Báo cáo khoa học: Structure, epitope mapping, and docking simulation of a gibberellin mimic peptide as a peptidyl mimotope for a hydrophobic ligand pot

Structure, epitope mapping, and docking simulation
of a gibberellin mimic peptide as a peptidyl mimotope
for a hydrophobic ligand
Takashi Murata
1,3
, Hikaru Hemmi
1
, Shugo Nakamura
2
, Kentaro Shimizu
2
, Yoshihito Suzuki
3
and
Isomaro Yamaguchi
3
1 National Food Research Institute, Kannondai, Tsukuba, Japan
2 Department of Biotechnology, Division of Agriculture and Agricultural Life Sciences, The University of Tokyo, Japan
3 Department of Applied Biological Chemistry, Division of Agriculture and Agricultural Life Sciences, The University of Tokyo, Japan
The mimotope is a structure that acts as a mimic of an
epitope recognized by an antibody. Because a com-
pound with the similar tertiary structure to the epitope
could work as a mimotope, peptidyl mimotopes could
be prepared even to an epitope composed of nonpept-
idyl molecules. Peptidyl mimics for carbohydrates and
double-stranded DNA have been reported [1–3]. It is
difficult to obtain sufficiently high titre antibodies when
using nonpeptidyl molecules such as carbohydrates,
because they elicit only a T-cell independent immune
response, while peptidyl molecules can raise high titre
antibodies in a T-cell dependent manner. These peptidyl

mimics of carbohydrates could thus potentially serve as
surrogate antigens in discovering vaccines to overcome
the T-cell independent immune response and to obtain
anticarbohydrate antibodies with high binding activity
[1]. Few reports have been made on peptidyl mimics for
other ligands, especially hydrophobic ones, with the
Keywords
solution structure; STD-NMR; docking
simulation; hydrophobic ligand; mimic
peptide
Correspondence
H. Hemmi, National Food Research
Institute, 2-1-12 Kannondai, Tsukuba, Ibaraki
305-8642, Japan
Fax: +81 29 8387996
Tel: +81 29 8388033
E-mail:
Note
The atomic coordinates for the 50 conform-
ers of peptide SD described in this paper
have been deposited with the Protein Data
Bank (PDB ID 1YT6). Chemical shifts for
peptide SD have been deposited in the
BioMagRes Bank as entry 6511.
(Received 9 June 2005, revised 29 July
2005, accepted 4 August 2005)
doi:10.1111/j.1742-4658.2005.04902.x
Using NMR spectroscopy and simulated annealing calculations, we deter-
mined the solution structure of the disulfide-linked cyclized decapeptide
ACLPWSDGPC (SD), which is bound to an anti-(gibberellin A

4
) mAb
4-B8(8) ⁄ E9 and was found to be the first peptidyl mimotope for a hydro-
phobic ligand. The resulting structure of the peptide showed a b-turn-like
conformation in residues three to seven and the region converges well
(average rmsd 0.54 A
˚
). The binding activity and the epitopes of the peptide
to the antibody were assessed using saturation transfer difference (STD)-
NMR experiments. We also conducted docking simulations between the
peptide and the mAb to determine how the peptide is bound to the mAb.
Resonances around the b-turn-like conformation of peptide SD (residues
3–5) showed strong STD enhancement, which agreed well with results from
docking simulation between peptide SD and the mAb. Together with the
commonality of amino acid residues of the mAb involved in interactions
with gibberellin A
4
(GA
4
) and peptide SD, we concluded that peptide SD
is bound to the antigen-binding site of mAb 4-B8(8) ⁄ E9 as a GA
4
mimic,
confirming evidence for the existence of peptide mimics even for hydropho-
bic ligands.
Abbreviations
GAs, gibberellins; mAb, monoclonal antibody; STD, saturation transfer difference; DQF, double-quantum-filtered; Fab, antigen binding
fragment.
4938 FEBS Journal 272 (2005) 4938–4948 ª 2005 FEBS
exception of the water-soluble ligands biotin [4] and

deoxynivalenol (DON) [5]. If peptidyl mimics for hydro-
phobic ligands become generally available, they could
work as ideal immunogens to create antibodies that pos-
sess high binding activities to various organic com-
pounds such as plant hormones.
In our previous paper [6], two types of homologous
peptides with two different successive amino acids in
the middle of peptides (underlined), ACLPW
SDGPC
(SD) and ACLPW
GTGPC (GT), were screened as
peptidyl mimotopes of a hydrophobic ligand gibberel-
lin A
4
(GA
4
) (Fig. 1) against mAb 4-B8(8) ⁄ E9 by a
phage display method using a disulfide constrained
phage display peptide library (Ph.D C7CTM phage
display peptide library kit), because disulfide-con-
strained peptide libraries have proved to be useful in
identification of structural epitopes. As far as we are
aware this is the first report on peptidyl mimics for
hydrophobic ligands. Both peptides are composed
mostly of hydrophobic amino acid residues. These are
cyclized, forming a disulfide cross-link between Cys2
and Cys10. Gibberellins (GAs), a class of plant hor-
mones, play important roles in various plant growth
phenomena, including seed germination, stem elonga-
tion, and flower development [7]. We assumed that the

peptides interacted with the mAb at the GA binding
site based on the observation that the binding of phag-
es displaying these peptides was replaced by antigen
GA
4
, but not by GA
4
methyl ester which is not recog-
nized by the mAb. To confirm that peptides are bound
to the antigen binding site of the mAb and to discuss
interactions between peptides and the mAb in detail, it
is essential to determine the conformation of the pep-
tides and then investigate their interactions with
atomic resolution. It is thus worthwhile to determine
the conformations of these peptides and to analyse the
interaction between peptides and the mAb to obtain
clear evidence that the peptides are real mimotopes of
GAs and to confirm the existence of peptidyl mimics
even for hydrophobic ligands.
In this paper, we first report the solution structure
of the GA-mimic peptide, peptide SD, by 2D NMR
methods. Second, we report epitope mapping of the
peptide against the mAb by saturation transfer differ-
ence (STD)-NMR methods. Finally, we report compu-
tational docking simulation between peptide SD and
the mAb, and discuss the interaction between the
GA-mimic peptide and the mAb.
Results
1D
1

H-NMR spectra of two synthetic cyclized
decapeptides
1D
1
H-NMR spectra of the two peptides
ACLPW
SDGPC (SD) and ACLPWGTGPC (GT)
clearly showed that both had three conformations,
based on the number of resonance signals (data not
shown). We could not separate the three conformers of
these peptides by reversed-phase HPLC in this study.
Ratios of the three conformers for peptide SD or pep-
tide GT were estimated to be 3 : 1.4 : 1 or 2 : 2 : 1,
based on the differential signal intensity of the resolved
side-chain NH resonance of tryptophane residue in
each conformer. We assigned the
1
H chemical shifts
only for peptide SD using 2D NMR experiments
because: (a) peptide SD and peptide GT have high
binding activity for the antibody [6]; and (b) resonance
signals in the 1D proton spectrum of the peptide GT
are more complicated for resonance assignment than
those in the 1D proton spectrum of peptide SD, a
result of the ratio of the three conformers of peptide
GT where the larger two are almost equal.
Resonance assignments of peptide SD
The sequence-specific assignments of the proton reson-
ance from the residue in the three conformers (denoted
the three conformers in order of the signal intensity in

1D
1
H spectrum as major conformer, minor conformer
1, and minor conformer 2) of peptide SD were made
using standard procedures [8] from 2D NMR spectra
collected at 20, 25, 30, and 35 °C. For assignments of
Pro residues Ha(i)–Hd(i +1 : Pro) (dad)orHa(i)–
Ha(i + 1 : Pro) (daa) NOEs were used instead of daN.
Both proline residues, Pro4 and Pro9, of the major con-
former showed strong dad NOEs, indicating that all
proline residues in the major conformer of peptide SD
have a trans configuration. Pro4 of minor conformer 1
and Pro9 of minor conformer 2, however, showed dad
NOEs, but Pro4 of minor conformer 2 and Pro9 of
minor conformer 1 did not. daa NOEs between Leu3
Ha and Pro4 Ha in minor conformer 2 or between Gly8
Fig. 1. Structures of gibberellin A
4
and cyclized decapeptide AC-
LPWSDGPC (SD).
T. Murata et al. Gibberellin mimics peptide-antibody recognition
FEBS Journal 272 (2005) 4938–4948 ª 2005 FEBS 4939
Ha and Pro9 Ha in minor conformer 1 were observed,
indicating that peptide linkages of Leu3–Pro4 in minor
conformer 2 and Gly8–Pro9 in minor conformer 1 exhi-
bit a cis configuration. Peptide SD thus exists in three
isoforms due to cis–trans isomerization about the pep-
tide linkages of Leu3–Pro4 and Gly8–Pro9. In STD-
NMR experiments, binding between all conformers of
peptide SD and mAb 4-B8(8) ⁄ E9 were observed as

detailed later. We therefore propose that the cis ⁄ trans
configuration of proline residues is basically not critical
to binding to the mAb. Proton peaks of all conformers
in the SD peptide were completely assigned (see
Table 1). Resonance assignments were extended by
determining stereospecific assignments of some methy-
lene protons to obtain high-precision NMR structures.
b-Methylene protons were stereospecifically assigned for
three of 10 residues in the major conformer of peptide
SD using information on
3
J
HaHb
coupling constants
qualitatively estimated from the short-mixing time
TOCSY spectrum combined with intraresidue NH-Hb
and Ha-Hb NOEs. Sequential- and medium-range NOE
connectivities and slowly exchanging amide protons in
the major conformer of peptide SD is summarized in
Fig. 2. Unfortunately, the two minor conformers of
peptide SD had concentrations too low to detect med-
ium-range NOEs. Therefore, we could not determine the
tertiary structures of the two minor conformers of pep-
tide SD.
Conformation of the major conformer of peptide
SD
The 3D structure of the major conformer of peptide
SD was determined by simulated annealing calcula-
tions using 49 NOE-derived distance restraints (inclu-
ding 11 intraresidue, 28 sequential-residue, and 10

medium range), four hydrogen bond restraints, and
eight dihedral angle restraints. Fifty conformations
that give low conformation energy and that give no
distance and dihedral angle violations greater than
0.5 A
˚
and 5 A
˚
, respectively, were obtained. Statistical
data for the 50 structures of the major conformer of
peptide SD are given in Table 2. The structures thus
obtained had good covalent geometry and stereochem-
istry, as evidenced by the low rmsd values for bond,
angle and improper from idealized geometry. The
Ramachandran plot confirmed the high quality of
these structures, which showed that 100% of / and w
angles are found within core and allowed regions. Fig-
ure 3A shows the resulting solution structures of the
major conformer of peptide SD, where these structures
are superimposed to give the best fit in space. The
rmsd value from the mean structure is 1.60 A
˚
for all
backbone atoms in the whole molecule, while the cor-
responding value is 0.54 A
˚
for all backbone atoms in
the region of residues 3–7. This data indicates that the
region from Leu3 to Asp7 converges very well in cal-
culated structures. Figure 3B shows the schematic

drawing of the lowest energy structure of the major
conformer of peptide SD among the 50 calculated
structures, which is well characterized by a b-turn-like
conformation in the sequence Leu3-Pro4-Trp5-Ser6.
Interactions of peptide SD with mAb 4-B8(8)/E9
by STD-NMR experiments
To investigate the interaction between peptide SD and
mAb 4-B8(8) ⁄ E9, we performed STD-NMR experi-
ments. The STD-NMR technique is a method of epi-
tope mapping by NMR spectroscopy. During the
experiment, resonances of the protein are selectively
saturated and the signals of a ligand that is specifically
bound to a target protein show changes in resonance
intensity and are observed in the difference NMR
spectrum, while those of nonassociating ligands are
cancelled out and not observed in the difference spec-
trum. The time course of saturation was determined by
plotting the STD amplification factor against satura-
tion time in the fixed concentration of peptide SD in
the presence of mAb 4-B8(8) ⁄ E9, since the absolute
magnitude of the STD effect depends on the concen-
tration of a ligand and saturation time [9]. Saturation
profiles of peptide SD showed that a 3-s saturation
time was sufficient for efficient saturation transfer from
a proton in the protein to that in peptide SD, and we
carried out STD-NMR experiments with a 3-s satura-
tion time for the epitope mapping of mAb 4-B8(8) ⁄ E9
(data not shown).
Figure 4 shows (A) the 1D
1

H-NMR spectrum of
peptide SD incubated with mAb 4-B8(8) ⁄ E9 at a ratio
of 100 : 1; and (B) the corresponding 1D STD spec-
trum. 1D STD-NMR signals of peptide SD were
assigned and some signals of the three conformers
overlapped. We confirmed STD-NMR signal assign-
ment by 2D STD-TOCSY spectra, and overlapping
signals were treated as a group to calculate their STD
intensity (Table 3). The integral value of the signal of
one of the b protons of Leu3 of minor conformer 1,
the largest STD intensity of peptide SD, was much lar-
ger than those of other STD signals, and thus this was
set to 200%. Table 3 shows the relative degree of sat-
uration of individual protons normalized to that of
one of the b protons of Leu3 of minor conformer 1.
STD enhancement was observed for all three conform-
ers, indicating that they all interact with the mAb. We
also found that the pattern of STD enhancement for
Gibberellin mimics peptide-antibody recognition T. Murata et al.
4940 FEBS Journal 272 (2005) 4938–4948 ª 2005 FEBS
Table 1.
1
H chemical shifts of three conformers in peptide SD (p.p.m., from 2,2-dimethyl-2-silapentane-5-sulfonate).
Residue
Major conformer Minor conformer 1 Minor conformer 2
NH HA HB Others NH HA HB Others NH HA HB Others
Ala1 – 4.10 1.50 – 4.10 1.48 – 4.10 1.57
Cys2 8.76 4.68 2.95(HB2)
a
, 3.12(HB3) 8.82 4.63 2.95, 3.12 8.61 4.75 2.98, 3.33

Leu3 8.65 4.52 0.72(HB3), HG 1.56 8.52 4.48 0.58, 1.16 HG 1.57 8.05 4.28 1.59 HG 1.43
1.23(HB2) HD 0.85, 0.92 HD 0.84, 0.93 HD 0.89
Pro4 – 4.24 1.80, 2.20 HG 1.95, 2.00 – 4.11 1.80, 2.19 HG 1.95, 2.01 – 4.44 1.68, 2.19 HD 1.93
HD 3.21, 3.69 HD 3.19, 3.69 HD 3.15, 3.30
Trp5 7.04 4.74 3.33, 3.49 HD1 7.22
HE1 10.31
HE3 7.67
HH2 7.30
HZ2 7.55
HZ3 7.23
6.88 4.72 3.34, 3.51 HD1 7.21
HE1 10.33
HE3 7.66
HH2 7.30
HZ2 7.56
HZ3 7.22
8.13 4.53 3.35 HD1 7.33
HE1 10.18
HE3 7.66
HH2 7.26
HZ2 7.51
HZ3 7.17
Ser6 7.62 4.42 3.77 7.44 4.58 3.67 7.74 4.39 3.62, 3.81
Asp7 8.48 4.77 2.74, 2.81 8.61 4.77 2.74, 2.86 8.20 4.60 2.70, 2.75
Gly8 7.92 4.05, 4.12 8.06 3.80, 4.02 8.00 3.97, 4.20
Pro9 – 4.43 1.93, 2.25 HG 2.01 – 4.58 1.93, 2.39 HG 2.15 – 4.43 1.94, 2.24 HG 2.00
HD 3.60 HD 3.57 HD 3.59
Cys10 8.16 4.44 3.06(HB2),
3.19(HB3)
8.25 4.48 3.02, 3.21 8.53 4.53 2.98, 3.26

a
b-Protons in parentheses were stereospecifically assigned.
T. Murata et al. Gibberellin mimics peptide-antibody recognition
FEBS Journal 272 (2005) 4938–4948 ª 2005 FEBS 4941
some residues among the three conformers such as
HE1 of Trp5 differed from one another (Table 3).
These results indicate that the conformational change
among the three conformers due to cis–trans isomeri-
zation at the position of Pro4 or Pro9 may affect the
difference in the pattern of STD enhancement. Strong
STD enhancement (> 60%) of all three conformers
was observed, however, only for residues, Leu3-Trp5,
constituting a b-turn-like structure, while C-terminal
residues Ser6-Cys10, have lower STD enhancement
(12–28%) except for NH of Ser6 in minor conformer 1
(Table 3), suggesting that the region from Leu3 to
Trp5 has more and tighter contacts to the surface of
the mAb.
Docking simulation between the conformation of
peptide SD and the mAb
Docking simulation of the conformation of peptide SD
obtained in this study to the crystal structure of mAb
4-B8(8) ⁄ E9 antigen binding fragment (Fab) [Protein
Data Bank (PDB) ID 1KFA] was performed by using
gold 2.1 software. Default parameters for the energy
function were used, including hydrogen bond energy
between the protein and ligand, van der Waals energy
between the protein and ligand and within the ligand,
and internal torsion energy for the ligand. To consider
Fig. 2. Summary of sequential and medium-range NOE connectivi-

ties observed for the major conformer of peptide SD. Bars, the size
of which indicates the NOE intensity (strong, medium, and weak),
represent sequential NOEs. Slow exchanging amide protons are
also represented as closed circles.
Table 2. Statistics for 50 NMR structures of peptide SD.
Number of restraints
Total distance restraints 53
Intraresidue 11
Sequential 28
Medium (1 < | i–j | < 5) 10
Long (| i–j | ‡ 5) 0
Hydrogen bond (2 per bond) 4
Total dihedral angle restraints 8
/ 5
w 0
v
1
3
rmsd from experimental restraints
NOE distance restraints (A
˚
) 0.0305 ± 0.0123
Dihedral angle restraints (degree) 0.3631 ± 0.1474
rmsd from ideal covalent geometry
Bonds (A
˚
) 0.0023 ± 0.00071
Angles (degree) 0.5644 ± 0.0294
Impropers 0.1668 ± 0.0529
/ and w in core and allowed regions (%)

a
100
rmsd relative to the mean structure (A
˚
)
Backbone (N, Ca and C¢ atoms) All non-H
Whole molecule (residues 1–10)
1.60 ± 0.46 2.22 ± 0.58
Core region (residues 3–7)
0.54 ± 0.19 1.47 ± 0.31
a
The program PROCHECK-NMR [30] was used for Ramachandran plot
analysis.
AB
N
C
C
N
A1
C2
L3
W5
S6
D7
P9
C10
P4
G8
Fig. 3. Superimposition of 50 structures (A) and ribbon diagram of
the lowest energy structure of the major conformer of peptide SD

(B). One disulfide bridge (Cys2–Cys10) and side chains of all resi-
dues are ball-and-stick representations. This figure was generated
using
MOLMOL [31].
A
B
Fig. 4. Reference NMR spectrum of mixture of peptide SD and
mAb 4-B8(8) ⁄ E9 in the ratio of 100 : 1 (A) and STD-NMR spectrum
of the same sample (B). Prior to acquisition, a 30 ms spin-lock
pulse was applied to remove residual protein resonance.
Gibberellin mimics peptide-antibody recognition T. Murata et al.
4942 FEBS Journal 272 (2005) 4938–4948 ª 2005 FEBS
the flexibility of peptide SD, 25 runs were executed for
each of the 50 NMR structures while fixing the main-
chain atoms of peptide SD and optimizing its side-
chain atoms. Among 1250 peptide SD-mAb
4-B8(8) ⁄ E9 Fab complex structures predicted in this
study, 904 complex structures with positive fitness
scores were selected, and clustered using the method of
Baker et al. [10]. During clustering, the rmsd of Ca
atoms was used as the measure of distance between
structures.
Three large clusters were found among 904 predicted
complex structures by the clustering analysis, and three
complex structures that were closest to each centre of
the three large clusters were obtained. We then com-
pared the fitness values of the three complex structures
to find the best model of the peptide SD-mAb
4-B8(8) ⁄ E9 complex structure. The fitness values of the
three complex structures were 51.26, 49.49, and 1.97.

The complex structure with the highest fitness value,
51.26, was obtained from the largest cluster, indicating
that this complex structure is the best model of all pre-
dicted complex structures in this study. Firstly, the
three complex structures obtained by clustering analy-
sis showed that peptide SD interacted with the antigen
binding site in mAb 4-B8(8) ⁄ E9 (Fig. 5A). Residues of
Pro4 and Trp5 of peptide SD in each of three complex
structures are located at almost the same positions in
the three complex structures, and these residues of
peptide SD showed hydrophobic interaction with mAb
4-B8(8) ⁄ E9 in complex structures. The three complex
structures also showed that amide proton of Ala33,
which is very important for binding with GA
4
, in mAb
4-B8(8) ⁄ E9 was located at a position to possibly form
hydrogen bonding with peptide SD. We speculated
from the three complex structures, however, that pep-
tide SD is mainly bound to mAb 4-B8(8) ⁄ E9 by hydro-
phobic interaction.
Next, we analysed the interaction between peptide SD
and mAb 4-B8(8) ⁄ E9 in detail using the best model pre-
dicted (Fig. 5B). In this model of peptide SD-mAb
4-B8(8) ⁄ E9 complex, two hydrogen bonds exist between
the antibody and the peptide: Ala33a NH–CO Pro4p
(where a denotes an antibody residue and p denotes a
peptide residue); and Thr53a OHsc–CO Trp5p (where sc
denotes side-chain) (Fig. 5B). We reported the crystal
structure of mAb 4-B8(8) ⁄ E9 with GA

4
previously [11].
In the complex structure, NH of Ala33 and NH of
Thr53 of the mAb formed hydrogen bonds with GA
4
.
The results indicate that peptide SD in the best complex
model interacts with very important residues, Ala33 and
Thr53, of the mAb for antigen recognition. This com-
plex model obtained from docking simulations in this
study thus appears to be extremely suitable.
As described, we found from STD-NMR experiments
in this study that the region Leu3-Trp5 of peptide SD is
an important epitope for interaction with the mAb. The
corresponding region of peptide SD in the three com-
plex models, also shown to interact with the mAb from
docking simulations in this study, is in good agreement
with the results obtained from STD-NMR experiments.
In complex models, the region from Ser6 to Cys10,
which showed the lower STD enhancement, had no
interactions with the mAb. We thus conclude that
Table 3. STD enhancement of peptide SD in the presence of
monoclonal antibody 4-B8(8) ⁄ E9. Resonance signals overlapping
between major conformer and minor conformer 1 are shown in ital-
ics. STD enhancement was normalized to the strongest enhance-
ment, Leu3 HB of minor conformer 1 (0.58 p.p.m.).
Resonances d (p.p.m)
STD enhancement
(%)
Ala-1 HA 4.10 28

Ala-1 HB 1.48–1.50 36
Cys-2 NH of minor conformer 1 8.82 6
Cys-2 HB 2.95, 3.12 46, 26
Leu-3 NH of minor conformer 1 8.52 46
Leu-3 HB2 of major conformer 1.23 50
Leu-3 HB3 of major conformer 0.72 68
Leu-3 HB of minor conformer 1 1.16 58
Leu-3 HB of minor conformer 1 0.58 200
Leu-3 HG 1.56–1.57 42
Leu-3 HD 0.84–0.85,
0.92–0.93
58
50
Pro-4 HB 1.80,
2.19–2.20
66
50
Pro-4 HB of minor conformer 2 1.68 44
Pro-4 HG ⁄ Pro9 HB ⁄ Pro9 HG of
major conformer
1.93–2.01 50
Pro-4 HD ⁄ Ser6 HB of
minor conformer 1
3.69–3.77 14
Pro-4 HD ⁄ Cys10 HB 3.19–3.21 24
Trp-5 NH of major conformer 7.04 108
Trp-5 NH of minor conformer 1 6.88 102
Trp-5 HB 3.33–3.51 34
Trp-5 HD1 ⁄ HZ3 7.21–7.23 80
Trp-5 HE1 of major conformer 10.31 20

Trp-5 HE1 of minor conformer 1 10.33 60
Trp-5 HE3 7.66–7.67 116
Trp-5 HH2 7.30 80
Trp-5 HZ2 7.55–7.56 76
Trp-5 HZ2 of minor conformer 2 7.51 64
Ser-6 NH of major conformer 7.62 20
Ser-6 NH of minor conformer 1 7.44 68
Ser-6 HB of major conformer 3.77 12
Asp-7 NH of minor conformer 1 8.61 24
Asp-7 HB 2.74,
2.81–2.86
22
22
Gly-8 NH of minor conformer 1 8.06 28
Pro-9 HB 2.25–2.39 24
Cys-10 HB 3.02–3.06 22
T. Murata et al. Gibberellin mimics peptide-antibody recognition
FEBS Journal 272 (2005) 4938–4948 ª 2005 FEBS 4943
complex models between peptide SD and the mAb
obtained from docking simulation in this study are suit-
able since epitopes of peptide SD to the mAb obtained
from STD-NMR are in good agreement with those
obtained from docking simulation.
Discussion
We previously identified two disulfide linked cyclized
decapeptides, SD and GT, which have affinity with
mAb 4-B8(8) ⁄ E9, an antibioactive GA antibody, by
screening a phage display peptide library [6]. In this
study, we performed NMR spectroscopic analysis of
the peptides to determine the conformation and an epi-

tope for mAb 4-B8(8) ⁄ E9 in order to obtain structural
information showing that the peptides are bound to
the antigen-binding site of the mAb as GA
4
mimics.
We first measured 1D
1
H-NMR spectra of peptides
SD and GT. 1D proton spectra of the peptides showed
that each of the two peptides has three cis ⁄ trans iso-
mers due to two proline residues; resonance signals in
1D proton spectra are complicated by resonance sig-
nals overlapping among the three isomers. Fortunately,
in the 1D proton spectrum of peptide SD, the intensity
of the resonance signals of one isomer (all trans-confi-
guration) is much stronger than those of the other two
isomers (one cis-configuration and one trans-configur-
ation). For the large isomer (major conformer) of pep-
tide SD, we therefore assigned resonance signals and
determined the solution structure by 2D NMR

Phe100BH
Leu97H
Leu98H
Tyr100AH
A

B
Fig. 5. Docking simulation models of pep-
tide SD and mAb 4-B8(8) ⁄ E9. (A) Electro-
static potential surface of mAb 4-B8(8) ⁄ E9
in complex with peptide SD. Three models
of peptide SD are shown in the wire model
(green). Surface electrostatic potentials of
mAb 4-B8(8) ⁄ E9 were calculated in
MOLMOL
[31], coloured by electrostatic potentials
with positive regions in blue and negative
regions in red. Some residues of peptide SD
or mAb 4-B8(8) ⁄ E9 described in text using
one-letter codes or three-letter codes,
respectively, are also represented. (B) Sche-
matic drawing of the interaction between
peptide SD and mAb 4-B8(8) ⁄ E9 in the
docking simulation model. The figure was
generated by
LIGPLOT [36].
Gibberellin mimics peptide-antibody recognition T. Murata et al.
4944 FEBS Journal 272 (2005) 4938–4948 ª 2005 FEBS
spectroscopy. We measured the STD-NMR spectrum
for the mixture of peptide SD and mAb 4-B8(8) ⁄ E9 to
investigate the interaction of peptide SD with the
mAb. We also performed the docking simulation using
the NMR structures of peptide SD and the crystal
structure of the mAb Fab in the complex with GA
4
to

analyse interactions between them in more detail.
The solution structure we determined for peptide SD
showed a b-turn-like conformation in residues 3–7 and
the region converges well (average rmsd 0.54 A
˚
). This
conformation would be stabilized by two intramolecular
Leu3 NH–Asp7 CO and Ser6 NH–Leu3 CO hydrogen
bonds. The b-turn motif has been observed in other
antigenic peptides free in solution [12–15] and bound
to antibodies [16–19]. The 12-residue carbohydrate-
mimetic peptide recognized by an antigroup B Strepto-
coccus antibody was recently reported to have a type I
b-turn both free and bound to the antibody [20]. The
turns present in the bound and free peptide are very sim-
ilar and residues forming this turn are recognized by the
mAb as demonstrated by STD-NMR experiments,
which indicates that the b-turn conformation may be an
important reason for the effective immunogenicity of
the peptide. In our study, bound conformation of pep-
tide SD has not been determined yet. However, peptide
SD has the b-turn-like conformation stabilized by two
hydrogen bonds when free and residues (Leu3-Trp5)
forming this turn are recognized by the mAb, as demon-
strated by STD-NMR experiments. We propose that the
b-turn-like conformation of peptide SD is important for
binding to the mAb, this being supported by the reason-
ably good simulated docking between peptide SD and
the mAb when fixing the b-turn-like conformation of
peptide SD. If we can monitor the changes in chemical

shifts of peptide SD on the addition of mAb using
15
N-labelled peptide, expected results will make it
clearer that the b-turn-like conformation of peptide SD
is important for binding to the mAb.
We previously determined the crystal structure of
the complex formed with the mAb 4-B8(8)⁄ E9 Fab
and GA
4
[11]. It shows that 3b-hydroxy and 6b-carb-
oxyl groups of GA
4
form hydrogen bonds with
Ala33H of the main chain, and with Thr53H of the
heavy chain, respectively. Furthermore, C ⁄ D rings of
GA
4
were in van der Waals’ contact mainly with the
aromatic side chain of Tyr100AH and Phe100BH of
the third complementarity-determining region of the
heavy chain in mAb 4-B8(8) ⁄ E9. Our complex model
between mAb 4-B8(8) ⁄ E9 and peptide SD in this study
shows that Pro4 CO and Trp5 CO of peptide SD are
hydrogen-bonded to Ala33H NH and to Thr53H side-
chain OH of the mAb, respectively. Furthermore, the
region composed of hydrophobic amino acid residues,
Leu3-Pro4-Trp5, of peptide SD form hydrophobic
interactiona with the hydrophobic surface of the mAb
including Tyr100AH and Phe100BH, as also demon-
strated by STD-NMR experiments (Figure 5; Table 3).

These results indicate that the peptide SD–mAb
4-B8(8) ⁄ E9 interaction is very similar to the GA
4
–mAb
4-B8(8) ⁄ E9 interaction. Previously [6], the binding of
phages having peptides SD and GT was not inhibited
by excess GA
4
methylester, which is not reactive with
mAb 4-B8(8) ⁄ E9, suggesting that the binding of the
peptides is tightly related to the binding property of
the mAb to its antigen. We therefore conclude that
peptide SD is a real mimotope of GA
4
for the mAb.
So far, quite a number of peptidyl mimics for carbo-
hydrates or double-stranded DNA have been prepared
[1–3]. Mimicry peptides for water-soluble ligands, bio-
tin and DON, have also been reported [4,5]. No
reports exist, however, on peptidyl mimics for hydro-
phobic ligands, such as GAs, except for our previous
study [6]. In this study, we confirmed that peptide
SD formed hydrophobic interactions with mAb
4-B8(8) ⁄ E9 as a GA
4
mimic. This is thus the first proof
that peptidyl mimics can be prepared even though the
ligands are hydrophobic, such as GAs. This finding
would provide further availability of peptidyl mimics
for other hydrophobic ligands as ideal immunogens to

create antibodies that possess high binding activities to
the organic compounds.
Experimental procedures
Sample preparation
Two synthetic cyclized decapeptides, ACLPWSDGPC (SD)
and ACLPWGTGPC (GT), made by the Fmoc method
were purchased from Bex (Tokyo, Japan). A disulfide bond
in these peptides was formed under oxidized conditions.
Synthetic peptides were dissolved in 50 mm phosphate buf-
fer (pH 5.0, 90% H
2
O ⁄ 10% D
2
O, v,v) to give a final con-
centration of  5mm for NMR experiments. To detect the
hydrogen bond, the peptide solution was lyophilized and
redissolved in an equal volume of 100% D
2
O. mAb
4-B8(8) ⁄ E9 was prepared as reported elsewhere [6]. The
mAb was dissolved in NaCl ⁄ P
i
(10% D
2
O), and concentra-
ted to  50 lm by centrifugation using Centricon-10 (Milli-
pore, Billerica, MA). The peptide was dissolved into the
mAb solution and the molar ratio of the peptide to mAb
was adjusted to 100 : 1 for STD-NMR experiments.
NMR spectroscopy

All NMR spectra were obtained on Bruker Avance
600 MHz and 800 MHz spectrometers at 293 K, 298 K,
T. Murata et al. Gibberellin mimics peptide-antibody recognition
FEBS Journal 272 (2005) 4938–4948 ª 2005 FEBS 4945
and 303 K. Standard Bruker software (xwinnmr 2.6) was
used to acquire and process NMR data. Water suppression
was performed using watergate sequence [21,22]. Chem-
ical shifts were referenced to internal 2,2-dimethyl-2-silapen-
tane-5-sulfonate at 25 °C. All 1D
1
H-NMR spectra were
recorded with 128 scans and 60 000 data points, and proc-
essed by zero-filling to 60 k points and multiplication by an
exponential function, followed by Fourier transformation.
Resonance signals were assigned based on 2D double-quan-
tum-filtered (DQF)-COSY, TOCSY, ROESY and NOESY
spectra. The 2D spectra were recorded with quadrature
detection in the phase-sensitive mode by time proportional
phase increment (TPPI) [23] and States-TPPI [24]. 2D spec-
tra were acquired with a spectra width of 15 p.p.m. in both
dimensions and 512 and 2048 complex points in both
dimensions. TOCSY spectra with a DIPSI-2 mixing
sequence were recorded with mixing times of 35, 60 or
80 ms. NOESY spectra were obtained with mixing times of
60, 100, 200, and 400 ms, and ROESY with a mixing time
of 100 ms. The high digital resolution DQF-COSY spec-
trum was recorded using 400 and 4096 complex points in
both dimensions. Slowly exchanging amide protons were
identified by lyophilizing peptide from a H
2

O solution, dis-
solving the peptide in D
2
O, and collecting sequential 2-h
2D TOCSY spectra. Before Fourier transformation, the
shifted sine-bell window function was applied to the t1 and
t2 dimensions. Peak picking and assignment were per-
formed with sparky (T. D. Goddard and D. G. Kneller,
sparky 3, University of California, San Francisco, CA).
1D and 2D STD-NMR spectra were performed as des-
cribed by Mayer and Meyer [9]. The time dependence of
the saturation transfer was determined by recording 1D
STD spectra with 1 k scans and saturation times from
0.25 s to 6.0 s. The irradiation power in all STD-NMR
experiments was set to  0.15 W. Relative STD values were
calculated by dividing STD signal intensities by the intensi-
ties of the corresponding signals in a reference spectrum of
the same sample recorded with 64 scans. All STD-NMR
spectra for epitope mapping were acquired using a series of
equally spaced 50 ms Gaussian-shaped pulses for saturation
with 1 ms intervals and the total saturation time of  3s.
The on-resonance irradiation of the protein was performed
at the chemical shift of )2.0 p.p.m., and the off-resonance
at 40 p.p.m. where no protein signal was present. Free
induction decay values with on- and off-resonance protein
saturation were recorded in alternative fashion. Subtraction
of the 1D STD spectra was achieved via phase cycling. Pro-
tein resonance was suppressed by application of a 30 ms
spin-lock pulse prior to acquisition. 2D STD-TOCSY spec-
tra with on- and off-resonance protein saturation were

recorded with 128 scans per t
1
increment in alternative fash-
ion. The 2D spectra were acquired with a spectra width of
15 p.p.m. in both dimensions, and 256 and 2048 complex
points in both dimensions. A MLEV-17 mixing time of
100 ms was applied in STD-TOCSY spectra.
Structure calculations
NOE-derived distance restraints were classified into three
ranges, 1.8–3.0 A
˚
, 1.8–4.0 A
˚
and 1.8–5.0 A
˚
, according to
the relative NOE intensities. Upper distance limits for
NOEs involving methyl protons and nonstereospecifically
assigned methylene protons were corrected appropriately
for centre averaging [25]. In addition, a distance of 0.5 A
˚
was added to the upper distance limits only for NOEs
involving methyl protons [26] after correction for centre
averaging. Torsion angle constraints on the backbone /
angle are usually derived from
3
J
HNHa
coupling constants
estimated from high digital resolution 2D DQF-COSY

spectra, and sequential and short-range NOEs. However,
we could not obtain / angle restraints from the DQF-
COSY spectra, because all
3
J
HNHa
coupling constants
observed were between 6 Hz and 8 Hz. The additional /
angle restraint of 100° ±80° was applied to residues for
which the intraresidue Ha-HN NOE was clearly weaker
than the NOE between HN and the Ha of the preceding
residue [27]. Five / angle restraints were obtained for the
peptide. Side-chain v
1
angles were determined by
3
J
HaHb
coupling constants qualitatively estimated from short-mix-
ing TOCSY connectivities combined with NH-Ha and
Ha-Hb NOEs [28]. Three v
1
angle restraints for the peptide
were obtained. v
1
angle restraints were normally restricted
to a ± 60° range from staggered conformations,
g + (+ 60°), t (180°)org–()60°). Hydrogen-deuterium
exchange experiments identified four hydrogen bond donors
for the peptide. Corresponding hydrogen bond acceptors

were determined based on NOE patterns observed for regu-
lar secondary structural regions and preliminary calculated
structures without restraints regarding hydrogen bonds.
Hydrogen bond constraints were applied to N–H and C¼O
groups: 1.7–2.4 A
˚
for the H-O distance and 2.7–3.4 A
˚
for
the N-O distance.
The peptide structures were calculated by simulated
annealing using torsion angle dynamics with the program
cns [29]. The structure calculation proceeded in two stages.
In the first stage, a low-resolution structure was preliminar-
ily determined using only NOE-derived distance restraints.
In the second stage, the same protocol was applied by add-
ing hydrogen bond restraints and dihedral angle restraints.
Additional NOE constraints were added in each round of
calculations, and restraints that were consistently violated
were removed. Additional NOE constraints were then
added and used in the final structure calculation. All subse-
quent numerical analyses were performed using procheck-
NMR [30] and molmol [31]. Structure figures were gener-
ated using molmol.
Docking simulation
Docking simulation of peptide SD to mAb 4-B8(8) ⁄ E9 Fab
was performed by using GOLD 2.1 software [32–35]. The
Gibberellin mimics peptide-antibody recognition T. Murata et al.
4946 FEBS Journal 272 (2005) 4938–4948 ª 2005 FEBS
structure of the Fab was obtained from the crystal structure

of the complex of the Fab and GA
4
determined in our pre-
vious work (PDB ID: 1KFA) [11]. The structure of the Fab
was fixed except for v angles of serine residues. The search
area was set within 10 A
˚
of the centroid of heavy atoms of
GA
4
in the crystal structure. Other calculation parameters
were set to the default values. To validate these calculation
conditions, we first tried the docking simulation of GA
4
to
the Fab. Ten simulation runs were executed and we found
that these calculation conditions could reproduce the con-
formation of GA
4
in good agreement with that in the crys-
tal structure (the average rmsd of all heavy atoms was
0.88 A
˚
). To consider the flexibility of peptide SD, 25 runs
were executed for each of the 50 structures of peptide SD
determined from 2D
1
H-NMR spectroscopic data and the
simulated annealing calculation above.
References

1 Monzavi-Karbassi B, Cunto-Amesty G, Luo P &
Kieber-Emmons T (2002) Peptide mimotopes as surro-
gate antigens of carbohydrates in vaccine discovery.
Trends Biotechol 20, 207–214.
2 Prinz DM, Smithson SL & Westerink MAJ (2004) Two
different methods result in the selection of peptides that
induce a protective antibody response to Neisseria
meningitidis serogroup C. J Immunol Methods 285, 1–14.
3 Deocharan B, Qing X, Beger E & Putterman C (2002)
Antigenic triggers and molecular targets for anti-double-
stranded DNA antibodies. Lupus 11, 865–871.
4 Devlin JJ, Panganiban LC & Devlin PE (1990) Random
peptide libraries: a source of specific protein binding
molecules. Science 249, 404–406.
5 Yuan Q, Pestka JJ, Hespenheide BM, Kuhn LA, Linz
JE & Hart LP (1999) Identification of mimotope pep-
tides which bond to the mycotoxin deoxynivalenol-spe-
cific monoclonal antibody. Appl Environ Microbiol 65,
3279–3286.
6 Sekimoto H, Suzuki Y & Yamaguchi I (2004) Identifica-
tion of peptidyl mimics of bioactive gibberellin recog-
nized by an antibody. Biosci Biotechnol Biochem 68,
2408–2410.
7 Davies PJ (1995) Plant Hormones: Physiology, Biochem-
istry and Molecular Biology, 2nd edn. Kluwer Academic
Publishers, the Netherlands.
8Wu
¨
thrich K (1986) NMR of Proteins and Nucleic Acids.
John Wiley & Sons, New York.

9 Mayer M & Meyer B (2001) Group epitope mapping by
saturation transfer difference NMR to identify segments
of a ligand in direct contact with a protein receptor.
J Am Chem Soc 123, 6108–6117.
10 Bonneau R, Strauss C & Baker D (2001) Improving the
performance of rosetta using multiple sequence align-
ment information and global measures of hydrophobic
core formation. Proteins 43, 1–11.
11 Murata T, Fushinobu S, Nakajima M, Asami O, Sassa
T, Wakagi T & Yamaguchi I (2002) Crystal structure of
the liganded anti-gibberellin A
4
antibody 4-B8(8) ⁄ E9
Fab fragment. Biochem Biophys Res Commun 293, 489–
496.
12 Dyson HJ, Rance M, Houghten RA, Lerner RA &
Wright PE (1988) Folding of immunogenic peptide frag-
ments of proteins in water solution. I. Sequence require-
ments for the formation of a reverse turn. J Mol Biol
201, 161–200.
13 Dyson HJ, Cross KJ, Houghten RA, Wilson IA, Wright
PE & Lerner RA (1985) The imunodominant site of a
synthetic immunogen has a conformational preference
in water for a type-II reverse turn. Nature 318, 480–483.
14 Chandrashekhar K, Profy AT & Dyson HJ (1991) Solu-
tion conformational preferences of immunogenic pep-
tides derived from the principal neutralizing
determinant of the HIV-envelope glycoprotein gp120.
Biochemistry 30, 9187–9194.
15 Campbell PA, McInnes C, Hodges RS & Sykes BD

(1995) Comparison of NMR solution structure of the
receptor binding domain of Pseudomonas aeruginosa pili
strains PAO, KB7, and PAK: implications for receptor
binding and synthetic vaccine design. Biochemistry 34,
16255–16268.
16 De Lorimier R, Moody MA, Haynes BF & Spicer LD
(1994) NMR-derived solution conformations of a hybrid
synthetic peptide containing multiple epitopes of envel-
ope protein gp120 from the RF strain of human immu-
nodeficiency virus. Biochemistry 33, 2055–2062.
17 Campbell PA, Wong WY, Irvin RT & Sykes BD (2000)
Interaction of a bacterially expressed peptide from the
receptor binding domain of Pseudomonas aeruginosa pili
strain PAK with a cross-reactive antibody: conforma-
tion of the bound peptide. Biochemistry 39, 14847–
14864.
18 Derrick JP, Maiden CJ & Feavers IM (1999) Crystal
structure of an Fab fragment in complex with a menin-
gococcal serosubtype antigen and a protein G domain.
J Mol Biol 293, 81–91.
19 Ghiara JB, Stura EA, Stanfield RL, Profy AT & Wilson
IA (1994) Crystal structure of the principal neutraliza-
tion site of HIV-1. Science 264, 82–85.
20 Johnson MA, Jaseja M, Zou W, Jennings HJ, Copie V,
Pinto BM & Pincus SH (2003) NMR studies of carbo-
hydrates and carbohydrate-mimetic peptides recognized
by an anti-group B Streptococcus antibody. J Biol Chem
278, 24740–24752.
21 Piotto M, Saudek V & Sklenar V (1992) Gradient-tai-
lored excitation for single-quantum NMR spectroscopy

of aqueous solutions. J Biomol NMR 2, 661–665.
22 Sklenar V, Piotto M, Leppik R & Saudek V (1993) Gra-
dient-tailored water suppression for
1
H-
15
N HSQC
experiments optimized to retain full sensitivity. J Magn
Reson 102, 241–245.
T. Murata et al. Gibberellin mimics peptide-antibody recognition
FEBS Journal 272 (2005) 4938–4948 ª 2005 FEBS 4947
23 Marion D & Wu
¨
thrich K (1983) Application of phase
sensitive two-dimensional correlated spectroscopy
(COSY) for measurements of
1
H-
1
H spin-spin coupling
constants in proteins. Biochem Biophys Res Commun
113, 967–974.
24 Marion D, Ikura M, Tschundin R & Bax A (1989)
Rapid recording of 2D NMR spectra without phase
cycling. Application to the study of hydrogen exchange
in proteins. J Magn Reson 85, 393–399.
25 Wu
¨
thrich K, Billetter M & Braun W (1983) Pseudo-
structures for the 20 common amino acids for use in

studies of protein conformations by measurements of
intramolecular proton-proton distance constraints with
nuclear magnetic resonance. J Mol Biol 169, 949–961.
26 Clore GM, Gronenborn AM, Nilges M & Ryan CA
(1987) Three-dimensional structure of potato carboxy-
peptidase inhibitor in solution. A study using nuclear
magnetic resonance, distance geometry, and restrained
molecular dynamics. Biochemistry 26, 8012–8013.
27 Clubb RT, Ferguson SB, Walsh CT & Wagner G (1994)
Three-dimensional solution structure of Escherichia coli
periplasmic cyclophilin. Biochemistry 33, 2761–2772.
28 Wagner G, Braun W, Havel TF, Schaumman T, Go N
&Wu
¨
thrich K (1987) Protein structures in solution by
nuclear magnetic resonance and distance geometry. The
polypeptide fold of the basic pancreatic trypsin inhibitor
determined using two different algorithms, DISGEO
and DISMAN. J Mol Biol 196, 611–639.
29 Bru
¨
nger AT, Adams PD, Clore GM, DeLano WL,
Gros P, Grosse-Kunstleve RW, Jiang J-S, Kuszewski
J, Nilges M, Pannu NS, Read RJ, Rice LM, Simon-
son T & Warren GL (1998) Crystallography & NMR
system: a new software suite for macromolecular
structure determination. Acta Crystallogr, Sect D 54,
905–921.
30 Laskowski RA, Rullmann JAC, MacArthur MW,
Kaptein R & Thornton JM (1996) AQUA and PRO-

CHECK-NMR: Programs for checking the quality of
protein structures solved by NMR. J Biomol NMR 8,
477–486.
31 Koradi R, Billeter M & Wu
¨
thrich K (1996) MOLMOL:
a program for display and analysis of macromolecular
structures. J Mol Graph 14 , 51–55.
32 Jones G, Willett P & Glen RC (1995) Molecular recog-
nition of receptor sites using a genetic algorithm with a
description of desolvation. J Mol Biol 245, 43–53.
33 Jones G, Willett P, Glen RC, Leach AR & Tayloe R
(1997) Development and validation of a genetic algo-
rithm for flexible docking. J Mol Biol 267, 727–748.
34 Nissink JWM, Murray CW, Hartshorn MJ, Verdonk
ML, Cole JC & Taylor RD (2002) A new test set for
validating predictions of protein–ligand interaction. Pro-
teins 49, 457–471.
35 Verdonk ML, Cole JC, Hartshorn MJ, Murray CW &
Taylor RD (2003) Improved protein-ligand docking
using GOLD. Proteins 52, 609–623.
36 Wallace AC, Laskowski RA & Thoronton JM (1995)
LIGPLOT: a program to generate schematic diagrams
of protein–ligand interactions. Protein Eng 8, 127–134.
Gibberellin mimics peptide-antibody recognition T. Murata et al.
4948 FEBS Journal 272 (2005) 4938–4948 ª 2005 FEBS