Biochemical and Biophysical Research Communications 355 (2007) 398403

www.elsevier.com/locate/ybbrc

Observation of multiple intermediates in a-synuclein bril formation

by singular value decomposition analysis
Tomoaki Kamiyoshihara a, Masaki Kojima a, Kenji Ueda b, Mitsuru Tashiro c,
Sakurako Shimotakahara d,*
a

School of Life Science, Tokyo University of Pharmacy and Life Science, Horinouchi, Hachioji, Tokyo 192-0392, Japan
b
Division of Psychobiology, Tokyo Institute of Psychiatry, 2-1-8 Kamikitazawa, Setagaya-ku, 156-8585, Japan
c
Department of Chemistry, College of Science and Technology, Meisei University, Hino, Tokyo 191-8506, Japan
d
School of Pharmacy, Tokyo University of Pharmacy and Life Science, Horinouchi, Hachioji, Tokyo 192-0392, Japan
Received 17 January 2007
Available online 6 February 2007

Abstract
One of the most well known characteristics for Parkinsons disease (PD) is a polymerization of wild-type or mutant a-synuclein into
aggregates and brils, commonly observed as Lewy bodies and Lewy neuritis in PD patients. Although numerous studies on a-synuclein
brillation have been reported, the molecular mechanisms of aggregation and brillation are not well understood yet. In the present
study, structural properties and propensities to form brils of wild-type, A30P, E46K, and A53T a-synucleins were investigated using
uorescence and circular dichroism (CD) methods. The results from these studies were analyzed using singular value decomposition
(SVD) method which estimates a number of conformationaly independent species for a given process. The time-dependent CD spectra
of the wild-type a-synuclein indicated a multi-step process in the bril formation, and SVD analysis using the time-dependent CD spectra
revealed that ve or nine intermediates were formed at the early stage of brillation.
2007 Elsevier Inc. All rights reserved.
Keywords: a-Synuclein; Fibrillation; Singular value decomposition; Intermediates

Parkinsons disease (PD) is one of the most common

movement disorders for elderly. It arises from the loss of
dopaminergic neurons in the substantia nigra and is
accompanied by the presence of intracellular inclusions
known as Lewy bodies and Lewy neuritis [1,2]. Structurally, Lewy bodies and Lewy neurites are composed of lamentous and granular materials. The causative factors of
various neurodegenerative disorders such as PD, Alzheimer disease and some forms of spongiform encephalopathy
are associated with the formation of abnormally aggregated proteins [3]. The aggregates, known as amyloid brils,
are characterized by their unique brous appearances, as

Corresponding author. Fax: +81 426 76 4542.

E-mail address: tashiro@ps.toyaku.ac.jp (S. Shimotakahara).

0006-291X/$ - see front matter 2007 Elsevier Inc. All rights reserved.
doi:10.1016/j.bbrc.2007.01.162

observed by electron microscope and b-strand rich structure [46].

a-Synuclein is a 140-amino acid protein consisting of an
acidic carboxyl-terminal region and six imperfect repeats
(consensus KTKEGV) distributed throughout most of
the N-terminal half of the polypeptide [7]. It is a highly soluble, heat-stable and natively unfolded protein [8,9] in vitro.
Protein expression is predominantly localized in the neurons of the central nerve system, particularly at presynaptic
terminals in close proximity to synaptic vesicles [7,1012].
Three dierent missense mutations in the a-synuclein gene,
A53T [13], A30P [14], and E46K [15], have been identied
in three instances of autosomal-dominantly inherited,
early-onset PD. Although a physiological role of a-synuclein is not well understood in neuronal cells, most of a-synuclein observed in vivo is accumulated as bril form in
Lewy bodies and Lewy neurites for PD. The mechanism

T. Kamiyoshihara et al. / Biochemical and Biophysical Research Communications 355 (2007) 398403

of amyloid bril formation of a-synuclein in vitro has been

extensively studied by Fink, Uversky, and other researchers
[1621]. These studies indicated that the hydrophobic
repeat region between 65 and 95 residues and the negative
charges at the C-terminal region may play a key role in
forming bril structures. A molecular species that is relatively compact but still unfolded was also found to be
involved in the early stage of bril formation [16].
Currently, a-synuclein is considered to be a major building block of the laments that comprises the proteinacious
pathological inclusions characteristic of many disorders,
including Lewy bodies and Lewy neurites in PD and
dementia with Lewy bodies [22,23]. The question is how
the essentially unstructured protein such as a-synuclein
can be transformed into the highly organized brils, forming the crossed b-sheet conformation. The past studies
indicated that the polymerization of a-synuclein is nucleation-dependent [24,25], and that a partially folded intermediate is generated in the process of a-synuclein bril
formation [16]. It has also been reported that a low pH
or an increase in temperature can also trigger the transformation of a-synuclein into a partially folded conformation
in vitro [16]. The strong correlation between the formation
of partially folded protein and the eciency of its bril formation suggests the presence of intermediates as precursors
in bril formation. Since the existence of transient precursors for aggregation has been reported in some cases, the
observation of transient intermediates during bril formation should be quite signicant for understanding the
molecular mechanism of PD as well as other transmissible
amyloidogenic diseases.
In the present study, structural properties and propensities to aggregate or form brils of the wild-type, A30P,
E46K, and A53T a-synucleins were investigated using uorescence and circular dichroism (CD) methods. These
mutants were of particular interest as their dierences in
brillation behaviors could give insights into the eect of
point mutations, which manifest as autosomal-dominantly
inherited, early-onset PD. Furthermore, in order to estimate the number of independent conformations formed
during the brillation process, rigorous analyses using singular value decomposition (SVD) method were performed
for the time-dependent CD spectra of the wild-type
a-synuclein.

399

heated to 80 C for 20 min and then centrifuged (13,000 rpm, 45 min,

4 C). During this step, the soluble a-synuclein remained in a supernatant.
The supernatant was applied onto a Superdex-75 gel ltration column
(Amersham Biosciences) with 20 mM potassium phosphate buer, pH 7.5,
containing 100 mM NaCl. The fractions containing a-synuclein was then
applied onto an anion exchange Resource-Q column (Amersham Biosciences) with 10 mM TrisHCl buer, pH 7.5. Samples were eluted with a
linear gradient of 00.4 M NaCl. The protein sample was concentrated for
bril formation using Amicon Ultra Centrifugal Filter device (Millipore)
with 5000 MWCO.
Amyloid bril formation and uorescence measurements. Amyloid bril
formation was carried out by incubating a-synuclein (3 mg/ml) in 50 mM
sodium phosphate buer, pH 7.5, containing 100 mM NaCl and 0.05%
azide at 25 and 50 C, with continuous stirring using a magnetic stirrer. A
progress of bril formation was monitored using thioavin-T (ThT,
Wako) binding assay [28] and its uorescence was measured with a Shimadzu RF-5300PC uorescence spectrophotometer. Aliquots of 3 lL
from the above bril solution at various incubation times were mixed with
a solution of 10 lM ThT in 20 mM TrisHCl buer, pH 8.0, to make up a
nal protein concentration of 3 lg/ml. After one minute of incubation,
ThT uorescence intensities were measured from 450 to 550 nm with
excitation at 450 nm. Every point was measured for at least four times and
the average values from four scans were used for further analysis.
Circular dichroism measurements. CD spectra were obtained with a
Jasco J-720 spectrophotometer. Spectra were recorded using a 1-mm
cuvette, scanning from 190 to 250 nm with a step size of 0.1 nm and the

Materials and methods

Expression and purication of a-synuclein. Human a-synuclein gene was
amplied from the previously described pET-a-synuclein [26] vector by
PCR method, subcloned into the NcoI and NdeI restriction sites of the
expression vector pET16b (Novagen), and expressed in Escherichia coli
BL21(DE3) (Novagen). Three mutant A30P, E46K, and A53T genes were
generated using a PCR method introducing a specic mutation at a given
position along the length of PCR-generated fragments [27]. All mutations
and subclonings were conrmed by DNA sequencing using Big Dye
sequencing kit (Applied Biosystems). After overnight expression, cells
were resuspended in the purication buer (20 mM TrisHCl, pH 8.0,
containing 20 mM EDTA, 200 mM NaCl), lysed by sonication (Branson
250D) and centrifuged (10,000 rpm, 60 min, 4 C). The supernatant was

Fig. 1. Amyloid bril formation of the wild-type (open diamond), A30P

(square), E46K (lled diamond), and A53T (triangle) mutant a-synucleins
monitored by ThT uorescence at 482 nm. Curve ttings were performed
using the sigmoid function. Fibrillations were carried out at (A) 25 and (B)
50 C.

400

T. Kamiyoshihara et al. / Biochemical and Biophysical Research Communications 355 (2007) 398403

Table 1
Propensities of amyloid bril formation of wild-type and A30P, E46K and
A53T mutant a-synucleins monitored by ThT uorescence at pH 7.5a
t1/2 (h)

Lag time (h)

Fluorescence maxb

Temperature (C):

25

50

25

50

25

50

WT
A30P
E46K
A53T

50
40
65
35

12
8
37
10

40
20
60
25

8.6
4.0
25
6.7

90
22
400
575

65
75
100
255

a
b

Data from analysis of Fig. 1.

Fluorescence was monitored at 482 nm.

scanning speed of 50 nm/min. For all spectra, an average of eight scans

was obtained. CD samples were prepared by diluting 10 lL of the above
protein sample (3.0 mg/ml) at various incubation times into 400 lL CD
buer (10 mM sodium phosphate, pH 7.5), resulting in a nal protein
concentration of 0.075 mg/ml.
Singular value decomposition. SVD analysis was carried out in a similar
manner as previously reported [29,30]. All the CD data were arranged in a
matrix form so that the value of (i,j) elements of the data matrix D corresponded to the wave length of the ith CD spectrum measured at the jth
incubation time. The data matrix was expressed as a linear combination of
orthogonal bases and their weighting factors:
D UWV T
Each column of matrix corresponds to the orthogonal base, and the product WVT determines the time-dependent weight for the individual base. W
is the diagonal matrix of the singular values, which are usually arranged by
magnitude in descending order. While the rank of the matrix D provides
the number of independent states in the ideal case without noise, the number of non-noise components must usually be estimated for the actual
data. This number was determined from the SVD results based on the
magnitudes of singular values, the shapes of the basis functions and
the autocorrelation of each base [29]. The autocorrelation C(Ui) reects
the signal-to-noise ratio of each base, and is dened as
CU i

M 1
X

U j;i U j1;i

j1

where Ui and Uj,i mean the ith column and (j, i) element of matrix U,
respectively.

Results and discussion

Fig. 2. Far-UV CD spectra of the wild-type a-synuclein at various
incubation times during the bril formation at incubation times of 0 h
(lled circle), 6 h (lled square), 12 h (lled diamond), 18 h (cross), 24 h
(lled triangle), 30 h (open square), 48 h (open triangle), and 120 h (open
diamond). The bril formation was carried out at 25 C.

Amyloid bril formation of the wild-type and mutant

a-synucleins
In order to closely observe the process of amyloid bril
formation, an accumulation of amyloid was monitored by

Fig. 3. The plot of time courses for the molar ellipticities at wavelength 195 nm (A,B) and 218 nm (C,D) for the wild-type (open diamond), A30P (square),
E46K (lled diamond), and A53T (triangle) a-synucleins during the bril formation. Curve ttings were performed using the sigmoid function. Incubation
was carried out at 25 C (A,C) and 50 C (B,D).

T. Kamiyoshihara et al. / Biochemical and Biophysical Research Communications 355 (2007) 398403

uorescence spectroscopy using ThT binding assay. This

method is commonly used to observe an amyloid bril formation of proteins, and in general, an increase in ThT uorescence intensity is considered to be a good indication of
lament accumulations [31]. The time-dependent changes
in ThT uorescence during incubation of the wild-type,
A30P, E46K, and A53T a-synucleins at 25 and 50 C are
shown in Fig. 1, and propensities for lament assembly
of these a-synucleins are summarized in Table 1. An
increase in the incubation temperature shortened half times
to reach saturation of bril formation and also lag periods
for initiation of amyloid bril formations (Table 1). These
observations were in good agreement with the previous
study [32], indicating an increase in the incubation temperature accelerated amyloid bril formation of the wild-type
and all mutant a-synucleins. For the eects of mutations,
A53T showed the highest values of ThT uorescence at
both incubation temperatures. The amounts of total bril
assemblies for A53T were about 6.4- and 3.5-fold larger
than those of the wild-type a-synuclein at 25 and 50 C,
respectively (Fig. 1). The amount of total bril assemblies
for A30P was also larger than that of the wild-type at
50 C, whereas only a quarter of assemblies for A30P was
accumulated compared with that for the wild-type at
25 C. E46K showed the longest lag periods at both incubation temperatures, which was contrary to the other two
mutants. Notably, the lag period of E46K at 50 C was
2.9-fold longer than that of the wild-type a-synuclein. This
mutant also showed 3.1-fold longer half time than that of
the wild-type a-synuclein at 50 C (Table 1). Although
these results from half times and lag periods suggested
the less propensity in brillation for E46K, the total accumulation of amyloid brils for E46K was larger than those
for the wild-type and A30P a-synucleins. In fact, E46K
showed 4.4-fold larger amount of assembly than the wildtype a-synuclein at 25 C.

401

mutants at 25 C (Fig. 3A and C). The time-dependent

changes in ellipticities for A30P at 195 and 218 nm were
nearly the same as those for the wild-type a-synuclein.
There was no isosbetic point observed in the time-dependent CD spectra (Fig. 2), indicating that the amyloid bril
formation for the wild-type a-synuclein was not a two-step
process, but a multi-step process with intermediate states.
Interestingly, an intersection point of the spectra at the
incubation time 0 and 6 h was clearly dierent from those
at the incubation time longer than 12 h. This result suggests
that the intermediate species could be accumulated at the
early stages of bril formation. To estimate the number
of conformationaly independent species in the brillation

CD spectra and SVD analysis

Fig. 2 shows the time course of CD spectra for the wildtype a-synuclein bril formation. During the process of
bril formation, the spectra pattern changed from random-coil form, as characterized by the negative ellipticity
at wavelength below 200 nm, to b-sheet structure, as characterized by the positive ellipticity at wavelength 195 nm
and the negative ellipticity at 218 nm. Since several studies
have suggested the presence of b-sheet structure in amyloid
brils, this change in CD spectra suggested that the amyloid formation was initiated at around 48 h of incubation
time at 25 C. The same transition was also observed for
A30P, E46K, and A53T mutants. To investigate the eect
of temperature, the time courses of molar ellipticities at
195 and 218 nm for the wild-type and three mutants were
analyzed at two dierent temperatures, 25 and 50 C
(Fig. 3). The amyloid bril formation of E46K was slower
than that of the wild-type at 50 C (Fig. 3B), and A53T
showed the fastest bril formation compared with the other

Fig. 4. SVD analyses for the CD data of the wild-type a-synuclein: (A)
Logarithm plot of singular values against ordinal numbers, and (B) the
SVD basis function. The rst ve bases are shown; 1st base (lled square),
2nd base (open triangle), 3rd base (cross), 4th base (open square), and 5th
base (lled triangle). The autocorrelation values of the bases are 1.000,
0.998, 0.997, 0.962, and 0.995 in the aforesaid order. (C) Same as (A),
except that the analyses were carried out using only the curves at the
incubation time 12 h and longer.

402

T. Kamiyoshihara et al. / Biochemical and Biophysical Research Communications 355 (2007) 398403

process, SVD analysis was carried out for the time-dependent CD spectra. As shown in Fig. 4A, there were discrete
intervals between 2 and 3, 5 and 6, and 9 and 10 in the magnitude of singular values, which indicated that the possible
number of independent components was two, ve or nine.
Since an isosbetic point could not be clearly identied in
the original CD spectra, it can be concluded that the number of independent species was ve or nine. The analysis by
small angle X-ray scattering measurements also supported
the presence of the multiple-state transition, which was also
observed for the three mutants (unpublished results). As
described above, the CD spectra at the incubation time
of 12 h and longer appeared to share an isosbetic point.
The SVD analyses using only these CD spectra revealed
the transition to be a two-state process (Fig. 4C). This
result suggests that the intermediate species that were
formed during brillation mainly accumulated at the early
stages of the amyloid formation. The multiple-state transition could involve partially folded intermediates and oligomeric intermediates leading to amyloid-like brils.
Acknowledgments
This study was supported by a Grant-in-Aid for Scientic Research (No. 17550085 for S.S. and 18550082 for
M.T.) from the Ministry of Education, Culture, Sports,
Science, and Technology.
References
[1] J.E. Galvin, V.M. Lee, M.L. Schmidt, P.H. Tu, T. Iwatsubo, J.Q.
Trojanowski, Pathobiology of the Lewy body, Adv. Neurol. 80 (1999)
313324.
[2] Q.L. Ma, P. Chan, M. Yoshii, K. Ueda, Alpha-synuclein aggregation
and neurodegenerative diseases, J. Alzheimers Dis. 5 (2003) 139148.
[3] B.S. Shastry, Neurodegenerative disorders of protein aggregation,
Neurochem. Int. 43 (2003) 17.
[4] S.Y. Tan, M.B. Pepys, Amyloidosis, Histopathology 25 (1994) 403
414.
[5] R.W. Carrell, D.A. Lomas, Conformational disease, Lancet 350
(1997) 134138.
[6] M. Sunde, C. Blake, The structure of amyloid brils by electron
microscopy and X-ray diraction, Adv. Protein Chem. 50 (1997) 123
159.
[7] K. Ueda, H. Fukushima, E. Masliah, Y. Xia, A. Iwai, M.
Yoshimoto, D.A. Otero, J. Kondo, Y. Ihara, T. Saitoh, Molecular
cloning of cDNA encoding an unrecognized component of
amyloid in Alzheimer disease, Proc. Natl. Acad. Sci. USA 90
(1993) 1128211286.
[8] W.S. Davidson, A. Jonas, D.F. Clayton, J.M. George, Stabilization
of alpha-synuclein secondary structure upon binding to synthetic
membranes, J. Biol. Chem. 273 (1998) 94439449.
[9] P.H. Weinreb, W. Zhen, A.W. Poon, K.A. Conway, P.T. Lansbury,
NACP, a protein implicated in Alzheimers disease and learning, is
natively unfolded, Biochemistry 35 (1996) 1370913715.
[10] J.M. George, H. Jin, W.S. Woods, D.F. Clayton, Characterization of
a novel protein regulated during the critical period for song learning
in the zebra nch, Neuron 15 (1995) 361372.
[11] G.S. Withers, J.M. George, G.A. Banker, D.F. Clayton, Delayed
localization of syneln (synuclein, NACP) to presynaptic terminals in
cultured rat hippocampal neurons, Brain Res. Dev. Brain Res. 99
(1997) 8794.

[12] A. Iwai, E. Masliah, M. Yoshimoto, N. Ge, L. Flanagan, H.A. de

Silva, A. Kittel, T. Saitoh, The precursor protein of non-A beta
component of Alzheimers disease amyloid is a presynaptic protein of
the central nervous system, Neuron 14 (1995) 467475.
[13] M.H. Polymeropoulos, C. Lavedan, E. Leroy, S.E. Ide, A. Dehejia,
A. Dutra, B. Pike, H. Root, J. Rubenstein, R. Boyer, E.S. Stenroos, S.
Chandrasekharappa, A. Athanassiadou, T. Papapetropoulos, W.G.
Johnson, A.M. Lazzarini, R.C. Duvoisin, G. Di Iorio, L.I. Golbe,
R.L. Nussbaum, Mutation in the alpha-synuclein gene identied in
families with Parkinsons disease, Science 276 (1997) 20452047.
[14] R. Kruger, W. Kuhn, T. Muller, D. Woitalla, M. Graeber, S. Kosel,
H. Przuntek, J.T. Epplen, L. Schols, O. Riess, Ala30Pro mutation in
the gene encoding alpha-synuclein in Parkinsons disease, Nat. Genet.
18 (1998) 106108.
[15] J.J. Zarranz, J. Alegre, J.C. Gomez-Esteban, E. Lezcano, R. Ros, I.
Ampuero, L. Vidal, J. Hoenicka, O. Rodriguez, B. Atares, V. Llorens,
E. Gomez Tortosa, T. del Ser, D.G. Munoz, J.G. de Yebenes, The
new mutation, E46K, of alpha-synuclein causes Parkinson and Lewy
body dementia, Ann. Neurol. 55 (2004) 164173.
[16] V.N. Uversky, J. Li, A.L. Fink, Evidence for a partially folded
intermediate in alpha-synuclein bril formation, J. Biol. Chem. 276
(2001) 1073710744.
[17] V.N. Uversky, A.L. Fink, Conformational constraints for amyloid
brillation: the importance of being unfolded, Biochim. Biophys.
Acta 1698 (2004) 131153.
[18] V.N. Uversky, A protein-chameleon: conformational plasticity of
alpha-synuclein, a disordered protein involved in neurodegenerative
disorders, J. Biomol. Struct. Dyn. 21 (2003) 211234.
[19] B.I. Giasson, I.V. Murray, J.Q. Trojanowski, V.M. Lee, A hydrophobic
stretch of 12 amino acid residues in the middle of alpha-synuclein is
essential for lament assembly, J. Biol. Chem. 276 (2001) 23802386.
[20] H. Miake, H. Mizusawa, T. Iwatsubo, M. Hasegawa, Biochemical
characterization of the core structure of alpha-synuclein laments, J.
Biol. Chem. 277 (2002) 1921319219.
[21] W. Hoyer, D. Cherny, V. Subramaniam, T.M. Jovin, Impact of the
acidic C-terminal region comprising amino acids 109140 on alphasynuclein aggregation in vitro, Biochemistry 43 (2004) 1623316242.
[22] M.G. Spillantini, M.L. Schmidt, V.M.-Y. Lee, J.Q. Trojanowski, R.
Jakes, M. Goedert, Alpha-synuclein in Lewy bodies, Nature 388
(1997) 839840.
[23] K. Arima, K. Ueda, N. Sunohara, S. Hirai, Y. Izumiyama, H.
Tonozuka-Uehara, M. Kawai, Immunoelectron-microscopic demonstration of NACP/alpha-synuclein-epitopes on the lamentous component of Lewy bodies in Parkinsons disease and in dementia with
Lewy bodies, Brain Res. 808 (1998) 93100.
[24] L. Narhi, S.J. Wood, S. Steavenson, Y. Jiang, G.M. Wu, D. Ana,
S.A. Kaufman, F. Martin, K. Sitney, P. Denis, J.C. Louis, J. Wypych,
A.L. Biere, M. Citron, Both familial Parkinsons disease mutations
accelerate alpha-synuclein aggregation, J. Biol. Chem. 274 (1999)
98439846.
[25] S.J. Wood, J. Wypych, S. Steavenson, J.C. Louis, M. Citron, A.L.
Biere, Alpha-synuclein brillogenesis is nucleation-dependent. Implications for the pathogenesis of Parkinsons disease, J. Biol. Chem. 274
(1999) 1950919512.
[26] M.A. Alim, M.S. Hossain, K. Arima, K. Takeda, Y. Izumiyama, M.
Nakamura, H. Kaji, T. Shinoda, S. Hisanaga, K. Ueda, Tubulin seeds
alpha-synuclein bril formation, J. Biol. Chem. 277 (2002) 21122117.
[27] R. Higuchi, B. Krummel, R.K. Saiki, A general method of in vitro
preparation and specic mutagenesis of DNA fragments: study of
protein and DNA interactions, Nucleic Acids. Res. 16 (1988) 7351
7367.
[28] H. Naiki, K. Higuchi, M. Hosokawa, T. Takeda, Fluorometric
determination of amyloid brils in vitro using the uorescent dye,
thioavin T1, Anal. Biochem. 177 (1989) 244249.
[29] M. Kojima, M. Tanokura, M. Maeda, K. Kimura, Y. Amemiya, H.
Kihara, K. Takahashi, pH-dependent unfolding of aspergillopepsin II
studied by small-angle X-ray scattering, Biochemistry 39 (2000) 1364
1372.

T. Kamiyoshihara et al. / Biochemical and Biophysical Research Communications 355 (2007) 398403
[30] H. Matsuura, S. Shimotakahara, C. Sakuma, M. Tashiro, H. Shindo,
K. Mochizuki, A. Yamaguchi, M. Kojima, K. Takahashi, Thermal
unfolding of ribonuclease T1 studied by multi-dimensional NMR
spectroscopy, Biol. Chem. 385 (2004) 11571164.
[31] W. Choi, S. Zibaee, R. Jakes, L.C. Serpell, B. Davletov, R.A.
Crowther, M. Goedert, Mutation E46K increases phospholipid

403

binding and assembly into laments of human alpha-synuclein,

FEBS Lett. 576 (2004) 363368.
[32] V.N. Uversky, H. Lee, J. Li, A.L. Fink, S. Lee, Stabilization of
partially folded conformation during alpha-synuclein oligomerization
in both puried and cytosolic preparations, J. Biol. Chem. 276 (2001)
4349543498.