Glycosaminoglycans have variable effects on α-synuclein aggregation and differentially affect the activities of the resulting amyloid fibrils

Parkinson's disease is mainly a sporadic disorder in which both environmental and cellular factors play a major role in the initiation of this disease. Glycosaminoglycans (GAG) are integral components of the extracellular matrix and are known to influence amyloid aggregation of several proteins, including α-synuclein (α-Syn). However, the mechanism by which different GAGs and related biological polymers influence protein aggregation and the structure and intercellular spread of these aggregates remains elusive. In this study, we used three different GAGs and related charged polymers to establish their role in α-Syn aggregation and associated biological activities of these aggregates. Heparin, a representative GAG, affected α-Syn aggregation in a concentration-dependent manner, whereas biphasic α-Syn aggregation kinetics was observed in the presence of chondroitin sulfate B. Of note, as indicated by 2D NMR analysis, different GAGs uniquely modulated α-Syn aggregation because of the diversity of their interactions with soluble α-Syn. Moreover, subtle differences in the GAG backbone structure and charge density significantly altered the properties of the resulting amyloid fibrils. Each GAG/polymer facilitated the formation of morphologically and structurally distinct α-Syn amyloids, which not only displayed variable levels of cytotoxicity but also exhibited an altered ability to internalize into cells. Our study supports the role of GAGs as key modulators in α-Syn amyloid formation, and their distinct activities may regulate amyloidogenesis depending on the type of GAG being up- or down-regulated in vivo.

accumulation or colocalization of GAGs with ␣-Syn amyloid deposits suggests the involvement of GAGs in mediating protein fibrillation. Holmes et al. (43) show the internalization and propagation of Tau and ␣-Syn seeds occurs in cells through cell-surface heparan sulfate proteoglycans (HSPG). However, the influence of GAGs on ␣-Syn aggregation and its associated biological activities remains elusive. In addition, how various GAGs with a different degree of sulfation/backbone affect the pathway(s) of ␣-Syn fibrillation and structure/functions of the resultant fibrils remains unexplored so far.
Here, we used three different GAGs and related polysaccharide/polymers to delineate how different GAGs influence the ␣-Syn aggregation pathway(s) as well as the structure/activities of resultant fibrils. The in vitro aggregation of ␣-Syn in the presence of GAGs demonstrates that each GAG and related polymers uniquely modulate ␣-Syn amyloid formation. ␣-Syn in the presence of GAGs produced morphologically different and less or nontoxic amyloid fibrils, which also exhibited altered internalization ability in cells. Using two-dimensional (2D) NMR, we studied GAG-protein interactions and showed that each GAG interacts with soluble ␣-Syn uniquely in a residue-specific manner. These GAGs also interact differentially with preformed fibrils and alter the fibrillar secondary structural composition. Overall, the present study helps to delineate the role of GAGs and related polymers as modulators in amyloid formation associated with PD and other amyloidrelated diseases.

Theoretical considerations for selecting various GAGs and polymers
To explore the interaction mechanism of GAGs with ␣-Syn and their role in aggregation, we used different GAGs and related polymers including heparin (Hep), chondroitin sulfate A (CSA), chondroitin sulfate B (CSB), dextran sulfate (Dext), chitosan, and polyvinyl sulfate (PVS) polymer. The chemical structure of these GAGs consists of a disaccharide-repeating unit of amino sugar moieties and uronic acid with different degrees of sulfation (Fig. 1A). We sought to understand how the chemical structure, backbone, and orientation of the functional groups of GAGs modulate ␣-Syn aggregation and its resultant fibril properties. For instance, we used CSA and CSB, which have identical chemical formulas and similar structure, except for the orientation of the 6Ј-COOH group of the disacchariderepeating unit, which is equatorial and axial in the case of CSA and CSB, respectively. To study the importance of the electrostatic interaction and the negatively charged sulfate group, chi- Glycosaminoglycans modulate ␣-synuclein fibrillation tosan, a positively charged carbohydrate polymer, was also included in this study. Further, PVS, which contains a long hydrocarbon chain of sulfate groups but lacks sugar backbone, was used to probe the significance of the amino sugar backbone of sulfated polymers. Of note, all of the GAGs under study contain a higher density of sulfate groups compared with the heparan sulfate found in the Lewy bodies.

Modulation of ␣-Syn aggregation kinetics by different GAGs
To investigate the effect of GAG concentration on ␣-Syn aggregation, four different molar ratios of protein and GAGs were used. The molar ratios of protein:GAG were 0.25, 0.5, 1, and 2, which we termed GAG 0.25, GAG 0.5, GAG 1, and GAG 2, respectively. We expressed and purified recombinant ␣-Syn and assessed the protein purity using SDS-PAGE analysis and MALDI-TOF spectrometry (Fig. S1). For the aggregation reaction, 300 M low-molecular-weight (LMW) protein (prepared as described under "Experimental procedures"), which contains mostly monomeric ␣-Syn (44), was incubated with GAGs/ polymers at the molar ratios described above. ␣-Syn alone was also incubated as a control. The protein samples were incubated at 37°C with slight agitation, and a thioflavin T (ThT)binding assay was performed at regular time intervals. ThT binding to amyloids generally follows a sigmoidal growth curve consisting of three distinct phases: (i) lag phase (very low ThT fluorescence), where proteins slowly associate to form an aggregation-competent nuclei; (ii) growth phase (rapid increase in ThT fluorescence), where the oligomeric nucleus gradually grows into fibrils; and (iii) stationary phase (ThT fluorescence becomes saturated), where fibril formation is completed and it remains in equilibrium with the monomeric protein. The sigmoidal growth curves obtained were fitted, and lag time was calculated using previously described equations (45) (Fig. 1A). The ThT fluorescence of ␣-Syn alone showed a sigmoidal growth curve with a lag time of 68 Ϯ 3 h. At low concentrations, Hep 0.25 and 0.5 accelerated ␣-Syn aggregation with lag times of 15 Ϯ 5 and 39 Ϯ 3 h, respectively. However, Hep at higher concentrations (Hep 1 and Hep 2) delayed the ␣-Syn aggregation with lag times of 85 Ϯ 5.5 and 91 Ϯ 9.8 h, respectively ( Fig. 1, Hep). In contrast to Hep, other GAGs accelerated the rate of ␣-Syn aggregation upon increase in their concentration (Fig. 1). Along with ThT-binding assay, the time-dependent structural transition was also monitored by circular dichroism (CD) (Fig. S2). At the beginning of incubation, ␣-Syn in the absence and presence of GAGs showed a random coil structure and subsequently got converted to ␤-sheet structure at the end of the aggregation (Fig. S2). To trace the structural transition and formation of ␤-sheet, we plotted CD intensity at 218 nm (specific for ␤-sheet) with time. Intensity at 218 nm increased with the incubation time for all the samples, indicating time-dependent ␤-sheet formation (Fig. S3). Also, we calculated the relative percentage of the secondary structures formed during the aggregation by CD deconvolution using CDPro software (46,47) (Fig. S4 and Table S1). The data showed time-dependent disappearance of the random coil structure and appearance of the ␤-sheet.
To examine the effect of a positively charged GAG/polymer on ␣-Syn aggregation, we studied ␣-Syn aggregation in the presence of chitosan. Similar to other GAGs, chitosan also resulted in faster aggregation kinetics. Although at the beginning of the incubation, ThT fluorescence increased rapidly, after some time, it formed a gel-like state that impeded our further measurement of ThT fluorescence (data not shown) and CD spectroscopy. However, at the end of the aggregation, we diluted and redissolved the gel in a higher volume of 20 mM Gly-NaOH buffer, pH 7.4 (containing 0.01% sodium azide) by vortexing. We recorded the CD spectra on day 10, where we found the presence of a ␤-sheet structure (Fig. S5). We further tested the effect of PVS and observed that it accelerated ␣-Syn aggregation rapidly. ␣-Syn at 300 M concentration incubated with PVS formed a ␤-sheet almost instantaneously and showed very high ThT binding to ␣-Syn aggregates. Therefore, using this high protein concentration (300 M), it was difficult to follow aggregation kinetics in the presence of PVS. Hence, we incubated 10 M ␣-Syn in the presence of different molar ratios of PVS. At a low molar ratio of PVS (PVS 0.25), no increment in ThT was observed even after 6000 min of incubation (Fig. 1B, PVS). Although in the case of PVS 0.5 and PVS 1, ␣-Syn showed nucleation-dependent polymerization, the aggregation rate was significantly higher compared to all other GAGs under study, despite the low ␣-Syn concentration used for PVS-mediated aggregation. In addition, PVS 2 completely eliminated the lag phase of aggregation, and after 10 min of incubation, it reached a plateau phase (Fig. 1B, PVS).
Taken together, all GAGs except heparin increased the aggregation rate of ␣-Syn with an increase in their concentrations. Heparin at low molar ratios (Hep 0.25 and 0.5) accelerated ␣-Syn aggregation, whereas at higher molar ratios (Hep 1 and 2), it delayed the aggregation process. To probe whether the nucleation of ␣-Syn aggregation is delayed at a high molar ratio of Hep, we performed static light scattering at 450 nm and time-dependent atomic force microscopy (AFM) during aggregation in the presence of Hep 0.25 and Hep 2 (Fig. 2, A and B). Static light scattering data (similar to ThT fluorescence kinetics) revealed that heparin at high concentrations (Hep 2) exhibited a longer lag phase compared to ␣-Syn alone and Hep 0.25 ( Fig. 1). Time-dependent morphological analysis by AFM showed that oligomer formation was delayed in the case of Hep 2 compared to both ␣-Syn control and Hep 0.25 (Fig. 2, A and  B). The inability of ␣-Syn to attain spontaneous nucleation in the presence of Hep was further confirmed by the addition of sonicated preformed fibrils (1 and 2% seed) into the ␣-Syn-Hep 2 reaction mixture after ϳ65 h of incubation. In the presence of seeds, a rapid increase in ThT binding was observed for ␣-Syn aggregation in the presence of Hep 2 (Fig. 2C). Therefore, the present data convincingly establish that higher concentration of Hep inhibits ␣-Syn aggregation by preventing the formation of the aggregation-competent oligomeric nucleus. Intriguingly, this phenomenon was observed only for heparin, and the trend was not observed for other GAGs under study. However, it is not clear at this point as to why this mechanism is unique only for heparin. It conceivably could be because its distinct backbone with negative charge density/distribution might influence ␣-Syn aggregation uniquely.
In the presence of CSB, the aggregation kinetics measured by ThT fluorescence showed three-state transitions, unlike other Glycosaminoglycans modulate ␣-synuclein fibrillation GAGs, which showed two-state kinetics (Figs. 1B, CSB, and 2D). It is possible that CSB stabilizes the aggregation intermediate in the middle of the kinetics observed as a temporary plateau phase, which eventually converts into fibrils. As shown in Fig. 2D, the CD data for this intermediate showed the presence of a stabilized helix-rich intermediate (48). This was further confirmed by AFM, which showed a globular oligomer-like structure at the beginning and end of the intermediate state (Fig. 2D, right  panel). Interestingly, this behavior was observed only for CSB and not for CSA, which shares structural similarities with CSB. This could be because of the unique CSB-␣-Syn interaction.
Further, to determine the degree of aggregation of ␣-Syn in the presence of GAGs, we analyzed the soluble protein content post-aggregation using 1D NMR ( Fig. S6 and Table S2). The difference between the starting concentration of protein used for fibrillation and the soluble protein left after aggregation (obtained by centrifugation) provided the degree of ␣-Syn fibrillation. The relative protein concentration after aggregation was calculated from the 1D NMR signal intensity. The results showed that 65% of the ␣-Syn fibrilized in the absence of GAGs/polymers, whereas in the presence of GAGs, the degree of aggregation increased by 20 -30% (Table S2). As ␣-Syn in the presence of chitosan formed a gel-like state upon incubation, it showed a very low soluble protein concentration in the super-natant. This could either be because of the high fibrillation rate in the presence of chitosan or entrapment of the remaining monomeric protein in the gel state.
To examine the differences in the oligomerization states of ␣-Syn in the presence of GAGs, the supernatant (obtained after centrifugation of the fibril solution) was analyzed by dynamic light scattering (DLS). The data showed that soluble ␣-Syn (obtained after fibril formation) exhibited higher-order soluble oligomers as evident from increased hydrodynamic radii. However, the oligomer size distribution profile varied among different GAGs (Fig. S7). This suggests that in the presence of GAGs/ polymers, equilibrium is shifted toward the formation of higher-order soluble aggregates, but the extent of oligomerization can vary depending upon the type of the GAG (Fig. S7).
In addition, to check the integrity of the protein in the absence and presence of GAGs/polymers, we performed SDS-PAGE analysis of the fibril samples after 10 days of incubation. The fibril samples did not show any degradation products (Fig.  S8), confirming the quality and integrity of all of the fibrils.

Effect of polymer length on ␣-Syn aggregation
Polymer length is one of the important factors that influence protein aggregation (13). We observed that with the increasing polymer length of the GAGs under study, the lag phase of ␣-Syn Glycosaminoglycans modulate ␣-synuclein fibrillation aggregation was decreased (Fig. S9, GAG 1 and GAG 2). This is in accordance with previous studies suggesting that polymers/ GAGs with less disaccharide units have less effect on fibrillation than GAGs with longer chain lengths (49,50). However, we found that this correlation was affected by several other factors such as protein:GAG molar ratio, charge density, the orientation of sulfate groups, etc. For instance, at 1:1 and 1:2 ratios of protein: GAG, with increasing polymer length, the lag phase of ␣-Syn aggregation is decreased (Fig. S9). However, at a low protein:GAG ratio (GAG 0.25), this correlation does not exist. Hep 0.25, despite being a low-molecular-weight GAG, accelerates the ␣-Syn aggregation kinetics with a lag time of 15 Ϯ 5 h, which is the shortest lag time among all GAGs studied at the same ratio (GAG 0.25).
Moreover, CSA and CSB have the same chain length but exhibit different aggregation patterns because of the different orientation of the charged groups on the sugar moiety. Previous findings by Antony et al. (13) suggest that polyamines accelerate ␣-Syn aggregation, depending on total charge, length, and concentration of the polyamine. However, in a later study of protein-polyamine complexes by NMR, they found that a greater polyamine charge, i.e. ϩ2 to ϩ5, correlates with increased affinity and aggregation propensity of ␣-Syn in the presence of polyamines, whereas variable chain length is of little or no importance (51). A study by Valle-Delgado et al. (22) emphasizes that the distribution of charged groups on GAGs is also one of the critical factors affecting A␤ aggregation. Therefore, the effect of GAGs on ␣-Syn aggregation kinetics could be because of the cumulative effect of concentration, density/orientation/distribution of charged groups, polymer length, and GAG backbone.

Residue-specific interaction between GAG and ␣-synuclein
We proposed that the observed differences in aggregation could be because of the different interaction patterns of GAGs/ polymers with ␣-Syn. To investigate this possibility, we performed 2D NMR using 15 N-labeled ␣-Syn in the presence of GAGs, with only ␣-Syn used as a control. ␣-Syn contains three distinct domains: (i) an amphiphilic N terminus, which binds to the negatively charged phospholipids; (ii) a central hydrophobic domain, which is responsible for aggregation; and (iii) a C terminus that is negatively charged and does not have any structural preference. To study the mode of interaction of ␣-Syn with GAGs/polymers, we recorded the 2D 1 H-15 N heteronuclear single-quantum coherent (HSQC) spectra in the presence and absence of GAGs. Previously published assignment values from the Biological Magnetic Resonance Bank (BMRB entry 16543) were used to assign the HSQC spectrum, which was further confirmed using 3D TOCSY-HSQC experiments. 15 N-Labeled ␣-Syn samples were titrated with various concentrations of GAGs, and a series of HSQC spectra were recorded to monitor the chemical shift perturbations originated as a result of the interactions. A quantitative analysis of the extent of interaction was done by calculating the chemical shift perturbations using the formula ((5⌬␦ 1 H N) ) 2 ϩ (⌬␦ 15 N 2 ) 2 ) 1/2 (45).
Hep at low concentrations (Hep 0.25 and 0.5) accelerate ␣-Syn aggregation, whereas higher concentrations (Hep 1 and 2) delay the fibrillation process. Therefore, we recorded NMR spectra of 15 N-labeled ␣-Syn in the presence of Hep at molar ratios of 0.25 and 2. The data indicate that the extent of interaction increases with the increase in Hep concentration. Higher peak shifts were observed for residues Lys-10, Lys-32, Lys-45, and Lys-60, which could be because of increased charge-based interactions (Figs. 3A and B, and S10). Significant perturbations were also observed for residues Ser-9, Thr-22, Thr-33, Thr-59, Thr-64, Thr-75, Thr-92, and Tyr-125, which could be because Glycosaminoglycans modulate ␣-synuclein fibrillation of hydrogen bonding between these residues and the oxygen of the Hep backbone. Moreover, residues from the N terminus showed more perturbation than those of the C terminus. This indicates that negatively charged Hep binds preferentially to the positively charged residues in the N terminus of ␣-Syn. CSA consists of disaccharide-repeating unit, and each unit contains two negatively charged sulfate ions on the same side of the polymeric chain (Fig. 1A). The observed chemical shift perturbations for CSA were less than for Hep. In fact, CSA did not show any preferred region for the binding and interacted throughout the protein (Figs. 3B and S10). CSA and CSB share identical chemical formulas and the same structure except for the orientation of the 6Ј-COOH group of the disaccharide-repeating unit. CSB also exhibited a pattern of interaction with ␣-Syn similar to that of CSA. However, Val-3, Ser-9, Ala-11, Lys-12, Glu-20, Gly-25, His-50, Ala-78, Tyr-125, and Tyr-133 showed noticeable chemical shift perturbations in the presence of CSB compared with CSA (Figs. 3B and S10). Furthermore, Dext exhibited larger interactions with ␣-Syn compared with other GAGs. It displayed strong interactions with the first 60 residues of the N terminus. The chemical shift perturbations at the C terminus and residues in the non-amyloid-␤ component region (70 -80) was also observed, albeit to a lesser extent. The observed high chemical shift differences at Lys-10, Lys-21, Lys-23, Lys-45, and Lys-60 could be because of the chargebased interaction between protein and Dext. In contrast to other GAGs, chitosan, a positively charged polymer, showed a strong preference for residues in the C terminus (110 -140) (Figs. 3B and S10). Significant chemical shift perturbations were observed in the negatively charged residues such as Glu-110, Glu-114, Asp-119, Asp-121, Glu-123, Glu-130, and Glu-131 because of the multiple charge-based interactions. The change in the chemical shift at Asn-122, Tyr-126, and Tyr-133 could be because of the hydrogen bonding between the free hydroxyl group of the polymeric chain and the ␣-Syn protein. Also, significant chemical shift perturbations were observed in the region of residues 51-53 ( Fig. S10). As discussed earlier, ␣-Syn at a high concentration (300 M) precipitates in the presence of PVS, and so NMR spectra could not be recorded for this sample. Taken together, the NMR data suggest that GAGs interact with a soluble form of protein through charge and/or hydrophobic interactions. These interactions increase the local concentration of ␣-Syn (as a result of a crowding effect) that leads to accelerated aggregation kinetics. Moreover, GAGs with a higher number of negative charges in the disaccharide-repeating unit showed stronger interactions than GAGs containing a lesser number of net negative charges. For instance, CSA and CSB have two negative charges, whereas Hep and Dext have four and six negative charges per disaccharide-repeating unit, respectively. This demonstrates that the extent of GAGprotein interaction increases with an increase in the net negative charge on the GAG backbone.

GAGs and related polymers modulate structure, morphology, and toxicity of ␣-Syn fibrils
The morphology of ␣-Syn fibrils formed in the presence of different GAGs and polymers was analyzed using transmission EM (TEM). ␣-Syn fibrils in the absence of GAGs/polymers showed an average fibrillar diameter of 13 Ϯ 2.2 nm comprising two to three filaments with turns and twists (Fig. 4, A, left panel, and B, and S11). In contrast, the fibrils formed in the presence of GAGs were relatively thinner and displayed morphological variations. ␣-Syn aggregates in the presence of Hep 0.5, 1, and 2 and Dext at all concentrations showed thin fibrils of ϳ6 -8 nm, whereas in the presence of Hep 0.25, ␣-Syn fibrils were relatively thicker, at 11 Ϯ 2.7 nm (Figs. 4, A, left panel and B, S11, and S12). ␣-Syn in the presence of CSA and CSB at all concentrations formed fibrils of ϳ5-6 nm in diameter. Laterally associated structures comprising four to six filaments were observed in the case of CSA 1 (Figs. 4, A, left panel, and B, and S11). ␣-Syn in the presence of chitosan showed significantly altered fibrillar morphology. Only small oligomeric structures (11 Ϯ 2.4 nm diameter) and very short fibrils were seen in chitosan 0.25, whereas laterally associated filaments (ϳ2-3 filaments of ϳ8 -9 nm in diameter) were observed for chitosan 0.5, 1, and 2 (Figs. 4, A, left panel, and B, and S12). In the case of PVS, ␣-Syn did not form any fibrillar structures but rather formed mostly amorphous, large clumps of aggregates (Figs. 4A, left panel, and S12). Furthermore, the secondary structures of the ␣-Syn fibrils formed in the presence of GAGs at equimolar concentrations (GAG 1) were characterized by FTIR. To this end, 10 l of ␣-Syn fibrils formed (after 10 days of incubation) in the absence and presence of various GAGs was spotted on a KBr pellet and dried under an IR lamp. The FTIR spectra were then acquired, processed, and analyzed as described under "Experimental procedures." The ␣-Syn fibrils (formed in the absence of any GAGs) mostly showed cross-␤-sheet conformation with a strong absorption peak around 1620 to 1640 cm Ϫ1 and a lowintensity peak around 1693 cm Ϫ1 (Fig. 4A, middle panel). ␣-Syn fibrils formed in the presence of CSB or Dext showed secondary structures similar to that of the ␣-Syn fibrils alone, with predominant absorption peaks characteristic of ␤-sheet conformation (in between 1620 -1640 cm Ϫ1 and 1690 cm Ϫ1 ). The ␣-Syn fibrils formed in the presence of Hep, PVS, and CSA displayed a significant increase in helical content (1654, 1650, and 1650 cm Ϫ1 , respectively) along with the cross-␤-sheet content. Major secondary structural changes were obtained in the case of chitosan, where helix (1650 cm Ϫ1 ) was the prominent secondary structure (Fig. 4A, middle panel). Further, to probe any alterations in the molecular structure of fibrils in the presence of GAGs, X-ray diffraction was performed (Fig. 4A, right panel). Differences in the packing and alignment pattern of amyloid fibrils can be obtained through diffraction data. The X-ray diffraction pattern of ␣-Syn fibrils in the presence of GAGs (GAG 1) showed ϳ4.7-Å meridional reflection, which represents the distance between two ␤-strands. However, in the case of ␣-Syn fibrils formed in the presence of PVS, a close doublet ϳ4.9 and ϳ4.7 Å was observed. ␣-Syn fibrils (in the absence of any GAGs) showed equatorial reflections ϳ8.2 and ϳ10 Å, which corresponds to the distance between two ␤-sheets in a cross-␤-sheet motif of amyloid fibrils (52). We observed variations in the equatorial reflections in each case, as depicted in Fig. 4A, right panel. Overall, the data suggest that GAGs contribute to remodeling of amyloid fibrils and alter the internal packing and arrangement of protofilaments in the fibrils. These structural Glycosaminoglycans modulate ␣-synuclein fibrillation variations may also reflect some fundamental differences in the biological properties of these fibrils.
Although ␣-Syn oligomers are more toxic than fibrils (53,54), various studies have suggested the toxicity of ␣-Syn fibrils and their potential role in PD (55)(56)(57). Next, we determined the toxicity of GAG-mediated ␣-Syn fibrils using an MTT reduction assay. The MTT assay is a method that is used routinely to measure the toxicity of extracellular amyloids (i.e. ␤-amyloid) as well as for intracellular amyloid-like ␣-Syn (53,54). For the MTT assay, fibrils formed at the end of aggregation (without sonication or other treatments) were added to the cultured SH-SY5Y cells at a concentration of 25 M. ␣-Syn fibrils and GAGs alone were used as controls. ␣-Syn fibrils formed in the absence of GAGs showed ϳ55% cell viability (Fig. 4C). ␣-Syn Glycosaminoglycans modulate ␣-synuclein fibrillation fibrils in the presence of Hep, CSA, CSB, and Dext formed less or nontoxic fibrils, and cell viability was slightly increased with the increase in GAG:protein stoichiometry (Fig. 4C). ␣-Syn fibrils in the presence of a lower concentration of chitosan exhibited a similar toxicity to that of ␣-Syn fibrils alone. However, ␣-Syn fibril toxicity decreased with an increase in chitosan concentration (chitosan 2). In the case of PVS, the toxicity of the resultant ␣-Syn aggregates increased with an increase in PVS concentration.

Effect of GAGs on preformed fibrils
We further explored the effect of GAGs/polymers on preformed ␣-Syn fibrils. To study this effect, the preformed ␣-Syn fibrils were incubated in the presence of GAGs, chitosan, and PVS (1:1 ratio, 150 M) for 1 h at room temperature. The incubation of the preformed fibrils with GAGs/polymers showed an increased binding with ThT except for chitosan, CSA, and CSB. In the case of Dext and PVS, a significant increase in ThT fluorescence was observed. A similar rise in the ThT fluorescence signal was also seen in the presence of Hep but to a lesser extent compared to Dext and PVS. ThT binding remained unaltered for CSA and CSB. Intriguingly, in the presence of chitosan, an almost 10-fold decrease in ThT was observed (Fig. 5A). The change in ThT fluorescence intensity in the presence of GAGs could have three reasons: (i) a change in the internal structure (remodeling) of ␣-Syn fibrils; (ii) a change in the amount of fibrils (i.e. shifting of the equilibrium between existing monomers toward fibrils; and (iii) the emergence of oligomers or Glycosaminoglycans modulate ␣-synuclein fibrillation other structural species. To study the possibility of change in the internal structure, CD and FTIR spectroscopy were carried out. CD spectroscopy of preformed ␣-Syn fibrils in the presence of GAGs did not show any significant change in ␤-sheet conformation, except for chitosan, where the ␤-sheet content decreased significantly (Fig. 5B). Structural differences in ␣-Syn fibrils due to the addition of GAGs/polymers were further analyzed by FTIR spectroscopy. The FTIR spectra were recorded in the range of the amide-1 region (1600 -1700 cm Ϫ1 ). ␣-Syn fibril showed peaks at 1626 and 1687 cm Ϫ1 suggestive of ␤-sheet content and 1662 cm Ϫ1 for ␤-turn. Upon the addition of Hep, along with the ␤-sheet, a helix peak was also observed (1655 cm Ϫ1 ). Preformed ␣-Syn fibrils in the presence of CSA showed ␤-sheet (peaks at 1625, 1633, and 1682 cm Ϫ1 ), ␤-turn (1671 cm Ϫ1 ), and a significant amount of helical content (1655 cm Ϫ1 ). Identical structural components were also observed in the presence of CSB, consisting of ␤-sheet (1624, 1633, and 1686 cm Ϫ1 ), ␤-turn (1669 cm Ϫ1 ), and helix (1657 cm Ϫ1 ). ␣-Syn fibril incubated with Dext showed high helical content (peak at 1650 cm Ϫ1 ) along with ␤-sheet (1623, 1632, and 1697 cm Ϫ1 ), and ␤-turn (1671 and 1679 cm Ϫ1 ). ␣-Syn fibrils in the presence of PVS showed high ␤-sheet (1623, 1633, and 1682 cm Ϫ1 ) and ␤-turn (1671 cm Ϫ1 ) content, while some helical character (1652 cm Ϫ1 ) also existed. Unlike other GAGs, chitosan displayed a significant change in fibrillar structure with the high amount of helical content (1655 cm Ϫ1 ) and reduced ␤-sheet (Fig. 5C). Overall, the variations in the secondary structural components observed by FTIR indicate remodeling of the ␣-Syn fibril structure in the presence of GAGs.
Secondly, the increase or decrease in ThT binding because of the addition of GAGs could also be because of the shifting of the monomer-fibril equilibrium toward fibrils, as stated above. Because after aggregation fibrils remain in equilibrium with monomers in the stationary phase, it is possible that the coexisting monomers in the fibrillar solution may get converted to fibrils in the presence of GAGs, which in turn would lead to increased ThT binding. To test this possibility, preformed ␣-Syn fibrils (post-incubation with GAGs/polymers) were centrifuged at 14,000 rpm for 1 h. The resultant fibril pellets in each case were redissolved and analyzed on 15% SDS-PAGE. Thereafter, using ImageJ software, densitometric analysis of the protein bands was performed, and the average intensities were calculated (Fig. 5, D and E). Strikingly, we found that there was no significant change in the amount of protein in fibrils formed after incubation with GAGs (except for chitosan). This suggests that GAGs (excluding chitosan) do not increase the amount of fibrils by monomer-fibril equilibrium shift. However, in the presence of chitosan, fibril content decreased significantly consistent with low ThT binding.
Lastly, we examined the possibility that because of the addition of GAGs in fibril solution, new oligomers and/or structural species might emerge resulting in an altered ThT binding. To test for this possibility, we performed DLS of the supernatant obtained after centrifugation of fibrils treated with GAGs. We found that ␣-Syn supernatant obtained from fibril solutions that were treated with all GAGs (except chitosan) showed an overlapping profile of oligomer size distribution with that of an untreated sample (Fig. 5F). This indicates that there was no emergence of new oligomers or structural species. However, in the case of chitosan, the emergence of higher-order soluble oligomers was observed. This indicates that the preformed fibrils that were treated with chitosan probably dissociated into a new type of oligomeric species, consistent with the ThT-binding assay and SDS-PAGE analysis of fibrils (Fig. 5, A and D). To further confirm these observations, the preformed fibrils incubated with chitosan were observed under AFM. Indeed, AFM revealed the disappearance of fibrillar morphology and the formation of oligomeric species (Fig. S13). Collectively, the data suggest that GAGs might not influence the monomer-fibril equilibrium shift or promote the emergence of new oligomers. Instead, treatment of preformed fibrils with GAGs modulates the internal fibril structure, as evident from FTIR, which in turn results in an altered ThT binding.
It is notable that irrespective of the structural change similar to Hep, the ThT fluorescence of the preformed ␣-Syn fibril treated with CSA and CSB remains unaltered (Fig. 5A). This could be because of fibril structure modulation by these GAGs resulting in an ineffective ThT binding. It is also possible that the presence of CSA and CSB in the fibril solution hinders the ThT binding to the fibrils. To validate this possibility, we performed a ThT-binding assay of freshly dissolved fibrils immediately upon the addition of CSA/B. We found that the ThT binding remained unaltered in the preformed fibrils alone sample and upon the addition of CSA/B (Fig. S14). This confirmed that CSA/B does not interfere with the binding of ThT to the fibrils. Therefore, the absence of higher ThT binding of preformed fibrils when incubated in the presence of CSA and CSB could be caused by structural modulation, as observed by FTIR spectroscopy (Fig. 5C).

Altered cellular uptake of ␣-Syn fibrils mediated by GAGs and polymers
Recent studies suggest that ␣-Syn amyloid fibrils are infectious, given that the exogenous addition of ␣-Syn fibrils leads to the formation of intracellular Lewy body-like inclusions in cells by seeding the endogenous soluble ␣-Syn (58). This secondary nucleation of ␣-Syn fibrils is not only responsible for increasing the fibril load and spreading the pathology inside the cells but may also result in the formation of toxic oligomers during this process (59). Because we found that different biopolymers interacted with soluble ␣-Syn or preformed ␣-Syn fibrils and produced structurally different fibrils, we speculated that they could also lead to altered cellular uptake/internalization. To study this hypothesis, FITC-labeled fibrils were prepared in the absence and presence of different GAGs. Next, FITC-labeled fibril seeds were generated by sonicating the fibrils for 3 min at 20% amplitude (pulse 3 s on/1 s off). The fibrillar morphology of the labeled fibrils before and after sonication was analyzed by TEM (Fig. S15A). The fibril seeds (for the fibrils formed in the presence of GAGs) were found to be of similar length (Fig.  S15B). Thereafter, GAG-mediated fibrillar seeds were allowed to internalize in SH-SY5Y cells at an optimized concentration (1 M) and time (24 h) (Fig. S16). We observed that CSA-, CSB-, and PVS-mediated fibril seeds were readily taken up by the cells and appeared as punctate structures inside the cell body (Fig. 6,  upper panel). On the contrary, seeds from fibrils formed in the Glycosaminoglycans modulate ␣-synuclein fibrillation

Glycosaminoglycans modulate ␣-synuclein fibrillation
presence of Hep and Dext did not internalize (Fig. 6, lower  panel). Chitosan showed entirely different internalization behavior among all other GAGs. Fibril seeds formed in the presence of chitosan did not internalize but got stuck on the surface of the cell membrane (Fig. 6, lower panel). Further, to quantify fibril internalization, fluorescence-activated cell sorting (FACS) was performed. FACS data suggested that internalization was maximum in the case of PVS (99.9%), closely followed by CSA (98.5%), ␣-Syn only (92%), and CSB (80%). In the case of chitosan, 91% of the cells showed a FITC-positive signal, but this signal was from chitosan fibrils adhered to the cell surface as observed in fluorescence bright field images. In contrast, Dext-and Hep-mediated fibrils showed a ϳ99% FITCnegative cell population (Fig. 6). These results suggest that GAGs and related polymers uniquely alter the internalization and cellular uptake of ␣-Syn fibrils.
To further explore the possibility that the amount and type of amyloid co-factors (in this case, GAGs) can also dictate the internalization, we estimated the amount of GAG incorporated within the fibrils with Alcian blue assay (60,61). We first generated the standard concentration curve by Alcian blue assay (Fig. S17) for each GAG (Hep, Dext, CSA, and CSB) using an established protocol (61) with slight modifications. From this standard curve, we analyzed the amount of GAGs remaining in the supernatant after fibril formation and calculated the percentage of GAGs incorporated within the fibrils. To do that, we centrifuged the fibrils (formed in the presence of each GAG at a protein:GAG molar ratio of 1:1, 300 M concentration) at 14,000 ϫ g for 1 h and removed the supernatant, which was then used for Alcian blue assay. The data showed that the amount of Hep, Dext, CSA, and CSB remaining in the supernatant (after fibril formation) was 210, 114, 107, and 97 M, respectively. Therefore, the percentage of Hep, Dext, CSA, and CSB incorporation within the fibrils was 30, 38, 64, and 67.5%, respectively (Table S3). Because the Alcian blue assay is only specific for sulfated GAGs, we could not determine the percentage of incorporation of PVS and chitosan in the fibrils. On comparing the amount of GAG incorporated in fibrils and the extent of fibril internalization, we found that less incorporation of Hep (30%) and Dext (38%) in the fibrils resulted in negligible or no cellular internalization, whereas high incorporation of CSA (64%) and CSB (67.5%) in the fibrils exhibited more internalization. The data suggest that the type and amount of GAG incorporated in the fibrils might also influence the surface topology and structure of the fibrils in a way that affects the internalization of fibril in cells.

Discussion
Glycosaminoglycans are important cellular components that have been reported to co-deposit with amyloids associated with various diseases including Parkinson's (24,25,62). In this study, we investigated how different GAGs and related polymers regulate the aggregation kinetics of ␣-Syn and the associated biological activity of the end-stage fibrils. Our data demonstrate that different GAGs not only modulate ␣-Syn aggregation kinetics but also their stoichiometry can affect the overall fibrillation pathway (Fig. 1). For all GAGs (except Hep), the rate of aggregation kinetics increases with an increase in GAG concen-tration. Intriguingly, Hep at low concentrations accelerates ␣-Syn aggregation, whereas at high concentrations it delays ␣-Syn aggregation and amyloid formation (Fig. 1B, Hep). At low concentration of heparin, it enables ␣-synuclein to form an aggregation-competent nucleus (where heparin acts as a template for the nucleus formation), resulting in faster aggregation. In contrast, at high concentrations it might form an aggregation-incompetent protein-heparin complex, possibly resulting in the formation of off-pathway intermediates to some extent, along with the on-pathway intermediate species. Therefore, the absence of an aggregation-prone nucleus in the case of the higher molar ratio of heparin significantly reduces the nucleation event and thereby the overall aggregation process. A similar observation is also reported by Ramachandran and Udgaonkar (38) regarding the Tau fibrillation pathway. This outcome implies that the intracellular concentration of Hep may be one of the controlling factors in ␣-Syn aggregation.
Although both CSA and CSB, with identical backbones and similar charges, accelerated ␣-Syn aggregation, CSB did not follow the classical sigmoidal aggregation kinetics (Figs. 1B and 2D). The biphasic kinetics behavior of CSB highlights that not only the backbone and charge but also the orientation of the functional group play a significant role in modulating the aggregation pathway of ␣-Syn. Similar findings also have been made for other amyloid protein/peptides, where the orientation of the sulfate groups is shown to have a different effect on A␤ aggregation (22). PVS, which does not contain any sugar backbone, accelerated ␣-Syn aggregation at an exceptionally high rate. This tendency could be because of the flexibility of the PVS backbone along with the high charge density of sulfate ion. The aggregates formed, however, lacked typical amyloid fibrillar morphology and showed amorphous-like globular structures (Figs. 4A and S12) with ␤-sheet conformation and ThT-binding capability (Fig. S2, PVS). In contrast to negatively charged GAGs and polymer, positively charged chitosan also accelerated the ␣-Syn kinetics and formed fibrils with high lateral association.
We have demonstrated that interactions between GAGs and soluble ␣-Syn govern the modulation of ␣-Syn aggregation. The 2D NMR studies revealed site-specific information, where each GAG showed a differential interaction with the soluble protein (Figs. 3 and S10). The NMR data of Hep and Dext binding with soluble ␣-Syn showed high chemical shift differences at the N terminus compared with the C terminus, suggesting preferential binding of GAG at the N terminus. Although low, the chemical shift perturbations were also observed throughout the ␣-Syn sequence. This suggests that Hep can also interact with the non-amyloid-␤ component region and the C terminus of ␣-Syn, or alternatively, higher binding to the N-terminal region and a subsequent change in ␣-Syn dynamics may also account for this observation. However, in the case of CSA and CSB, the chemical shift perturbations were found throughout the sequence, suggesting that these GAGs either interact nonspecifically with the complete ␣-Syn sequence or bind to the N terminus and influence the dynamics of the entire protein (63,64). In contrast, chitosan showed a high level of perturbation in the C terminus, comprising mostly acidic residues, and facilitated the aggregation by restricting the flexibility of this domain (51). Moreover, as per the available reports, the C terminus of Glycosaminoglycans modulate ␣-synuclein fibrillation ␣-Syn plays a protective role against ␣-Syn aggregation (65)(66)(67). It has been shown that C-terminal truncation results in faster fibril formation and this C-terminal domain regulates the fibrillation of ␣-Syn (65). GAGs and related polymers, therefore, accelerate the aggregation of ␣-Syn by perturbing the long-range interactions (either by directly binding to C/N terminus or indirectly by influencing the structural dynamics). Further, we found that each GAG uniquely modulates the structure and morphology of the resultant fibrils. Structural variation might be the cause of less toxicity observed for the fibrils formed in the presence of GAGs (except PVS and chitosan) (Fig. 4C). This suggests that GAGs not only can modulate the structure of ␣-Syn fibrils but also have the ability to diminish their inherent toxic property (Fig. 4).
GAGs are integral components of the extracellular matrix and may come in contact with amyloid fibrils from cell inclusion bodies that get released into the extracellular environment (68). This finding prompted us to study the effect of GAGs on preformed fibrils. Strikingly, we found that GAGs remodel the internal fibril structure (Fig. 5). However, contrasting results were obtained in the case of chitosan, which dissociated the preformed fibrils into oligomeric species. We noted that chitosan incubated with ␣-Syn monomer resulted in the formation of fibrils, but its effect on preformed fibrils was different. A possible explanation could be that chitosan upon interaction with the C terminus in the monomeric state minimizes the electrostatic repulsion between the negatively charged ␣-Syn molecules, thereby increasing the local concentration of the protein and facilitating fibril formation. However, dissociation of fibrils occurred on incubating chitosan with preformed fibrils. It is possible that the interaction of positively charged chitosan with the negatively charged C terminus of preformed ␣-Syn fibrils could be much stronger compared with inter-protein interactions, resulting in fibril disintegration (Fig. 5).
Previously it has been shown that fibrils of prion and Tau protein internalize through cell-surface HSPGs (43,69). Also, it is established that ␣-Syn amyloid fibrils can internalize and seed their endogenous counterparts in cells (58). We studied whether different GAGs alter the uptake of ␣-Syn fibrils in cells. The fibrils formed in the presence of Hep and Dext did not internalize, whereas a different extent of internalization was observed with fibrils formed in the presence of CSA, CSB, and PVS (Fig. 6). This shows that the GAG-mediated alteration of fibrillar structure could affect the ability for cellular internalization of amyloid fibrils. In addition, previous studies have shown that Hep attenuates the cellular binding and uptake of Tau, ␣-Syn, and A␤ fibrils (43). Hep binds directly to cellular HSPGs and prevents the internalization of amyloid fibrils (43,70). Apart from the attenuation of the receptor-mediated uptake of fibril in the presence of specific GAGs (Hep and its mimetics), the amount and type of GAGs incorporated in the fibrils might also dictate the cellular internalization.
In conclusion, we have demonstrated the diverse mechanisms exhibited by different GAGs on ␣-Syn fibrillation. We showed that subtle changes in GAG backbone and charge density lead to significant alteration in the fibril property, which could direct the fibrillation process in vivo. Thus, our findings significantly contribute toward unfolding the role of GAGs in the etiology of Parkinson's disease and associated synucleinopathies.

Chemicals and reagents
All chemicals used in the study were obtained from Sigma except for low-molecular-weight heparin (purchased from Thermo Fisher Scientific) and 15 N-labeled ammonium chloride from Cambridge Isotope Laboratories, Inc. The Milli-Q system (Millipore Corp., Bedford, MA) was used for double distillation and deionization of water.

Protein expression and purification
␣-Syn plasmid (pRK172) (71) was used for protein expression. ␣-Syn protein was expressed and purified as per the established protocol described by Volles and Lansbury (72) with slight modifications (44,73).

Preparation of low-molecular-weight (LMW) ␣-Syn
Low-molecular-weight protein (consisting mostly of the monomeric species) was prepared for all the aggregation studies. LMW was prepared according to the previously described method (48,73).

Amyloid fibril formation
Aggregation of ␣-Syn was performed in the presence of Hep, Dext, CSA, CSB, chitosan, and PVS at four different concentrations. Low-molecular-weight Hep (ϳ5 kDa), Dext (ϳ17 kDa), CSA (ϳ18 kDa), CSB (ϳ18 kDa), and PVS (ϳ170 kDa) were dissolved in 20 mM Gly-NaOH buffer, pH 7.4 (containing 0.01% sodium azide). Chitosan (ϳ11.9 kDa) was dissolved in prewarmed 20 mM Gly-NaOH buffer by adding a few drops of glacial acetic acid. The pH of all the GAG solutions (except for chitosan) was then adjusted to 7.4 by adding a few microliters of 2 N NaOH. The pH of chitosan was adjusted in the range of 6.5 to 7.0, as it forms a gel and precipitates out at pH 7.4. The aggregation kinetics of 300 M ␣-Syn was carried out with different concentrations of GAG, and only ␣-Syn was used as a control. Four different molar ratios of ␣-Syn and GAG (1:0.25, 1:0.5, 1:1, and 1:2) were used. The protein solutions were kept in microcentrifuge tubes for rotation at 50 rpm in an EchoTherm model RT11 rotating mixer (Torrey Pines Scientific) in a 37°C incubator. Fibril formation was monitored continuously by ThT fluorescence assay and secondary structural changes by CD spectroscopy. The morphology of the end-stage fibrils was analyzed by TEM.

ThT fluorescence assay
All samples kept for incubation were diluted to get the concentration of 7.5 M in 200 l of 20 mM Gly-NaOH buffer, pH 7.4 with 0.01% sodium azide followed by the addition of 2 l of 1 mM ThT solution. The stock of 1 mM ThT was prepared in 20 mM Tris-HCl buffer, pH 8.0, containing 0.01% sodium azide before use. The ThT fluorescence assay was done at an excitation wavelength of 450 nm and emission in the range of 465 to 500 nm with a slit width of 5 nm (both for excitation and emis-Glycosaminoglycans modulate ␣-synuclein fibrillation sion) using a FluoroMax-4 spectrofluorometer (HORIBA-Jobin Yvon). The ThT fluorescence obtained for all the samples was plotted at 480 nm with respect to incubation time, and Origin 8.0 software was used to fit the data to a sigmoidal curve.
The calculations for lag time were done as per the published protocol (45) using the following equation, where y refers to ThT fluorescence at a particular time point, y max refers to the maximum ThT fluorescence, y 0 refers to ThT fluorescence at t 0 , and t lag ϭ t1 ⁄ 2 Ϫ 2/k.
The aggregation kinetics was repeated three times, and a representative single set is shown of three sets performed independently. Lag time calculations show mean Ϯ S.D. values.

CD spectroscopy
The secondary structural transition during aggregation by ␣-Syn alone and ␣-Syn in the presence of various GAGs and polymers was monitored using CD spectroscopy. For spectral measurement, a quartz cell (Hellma Cells, Forest Hills, NY) with a 0.1-cm path length was used to contain 200 l of 7.5 M protein samples (prepared in 20 mM Gly-NaOH, pH 7.4, with 0.01% sodium azide). Subsequent spectra were acquired using a CD spectrometer (JASCO-810) in the wavelength range of 198 to 260 nm at 25°C. The processing of raw data was done by buffer subtraction and smoothing as per the manufacturer's recommendations. Three independent experiments were done for all samples. The CD spectra were deconvoluted using CDPro software (46) as per the previously published protocol (48).

Protein estimation by 1D NMR spectroscopy
␣-Syn in the presence of various GAGs was incubated at 37°C for 10 days for aggregation and fibril formation as described above. To determine the degree of aggregation in each case by 1D NMR, fibrils were centrifuged at 14,000 ϫ g for 1 h. As aggregation experiments were carried out in 20 mM Gly-NaOH, pH 7.4, this concentration of glycine showed a strong signal in NMR spectra. Therefore, supernatants removed following centrifugation were dialyzed extensively against MQ water using a 10-kDa molecular weight cutoff (MWCO) mini dialysis unit for 36 h. The relative amount of protein left in the supernatant was determined from the 1D NMR spectra of the protein. The NMR spectra were measured in a Bruker Avance 750-MHz spectrometer equipped with a 5-mm liquid state probe, and D 2 O was used as a lock solvent. For spectral measurements, 512 scans were recorded for each sample at 288 K. The parameters were kept the same throughout the experiment. ␣-Syn and GAGs of known concentrations (100 M) were used as controls. After recording the scans, the signal of the test sample (unknown concentration) was compared with that of a standard (reference) with a known concentration. Thereafter, a unique peak in the NMR spectra of ␣-Syn (from supernatant) was selected such that the chosen peak was unaltered in the presence of GAGs/polymers compared with the ␣-Syn control (known concentration) sample. The area under the peak was integrated, and the concentration of soluble ␣-Syn in the supernatant was determined. The processing and analyzing of the data was done using TopSpin (version 3.5). The difference between the starting concentration of protein for fibrillation (i.e. 300 M) and the soluble protein left after aggregation provided the degree of ␣-Syn aggregation in the presence of each GAG. The experiment was done twice from two independent sets.

Dynamic light scattering
␣-Syn was incubated in the presence of GAG (GAG 0.25, GAG 0.5, GAG 1, and GAG 2) at 37°C for 10 days. Fibril formation was confirmed by CD and TEM analysis at the end of the aggregation. Thereafter, fibrils were centrifuged at 14,000 ϫ g for 1 h and supernatant was removed. 10 l of supernatant from each sample was aliquoted and diluted 10ϫ in filtered 20 mM Gly-NaOH buffer. Thereafter, measurements were taken at 25°C using DynaPro (Wyatt Technology). The hydrodynamic radii (regularization results) of ␣-Syn in the absence and presence of different GAGs were compared. Measurements were acquired from two independent experimental sets of data.

Static light scattering
Light scattering was performed during aggregation kinetics of ␣-Syn alone and in the presence of Hep 0.25 and Hep 2. Samples were aliquoted from reaction tubes at regular intervals and diluted to 10 M in 400 l of 20 mM Gly-NaOH buffer, pH 7.4. The measurements were recorded at excitation and emission wavelengths of 450 nm with a 5-nm slit for 60 s in a FluoroMax-4 spectrofluorometer (JASCO) instrument. The scattering intensity at 30 s was plotted. Three independent experiments were done, and a representative plot is shown.

Atomic force microscopy
Changes in the morphology of LMW ␣-Syn incubated in the presence and absence of Hep was studied using AFM (Asylum Research). To do this, a 20 M sample (aliquoted at regular intervals during aggregation kinetics) was spotted on a freshly cleaved mica sheet. The sample was allowed to incubate for ϳ5 min on a mica sheet at room temperature. This was followed by washing the mica sheet with distilled water to get rid of any unbound proteins/aggregates and subsequent drying under vacuum. The AFM imaging was carried out with AFM operated in tapping mode using a silicon cantilever. It was carried out with a scan rate of 1.0 Hz by scanning 3-4 different areas/sample.

Seeding of ␣-Syn in the presence of Hep
300 M LMW ␣-Syn was incubated with slight agitation (50 rpm) at 37°C in the presence and absence of 600 M Hep (protein:Hep, 1:2). Separately, ␣-Syn seeds were prepared by sonicating the preformed ␣-Syn fibrils at 20% amplitude for 3 min (3 s on/1 s off). After ϳ65 h of aggregation, the ␣-Syn seeds were added to the incubated ␣-Syn solution containing Hep. In the final mixture, the seed concentrations used were 1 and 2% (v/v), respectively. The aggregation kinetics and fibril formation were further monitored by ThT fluorescence assay as described above. The ThT signal for only buffer and in the presence of the seeds was measured and subtracted from the ThT fluorescence value for ␣-Syn in the presence and absence of Glycosaminoglycans modulate ␣-synuclein fibrillation Hep from each time point. Henceforth, ThT fluorescence at different incubation periods was plotted at 480 nm. Only 2% of the seeds were used as a control.

Transmission EM
40 M protein solutions were spotted on carbon-coated Formvar grids (Electron Microscopy Sciences, Hatfield, PA) followed by incubation at room temperature for 5 min. Thereafter, the grids were washed twice with autoclaved MQ water, and the remaining water was gently wiped off with Whatman filter paper. This was followed by staining of all the samples with 10 l of filtered aqueous uranyl formate solution (1% (w/v)) for another 5 min. Finally, the grids were air-dried for 5 min. The images were acquired at 120 kV with ϫ43,000 magnification using a TEM (TECNAI12 D312 FEI).
For analysis of fibril diameter and length, 120 counts were randomly taken from at least 10 different fields. The length and diameter were measured with ImageJ software (National Institutes of Health), which then was used to generate the box plot using Origin Pro (version 8).

NMR study
Two-dimensional 1 H-15 N HSQC spectra were recorded for ␣-Syn protein samples in 20 mM Gly-NaOH buffer, pH 7.4 (containing 0.01% sodium azide), with a (90:10) H 2 O/D 2 O ratio. The experiment was performed with N 15 -labeled ␣-Syn both in the presence and absence of GAG at the different molar ratios. The protein sample was incubated with various GAGs for 1 h at room temperature before the spectra were recorded. All NMR spectra were recorded with 150 M LMW protein solution (as described above under "Preparation of low-molecular-weight (LMW) ␣-Syn") at 15°C in a Bruker Avance 800-MHz spectrometer with a triple resonance probe equipped with deuterium decoupling and gradient facility. 4,4-Dimethyl-4-silapentane-1-sulfonic acid (DSS) was used as an internal reference for the calibration of proton chemical shifts. The nitrogen chemical shift was done indirectly as per the Biological Magnetic Resonance Bank protocols (74). Processing and analysis of all spectra were done using TopSpin (version 2.1) and Sparky 3.113, respectively. Calculations for chemical shift perturbations were carried out using ((5⌬␦ 1 H N) ) 2 ϩ (⌬␦ 15 N) 2 ) 1 ⁄2 to recognize significant perturbations originating because of interaction with GAG.

FTIR spectroscopy
For the FTIR study, a thin KBr pellet was made and kept under an IR lamp for a few minutes. Then 10 l of sample was spotted and allowed to dry. The dried pellet was placed in a transmission holder, and spectra were recorded subsequently using a Bruker VERTEX 80 instrument with a deuterated triglycine sulfate (DTGS) detector. A range of 1800 to 1500 cm Ϫ1 was selected to record IR spectra, and 32 scans were taken for each spectrum at a resolution of 4 cm Ϫ1 . The Fourier selfdeconvolution method was used to deconvolute raw data corresponding to an amide-I region (1700 -1600 cm Ϫ1 ). This deconvoluted spectrum was processed using a Lorentzian curve-fitting procedure, and integration was done using Opus-65 software (Bruker, Leipzig, Germany) as per the man-ufacturer's instructions. Three independent sets of FTIR spectra were measured for each sample.

X-ray diffraction
For X-ray diffraction sample preparation, a 0.7-mm capillary tube was used in which the fibril sample was loaded and dried for 30 min under vacuum. Then, the capillary tube containing the dried sample was placed in the X-ray beam path with the sample-to-detector distance set as 200 mm and exposure time at 120 s. Finally, images of each fibril sample were taken using a Rigaku R-AXIS IVϩϩ detector (Rigaku) and analyzed using Adxv software.

MTT assay
To study the cytotoxicity of GAG-mediated ␣-Syn fibrils, MTT assay was done. To do so, the neuroblastoma cell line SH-SY5Y was cultured in Dulbecco's modified Eagle's medium (HiMedia) with 1% antibiotic mixture (HiMedia) and 10% fetal bovine serum (HiMedia) at 37°C in 5% CO 2 . Once the cells reached 80% confluency, they were trypsinized with 0.25% trypsin-EDTA for 3 min at 37°C. The trypsin was neutralized with an equal amount of growth media, and cells were harvested by centrifugation. Finally, after the cells were counted, they were seeded in a 96-well plate (Corning) at a density of 4 ϫ 10 4 cells/well and incubated for 24 h at 37°C in 5% CO 2 . After that, the cells were treated for 24 h with ␣-Syn fibrils formed in the absence and presence of various GAGs. After a 24-h incubation, MTT reagent (Sigma) (dissolved in PBS and filtered) was added at a concentration of 0.5 mg/ml/well and incubated for another 4 h. This was followed by the addition of 100 l of solubilization buffer (50% N,N-dimethylformamide and 20% SDS, pH 4.6) for the dissolution of the reduced MTT reagent, and the mixture was incubated overnight. The experiments were done in triplicates. Absorbance was taken at 560 and 690 nm (for background correction) using a SpectraMax M2 e microplate reader (Molecular Devices). GAG solution and buffer alone were used as controls.

Labeling of ␣-Syn with FITC
Labeling of the LMW solution of ␣-Syn was carried out with a 3 molar excess of amine-reactive fluorescent dye, i.e. FITC ("isomer I") according to the manufacturer's instructions (Invitrogen). Extensive dialysis (ϳ36 h) in PBS, pH 7.4, was done using a mini dialysis unit (10,000 MWCO, Millipore) to remove any free unreacted dye from the reaction with an intermittent buffer exchange every 3 h until 40 h. After that, the labeled protein solution was lyophilized and subsequently stored at Ϫ20°C until further use. The concentration and degree of labeling of the protein were calculated according to the manufacturer's recommendation. Finally, FITC-labeled ␣-Syn fibrils and seeds in the presence of GAGs were prepared, the detailed procedures of which are described in the supporting information.

Internalization of fibrils in cells
To study the internalization ability of ␣-Syn fibrils (formed in the presence of GAG), an internalization assay was done. The human neuroblastoma cell line SH-SY5Y was seeded at a cell Glycosaminoglycans modulate ␣-synuclein fibrillation density of 5 ϫ 10 3 at 37°C for 24 h in a 5% CO 2 -containing humidified chamber. For the internalization experiment, FITC-labeled sonicated fibrils of ␣-Syn formed in the presence of GAG were used. After that, 1 M labeled ␣-Syn fibrils seeds were added to the SH-SY5Y and incubated for 24 h. Prior optimization of the fibril seed concentration (1 M) and time (24 h) was done with only labeled ␣-Syn fibril (discussed in the supporting information). Control cells with unlabeled ␣-Syn were maintained simultaneously. After 24 h, the treated cells were fixed with 4% paraformaldehyde solution and kept at 4°C. Finally, the amount of internalized ␣-Syn fibrils was quantified by FACS and imaged using a Leica DMi8 microscope in DIC and fluorescent mode.

GAG estimation by Alcian blue assay
The Alcian blue assay was performed as per the previously published protocol (61) with slight modifications. Briefly, Alcian blue dye stock was prepared in 0.018 M H 2 SO 4 using Alcian blue 8XG solution (a 1 ⁄ 4 dilution of original dye solution). The working dye solution was prepared containing 0.25% Triton X-100, 0.018 M H 2 SO 4 , and 10% of the dye stock. Solution A, containing 10 l of 0.027 M H 2 SO 4 , 0.375% Triton X-100, and 4 M guanidine HCl, was added to 10 l of GAG standard (prepared in MQ water) along with 100 l of working dye solution. This was followed by a 5-min incubation and centrifugation for 10 min at 16,000 ϫ g at 4°C. The supernatant was then discarded, and the pellet was redissolved in 200 l of 8 M guanidine-HCl by vigorous vortexing. Finally, the solution was transferred to a 96-well plate, and absorbance was taken at 600 nm in a SpectraMax M2 e plate reader spectrophotometer. The unknown concentration of the GAGs in the supernatant was determined by a standard curve. The experiment was done in duplicates from two independent sets.