Tài liệu Báo cáo Y học: BIGH3 (TGFBI) Arg124 mutations influence the amyloid conversion of related peptides in vitro Implications in the BIGH3-linked corneal dystrophies pptx

BIGH3 (TGFBI) Arg124 mutations influence the amyloid conversion
of related peptides
in vitro
Implications in the BIGH3-linked corneal dystrophies
Clair-Florent Schmitt-Bernard
1,2
, Alain Chavanieu
3
, Gudrun Herrada
3
, Guy Subra
3
, Bernard Arnaud
4
,
Jacques G. Demaille
1
, Bernard Calas
3
and A
´
ngel Argile
´
s
1
1
Institut de Ge
´
ne
´
tique Humaine, CNRS UPR 1142, Montpellier, France;

2
Antigone Ophtalmologie, Montpellier, France;
3
Centre de
Biochimie Structurale, CNRS UMR 5048, Universite
´
Montpellier, Montpellier, France;
4
Service d’Ophtalmologie, CMC Gui de
Chauliac, Montpellier, France
Amyloid deposits with Arg124 mutated TGFBI protein
have been identified in autosomal dominant blinding corneal
dystrophies. We assessed in vitro the mechanisms determin-
ing TGFBI protein amyloid transformation involving
mutations of Arg124. Eight peptides synthesized following
the TGFBI protein sequence, centered on codon Arg124
holding the previously reported amyloidogenic mutations
and the respective controls were studied. Cys124 and His124
mutated peptide preparations contained significantly higher
amounts of amyloid than the native peptide. Blocking the
SH group of Cys124 and deleting the first four NH
2
-terminal
amino acids including Val112-Val113 resulted in a decrease
in amyloid fibril formation while deletion of the nine
CONH
2
-terminal residues increased amyloid fibril concen-
tration. Fourrier transformed-infrared spectroscopy analysis
of the different peptide solutions showed an increase in

b-pleated sheet structures in those with enhanced amyloid
yielding. We designed a peptide (BB1) likely to counteract
the role of Val112-Val113 in amyloid fibril formation.
Incubation of Cys124 peptide with BB1 indeed resulted in a
35% inhibition of amyloid fibril formation.
Our results are in keeping with the clinical observations of
Arg124 mutation-linked amyloidosis and show the import-
ance of Val112–Val113, disulfide and hydrogen bonding in
increasing the b-pleated conformation and amyloid forma-
tion. These findings shed new light on the molecular mech-
anisms of TGFBI protein amyloidogenesis and encourage
further research on the use of specifically designed peptides
as putative therapeutic agents for these disabling diseases.
Keywords: amyloidosis; keratoepithelin; lattice corneal dys-
trophy; granular corneal dystrophy; synthetic peptide.
Hereditary corneal dystrophies are a cause of blindness.
These dystrophies are characterized by a progressive
alteration of the particular structure of the cornea resulting
in loss of its transparency. Based upon their clinical
characteristics, hereditary corneal dystrophies form a
distinctive group of corneal diseases. Some of them involve
the corneal stroma where deposits begin to appear
during the first decades of life and severely impair visual
acuity in adulthood. Their therapy is restricted to keratopl-
asty and phototherapeutic keratectomy by Excimer laser.
Unfortunately, the benefits of these therapies remain
transient as recurrence of the deposits is the rule.
Genetic studies have recently confirmed that a group of
hereditary corneal dystrophies have a common molecular
mechanism: the involvement of the BIGH3 (TGFBI,

transforming growth factor b-induced) gene [1]. Specific
BIGH3 mutations have been linked to particular forms of
the disease in this group of dystrophies. Autosomal
dominant BIGH3-linked corneal dystrophies may present
amyloid deposits, granular deposits or a mixture of both
(granular and amyloid).
The BIGH3 gene encodes for a 683 amino-acid protein
inducible by TGFb, the TGFBI protein also known as
big-h3. It is a prominent constituent of the cornea, skin, and
matrix of many connective tissues [2]. It is a secreted protein
with an amino-terminal secretory sequence, a carboxy-
terminal Arg-Gly-Asp sequence and four homologous
domains of 140 amino acids [2]. The TGFBI protein, as
other homologous proteins, may interfere in the cell
adhesion process. The Arg-Gly-Asp sequence is known to
act as a ligand recognition site for integrins. The particular
integrins with binding capacities for the corneal TGFBI
protein remain to be fully identified. Kim et al. [3] have
recently shown that alpha3-beta1 integrins bind to TGFBI.
Two major sites for mutation have been recognized in the
BIGH3 gene as inducing four distinct hereditary corneal
dystrophies. These mutation sites are located at codon
Arg124 and codon Arg555 [4]. Other mutations in the
BIGH3 gene have been occasionally reported [5–11].
Mutations in Arg555 are responsible for corneal dystrophy
of Bowman’s layer type 2 (CDB2, Thiel-Behnke corneal
Correspondence to C F. Schmitt-Bernard, IGH CNRS UPR 1142,
141, rue de la Cardonille, F-34396 Montpellier cedex 5, France.
Fax: + 33 4 67 42 39 73, Tel.: + 33 4 67 42 09 83,
E-mail:

Abbreviations:BIGH3,beta-inducedgene-human3;big-h3, BIGH3
gene product; TGFBI, transforming growth factor beta-induced gene;
TGFb, transforming growth factor beta; LCD, lattice corneal dys-
trophy; CDB, corneal dystrophy of Bowman’s layer; GCD, granular
corneal dystrophy; ThT, thioflavin T.
(Received 5 March 2002, revised 6 August 2002,
accepted 23 August 2002)
Eur. J. Biochem. 269, 5149–5156 (2002) Ó FEBS 2002 doi:10.1046/j.1432-1033.2002.03205.x
dystrophy, mutation Arg555Gln) and Granular corneal
dystrophy type 1 (GCD1, mutation Arg555Trp). Codon
Arg124 seems to be particularly critical in corneal dystro-
phies as four different phenotypes are associated with four
different mutations of this residue. These mutations are:
Arg124Cys (R124C) [1] in lattice corneal dystrophy type 1
(LCD1) characterized by amyloid deposits, Arg124Ser
(R124S) [12] in a phenotypic variant of granular corneal
dystrophy type 1 (GCD1), Arg124His (R124H) [1] in
granular corneal dystrophy type 2 (GCD2, Avellino
dystrophy) a mixed type of amyloid and granular deposits,
and Arg124Leu (R124L) [13] in corneal dystrophy of
Bowman’s layer type 1 (CDB1, GCD3, Reis-Bu
¨
cklers
corneal dystrophy) a phenotypic variant of GCD1 charac-
terized by superficial granular deposits. The biochemical
mechanisms responsible for the alteration of protein behav-
ior following mutations at codon Arg124 remain unknown.
We have recently described an effective in vitro system to
produce amyloid fibrils from TGFBI protein 110)131
derived peptides [14]. In the present report, we used our

in vitro system to elucidate the mechanisms involved in
amyloidogenesis. We analyzed the in vitro behavior of
several peptides holding the TGFBI protein codon 124,
whose sequences were selected according to previous genetic
reports of corneal dystrophies linked to TGFBI protein
Arg124 mutations. Following our results, that are in
accordance with the clinical observations of Arg124 muta-
tion linked amyloidosis, we designed a short peptide (BB1)
with predicted capacity to inhibit amyloid fibril formation
and tested its inhibitory effect on amyloid fibril formation
from codon 124 mutated peptides in vitro.
EXPERIMENTAL PROCEDURES
Synthesis and purification of peptides
Eight peptides derived from the TGFBI protein were
synthesized (Table 1): four 22 amino-acid long peptides
comprising codon 124 in its native (R110)131) and mutated
forms (C110)131, H110)131, S110)131); two 18 amino-
acid long peptides (R114)131, C114)131) missing amino
acids 110–113 (Leu-Gly-Val-Val); one 13 amino-acid long
peptide (big-h3 110)122) excluding codons 123 and 124 at
the CONH
2
-terminal end; and one peptide homologous to
C110)131 in which the SH group was blocked by using
Fmoc-L-cystein acetamidomethyl (Acm) to prevent disul-
fide bonding (C110)131Acm). The mutated peptides were
synthesized following previous reports demonstrating the
presence of C, H, and S in the corresponding clinical
forms of corneal dystrophies (LCD1, GCD2 and GCD1,
respectively).

Chemicals
Trifluoroacetic acid and acetonitrile (HPLC grade) were
purchased from SDS (Peypin, France). All compounds
used for solid-phase peptide synthesis (solvents, resins
and protected amino acids) were from PE Biosystem
(Framingham, USA).
Peptide synthesis
Peptide synthesis was carried out at a 0.2-mmol scale using a
continuous flow apparatus (PE Biosystem, Pioneer,
Framingham, USA) starting from Fmoc-PAL-PEG-PS
resins. The coupling reaction was performed with 0.5
M
of
HATU in presence of 1
M
of DIEA. Protected group
removal and final cleavage were carried out with trifluoro-
acetic acid/H
2
O/EDT/phenol [14] and crude peptides were
purified by reverse-phase HPLC on a C18 semipreparative
column. Electrospray ionization mass spectra were in
complete agreement with the proposed structure.
In vitro
processing of the peptides
The peptides were processed as previously described [14].
Briefly, 0.1 lmolofeachpeptidewasdilutedin250lLof
1/15
M
phosphate-buffered solution (pH 7.4). The solutions

were placed in the dialysis well of a Fast Dialyzer
Interbiotech (Interchim S.A., Montluc¸ on, France) mounted
with a dialysis membrane of 1000 Dalton cut-off. The
sample solutions were dialyzed at 4 °C for 72 h against 1/15
M
phosphate-buffered solution (pH 7.4). The dialysate was
exchangedevery24hwithafreshbufferedsolution.
Spontaneous fibril formation was assessed by the dilution
of 0.5 lmol of each peptide in 50 lL of distilled water and
studied on day 1, day 3, day 7, and day 15.
Congo red stain
Pellets obtained by centrifugation of the protein solution at
10
5
g,for1hat4°C were resuspended in 40 lL of distilled
water. The samples were spread onto glass microscope slides
with gelatin. The smears were air-dried, fixed with 95%
ethanol for 5 min and stained with Congo red (Sigma-
Aldrich Chemicals, Inc). The presence of amyloid fibrils was
confirmed by viewing the typical apple-green birefringence
under plane polarized light using a Zeiss microscope (Carl
Zeiss, Go
¨
ttingen, Germany).
Thioflavine T fluorescence analysis
The studies were performed essentially as described in
previous reports [16]. This technique has been proved to be
adequate in amyloid quantification in other in vitro systems
[17]. Thioflavine T (ThT) was purchased from Sigma-
Aldrich Chemicals, Inc. (St Quentin Fallavier, France). The

peptide solution was dialyzed against 1/15
M
phosphate-
buffered solution at pH 7.4 and ultracentrifugated at 10
5
g,
for 1 h, at 4 °C. The pellets were resuspended in 50 m
M
glycine-NaOH pH 8.5 containing 100 l
M
ThT in an
assay volume of 500 lL, and processed immediately.
Table 1. Synthetic peptides designed for the study.
Peptide H
2
N-sequence-CONH
2
R110)131 LGVVGSTTTQLYTDRTEKLRPE
R114)131 GSTTTQLYTDRTEKLRPE
C110)131 LGVVGSTTTQLYTDCTEKLRPE
C114)131 GSTTTQLYTDCTEKLRPE
C110)131Acm LGVVGSTTTQLYTDCTEKLRPE
Acm
H110)131 LGVVGSTTTQLYTDHTEKLRPE
S110)131 LGVVGSTTTQLYTDSTEKLRPE
big-h3 110)122 LGVVGSTTTQLYT
5150 C F. Schmitt-Bernard et al. (Eur. J. Biochem. 269) Ó FEBS 2002
Fluorescence spectroscopy was performed on a Quanta-
Master System Spectrofluophotometer (Photon Technology
International, Monmouth Junction, NJ, USA) at 25 °Cas

described by Naiki et al. [16] with excitation at lambda-
ex ¼ 450 nm and emission spectra at lambda-em ¼
482 nm.
Electron transmission microscopy
The samples were applied to a formvar carbon-film-coated
copper grid and then negatively stained with 1% uranyl
acetate for 60 s. The specimens were viewed on a Hitashi
H-7000 electron microscope (Hitashi LTD, Tokyo, Japan)
with an acceleration voltage of 75 kV.
Fourier-Transform Infrared Spectroscopy (FT-IR)
The secondary structure of the peptides was studied with
FT-IR spectroscopy of the amide I region performed in
both H
2
OpH7.4andD
2
Oat25°C immediately after
their dilution, and 24 h after their suspension in H
2
O
[18,19]. D
2
O was used to prevent intermolecular and
intramolecular hydrogen bond formation. Infrared spec-
tra were collected on a Perkin-Elmer spectrum-one IR
spectroscope (PE Applied Biosystems, Foster City, CA,
USA). Absorbance was plotted against the wave number.
The spectrograms were Fourier-deconvoluted and the
secondary structure was determined by Gaussian curve-
fitting. The calculation of each fraction of the total band

area over the curve was performed with an overlap
method after baseline correction. Anti-parallel b-pleated
structures were quantified at 1625 cm
)1
wave number
and antiparallel b-aggregates were determined at
1685 cm
)1
.
Beta-breaker methods
Two peptides, five amino-acid long, were synthesized as
described above. Their sequences were decided on the
grounds of the results of the amyloid fibril experiments. BB1
was designed to interfere with the Val112–Val113 of the
C110)131 peptide and consisted of H
2
N-LPVVD-CONH
2
.
An unspecific peptide BB2 (H
2
N-LPFFD-CONH
2
)was
used as control for comparison to evaluate the effect of BB1.
1 l
M
of each peptide was incubated with 0.05 l
M
of

C110)131 in an assay volume of 50 lL, at 37 °C. Separate
experiments were performed and independently processed
on day 1, day 3, day 7, and day 14 by Thioflavine T
fluorescence analysis in order to determine the amount of
amyloid fibril formation.
RESULTS
Role of amino-acid 124
Congo red stained smears of peptides R110)131,
C110)131, H110)131, and S110)131 dialyzed against a
phosphate-buffered solution 1/15
M
pH 7.4 displayed
significant amounts of birefringent material. Transmission
electron microscopy showed the fibrillar pattern of this
material confirming the amyloid nature. The yield of
amyloid was clearly different depending on the considered
peptide (Fig. 1). Large quantities of amyloid were found
with C110)131 and H110)131 peptides, while amyloid was
scarcer with S110)131, and R110)131.
The amount of amyloid was quantitatively determined
using ThT spectrofluorometry as it can be expressed as the
fluorescence emission spectra at lambda-em ¼ 482 nm
following excitation at lambda-ex ¼ 450 nm in comparison
Fig. 1. Dialysis-based amyloid fibril formation. Left hand side show the
Congo red staining and right hand-side panels show polarized light
microscopy of the same samples. The slides included correspond to (A)
material obtained from dialysis of the C110–131 peptide (B) material
obtained from dialysis of the H110–131 peptide (C) material obtained
from dialysis of the S110–131 peptide (D) material obtained from
dialysis of the R110–131 peptide (E) material obtained from dialysis of

the C114–131 peptide and (F) material obtained from dialysis of the
big-h3 110–122 peptide. This morphological analysis showed more
dichroic material in solutions containing big-h3 110–122 and
C110–131 peptides than in those containing H110–131, S110–131,
R110–131, and C114–131 peptides.
Ó FEBS 2002 TGFBI protein Arg124 mutations and amyloid formation (Eur. J. Biochem. 269) 5151
to the baseline, which corresponds to ThT auto-fluores-
cence. ThT fluorescence was the highest for C110)131, and
diminished significantly for each of the following peptides
H110)131, S110)131, and R110)131, respectively (Fig. 2).
No amyloid material was observed with the same
peptides suspended in distilled water without NaCl/P
i
dialysis at 2 and 24 h. A discrete red/green dichroism was
noticeable by day 3 on smears with the C110)131 peptide,
while R110)131, H110)131, and S110)131 remained
negative. Spontaneous amyloid formation was only ob-
served with prolonged incubation periods. ThT fluorescence
showed increasing emission spectra until day 15 (Fig. 3). By
this time point, ThT emission spectra were significantly
higher for C110)131, and H110)131, compared to that of
S110)131 and R110)131. Therefore, the results of both
spontaneous and dialysis-based precipitations showed the
importance of the residue 124 in determining its tendency to
precipitate into amyloid fibrils.
Analysis of the secondary structure of both R110)131
(Fig. 4A) and C110)131 (Fig. 4B) in H
2
O revealed a switch
in the IR spectrograms from an antiparallel b-pleated sheet

(1625 cm
)1
)tob-aggregation (1685 cm
)1
) when comparing
the baseline preparations to those at 24 h. Further,
comparison of C110)131 to R110)131 showed that the
former had approximately 30% more b-sheet conformation
than R110)131 both at baseline and by 24 h.
Role of the Cys–Cys disulfide bonds
The role of disulfide bonding was examined by comparing
amyloid formation of C110)131 to that of the same peptide
with blocked sulfated radicals of Cys, the C110)131Acm
peptide. Blocking the Cys residues resulted in a 50%
decrease in ThT signal, suggesting that the increased
tendency to form amyloid fibrils observed for C110)131
peptide is, at least in part, due to disulfide bonding
(Fig. 5B). Blocking Cys–Cys disulfide bonding resulted in
a similar decrease in amyloid fibril formation than trunca-
ting the NH
2
-terminals sequence of the C110)131 peptide
(peptide C114)131) suggesting both these elements are of
equal importance for the formation of amyloid from the
C110)131 peptide.
Role of the NH
2
-terminal sequence
The role of the NH
2

-terminal sequence of the 124 centered
peptides included in this study was analyzed by comparing
the 110–131 peptides to the 114–131 counterparts. The ThT
emission spectrum from R110)131 was significantly higher
than that of the same peptide without 110Leu-Gly-Val-
Val113 (R114)131) (Fig. 5A). A similar decrease in amyloid
fibril formation was observed when comparing C110)131
to its corresponding peptide lacking the NH
2
-terminal
110Leu-Gly-Val-Val113 (C114)131) (these results are illus-
trated morphologically in Fig. 1, and quantitatively in
Fig. 5B).
On FT-IR spectroscopy, C114)131 appeared to have less
b-sheet component than C110)131 (6% and 23%, respect-
ively) confirming that the NH
2
-terminal sequence of
C110)131 participates in the b-conformation of the peptide
(Figs 4D and 6). These results were observed both in D
2
O
as well as in H
2
O.
Role of the CONH
2
-terminal sequence
The participation of the CONH
2

-terminal sequence of the
TGFBI protein derived peptides on amyloid precipitation
was assessed by comparing the 22 amino-acid-long peptide
centered on codon 124 (big-h3 110)131) to its counterpart
missing residues 123–131 (big-h3 110)122). The amount of
amyloid material obtained with big-h3 110)122 peptide was
significantly higher than that obtained with the full length
big-h3 110)131 peptide. This was clearly supported by
morphological studies (Fig. 1E) and confirmed by quanti-
tative analysis (Fig. 2). These differences were observed
from dialysis as well as from spontaneous amyloid fibril
formation studies, and were confirmed by FT-IR spectros-
copy studies (Figs 4C and 6). big-h3 110)122 displayed a
Fig. 2. ThT fluorescence profiles of synthetic peptides in solution after
dialysis. The excitation wavelength was fixed at lambda-ex ¼ 450 nm.
ThT fluorescence was maximal for a lambda-em wavelength of 482 nm
and this value was used as the reference for the quantitative meas-
urements of the amount of amyloid material contained in each peptide
solution. ThT alone was included as for the auto-fluorescence control.
big-h3 110)122 and C110)131 produced the most amyloid material.
Lesser amounts of amyloid were obtained with H110)131, S110)131,
and R110)131 solutions, respectively.
Fig. 3. ThT fluorescence kinetic analysis of spontaneous amyloid preci-
pitation. big-h3 110)122 appears to have the highest amyloidogenic
properties when compared withthe big-h3 110)131 peptides. C110)131
and H110)131 show significant amyloid fibril formation from day 1 to
day 15 in contrast to S110)131 and R110)131. Log scale Y axis.
5152 C F. Schmitt-Bernard et al. (Eur. J. Biochem. 269) Ó FEBS 2002
higher beta-content on FT-IR spectrograms than its full-
length counterpart, both in D

2
OandH
2
O.
Kinetic studies also showed a higher propensity of big-
h3 110)122 to form amyloid fibrils when compared to the
full-length peptide (big-h3 110)131). Accelerated amyloid
fibril formation was observed in experiments with spon-
taneous amyloid precipitation, without dialysis. None of
the TGFBI protein110)131 peptides analyzed (C110)131,
H110)131, S110)131 and R110)131) displayed any
significant amount of amyloid fibrils by 24 h. By contrast,
big-h3 110)122 peptide displayed a faint red/green biref-
ringence on Congo red smears by 2 h, and by 24 h a
large amount of dichroic material was observed (Fig. 7A).
The amyloid nature of the big-h3 110)122 deposits was
confirmed by electron microscopy showing bundles of
8–10 nm wide filaments (Fig. 7B).
Fig. 4. Secondary structure determination of the peptides by FT-IR spectroscopy in H
2
O. The peptides included are the following: R110)131 (A),
C110)131 (B), big-h3 110)122 (C) as well C110)131 and C114)131 (D). Anti-parallel b-pleated structures were quantified at 1625 cm
)1
wave
number and antiparallel b-aggregates were determined at 1685 cm
)1
wave number. A significant depression in the spectrogram can be observed at
1685 cm
)1
wavenumber by 24 h in R110)131 (A) and C110)131 (B) peptide solutions (arrows) demonstrating their tendency to form b-aggregates.

Calculation of the ratio between the total surface area and the surface area at 1625 cm
)1
and 1685 cm
)1
shows that C110)131 contains
approximately 30% more b-structures than R110)131 both at baseline and at 24 h indicating the higher tendency of C110)131 to adopt a b-pleated
conformation. (C) big-h3 110)122 (the peptide missing the eight residues of the CONH
2
-terminal end) displayed a strong formation of antiparallel
b-sheet conformation at 24 h confirming its high propensity to adopt a b-pleated conformation when compared to the other peptides analyzed
(arrow) (profound depression at 1625 cm
)1
wavenumber in the solid line plot corresponding to the peptide solution at 24 h). (D) Comparison of
C114)131 (missing the hydrophobic Val112–Val113) and C110)131 peptides at 24 h. It can be observed that C114)131 has significantly less
b-pleated structures than C110)131 (1625 cm
)1
and 1685 cm
)1
wavenumber differences arrows).
Fig. 5. ThT fluorescence quantitative analysis comparing big-h3 110)131 and big-h3 114)131 peptides after dialysis. R114)131 and C114)131
produce less amyloid than R110)131 (A) and C110)131 (B), respectively, proving the NH
2
-terminal sequence is involved in the amyloid conversion
of both the big-h3 110)131 peptides. A similar difference is apparent between C110)131 and C110)131 Acm suggesting that disulfide bonding also
accounts for amyloid formation.
Ó FEBS 2002 TGFBI protein Arg124 mutations and amyloid formation (Eur. J. Biochem. 269) 5153
ThT analysis confirmed the high propensity of big-
h3 110)122 to form amyloid spontaneously as it displayed
the highest ThT signal from day 1 to day 15 and had the
fastest aggregation kinetics (Fig. 3). The higher amyloi-

dogenicity of big-h3 110)122 when compared to its full-
length counterpart was also evident from dialysis-induced
amyloid fibril formation (Figs 2E and 3 illustrate this point
by Congo red stained smears and ThT fluorescence studies,
respectively).
Role of hydrogen bonds
The role of hydrogen bonds in amyloidogenesis was
analyzed by comparing FT-IR spectrograms in H
2
Oand
in D
2
O, known to prevent hydrogen bond formation. The
FT-IR spectrogram of C110)131 in D
2
O displayed signi-
ficantly less b-pleated sheet structures than the same peptide
analyzed in H
2
O (Figs 4B and 6). A similar decrease in
b-pleated sheet structures in D
2
O was observed with
R110)131 and C114)131 peptides (Fig. 6). Comparing
the spectrograms of the different peptides in D
2
O showed
that big-h3 110)122 had the highest percentage of b-sheet
content (34%), while R110)131, C110)131, and C114)131
had 15%, 23% and 6%, respectively.

These data strongly suggest that hindering hydrogen
bond formation decreases the b-pleated sheet structures of
the peptides, thereby limiting amyloid fibril formation.
Effect of the b-pleated sheet inhibitor
on amyloid fibril formation ‘
in vitro
’
The BB1 peptide, specifically designed to hinder amyloid
formation, was incubated with C110)131. The presence of
the BB1 peptide resulted in a 35% reduction in the yielding
of amyloid fibrils in vitro throughout the complete length of
the experiment (Fig. 8). Conversely, the BB2 peptide did not
interfere with C110)131 amyloid fibril formation. The BB2
ThT fluorescence curve matched the C110)131 ThT
fluorescence curve from day 1 to day 14. Therefore, BB1
peptide was able to inhibit amyloid fibril formation, and this
inhibition was specific as BB2 control peptide did not
modify amyloid yielding.
DISCUSSION
We previously presented an effective dialysis-based in vitro
system to study the formation of amyloid fibrils from
TGFBI protein-related synthetic peptides [14]. Based on this
system, we have now assessed the influence of the structure
of mutant TGFBI protein related peptides centered on
codon 124 on amyloid formation. From our data it becomes
apparent that (a) the particular amino-acid occupying the
position 124, (b) the presence of Val–Val at positions
112–113, and (c) the existence of disulfide and hydrogen
bonding, are of utmost importance in determining the
tendency of TGFBI protein related peptides to adopt a

b-pleated sheet configuration and to form amyloid. The
CONH
2
-terminus of the peptides analyzed has an inhibitory
Fig. 6. Secondary structure determination of the peptides by FT-IR
spectroscopy in D
2
O. Spectrograms of solutions containing the same
peptides included in Fig. 4 were analyzed in D
2
O (known to prevent
intra- and intermolecular hydrogen bonds). Note the absence of
inflexion of the curves both at 1625 and 1685 cm
)1
wavenumber, when
compared to the same peptides analyzed in H
2
O, showing that
hycdrogen bonding participates in the b-pleated conformation of the
peptides. The relative areas of the peaks around these wavenumbers
expressed as percentage of the total area over the curve were clearly
diminished for all the peptides (see the text).
Fig. 7. Spontaneous amyloid fibril formation.
(A) Congo red staining and polarized light of
big-h3 110)122 after 24 h of suspension in
distilled water showing a large amount of
amyloid material (· 125). (B) Electron
micrograph of the negatively stained material
from big-h3 110)122 spontaneous precipita-
tion on day 1 displaying bundles of 8–10 nm

wide fibrillar structure undistinguishable from
amyloid fibrils (· 60 000, bar ¼ 50 nm).
5154 C F. Schmitt-Bernard et al. (Eur. J. Biochem. 269) Ó FEBS 2002
effect on amyloid formation. Although the behavior of a
synthetic peptide in vitro is not identical to that of the
complete protein in vivo, our findings are in perfect
agreement with the phenotype–genotype associations
observed in clinical reports and identify some important
factors determining the appearance of amyloidosis. Our
system also allowed us to test a small peptide specifically
designed to prevent amyloid fibril formation from the
TGFBI protein related peptides. The results of these
experiments were very encouraging.
Concerning the importance of codon 124, our data
clearly show that of all the peptides analyzed C110)131 is
the peptide with the greatest tendency to form amyloid
fibrils. The amount of amyloid material produced was the
highest and kinetic studies showed that it was the fastest to
precipitate into amyloid fibrils. On the contrary, the
mutated S110)131 and the native R110)131 proved to
have the least tendency to form amyloid fibrils as shown by
ThT fluorescence analysis, while H110)131 presented an
intermediate degree of amyloid fibril formation.
Cysteine has the capacity to form disulfide bonds. As
disulfide bonding of the protein may participate in the
appearance of aggregates and predispose to the formation
of amyloid fibrils, we explored the consequences of
preventing disulfide bonding on amyloid transformation.
Blocking the Cys residues decreased amyloid formation by
approximately 50%, demonstrating that Cys–Cys bond

formation does facilitate amyloid precipitation, but is not
the only factor responsible for the amyloidogenicity of
C110)131 peptide. The participation of other factors in
amyloidogenesis, unrelated to disulfide bonding, is further
supported by the appearance of amyloid fibrils with
H110)131, and to a lesser extent with the other peptides
studied.
Amyloid deposits are characteristically insoluble. Hydro-
phobicity is an important feature in the formation of
b-pleated insoluble aggregates [20,21]. The NH
2
-terminal
sequence of the peptides included comprises two valine
residues at positions 112 and 113. Our studies with the
peptides lacking the residues 110–113 (excluding the two
valine amino acids), showed a decrease in the formation of
amyloid fibrils, suggesting the importance of the hydropho-
bic NH
2
-end of the peptides. In the same way, removing the
CONH
2
-terminus of the peptides resulted in increased
amyloid fibril formation. This increase may be due to a
direct inhibitory effect of this nine-amino-acid sequence, or
to disequilibrium in the hydrophobic-hydrophilic balance of
the truncated peptide. Based on the importance of the two
Val residues in amyloid fibril formation we designed BB1, to
interfere hydrophobic interactions. The positive results
obtained with BB1 further confirm the importance of

Val112-Val113 hydrophobic sequence in amyloid fibril
formation from TGFBI protein derived peptides.
Finally, hydrogen bonding also seems to participate
substantially in the formation of amyloid fibrils as demon-
strated by our studies on the secondary structure of the
peptides, both in H
2
O (allowing hydrogen bonding) and
D
2
O (preventing from hydrogen bonding) with less
b-pleated structures under the milieu preventing hydrogen
bonding.
Recent clinical reports are challenging some of the
classically accepted axioms on corneal dystrophy. Diseased
corneas of mixed granular corneal dystrophy type 2
(GCD2), associated with the R124H mutation, contain
substantial amounts of amyloid deposits when carefully
studied [22]. More surprisingly, granular corneal dystro-
phies of early and late onset, with pure granular deposits on
clinical examination, have also been found to contain
amyloid deposits in the cornea when assessed by electron
transmission microscopy [23]. In that report, amyloid
deposits were only identified in 1 out of the 2 affected
family members [24]. Therefore, amyloid fibrils may be
present in the deposits of supposedly pure granular dystro-
phies while undetected clinically or with Congo red staining
and light microscopy. The results reported in our study are
in keeping with these recent clinical reports, as they
demonstrate the capacity of forming amyloid fibrils in vitro

for all the peptides assessed, including the native form, when
the peptide solution was allowed to interact for longer
incubation periods.
As well as the precise factors we demonstrate to
participate in amyloid fibril formation, other factors may
also contribute in TGFBI protein amyloidogenesis. The first
is the putative alteration in the degradation of the TGFBI
protein that would allow amyloid fibril formation. It has
been recently proven that corneas of patients affected by
LCD1, GCD1, and GCD2 have an abnormal pattern of
TGFBI protein content compared with normal corneas,
suggesting the existence of an abnormal degradation of the
protein [22,24]. The second is the putative participation of
local factors on the appearance of the disease. The TGFBI
protein, in addition to the cornea, is abundantly expressed
in the skin. However, the clinical manifestations of BIGH3
related dystrophies are restricted to the cornea and we failed
to find amyloid deposits in the skin in LCD1, even when
carefully checked using electron microscopy [5]. Further, it
has been established that cultured fibroblasts bearing the
R124C mutation do not produce amyloid spontaneously
[22]. Therefore, the involvement of the cornea may be the
consequence of specific constitutive and/or physiological
factors of this tissue that would enhance amyloid formation,
as the presence of the same protein precursor in other
tissues does not result in amyloid deposits. These factors
may include interactions with other constitutive molecules
(glycosaminglycans, [25]), and/or the particular physiology
of the cornea, which acts as a semipermeable membrane
allowing ion and water transfers as in our in vitro system.

Fig. 8. Effect of the BB1 and BB2 peptides on amyloid fibril formation.
Amyloid fibril formation from C110)131 peptide was significantly
reduced in the mix containing BB1 (LPVVD) (35% when compared
with the control curve of C110)131 in the absence of BB1). The mix
with the nonspecific BB2 peptide (LPFFD) displays a ThT fluores-
cence curve matching the C110)131 control curve thus showing no
interference of BB2 with C110)131 amyloid fibril formation.
Ó FEBS 2002 TGFBI protein Arg124 mutations and amyloid formation (Eur. J. Biochem. 269) 5155
These local factors may also explain the corneal involve-
ment in other forms of amyloidosis [26].
In summary, autosomal dominant corneal diseases linked
to mutations in the BIGH3 gene are a prominent cause of
corneal blindness in adulthood. Conventional therapies
remain unsatisfactory due to the high rate and short delay of
recurrence after penetrating keratoplasty. Our findings
explain at the molecular level the disparities observed
among the different types of BIGH3 R124 mutation-linked
corneal dystrophies and suggest that designing specific
peptides able to inhibit the b-pleated conformational
organization of TGFBI protein related peptides is an
obvious route to successful therapy. The results of the first
molecule with this type of capacity are reported.
ACKNOWLEDGEMENTS
We want to thank Thierry Durroux (CCIPE, Montpellier, France) for
providing the Fluorescence spectrometer, Florence Trebillac (CRIC,
Montpellier, France) for her help in electron transmission microscopy,
and Dr Peter G. Kerr for his help in correcting this manuscript.
REFERENCES
1. Munier, F.L., Korvatska, E., Djemai, A., Le Paslier, D., Zografos,
L., Pescia, G. & Schorderet, D.F. (1997) Kerato-epithelin muta-

tions in four 5q31-linked corneal dystrophies. Nat. Genet. 15,247–
251.
2. Skonier, J., Neubauer, M., Madisen, L., Bennett, K., Plowman,
G.D. & Purchio, A.F. (1992) cDNA cloning and sequence analysis
of beta ig-h3, a novel gene induced in a human adenocarcinoma
cell line after treatment with transforming growth factor-beta.
DNA Cell. Biol. 11, 511–522.
3. Kim,J.E.,Kim,S.J.,Lee,B.H.,Park,R.W.,Kim,K.S.&Kim,
I.S. (2000) Identification of motifs for cell adhesion within the
repeated domains of transforming growth factor-beta-induced
gene, beta ig-h3. J. Biol. Chem. 275, 30907–30915.
4. Korvatska, E., Munier, F.L., Djemai, A., Wang, M.X., Frueh, B.,
Chiou, A.G., Uffer, S., Ballestrazzi, E., Braunstein, R.E.,
Forster, R.K., Culbertson, W.W., Boma, H., Zografos, L. &
Schorderet, D.F. (1998) Mutation hot spots in 5q31-linked corneal
dystrophies. Am.J.Hum.Genet.62, 320–324.
5. Schmitt-Bernard, C.F., Guittard, C., Arnaud, B., Demaille, J.,
Argiles, A., Claustres, M. & Tuffery-Giraud, S. (2000) BIGH3
exon 14 mutations lead to intermediate type I/IIIA of lattice
corneal dystrophies. Invest. Ophthalmol. Vis. Sci. 41, 1302–1308.
6. Schmitt-Bernard, C.F., Claustres, M., Arnaud, B., Demaille, J. &
Argiles, A. (2000) Lattice corneal dystrophy. Ophthalmology 107,
1613–1614.
7. Endo,S.,Nguyen,Y.H.,Fujiki,K.,Hotta,Y.,Nakayasu,K.,
Yamaguchi, T., Ishida, N. & Kanai, A. (1999) Leu518Pro muta-
tion of the beta ig-h3 gene causes lattice corneal dystrophy type I.
Am. J. Ophthalmol. 128, 104–106.
8. Fujiki, K., Hotta, Y., Nakayasu, K., Yokoyama, T., Takano, T.,
Yamaguchi, T. & Kanai, A. (1998) A new L527R mutation of the
beta igh3 gene in patients with lattice corneal dystrophy with deep

stromal opacities. Hum. Genet. 103, 286–289.
9.Rozzo,C.,Fossarello,M.,Galleri,G.,Sole,G.,Serru,A.,
Orzalesi, N., Serra, A. & Pirastu, M. (1998) A common big-h3
gene mutation (Df540)inalargecohortofsardinianReis-Bu
¨
cklers
corneal dystrophy patients. Hum. Mut. 12, 215–216.
10. Yamamoto, S., Okada, M., Tsujikawa, M., Shimomura, Y.,
Nishida, K., Inoue, Y., Watanabe, H., Maeda, N., Kurahashi, H.,
Kinoshita, S., Nakamura, Y. & Tano, Y. (1998) A kerato-epi-
thelin (beta ig-h3) mutation in lattice corneal dystrophy type IIIA.
Am.J.Hum.Genet.62, 719–722.
11. Munier,F.L.,Frueh,B.E.,Othenin-Girard,P.,Uffer,S.,Cousin,
P.,Wang,M.X.,Heon,E.,Black,G.C.,Blasi,M.A.,Balestrazzi,
E., Lorenz, B., Escoto, R., Barraquer, R., Hoeltzenbein, M.,
Gloor, B., Fossarello, M., Singh, A.D., Arsenijevic, Y., Zografos,
L. & Schorderet, D.F. (2002) BIGH3 mutation spectrum in cor-
neal dystrophies. Invest. Ophthalmol. Vis. Sci. 43, 949–954.
12. Stewart, H.S., Ridgway, A.E., Dixon, M.J., Bonshek, R., Parveen,
R. & Black, G. (1999) Heterogeneity in granular corneal dystro-
phy: Identification of three causative mutations in the TGFBI
(BIGH3) gene – lessons for corneal amyloidogenesis. Hum. Mut.
14, 126–132.
13. Mashima, Y., Nakamura, Y., Noda, K., Konishi, M., Yamada,
M., Kudoh, J. & Shimizu, N. (1999) A novel mutation at codon 124
(R124L) in the BIGH3 gene is associated with a superficial variant
of granular corneal dystrophy. Arch. Ophthalmol. 117, 90–93.
14. Schmitt-Bernard, C.F., Chavanieu, A., Derancourt, J., Arnaud,
B., Demaille, J.G., Calas, B. & Argiles, A. (2000) In vitro creation
of amyloid fibrils from native and Arg124Cys mutated

BIGH3
(110)131)
peptides, and its relevance for lattice corneal
amyloid dystrophy type 1. Biochem. Biophys. Res. Commum. 273,
649–653.
15. King, D.S., Fields, C.G. & Fields, G.B. (1990) A cleavage method
which minimizes side reactions following Fmoc solid phase pep-
tide synthesis. Int. J. Pept. Protein Res. 36, 255–266.
16. Naiki, H., Higuchi, K., Hosokawa, M. & Takeda, T. (1989)
Fluorometric determination of amyloid fibrils in vitro using the
fluorescent dye, thioflavin T1. Anal. Biochem. 177, 244–249.
17. Yamaguchi, I., Hasegawa, K., Takahashi, N., Gejyo, F. & Naiki,
H. (2001) Apolipoprotein E inhibits the depolymerization of beta
2-microglobulin–related amyloid fibrils at a neutral pH.
Biochemistry 40, 8499–8507.
18. Haris, P.I. & Chapman, D. (1994) Analysis of polypeptide and
protein structures using Fourier transform infrared spectroscopy.
Methods Mol. Biol. 22, 183–202.
19. Szabo,Z.,Klement,E.,Jost,K.,Zarandi,M.,Soos,K.&Penke,
B. (1999) An FT-IR study of the beta-amyloid conformation:
standardization of aggregation grade. Biochem. Biophys. Res.
Commun. 265, 297–300.
20. Glenner, G.G. (1980) Amyloid deposits and amyloidosis: the
beta-fibrilloses (first of two parts). N.Engl.J.Med.302,
1283–1292.
21. Glenner, G.G. (1980) Amyloid deposits and amyloidosis: the
beta-fibrilloses (second of two parts). N.Engl.J.Med.302,
1333–1343.
22. Korvatska, E., Henry, H., Mashima, Y., Yamada, M., Bachmann,
C., Munier, F.L. & Schorderet, D.F. (2000) Amyloid and

non-amyloid forms of 5q31-linked corneal dystrophy resulting
from kerato-epithelin mutations at Arg-124 are associated with
abnormal turnover of the protein. J. Biol. Chem. 275, 11465–
11469.
23. Akhtar, S., Meek, K.M., Ridgway, A.E., Bonshek, R.E. & Bron,
A.J. (1999) Deposits and proteoglycan changes in primary and
recurrent granular dystrophy of the cornea. Arch. Ophthalmol.
117, 310–321.
24. Takacs,L.,Boross,P.,Tozser,J.,Modis,L.,JrToth,G.&Berta,
A. (1998) Transforming growth factor-beta induced protein,
betaIG-H3, is present in degraded form and altered localization in
lattice corneal dystrophy type I. Exp. Eye Res. 66, 739–745.
25. Snow, A.D. & Kisilevsky, R. (1985) Temporal relationship
between glycosaminoglycan accumulation and amyloid deposition
during experimental amyloidosis: a histochemical study. Lab.
Invest. 53, 37–44.
26.Gorevic,P.D.,Munoz,P.C.,Gorgone,G.,Purcell,J.J.Jr,
Rodrigues,M.,Ghiso,J.,Levy,E.,Haltia,M.&Frangione,B.
(1991) Amyloidosis due to a mutation of the gelsolin gene in an
American family with lattice corneal dystrophy type II. New Eng.
J. Med. 325, 1780–1785.
5156 C F. Schmitt-Bernard et al. (Eur. J. Biochem. 269) Ó FEBS 2002