of 16

Agrin Binds to β-Amyloid (Aβ), Accelerates Aβ Fibril Formation, and Is Localized to Aβ Deposits in Alzheimer's Disease Brain

12 views
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Share
Description
Agrin Binds to β-Amyloid (Aβ), Accelerates Aβ Fibril Formation, and Is Localized to Aβ Deposits in Alzheimer's Disease Brain
Transcript
  Agrin Binds to  -Amyloid (A  ),AcceleratesA  Fibril Formation, and Is Localized toA  DepositsinAlzheimer’s Disease Brain Susan L. Cotman,* Willi Halfter, † and Gregory J. Cole* ,1 *Neurobiotechnology Center and Department of Neuroscience, The Ohio State University,184 Rightmire Hall, 1060 Carmack Road, Columbus, Ohio 43210; and  † Department of Neurobiology, University of Pittsburgh, Pittsburgh, Pennsylvania 15261 Agrin is an extracellular matrix heparan sulfate proteogly-can (HSPG) well known for its role in modulation of theneuromuscular junction during development. Althoughagrin is one of the major HSPGs of the brain, its functionthere remains elusive. Here we provide evidence suggest-ing a possible function for agrin in Alzheimer’s diseasebrain. Agrin protein binds the amyloidogenic peptide A  (1–40) in its fibrillar state via a mechanism that involvesthe heparan sulfate glycosaminoglycan chains of agrin.Furthermore, agrin is able to accelerateA  fibril formationand protect A   (1–40) from proteolysis,  in vitro.  Support-ing a biological significance for these  in vitro   data, immu-nocytochemical studies demonstrate agrin’s presencewithinsenileplaquesandcerebrovascularamyloiddepos-its, and agrin immunostained capillaries exhibit pathologi-cal alterations in AD brain. These data therefore suggestthat agrin may be an important factor in the progression ofA   peptide aggregation and/or its persistence in Alzhei-mer’s disease brain. INTRODUCTION Alzheimer’sdisease(AD)isaprogressiveneurodegen-erative disorder characterized by pathological lesionsincluding senile plaques, congophilic angiopathy, andneurofibrillary tangles (reviewed by Selkoe, 1991). The  -amyloid peptide (A  ) has been shown to be the majorcomponent of senile plaques and congophilic angiopa-thy (Glenner and Wong, 1984; Masters  et al.,  1985), andthe discovery of familial AD mutations that causeincreases in A   production has illuminated the impor-tance of A   deposition in AD (reviewed by Selkoe,1996).A  is a 39- to 42-amino-acid peptide that arises fromproteolytic processing of the amyloid precursor protein(APP) (Goldgaber  et al.,  1987; Kang  et al.,  1987; Robakis et al.,  1987; Tanzi  et al., 1987). The A   peptide that isfound in the senile plaques and cerebrovascular depos-its exists as a multimeric aggregate with a fibrillarappearance (Roher  et al.,  1986). Therefore, the processesthat lead to fibrillar A  accumulation are a major focusin Alzheimer’s disease research. Several other mol-ecules have also been shown to be associated withamyloid deposits, and  in vitro  studies of fibrillogenesissuggest they may be important in the aggregation andpersistence of theA  fibrils  in vivo.  These includeApo E(Wisniewski  et al.,  1994; Wood  et al.,  1996), laminin(Bronfman  et al.,  1998; Monji  et al.,  1998), AchE (Alvarez et al.,  1995), and proteoglycans (Castillo  et al.,  1997).HSPGs are a subclass of proteoglycans of particularinterest in the A   aggregation process. Immunohisto-chemical studies have localized HSPGs to amyloid-containing lesions, including congophilic angiopathy,senileplaques,diffuseamyloiddeposits,andneurofibril-lary tangles (NFTs) (Snow  et al.,  1988; Su  et al.,  1992;Donahue et al., 1999).Adirectbindinginteractionoccurs between the  -amyloid peptide and the HSPG perlecan,and this interaction accelerates the rate of A   fibrilformation  in vitro  (Castillo  et al.,  1997). Furthermore,perlecan has been shown to maintain fibril stability  invitro  (Castillo  et al.,  1997) and to affect the proteolyticsensitivity of the fibrillar form of A   (Gupta-Bansal  etal.,  1995). Perhaps the most striking evidence support-ing the importance of HSPGs in AD pathophysiology isthe  in vivo  demonstration that coinfusion of HSPG withA   into rat brain results in A   deposits that become 1 To whom correspondence and reprint requests should be ad-dressed. Fax: (614)292-5379. E-mail: cole.115@osu.edu. MCN Molecular and Cellular Neuroscience   15,  183–198 (2000)doi:10.1006/mcne.1999.0816,  available online at http://www.idealibrary.com on  183 1044-7431/00 $35.00Copyright   2000 byAcademic PressAll rights of reproduction in any form reserved.  compacted and stellate with age. This morphology issimilar to the features of senile plaques in AD (Snow  etal.,  1994a). In this study, control infusions (A   alone)onlyresultedinthedevelopmentoffibrillarA  deposits50% of the time, and deposits never developed theAD-like features.In the studies presented here, we have expanded the biochemicalcharacterizationofA  -proteoglycaninterac-tions and fibrillogenesis by studying another major brain heparan sulfate proteoglycan, agrin. This HSPG isa large extracellular matrix molecule identified as animportant factor in synaptogenesis at the neuromuscu-lar junction (Nitkin  et al.,  1987; Reist  et al.,  1987; Wallace,1989). Expression studies show that agrin is one of themajor HSPGs in the extracellular matrix of the CNS, being expressed by both neurons and glia in develop-ment and into adulthood (Kroger  et al.,  1996; Halfter  etal.,  1997; Cohen  et al.,  1997; Ji  et al.,  1998). Althoughagrin’s function in brain is not very well understood,studiesinourlaboratorieshavesuggestedpossiblerolesin cell–cell interactions and regulation of the extracellu-lar matrix in the developing nervous system (Burg  et al., 1995, Storms  et al.,  1995, Cotman  et al.,  1999). Thedemonstration that agrin induces cAMP response ele-mentbindingprotein(CREB)phosphorylationinhippo-campal neurons also suggests a role in regulatingsynaptic efficacy (Ji  et al.,  1998). Most recently, Donahue et al.  (1999) have described abnormal distribution of agrin in AD brain, as compared to normal aged brain,suggestingagrin,likeotherHSPGs,maybeimportantinAlzheimer’s disease.In this report, we have begun to address the questionof whether agrin is involved in AD pathophysiology.Using a solid-phase immunoassay, we have examined apotential binding interaction between agrin and  -amy-loid. Furthermore, using a variety of   in vitro  assays, wehave addressed whether this interaction may havefunctional consequences on A   biochemistry and inAlzheimer’s disease. RESULTS FibrillarA  (1–40) Binds ImmobilizedAgrin via a Heparan Sulfate-Dependent Mechanism  By using a solid-phase immunoassay, we examinedthe binding interaction between agrin and one of theA  peptides, A   (1–40), found in the senile plaques of Alzheimer’s disease. It has previously been shown thatheparin and HSPGs preferentially bind the fibrillaraggregated state of the A  peptide (Gupta-Bansal  et al., 1995; Castillo  et al.,  1997; Watson  et al.,  1997). In order todetermine whether an agrin–A   interaction is alsogoverned by the conformational state of the peptide,fibrillar and nonfibrillar A   (1–40) were prepared foruse in the solid-phase assay. The fibrillar A  (1–40) wasprepared by resuspending commercially obtained pep-tide in water, and aging it for 1 week at 37°C. Thenonfibrillar A   (1–40) was prepared by treating thefreshly resuspended peptide with the solvent, hexafluo-roisopropanol (HFIP). This solvent promotes the nonfi- brillar conformation by promoting or stabilizing  -heli-ces of theA  peptide (Barrow et al.,  1992). TheA  (1–40)preparations were tested using the Thioflavin T methodfor fibril detection (Levine, 1993) and results confirmedthe nonfibrillar and fibrillar content of those samples(data not shown). A   (1–40) specifically bound theimmobilized agrin when it was in a fibrillar state, withno binding observed for the nonfibrillar A  (1–40) (Fig.1). Thus, the agrin–A  interaction is consistent with thepreviously described mechanism for HSPG interactionswith A  , exhibiting a preference for the fibrillar confor-mation.Thebindingbetween  -amyloidandHSPGsisknownto be mediated by the heparan sulfate glycosaminogly-can (GAG) chains (Castillo  et al.,  1997), which would beexpected since it is well known that A   interacts withheparin and heparan sulfate (Leveugle  et al.,  1994;Watson,  et al.,  1997). Furthermore, it has been demon-strated that agrin exists as an HSPG in human brain(Halfter  et al.,  1997; Donahue  et al.,  1999). Therefore, wewanted to determine whether the binding we observedin the solid-phase assay was occurring via a heparansulfate-dependent mechanism. In previous studies, wehave used nitrous acid to chemically degrade the hepa-ran sulfate side chains of agrin (Tsen  et al.,  1995; Cotman et al., 1999).Thismethodofsidechainremovalhasgivenconsistent results with other methods, with respect todetermining the extent of heparan sulfate-mediated binding. Furthermore, short-term exposure to the ni-trous acid did not affect core protein interactions, sincetwo core-binding proteins, merosin and tenascin, stillexhibited binding to nitrous acid-treated agrin (Cotmanand Cole, unpublished observations; Cotman  et al., 1999). Upon nitrous acid treatment of agrin, we wereable to almost completely diminish binding of thefibrillar A  (1–40) to the immobilized agrin, suggestingthat this interaction indeed occurs in a heparan sulfate-dependent manner (Fig. 1).As with the untreated agrin,nonfibrillar A   (1–40) did not bind the nitrous acid-treated agrin. 184  Cotman, Halfter, and Cole   Agrin Protects FibrillarA  (1–40) from Proteolysis  Previous studies have shown that Engelbreth–Holm–Swarm (EHS) mouse HSPG (perlecan) can protect fibril-lar A   from proteolysis (Gupta-Bansal  et al.,  1995). Wehave used an  in vitro  protease assay, using the nonspe-cific protease papain, to determine whether the bindinginteraction between agrin and A   (1–40) results inprotection of the peptide from proteolysis.In preliminary experiments, we observed differentialrates of protease degradation in the nonfibrillar versusthe fibrillar preparations of theA  (1–40). Within 1 h, allnonfibrillar peptide was degraded when examined on aTris-tricineSDS–PAGEgel.Incontrast,wewerestillableto observe fibrillar peptide after one hour of exposure tothe protease (data not shown). Based upon these obser-vations, we examined agrin’s effects on the proteasesensitivity of A   (1–40) over time courses appropriatefor the individual preparations.When fibrillar A   (1–40) was treated with papain, itwas degraded to undetectable levels after 2 h, asdetermined by Coomassie blue staining. In contrast,when A   (1–40) was preincubated with agrin, peptidewasstillpresentafter2and4hofpapaintreatment(Fig.2). Furthermore, the altered proteolytic sensitivity of theA   (1–40) peptide was specific to agrin, since laminin,another A  -binding and plaque-associated molecule(Murtomaki  et al.,  1992; Bronfman  et al.,  1998), was FIG. 1.  Agrin binds to fibrillar, but not nonfibrillarA  (1–40).Agrin was immobilized on 8-well strip plates via the 6D2 mAb, and an interactionwithA  (1–40)wasexaminedusinganenzyme-linkedimmunosorbentassay(ELISA).RepresentativebindingprofilesareshownwithfibrillarA  (1–40) and nonfibrillar A  (1–40). To determine whether binding occurred in a heparan sulfate-dependent mechanism, nitrous acid was used tochemically degrade the heparan sulfate GAG chains of agrin prior to immobilization (closed circles, untreated agrin; open circles, nitrousacid-treated agrin). FIG. 2.  Agrin binding to A  (1–40) diminishes A  (1–40) proteolyticsensitivity. Nonfibrillar and fibrillar A   (1–40) (2 µg) were preincu- bated alone, with agrin (4 µg), or laminin (4 µg), and subsequentlytreated with 0.02 U of papain for the times indicated. Samples wereanalyzed by 16.5% Tris-Tricine SDS–PAGE. A   (1–40) ran at itsexpected apparent molecular mass (4.3 kDa) as indicated by the bandin the untreated A   (1–40) lane. Note the same size bands whichappear at the 2- and 4-h times when fibrillar A   (1–40) is pretreatedwith agrin. Agrin Interactions and Localization with A   185  unable to protect the peptide from papain digestion(Fig. 2). No protection of nonfibrillar A   (1–40) wasobserved, since A  (1–40) preincubated with agrin wascompletely digested after only 15 min of exposure topapain (Fig. 2). AgrinAcceleratesA  (1–40) Fibril Formation  Since other proteoglycans have been shown to alterA   fibril formation  in vitro  (Castillo  et al.,  1997), wethought it important to examine agrin’s effects on A  fibril formation. Fibril formation was studied using aseeding method previously described by Wood  et al. (1996). This method treats A   (1–40), which has beendisaggregated with HFIP, with a ‘‘seed’’ of previouslyformed A   (1–40) fibrils such that fibril formation isstimulated.Abenefit of studying fibril formation by thismethod is that the rate of fibril formation is highlyreproducible with a given concentration of fibrillar seed(Wood  et al.,  1996). Commercially obtained preparationsof A  (1–40), which spontaneously form fibrils as well,varywidelyintheirinitialseedcontent,evenfromlottolot. Thus fibril formation rates in commercial prepara-tions vary widely (Howlett  et al.,  1995).Additionally, byadding a low concentration of seed relative to theamounts typically obtained in commercial preparations,one can slow down the rate of fibril formation, thuscreating a window where molecules that may affect therate can easily be studied.Using the Thioflavin T method for fibril detection(Levine, 1993), we examined agrin’s affects onA  (1–40)fibril formation at 37°C, over a 1-week period. A 0.1µg/ml seed of fibrillar A   (1–40) was added to thereactions, and aliquots were taken throughout the weekfor Thioflavin T quantitation of fibrils. Three separateexperiments were performed usingA  (1–40) alone andA  (1–40) plus agrin in a 250:1 molar ratio, respectively.Agrin significantly accelerated the rate of fibril forma-tion, even as early as 3 h after fibril formation wasinitiated (Fig. 3A). Fluorescence measurements at 3 hindicated a 4.3-fold increase ( P  0.01) when agrin wascoincubated with A   (1–40). Throughout the 1-weekperiodofincubation,fluorescencemeasurementscontin-ued to indicate a statistically significant acceleration of fibril formation when agrin was present (Fig. 3A).Agrinincubated alone showed no intrinsic fluorescence withThioflavinT,indicatingthattheaugmentationinfluores-cence intensity was due to A  -agrin interactions (datanot shown).Inordertoexaminethemorphologyoftheaggregates beingformedduringthefibrilformationassays,sampleswere examined using transmission electron microscopy(TEM). Reactions were set up as described for theThioflavin T assay and were incubated at 37°C for 2days. These were subsequently negatively stained with2% uranyl acetate and examined by TEM. After 2 days,when only a small amount of fluorescence was detectedin the A   (1–40) sample (†, Fig. 3A), only short, thinfibrils could be detected by TEM (data not shown).Much of the 2-day material in the A   (1–40) sample FIG. 3.  Acceleration ofA  (1–40) fibril formation by agrin.A  (1–40)(50 µM) was incubated in PBS, pH 7.4, at 37°C, with (closed circles) orwithout(opencircles)200nMagrin.Aliquots(5µl)weretakenat3,24,48, 96, and 168 h, and fibril content was measured by Thioflavin Tfluorometry (A).Agrin significantly accelerated fibril formation ofA  (1–40)throughoutthe1-weekincubation.Dataisshownasmean  SDfrom three experiments. Significance was calculated using pairedStudent’s t tests(* P  0.01;** P  0.001).Transmissionelectronmicros-copy (TEM) was used for visualization of the samples at 48 h (†, A).SampleswereadsorbedontoFormvar-coatedgridsfor30secondsandsubsequently stained with 2% uranyl acetate (2 min) for examinationon TEM. TEM revealed that only a small amount of the material in theA  (1–40) sample was fibrillar at this time point, in accordance withfluorescence measurements at 48 h. Most of the material appeared asnonfibrillar, amorphous aggregates (B). In contrast, the material fromthe A  (1–40) plus agrin samples appeared as fibrillar aggregates (C).Original magnification of micrographs, 112,000  . 186  Cotman, Halfter, and Cole   consistedofamorphousaggregates,whicharecharacter-istic of nonfibrillar A   (1–40) (Watson  et al.,  1997) (Fig.3B). In contrast, many fibrils were readily detectablewhen agrin was coincubated withA  (1–40) (Fig. 3C).In a separate assay, we examined agrin’s effects onA  (1–40) fibril formation using a solubility assay. Takingadvantage of the insolubility of A   (1–40) fibrils, reac-tions that had been incubated for 1 week at 37°C werecentrifuged to pellet any insoluble material that hadformed. Supernatants and washed pellets were thenexamined by SDS–PAGE. Coincubation of A   (1–40)with agrin resulted in an increase in insoluble peptidewhen compared to A  (1–40) alone, as indicated by theintensity of the 4.3-kDa band corresponding to A  (1–40) (Fig. 4A). Furthermore, examination of agrin byWesternblotrevealsthata400-to600-kDaimmunoposi-tive smear, which is representative of agrin’s appear-ance on SDS–PAGE (Tsen  et al.,  1995), is localized to thepellet when coincubated withA  (1–40) (Fig. 4B). Thesedata strongly suggest a physical association betweenagrinandthepelletableA  (1–40),sinceagrinincubatedalone does not appear in the pellet fraction (Fig. 4B).These data support our results in the solid-phase immu-noassaywherefibrillarA  (1–40)isabletobindimmobi-lized agrin (Fig. 1).A  (1–40)sampleswerealsoexaminedbyTEMafter1week of incubation at 37°C in order to further study themorphology of fibrils formed when A   (1–40) is incu- batedaloneversuscoincubatedwithagrin.Asdescribedfor the studies reported in Fig. 4, samples were centri-fuged to collect insoluble A   fibrils, resuspended inwater, and negatively stained for TEM. Several differ-ences between A  (1–40) and A  (1–40)–agrin samplescould be noted from these micrographs. First, whenrandom fields were examined from the A   (1–40)samples, we consistently observed less fibrillar materialversus the samples that had been coincubated withagrin (Figs. 5A and 5B). This is in accordance with theincreased abundance of peptide in the pellet when A  (1–40) is coincubated with agrin, as observed by SDS–PAGE (Fig. 4A). Second, the fibrils observed in thesamples coincubated with agrin were significantlygreater in length than those observed in the controls(Figs. 5C and 5D). The longest fibrils observed in theA  -only samples and the A   plus agrin samples aredemonstrated in the Figs. 5E and 5F micrographs,respectively. Note that these fibrils have been presentedat the same magnification. Finally, fibrils in the samplesthat were coincubated with agrin appear to laterallyaggregate and bundle. This conclusion is suggested bythe appearance of regions within thick bundles whereseveral thinner fibrils are aligned in a parallel fashion.This is demarcated by the arrows in Fig. 5F. Figure 5G isone of several large accumulations of fibrils that wasobserved in theA  (1–40) plus agrin samples. Agrin Localizes to   -Amyloid Deposits in Alzheimer’s Disease Brain  Recent findings have demonstrated altered agrinimmunostaining within the parenchyma and microvas-culature of Alzheimer’s disease brain. Agrin was local-ized to diffuse and neuritic plaques, neurofibrillary FIG. 4.  Coincubation of agrin with A  (1–40) increases A  ’s insolu- bility and agrin becomes localized to the insoluble fraction. A  (1–40)wasincubatedwithorwithoutagrin,asdescribedinthelegendofFig.3, and samples were centrifuged to pellet the insoluble material. Tovisualize A   (1–40), samples were separated on a 16.5% Tris-TricineSDS–PAGE gel and stained with Coomassie blue. Comparison of the4.3-kDa band (corresponding to A  ) from supernatant fraction andpellet fractions reveals more pelletable A   when coincubated withagrin. Additionally, there was a concomitant decrease in A   bandintensity from the supernatant fraction of the A  (1–40)-agrin samplewhen compared to supernatant from A  (1–40) alone (A). To examineagrin’s localization in the centrifuged fractions, samples were sepa-rated on a 4% SDS–PAGE gel and Western blotted. The blot wasprobed for agrin using the 6D2 mAb.An immunopositive smear from400 to 600 kDa (representative of agrin’s appearance when separated by SDS–PAGE) was seen in the pellet fraction when agrin wascoincubated with A   (1–40). No agrin was observed in the pelletfraction whenA  (1–40) was omitted from the incubation (B). Agrin Interactions and Localization with A   187
Related Search
Advertisements
Related Docs
View more...
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks