Agrin binds α-synuclein and modulates α-synuclein fibrillation

Abstract

Recent studies have begun to investigate the role of agrin in brain and suggest that agrin’s function likely extends beyond that of a synaptogenic protein. Particularly, it has been shown that agrin is associated with the pathological lesions of Alzheimer’s disease (AD) and may contribute to the formation of β-amyloid (Aβ) plaques in AD. We have extended the analysis of agrin’s function in neurodegenerative diseases to investigate its role in Parkinson’s disease (PD). α-Synuclein is a critical molecular determinant in familial and sporadic PD, with the formation of α-synuclein fibrils being enhanced by sulfated macromolecules. In the studies reported here, we show that agrin binds to α-synuclein in a heparan sulfate-dependent (HS-dependent) manner, induces conformational changes in this protein characterized by β-sheet structure, and enhances insolubility of α-synuclein. We also show that agrin accelerates the formation of protofibrils by α-synuclein and decreases the half-time of fibril formation. The association of agrin with PD lesions was also explored in PD human brain, and these studies shown that agrin colocalizes with α-synuclein in neuronal Lewy bodies in the substantia nigra of PD brain. These studies indicate that agrin is capable of accelerating the formation of insoluble protein fibrils in a second common neurodegenerative disease. These findings may indicate shared molecular mechanisms leading to the pathophysiology in these two neurodegenerative disorders.

Introduction

Alzheimer’s disease (AD) and Parkinson’s disease (PD) represent the two most common neurodegenerative disorders in humans. Recent studies demonstrate that these, and many other human diseases, are protein conformational disorders characterized by the formation of unstable protein intermediates that lead to the formation of amyloid protein fibrils and neuronal degeneration (Carrell and Lomas, 1997; Soto, 2003; Uversky and Fink, 2004). In AD, β-amyloid (Aβ) peptide undergoes aggregation to form highly insoluble fibrils that accumulate as extracellular plaques (Glenner and Wong, 1984; Tanzi et al., 1987). A number of different mechanisms contribute to Aβ aggregation, including amyloid precursor protein mutations (Citron et al., 1992; Cai et al., 1993; Suzuki et al., 1994) and interaction of Aβ with various proteins that include ApoE4 (Strittmatter et al., 1993) and heparan sulfate proteoglycans (HSPGs) (Snow et al., 1994). Aβ fibrils and protofibrils are neurotoxic (Pike et al., 1991; Lambert et al., 1998; Hartley et al., 1999) and are considered an initiating factor in the progression of AD to later pathologies.

PD is a neurodegenerative disease characterized by loss of dopaminergic neurons in the substantia nigra. The aggregation and fibrillation of α-synuclein, a 14-kDa presynaptic protein, is thought to be a causative factor for PD (Polymeropoulos et al., 1997; Kruger et al., 1998; Uversky and Fink, 2002, 2004; Uversky, 2003). Fibrillar α-synuclein is a major proteinacous component of Lewy bodies and Lewy neurites (Spillantini et al., 1997; Wakabayashi et al., 1997; Bayer et al., 1999), the primary pathological lesions in PD. The aggregation of α-synuclein is potentiated by a number of different factors, including α-synuclein mutations (El-Agnaf et al., 1998a; Narhi et al., 1999; Li et al., 2001, 2002) and interactions with charged macromolecules, with the extent and rate of fibrillation being increased by sulfated molecules, such as heparin (Uversky and Fink, 2002, in press; Cohlberg et al., 2002; Uversky, 2003).

Agrin is an extracellular matrix and transmembrane HSPG in the central nervous system (CNS) (Tsen et al., 1995; Cohen et al., 1997; Halfter et al., 1997). Recent studies have shown that agrin localizes to all lesion types in AD (Donahue et al., 1999; Verbeek et al., 1999; Cotman et al., 2000) and potentiates Aβ fibril formation (Cotman et al., 2000). In light of evidence that HS may also contribute to the pathophysiology of PD, we considered it important to examine agrin’s role in regulating the aggregation state of α-synuclein. An ability of agrin to modulate both α-synuclein and Aβ fibrillation would raise the possibility that these two neurodegenerative diseases might share similar molecular mechanisms that contribute to their respective protein aggregations.

In this study, we examined the interaction of agrin with α-synuclein and demonstrated that agrin binds to and promotes α-synuclein insolubility and accelerates the formation of α-synuclein protofibrils and fibrils. We also show that agrin and α-synuclein colocalize in Lewy bodies in pigmented neurons in the substantia nigra of PD brain. These data indicate that agrin may contribute to the etiology of PD by modulating the aggregation state of α-synuclein in dopaminergic neurons.

Results

Agrin binds to α-synuclein

In view of recent studies that have shown the capability of sulfated macromolecules, including heparin, to regulate α‐synuclein fibrillation (Cohlberg et al., 2002), we initiated experiments to assess whether agrin would be one CNS HSPG capable of interacting with α-synuclein. Our rationale for these studies was that significant heterogeneity in binding has been demonstrated for proteoglycans, with the same proteoglycan exhibiting differences in binding specificity (Sanderson et al., 1994; Cotman et al., 1999; Knox et al., 2002) or changes in proteoglycan glycosylation modifying its binding specificity for the same heparin-binding proteins (Sanderson et al., 1994; Herndon et al., 1999). Thus, it is conceivable that despite α-synuclein fibrillation being regulated by sulfated macromolecules, agrin may not be the sulfated macromolecule responsible for, or capable of, modulating α-synuclein fibrillation.

Agrin binding to α-synuclein was assessed using an enzyme-linked immunosorbent assay (ELISA), following protocols used previously in our laboratory to examine agrin binding to Aβ (Cotman et al., 2000). Agrin monoclonal antibodies (MAbs) were adsorbed to an ELISA well, and purified chick vitreous body agrin was bound to the antibody-coated well. Recombinant α-synuclein (10, 20, or 40 µg) was then added to the immobilized agrin, and binding was quantified using a goat antiserum to α-synuclein. As shown in Figure 1A, α-synuclein does bind to immobilized agrin in a dose-dependent manner.

As an alternative approach to assess agrin binding to α-synuclein, agrin was incubated with α-synuclein, and an anti-agrin MAb was employed to capture agrin and any associated α-synuclein (Figure 1B). These studies also demonstrated that α-synuclein binds agrin, because the captured agrin contained associated α-synuclein. When α-synuclein was incubated alone with the agrin antibody-coated wells, binding of α-synuclein to agrin MAb was not observed (data not shown).

Agrin binding to α-synuclein alters α-synuclein solubility

Previously we have shown that when agrin is incubated in solution with Aβ peptide, we can detect binding of agrin to Aβ, with a concomitant decrease in Aβ and agrin solubility (Cotman et al., 2000). We therefore reasoned that if agrin is also capable of modulating α-synuclein fibrillation as a result of binding to α-synuclein, then it should also affect α-synuclein solubility. To address this question, agrin (50 µg/mL) was incubated with α-synuclein (1 mg/ml) without agitation for 7 days, followed by centrifugation to assess changes in agrin and α-synuclein solubility. These experiments demonstrated that although α-synuclein incubated alone did not form significant insoluble aggregates, the incubation of α-synuclein with agrin did result in significant insoluble aggregates. When the soluble (supernatant) and insoluble (pellet) fractions were analyzed by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) and immunoblotting, we observed a pronounced decrease in solubility of α-synuclein when coincubated with agrin (Figure 2). We also observed a shift to higher molecular weight of the α-synuclein incubated with agrin, suggesting enhanced aggregation of the α-synuclein, which is even resistant to dissociation under reducing conditions. Similar shifts in the molecular mass of α-synuclein have been observed previously with aggregation and fibrillation (Gosavi et al., 2002). When agrin was treated with nitrous acid or heparitinase, resulting in the degradation of HS chains, and then incubated with α-synuclein, we did not observe a decreased solubility of α-synuclein (data not shown). This indicates that agrin’s ability to regulate α-synuclein solubility is dependent on its HS chains. Interestingly, we also observed that agrin incubated with α-synuclein becomes insoluble as a result of binding to α-synuclein, and only a subset of agrin that is more highly glycosylated is associated with the insoluble fraction (Figure 2). Thus, these data suggest that a subset of agrin molecules, which are more heavily glycosylated, bind to α-synuclein and promote its aggregation.

Agrin binding induced conformational changes in α-synuclein

Numerous studies have provided evidence for the interaction of HSPGs with amyloid proteins, with HSPGs being capable of inducing conformational changes in proteins. This conformational change is characterized by augmented β-sheet structure that leads to the formation of amyloid fibrils. In neurodegenerative diseases, McLaurin et al. (1999) used circular dichroism (CD) and a marine sponge HSPG to show that heparan sulfate glycosaminoglycans (HSGAGs) induce a conformational change in Aβ peptide, changing Aβ from a peptide with random structure to primarily β-sheet structure. We have documented using CD that agrin induces a conformational change with β-sheet structure in Aβ (Cole and Liu, in press), indicating that agrin may be a physiologically relevant HSPG capable of modulating the structure of amyloid-associated proteins. To determine whether agrin binding to α-synuclein may also lead to alteration in α‐synuclein conformation, we analyzed α-synuclein structure by CD in the presence and absence of agrin. We employed α-synuclein samples incubated in 250 M excess in the absence or presence of agrin without agitation, because agitation is known to accelerate the formation of amyloid fibrils (Cohlberg et al., 2002). We observed marked changes in α-synuclein conformation as the result of binding to agrin, which could be detected by CD immediately following the addition of agrin (Figure 3A). Agrin structure as elucidated by CD is primarily α-helical in conformation (data not shown), and agrin present at the molar concentration (40 nM) used for the analysis of α-synuclein conformation was shown to lack β-sheet structure (Figure 3A). CD spectra of α-synuclein in the presence of agrin exhibited a shift in wavelength that is characteristic of β-sheet structure in proteins (36). With increasing incubation times in the absence of agitation, it could be observed that agrin continued to induce a conformational change in α-synuclein, with α-synuclein adopting β-sheet structure in the absence of agrin by 5 days of incubation (Figure 3D). Thus, it is apparent that agrin acts as a catalyst to accelerate conformational changes in α-synuclein.

Agrin binding to α-synuclein accelerates the formation of protofibrils and fibrils

To ascertain whether the decrease in solubility of α-synuclein in the presence of agrin results from the formation of insoluble fibrils and whether changes in α-synuclein structure as measured by CD is also the result of fibrillation, we assessed α-synuclein fibrillation by Thioflavin T (ThioT) fluorescence. This dye complexes with proteins that form cross β-sheet structures in amyloid fibrils, with the level of fluorescence being proportional to the extent of amyloid fibril formation (Naiki et al., 1989; LeVine, 1993). We have shown previously that binding of agrin to Aβ peptide leads to enhanced ThioT fluorescence as a result of agrin accelerating the formation of Aβ fibrils (Cotman et al., 2000). In our present studies α-synuclein was incubated with agitation to accelerate fibril formation (Cohlberg et al., 2002). As shown in Figure 4, the addition of agrin to α-synuclein solutions accelerates the formation of fibrils as assessed by ThioT fluorescence. These experiments indicate a reduced lag time (nucleation) for fibril formation by α-synuclein in the presence of agrin as well as a decreased half-time of fibril formation (Figure 4).

To extend these observations, we analyzed α-synuclein fibrillation in the presence or absence of agrin using electron microscopy (EM). We analyzed samples incubated with or without agrin for different time periods, and samples were selected based on ThioT fluorescence, to compare α‐synuclein samples, with or without agrin, that had equivalent levels of ThioT fluorescence. The rationale for this approach was to show that α-synuclein in the presence of agrin exhibited accelerated protofibril and fibril formation when compared with α-synuclein alone. These data show that after 4 h of incubation with agitation, small punctate aggregates of α-synuclein are already detectable when agrin is present, consistent with agrin accelerating the aggregation process (Figure 5). These aggregates likely correspond to oligomers and protofibrils, as shown in other studies (Lee and Lee, 2002). The acceleration in the formation of protofibrils of α-synuclein in the presence of agrin is also evident when α-synuclein incubated with agrin for 12 or 28 h is compared with α-synuclein alone for 18 or 32 h (Figure 5). Thus, α-synuclein protofibrils in the presence of agrin at 12 h are similar in structure to α-synuclein protofibrils in the absence of agrin at 18 h of incubation, and agrin/α‐synuclein protofibrils at 28 h resemble α-synuclein protofibrils at 32 h of incubation. The formation of mature amyloid fibrils by α-synuclein is also accelerated and enhanced when α-synuclein incubated with agrin is compared with α-synuclein alone after 65 h of incubation (Figure 5). Collectively, the EM and ThioT data indicate that agrin is accelerating and enhancing the formation of α-synuclein amyloid fibrils, suggesting a possible critical role for agrin in the pathophysiology of PD.

Immunohistochemical analysis of agrin distribution in PD brain

The ability of agrin to bind α-synuclein and accelerate its fibrillation raises the crucial question of whether agrin is associated with α-synuclein in the pathological lesions of PD. This would suggest the interesting possibility that agrin might contribute to the pathophysiology of this neurodegenerative disease.

To address the question of whether agrin becomes associated with α-synuclein in PD brain, we analyzed agrin and α-synuclein distribution in PD and age-matched control human brain tissue, using double labeling of tissue sections. The antisera we used in these experiments were a rabbit antiserum to recombinant human agrin that had been immunopurified using a human agrin affinity column and shown to react only with agrin in human brain (Cotman et al., 2000) and a goat antiserum to an α-synuclein synthetic peptide. When the substantia nigra of PD patients is analyzed using these polyclonal antisera to either agrin or α-synuclein, we observe localization of agrin to α-synuclein-positive Lewy bodies in dopaminergic neurons in the substantia nigra (Figure 6). Only a subset of neurons was found to be immunopositive for α-synuclein (Figure 6A and D), and these neurons were also stained using anti-agrin antiserum (Figure 6B and D). Using 4′-6-diamidino-2-phenyindole (DAPI) staining, it can also be seen that only a small subset of neurons contain Lewy bodies immunoreactive for agrin and α-synuclein (Figure 6C). Importantly, the immunopositive neurons, when analyzed by phase contrast, were identified as pigmented neurons (Figure 6E), the target of PD lesions in the substantia nigra. Similar labeling of pigmented neurons, with either α-synuclein antiserum or agrin antiserum, was observed when sister sections were labeled individually, indicating that the detection of agrin immunoreactivity in α-synuclein-positive neurons was specific (data not shown). It can also be observed that structures with an appearance similar to Lewy neurites are stained by both α-synuclein and agrin antisera (Figure 6A–D). Analysis of a second PD brain sample is shown in Figure 6H–J and again shows at higher magnification that α-synuclein and agrin colocalize in Lewy bodies in substantia nigral neurons. In age-matched control patients, we did not observe detectable levels of agrin in neurons and primarily observed agrin associated with the microvasculature (Cotman et al., 2000; data not shown). These data raise the question of how agrin becomes localized intracellularly in neuronal lesions, such as neurofibrillary tangles in AD and Lewy bodies and neurites in PD, and suggests possible alterations in agrin trafficking or endocytosis.

Modulation of α-synuclein cytotoxicity by agrin

The precise role of α-synuclein in the pathogenesis of PD and other neurodegenerative diseases characterized by α‐synuclein aggregation is unknown, although α-synuclein has been shown to be cytotoxic toward neurons. Overexpression of wild-type and mutant α-synuclein has been shown to cause dopaminergic neuronal cell death in both primary and established neuronal cultures (Zhou et al., 2000, 2002). Extracellular administration of α-synuclein has also been shown to induce neuronal cell death, with fibrillar α-synuclein exhibiting greater cytotoxicity (El-Agnaf et al., 1998b), although the physiological significance of this cytotoxicity may be questioned because α-synuclein is an intracellular protein. To ascertain whether agrin may modulate the ability of α-synuclein to induce cell death, as a result of modulating α-synuclein fibrillation, we examined the effect of agrin on α-synuclein-mediated cytotoxicity. In these experiments, α-synuclein was incubated for 1–7 days with or without agrin, and fibrillation of α-synuclein was quantified using ThioT binding. α-Synuclein preparations were then added to the medium of human neuronal SH-SY5Y cell cultures, and after 12 or 48 h exposure to α‐synuclein or α-synuclein-agrin fibrils, cytotoxicity was quantified by the release of lactate dehydrogenase (LDH) from cell cultures. We found that agrin had modest effects on α-synuclein cytotoxicity, augmenting α-synuclein cytotoxicity when assessed after a 12-h exposure (data not shown). However, these data were not statistically significant, indicating that agrin’s effects on α-synuclein fibrillation do not considerably modulate α-synuclein cytotoxicity.

Discussion

This study was undertaken to ascertain the role of the HSPG agrin in a second common neurodegenerative disease characterized by protein misfolding, PD. Recent studies have shown that agrin is associated with all lesion types found in AD (Donahue et al., 1999; Verbeek et al., 1999; Cotman et al., 2000), which include Aβ senile plaques, Aβ deposits in the microvasculature, and neurofibrillary tangles. The presence of agrin in Aβ deposits in the microvasculature has been suggested to lead to increased blood brain barrier permeability (Berzin et al., 2000), which may contribute to the etiology of AD. Agrin has also been shown to bind to Aβ and to accelerate fibril formation by Aβ, suggesting a possible role for agrin in the formation of AD pathologies. HSPGs, or at least heparin or HSGAGs, have also been suggested to contribute to the etiology of other protein conformational disorder diseases. Thus, HSPGs have been localized to the early stages of plaque formation in mouse and hamster models of scrapie prion disease (Snow et al., 1990; McBride et al., 1998), are associated with prion protein in human prion diseases (Snow et al., 1989, 1990), bind islet amyloid polypeptide and localize to amyloid plaques in type 2 diabetes (Young et al., 1992; Castillo et al., 1998), and are associated with inflammation-associated amyloid deposits (Snow et al., 1987, 1991). HSPGs have been shown to accelerate the formation of amyloid plaques in these protein conformational diseases, and HSPG binding to amyloid-associated proteins leads to conformational changes in these proteins that ultimately result in amyloid fibril formation (McCubbin et al., 1988; McLaurin et al., 1999; Cole and Liu, in press). Sulfated macromolecules, such as heparin and HS, have also been shown to bind to α-synuclein and to accelerate α‐synuclein fibril formation (Cohlberg et al., 2002). The heparin GAG chains were also shown to become incorporated into the α-synuclein fibrils (Cohlberg et al., 2002), indicating the importance of heparin-like molecules in regulating α-synuclein fibrillation. The presence of HSPGs has also been suggested in the amyloid lesions of PD, as fibroblast growth factor (FGF) binding to Lewy bodies is eliminated by heparinase treatment (Perry et al., 1992).

Based on these experimental observations, we carried out studies to evaluate the role agrin may play in α-synuclein fibrillation and the etiology of PD. Using a variety of experimental approaches, we demonstrate that agrin binds to α-synuclein. Of particular interest is our demonstration that incubation of α-synuclein with agrin leads to enhanced insolubility of α-synuclein as a result of fibril formation, with agrin becoming incorporated into the fibrils based on the ability to sediment agrin with α-synuclein fibrils. In addition, we observe that only a subset of agrin is associated with the α-synuclein fibrils when analyzed by SDS–PAGE and immunoblotting, with this agrin being higher in molecular weight. Although not confirmed in these studies, it is likely that this high molecular weight fraction of agrin contains either longer HSGAG chains or larger numbers of HSGAG chains. The agrin used in these studies was purified from embryonic chick vitreous bodies, and chicken agrin has been shown to contain three consensus GAG attachment sites (Denzer et al., 1995; Tsen et al., 1995). Our recent studies have shown that agrin contains two clusters of potential GAG attachment sites, with the one cluster containing multiple sites for the addition of HS (Winzen et al., 2003), indicating that subsets of glycosylated agrin are likely to exist in brain, with these subsets of agrin possessing distinct functions that can include binding to α-synuclein. It is unknown whether agrin HSGAG structure is altered in neurodegenerative diseases, such as AD or PD, or whether a more glycosylated subset of agrin possesses higher affinity for proteins, such as α-synuclein. Previous analyses of HSGAG structure in AD have been inconclusive. An analysis of brain HSGAG structure indicated no change in HSGAGs in AD brain (Lindahl et al., 1995), although AD fibroblasts did exhibit changes in HSGAG structure (Zebrower et al., 1992). In addition, HSGAG sulfation has been shown to be altered in inflammation-associated amyloidosis (Lindahl and Lindahl, 1997). Thus, it will be of interest to analyze agrin’s HSGAG structure in diseases, such as PD, to ascertain whether changes in agrin HSGAG structure may contribute to α-synuclein binding and the formation of amyloid fibrils.

Because we were able to demonstrate that agrin is capable of binding α-synuclein, we investigated whether agrin would modulate α-synuclein fibril formation, as has been shown for Aβ in AD (Cotman et al., 2000), and whether agrin would be capable of inducing conformational changes in α-synuclein, as has been shown for Aβ (Cole and Liu, in press). These studies followed established protocols to quantify α-synuclein fibrillation and to measure the kinetics of α-synuclein fibril formation (Cohlberg et al., 2002). Our data show that agrin can induce rapid alterations in α-synuclein structure, with β-sheet structure becoming pronounced as a result of agrin binding. We also demonstrate that agrin accelerates α-synuclein fibrillation by decreasing the lag time and half-time of α-synuclein fibril formation, which is suggestive of agrin regulating the nucleation of α-synuclein fibrils. Fibrillation of α-synuclein has been shown to be a nucleation-dependent process (Wood et al., 1999; Uversky et al., 2001), with fibrillation occurring in progressive steps. Thus, small punctate aggregates of α-synuclein are first observed (Gosavi et al., 2002), which are likely the cellular equivalent of protofibrils (Lee and Lee, 2002). The protofibrils are intermediates in the process of α-synuclein fibrillation (Goldberg and Lansbury, 2000), with large fibrillar inclusions of α-synuclein subsequently forming in cells (Gosavi et al., 2002). Our EM analysis of α-synuclein fibril formation shows that the presence of agrin accelerates the formation of small punctate aggregates of α-synuclein when analyzed at 4 h of incubation, as well as more readily discernible protofibrils when analyzed at 12–18 h of incubation. Formation of fibrillar aggregates was also enhanced in the presence of agrin when α-synuclein samples were analyzed for 65 h of incubation by EM. Thus, agrin is capable of playing a critical role in the regulation of α-synuclein fibrillation, much as has been shown for Aβ peptide.

The experiments conducted here were extended to examine the distribution of agrin in PD brain, which is important in view of our demonstration that agrin can bind α-synuclein and regulate its fibrillation. Analysis of the substantia nigra of PD patients using confocal fluorescence microscopy indicates colocalization of agrin and α-synuclein in Lewy bodies, the pathological lesion of PD. Importantly, we observed the expression of agrin and α-synuclein in pigmented neurons of the substantia nigra, providing additional support for the conclusion that agrin is associated with Lewy body lesions in dopaminergic neurons of the substantia nigra. These results provide support for the previous suggestion that HSPGs are a component of Lewy bodies, because FGF-binding sites in Lewy bodies were shown to be heparinase sensitive (Perry et al., 1992). However, although previous studies suggested that heparinase-sensitive FGF-2-binding sites were present in only a fraction of Lewy bodies in PD brain (Perry et al., 1992), our studies suggest that agrin may be associated with a majority of α-synuclein-positive Lewy bodies, at least in the four patient samples analyzed in our study. It should be noted, however, that we did not observe all Lewy bodies as being agrin positive, and we did analyze one PD patient with only sparse agrin staining (data not shown). It therefore remains possible that there is a variable expression of agrin in the lesions of PD, with some patients exhibiting a disease state that is characterized by only a subset of lesions containing agrin. Accordingly, although agrin has been shown to be associated with all pathological lesions in AD (Donahue et al., 1999; Verbeek et al., 1999; Cotman et al., 2000), only a subset of lesions are agrin positive in hereditary cerebral hemorrhage with amyloidosis (Dutch type; van Horssen et al., 2001).

It is also important to note that recent studies by van Horssen et al. (2004) analyzed the distribution of HSPGs in PD brain and demonstrated an absence of agrin immunoreactivity, as well as other HSPG core proteins, in Lewy bodies in PD brain. This led to the conclusion that HSPGs are not localized to Lewy bodies in PD and will not play a significant role in the pathophysiology of PD (van Horssen et al., 2004). However, the agrin antibody used in these studies was a MAb, compared with a polyclonal antiserum generated against human agrin used in our studies described here. In light of the demonstration that heparinase-sensitive FGF-binding sites are present in Lewy bodies (Perry et al., 1992), the wealth of evidence that HSPGs are a shared component of amyloid deposits in protein conformational disorders (Ancsin, 2003; Cole and Liu, in press), and that our agrin antiserum specifically stains Lewy bodies in PD, it is clear that additional studies are warranted to further characterize the potential role of agrin, or other HSPGs, in a protein conformation disorder, such as PD.

We think it is also important to consider the physiological significance of HSPGs, such as agrin, being localized to intracellular lesions, such as Lewy bodies, because HSPGs are typically thought of as cell surface and extracellular matrix molecules. However, there is a wealth of evidence that demonstrates an intracellular localization of HSPGs, especially in disease states (Perry et al., 1992; Su et al., 1992; Odawara et al., 1998; Donahue et al., 1999; Roskams et al., 1999; Verbeek et al., 1999; Cotman et al., 2000; Lundquist and Schmidtchen, 2001). The demonstration of the intracellular localization of HSPGs includes numerous neurodegenerative diseases, which include AD (Su et al., 1992; Donahue et al., 1999; Verbeek et al., 1999; Cotman et al., 2000) and PD (Perry et al., 1992). In particular, our laboratory and others have shown that HSPGs, such as agrin, are associated with neurofibrillary tangles in AD (Verbeek et al., 1999), and it has also been shown that HSPGs associate with Pick bodies in Pick disease (Odawara et al., 1998). It has also been shown recently that HSPGs colocalize intracellularly in neurons with b-amyloid cleavage enzyme 1 (Scholefield et al., 2003). It is therefore conceivable that in pathological conditions associated with a disease, such as PD, agrin becomes localized intracellularly, where it could function as a pathological chaperone to affect α-synuclein aggregation. It has recently been documented that proteins, such as torsinA, which are molecular chaperones that are not normally cytosolic, can contribute to α-synuclein aggregation (Shashidharan et al., 2000), and thus it is tempting to speculate that an HSPG, such as agrin, could also have a similar function when subjected to an abnormal subcellular localization. Alternative mechanisms by which agrin could modulate α-synuclein aggregation can be considered in light of recent evidence that α-synuclein is localized intracellularly in secretory vesicles and is secreted from neuronal cells in an endoplasmic reticulum-/Golgi‐independent pathway (Lee et al., 2005). It is therefore tempting to speculate that agrin could become associated with α-synuclein in intracellular secretory vesicles, or once secreted from, cells α-synuclein could bind agrin, with an α‐synuclein/agrin complex subsequently being endocytosed in PD.

In view of evidence that overexpression of either wild-type or mutant α-synuclein in neurons leads to cell death (Zhou et al., 2000, 2002) and that extracellular exposure of cells to fibrillar α-synuclein promotes cell death (El-Agnaf et al., 1998b), we examined the effect of agrin on α-synuclein-mediated neuronal toxicity. We expected to observe a noticeable enhancement of α-synuclein cytotoxicity by agrin as a result of augmented formation of α-synuclein protofibrils and fibrils in the presence of agrin, because α-synuclein cytotoxicity has been shown to be fibril-dependent, at least when α-synuclein is exposed to cells in the extracellular environment (El-Agnaf et al., 1998b), and oligomers of α‐synuclein have been suggested to mediate neuronal cytotoxicity (Volles et al., 2001). An important caveat to these studies, however, is that exposure of cells to α-synuclein in the extracellular environment may not be physiologically relevant, because α-synuclein is an intracellular protein. Our studies indicated that although agrin had modest influence on α-synuclein cytotoxicity, the effects were not statistically significant. These data therefore indicate that agrin’s ability to accelerate α-synuclein fibrillation does not impact extracellular α-synuclein cytotoxicity. However, it remains to be determined whether agrin’s binding to intracellular α-synuclein will modulate α-synuclein cytotoxicity.

It is of interest that both AD and PD are members of a family of neurodegenerative diseases referred to as protein conformational disorders (Carrell and Lomas, 1997; Soto, 2003), as these diseases are characterized by the formation of unstable protein intermediates that form amyloid fibrils, and that ultimately lead to neuronal cell death (Carrell and Lomas, 1997; Uversky and Fink, in press). Our data showing that agrin can regulate the formation of amyloid fibrils by both α-synuclein and Aβ raise the interesting possibility that agrin may function as a pathological chaperone capable of binding to the unstable protein intermediates in these diseases, driving the reaction toward the formation of aggregates, including protofibrils and amyloid fibrils. The demonstration that HSPGs are present in immature prion protein plaques in scrapie mice (McBride et al., 1998) and hamsters (Snow et al., 1990) and in Creutzfeld–Jakob disease (Snow et al., 1989, 1990) lends credence to this hypothesis, as prion diseases are also protein conformational disorders (Carrell and Lomas, 1997; Soto, 2003). However, it remains to be determined whether agrin is present in prion plaques and is capable of binding prion protein.

Materials and methods

Analysis of agrin–α-synuclein binding

Agrin binding to α-synuclein was assessed using an ELISA, according to previously published protocols (Cotman et al., 2000). Agrin MAbs (6D2 and 3A12, 50 µg/mL) were adsorbed to an ELISA well, and chick vitreous body agrin (1 µg) was bound to the well. Chick vitreous body agrin was purified from E14 chick vitreous bodies, as previously described (Cotman et al., 1999). Recombinant human α‐synuclein was purified, as previously described (Uversky et al., 2001), and 10, 20, or 40 µg in 100 µL of phosphate-buffered saline (PBS)–2% bovine serum albumin (BSA) was added to the well and incubated with the agrin for 2 h. Wells were washed three times with PBS between incubations. Binding of α-synuclein was visualized by incubating wells with polyclonal goat antisera to α-synuclein (Biodesign International, Saco, ME), followed by biotinylated anti-goat antiserum and avidin–biotin reagent (Vector Laboratories, Burlingame, CA). Wells were then incubated with 3,3′,S,S′-tetramethylbenzidine peroxidase substrate (Sigma, St. Louis, MO) and read at 450 nm using an ELISA plate reader.

Agrin binding to α-synuclein was also assessed by incubating chick vitreous agrin (5 µg) with 50 µg α-synuclein at 37°C without agitation for 7 days. Aliquots of this reaction mixture diluted in 100 µL of PBS–2% BSA, corresponding to either 9 or 15 µg of α-synuclein and 0.9 or 1.5 µg agrin, were then added to 96-well plates coated with agrin MAb, as described above. Following a 2-h incubation, α-synuclein polyclonal antiserum was added to visualize α-synuclein “captured” by binding of agrin to the agrin antibody coated well. α-Synuclein bound to the agrin was then quantified, as described above. α-Synuclein in the absence of agrin was not observed to bind to the agrin MAb in this assay.

Analysis of agrin’s effects on α-synuclein solubility

An alternative approach to assess the ability of agrin to bind to α-synuclein was to incubate agrin with α-synuclein at 37°C without agitation for 7 days, to determine whether agrin could regulate α-synuclein solubility. These studies were carried out, as previously described, in our laboratory using Aβ peptide (Cotman et al., 2000). Briefly, α-synuclein (100 µg; 1 mg/mL) was incubated with chick agrin (5 µg; 50 µg/mL) for 7 days at 37°C without agitation. Reaction mixtures were then centrifuged at 13,000 × g for 30 min, supernatants were saved (soluble fraction), and pellets (insoluble fraction) were washed once with PBS. Aliquots of soluble and insoluble fractions were then analyzed by SDS–PAGE for the presence of agrin or α-synuclein, using either a 4% gel (agrin) or 15% gel (α-synuclein), followed by western blotting using a mouse MAb to agrin or a goat antiserum to α-synuclein. Because of the small amounts of agrin that became associated with the insoluble fraction, significantly smaller amounts of insoluble α-synuclein, when compared with soluble α-synuclein, were analyzed by SDS–PAGE.

Analysis of α-synuclein conformation by CD

CD analysis of α-synuclein as a result of agrin binding was assessed, according to published methods (McLaurin et al., 1999). α-Synuclein (100 µM) was incubated with or without agrin (0.5 µM; 300 ng/µL) in 50 mM PBS at 37°C for up to 6 days without agitation. Samples were 10 times diluted immediately before CD spectra being recorded on a Jasco Circular Dichroism Spectrometer (Model J-600; Tokyo, Japan) at room temperature. Spectra were scanned from 200 to 260 nm, with 1 mm cell pathlength, 1 nm bandwidth, 0.4 nm steps, and 100 nm/min speed. Every sample was scanned eight times to eliminate the background to noise signal.

Analysis of effects of agrin on α-synuclein fibrillation

To assess whether agrin might modulate the kinetics of fibril formation by α-synuclein, we quantified the formation of fibrils by α-synuclein based on ThioT fluorescence. Assay solutions contained protein at a concentration of 0.5 mg/mL (35 µM) in 20 mM Tris–HCl and 0.1 M NaCl, pH 7.5, at room temperature, containing 20 µM ThioT with or without agrin (12 µg/mL). A volume of 150 µL of the mixture was pipetted into a well of a 96-well plate (white plastic, clear bottom), and a 1/8th internal diameter Teflon sphere (McMaster-Carr, Los Angeles, CA) was added. Each sample was run in triplicate or quadruplicate. The plates were sealed with Mylar plate sealers (Dynex, Chantilly, VA). The plate was loaded into a fluorescence plate reader (Fluoroskan Ascent) and incubated at 37°C with shaking at 150 rpm with a shaking diameter of 20 mm. The samples were shaken to speed the rate of fibril formation. The fluorescence was measured at 15-min intervals with excitation at 450 nm and emission at 485 nm, with a sampling time of 100 ms. Data from replicate wells were averaged before plotting fluorescence versus time and fit to an empirical equation (Nielsen et al., 2001). Effects of agrin on α-synuclein fibrillation were also determined by electron microscopic analysis of fibrils, using previously described methods (Cohlberg et al., 2002).

Immunohistochemical localization of agrin in PD brain

Human brain tissue from PD patients and age-matched controls were obtained from the Harvard Brain Tissue Resource Center (Belmont, MA). Paraffin sections were deparaffinized in xylene and rehydrated through an ethanol to water series. Sections were then treated with 88% formic acid for 5 min, rinsed in water, and then processed for immunohistochemistry, as previously described (Cotman et al., 2000). For confocal microscopy, paraffin sections of human brain that had been deparaffanized and rehydrated were blocked with 5% donkey serum in PBS, and then incubated overnight at 4°C with primary antibodies to agrin. The polyclonal antiserum against agrin was prepared to a fusion protein of recombinant agrin (Cotman et al., 2000) and was immunopurified in an affinity column containing immobilized human agrin protein. This antiserum has been shown to only react with agrin in human brain (Cotman et al., 2000). Following the 2-h incubation with Cy3-conjugated anti-rabbit IgG (Jackson Immunobiologicals, West Grove, PA), sections were incubated overnight at 4°C with anti-α-synuclein (Biodesign International) in PBS–2% BSA. Sections were then incubated for 2 h with Cy2-conjugated anti-goat IgG (Jackson Immunobiologicals). Fluorescence was visualized using either Nikon C1 confocal microscopy imaging system or Nikon Diaphot-inverted microscope (Nikon, Melville, NY).

Analysis of α-synuclein neuronal cytotoxicity

To assess a potential role for agrin in α-synuclein cytotoxicity, we followed previously published protocols that analyzed the effect of extracellular administration of α‐synuclein on neuronal cytotoxicity (El-Agnaf et al., 1998b). α-Synuclein (110 µM) was incubated in the presence or absence of agrin (440 nM) for 1–7 days, to generate α‐synuclein with differing fibrillation states. The fibrillation state of α-synuclein was assayed by ThioT binding before the use in cytotoxicity assays. Cultures of SH-SY5Y cells (7500 cells/well) were assayed for cytotoxicity in 96-well plates, by adding aliquots of α-synuclein or α-synuclein–agrin (final concentration of 10 µM α-synuclein/well) and incubating for 12–48 h. To quantify cell death, we incubated cultures with CytoTox-One (Promega, Madison, WI), according to the manufacturer’s protocols, and cell death was quantified by the measurement of LDH release using a fluorimeter with excitation at 544 nm and emission at 590 nm. Triplicate wells were used per treatment per experiment.

Acknowledgments

This work was supported by NIH grants NS33981 and NS39985.

References

Author notes

²Biomedical/Biotechnology Research Institute, North Carolina Central University, Durham, NC 27707; ³Department of Chemistry and Biochemistry, University of California, Santa Cruz, CA 95064; and ⁴Department of Neurobiology, University of Pittsburgh, Pittsburgh, PA 15261