Amyloid fibril formation by human stefin B: influence of pH and TFE
on fibril growth and morphology
EVA ZˇEROVNIK1, MIHA SˇKARABOT2, KATJA SˇKERGET1, SILVA GIANNINI3*,
VERONIKA STOKA1, SASˇA JENKO-KOKALJ1, & ROSEMARY A. STANIFORTH3
1Department of Biochemistry, Molecular and Structural Biology,2Department of Condensed Matter Physics, Joz ˇef Stefan
Institute, 1000 Ljubljana, Slovenia, and3Department of Molecular Biology and Biotechnology, Krebs Institute, University of
Sheffield, S10 2TN Sheffield, UK
Keywords: Amyloid fibrils, cystatins, flow cytometry, lag phase oligomers, morphology by AFM, TFE effect on the kinetics
HFIP¼1,1,1,3,3,3-hexafluoroisopropanol; CM-papain¼carboxy-methylated papain; PBS¼phosphate-buffered saline;
FPLC¼fast protein liquid chromatography
FSC¼forward scatter; SSC¼side scatter; C.R.¼Congo red; AFM¼atomic force microscopy;
electronmicroscopy; GA¼glutaraldehyde; TFE¼2,2,2-trifluoroethanol; EtOH¼ethanol;
As shown before, human stefin B (cystatin B) populates two partly unfolded species, a native-like state at pH 4.8 and a
structured molten globule state at pH 3.3 (high ionic strength), from each of which amyloid fibrils grow. Here, we show that
the fibrils obtained at pH 3.3 differ from those at pH 4.8 and that those obtained at pH 3.3 (protofibrils) do not transform
readily to mature fibrils. In addition we show that amorphous aggregates are also a source of fibrils. The kinetics of amyloid
fibril formation at different trifluoroethanol (TFE) concentrations were measured. TFE accelerates fibril growth at
predenaturational concentrations of the alcohol. At concentrations higher than 10%, the fibrillar yield decreases
proportionately as the population of an all a-helical, denatured form of the protein increases. At an optimum TFE
concentration, the lag and the growth phases are observed, similarly to some other amyloidogenic proteins. Morphology of
the protein species at the beginning and the end of the reactions was observed using atomic force microscopy and
transmission electron microscopy. Final fibril morphologies differ depending on solvent conditions.
Human stefins belong to family I25A of cystatins
(http://merops.sanger.ac.uk/), the cysteine proteinase
inhibitors [1,2]. The two intracellular cystatins,
stefins A and B, play an important role in cell
physiology. They are involved in regulating cell death
and survival [3–5], which are key processes in
neurodegeneration and cancer, respectively. More
specific pathologies are epilepsy syndrome – the
progressive myoclonus epilepsy of type 1 (EPM1) in
the case of stefin B (cystatin B) [6–8] and hereditary
cystatin C amyloid angiopathy (HCCAA) in the case
of cystatin C [9,10]. Wild-type cystatin C has been
found as a component of amyloid plaques in
Alzheimer’s disease . Similarly, stefin B, together
with stefin A and some cathepsins, was identified in
the core of amyloid plaques of various origin .
Stefin B served as a very suitable model for studies
of amyloid fibril formation [13–15], membrane
interactions  and amyloid-induced toxicity (Cˇeru
et al. 2007, submitted for publication). We have
shown that the in vitro amyloid fibrils formed by this
protein have all the characteristics of other amyloid
fibrils: they are straight, over 1 mm long, bind Congo
red and thioflavin T , they show the character-
istic cross-b X-ray diffraction pattern  and have
the expected dimensions of 7–14 nm in width .
A more stable yet homologous protein, human stefin
A, can also form amyloid fibrils in vitro  but only
under very stringent conditions. Namely, the fibrilla-
tion of stefin A can be started after heating the
protein to predenaturational temperatures of around
formation . That heating is a prerequisite
for fibrillation was earlier reported for another
Correspondence: Dr Eva Zˇerovnik, Department of Biochemistry, Molecular and Structural Biology, Jozef Stefan Institute, Jamova 39, 1000 Ljubljana, Slovenia.
Tel: þ386 1 477 3753. Fax: þ386 1 257 35 94. E-mail: email@example.com
*Department of Chemistry, University of Cambridge, Lensfield Road, Cambridge CB2 1EW, UK
Amyloid, September 2007; 14(3): 237–247
ISSN 1350-6129 print/ISSN 1744-2818 online ? 2007 Informa UK Ltd.
structurally homologous protein, monellin .
Chimeras between stefin A and B were prepared,
which retained the propensity of stefin B to form
fibrils and others which appeared resistant, similarly
to stefin A . In accordance with an important
role for domain swapping prior to fibril formation,
the in vitro-induced fibrillation of wild-type cystatin
C was inhibited when domain swapping was pre-
For stefin B, not only morphologies, but general
traits of the mechanism were found similar to those
reported for other amyloidogenic proteins or pep-
tides, folded or unfolded, including Ab peptide 
and a-synuclein . It was reported that pH
dependence of fibrillation is shared among Ab, prion
and stefin B [23,24]. In this work we present some
additional data gathered thus far on amyloid fibril
formation by this protein.
The effect of trifluoroethanol (TFE) concentra-
tion on the rate of amyloid fibril formation was
measured. A typical bell-shaped dependence was
decrease of the rate with increasing TFE (denatur-
ant) concentration. The morphology of aggregates
and amyloid fibrils was monitored by the comple-
mentary techniques of atomic force microscopy
(AFM) and transmission
(TEM). Our results show that species in the lag
phase as well as the final fibril morphology differ
when the reaction is triggered from different initial
states induced by pH and salt and that they cannot
easily transform into each other.
was expressed in Escherichia coli as described
. Purification steps consisted of CM-papain
Sepharose-affinity chromatography and gel filtration
on Sephacryl-200 (Amersham Pharmacia, Uppsala,
Sweden). Protein concentration was evaluated using
an extinction coefficient of 5300 M71cm71
280 nm. Typical yields were450 mg/l rich culture.
All chemicals were of analytical grade and obtained
from BDH (VWR International, West Chester,
PA, USA) except thioflavin T (ThT) (Aldrich,
(TFE), 499% pure (Fluka, Buchs, Switzerland).
Double distilled water was used and solvents were
filtered through 0.22 mm filters.
human stefinB (C3S,E31Y)
Stefin B was dissolved to a standard concentration of
34 mM. Buffers were: 0.015 M glycine/HCl, 0.26 M
sodium sulfate, pH 3.3; 0.015 M sodium acetate,
0.15 M NaCl, pH 4.80; 0.01 M sodium phosphate,
0.12 M NaCl, pH 6.3. TFE was added to various
concentrations (vol/vol%) as indicated. To exclude
the effect of changed salt, which remained in solution
when diluting the protein from pH 3.3 (0.26 M
Na2SO4) to pH 4.8, a control experiment was
performed at pH 4.8, 10% TFE and final 0.13 M
Thioflavin T fluorescence
The Perkin Elmer model LS 50 B luminescence
spectrometer was used for measuring the fluores-
440 nm and emission spectra were recorded from
455 to 600 nm. The protein solution at 34 mM, in
which fibrils were growing, was added in a buffer
with ThT dye dissolved (25 mM sodium phosphate,
pH 7.5, 15 mM ThT), immediately before the
measurement. The final protein concentration (on
5x dilution) was 6.8 mM. The ThT dye was
previously shown to monitor amyloid fibrils and that
it did not bind to the native monomeric, dimeric or
denatured states of stefin B.
Transmission electron microscopy
Protein samples (20 ml of 34 or 45 mM protein
solution, diluted 10–50 times if appropriate) were
applied on a Formvar and carbon-coated grid.
After 5 min the sample was soaked away, the grid
was washed with PBS buffer (pH 7.3), then
stained with 1% uranyl acetate. Samples were
observed with a Philips CM 100 transmission
electron microscope at 80 kV and magnifications
from 10000x to 130000x. Images were recorded
by Bioscan CCD camera Gatan, using Digital
Atomic force microscopy
A sample 20 ml of stefin B (45 mM protein concen-
tration) was spread across freshly cleaved mica
surface (1 cm2), incubated for 5 min and gently
washed with pH 4.8 solvent and deionized water.
Excess water was removed with a stream of nitrogen.
Multimode scanning probe microscope (Digital
Instruments) operating in tapping mode. Silicon
cantilevers with a typical tip radius of 20 nm were
used (Nanosensors). Scanning parameters were
different for individual tips and samples, but good
resolution was obtained for low drive amplitudes,
typically around 3 nm. Images were taken at a scan
rate of 1.5 Hz and the image resolution was
E. Zˇerovnik et al.
Glutaraldehyde (GA) cross-linking was carried out
using standard procedures . Samples with final
protein concentrations of 10 mM were reacted for
25 min at room temperature with either 0.6 wt/vol%
GA, 0.06% GA or, in the absence of GA as a control.
The reaction was quenched by adding 1/8 volume of
1M Tris buffer, pH 8.3. The cross-linked samples
were analyzed on a 15% SDS-page gel, with heating
and with no heating prior to application. As controls,
monomeric and dimeric stefin A were cross-linked
under the same conditions.
Flow cytometric experiments were carried out as
described . Briefly, 1-ml aliquotes of 34 mM
stefin B at pH 4.8 in the presence or absence of 10%
TFE were prefiltered through a 0.22-mm membrane
and subjected to fibrillation at room temperature. At
various time points (0 to 90 days) the samples were
monitored on a Becton-Dickinson FACScan Flow
cytometer. On all samples ThT fluorescence was
recorded, in parallel. Specifically, after the first-flow
cytometric acquisition (sample autofluorescence), an
amount of ThT was added (to 10 mM final con-
centration) to stefin B sample and subjected to a
second-flow cytometric acquisition. For each condi-
tion we measured a forward scatter (FSC), side
scatter (SSC) and fluorescence signal (FL2) from the
photomultiplier corresponding to 530 nm bandpass
filter. A threshold value was set at channel 52 using
the FSC data to remove the signals arising from dust
and particulate debris. Ten thousand data points
were collected for each condition.
Results and discussion
The lag phase
We have previously shown that human stefin B forms
amyloid-like fibrils at pH 4.8 and at pH 3.3, 0.26 M
sodium sulfate [13,28], conditions which lead to
native-like and structured molten globule intermedi-
ates, respectively . A prolonged lag phase in a
time frame of weeks is observed at pH 4.8 at room
temperature, with no alcohol added. Therefore, TFE
was used to accelerate the lag phase at pH 4.8 and
room temperature to a time frame of days.
Fibril morphologies obtained during the lag phase
using TEM have been shown before [13,16,30]. At
pH 4.8 in the lag phase a typical granular aggregate
was observed, which consisted of closed or open
arrangements of loosely attached oligomers. In
addition to the granular aggregate, globular particles
and chains composed of such particles could be seen
during the lag phase at pH 3.3, high ionic strength
(0.26 M sodium sulfate). AFM images of such an
aggregate are shown in Figure 1.
Figure 1. (A,B) Lag-phase species by AFM. The prefibrillar
aggregates in the buffer of pH 3.3 (0.26 M disodium sulfate).
Image was taken after 24 h of standing at room temperature,
protein concentration was 34 mM. Two magnifications are shown
as (A) and (B). The left part of each AFM image represents height
variation and the right part represents tapping mode amplitude
variations. The height bar is the same for both images.
Figure 2. Oligomer composition of the lag-phase aggregates. SDS-
PAGE analysis of stefin B samples taken during the lag phase and
cross-linked with glutaraldehyde (GA): lanes 1, 2, 3, stefin B at pH
7.3 cross-linked with 0.6% GA, 0.06% GA and not cross-linked at
all, respectively; lane M, molecular weight standards (as indi-
cated); lane 4, stefin B at pH 3.3, 0.26 M Na2SO4; lane 5, stefin B
at pH 4.8, 10% TFE; lane 6, stefin B at pH 4.8, no TFE, all cross-
linked with 0.6% GA.
Amyloid fibril formation by human stefin B
By cross-linking and SDS/PAGE (Figure 2) the
only oligomer observed in the soluble fraction at
pH 4.8 in the absence of the alcohol, was a dimer
(Figure 2, lane 6). In the cross-linking SDS-PAGE
experiment and gel filtration performed on the
soluble fractions at pH 4–8, 10% TFE or those at
pH 3.3, 0.26 M sodium sulfate, dimers were the
predominant species also. Under these latter condi-
tions, however, tetramers and some higher oligomers
were also present, producing a typical ladder of
oligomers (Figure 2, lanes 4 and 5). Oligomerization
of stefin B was also observed at pH 7.3, where fibrils
have not been detected (Figure 2, lane 1).
Early precipitation events have been observed
within an hour of triggering the reaction from the
native-like acid intermediate at pH 4.8, 378C (10%
TFE), which represents a very early phenomenon
within the lag time. These visible aggregates, which
do not bind ThT, eventually disperse but represent
very large species with dimensions in the range of
tens of microns. As observed for other proteins ,
fibrillar species appear on the boundaries of these
amorphous clumps (Figure 3A–C). This suggests
that the amorphous aggregates, similarly to the
granular ones [13,16] may be the source of nuclea-
tion sites for amyloid fibril formation. Consistent
with cross-linking and SDS-PAGE results (Figure 2)
dimers are observed by the size-exclusion FPLC as
the oligomers, which temporarily arise from the lag-
phase aggregates before fibrils form (Fig. 3D).
Effect of TFE on the rate of fibrillation
TFE titrations of stefin B at several pH values had
been carried out previously [32,33]. It has been
demonstrated that the native-like intermediate at pH
Figure 3. Amorphous aggregation and dimer formation. (A) Electron micrographs of the large amorphous aggregates occuring within an
hour of triggering the reaction at pH 4.8, 10% TFE and 378C. The scale bar represents 0.5 mm. With time, small fibrillar species appear at
the edges of the aggregates (B) and eventually mature fibers predominate on the EM grids (C). (D) Time course of the recovery of a dimeric
species of stefin B during the lag and growth phases monitored by size exclusion chromatography at pH 4.8, 10% TFE (ordinate represents
intensity of absorbance at 280 nm as measured by FPLC). A species with a retention time of 22.5 min corresponding to an estimated
molecular weight of 22 kDa (dimers) becomes the dominant species but disappears suddenly when the elongation-growth phase begins. The
figure is reproduced from reference 30, with permission from NOVA Science Publ., New York.
E. Zˇerovnik et al.
4.8 is more labile towards TFE than the ‘‘struc-
tured’’ molten globule at pH 3.3 (0.26 M sodium
sulfate). Namely, at pH 4.8, concentrations of TFE
above 10% are already denaturating to the protein,
and at 15% TFE the protein lies in the transition
region to the all a-helical denatured state . In
contrast, even 20% TFE remains predenaturational
for the structured molten globule at pH 3.3 .
The time course of stefin B amyloid fibril
formation was monitored by thioflavin T (ThT)
binding . Figure 4A shows fibrillation reactions
triggered at pH 4.8 and observed at varying TFE
concentrations. Increasing the concentration of TFE
dramatically reduces the lag phase, but decreases
both the rate of the elongation phase and the final
yield of fibrils. At TFE concentrations greater than
25%, the fibril yield approaches zero. At this
concentration of the alcohol the majority of the
protein species populated at equilibrium have under-
gone a transition to an a-helical denatured state ,
which as we show here, does not lead to amyloid
fibrils any more.
The effect of TFE has not been extensively studied
and, we could find only two reports on the effect of
TFE concentration on the rate of fibrillation [36,37].
Fezoui and Teplow , argue that ‘‘formation of a
partially folded, a helix-containing intermediate’’,
which occurs with natively unfolded peptides,
genesis of natively folded proteins’’. However, in
Figure 4. Effects of TFE, pH and salt on amyloid-fibril growth. (A) ThT fluorescence intensity at 480 nm is plotted against time of
fibrillation, at different TFE concentrations ranging from 5 to 30 (vol/vol%), 24+18C. (B) ThT fluorescence recorded at 482 nm vs. time of
fibril growth initiated from pH 3.3, 0.26 M sodium sulfate by dilution to pH 4.8, 0.13 M sodium sulfate after 3 weeks, from the final
protofibrils (filled squares) or after 24 h, from the lag phase (filled triangles). Control experiment at pH 4.8, 0.13 M sodium sulfate is also
Amyloid fibril formation by human stefin B
amyloidogenesis of the globular protein, stefin B, we
observe the same TFE effect, i.e. maximal rate of
fibril growth from an a-helical intermediate state and
no more growth from the all a-helical denatured
state. Perhaps, the mechanism of amyloidogenesis of
natively unfolded and natively folded proteins
involves a common intermediate, with additional
(non-native) a-helical structure?
Two papers appeared [38,39] in which the authors
explain the effect of TFE on the rate of fibril
formation. Both papers stress the importance of the
solvent properties of the alcohol. According to Goto
and co-workers , alcohols such as EtOH, TFE
and HFIP initially accelerate amyloid fibril forma-
tion, regardless of the protein, due to the fact that the
alcohol concentration approaches the ‘‘clustering
concentration’’. This happens for TFE at about
40% and indeed, in the set of proteins they studied,
the concentration of TFE needed for the maximal
acceleration was 20–30%. In the paper by Winter
and colleagues  the bell-shaped dependence,
which in the case of insulin had a maximum from 10
to 15% of TFE, was explained by solvent behavior of
the alcohol, which perturbs water structure in a
cosmotrophic manner, thereby, enforcing protein
hydration and promoting folding/aggregation.
We have recently studied titration by TFE of stefin
B wild type and two EPM1 mutants, among them an
N-terminal fragment  and have shown that very
different TFE concentrations are needed for unfold-
ing and amyloid-fibril formation. The chimeras
between stefin A and B were also prone to form
amyloid fibrils at very different TFE concentrations
. Therefore, we conclude that protein stability,
hydrophobicity and H-bonding potential are at least
equally important as the changed solvent properties
provided by TFE.
Dead-end pathway to protofibrils at pH 3.3?
Knowing that protofibrils in some other proteins
have been shown to be off-pathway [41,42] we
started a reaction from protofibrils (the final mor-
phology after 3 weeks at pH 3.3, 0.26 M sodium
sulfate) by changing pH to 4.8 (10% TFE, 0.13 M
sodium sulfate). To exclude the effect of changed
ionic strength, a control experiment of fibril growth
was performed at pH 4.8, 10% TFE and final
0.13 M sodium sulfate.
One would expect that the kinetics were acceler-
ated if the protofibrils combined directly into longer
fibrils. As shown in Figure 4B (see legend), no fibril
elongation was observed when protofibrils (PF) were
diluted into the fibril growth-promoting solvent.
Interestingly, the arrest in fibril growth at pH 4.8
was not observed when the reaction was started after
24 h at pH 3.3, which is during the lag phase. This
experiment shows the lag-phase species (Figure 1) to
be up-stream of the cross-road between the two
pathways leading to protofibrils or fibrils.
The above result suggests that the protofibrils are a
dead-end pathway, and agrees with flow cytometry
Figure 5. Flow cytometric analysis of stefin B fibril formation. An
example of stefin B fibrils at 34 mM protein concentration. For this
particular experiment stefin B was left at pH 4.8, room
temperature, for more than 3 months. All data were collected
using a Becton-Dickinson FACScan Flow cytometer as described
under Methods. (A) Plot of forward scattering (FSC) vs. side
scattering (SSC). The graph shows arbitrary barriers for small
fibrils, larger fibrils and aggregates. The remaining structure which
did not fulfill this criteria was labeled as other. (B) ThT
fluorescence in FL2 (fluorescence channel with a 530 nm
bandpass filter) associated with stefin B fibril formation. Flow
cytometric acquisition was carried out after addition of ThT buffer
to stefin B sample (to a final ThT concentration of 10 mM).
E. Zˇerovnik et al.
and TEM/AFM imaging where the appearance and
morphology of fibrils were different at pH 3.3,
0.26 M sodium sulfate, to that at pH 4.8. It also is
consistent with CD spectra measurements. When
CD spectra were measured in the course of the
fibrillation reactions at pH 3.3 or 4.8, a typical CD
spectrum for a ‘‘structured molten globule’’ re-
mained throughout the fibrillation at pH 3.3.
However, at pH 4.8 a native-like spectrum (in the
lag phase) changed into one resembling the pH 3.3
spectrum (more a-helical shape with minima at
208 and 220 nm), during the growth phase, and
transformed into a b-sheet like spectrum at later
times (the plateau phase) . This suggests that
early intermediates in the fibrillation reaction can
gain some a-helical structure but that species
populated at later times at pH 3.3, which preserve
this structural feature are no longer assembly
competent. It is noteworthy that oligomeric a-helical
intermediates were also detected during Ab peptide
Proportion of aggregates vs. fibrils
Concomitant decrease of aggregates with increase of
fibrillar material is observed by TEM and AFM.
Intuitively, one would judge that all the aggregates
transform into fibrils. To obtain a more quantitative
view, the ratio of aggregates relative to fibrils was
determined using flow cytometry .
A typical histogram obtained by flow cytometry at
pH 4.8 in the absence of alcohol is shown in Figure
5A. The linear dependence of forward scattering
(FSC) vs. side scattering (SSC) is characteristic of
fibrillar aggregates whereas amorphous aggregates
show a stochastic distribution . The quantitation
of stefin B fibrils and aggregate formation was
obtained from the arbitrary labeling as indicated in
Figure5A. Fibrilswere dissectedassmall
(FSC?500) and large (FSC: 500–1000). Amor-
phous aggregates share a low forward scatter
(FSC?200) and an increasing side scatter (SSC:
100–1000). The remaining structure which did not
fulfill these criteria was labeled as ‘other’. Figure 5B
is depicts ThT dye fluorescence, collected in FL2
channel. All fibrillar particles bind ThT, as expected
Table I shows the percentage of smaller vs. larger
fibrils, amorphous aggregate and the remaining
structures. This was collected as a function of time
at pH 4.8 and pH 4.8 with 10% TFE. In the lag
phase, both at pH 4.8 or 4.8 with 10% TFE, no
major changes over time were detected. There
was only a small amount of the aggregates at pH
4.8 in the absence of the alcohol. After matura-
tion (Table I), the amount of smaller fibrils di-
minished (from 80 to 50%) while the amount of
the larger fibrils and the remaining structures
increased. Importantly, in experiments at pH 4.8,
flow cytometry suggests that amorphous aggregates
constitute in fact only a very small proportion
(50.2%) of the protein sample at any time point.
It is possible that they appear more prevalent by
electron microscopy as a result of a preferential
adherence to carbon grids.
Fibrillar material with different properties was
observed at the plateau of the reaction at pH 3.3 (not
shown). The relation of forward to side scatter was
linear; however, the side scatter was weaker in
absolute values. This we interpret as indicative of
thinner and shorter fibrils, which is confirmed by
AFM (next section).
flow cytometry data shows that a pool of fibril-
competent conformations is formed initially as some
dead end of folding and that these conformations do
not transform into each other to any significant
Table I. Flow cytometry of amyloid fibrillation. Flow cytometric measurements and analysis were carried out as described under Methods.
The quantitation of stefin B fibril formation, namely fibrils:aggregates ratio resulted from the arbitrary labeling as indicated in Figure 5A.
Fibrils were dissected as small (FSC?500) and large (FSC: 500–1000). Amorphous aggregates share a low forward scatter (FSC?200)
and an increasing side scatter (SSC: 100–1000). The remaining structure which did not fulfill these criteria was labeled as ‘‘other’’. Fibril
formation was proved by monitoring thioflavin T (ThT) fluorescence. All results are expressed as percentages. Relative errors are around
10% of the values.
other conditionsTime Small fibrils (%)Larger fibrils (%)Aggregates‘‘Other’’ ThT (%)
4.8 0–24 h
4.8, 10% TFE0–24 h
Amyloid fibril formation by human stefin B
Fibril morphologies as observed by AFM
AFM provides three-dimensional images of the
objects and has been used widely to visualize amyloid
fibrils [41,44–46]. In interpreting the fibril dimen-
sions by AFM, one should be aware that lateral
dimensions can appear bigger due to the finite tip
size. It therefore should be kept in mind that height is
measured more accurately by AFM while more
reliable lateral dimensions (width and length) can
be obtained by TEM.
In Figure 6A and B, the morphologies at 24 h and
at the plateau of the reaction at pH 3.3, 0.26 M
sodium sulfate, which is 3 weeks, are shown.
Initially, ellipsoid particles of a diameter around
12 nm can be seen (Figure 6A). These were analyzed
more thoroughly in our other work on the dimen-
sions of cystatin ‘‘toxic oligomer’’ (Cˇeru et al. 2007,
submitted for publication). In Figure 6C and D, the
reaction of the fibril growth at pH 4.8 and 10% TFE
is shown. Morphologies at 24 h and a final point
(more than 3 weeks) were imaged. Initially, shorter,
and circular arrangements – rings (the latter not
presented) were seen, which disappeared with
time. Finally, the longer, unbranched fibrils prevail
In Figure 7A–C, AFM images of mature stefin B
fibrils under different incubation conditions are
compared. Similar TEM data were also obtained
[30, chapter 6] and are not replicated here. At pH
4.8, the fibrils appear long, straight and un-
branched (Figure 7A). The same reaction in the
morphology but the fibrils often clump together,
causing some distortion (Figue 7B). Dimensions of
the fibrils are 14 nm in diameter and up to several
mm in length. Narrow filaments of 7 nm are also
observed, supporting the hypothesis that the fibrils
are made of intertwined filaments, with a regular
helicalperiodicityof 26 nm
reaction is triggered from the structured molten
globule at pH 3.3, 0.26 M Na2SO4, the final fibrils
appear shorter and less regular than at pH 4.8
(Figure 7C), resembling protofibrils. At pH 3.3,
0.26 M Na2SO4and 20% TFE, the majority of the
sample aggregated in an amorphous manner (TEM
data in [30, chapter 6]) despite the fact that this
concentration of the alcohol corresponds to opti-
mum predenaturational conditions for the struc-
tured molten globule of stefin B populated at this
It can be seen that amyloid-like fibrils can form
from stefin B under a variety of conditions, however,
those which allow the slowest nucleation and growth
(pH 4.8, room temperature) produce the longest and
most regular fibrils. This often is not the case in
physiological conditions (36–418C, glycosaminopro-
teoglycans, membrane components), where initial
amorphous aggregation is likely.
Figure 6. AFM data on the time course of morphology changes.
Stefin B dissolved in the pH 3.3 buffer 0.26 M disodium sulfate at
34 mM protein concentration (A) after 24 h (B) at the end of
reaction, which is 3 weeks. Stefin B dissolved in the pH 4.8 buffer
with 10% TFE at 34 mM protein concentration (C) after 24 h (D)
at the end of reaction (more than 3 weeks). The fibrillation
reactions took place at room temperature (24+18C). The left part
of the AFM images represents height variation and the right part
represents tapping mode amplitude variations. The height bar is
the same for both images.
E. Zˇerovnik et al.
Mechanisms of amyloid-fibril formation of various
proteins might differ in details ; however,
common features are observed for most proteins.
Fibrils of stefin B can grow from granular aggregate
(pH 4.8), globular oligomers (pH 3.3) or amorphous
aggregate (4378C, 410% TFE), all accumulating
in the lag phase, in different proportions under
different solution conditions. Massive amorphous
aggregates (in parallel to the granular ones) preced-
ing amyloid-fibril formation, which accumulate at
pH 4.8 (10% TFE) at somewhat elevated tempera-
tures (Figure 3), have been observed for other
amyloidogenic proteins before, and may represent
nucleation sites for amyloid-fibril formation .
Under other conditions however, flow cytometry
suggests that amorphous aggregates are not a
significant proportion of the different species popu-
lated during amyloid formation and so their role is
Different morphological species are observed in
the course of amyloid fibrillation reaction of stefin B.
By varying pH, TFE concentration and temperature,
the outcome of fibrillation can be modified. Whereas
at low pH (3.3), the reaction yields ‘‘off-pathway’’
protofibrils, higher pH (4.8) yields mature amyloid
fibrils. Intriguingly, the time course for morphologi-
cal changes occurring during the amyloid fibril
formation by this protein is reminiscent of that
described for Ab [21,48–50], transthyretin  and
some other amyloidogenic proteins [4,41,44–46,52].
The importance of suitable conditions for studying
prefibrillar states is self-evident since in the Ab case
, which causes Alzheimer’s disease, but also for a
number of other proteins, oligomers and protofibrils
have been shown to be more toxic for the cell than
mature amyloid [54,55].
The effect of TFE [36–39] on the fibril growth by
stefin B is promoted at predenaturational TFE
concentrations but not at denaturational concentra-
tions. Similarly, fibril growth is maximized and the
lag phase decreased at predenaturational tempera-
tures (the melting temperature is ca. 408C at pH 4.8)
and, pH 4.8 favors fibril formation better than the
more destabilizing pH 3.3. Together, this argues for
a partially unfolded intermediate as a precursor for
fibril initiation and suggests that structure-promoting
conditions are essential for the correct maturation of
For several cystatins studied to date, 3D domain-
swapped dimers and in some cases, tetramers are
formed under conditions which favor fibrillation
[17,20,56–59]. We have shown previously that the
formation of domain-swapped dimers is linked to
nearly complete unfolding [17,57]. For the close
relative of stefin B, stefin A, domain-swapped dimers
form upon cooling from predenaturational tempera-
tures with a high energy of activation in a range of
protein unfolding reactions . Thus, domain-
swapped dimers [56,57] and tetramers of cystatins
[58,59] are likely the ones which accumulate in the
lag phase as a result of partial unfolding. Their actual
role in fibril initiation, fibril growth and maturation is
the subject of further studies.
The authors are grateful to Ing. Luise Kroon Zˇitko
for help in expression and isolation of recombinant
human stefin B, to Dr Manca Kenig for the
Figure 7. AFM images of stefin B amyloid fibrils in different
solvents. Samples were analyzed in the final stage of growth
(plateau region) in each solvent: (A) pH 4.8, (B) pH 4.8 with 10%
TFE and (C) pH 3.3, 0.26 M Na2SO4. The left part of each AFM
image represents height variation and right part represents tapping
mode amplitude variations. The height bar is the same for all three
Amyloid fibril formation by human stefin B
cross-linking/SDS experiment and to Uros ˇ Gregorc
MSc for assistance with flow cytometry measure-
ments. For continuous TEM measurements (pre-
sented in references 13,16,30) we are indebted to Dr
Marus ˇa Pompe Novak and MSc Magda Tus ˇek
Zˇnidari? c from the National Institute of Biology
(NIB), Ljubljana. This work was funded by the
Ministry of Higher Education, Science and Technol-
ogy of the Republic Slovenia (through the Slovenian
Research Agency – ARRS) and the BBSRC, UK.
R.A.S. is a Royal Society university research fellow.
1. Turk V, Bode W. The cystatins; protein inhibitors of cysteine
proteinases. FEBS Lett 1991;285:213–219.
2. Turk B, Turk D, Salvesen GS. Regulating cysteine protease
activity: essential role of protease inhibitors as guardians and
regulators. Curr Pharm Des 2002;8:1623–1637.
3. Jones B, Roberts PJ, Faubion WA, Kominami E, Gores GJ.
Cystatin A expression reduces bile salt-induced apoptosis in a
rat hepatoma cell line. Am J Physiol 1998;275:G723–723.
4. Pennacchio LA, Bouley DM, Higgins KM, Scott MP,
Noebels JL, Myers RM. Progressive ataxia, myoclonic
epilepsy and cerebellar apoptosis in cystatin B-deficient mice.
Nature Genet 1998;20:251–258.
5. Lieuallen K, Pennacchio LA, Park M, Myers RM, Lennon
GG. Cystatin B-deficient mice have increased expression of
apoptosis and glial activation genes. Hum Mol Genet
6. Pennacchio LA, Lehesjoki AE, Stone NE, Willour VL,
Virtaneva K, Miao J, D’Amato E, Ramirez L, Faham M,
Koskiniemi M, Warrington JA, Norio R, de la Chapelle A,
Cox DR, Myers RM. Mutations in the gene encoding cystatin
B in progressive myoclonus epilepsy (EPM1). Science
7. Bespalova IN, Adkins S, Pranzatelli M, Burmeister M. Novel
cystatin B mutation and diagnostic PCR assay in an
Unverricht-Lundborg progressive myoclonus epilepsy patient.
Am J Med Genet 1997;74:467–471.
8. Kagitani-Shimono K, Imai K, Okamoto N, Ono J, Okada S.
Unverricht-Lundborg disease with cystatin B gene abnormal-
ities. Pediatr Neurol 2002;26:55–60.
9. Jensson O, Palsdottir A, Thorsteinsson L, Arnason A,
Abrahamson M, Olafsson I, Grubb A. Cystatin C mutation
causing amyloid angiopathy and brain hemorrhage. Biol
Chem Hoppe Seyler 1990;371(Suppl.):229–232.
10. Bjarnadottir M, Nilsson C, Lindstro ¨m V, Westman A,
Davidsson P, Thormodsson F, Blo ¨ndal H, Gudmundsson
G, Grubb A. The cerebral hemorrhage-producing cystatin C
variant (L68Q) in extracellular fluids. Amyloid: J Protein
Folding Disord 2001;8:1–10.
11. Deng A, Irizarry MC, Nitsch RM, Growdon JH, Rebeck GW.
Elevation of cystatin C in susceptible neurons in Alzheimer’s
disease. Am J Pathol 2001;159:1061–1068.
12. Ii K, Ito H, Kominami E, Hirano A. Abnormal distribution of
cathepsin proteinases and endogenous inhibitors (cystatins) in
the hippocampus of patients with Alzheimer’s disease,
parkinsonism-dementia complex on Guam, and senile de-
mentia and in the aged. Virchows Arch A Pathol Anat
13. Zˇerovnik E, Pompe-Novak M, Sˇkarabot M, Ravnikar M,
Mus ˇevi? c I, Turk V. Human stefin B readily forms amyloid
fibrils in vitro. Biochim Biophys Acta 2002;1594:1–5.
14. Jenko S, Sˇkarabot M, Kenig M, Gun? car G, Mus ˇevi? c I,
TurkD, ZˇerovnikE. Different
amyloid fibrils by two homologous proteins – human stefins
A and B: searching for an explanation. Proteins 2004;55:417–
15. ZˇerovnikE,Zavas ˇnik-Bergant
Pompe-Novak M, Sˇkarabot M, Goldie K, Ravnikar M,
Mus ˇevi? c I, Turk V. Amyloid fibril formation by human stefin
B in vitro: immunogold labelling and comparison to stefin A.
Biol Chem 2002;383:859–863.
16. Anderluh G, Gutierrez-Aguirre I, Rabzelj S, Cˇeru S,
Kopitar-Jerala N, Ma? cek P, Turk V, Zˇerovnik E. Interaction
of human stefin B in the prefibrillar oligomeric form
with membranes – correlation with cellular toxicity. FEBS J
17. Jerala R, Zˇerovnik E. Accessing the global minimum
conformation of stefin A dimer by annealing under partially
denaturing conditions. J Mol Biol 1999;291:1079–1089.
18. Konno T, Murata K, Nagayama K. Amyloid-like aggregates
of a plant protein: a case of a sweet-tasting protein, monellin.
FEBS Letts 1999;454:122–126.
19. Kenig M, Jenko-Kokalj S, Tus ˇek-Zˇnidari? c M, Pompe-Novak
M, Gun? car G, Turk D, Waltho JP, Staniforth RA, Avbelj F,
Zˇerovnik E. Folding and amyloid fibril formation for a series
of human stefins’ chimeras: any correlation? Proteins 2006;
20. Nillson M, Wang X, Rodziewicz-Motowidlo S, Janowski R,
Lindstro ¨m V, O¨nnerfjord P, Westermark G, Grzonka Z,
Jaskolski M, Grubb A. Prevention of domain swapping
inhibits dimerization and amyloid fibril formation of cystatin
C. J Biol Chem 2004;279:24236–24245.
21. Lomakin A, Teplow DB, Kirschner DA, Benedek GB.
Kinetic theory of fibrillogenesis of amyloid beta-protein. Proc
Natl Acad Sci USA 1997;94:7942–7947.
22. Munishkina LA, Phelan
Conformational behavior and aggregation of a-synuclein
in organic solvents: modeling the effects of membranes.
23. Matsunaga Y, Zˇerovnik E, Yamada T, Turk V. Conforma-
tional changes preceding amyloid fibril formation of amyloid-
beta and stefin B; parallels in pH dependence. Curr Med
24. Matsunaga Y,Zˇerovnik
Conformational changes preceding amyloid fibril formation
of amyloid-beta, prion protein and stefin B; parallels in pH
dependence. Med Chem Rev Online 2005;2:359–367.
25. Jerala R, Trstenjak M, Lenar? ci? c B, Turk V. Cloning a
synthetic gene for human stefin B and its expression in E. coli.
FEBS Lett 1988;239:41–44.
26. Craig WS. Determination of quaternary structure of an active
enzyme using chemical cross-linking with glutaraldehyde.
Methods Enzymol 1988;156:333–345.
27. Wall J, Solomon A. Flow cytometric characterization of
amyloid fibrils. Methods Enzymol 1999;309:460–466.
28. Zˇerovnik E, Turk V, Waltho JP. Amyloid fibril formation
by human stefin B: influence of the initial pH-induced
intermediate state. Biochem Soc Trans 2002;30:543–547.
29. Zˇerovnik E, Jerala R, Kroon-Zˇitko L, Turk V, Lohner K.
Characterization of the equilibrium intermediates in acid
denaturation of human stefin B. Eur J Biochem 1997;
30. Zˇerovnik E, Giannini S, Stoka V, Tus ˇek-Zˇnidari? c M,
amyloid fibrillation: stefin B as a good model protein.
In: Zˇerovnik E, Kopitar-Jerala N, Uversky VU, editors.
human stefins and cystatins. New York: Nova Science,
2006. pp 97–114.
C, Uversky VN,FinkAL.
E, YamadaT, TurkV.
E. Zˇerovnik et al.
31. Smith DP, Jones S, Serpell LC, Sunde M, Radford SE. A Download full-text
systematic investigation into the effect of protein destabilisa-
tion on beta 2-microglobulin amyloid formation. J Mol Biol
32. Zˇerovnik E, Virden R, Jerala R, Kroon-Zˇitko L, Turk V,
Waltho JP. Differences in the Effects of TFE on the folding
pathways of human stefins A and B. Proteins 1999;36:205–
33. Zˇerovnik E, Kenig M, Waltho JP, Staniforth RA. Conclusions
on the mechanism of protein folding from stefins and cystatins
studies. In: Uversky VU, Zˇerovnik E, Kopitar-Jerala N,
editors. The molecular anatomy and physiology of proteins:
human stefins and cystatins. New York: Nova Science. 2006.
34. Klunk WE, Jacob RF, Mason RP. Quantifying amyloid
by congo red spectral shift assay. Methods Enzymol 1999;
35. Hamada D, Goto Y. The equilibrium intermediate of
b-lactoglobulin with non-native a-helical structure. J Mol
36. Fezoui Y, Teplow DB. Kinetic studies of amyloid beta-protein
fibril assembly. Differential effects of alpha-helix stabilization.
J Biol Chem 2002;277:36948–36954.
37. Chiti F, Taddei N, Bucciantini M, White P, Ramponi G,
Dobson CM. Mutational analysis of the propensity for
amyloid formation by a globular protein. EMBO J 2000;
38. Yamaguchi K, Naiki H, Goto Y. Mechanism by which the
amyloid-like fibrils of b2-microglobulin fragment are induced
by fluorine-substituted alcohols. J Mol Biol 2006;363:279–
39. Grudzielanek S, Jansen R, Winter R. Solvational tuning of the
unfolding, aggregation and amyloidogenesis of insulin. J Mol
40. Rabzelj S, Turk V, Zˇerovnik E. In vitro study of stability and
amyloid fibril formation of two mutants of human stefin B
(cystatin B) occurring in patients with EPM1. Protein Sci
41. Kad NM, Myers SL, Smith DP, Smith DA, Radford SE,
Thomson NH. Hierarchical assembly of b2-microglobulin
amyloid in vitro revealed by atomic force microscopy. J Mol
42. Gosal WS, Morten IJ, Hewitt EW, Smith DA, Thomson NH,
Radford SE. Competing pathways determine fibril morphol-
ogy in the self-assembly of beta2-microglobulin into amyloid.
J Mol Biol 2005;351:850–864.
43. Kirkitadze MD, Condron MM, Teplow DB. Identification
and characterization of key kinetic intermediates in amyloid
beta-protein fibrillogenesis. J Mol Biol 2001;312:1103–1119.
44. Souillac PO, Uversky VN, Fink AL. Structural transforma-
tions of oligomeric intermediates in the fibrillation of the
immunoglobulin light chain LEN. Biochemistry 2003;42:
45. Kad NM, Thomson NH, Smith DP, Smith DA, Radford SE.
b2-Microglobulin and its deamidated variant, N17D, form
amyloid fibrils with a range of morphologies in vitro. J Mol
46. Malisauskas M, Zamotin V, Jass J, Noppe W, Dobson CM,
Morozova-Roche LA. Amyloid protofilaments from the
calcium-binding protein equine lysozyme: formation of ring
and linear structures depends on pH and metal ion
concentration. J Mol Biol 2003;330:879–890.
47. Zˇerovnik E. Amyloid fibril formation; proposed mechanisms
and relevance to conformational disease. Eur J Biochem
48. Lomakin A, Teplow DB, Kirschner DA, Benedek GB.
Kinetic theory of fibrillogenesis of amyloid beta-protein. Proc
Natl Acad Sci U S A 1997;94:7942–7947.
49. Yong W, Lomakin A, Kirkitadze MD, Teplow DB, Chen SH,
Benedek GB. Structure determination of micelle-like inter-
mediates in amyloid beta-protein fibril assembly by using
small angle neutron scattering. Proc Natl Acad Sci USA
50. Tjernberg LO, Pramanik A, Bjorling S, Thyberg P, Thyberg J,
Nordstedt C, Berndt KD, Terenius L, Rigler R. Amyloid
beta-peptide polymerization studied using fluorescence corre-
lation spectroscopy. Chem Biol 1999;6:53–62.
51. Serag AA, Altenbach C, Gingery M, Hubbell WL, Yeates TO.
Identification of a subunit interface in transthyretin amyloid
fibrils: evidence for self-assembly from oligomeric building
blocks. Biochemistry 2001;40:9089–9096.
52. Chamberlain AK, MacPhee CE, Zurdo J, Morozova-Roche
LA, Hill HA, Dobson CM, Davis JJ. Ultrastructural
organization of amyloid fibrils by atomic force microscopy.
Biophys J 2000;79:3282–3293.
53. Lambert MP, Barlow AK, Chromy BA, Edwards C, Freed R,
Liosatos M, Morgan TE, Rozovsky I, Trommer B, Viola KL,
Wals P, Zhang C, Finch CE, Krafft GA, Klein WL.
Diffusible, nonfibrillar ligands derived from Abeta1-42 are
potent central nervous system neurotoxins. Proc Natl Acad
Sci U S A 1998;95:6448–6553.
54. Lashuel HA, Hartley D, Petre BM, Walz T, Lansbury PT Jr.
Neurodegenerative disease: amyloid pores from pathogenic
mutations. Nature 2002;418:291.
55. Walsh DM, Hartley DM, Kusumoto Y, Fezoui Y, Condron
MM, Lomakin A, Benedek GB, Selkoe DJ, Teplow DB.
Amyloid beta-protein fibrillogenesis. Structure and biological
56. Janowski R, Kozak M, Jankowska E, Grzonka Z, Grubb A,
Abrahamson A, Jaskolski M. Human cystatin C, an amyloi-
dogenic protein, dimerizes through three-dimensional domain
swapping. Nat Struct Biol 2001;8:316–320.
57. Staniforth RA, Giannini S, Higgins LD, Conroy MJ,
Hounslow AM, Jerala R, Craven CJ, Waltho JP. Three-
dimensional domain swapping in the folded and molten-
globule states of cystatins, an amyloid-forming structural
superfamily. EMBO J 2001;20:4774–4781.
58. Sanders A, Jeremy Craven C, Higgins LD, Giannini S,
Conroy MJ, Hounslow AM, Waltho JP, Staniforth RA.
Cystatin forms a tetramer through structural rearrangement
of domain-swapped dimers prior to amyloidogenesis. J Mol
59. Jenko Kokalj S, Gun? car G, Sˇtern I, Morgan G, Rabzelj S,
Kenig M, Staniforth RA, Waltho JP, Zˇerovnik E, Turk D.
Essential role of proline isomerization in stefin B tetramer
formation. J Mol Biol 2007;366:1569–1579.
Amyloid fibril formation by human stefin B