. 39
( 45 .)


tion of a glycosyl phosphatidylinositol anchor (shaded areas). A polymorphism at
codon 129 (M/V) is common in the Caucasian population. Pathogenic mutations at
codon 102 (P -+ L) and codon 145 (Y + stop) are in phase with M129, whereas
mutations at codons 105 (P + L), 117 (A -+ V), 198 (F + S), and 217 (Q -+ R) are in
phase with V129.

although in a few instances (e.g., some patients with the codon 117 mutation)
the cerebellum is not involved. Amyloid deposition usually occurs in the
neuropil. However, in the case of the 145 mutation it takes place primarily in
the wall of parenchymal and leptomeningeal vessels. This neuropathologic phe-
notype is distinct from that of GSS disease and has been designated as PrP
Priori Protein Amy/o&

cerebral amyloid angiopathy (PrP-CAA) (7) In addition to amylold deposits,
many patients with P + L substitution at codon 102 show severe sponglform
changes m the cerebral cortex, whereas patients with codon 145, 198, or 2 17
mutation display neurofibrlllary tangles composed of paired helical filaments
m neo- and archlcortex as well as several subcortlcal nuclei (6-20).
Amylold deposltlon IS consistently accompanied by hypertrophy and prohf-
eration of astroglial and mlcroglial cells, neurltlc abnormalities, and neuronal
loss leading to a vartable degree of atrophy of affected regions (6). The close
topographical relattonshlp between amyloid deposits and tissue changes sug-
geststhat PrP amyloid plays a role in the pathogenesls of nerve cell degeneration
and ghal cell reaction On this ground, we carried out studies aiming to define
the composltlon of amylold fibrlls m GSS families with different PRNP muta-
tions, to Identify the PrP sequencesthat are critical for amylold formation, and to
evaluate the biological effects of PrP peptldes on nerve and gllal cells in vitro
2. The Amyloid Protein in GSS Disease
The biochemical composition of PrP amylold was first determined m brain
tissue samples obtained from patients of the Indiana kindred of GSS (I I, 12),
carrying a F + S substltutlon at PrP residue 198 in couplmg phase with V129
(13,14). Amylold cores were isolated by a procedure combining buffer extrac-
tion, sieving, collagenase digestion, and sucrose density gradient (Fig. 3). Pro-
teins were extracted from amylold fibrlls with formic acid, purified by gel
filtration chromatography and reverse-phase high-performance liquid chroma-
tography (HPLC), analyzed by sodium dodecyl sulfate-polyacrylamlde gel elec-
trophoresls (SDS-PAGE) and rmmunoblot, and sequenced as described m detail
below. The amylold preparations contained two major peptldes of -11 and -7
kDa (Fig. 4) spanning residues 58-150 and 8l-l 50 of PrP, respectively (11,22).
Sequence analysis showed that these peptides had ragged N- and C-termmt.
The finding that the amylold protein was an N- and C-terminal truncated
fragment of PrP was verified by lmmunostammg brain sections with antisera
raised against synthetic peptldes homologous to residues 2340, 90-102,
127-147, and 220-23 1 of human PrP, and residues 15--40,90-l 02, and 220-232
of hamster PrP. The amyloid cores were strongly lmmunoreactlve with the
antisera to the mid-region of the molecule, whereas only the periphery of the
cores was mm-mnostained by antibodies to N- or C-terminal domains (6,15).
In GSS 198, the amylold protein does not include the region containing the
ammo acid substitution. To establish whether amylold peptldes originate from
mutant or both mutant and wild-type PrP, we analyzed patients heterozygous at
codon 129 (M/V) and used V129 as a marker of the mutant allele, since m this
family VI 29 is in phasewith mutant S198.Amino acid sequencmgand electrospray
massspectrometry of peptldes generatedby digestion of the amyloid protein with
268 Tagliavm et al.

Brain Tissue
homogenize (TBS, 1% %hB X-100)

* sieve (1 mm nylon mesh)
l centrifuge uw.Nl g, 20 min)
homogenize (0.6 XI, 0.6% TAton X-100)
l M

cenMivll0 (10,000 g, 20 min)
homogenize (1.5 M KCI, 0.6% T&on X-100)

sieve (105 um nylon mesh)
+ l

centrifugci ClO.600 g, 20 min)

digest with collagenase overnight
centrifuge (100,000 g, 45 min)


+ wash (50 mM Tns)

Sucrose Gradient

* cedrifuge (120,000 g, 2 hr)

Plaque Cores

Fig. 3. Preparative scheme for extractlon of amylold cores from bram ussue

endoprotemaseLys-C showed that the samplescontamed only pepttdeswith V 129,
suggesting that only mutant PrP was mvolved m amyloid formation (Fig. 5A) (I 2)
Subsequently, we characterized the amyloid protein in GSS kmdreds with
other PRNP mutattons (i.e., A + V at codon 117 and Q + R at codon 2 17) and
found that the smallest amylotd subunit was a -7 kDa N- and C-terminal trun-
cated fragment of PrP of stmllar size and sequence and was derived from the
mutant allele (Fig 5B) (12). In patients with GSS 117, the amylotd protem
contained the mutant V 117 (16)
3. Assembly and Conformation of PrP Peptides In Vitro
To determine which restduesare important for polymertzatton of PrP pepttdes
mto amylotd fibrtls and which condmons promote peptide assembly, we mvestt-
gated fibrtllogenests m vitro usmg synthetic peptides homologous to consecuttve
segments of the amylotd protein purified from GSS brains (I.e., the octapepttde
repeat region, restdues 89-l 06, 106-l 26, and 127-147) We found that pepttdes
wtthm the PrP sequence 106-147 readily assembled into fibrtls, whereas pep-
tides correspondmg to the N-termmal segment of the amylotd protem drd not
Prion Protein Amyloids 269





: 30

0.7 21.5

06. 6.c





f .:., , , , , , , , , ,
0 50 loo
VoEe (ml)

Fig. 4. Partial purification of GSS amyloid proteinsby gel filtration chromatography.
Elution profile of proteinsextracted by formic acid from amyloid coresand fractionatedon
a SephadexG- 100column. Protein peakswere collected asindicated (fractions 1-7) and
subjectedto immunoblot analysiswith the monoclonalantibody 3F4 to residues109-l 12
of humanPrP (inset). Gel filtration yielded two major peaks:the void volume (fraction 1)
and a low-mol-wt peak (fraction 6) that was presentasa broadband centeredat 7 kDa on
SDS-polyacrylamide minigels. In addition, minor intermediate peaks (fractions 2-5)
were presentin the chromatograms.Fractions4 and 5 containedan 11-kDa PrP fragment.

(I 7). In particular, the peptide homologous to residues 106-126 was extremely
fibrillogenic and formed dense meshworks of straight filaments ultrastructurally
similar to those observed in GSS patients (Fig. 6A). The fibrillary assemblies
were partially resistant to proteinase K digestion, exhibited tinctorial and optical
properties of in situ amyloid (i.e., birefringence under polarized light after Congo
red staining and yellow fluorescence after thioflavine S treatment), and showed
an X-ray diffraction pattern consistent with that of native amyloid fibrils, with
reflections corresponding to H-bonding between antiparallel P-sheets(2 7,18).
Circular dichroism (CD) spectroscopy revealed that peptide 106-l 26 is able
to adopt different conformations in relation to the microenvironment (Fig.
270 Taglia vini et al.

SP REPEATS kd 129 F ;98 s.s

w 81 D 144
G 82 E 146
Y 150

v 129

SP REPEATS M 129 0 117 cc

v 129 R 217
I --------

W 81 v 145
G 82 E 146

Amyloid k?ii
v 129

Fig. 5. Schematic representation of the PrP molecule and the 7-kDa amyloid pro-
tein in GSS patientswith mutationsat codon 198(A) or codon 217 (B) and heterozy-
gous M/V at codon 129. The amyloid peptides have ragged N and C termini, and
contain only V 129, suggesting that only mutant PrP is involved in amyloid formation.

6B,C) (19). It showed primarily a P-sheet secondary structure in phosphate
buffer, pH 5.0, a combination of P-sheet and random coil in phosphate buffer,
pH 7.0, a random coil conformation in deionized water, and an a-helical struc-
ture in the presence of micelles formed by a 5% SDS solution. The addition of
a-helix stabilizing solvents (e.g., trifluoroethanol or hexafluoropropanol) to a
solution of peptides in deionized water induced a conformational shift from
random coil to a-helix, but did not modify the P-sheet structure of the peptide
previously suspended in phosphate buffer, pH 5.0 (19). These data suggest that
the PrP region including residues 106-126 may feature in the conformational
transition from normal to abnormal PrP.
4. Biological Effects of PrP Peptides In Vitro
To test the hypothesis that the accumulation of PrP amyloid protein may be
the direct cause of nerve cell degeneration and glial cell reaction in GSS dis-
ease and PrP-CAA, we investigated the effects of the exposure of primary
Prion Protein Amyloids 271

190 240 190 240

Fig. 6. Assembly and conformation of a synthetic peptide homologous to residues
106126 of human PrP. (A) Electron micrograph of fibrils generated in vitro by the
peptide. The fibrils are straight, unbranched, and have a diameter of -8 nm. Bar: 100
nm. (B,C) Circular dichroism spectra of PrP 106-126. The peptide adopts predomi-
nantly a P-sheet conformation when suspended in 200 mA4 phosphate buffer, pH 5.0
(B), and an a-helical structure when dissolved in 50% trifluoroethanol (C).

and astrocytes to the synthetic peptides used for fibril-
cultures of neurons
logenesis studies.
The prolonged exposure of hippocampal neurons to micromolar concentra-
tions of PrP 106-126 resulted in marked neuronal loss (Fig. 7A,B) (20). Con-
versely, the other PrP peptides and a scrambled sequence of PrP 106-I 26 were
not effective. The neurotoxic effect of PrP 106-126 on cultured neurons was
dose-dependent; the toxic response was first detected at a concentration of
20 pM, was statistically significant at 40 @I, and resulted in virtually com-
plete neuronal loss at 60 or 80 pJ4. Fluorescence microscopy following treat-
ment with DNA-binding fluorochromes (e.g., Hoechst 33258; Sigma, St. Louis,


. 39
( 45 .)