and absence of D-GalNAc and ⌬UA-GalNAc. The monosaccharides D-galacturonic acid, D-GalNAc-6S, and D-GalNAc-4S,6S were all effective at competing with all chondroitin sulfate GAGs for A␤ binding at a 1:10 ratio (by weight). These results suggest that the alterations in fibrous structure detected by electron microscopy and fluorescence studies can be attributed to the binding of these monosaccharides to the GAG binding site on A␤40 and A␤42. These observations further suggest that the monosaccharides D-galacturonic acid, D-Gal-NAc-6S, and D-GalNAc-4S,6S are sufficient to induce A␤ binding and structural transitions associated with A␤-GAG interactions. It was not surprising to find that D-GalNAc-4S competed poorly with all GAGs for A␤ binding because the intact chondroitin-4-sulfate is the least effective of all the chondroitin sulfate GAGs at inducing the structural transitions necessary for fibril formation and aggregation (18).
The chondroitin sulfate-derived disaccharides had variable abilities to compete for A␤ binding with ⌬UA-Gal-4S,6S being the most effective (Table I). The differences in binding of the disaccharides reflect the varying abilities of the chondroitin sulfate GAGs to bind, induce a structural change in A␤, and enhance lateral aggregation. One corollary to our results is that we cannot rule out the possibility that the disaccharides induced a conformational change in A␤ that allowed the intact GAG chain to elicit binding between A␤ fibrils as has been previously suggested for GAG binding to preformed fibers (18). The chondroitin sulfate-derived monosaccharides and disac-charides binding strengths, as determined by the extent of competition, may reflect the fluctuation of A␤ binding to surfaces with slight variations in charge distribution. These characteristics have been reported previously for myo-inositol and its phosphorylated analogues as well as alterations in the distribution and surface charge of the antibiotic rifampicin (28,31).
Chondroitin Sulfate-derived Saccharides Inhibit Heparan Sulfate GAG Binding to A␤-The similarities in GAG structure between chondroitin sulfate and heparan sulfate GAGs previously have stimulated the suggestion that proteins that bind to chondroitin sulfate should interact with heparan sulfate and vice versa (32)(33)(34). The basic fibroblast growth factor of the heparan sulfate-binding proteins, platelet-derived factor 4, and fibronectin react weakly with dermatan sulfate, whereas heparin cofactor II and hepatocyte growth factor have a comparable high affinity for both heparan sulfate and dermatan sulfate (35)(36)(37)(38)(39). To determine whether the chondroitin sulfate-derived monosaccharides could compete with other GAGs for A␤ binding, we repeated the competitive inhibition ELISAs using heparin (Table II). It is speculated that polymeric GAGs bind to the same region or structural motif in A␤ (40); therefore it was not unexpected to find that the monosaccharide D-GalNAc-4S,6S could compete to the same extent with heparin as was seen for dermatan sulfate. Our results for heparin competition illustrate that the competition detected between heparin and the chondroitin sulfate-derived saccharides is independent of the detection antibody, as both the A␤-specific antibodies, 6E10 and 4G8, detect a similar concentration-dependent inhibition (Table II). These results suggest that development of an inhibitor for GAG binding to A␤ could represent an agent to block A␤-proteoglycan interactions.
Further investigation into the ability of chondroitin sulfatederived saccharides to inhibit heparan sulfate binding to A␤ demonstrated similar results to those of both chondroitin and  dermatan sulfate (Table III). None of the nonsulfated monosaccharides or disaccharides could inhibit A␤ binding to both heparan sulfate and keratan sulfate; these results are similar to those for dermatan sulfate competition studies. The iduronic acid-2-sulfated disaccharides were unable to compete for heparan sulfate binding, whereas ⌬UA-2S-GalNAc-4S and ⌬UA-2S-GalNac-6S were able to compete for keratan sulfate binding. These results suggest that subtle changes in the GAG backbone and distribution of sulfation have significant effects on the ability of chondroitin sulfate-derived saccharides to compete for GAG binding sites. The disaccharides ⌬UA-GalNAc-4S, ⌬UA-GalNAc-6S, and ⌬UA-GalNAc-4S,6S all competed with heparan sulfate for A␤ binding with the disulfated derivative being the most effective (Table III). As was seen for competition for chondroitin sulfate GAGs, the monosaccharide D-GalNAc-4S,6S competed with high affinity with both dermatan sulfate and heparan sulfate. These results suggest that a therapeutic approach based on this structural motif may inhibit A␤ binding to all GAGs present in the central nervous system. Cumulatively, our results demonstrate that chondroitin sulfate-derived monosaccharides represent the minimal GAG subunit required for A␤ binding and that lateral aggregation between A␤ fibers or the transition of protofilaments into mature amyloid fibers requires a sulfated GAG disaccharide. These results suggest that the size constraints of the monosaccharide are insufficient to facilitate the association of fibers but are sufficient to bind A␤. Development of drugs based on these monosaccharide compounds will have to take into account the potential for stabilization of toxic A␤ intermediates. We have previously shown that A␤42 is stabilized in a nontoxic oligomer in the presence of inositol stereoisomers; this illustrates the potential for drug design based on the present methodology (28,31,41). Alternatively, GAG-derived disaccharides may represent a template in which to develop drugs that will decrease available monomer in situ by accelerating precipitation of A␤ fibers. These studies further emphasize the importance of investigations into the design of GAG memetics as potential amyloid therapeutics.