Association of Heparan Sulfate Proteoglycan with the Neurofibrillary Tangles of Alzheimer's Disease
The major intracytoplasmic lesion of Alzheimer's disease is the neurofibrillary tangle (NFT), which is primarily composed of paired helical filaments (PHFs). The mechanism responsible for the formation of PHFs, as well as their insolubility and apparent heterogeneity, is unknown. We found that basic fibroblast growth factor (bFGF) binds to heparinase-sensitive sites in NFTs. bFGF binding is due to a heparan sulfate proteoglycan (HSPG) immunocytochemically identified in NFTs. In the presence of polycations (e.g., Ca2+), HSPG will bind to free carboxyl groups in NFT proteins. HSPG binding may play a role in transforming normal soluble proteins into insoluble PHFs.
The Journal of Neuroscience, November 1991, If(11): 3679-3683
Association of Heparan Sulfate Proteoglycan with the Neurofibrillary
Tangles of Alzheimer’s Disease
George Perry,’ Sandra L. Siedlak,’ Peggy Richey,’ Mitsuru Kawai,’ Patrick Gras,’ Rajesh N. Kalaria,’ Pamela
G. Galloway,’ Jan Miriam Scardina,* Barbara Cordell,2 Barry D. Greenberg,3 Steven R. Ledbetter, and
‘Division of Neuropathology, Institute of Pathology, Case Western Reserve University, Cleveland, Ohio 44106, 2California
Biotechnology, Mountain View, California 94043, and 3Upjohn Company, Kalamazoo, Michigan 49001
The major intracytoplasmic lesion of Alzheimer’s disease is
the neurofibrillary tangle (NFT), which is primarily composed
of paired helical filaments (PHFs). The mechanism respon-
sible for the formation of PHFs, as well as their insolubility
and apparent heterogeneity, is unknown. We found that ba-
sic fibroblast growth factor (bFGF) binds to heparinase-sen-
sitive sites in NFTs. bFGF binding is due to a heparan sulfate
proteoglycan (HSPG) immunocytochemically identified in
NFTs. In the presence of polycations (e.g., Ca*+), HSPG will
bind to free carboxyl groups in NFT proteins. HSPG binding
may play a role in transforming normal soluble proteins into
The composition and mode of formation of the neurofibrillary
tangle (NFT’), a distinctive neuronal lesion, composed of paired
helical filaments (PHFs) in Alzheimer’s disease (AD), is con-
troversial (Perry, 1987). Two features that make the understand-
ing of the NFT difficult are the complexity of their composition
and their insolubility. Although it has recently been demon-
strated that a soluble form of PHF is solely composed of the
microtubule-associated protein tau (Greenberg and Davies, 1990;
Lee et al., 1991), NFTs are largely insoluble and contain other
antigens distinct from tau, including neurofilaments (Gambetti
et al., 1987; Mulvihill and Perry, 1989) and tropomyosin (Gal-
loway et al., 1990). The mechanism responsible for the incor-
poration of these presumably associated proteins into NFTs is
In a previous study of the role of growth factors in the for-
mation of senile plaques, we found that basic fibroblast growth
factor (bFGF) binds to NFTs (Perry et al., 1990a,b; Siedlak et
al., 199 1). We now show that bFGF is bound to a heparan sulfate
proteoglycan (HSPG) that is also present in NFTs. We found
that HSPG binds to the free carboxyl groups of the PHFs and
suggest that similar interactions may mediate the binding of
other proteins to NFTs and render them insoluble.
Received Apr. 23, 1991; accepted June 10, 1991.
We thank Sandy Bowen for manuscript preparation and Drs. L. Culp and T.
Rosenheny for advice. This work was supported by NIH Grants AG007552,
AG009287, and AG00795. G.P. is the recipient of NIH Research Career Devel-
opment Award K04-AG00415, and P.C. and M.K. are fellows of the Fogarty
Correspondence should be addressed to George Perry, Ph.D., Division of Neu-
ropathology, Case Western Reserve University, 2085 Adelbert Road, Cleveland,
Copyright 0 199 1 Society for Neuroscience 0270-6474/91/l 13679-05$05.00/O
Materials and Methods
Tissue. The hippocampal gyrus was taken from nine pathologically con-
firmed (Khachaturian, 1985) cases of AD (mean age, 82 yr; range, 73-
88 yr) and two controls (60 and 64 yr) and placed in methacam (meth-
anol 6 : chloroform 3 : acetic acid l), a fixative that makes no covalent
modification, for 24-48 hr before dehydration and embedding in par-
affin. Tissue from two AD cases (69 and 73 yr) was frozen in liquid
nitrogen-chilled isopentane. We also used a Bouin’s-Hollande fixed,
paraffin-embedded sample taken from the frontal cortex of an AD case
(74 yr) taken at biopsy (gift of Dr. S. Chou, Cleveland Clinic Founda-
tion). Paraffin-embedded sections were cut at 6 pm, and frozen sections,
at 10 Wm.
Antibodies. The following antibodies were used: (1) monoclonal an-
tibody (Ig fraction) to an epitope located in the carboxyl terminal of
bFGF, 48.1 (J. M. Scardina, unpublished observations), and (2) rabbit
antiserum to an HSPG core protein (Ledbetter et al., 1987; Noonan et
Protein preparations. The following proteins were used: (1) human
bFGF (154 amino acid form) recombinantly produced in Escherichiu
co/i and purified to homogeneity by using anion exchange chromatog-
raphy, (2) heparan sulfate proteoglycan core protein isolated from Engel-
breth-Helm-Swarm tumor (Ledbetter et al., 1985. 1987: Noonan et al.,
1988) and (3) ovalbumin obtained from Sigma.
Preparation of peroxidase-protein conjugates. Peroxidase-conjugated
proteins were prepared by coupling horseradish peroxidase (HRP) to
ovalbumin (as a control), bFGF, or HSPG core protein. Briefly, proteins
were dialyzed into 0.9% NaCl in 1 M NaHCO,, pH 9.5, and incubated
for 16 hr at room temperature (RT) with activated peroxidase Pierce
at a ratio adjusted for the molecular weight of the protein to be coupled.
Excess reactive groups were blocked with 0.2 M Tris, and 1% bovine
serum albumin, followed by dialysis in 150 mM NaCl and 50 mM Tris,
Staining. Sections were immunostained with the antibodies by em-
ploying the peroxidase-antiperoxidase procedure (Stemberger, 1986)
with 3,3’-diaminobenzidine as cosubstrate. Prior to immunostaining,
sections were treated with 3% H,O, in methanol for 30 min to block
endogenous peroxidase. HRP-protein conjugates were visualized by
development for peroxidase activity as above. In some experiments,
tissue sections were treated with 0.2 M NaOH at RT for 1 hr prior to
staining. In control experiments, primary antibodies were omitted and
the results compared with our experimentals. The extent of ligand bind-
ing was evaluated by comparing staining intensities.
Competition, adsorption, and binding experiments. In order to ascer-
tain the specificity of bFGF binding to NFI in sections, we incubated
bFGFat 4°C for 22 hr with either 100 &ml heparin, 5 mg/ml polylysine,
or 30 &ml protamine and then applied the solutions to the sections
followed by immunostaining with monoclonal antibody to bFGF 48.1.
In other exneriments. we aDDlied 0.5 &ml bFGF-HRP together with
various concentrations of bPGF (0.15 ng to 160 &ml serially diluted
1:4) to the section and incubated for 16 hr at 4°C. In adsorption ex-
periments, the antibody was incubated with antigen for 16 hr at 4°C
before application to tissue sections. Binding of bFGF or HSPG to NFI
without HRP labeling was assessed by incubating the tissue section with
ligand, rinsing with the same buffer used for incubation, hxing with 3.7%
formaldehyde for 5 min, and then immunostaining with an antibody
to the ligand, but at a greater dilution that does not identify endogenous
3660 Perry et al. - HSPG in Neurofibrillary Tangles
1. bFGF binds to neurofibrillary tangles
and neuropil threads
well as to amyloid deposits (*). In control sections (not shown), bFGF
binding was limited to blood vessels and the few age-related NITS.
Scale bar, 25 Nrn.
NFfs recognized by an antibody to HSPG, note that the
NPTs recognized are intraneuronal. Nucleus is indicated by
Additionally, intraneuronal granules are stained. No staining of NFTs
or granules was seen in the absence of primary antibody. Scale bar, 25
HSPG. In control experiments, the sections were stained with the same
antibody dilution minus the ligand preincubation.
pH and Ca2+ dependence of bFGF and HSPG binding.
of pH on bFGF or HSPG binding to tissue sections was evaluated by
incubating ligand or ligand-HRP at pH 3.1,4.1, 5.0, 6.0, 7.0, 7.6, 8.0,
9.0, 9.9, and 10.8 by using 0.1
Bis-Tris-HCl from pH 3.1 to 6.0, 0.1
Tris-HCl for pH 7.0-9.0, and 0.05
fonic acid-HCl for pH 9.9 and 10.8 as buffers. The dependence on Ca2+
was determined by using CaZ+/EGTA buffers and ratios of 0.3, 0.5,
0.75, 0.9, 1.0, 1.05, 1.25, 1.5, and 3.0 with 2 mM EGTA (Potter and
Gergely, 1975) in 150 mM NaCl and 50 mM Tris-HCl, pH 7.85. Greater
than 1 mM free Ca*+ solutions omitted EGTA.
Protein and sugar mod@ication treatments.
sections were incubated with one of the following: (1) 0.05 U/ml of
either heparinase (0.05 U/ml, ICN, or 1 U/ml, Sigma), heparinase 2 (1
U/ml, Sigma), heparitinase (0.05 U/ml, ICN or Seikagaku), or a com-
bination of heparinase and heparitinase at RT for 22 hr in 20 mM CaCl,
and 10 mM Tris, pH 7.0 (Schubert et al., 1988). The specificity of
heparinase or heparitinase treatment for heparan removal from sections
was shown by comparison with the effect of 0.0 1 &ml chondroitinase
ABC (ICN) in 20 mM CaCl, and 10 mM Tris, pH 8.0, incubated at RT
for 22 hr or 400 &ml trypsin (Worthington) at 37°C for 10 min in 300
mM NaCl, 20 mM CaCl,, and 50 mM Tris, pH 7.6. As controls for any
residual protease activity in the heparinase preparation, the inhibitor
0.5 mM leupeptin, 1 mM phenylmethylsulfonyl fluoride, or exogenous
protein (1 mg actin/ml) was used. (2) Two percent periodic acid for 16
hr at RT to modify sugar residues (Behrouz et al., 1989); (3) nitrous
acid consisting of 0.24
sodium nitrite in 1.8
acetic acid for 90 min
at RT to modify N-linked sulfates (Hirabayashi et al., 1989); (4) 4
U/ml neuraminidase (Sigma. tvoe VI) in 0.2
acetate. DH 5.4. for 2
hr at 37°C to remove sia!ic &l-residues (Szumanska et al., 1987); (5)
ethanolamine with the addition of 20 mM 1 -ethyl-3-(3-dimethyl-
aminopropyl) carbodiimide (carbodiimide) after 10 min at 37°C to
block and convert carboxyl groups to secondary amines (Taniuchi et
al., 1986); (6) 10 mM paraformaldehyde with 60 mM sodium cyano-
borohydride, 10 mM EDTA, and 10 mM phosphate, pH 7.0, for 2 hr at
37°C to block amino groups (Peterson et al., 1979); (7) 400 &ml alkaline
phosphatase (Sigma, type VII-S, bovine intestinal) in 0.1
8.0, with 0.0 1
phenylmethylsulfonyl fluoride for 18 hr followed by 2
U/ml acid phosphatase (Sigma; prostatic, 3200 U/mg) in 0.1
pH 5.0, for 3 hr at 37°C (Ueda et al., 1990); or (8) an unembedded
methacam-fixed tissue block of 3 mm thickness treated with 50% hy-
drofluoric acid for 48 hr at 4°C (Mayor et al., 1990) to remove phosphate
groups followed by paraffin embedding.
bFGF bound to NFTs (Fig. 1) as well as to amyloid deposits of
senile plaques, an extracellular filamentous lesion, and vessel
walls in both control and AD cases. However, the binding was
more prominent in the AD cases because of the larger number
of NFTs and senile plaques. bFGF binding, regardless as to
whether it was with NFTs, amyloid deposits, or vessels, had the
characteristics shown in Table 1. bFGF showed saturable bind-
ing that appeared to be charge dependent since CaZ+ or other
polycations as well as anions inhibited the binding. Since bFGF
is known to bind to its receptor as well as to HSPG (Rillcin and
Moscatelli, 1989; Yayon et al., 1991), bFGF binding to NFTs
suggested that HSPGs might be present. The blocking of bFGF
binding by treatment of tissue with heparinases or heparitinase
Table 1. Characteristics of bFGF and HSPG binding to NFTs
c9.0 X lfk6
Figure3. Intraneuronal NFIs, as well as extracellular NFl3 (not shown),
bind HSPG (HSPG-HRP). NITS and intraneuronal granules recognized
in neuronal cell bodies containing a nucleus are indicated by arrowheads.
Differential interface contrast microscopy. Scale bar, 25 pm.
and the finding that an antiserum to an HSPG immunoreacted
with NFTs (Fig. 2) supported bFGF interaction with HSPG in
In order to understand how HSPGs are incorporated into
NFTs, we characterized the HSPG-binding sites in NFTs by
studying the properties of HSPG binding to NFIs (Fig. 3). The
binding characteristics (Tables 1 and 2) suggest that CaZ+ acts
as a bridge between anionic groups on HSPG, for example,
removal of N-linked sulfate groups by nitrous acid or sugar
residue oxidation with periodate blocked HSPG binding to
and anions in NFTs. Consistent with the nonspecific nature of
the polycation requirement, protamine (30-100 &ml) can sub-
The Journal of Neuroscience, November 1991, 1 f(11) 3681
stitute for Ca2+. The Ca*+ inhibition of bFGF binding to HSPG
(anion) in NFTs, is probably mediated by the cationic region
of bFGF (Seno et al., 1990). HSPG binding to NFTs is probably
to cationic groups in NFT, candidates include sugar residues of
P-component, a serum glycoprotein (Kalaria and Grahovac,
1990; Kalaria et al., 199 l), HSPG, phosphate, or carboxyl res-
idues in PHF proteins. To distinguish among these possibilities,
we treated sections with a variety of reagents that alter either
sugar residues or amino acid side chains (Table 2). The same
treatments were performed to characterize bFGF-binding sites.
Alteration of sugar residues with reagents such as periodic acid
or nitrous acid, neuraminidase, or heparinases generally in-
creased HSPG binding to NFT. Thus, not only did HSPG not
seem to be bound to sugar residues in NFTs, but the removal
of sugar residues seemed to unmask new binding sites. In con-
trast, sugar-modifying treatments generally reduced bFGF bind-
ing. Moreover, treatment with formaldehyde-cyanoborohy-
dride, to block amino groups, or with phosphatases or
hydrofluoric acid (Mayor et al., 1990), to remove phosphate
groups, did not decrease HSPG binding, indicating that free
amino and phosphate groups in NFTs did not play a role in
HSPG binding. Treatment with carbodiimide, which had no
effect on bFGF binding, blocked HSPG binding, indicating that
free carboxyl groups are required for HSPG binding to NFI.
A perplexing aspect of HSPG binding was that although HSPG
is a protein of the extracellular matrix, HSPG was found in
intraneuronal NFTs. Two alternative possibilities are that the
association of HSPG with NFTs has occurred during postmor-
tem autolysis or after extracellularization of NFTs. Both pos-
sibilities are unlikely because identical findings were obtained
with the tissue obtained at biopsy (not shown), and we found
that neurons with abundant cytoplasm and the nucleus had
HSPG immunoreactivity (Fig. 2) and HSPG-binding sites (Fig.
3). Therefore, HSPG must be associated with intracellular NFTs
It has been well established that extracellular amyloid deposits
of AD associate with molecules present in the extracellular ma-
trix (Snow et al., 1988; Snow and Wight, 1989). However, there
Table 2. Effect of modifying treatments on bFGF and HSPG binding to NFTs
Treatment Modification bFGF
Heparinase 2 Heparan removal
Heparitinase Heparan removal
heparinase Heparan removal
Heparitinase + heparitinase
with protease inhibitor Heparan removal
Chondrotinase ABC Chondroitin removal
Neuraminidase Sialic acid removal
Periodic acid Sugar oxidation
Nitrous acid N-linked sulfate removal
Carbodiimide Carboxyl blockage
Formaldehyde Amino blockage
Phosphatases Phosphate removal
Hydrofluoric acid Phosphate removal
Effect on binding: T, increased; -, no effect; 1, decreased.
3682 Perry et al. * HSPG in Neurofibrillaty Tangles
is considerable evidence that a glycoprotein, possibly a proteo-
glycan, is also associated with NFTs (Mann et al., 1988; Ro-
senblatt et al., 1989; Snow et al., 1989, 1990). The present study
confirms and extends these findings by characterizing the bio-
chemical basis for HSPG interaction with NFT. We have found
that HSPG- and bFGF-binding sites are present in NFTs. Both
these components are likely to be bound to an anion in NFTs
provided by HSPG for bFGF and by carboxyl residues for HSPG.
HSPG incorporation into NFTs occurs intraneuronally and is
therefore not an artifact resulting from contamination of extra-
The presence of HSPG in NFTs may provide an explanation
for two puzzling features of these inclusions: number of the
components and the insolubility. HSPG is present in all known
amyloid filaments in systemic and cerebral amyloidoses, and
HSPG deposition is concomitant with amyloid filament for-
mation (Snow et al., 1987a). Although amyloid filaments form
in vitro by self-assembly, in vivo HSPG is likely to be required
as an initiator of amyloid formation because of the low con-
centration of free amyloidogenic subunits (Kisilevsky and Snow,
1988; Snow and Wight, 1989). HSPG presumably plays this role
because of its high negative charge, which allows it to concen-
trate and protect ligands from proteolysis while making insol-
uble complexes, as it apparently does for bFGF (Schreiber et
al., 1985; Gospodarowicz and Cheng, 1986; Saksela et al., 1988).
HSPG may play a similar role in the formation of the ab-
normal filaments of NFTs. It is known that HSPG can bind
proteins other than bFGF, some of these proteins such as am-
yloid precursor protein (Schubert et al., 1988, 1989; Klier et al.,
1990), the asymmetric form of AChE (Brandon et al., 1985;
Mesulam and Moran, 1987) and P-component (Hamazaki, 1987;
Duong et al., 1989; Kalaria and Grahovac, 1990; Kalaria et al.,
199 1) have been found in NFT’s. Therefore, the compositional
complexity of NFTs may be due to the binding of proteins with
multiple binding sites triggered by the presence of HSPG, a
phenomenon similar to that occurring in the extracellular matrix
in which various proteins self-assemble through multiple hetero-
and homologous bonds (Yurchenco and Johannes, 1990). More-
over, the binding of HSPG to carboxyl groups may explain why
the identified PHF constituents, tau, neurofilaments, and tropo-
myosin, are abundant in glutamate and aspartate residues
(Cleveland et al., 1977; Geisler et al., 1983; Matsumura et al.,
198 5). Consistent with this interpretation, antibodies to tau block
HSPG binding (S.L. Siedlak and G. Perry, unpublished obser-
vations). While HSPG has not been detected in enriched PHF
fractions (Sparkman et al., 1990; Lee et al., 199 1; P. Mulvihill
and G. Perry, unpublished observations), its association with
PHF may depend on polycationic bridges not maintained in the
isolation procedure. Although PHF may be primarily a polymer
of tau, PHF formation may require incorporation of multiple
components that are rendered insoluble as they are incorporat-
ed. HSPG’s interaction with PHF proteins may provide the key
to the understanding of these mechanisms.
Cleveland DW, Hawo SY, Kirschner MW (1977) Physical chemical
properties of purified tau factor and the role of tau in microtubule
assembly. J Mol Biol 116:227-247.
Duong T, Pommier EC, Scheibel AD (1989) Immunodetection of
human serum amyloid P-component in Alzheimer’s disease. Acta
Neuropathol (Berl) 78429437.
Galloway PG, Mulvihill P, Siedlak S, Mijares M, Kawai M, Padgett H,
Kim R, Perry G (1990) Immunochemical demonstration of tropo-
myosin in the neurofibrillary pathology of Alzheimer disease. Am J
Gambetti P, Autilio-Gambetti L, Manetto V, Perry G (1987) Com-
position of paired helical filaments of Alzheimer’s disease as deter-
mined by specific probes. In: Banbury report 27, Molecular neuro-
pathology of aging (Davies P, Finch CE, eds), pp 309-320. Cold Spring
Harbor, NY: Cold Spring Harbor Laboratory.
Geisler M, Kaufman E, Fischer S, Plessmann U, Weber K (1983)
Neurofilament architecture combines structural principles of inter-
mediate filaments with carboxyl-terminal extensions increasing in size
between triplet nroteins. EMBO J 2: 1295-1302.
Gospodarowicz 6, Cheng J (1986) Heparin protects basic and acidic
bFGF from inactivation. J Cell Physiol 126465484.
Greenberg SG. Davies P (1990) A nrenaration of Alzheimer oaired
helical ilaments that displays ‘distinct; proteins by polyacrylamide
electrophoresis. Proc Nat1 Acad Sci USA 875827-5831.
Hamazaki H (1987) Ca2+ mediated association of human serum am-
yloid P-component with heparan sulfate and dermatan sulfate. J Biol
Chem 262: 1456-1460.
Hirabayashi Y, Shimizu S, Yamada K (1989) A nitrous acid procedure
as a selective histochemical means of eliminating the N-sulfate of
glycoconjugates. Histochem J 2 1:687-692.
Kalaria RN, Grahovac I (1990) Serum amyloid P immunoreactivity
in hippocampal tangles, plaques and vessels: implications for leakage
across the blood-brain barrier in Alzheimer disease. Brain Res 5 16:
Kalaria RN, Galloway PG, Perry G
(199 1) Widespread serum amyloid
P immunoreactivity in cortical amyloid deposits and the neurofi-
brillary pathology of Alzheimer’s disease and related disorders. Neu-
ropathol Appl Neurobiol 17: 189-20 1.
Khachaturian ZS (198 5) Diagnosis of Alzheimer’s disease. Arch Neu-
rol 42:1097-l 105.
Kisilevsky R, Snow A (1988) The potential significance of sulphated
glycosaminoglycans as a common constituent of all amyloid: or, per-
haps amyloid is not a misnomer. Med Hypotheses 26:23 l-236.
Kher FG, Cole G, Stallcup W, Schubert D (1990) Amyloid p-protein
precursor is associated with extracellular matrix. Brain Res 5 15:336-
Ledbetter SR, Tyrell B, Hassell JR, Horigan EA (1985) Identification
of the precursor protein to basement membrane heparan sulfate pro-
teoglycans. J Biol Chem 260:8 106-8 113.
Ledbetter SR, Fisher LW, Hassell JR (1987) Domain structure of the
basement membrane heparan sulfate proteoglycan. Biochem 26:988-
Lee VMY, Balin BJ, Otvos L, Trojanowski JQ (1991) A68: a major
subunit of paired helical filaments and derivatized forms of normal
Mann DMA, Bonshek RE, Marcyniuk B, Stoddart RW, Torgerson E
(1988) Saccharides of senile plaques and neurofibrillary tangles in
Alzheimer’s disease. Neurosci Lett 85277-282.
Matsumura F, Yamashiro I, Matsumura S (1985) Purification and
characterization of multiple isoforms of tropomyosin from rat cul-
tured cells. J Biol Chem 260:13851-13859.
Mayor S, Menor AK, Cross GAM, Ferguson MAJ, Dwek RA, Rade-
macher TW (1990) Glycolipid precursors for the membrane anchor
of Trypanosoma brucei variant surface glycoproteins. J Biol Chem
Mesulam MM, Moran MA (1987) Cholinesterases within neurofi-
brillary tangles related to age and Alzheimer’s disease. Ann Neurol
Behrouz N, Defossez A, Delacourte A, Hublau P, Mazzucca M (1989) Mulvihill P, Perry G (1989) Immunoaffinity demonstration that paired
Alzheimer disease: glycolytic pretreatment dramatically enhances im-
helical filaments of Alzheimer disease share epitopes with neurofi-
munolabelling of senile plaques and cerebrovascular amyloid sub-
laments, MAP2 and tau. Brain Res 484: 150-l 56.
stance. Lab Invest 61:576-583.
Noonan DM, Horigan EA, Ledbetter SR, Vogeli G, Sasaki M, Yamada
Brandon E, Maldonado M, Garrido J, Inestrosa NC (1985) Anchorage
Y, Hassel JR (1988) Identification of cDNA clones encoding dif-
of collagen-tailed acetylcholinesterase to the extracellular matrix is
ferent domains of the basement membrane heparan sulfate proteo-
mediated by heparan proteoglycans. J Cell Biol 101:985-992.
glycan. J Biol Chem 263:16379-16387.
The Journal of Neuroscience, November 1991, 7 7(11) 3663
Perry G (ed) (1987) Alterations in the neuronal cytoskeleton in Alz-
heimer disease, pp 229. New York Plenum.
Perry G, Siedlak S, Kawai M, Cras P, Mulvihill P, Cordell B, Miriam
Scardina J, Gambetti P (1990a) Neurofibrillary tangles, neuropil
threads and senile plaques all contain abundant binding sites for basic
fibroblast growth factor (@FGF). J Neuropathol Exp Neurol49:3 18.
Perry G, Siedlak S, Mulvihill P, Kawai M, Cras P, Cordell B, Miriam
Scardina J, Ledbetter S, Greenberg B, Gambetti P (1990b) Neu-
rofibrillary tangles contain heparan sulfate proteoglycans. Neurobiol
Peterson DT, Merrick WC, Safer B (1979) Binding and release of
radiolabeled eukaryotic initiation factors 2 and 3 during 805 initiation
comnlex formation. J Biol Chem 254:2509-25 16.
Potter JP, Gergely J (1975) The calcium and magnesium binding sites
on troponin and their role in the regulation of myofibrillary adenosine
triphosphate. J Biol Chem 250:4628-4633.
Rifkin DB, Moscatelli D (1989) Recent developments in the cell bi-
ology of basic fibroblast growth factor. J Cell Biol 109: l-6.
Rosenblatt DE, Geula C, Mesulam MM (1989) Protease nexin I im-
munostaining in Alzheimer’s disease. Ann Neurol 26:628-634.
Saksela K, Moscatelli D, Sommer A, Ratkin DB (1988) Endothelial
cell-derived sulfate binds basic fibroblast growth factor and protects
it from nroteolvtic dearadation. J Cell Biol 107:741-75 1.
Schreiber AB, Kenney J; Kowalski WJ, Friesel R, Mehlman T, Maciag
T (1985) Interaction of endothelial cell growth factor with heparin:
characterization by receptor and antibody recognition. Proc Nat1 Acad
Sci USA 82:6138-6142.
Schubert D, Schroeder R, LaCorbiere M, Saitoh T, Cole G (1988)
Amyloid fl protein is possibly a heparan sulfate proteoglycan core
protein. Science 241:223-224.
Schubert D, LaCorbiere M, Saitoh T, Cole G (1989) Characterization
of an amyloid precursor protein that binds heparin and contains
tyrosine sulfate. Proc Nat1 Acad Sci USA 86:2066-2069.
Seno M, Sasada R, Kurokawa T, Igarashi K (1990) Carboxyl-terminal
structure of basic fibroblast growth factor significantly contributes to
its affinitv for henarin. Eur J Biochem 188:239-245.
Siedlak SL,Cras P, Kawai M, Richey P, Perry G (199 1) Basic fibroblast
growth factor binding is a marker for extracellular neurofibrillary
tangles in Alzheimer’s disease. J Histochem Cytochem 39:899-904.
Snow AD, Wight TN (1989) Proteoglycans in the pathogenesis of
Alzheimer’s disease and other amyloidoses. Neurobiol Aging lo:48 l-
Snow AD, Willmer J, Kisilevsky R (1987a) Sulfated glycosaminogly-
cans: a common constituent of all amyloids? Lab Invest 56: 120-l 23.
Snow AD, Willmer J, Kisilevsky R (1987b) A close ultrastructural
relationship between sulfated proteoglycans and AA amyloid fibrils.
Lab Invest 57:687-698.
Snow AD, Mar H, Nochlin D, Kimata K, Kato M, Suzuki S,