The EMBO Journal vol.4 no.11 pp.2757 -2763 1985
Neuronal origin of a cerebral amyloid: neurofibriliary tangles of
Alzheimer's disease contain the same protein as the amyloid of
plaque cores and blood vessels
Colin L.Masters1'2, Gerd Multhaup3, Gail Simms', Jutta
Pottgiesser3, Ralph N.Martins1 and Konrad Beyreuther3
'Laboratory of Molecular and Applied Neuropathology, Neuromuscular
Research Institute, Department of Pathology, University of Western
Australia, Western Australia 6009, and 2Department of Neuropathology,
Royal Perth Hospital, Perth, Western Australia 6001, and 3Institute for
Genetics, University of Cologne, Weyertal 121, D-5000 Cologne 41, FRG
Communicated by K.Beyreuther
The protein component of Alzheimer's disease amyloid
[neurofibrillary tangles (NFI), amyloid plaque core and con-
gophilic angiopathy] is an aggregated polypeptide with a
subunit mass of4 kd (the A4 monomer). Based on the degree
of N-terminal heterogeneity, the amyloid is first deposited in
the neuron, and later in the extracellular space. Using anti-
sera raised against synthetic peptides, we show that the N ter-
minus of A4 (residues 1-11) contains an epitope for
neurofibrillary tangles, and the inner region of the molecule
(residues 11-23) contains an epitope for plaque cores and
vascular amyloid. The non-protein component of the amyloid
(aluminum silicate) may form the basis for the deposition or
amplification (possible self-replication) of the aggregated
amyloid protein. The amyloid of Alzheimer's disease is similar
in subunit size, composition but not sequence to the scrapie-
associated fibril and its constituent polypeptides. The sequence
and composition ofNFT are not homologous to those of any
of the known components of normal neurofilaments.
Key words: protein sequence/h.p.l.c./scrapie/aluminum silicate/
Alzheimer's disease (AD) is the commonest form of cerebral
degeneration leading to dementia in late adult life. The peak in-
cidence occurs in the eighth decade, and the average duration
of illness is 6 years, steadily progressing without remission or
relapse. The pathognomonic changes in the AD brain are the for-
mation of fibrillar amyloid deposits in intracellular and ex-
tracellular locations: intracellular neurofibrillary tangles (NFT)
composed of paired helical filaments (PHF) within neurons and
within pre-synaptic axonal terminals surrounding extracellular
deposits of amyloid plaque cores (APC), and a variable degree
of amyloid congophilic angiopathy (ACA). There
widespread neuronal loss and gliosis in areas affected by NFT
and APC formation. These typical changes ofAD also occur in
the brains of aged individuals with Down's syndrome (DS).
Although the cause ofAD is unknown, it shares many features
with the unconventional virus diseases [Creutzfeldt-Jakob disease
(CJD), kuru and scrapie] (Masters et al., 198la, 198lb), but not
least of which are the occurrence of amyloid plaques (Masters
et al., 1981b); the similarities in morphology between amyloid
fibrils, PHF and scrapie-associated fibrils (SAF) (Merz et al.,
1981; DeArmond et al., 1985), the Congo Red binding proper-
ties of SAF (Prusiner et al., 1983); the immunological cross-
reactivity between scrapie-associated proteins and scrapie amyloid
©C IRL Press Limited, Oxford, England.
plaques (Bendheim et al., 1984; DeArmond et al., 1985); the
observation that a method used to purify SAF when applied to
an AD brain will yield PHF (Rubenstein et al., 1985); and that
the size and amino acid composition of the scrapie-associated pro-
teins are consistent with amyloid proteins in general (Multhaup
et al., 1985).
As a first step in understanding the pathogenesis of AD, we
recently isolated and characterized the protein component of the
APC (Masters et al., 1985), and found it to consist ofaggregates
of a 4-kd polypeptide (A4 monomer, A8 dimer, A16 tetramer,
A64 hexadecamer), similar in amino acid composition and se-
quence to ACA (Glenner and Wong, 1984). However the APC-
A4 was found to have ragged N termini, in contrast to ACA-A4
which was reported as being composed of only the full length
major sequence. The same A4 protein was found to be the con-
stituent of the APC and ACA ofDS (Masters et al., 1985; Glen-
ner and Wong, 1984).
We have now examined the composition ofpurified NFT-AD,
and report that they contain the same A8, A16 and A64 oligomeric
species of the A4 monomer as found in APC and ACA. The N
termini of the NFT protein are even more ragged than found for
APC. From these observations we can draw several conclusions
on the probable pathogenesis of NFT, APC and ACA. Since the
subunit protein of SAF is now also known in some detail
(Multhaup et al., 1985; Oesch et al., 1985), comparisons can
be drawn between the various forms of cerebral amyloid proteins.
Results and Discussion
The enriched fractions of NFT (Figure 1) contain lipofuscin and
collagen as the major contaminants. Some APC and fragments
thereof are seen by electron microscopy, but overall 50-90%
of the fraction consists of NFT, as judged by light and electron
microscopy. The purity of the preparations varies from case to
case, and from those individuals with the greatest numbers of
NFT (enumerated by histology of the contralateral hemisphere)
the highest and cleanest yields of NFT are obtained.
NFT and APC are composed of the A4 protein subunit
The protein component of NFT was extracted with 100% for-
mic acid. To assess the degree of protein solubilization, we us-
ed quantitative amino acid analysis of HCl-hydrolysates of NFT
before and after formic acid extraction. More than 90% of the
NFT protein was soluble in formic acid, and the amino acid com-
positions of total NFT protein before and after extraction were
almost indistinguishable (Table I). SDS (0.1-10%) does not
solubilize the NFT proteins in their native state, a property shared
with the APC proteins (Masters et al., 1985). However, once
acid and thus
proteinaceous components, these proteins are soluble in 0.1 %
SDS buffers, but are still not dissociated into monomeric subunits,
until treated with 6 M urea. When electrophoresed in SDS-urea
polyacrylamide gels, the 4-kd monomer is seen, identical to the
APC-A4 protein (Figure 2).
separated from non-
C.L.Masters et al.
elongated and flame-shaped, typical of intracellular accumulations of NFT. (Magnification X 550).
1. An enriched fraction NFT stained with C.ongo Red and viewed with polarization light microscopy. Most of the negatively birefringent material
Table I. Amino acid composition of NFT and APC proteins
Number of residues per monomer
H.p.l.c. fractions of formic
proteind (7 kd)
aFor 2.5 jig of NFT protein.
bTryptophan was not determined.
cCalculated composition of the A4 subunit of APC from AD and DS individuals (Masters et al., 1985).
dDeduced for the CLAC strain of the 7-kd deglycosylated hamster scrapie-associated protein (Multhaup et al., 1985).
NFT- andAPC-A4 subunits aggregate into dimers, tetramers and
tion chromatography, the NFT proteins have mol. wts. of 64,
16 and 8 kd (A64, A16, A8) (Figure 3a). The NFT-A64 is
dissociated into NFT-A8 after formic acid treatment prior to re-
chromatography (Figure 3c). NFT-A16 can also be broken down
to A8 species after re-extraction with formic acid (data not
After solubilization in formic acid, the NFT protein can be
chromatographed in 0.1 % SDS buffers. By h.p.l.c. gel permea-
Alzheimer's disease amyloid protth
Fig. 2. Gel electrophoresis of NFT and APC proteins. SDS/urea
polyacrylamide gel electrophoresis of formic acid extracts of oxidized NFT
proteins (12.2jig,lane 1), oxidized APC proteins (9.6 peg, lane 2), NFT
proteins (12.2isg, lane 3) and APC proteins (9.6jig,lane 4). Lanes Ml
and M2 are marker proteins. Both NFT and APC proteins behave in the
same electrophoretic manner - most of the protein migrates as the A4
shown). Thus, NFT-A64, A16 and A8 appear to be hexadecamers,
tetramers and dimers of 4-kd subunits. The APC proteins have
identical aggregational properties (Masters et al., 1985 and Figure
3). A8 dimers of both NFT and APC are not dissociated into
monomers by either SDS or formic acid. But urea and SDS
together will lead to dissociation into A4 monomers (Figure 2).
A8 dimers are seen as faint bands in SDS-urea gels of freshly
prepared material (Figure 2) and as more prominent bands from
preparations that have been left at room temperature for some
time (Masters et al., 1985).
Are the amyloid fibril proteins stabilized by interchain S-S
bridges? The amino acid compositions of NFT- and APC-A4 in-
dicate the presence of one half cystine per chain (Table I). Per-
formic acid treatment of NFT and APC proteins prior to gel
electrophoresis does not cause any appreciable change, suggesting
that the native tangle and plaque core A4 proteins are devoid of
interchain disulfide linkages (Figure 2). Both APC- and NFT-
A4 monomers show the same pH- and concentration-dependent
aggregation. At pH values > 7, higher aggregates dominate and
little A16 and A8 species are detected by h.p.l.c. in buffers con-
taining SDS. This aggregation tendency is completely changed
at pH values < 7, and almost no A64 species are seen if the pro-
tein concentration is kept below 0.1 mg/mi. Above this concen-
tration, the excess protein almost exclusively occurs as an increase
in the A64 peak, whereas the A8 and A16 peaks remain relatively
The amino acid composition and sequences ofNFTandAPCpro-
teins are the same
The amino acid composition of NFT-A4 (as deduced from the
compositions of NFT-A64, A16, A8 and whole NFT) is very
similar to APC-A4 (Table I). This alone suggested that the NFT-
A4 and APC-A4 were the same protein. Gas-liquid solid phase
sequencing of unfractionated NFT protein and NFT-A8 confirms
the identity of the NFT and APC proteins (Figure 4). The N-
terminal sequence of the NFT proteins shows a large degree of
heterogeneity (Figure 4); more heterogeneous, in fact, than seen
for the APC protein. If, as reported by Glenner and Wong (1984),
Fig. 3. H.p.l.c. gel permeation chromatography of NFT and APC proteins.
obtained by formic acid extraction of purified whole NFT or APC were
lyophilized, re-dissolved in 50 M1 of h.p.l.c. buffer and chromatographed at
a flow-rate of 0.2 ml/min. The peaks eluting at 57.0 min, 68.1 min and
71.7 min correspond to proteins of mol. wt. 64 kd (A64), 16 kd (A16) and 8
kd (A8), as deduced from the elution times of marker proteins. The peak at
110 min is the salt front. Protein peaks were collected, lyophilized, re-
dissolved in formic acid, lyophilized again and taken up in 50 ,ul of h.p.l.c.
buffer for re-chromatography. (C) is the re-chromatographic profile of the
NFT-A64 peak from (A). (D) is the re-chromatographic profile of the APC-
A64 peak from (B).
tg of NFT protein (A) and 12.5 yg of APC protein (B)]
the ACA protein exhibits complete N-terminal homogeneity
(Figure 4), then we may postulate that the NFT protein is chrono-
logically the oldest, the APC protein is intermediate, and that
the ACA protein is the most recently deposited form of the three
types of cerebral amyloid. One ought not to forget that the amy-
loid proteins of AD may have age ranges extending over two
orders of magnitude from 50 to 5000 days, at least based on the
known durations of the clinical disease. The N-terminal hetero-
geneity also excludes the possibility that residues 1-11 of the
A4 molecule are involved in the aggregational properties of the
NFT, APC and ACA share differential antigenic determinants
within the A4 monomer
Most rabbits immunized with whole APC eventually respond by
producing antibodies which react with both APC and ACA
.L.Masters et al.
(Figure 5a, b). Two rabbits immunized with APC [one with
alkali-degraded material, the other with sonicated APC coupled
to keyhole limpet haemocyanin (KLH)] responded by making
antibodies to NFT and not APC or ACA, as did one rabbit im-
munized with the h.p.l.c.-derived A16/A8 peaks from an APC
preparation (Figure Sc). The NFT stained in the sections ofAD
brain are restricted to the cell soma, and appear to be a select
population, as some NFT within the hippocampal regions were
not reactive. Antisera raised against the synthetic peptide cor-
responding to the intact N terminus of the A4 monomer [SP
A4(1-11)] also stained NFT in AD brain sections. However,
antisera raised against the synthetic peptide corresponding to the
middle portion ofthe A4 monomer [SP A4(1 1-23)] stained APC
and ACA and not NFT.
From these experiments we conclude first that NFT, APC and
ACA share antigenic determinants and that APC and ACA are
'antigenically more closely related to each other than to NFT,
and second that there are at least two epitopes on the A4
Asp - Ala- Glu - Phe-Arg- His-Asp-Ser-Gly- Tyr -Glu - Val - His-His-Gln---
100% Asp-Ala-Glu-Phe-Arg-His-Asp-Ser-Gly-Tyr-Gln-Val-His-His-Gtn ---
Fig. 4. N-terminal amino acid sequences of the 4 kd proteins (A64) from
NFT, APC and ACA. Purified NFT-A8 from three independent experiments
(with 6 yg, 8 yg and 12 yg protein) was obtained from h.p.l.c.-purified
material as shown in Figure 3. The sequences of NFT-A8 as presented here
differ from those of total NFT protein extracts in that the latter show a
constant background of proline and glycine residues accounting for -10%
of the material applied onto the glass fiber disk of the sequencer. We
interpret this as being due to collagen contaminants which co-purify with the
NFT. The sequences of APC-A4 and ACA-A4 are taken from Masters et al.
(1985) and Glenner and Wong (1984), respectively. The APC-A4 sequence
up to residue number 28 is now known (Masters et al., 1985).
monomer: the N-terminal region contains an antigenic site which
is exposed in intact NFT and in the A16/A8 species; the subunits
of the APC and ACA fibrils are assembled in such a manner
that the central portion of the monomer (at least in the regions
of residues 11 -23) is exposed. The full unravelling of the an-
tigenic structure of NFT, APC and ACA will prove to be more
complex than what has been presented above.
Selkoe and Abraham (1985) recently observed that rabbit an-
tisera to whole NFT also react with isolated APC. Their data
therefore confirm our observations that NFT, APC and ACA
share antigenic determinants. It is likely that multiple epitopes
exist [perhaps related to the N-terminal heterogeneity or to post-
translational modification, such as phosphorylation as postulated
by Sternberger et al., (1985)] which may partly explain the con-
flicting results of others who have found that antisera to normal
brain components (neurofilaments, microtubules, and vimentin)
cross-react with NFT (see, for example, Perry et al., 1985).
Monoclonal antibodies which recognize NFT and other normal
brain components (Grundke-Iqbal et al., 1985), react by Western
blotting with a series of polypeptides derived from NFT solubiiz-
ed in SDS. These polypeptides range between 16 and 65 kd, and
show a remarkable 'step-ladder' effect in which the bands differ
equally by 4-5 kd. Since these monoclonal antibodies also
recognize the A64 component of a PHF preparation (R.Ruben-
stein, personal communication), it is likely that these multiple
polypeptides differing by 4-5 kd represent the oligomeric state
of the NFT subunits in which the A4 monomer has been
dissociated in a step-wise fashion.
There is a non-proteinaceous component in APC and NFT
Extraction of purified NFT or APC with formic acid leads to
the complete removal of the A4 protein, but leaves behind a vis-
ible, insoluble residue. We have analyzed APC by energy disper-
sive X-ray analysis (both whole APC and their formic acid
residues), and found them to contain silicon, aluminum, sodium
(Schroder, Multhaup, Kisters, Masters and Beyreuther,
preparation). While in the center of the APC there is evidence
ofaluminum silicates containing sodium or other monovalent ions
(possibly in the form of 'clay-like' deposits of montmorillonites),
at the outer region ofthe APC, silicon salts are possibly the main
Fig. 5. Immunocytochemistry of antibodies raised against whole APC and h.p.l.c.-derived pooled APC-A16/A8 species. Rabbits immunized with whole APC
respond by producing antibodies to APC (a) and to the ACA (b). An antiserum raised against APC-A16/A8 species reacts with NFT (c). The sections are of
AD cortex, fixed in formalin, embedded in paraffin, using the unlabelled peroxidase-anti-peroxidase technique. Primary antisera were used at a dilution of
1:100. Pre-immune sera at 1:50 dilution were unreactive. (Scale bar 60 Mm).
Alzheimer's disease amyloid proteins
The neuronal origin of the cerebral amyloid
While it has been suggested for many years that NFT, APC and
ACA have a common origin, the results outlined above clearly
show this to be the case. If one accepts that the relative degrees
of N-terminal heterogeneity are a reflection of the age of the
deposited amyloid protein, then the conclusion is inescapable that
the A4 protein is derived from the neuron and deposited first in
its soma as a tangle and in its axonal terminals as the neuritic
degeneration accompanying the plaque. A sequence of events is
schematically presented in Figure 6, where after the initial deposi-
tion in the neuron, the amyloid spills into the extracellular space
to form the APC. Finally, and irregularly, the A4 protein is
deposited around and within blood vessels to form ACA. The
alternate hypothesis, that the A4 protein is derived from an
hematogenous precursor, and then transported to the neuron, now
seems quite unlikely.
Although the A4 protein does not show any sequence homology
with any of the known components of normal neurofilments
(Geisler et al., 1985), it may be anticipated that A4 is derived
from a component of the cytoskeleton. It may derive from the
accumulation of a normal or abnormal gene product (possibly
neurofilaments - Gajdusek, 1985) or it may represent a degrada-
tion product of some other component of the neuronal
cytoskeleton (such as microtubule-associated protein - Gray,
:NEUROFIRRILLARY TANGLE (NFT)
IN PERIKARYON OF NEURON
The size of A4 is unusual for a primary translational product
of a gene, and therefore A4 is most likely derived from a precur-
sor protein. However, post-translational processing ofthe putative
A4 precursor does not necessarily depend on a proteolytic activity.
The intriguing association of aluminum silicates with these
amyloid proteins (Nikaido et al., 1972; Perl and Brody, 1980)
deserves further study. The montmorillonites may form a self-
replicating system, structurally resembling fl-sheets; they may
bind polypeptides with a low content ofbasic amino acid residues,
and are capable of hydrolysing peptide bonds (Weiss, 1981). That
the neuronal amyloid proteins can assume paracrystalline arrays
(Metuzals et al., 1981) is consistent with an inorganically based
matrix. Although we have no evidence that the montmorillonite-
like inorganic residues are involved in the processing of the A4
protein or its precursor, such a concept does lead to interesting
speculation on the origin of aluminum deposits in the NFT of
the Guam dementia complex (Garruto et al., 1984), and even
into the realm of self-replicating proteins (for example, the
generation of amyloid proteins from precursors).
The structural basis of cerebral amyloid filaments
By electron microscopy, intracellular NFT are composed of
paired helical filaments (PHF), on average of larger diameter
and of more regular periodicity than the extracellular filaments
which form APC and ACA (Merz et al., 1983). If the same pro-
tein monomers are the subunits for NFT, APC and ACA, why
NFT IN PR'
!(b) AMYLOID PK
0D CONGOPHILIC ANGIOPATHY (ACA):
ILTRATION OF SMALL ARTERIOLES
Fig. 6. Schematic diagram of the relationships between NFT, APC and ACA. Paired helical filaments are formed first in the intracellular space of the neuron,
and coalesce to form neurofibrillary tangles (NFT) in the soma and axonal processes. At the same time or a little later, the same precursor proteins [either in
the dimeric (A8) or higher aggregated forms] are deposited in the extracellular space associated with the pre-synaptic axonal processes. These deposited
proteins form the amyloid fibrils which may either crystallize within the center of a plaque to form an APC, or the fibrils or their precursor subunits may
migrate and accumulate around and within the walls of small blood vessels (ACA).
C.L.Masters et al.
then are there discernible morphological differences? We suspect
that the intracellular NFT protofilaments are assembled from the
A4/A64 units with a different packing density than the extracellular
APC/ACA protofilaments. The tightest packing is expected for
the full-size ACA-A4 monomers for which no ragged N termini
were reported. Indeed, APC and ACA filaments are tightly pack-
ed, being straight and single stranded in contrast to the twisted
helical filaments or ribbons which comprise the NFT (Wisniewski
and Wen, 1985; Wischik et al.,
cumulating in the swollen axonal terminals surrounding the APC
are of intermediate size and are heterogeneous (Yoshimura,
1984). Some appear to consist of PHF which are continuous with
straight, single-stranded filaments. We suggest that this struc-
tural variation is a result of factors such as the N-terminal
heterogeneity which is most prominent for NFT-A4, thus allow-
ing for a greater degree of helicity, or the local environment (ionic
strength, pH) in which the filaments are assembled.
It is possible that the His-His sequence (residues 13- 14), but
not His-7, are responsible for the pH dependency of A4 aggrega-
deprotonation favors aggregation, factors which could easily be
influenced by the milieu in which the protofilaments are
assembled. The suggested3-pleated sheet structures of these
amyloid proteins would explain the thread-like aggregation of
the A4 monomers. These monomers, with -40 residues, are
unable by themselves to form a f-barrel structure, the closed-
end fl-structure of secreted proteins with fl-pleated sheets (Schulz
and Schirmer, 1984). Such a fl-barrel structure would require
- 100 residues to form a minimum of six f-strands. A4 may be
considered as half a fl-barrel with a high tendency for dimeriza-
tion to A8. The highly stable dimers (which may represent the
'native' or soluble form of the protein in the cytoplasm of
neurons) form tetramers. Packing of four tetramers (A64 hex-
adecamer) would then result in the formation of one segment (disk
or sphere) of a protofilament. It has previously been suggested
that the amyloid fiber in turn is composed of four or eight pro-
tofilaments (Wisniewski and Wen, 1985) or a six-stranded rib-
bon (Wischik et al., 1985).
The amyloidfilaments and proteins of AD are related to the
filaments and polypeptides associated with scrapie
There are several striking similarities between the A4 proteins
ofAD and the filaments and polypeptides associated with scrapie.
In additionto thepreviously
resemblances between the APC- and NFT-AD filaments with
SAF (Merz et al., 1981; Rubenstein et al., 1985), the chemical
inactivation profile of scrapie is similar to the solubility profile
for the AD amyloid proteins (Masters et al., 1985). To these,
we can now add the strong similarity in amino acid composition
between the AD-A4 monomer and the 7-kd deglycosylated
scrapie-associated protein (Table I and Multhaup et al., 1985).
In length, the AD-A4 monomer is only - 15 amino acids shorter
than the scrapie-associated protein. The scrapie-associated pro-
tein also has ragged N termini (Multhaup et al., 1985; Oesch
et al., 1985).
However there is no sequence homology between AD-A4 and
the scrapie polypeptides. The corresponding precursor proteins
are clearly distinct. A further difference is that the scrapie-
associated protein is glycosylated (to give an apparent mol. wt.
of 30 kd) but AD-A4 is not (Masters et al., 1985; Multhaup et
Overall, there are sufficient similarities between the AD and
scrapie amyloid filaments and proteins to suspect that they are
1985). The filaments ac-
derived by mechanisms which have in common the generation
of self-aggregating polypeptides of low mol. wt. These mechan-
isms include mistranslation or altered splicing or processing of
translation products. If the scrapie filaments and proteins are an
integral part or a direct effect of the infectious agent, it follows
that AD is also an infectious process similar to scrapie. Alter-
natively, it is still conceivable that the amyloid filaments and pro-
teins of both AD and scrapie are pathologic by-products of
independent disease processes. This would be consistent with the
recent observation that the scrapie-associated protein is encoded
by a host gene and is expressed in both normal and infected
animals (Oesch et al., 1985). To date, AD remains a non-trans-
missible disease as judged by long-term animal inoculation
Materials and methods
NFT were obtained from six cases of AD, using a modification of the method
used to purify APC (Masters et al., 1985). The starting material in each case
was fresh frozen cerebral cortex, from which the leptomeninges and largerblood
vessels had been removed. The cortex was homogenized and extracted in high
salt and detergent buffers, digested with pepsin, and centrifuged over a discon-
tinuous sucrose gradient (Masters et al., 1985). The 20-30% sucrosegradient
interface was diluted in 50 mM Tris, 1% SDS, 10 mM EDTA, pH 7.6, and
incubated at room temperature for 10 min with mixing. After centrifugation at
200 g for 10 min to pellet the APC, the supernatant was centrifuged at 10 000
g for 30 min. The pellet was taken up in the Tris/SDS/EDTA buffer, andlayered
over a sucrose step gradient (1.0 M, 1.2 M, 1.4 M, 2.0 Msucrose). Thisgra-
dient is centrifuged at 150 000 g for 2 h at 4°C. NFT are recovered at the
1.4- 2.0 M sucrose interface, diluted with distilled water, pelleted at 10 000g,
washed in 2% SDS, then resuspended in Tris-buffered saline and frozen for fur-
ther analysis. Contaminants, as judged by electron microscopy, consistchiefly
of lipofuscin, collagen, amyloid plaque cores and occasional bacteria. Control
brains (from individuals without recognizable NFT or APC by light microscopy)
were similarly processed, and yielded fractions containing lipofuscin only.
Polyacrylamide gel electrophoresis
The formic acid extracts of NFT and APC were lyophilized, dissolved insample
buffer containing 6 M urea but no thiols, and heated for 30 min at37°Cbefore
loading on the gels. Oxidation was achieved with performic acid. The slabgels
[15% (w/v) polyacrylamide, 1 mm thick] were made from stock solutions con-
taining 6 M urea and run at 28 mA constant current. The protein bands were
stained with Coomassie brilliant blue R.
The protein components of NFT and APC were dissolved in 90% formic acid,
lyophilized and re-dissolved in 50ytl ofh.p.l.c. buffer (0.1S% SDS, 150mM sodium
phosphate, pH 6.8). Chromatography was performedon twoanalytical 125I pro-
tein columns (Waters) (30 cm x 7.8 mm) connected in tandem, with aguard
column (3 cm x 2 mm) filled with 125I bulk packing phase.The flow-rate was
0.2 ml/min, and the protein peaks were detected byabsorbance at 214 nm. For
amino acid sequencing, protein peaks were lyophilized, thenprecipitatedwith
methanol (Wessel and Flugge, 1984) to remove excess detergent.
Amino acid analysis
NFT proteins (1-4jig)were hydrolyzed and then analyzedon an automated
amino acidanalyzer (Beckman 121 M).
Protein sequence analysis
Samples were dissolved in 3011of formic acid and dried on theglassfiber disks
of a gas-liquid solid phase protein sequencer (Applied Biosystems,Model470A).
The filters were pre-loaded with 1.5 mg ofpre-conditioned Polybrene (Aldrich).
Sequencing was performed as described (Beyreuther et al., 1983).
Immunizations and immunocytochemical characterizations
Rabbits (young adult New Zealand White or Semi-lops) were immunized with
various APC or NFT preparations, following the usual schedule ofprimaryin-
oculation in complete Freund's adjuvant thenboostingwithantigeninincomplete
Freund's adjuvant. Inoculations were made at multiple sitessubcutaneouslyover
the rabbit's back.
More than 18 rabbits are being used in these experiments. Theantigenic prepara-
tions included the following. (i) APC (-2 x 105 APC for each dose) which
had been (a) left intact; (b) sonicated and coupledto KLH; (c) degradedwith
alkali (1.0 M NaOH); (d) solubilized in 80% phenolandcoupledto KLH; (e)
Alzheimer's disease amyloid proteins
solubilized in formic acid and coupled to KLH. (ii) The h.p.l.c.-derived frac-
tions corresponding to a pool of the A16 and A8 peaks (coupled to KLH) from
either APC or NFT starting material. Approximately 5 yg of protein at each in-
oculation was used. (iii) Synthetic peptides corresponding to residues 1- 11 [SPA4
(I -I 11)] or residues 11 -23 [SPA4( 1 -23)] of the A4 monomer were made,
and coupled to KLH; these synthetic peptides were inoculated in doses of 250,g.
The rabbits were bled before the immunizations commenced and then after
the second and each subsequent boost, and their sera were initially screened by
immunocytochemistry, using paraffin-embedded sections of formalin fixed AD
or control brain. The sera were used at a dilution of 1:100, using the peroxidase-
anti-peroxidase (PAP) technique. Protease digestion (Protease Type VII from
Bacillus ainvloliquefaciens, Sigma P5255, 0.5 mg/ml, at 37°C for 10 min) of
the sections was found to enhance the reactivity of some of the antisera towards
Wisniewski,H.M. and Wen,G.Y. (1985) Acta. Neuropathol., 66, 173-176.
Yoshimura,N. (1984) Clin. Neuropathol., 3, 22-27.
Received on 7 August 1985
We thank Dr. Neville Hills and the Perth City Mortuary forensic pathologists
for providing material; Klaus Neifer, Regine Hanssen, Trudy Parker and Nicola
Weinman for skillful technical assistance; Dr. R. Terry for helpful criticism. This
research was supported by grants from the National Health and Medical Research
Council of Australia (C.L.M.), the Telethon and Royal Perth Hospital Research
Foundations (C.L.M.), the Deutsche Forschungsgemeinschaft (SFB 74 to K.B.),
the Bundesministerium fur Forschung und Technologie (K.B.) and the Fonds
der Chemischen Industrie (K.B.).
Bendheim,P.E., Barry,R.A., DeArmond,S.J., Stites,D.P. and Prusiner,S.B.
(1984) Nature, 310, 418-421.
Beyreuther,K., Bieseler,B., Bovens,J., Dildrop,R., Neifer,K., Stuber,K., Zaiss.S.,
Ehring,R. and Zabel,P. (1983) in Tschesche,H. (ed.), Modem Methods in Pro-
tein Chemistry, Walter de Gruyter, Berlin/NY, pp. 303-355.
DeArmond,S.J., McKinley,M.P., Barry,R.A., Braunfeld,M.B., McColloch,J.R.
and Prusiner,S.B. (1985) Cell, 41, 221-235.
Gajdusek,D.C. (1985) N. Engl. J. Med., 312, 714-719.
Garruto,R.M., Fukatsu,R., Yanagihara,R.Y., Gajdusek,D.C., Hook,G. and
Fiori,C.E. (1984) Proc. Natl. Acad. Sci. USA, 81, 1875-1879.
Geisler,N., Plessmann,U. and Weber,K. (1985) FEBS Lett., 182, 475478.
Glenner,G.G. and Wong,C.W. (1984) Biochem. Biophvs. Res. Commun., 120,
Gray,E.G. (1985) Neuropath. Appl. Neurobiol., in press.
Grundke-Iqbal,I., Wang,G.P., Iqbal,K., Tung,Y.-C. and Wisniewski,H.M. (1985)
Acta. Neuropathol., in press.
Masters,C.L., Gajdusek,D.C. and Gibbs,C.J.,Jr. (1981a) Brain, 104, 535-558.
Masters,C.L., Gajdusek,D.C. and Gibbs,C.J.,Jr. (1981b) Brain, 104, 559-587.
Masters,C.L., Simms,G., Weinman,N.A., Multhaup,G., McDonald,B.L. and
Beyreuther,K. (1985) Proc. Natl. Acad. Sci. USA, 82, 4245-4249.
Merz,P.A., Somerville,R.A., Wisniewski,H.M. and Iqbal,K. (1981) Acta
Neuropathol., 54, 63-74.
Merz,P.A., Wisniewski,H.M., Somerville,R.A., Bobin,S.A., Masters,C.L. and
Iqbal,K. (1983) Acta. Neuropathol., 60, 113-124.
Metuzals,J., Montpetit,V. and Clapin,D.F. (1981) Cell Tissue Res., 214, 455482.
Multhaup,G., Diringer,H., Hilmert,H., Prinz,H. and Beyreuther,K. (1985) EMBO
J., 4, 1495-1501.
Nikaido,T., Austin,J., Trueb,L. and Rinehart,R. (1972) Arch. Neurol., 27,
Oesch,B., Westaway,S., Walchli,M., McKinley,M.P., Kent,S.B.H., Aeber-
sold,R., Barry,R.A., Tempst,P., Teplow,D.B., Hood,L.E., Prusiner,S.B. and
Weissmann,C. (1985) Cell, 40, 735-746.
Perl,D.P. and Brody,A.R. (1980) Science (Wash.), 208, 297-299.
Perry,G., Rizzuto,N., Autilio-Gambetti,L. and Gambetti,P. (1985) Proc. Natl.
Acad. Sci. USA, 82, 3916-3920.
Prusiner,S.B., McKinley,M.P., Bowman,D.A., Bolton,D.C., Bendheim,P.E.,
Groth,D.F. and Glenner,G.G. (1983) Cell, 35, 349-358.
Rubenstein.R., Kascsak,R.J., Merz,P.A., Wisniewski,H.M., Carp,R.I. and
Iqbal.K. (1985) Brain Res., in press.
Schulz.G. and Schirmer,H. (1984) Principles ofProtein Structure, published by
Selkoe,D.J. and Abraham,C. (1985) Neurology, 35, (Suppl. 1), 217.
Sternberger,N.H., Sternberger,L.A. and Ulrich,J. (1985) Proc. Natl. Acad. Sci.
USA, 82, 42744276.
Weiss,A. (1981) Angew. Chein. Engl. Ed., 20, 850-860.
Wessel,D. and Flugge,U.I. (1984) Anal. Biochem., 138. 141-143.
Wischik,C.M., Crowther,R.A., Stewart,M. and Roth.M. (1985) J. Cell Biol.,