Keywords copper-binding proteins; cystatin; inhibition of amyloid fibril formation; oligomers; protein aggregation; stefin B Correspondence E.. There is no difference in the affinity of c
Trang 1role in the inhibition of amyloid fibrillation
Eva Zˇerovnik1, Katja Sˇ kerget1
, Magda Tusˇek-Zˇnidaricˇ2, Corina Loeschner3, Marcus W Brazier3 and David R Brown3
1 Department of Biochemistry and Molecular Biology, Jozˇef Stefan Institute, Ljubljana, Slovenia
2 Department of Plant Physiology and Biotechnology, National Institute of Biology, Ljubljana, Slovenia
3 Department of Biology and Biochemistry, University of Bath, UK
Common features of many neurodegenerative diseases
are misfolding, aggregation and amyloid fibril
forma-tion of a pathological mutant (in inherited diseases), or
of a normal protein or its normal variant (in sporadic
cases) Amyloid fibril formation is regarded as a generic
property, common to most proteins [1,2], encouraging
the study of proteins not involved in any pathology
Amyloid-induced toxicity has also been proposed to be
a generic phenomenon [3], with prefibrillar oligomers as
the most likely toxic agent Sequestering of the fibrils
into intracellular inclusions or extracellular plaques
might actually be beneficial, as it is now believed that
soluble oligomers are the cause of the initial insult rather than the insoluble fibrous material
Environmental factors, among them metal ions, are believed to contribute to the onset of Parkinson’s, Alzheimer’s and prion disease The influence of metal ions on the underlying process of amyloid fibril forma-tion remains controversial In a number of cases copper, like other redox active metals, has been shown to pro-mote aggregation or polymerization However, there are recent reports that binding of Cu2+and Zn2+, but not Fe3+, to amyloid-b peptide retards amyloid fibril formation [4]
Keywords
copper-binding proteins; cystatin; inhibition
of amyloid fibril formation; oligomers;
protein aggregation; stefin B
Correspondence
E Zˇerovnik, Department of Biochemistry
and Molecular Biology, Jozˇef Stefan
Institute, Jamova 39, 1000 Ljubljana,
Slovenia
Fax: +386 1477 3984
Tel: +386 1477 3753 ⁄ 3900
E-mail: eva.zerovnik@ijs.si
David R Brown, Department of Biology and
Biochemistry, University of Bath, Claverton
Down, Bath, BA2 7AY, UK
Fax: +44 1225 386779
Tel: +44 1225 383133
E-mail: bssdrb@bath.ac.uk
(Received 3 May 2006, revised 16 July
2006, accepted 18 July 2006)
doi:10.1111/j.1742-4658.2006.05426.x
We show that human stefin B, a protease inhibitor from the family of cystatins, is a copper binding protein, unlike stefin A We have used isothermal titration calorimetry to directly monitor the binding event at
pH 7 and pH 5 At pH 7 stefin B shows a picomolar affinity for copper but at pH 5 the affinity is in the nanomolar range There is no difference
in the affinity of copper between the wildtype stefin B (E31 isoform) and a variant (Y31 isoform), whereas the mutant (P79S), which is tetrameric, does not bind copper The conformation of stefin B remains unaltered by copper binding It is known that below pH 5 stefin B undergoes a conform-ational change and amyloid fibril formation We show that copper binding inhibits the amyloid fibril formation and, to a lesser degree, the initial aggregation Similarities to and differences from other copper binding amy-loidogenic proteins are discussed
Abbreviations
AFM, atomic force microscopy; ITC, isothermal titration calorimetry; SEC, size exclusion chromatography; TEM, transmission electron microscopy; TFE, 2,2,2 trifluorethanol; ThT, thioflavin T.
Trang 2Using human stefin B as a model for studying the
mechanism of amyloid fibril formation [5,6], it has
been found that fibrillation starts with a lag phase and
continues with a fibril growth reaction The lag phase,
in which granular (micellar-like) aggregate accumulates
[5], can be reduced by increasing the temperature, by
adding the organic solvent trifluoroethanol (TFE) or
by seeding [6] The time course for morphological
changes occurring during the amyloid fibril formation
by human stefin B is reminiscent of that described
for other amyloidogenic proteins, including amyloid-b
peptide [7] By following the kinetics of stefin B fibril
formation, conditions were defined where the protein
exists in the form of prefibrillar oligomers⁄ aggregates,
which persist during the lag phase The prefibrillar
forms were shown to be cytotoxic and to interact with
acidic phospholipids [8]
Human stefin B, officially termed cystatin B
(sub-family A, (sub-family I25 of cystatins following the
MEROPS classification [9]), is a cysteine protease
inhibitor [10,11] The structure and function of this
protein may be relevant to both amyloid fibril
forma-tion and metal binding Stefin B is homologous to a
closely related protein, stefin A Crystal structures of
stefin B in complex with papain [12] and of stefin A in
complex with cathepsin H [13] have been determined
The solution structure of free stefin A is also known
[14] Domain-swapped dimers have been shown for
ste-fin A and for cystatin C [15–17] Domain swapping
may have a role in amyloid fibril formation of this
family of proteins [16]
Stefin B is expressed widely in human tissue and is
thought to act as an inhibitor of the lysosomal
cathep-sins Alternative functions are possible, as the protein
was found as part of a multiprotein complex of
unknown function, specific to the cerebellum [18] It is
located not only in the lysosomes and in the
cyto-plasm, but also in the nucleus [19] Lack of expression
of stefin B is associated with signs of cerebellar
gran-ular cell apoptosis, ataxia and myoclonus as shown in
studies of stefin B deficient mice [20] Genes involved
in the activation of glial cells were overexpressed in
such mice [21] Stefin B (cystatin B gene) is also tightly
linked to epilepsy Mutations in this gene [22,23],
which lead mainly to lower protein expression, result
in progressive myoclonus epilepsy of the Unverricht–
Lundborg type The protein was reported to be
overex-pressed after seizures [24], implicating its
neuroprotec-tive role, similarly to that of cystatin C [25] Similarly
to cystatin C [26], it was found as a constituent of
senile plaques of different disease origin [27]
In the current study the ability of stefin B to bind
copper was assessed We used isothermal titration
calorimetry to monitor the binding event at pH 7 and
pH 5 It was found that the protein binds two Cu2+ atoms with high affinity whereas the mutant P79S, which is tetrameric, does not It also was shown that the presence of equimolar to three-fold molar excess of
Cu2+inhibits the fibrillation propensity of the protein This was demonstrated by thioflavin T fluorescence and electron microscopy
Results
Measurement of copper binding by isothermal titration calorimetry
One of the most widely accepted methods for deter-mining the affinity of a ligand for a protein is iso-thermal titration calorimetry (ITC) We used ITC to determine the affinity of copper for human stefins Recombinant stefin proteins were dissolved in 5 mm Mes buffer at either pH 7 or pH 5 The proteins ana-lyzed were stefin A, stefin B (E31 isoform), a variant
of the protein (variant 2) with a change from E to Y
at amino acid residue 31 (Fig 1), and a mutant form
of the variant (P79S) Copper was found to bind to stefin B at pH 7 but not to stefin A (Fig 2) The bind-ing isotherm data were fitted to sequential bindbind-ing site parameters and the best fit, producing the smallest v2 values, indicated two binding sites, both with affinities
in the picomolar range at pH 7 (Table 1) No optimal
fit was found for stefin A, indicating that the protein has no specific affinity for copper
Further analysis showed that stefin B, again unlike stefin A, also binds copper at pH 5 but that the affin-ity is by two orders of magnitude less, in the nano-molar range (Fig 3, Table 1) Additional ITC experiments were carried out with the variant form of stefin B and its mutant form, P79S The variant stefin B (Y31 isoform) binds Cu2+ with similar affinity to that
of the more common E31 isoform (Table 1) However, the mutant form of the variant, P79S, shows no specific copper binding at either pH (Fig 3, Table 1)
Conformation and stability in the presence and absence of copper
Stefin B is a predominantly b-sheet protein with five strands wrapping around an a-helix The far UV CD spectra in Fig 4A reveal small differences in intensity and shape between stefins A and B and the P79S mutant The two isoforms of stefin B have exactly the same far UV CD Regardless of the sequence differences (as highlighted in Fig 1), the secondary and tertiary structures of stefins A and B are the same, as determined
Trang 3from the 3D structures [12,14] Therefore, the
differ-ences in the far UV CD must be due to the contribution
of tyrosines to this region of the spectrum [28]
Figure 4B shows the near UV CD spectra of
stefin A, stefin B, and the P79S mutant It can be seen
that spectra of stefin B and the P79S mutant are very
similar in shape whereas that of stefin A is different
This is accounted for by the different aromatic amino
acid content (Fig 1) The similar shapes of the stefin B
and P79S spectra provides evidence for similar 3D
structure and correct folding of the mutant The lower
intensity may arise from partitioning of the protein
into an aggregated state
The effect of copper binding on the secondary
struc-ture was determined using CD in the far UV To
prepare proteins without Cu2+, part of each protein
solution was exchanged by ultrafiltration with the chelating buffer at pH 7 and then diluted to the appro-priate concentration The other part was diluted directly into buffer with 50 lm CuSO4 The far UV
CD spectra of stefin B (E31 isoform) were the same in the presence and absence of copper (Fig 5A) This indicates that copper binding has no apparent effect
on the structure of the protein Similarly, the presence
of Cu2+had no effect on the CD spectrum of variant
2 (Fig 5B) For comparison, the spectra of the P79S mutant and of stefin A were recorded with and with-out copper (Fig 5C,D) The latter two proteins do not bind Cu2+ Being aware of the pitfalls of such an ana-lysis for proteins with unusual aromatic contribution
to the far UV CD, the secondary structure estimates were calculated from the far UV CD spectra using
-6
-4
-2
0
2
0 10 20 30 40 50 60 70
Time (min)
-0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
-2
0
Molar Ratio
kcal/mole of injectant kcal/mole of injectant
-2 -1 0
0 10 20 30 40 50 60 70
Time (min)
-0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 -4
-2 0
Molar Ratio
Fig 2 Comparison of copper binding by ste-fin A and steste-fin B using ITC Protein sam-ples were prepared at 35 l M in 5 m M Mes
pH 7 buffer The ligand used for injection was Cu2SO4equilibrated with a four molar excess of glycine Cu 2+ was applied up to a five molar excess The top panel represents the data from the titration as a series of peaks corresponding to the heat change (lcalÆs)1) with each injection The bottom panel is a plot of heat change on ligand addi-tion (kcalÆmole)1) against the ligand ⁄ stefin molar ratio The background heat change from the Cu 2+ ⁄ Gly mixture injected in the Mes buffer was subtracted from the raw data Data of one representative experiment each is shown.
Wt mmsgapsatq pataetqhia dqvrsqleek enkkfpvfka vsfksqvvag tnyfikvhvg dedfvhlrvf qslphenkpl
Var2 mmsgapsatq pataetqhia dqvrsqleek ynkkfpvfka vsfksqvvag tnyfikvhvg dedfvhlrvf qslphenkpl
P79S mmsgapsatq pataetqhia dqvrsqleek ynkkfpvfka vsfksqvvag tnyfikvhvg dedfvhlrvf qslphenksl
StA mipgglseak patpeiqeiv dkvkpqleek tnetygklea vqyktqvvag tnyyikvrag dnkymhlkvf kslpgqnedl
Wt tlsnyqtnka khdeltyf
Var2 tlsnyqtnka khdeltyf
P79S tlsnyqtnka khdeltyf
StA vltgyqvdkn kddeltgf
Fig 1 Comparison of stefin sequences Shown are the primary amino acid sequences of the three stefin B proteins studied and that of stefin A The potential copper binding site with four histidine residues is shown in the boxes Differences between the wildtype stefin B, variant
2 and the P79S mutant of the variant are shown by the bold, underlined letters All four proteins of 98 amino acids are approximately 11 kDa.
Trang 4dichroweb online software [29,30] There was no
dif-ference when comparing the effects of Cu2+, which
supports the conclusion that copper binding does not
alter the secondary structure of the protein
Near UV CD spectra are a better measure of protein tertiary structure and thus the solution conformation There are cases where the tertiary structure can denature with no substantial change in the secondary structure, leading to intermediate states Similar obser-vations were made for stefin B previously [31] Near
UV CD spectra of the protein at pH 7 and pH 5 with
no metal bound and in its presence were recorded (Fig 5E,F) The spectra show that stefin B is sensitive
to Cu2+ at pH 7 where a significant decrease in ellip-ticity is detected, whereas at pH 5 this does not seem
to be the case Lower intensity of the CD signal at
pH 7 in presence of Cu2+ (Fig 5E) may, similarly to P79S (Fig 4B), arise from enhanced protein aggrega-tion rather than a conformaaggrega-tional change
To assess protein stability, thermal denaturation of the protein in presence of Cu2+ or in its absence was recorded at 210 nm (at a protein concentration of around 20 lm; not shown) and there was no difference
in the temperature of half-denaturation Performing thermal denaturation at 277 nm (which is only possible
at around 100 lm protein concentration) has shown
Molar Ratio -0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
Molar Ratio
kcal/mole of injectant kcal/mole of injectant
kcal/mole of injectant kcal/mole of injectant
-2
0
-1
-0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
Stefin A pH5 -2
0
-1
-0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 -0.20
-0.15 -0.10 -0.05 0.00 0.05 0.10
Molar Ratio
P79S pH7
-0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 -0.20
-0.15 -0.10 -0.05 0.00 0.05 0.10
Molar Ratio P79S pH5
Variant 2 pH7 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 -4
-2 0
Molar Ratio
Stefin B pH5
Molar Ratio
Variant 2 pH5 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 -4
-2 0
Fig 3 Effects of pH and primary sequence
on stefin copper binding Further analysis of
stefin proteins (35 l M ) with ITC to assess
affinity of copper binding Experiments were
carried out at either pH 7 or pH 5 using
5 m M Mes as a buffer Shown are stefin A
and stefin B at pH 5 and variant 2 of
ste-fin B and the P79S mutant of the variant at
both pH 7 and pH 5 Each panel is a plot of
heat change on ligand addition (kcalÆmole)1)
against the ligand ⁄ stefin molar ratio for one
of the proteins analysed The background
heat change from the Cu 2+ ⁄ Gly mixture
injected in the Mes buffer was subtracted
from the raw data Data of one
representa-tive experiment each is shown.
Table 1 Copper affinity for stefin proteins as determined by ITC.
Values shown for affinity are those from the best fit of two sites.
nsd ¼ no site detected Values are in units of M )1 and are the
averages of three measurements The standard error was less than
5% of the value for each measurement Shown are the two values
for a sequential two site fit Fitting for one or three sites resulted in
two values at least two orders of magnitude higher.
Binding
Stefin B (Variant 2)
Stefin B (P79Sb)
pH 7
7.6 · 10 9
–
pH 5
1.8 · 10 8
nsd
Trang 5that the protein aggregates heavily in the presence of
Cu2+ at pH 7 and to a lower extent at pH 5
There-fore, no comparison of the stability of the tertiary
structure to the effect of Cu2+could be made
Influence of copper on oligomerization,
aggregation and amyloid fibril formation
First, we probed the effect of copper binding on
oligomer formation The wildtype stefin B has
well-defined oligomers, which can be separated by size
exclusion chromatography (SEC) The isolated
mono-mer was incubated at pH 7 or at pH 5, in the presence
or absence of Cu2+, and after longer times (1 week at
room temperature and 4 days at 36C, respectively),
samples were taken for the SEC analysis The results
can be described as follows: at a lower temperature no
difference in the ratio between the oligomers is seen,
whereas after incubation at 36C there is a marked
shift from the monomer towards the dimer (and some
tetramer) at pH 7 At pH 5 the protein undergoes
complete dimerization even with no Cu2+ present A
conclusion can be made that copper binding facilitates
dimerization already at neutral pH (which would be expected to promote further oligomerization⁄ aggrega-tion)
It has been shown previously that 10% 2,2,2 trifluor-ethanol (TFE) is the optimal concentration needed to accelerate fibril growth by stefin B at pH 5 [5,6] In the present series of experiments we compared fibrillation
of stefin B wildtype (E31 isoform), of stefin B variant
2 (Y31 isoform) and the mutant P79S of the variant, all at pH 5.0 with 10% TFE, 25C, in the presence of
Cu2+or with no Cu2+present Two concentrations of CuSO4 in the buffer were used (50 and 150 lm) giving
1 : 1 and 1 : 3 protein to Cu2+ratios
Figure 6 and Table 2 show the outcome of the fibril-lation assays with and without Cu2+ present in the medium Fibrillation of stefin B wildtype (E31 isoform) was first recorded at 40C, at pH 7, where no fibrilla-tion was observed, and at pH 5 at 40C (Fig 6A) The thioflavin T (ThT) intensity increased to some extent under these latter conditions, reflecting fibril growth, but much less than when TFE was added (Fig 6B) Fibrillation of all the three proteins (stefin B wildtype, variant 2 and the P79S mutant) at the stand-ard assay conditions (pH 5, 10% TFE, 25C), are plotted in Fig 6B–D It can be seen that Cu2+ inhib-ited fibril growth in all cases: with stefin B wildtype (Fig 6B), with stefin B variant 2 (Fig 6C) and even with the P79S mutant (Fig 6D) A very similar overall picture was obtained with three-fold Cu2+ excess (not shown)
The results were normalized in such a way that the maximal value of ThT fluorescence intensity was taken
as 100% From these, for each reading of ThT fluores-cence the percentage of inhibition of the fibril growth was obtained The percentage of inhibition (Table 2) is correlated with Cu2+ concentration, and is higher at
1 : 3 protein to Cu2+ molar ratio than at 1 : 1 The P79S mutant, which does not bind Cu2+ and is tetra-meric, also does not fibrillate to the same extent as wildtype or variant 2, but some inhibition by Cu2+ is still observed (Fig 6D, Table 2) Regardless, this does not seem to depend on Cu2+concentration
Transmission electron microscopy (TEM) data were collected at 9000 min of fibrillation, which is the time where maximal change in ThT fluorescence occurs (Fig 6B,C) The TEM images (Fig 7) confirm that a three-fold molar excess of Cu2+ (Fig 7B,D) markedly reduces the amount of fibrils in comparison to granular aggregate in stefin B and in the variant In comparison with earlier studies it seems that even aggregation is inhibited and not only fibril formation This will be discussed in view of a proposed higher toxicity of the aggregates in comparison to the mature fibrils
far UV CD spectra
-8000
-6000
-4000
-2000
0
2000
4000
6000
8000
nm
stefin A stefin B P79S
near UV CD spectra
-40
-20
0
20
40
60
80
100
120
nm
stefin A stefin B P79S
B
A
Fig 4 Circular dichrosim spectra of stefin B and of stefin A (A) Far
UV CD spectra of stefin A, stefin B and the P79S mutant (B) Near
UV CD spectra of stefin A, stefin B and the P79S mutant.
Trang 6Cystatins and neurodegenerative disease
Cystatin C is a well-known amyloidogenic protein The
L68Q variant is associated with a hereditary form
of cerebral amyloid angiopathy that results in a
fatal brain hemorrhage [32] Wildtype cystatin C has
been found as a component of amyloid plaques in Alzheimer’s disease [26] and shown to inhibit amyloid fibril formation of amyloid-beta [33] In searching the literature we found a report of stefins A and B together with some cathepsins being found in the core
of senile plaques of different disease origin [27] Cystatins have been found to be important in neuro-degeneration and in neuroregeneration Cystatin C
-8000
-6000
-4000
-2000
0
2000
4000
6000
8000
nm
wt stB with Cu
wt stB without
-6000 -4000 -2000 0 2000 4000 6000 8000 10000 12000
nm
var2 stB with Cu var2 stB without
-3000
-2000
-1000
0
1000
2000
3000
4000
5000
6000
nm
P79S with Cu P79S without
C
-8000 -6000 -4000 -2000 0 2000 4000 6000
nm
stA with Cu stA without
D
-2
-1
0
1
2
3
4
5
nm
pH7 pH7Cu
E
-2 -1 0 1 2 3 4 5
nm
pH5 pH5Cu
F
Fig 5 Circular dichroism spectroscopy as a function of Cu2+concentration (A) Far UV CD spectra of stefin B wildtype (E31 isoform) in pres-ence of Cu 2+ and without Cu 2+ Measurements in the far UV were carried out at 25 C at pH 7.3 (NaCl ⁄ P i buffer) (1 mm rectangular cell, bandwidth 1 nm, each 1 nm for 5 s) (B) Far UV CD spectra of stefin B variant 2 (Y31 isoform) in the presence and without Cu 2+ (C) Far UV
CD spectra of the P79S mutant in the presence and without Cu2+ (D) Far UV CD spectra of stefin A in the presence and without Cu2+ (E) Near UV CD spectra of stefin B wildtype in the presence of Cu 2+ and without Cu 2+ at pH 7 Measurements in the near UV were collected
at 20 C using a 10 mm rectangular microcell, bandwidth 0.5 nm, collecting data each 0.5 nm for 3 s (F) Near UV CD spectra of stefin B wildtype in the presence of Cu2+and without Cu2+at pH 5.
Trang 7was reported to modulate neurodegeneration and
neurogenesis following status epilepticus in mouse [25]
Mutations in human stefin B (cystatin B gene; CSTB)
were identified as a cause of the progressive myoclonus
epilepsy of the Unverricht–Lundborg type In studies
of CSTB-deficient mice, lack of this inhibitor was found to be associated with signs of cerebellar granular cell apoptosis [20] The mice develop progressive ataxia and myoclonic seizures and undergo an extensive loss
of Purkinje cells They provide a reasonably good model for the disease The transcripts that were consis-tently increased in brain tissue from CSTB-deficient mice encode proteins involved in responding to neuronal damage [21], i.e., genes which code for increased proteolysis, apoptosis and glial cell activation Copper homeostasis is important in the brain, there-fore the role of copper binding or loss of its binding could be related to specific cerebellar function(s) of stefin B [18], which remains to be seen by more in vivo studies
Stefin B as a copper binding protein
We have demonstrated that human stefin B is a high affinity copper binding protein It exerts two high affinity biding sites in the picomolar range at pH 7
pH=5, 40 o C
-200
0
200
400
600
800
1000
Time (min)
wt stB without
wt stB with Cu
pH=5, 25 o C, 10%TFE
-200 0 200 400 600 800 1000
Time (min) ThT fluorescence / 480 nm ThT fluorescence / 480 nm
wt stB without
wt stB with Cu
B A
ThT fluorescence / 480 nm ThT fluorescence / 480 nm
pH=5, 25 o C, 10% TFE
-200
0
200
400
600
800
1000
Time (min)
var2 stB without var2 stB with Cu
pH=5, 25 o C, 10% TFE
-200 0 200 400 600 800 1000
Time (min)
mut P79S without mut P79S with Cu
Fig 6 Inhibition of fibrillation of stefin B by Cu 2+ as probed by ThT fluorescence For experimmental detail see Experimental procedures Final protein concentration was in all cases 45 l M and final concentration of Cu2+46 l M , leading to 1 : 1 of protein to Cu2+ratio Results for 1 : 3 pro-tein to Cu 2+ ratio have also been obtained (not shown) (A) Stefin B wildtype (E31 isoform) at pH 5, 40 C, 0 and 50 l M Cu 2+ in the buffer (B) Stefin B wildtype (E31 isoform) at pH 5, 10% TFE, 25 C, 0 and 50 l M Cu 2+ in the buffer (C) Stefin B variant 2 (Y31 isoform) at pH 5, 10% TFE, 25 C, 0 and 50 l M Cu2+in the buffer (D) P79S mutant of variant 2 at pH 5, 10% TFE, 25oC, 0 and 50 l M Cu2+in the buffer.
Table 2 Inhibition of fibrillation of stefin B proteins by Cu 2+
Con-centration of the protein was normally 45 l M while final
concentra-tions of Cu2+in solution were 46 l M and 138 l M , which gives 1 : 1
and 1 : 3 protein to Cu 2+ molar ratios, respectively.
Protein ⁄ variant [Cu 2+ ] (l M ) Solvent composition % of inhibition
Trang 8The affinity for these sites is decreased with decreased
pH (Figs 2 and 3, Table 1) In comparison, human
stefin A, which has the same 3D structure, does not
bind copper Although the structure of the two
pro-teins is almost identical, the sequences differ in a
num-ber of places, but in particular, stefin B has a numnum-ber
of histidines in the C-terminus (box in Fig 1) at sites:
92, 75, 66 and 58 As histidine residues are central to
copper binding in many proteins they probably form
part of the copper binding sites in this protein
Although there are four histidines in the C-terminal,
another histidine is present at position 18, and given the folded state of the protein it is possible that this residue could play a role in copper binding No His residues are located at the homologous sites in human cystatin C (sequences were aligned), suggesting that copper binding might be specific to stefin B among the three human cystatins
Further support for the premise that the C-terminus could be the copper binding domain comes from stud-ies of the P79S mutant, which differs from the copper binding forms of stefin B by one amino acid residue
Fig 7 Inhibition of amyloid fibril growth as observed by TEM Samples at 34 l M protein concentration were incubated at 25 C in the stand-ard fibrillation buffer and after 9000 min (the plateau level of the reaction) they were prepared for TEM measurement (A) Stefin B variant 2 (Y31 isoform) prepared in chelated buffer pH 5, 10% TFE (B) Stefin B variant 2 (Y31 isoform) in the same buffer with 100 l M final Cu2+ (C) Stefin B wildtype (E31 isoform) prepared in chelated buffer pH 5, 10% TFE (D) Stefin B wildtype (E31 isoform) in the same buffer with
100 l M final Cu 2+ concentration.
Trang 9change but lacks any copper binding capacity This
point mutation lies within the C-terminal region and
therefore could affect the structure of this part of the
protein However, it should not be dismissed that the
P79S mutant differs from the other stefin variants in
that it forms a tetramer Analysis of the far UV CD
spectra (Fig 4A) by dichroweb [29,30] suggests only
negligible change in the secondary structure between
the wildtype stefin B or variant stefin B and the P79S
mutant of the variant, consistent with tetramerization
Near UV CD spectra of stefin B and the P79S mutant
are also very similar (Fig 4B), consistent with proper
folding of the tetramer comparable to the wildtype
protein It is possible, however, that a new interface
formed in the tetramer would disrupt copper binding
Inhibition of amyloid fibril formation by stefin B
in presence of copper
The mechanism of amyloid fibril formation of cystatins
is being studied [5,6] It is proposed that domain
swapping is followed by tetramerization and further
oligomer formation [16,34], which accumulate into the
so-called ‘critical oligomers’ [35] and then grow into
protofibrils and mature fibrils Therefore, inhibition by
Cu2+ of amyloid fibril formation of human stefin B
could result from loss of correct Cu2+ coordination
before the stage of tetramerization Possibly, loss of
copper binding could still allow domain swapping to
occur It is of interest that a N-terminally truncated
pri-on protein, lacking the copper binding domain is
cap-able of domain swapping and forms a dimer as revealed
by crystal structure analysis [36] In our case, SEC data
collected for samples at pH 7 have confirmed that
cop-per binding shifts the equilibrium towards the dimer
and that the monomer remains monomeric in its
absence At pH 5, the protein is dimeric even with no
Cu2+bound
It has been shown that stefin B is very much prone
to undergo oligomerization and amyloid fibril
forma-tion [5,6,31] Therefore, we probed the effect of Cu2+
(loss of copper binding) on fibril formation of this
pro-tein Our results show that upon copper binding
amy-loid fibrillation gets less (this is judged by ThT
fluorescence intensity and TEM; see below) In
partic-ular, our data shows that copper binding inhibits
amy-loid fibril growth of the two stefin B isoforms but does
not affect the P79S mutant, which is purely tetrameric
This seems to suggest that Cu2+ could stabilize the
protein and thus inhibit amyloidogenesis before the
point of tetramerization This would suggest that
initial aggregation is critically dependent on domain
swapping, which is a step prior to tetramerization
[16,34] However, delay in tetramer formation could still lead to granular aggregate formation (Fig 7B,D)
It has been shown that prefibrillar oligomers may be more toxic than the fibrils themselves [37,38] In the case of stefin B (Fig 7) it seems that not only fibrilla-tion is diminished but also the amount of granular aggregate (Fig 7B,D) This is judged from our previ-ous observations of the lag phase granular aggregate obtained at the same protein concentration [5,39] Amyloid fibril formation of the N-terminal fragment
of stefin B up to residue 68, as observed in some patients with Unverricht–Lundborg type 1 progressive myoclonus epilepsy has shown an increased amyloido-genic potential, as reported by Rabzelj et al [39] Not-withstanding problems with folding of the fragment [39], which stays unfolded, this seems to support our premise that loss of copper binding (residues 92 and
75 are lost) contributes to the progress of amyloido-genesis
Effect of copper binding on amyloid formation
of other proteins
A number of other amyloidogenic proteins have been shown to be copper binding proteins Copper binding
is sometimes specific and at other times nonspecific In particular, copper like other redox active metals has been shown to promote aggregation or polymerization
in a number of cases Although the prion protein binds copper in its native conformation [40], the presence of copper has also been shown to accelerate aggregation
of the protein [41] or increase the infectivity of prion isolates However, this kind of interaction is nonspe-cific Other studies have suggested that specific binding
to the prion protein stabilizes its structure and pre-vents intermediate, partially unfolded states that could result in a major conformational change [42] Loss of appropriate metal binding and substitution with a dif-ferent metal could initiate such conformational chan-ges in the prion protein [43] Recently, it has been shown that copper binding to alpha synuclein causes aggregation and fibril formation [44] It is well known that alpha synuclein is one of the natively unfolded proteins Copper also binds to the amyloid precursor protein and to the cleavage product amyloid-beta [45,46] However, in this case, binding of copper to the amyloid precursor protein is thought to prevent clea-vage by beta-secretase [47] while interaction between copper and amyloid-beta initiates polymerization [48] Recent studies have shown that Cu2+and Zn2+ bind-ing to amyloid-beta (1–40) peptide, in distinction to
Fe3+, retards amyloid fibril formation [4] There were also reports that prefibrillar aggregation was promoted
Trang 10by these ions [41] which would not be beneficial in the
light of higher toxicity of such aggregates [3]
The discovery that copper binding to stefin B
inhib-its fibril formation but does not prevent aggregation to
prefibrillar oligomers is quite significant Both these
facts are in accordance with amyloid-beta [4] and prion
studies [41], respectively We propose that in globular
proteins which bind Cu2+specifically, the metal
bind-ing can be protective against amyloid fibril formation
Therefore, maintaining correct metal ion protein
inter-actions might be key to whether such proteins are able
to enter an amyloidogenic pathway However, copper
binding, most likely nonspecific, does not always
pre-vent prefibrillar aggregate formation, which may be
even more toxic [3]
Recently, Miranker and coworkers [49] indicated
that b-2 microglobulin aggregated more heavily in the
presence of Cu2+ (but not Ni2+) They have shown
that, due to high affinity copper binding to a
conform-ationally changed monomer M*, equilibrium is shifted
to more oligomers Thus, in their case, specific copper
binding to oligomers accelerated amyloid aggregation
(measured by ThT fluorescence) In our case, the
monomer and dimer seem to bind Cu2+, whereas the
tetramer looses this ability, which makes it logical that
the fibrillation (which starts with oligomer formation)
would be inhibited
Conclusion
We show two per se interesting and novel facts: (a)
that human stefin B (cystatin B) is a high affinity
copper binding protein, and (b) that amyloid fibril
formation of this protein is diminished in presence
of Cu2+ ions Another interesting finding is that
copper binding gets less strong at pH 5, under fibril
promoting conditions and that a mutant, which was
shown to be tetrameric, does not bind copper
Per-haps all these facts are not related and have no
rele-vance to in vivo function of the protein and even
less to amyloid fibrillation As for the function
[10,11,18], this is open to more research The protein
decreases apoptosis not only by protease inhibition,
as shown by gene knockout studies [20,50] As
apop-tosis is highly connected to either oxidative stress
and⁄ or protein aggregation, alternative function
(mis-function) of this protein could be researched in those
directions
A broader implication for future research is that
understanding what causes the loss of appropriate
metal binding might be crucial for the understanding
of the role of amyloidogenic proteins in a number of
neurodegenerative disorders
Experimental procedures
Materials 2,2,2 Trifluorethanol was from Fluka (Buchs, Switzerland) and thioflavin T from Aldrich (St Louis, MO, USA) Other chemicals were from Sigma (St Louis, MO, USA), Carlo Erba (Milano, Italy), Serva (Westbury, NY, USA) and Merck (Darmstadt, Germany)
Recombinant proteins Recombinant human stefin B variants were produced in Escherichia coliand isolated as described [51,52]
Isothermal titration calorimetry measurements All measurements were made on a Microcal VP-Isothermal Titration Calorimeter instrument as previously described [53] Briefly, a time course of injections of a ligand to a macromolecule or vice versa were made in an enclosed reaction cell maintained at a constant temperature The instrument measured the heat generated or absorbed as the ligand-macromolecule reaction occured A binding isotherm was fitted to the data, expressed in terms of the heat change per mole of ligand against the ligand to macromolecule ratio From the binding isotherm values for the reaction stoichiometry, association constants Ka, the change in enthalpies H and change in entropies S were obtained All solutions were filtered through a 0.22 lm filter and degassed prior to use All measurements were made in a buffer consisting of 5 mm Mes at either pH 5 or pH 7 Solutions were treated with the chelex medium to remove trace metals, according to the manufacturer’s instructions (Sigma)
Direct titration of protein solutions with aqueous copper salts was avoided because of the nonphysiological nature of such interactions; these may be avoided by the use of a copper chelate Copper(II) forms a bis glycine complex, Cu(Gly)2, in the presence of excess glycine Therefore a copper⁄ glycine ratio of 1 : 4 was used by dissolving 3.0 mm copper(II)chloride and 12 mm glycine in chelex treated water The excess glycine ensured that titrated copper was either chelated to glycine or incorporated into the protein, avoiding aqueous copper in the reaction cell It also acted
as a competitor to nonspecific protein copper interactions
A control ITC experiment of titrating stefin B protein with glycine alone was performed and no binding was observed
In our hands, as others [54], the most reproducible data was obtained from injections of the metal into a solution of the protein Typically an initial injection of 2 lL copper chelate solution was followed by a further 29 injections of
4 lL of Cu(II) into the protein in the sample cell stirred
at 300 r.p.m Injections were separated by 120 s to allow equilibration and sample temperature was maintained at