Structural basis for cyclodextrin recognition by
Thermoactinomyces vulgaris cyclo⁄ maltodextrin-binding
protein
Takashi Tonozuka
1
, Akiko Sogawa
1
, Mitsugu Yamada
2
, Naoki Matsumoto
1
, Hiromi Yoshida
2,3
,
Shigehiro Kamitori
2,3
, Kazuhiro Ichikawa
1
, Masahiro Mizuno
1,
*, Atsushi Nishikawa
1
and
Yoshiyuki Sakano
1
1 Department of Applied Biological Science, Tokyo University of Agriculture and Technology, Japan
2 Graduate School of Medicine, Kagawa University, Japan
3 Information Technology Center, Kagawa University, Japan
Cyclodextrins (CDs) are cyclic a-1,4-glucans, and the
central cavity of CDs can host a large number of che-
micals by hydrophobic interaction [1]. A thermophilic

mined. Like Escherichia coli maltodextrin-binding protein (EcoMBP) and
other bacterial sugar-binding proteins, TvuCMBP consists of two domains,
an N- and a C-domain, both of which are composed of a central b-sheet
surrounded by a-helices; the domains are joined by a hinge region contain-
ing three segments. c-Cyclodextrin is located at a cleft formed by the two
domains. A common functional conformational change has been reported
in this protein family, which involves switching from an open form
to a sugar-transporter bindable form, designated a closed form. The
TvuCMBP–c-cyclodextrin complex structurally resembles the closed form
of EcoMBP, indicating that TvuCMBP complexed with c-cyclodextrin
adopts the closed form. The fluorescence measurements also showed that
the affinities of TvuCMBP for cyclodextrins were almost equal to those for
maltooligosaccharides. Despite having similar folds, the sugar-binding site
of the N-domain part of TvuCMBP and other bacterial sugar-binding pro-
teins are strikingly different. In TvuCMBP, the side-chain of Leu59 pro-
trudes from the N-domain part into the sugar-binding cleft and orients
toward the central cavity of c-cyclodextrin, thus Leu59 appears to play the
key role in binding. The cleft of the sugar-binding site of TvuCMBP is also
wider than that of EcoMBP. These findings suggest that the sugar-binding
site of the N-domain part and the wide cleft are critical in determining the
specificity of TvuCMBP for c-cyclodextrin.
Abbreviations
CD, cyclodextrin; EcoMBP, Escherichia coli maltodextrin-binding protein; Mol, molecule; SeMet, selenomethionine; TliTMBP, Thermococcus
litoralis trehalose ⁄ maltose-binding protein; TvuCMBP, Thermoactinomyces vulgaris cyclo ⁄ maltodextrin-binding protein.
FEBS Journal 274 (2007) 2109–2120 ª 2007 The Authors Journal compilation ª 2007 FEBS 2109
of TVA I and TVA II [4,5] and TVA II complexed
with CDs [6]. To find the proteins physiologically rela-
ted to these enzymes, the flanking regions of the genes
were sequenced. A gene homologous to those of the
bacterial sugar-binding protein family was found to be

lose ⁄ maltose-binding protein (TliTMBP) [20], Pyrococ-
cus furiosus maltodextrin-binding protein [21] and
Alicyclobacillus acidocaldarius maltose ⁄ maltodextrin-
binding protein [22], have been determined. These pro-
teins share a common structural motif that consists of
two domains, joined by a hinge region, which sur-
round a sugar-binding site [11,15]. A common drastic
conformational change is found in this protein family,
which participates in switching from an open form to
a closed form [16]. In EcoMBP, the closed form has
been observed in the complexes with linear maltooligo-
saccharides, such as maltose, maltotriose and malto-
tetraose [17], and this form is capable of interacting
with the MalFGK
2
sugar-transporter complex. In
contrast, the open form does not have the ability to
perform the proper interaction with the MalFGK
2
sugar-transporter complex. Interestingly, EcoMBP
adopts the open form in the unliganded protein but
also in the complex with b-CD [18].
Here we present the crystal structure of TvuCMBP
complexed with c-CD. Unlike EcoMBP complexed
with b-CD, the Tvu CMBP–c-CD complex was deter-
mined as the closed form. The structure provides
evidence that the architecture of TvuCMBP is well
optimized for interacting with the central hydrophobic
cavity of c-CD.
Results and Discussion

be a flexible linker, which allows the proteins to inter-
act with carbohydrates as well as the membrane-bound
transport proteins [22,24].
An omit map shows that one c-CD binds to each
TvuCMBP molecule (Fig. 1A). Although noncrystallo-
graphic symmetry restraints were not applied in the
late stage of the refinement, c-CD was found to form
the same contacts with TvuCMBP in Mol-A–D. The
rmsd between Mol-A and Mol-B, Mol-A and Mol-C,
Mol-A and Mol-D are 0.77, 0.94, and 0.74 A
˚
, respect-
ively, for all atoms, and 0.42, 0.55, and 0.43 A
˚
,
respectively, for main-chain atoms, suggesting that the
four structures are almost identical. To facilitate des-
cription, the following depiction is based on Mol-A.
Overall structure of TvuCMBP
The bacterial sugar-binding proteins have been reported
to share a common structural motif [11,15]. Like other
bacterial sugar-binding proteins [15–22], TvuCMBP
consists of two domains, the N-domain (residues 17–127
Structure of cyclo ⁄ maltodextrin-binding protein T. Tonozuka et al.
2110 FEBS Journal 274 (2007) 2109–2120 ª 2007 The Authors Journal compilation ª 2007 FEBS
and 283–330) and the C-domain (residues 131–279 and
334–397) (Fig. 1B,C). Both domains have similar archi-
tectures; a b-sheet is located at the center, surrounded
by a-helices. The two domains are joined by a hinge
region, which contains three segments (residues 128–

bones of their N-domains are completely different
(Fig. 2B).
The C-domain parts of the sugar-binding sites
of TvuCMBP and related proteins
As the sugar-binding site is formed by N- and
C-domains, residues involved in sugar binding are
grouped into two parts, the N-domain part and the
Table 1. Data collection and refinement statistics.
Derivative (SeMet)
Native
Peak Edge
Data collection
Beamline PF BL-5 A PF-AR NW-12
Wavelength 0.97932 0.98000 1.0000
Space group C2 C2
Cell dimensions
a (A
˚
) 85.3 167.4
b (A
˚
) 49.3 95.3
c (A
˚
) 87.6 117.1
b (°) 94.9 131.6
Resolution range (A
˚
) 50–2.30 (2.38–2.30)
b

Protein 11 856
Ligand 352
Water 539
R
cryst
(%) 21.8
R
free
(%) 26.8
rmsd
Bond length (A
˚
) 0.007
Bond angles (°) 1.40
Average B
Protein 26.0
Ligand 31.5
Water 28.1
a
R
merge
¼ SS|Ii–<I>| ⁄S<I>0.
b
The values for the highest resolution shells are given in parentheses.
T. Tonozuka et al. Structure of cyclo ⁄ maltodextrin-binding protein
FEBS Journal 274 (2007) 2109–2120 ª 2007 The Authors Journal compilation ª 2007 FEBS 2111
C-domain. The structure of c-CD shows a sliced coni-
cal form; the OH-2 and OH-3 hydroxyl groups of all
glucose residues are positioned on one side, and the
OH-6 hydroxyl groups are located on the other side.

ilar conformations to Glc2¢ and Glc4¢ of maltotetraose
bound to EcoMBP (Fig. 4D). In the case of trehalose
(labeled from Glc4¢¢ to Glc5¢¢, as shown in Fig. 4C)
bound to TliTMBP, although the conformations of
Glc4 of TvuCMBP and corresponding glucose residues
bound to EcoMBP (Glc4¢) and TliTMBP (Glc4¢¢) are
similar, those between EcoMBP and TliTMBP are clo-
ser than those between TvuCMBP and TliTMBP, and
neither the first (Glc4¢¢) nor the second residues
(Glc5¢¢) of trehalose bound to TliTMBP strictly fit to
the glucose residues of c-CD bound to TvuCMBP.
These findings indicate that the sugar-binding mecha-
nisms of the C-domains of TvuCMBP and EcoMBP are
relatively conserved, whereas the different architecture of
the C-domain of Tli TMBP may be more s uitable for the
specific b inding to small oligosaccharides lik e trehalose.
Comparison of the N-domain parts of the
sugar-binding sites of TvuCMBP, EcoMBP
and TliTMBP
The N-domain part of the sugar-binding site consists of
three loops, region N-I (residues 25–33), region N-II
(residues 56–61) and region N-III (residues 80–85)
(Figs 3B and 5A). Compared with TvuCMBP, EcoMBP
and TliTMBP, the positions and the conformations of
the three regions are strikingly different (Fig. 5A–C).
Fig. 1. Three-dimensional structure of TvuCMBP complexed with
c-CD. (A) Stereoview of the omit map electron density for c-CD
bound to Mol-A with 2.0 r contoured level. The omit map was cal-
culated from the coefficients of the (F
obs

are from the hinge region (b-strands located at the bot-
tom of the sugar-binding cleft), appear to be most
responsible for the binding to trehalose (Fig. 5C). No
aromatic residues equivalent to Tyr121 and Trp295 of
TliTMBP are found in TvuCMBP and EcoMBP. The
primary structures of the sugar-binding sites of the
three proteins were aligned based on the structural
comparison (Fig. 6). In TliTMBP, the conserved resi-
dues are found to be few. Between TvuCMBP and
EcoMBP, many residues, including Leu and Trp, seem
to be conserved, but the positions and the conforma-
tions of the residues at regions N-I–III are different, as
described above.
The capacities of the sugar-binding sites of
TvuCMBP, EcoMBP and TliTMBP, where all of the
conformations are the closed form, were compared
(Fig. 7A–C). The cleft of the sugar-binding site of
TvuCMBP is the widest among the three sugar-binding
proteins (Fig. 7A). Although the side-chain of Leu59 is
located in the cleft, the sugar-binding site around
Leu59 is wide open. Lys229 and Glu361 form protru-
sions at the entrance of the cleft, and the distance
between Nf of Lys229 and Oe1 of Glu361 is 16.7 A
˚
.
On the other hand, the width of the sugar-binding cleft
of EcoMBP is apparently narrower than that of
TvuCMBP (Fig. 7B). Similar protrusions, which are
formed by Asp209 and Arg344, are observed at the
entrance of the cleft of EcoMBP, but the distance

values of K. oxytoca
CymE is highly specific for CDs, while TvuCMBP
Fig. 2. Superposition of the Ca backbones.
The figure was generated using
RASTOP. (A)
Stereoview of the Ca backbone of
TvuCMBP–c-CD complex (blue), which are
superimposed on those of EcoMBP–malto-
tetraose (yellow; PDB ID, 4MBP) and
TliTMBP–trehalose complex (magenta; PDB
ID, 1EU8). (B) Comparison of the Ca back-
bones of TvuCMBP–c-CD complex (blue)
and EcoMBP–b-CD (orange; PDB ID,
1DMB). C-domains of the two structures
were superimposed. CDs are represented
as stick models.
T. Tonozuka et al. Structure of cyclo ⁄ maltodextrin-binding protein
FEBS Journal 274 (2007) 2109–2120 ª 2007 The Authors Journal compilation ª 2007 FEBS 2113
shows the high affinities for not only CDs but also
higher maltooligosaccharides.
It is impossible to determine, however, whether
TvuCMBP adopts the open form or the closed form
with the sugars listed in Table 2 by this experiment.
Although the experimental conditions were different,
the K
d
values of EcoMBP for maltose, maltotriose,
maltotetraose, and b-CD are reportedly 1.0, 0.2, 1.6
and 1.0 lm, respectively [13], indicating that the K
d

illustrated.
Structure of cyclo ⁄ maltodextrin-binding protein T. Tonozuka et al.
2114 FEBS Journal 274 (2007) 2109–2120 ª 2007 The Authors Journal compilation ª 2007 FEBS
TliTMBP, both of which engage in binding to linear
maltooligosaccharides. The most remarkable feature is
that Leu59 protrudes into the sugar binding cleft
(Fig. 5A), which enables TvuCMBP to interact effi-
ciently with the hydrophobic cavity of CDs. The
hydrophobic environment provided by Leu58 and
Leu59 could also promote to incorporate CDs into
the sugar-binding cleft of TvuCMBP. In addition, the
wide cleft of TvuCMBP (Fig. 7A) is large enough to
accommodate CDs. These findings indicate that the
architecture of TvuCMBP is suitable for binding to
c-CD.
Fig. 4. Stereoview of four regions (regions C-I–C-IV) located at C-domain involving the sugar binding. Regions C-I, C-II, C-III and C-IV are blue,
yellow, magenta and red, respectively. The ligands shown in (A–C) are in gray. The figures were generated using
MOLSCRIPT [40] and
RASTER3D [42]. (A) TvuCMBP–c-CD complex. The glucose residues of c-CD are labeled from 1 to 8. (B) EcoMBP–maltotetraose complex. The
glucose residues of maltotetraose are labeled from 1¢ to 4¢. (C) TliTMBP–trehalose complex. The glucose residues of trehalose are labeled
from 4¢¢ to 5¢¢. (D) A superimposition of c-CD bound to TvuCMBP (blue) and maltotetraose bound to EcoMBP (orange). The two structures
(A) and (B), are superimposed and the portions of c-CD and maltotetraose are illustrated.
T. Tonozuka et al. Structure of cyclo ⁄ maltodextrin-binding protein
FEBS Journal 274 (2007) 2109–2120 ª 2007 The Authors Journal compilation ª 2007 FEBS 2115
The positions of the major aromatic residues located
at the C-domain part are conserved between
TvuCMBP and EcoMBP, but b-CD is not a proper lig-
and for EcoMBP, and in fact, glucose residues of
b-CD undergo stacking with aromatic residues derived
from C-domain, Tyr155, Trp230, and Trp340, whereas

are from the hinge region, are shown in
pink. The figures were generated using
MOLSCRIPT [40] and RASTER3D [42].
Fig. 6. Alignment of the primary structures
of regions N-I–III and C-I–IV. Identical amino
acid residues are shown in white on black.
The numbering of the amino acid sequences
is given.
Structure of cyclo ⁄ maltodextrin-binding protein T. Tonozuka et al.
2116 FEBS Journal 274 (2007) 2109–2120 ª 2007 The Authors Journal compilation ª 2007 FEBS
was also employed. A DNA fragment encoding the mature
TvuCMBP was prepared by polymerase chain reaction
using a plasmid, pTP-TVE [7], and oligonucleotides, 5¢-
GGG AAT TCC ATA TGT GCG GGC CAA AGC GGG
ATC CC-3¢ and 5¢-GTT TTC CCA GTC ACG ACG TTG
T-3¢, which have restriction sites of NdeI and EcoRI sites,
respectively, to facilitate cloning of the fragment. The
amplified fragment was digested with the enzymes NdeI and
EcoRI, and inserted into the NdeI and EcoRI sites of
pET21a, resulting in the plasmid pETCBP. The sequence of
the construct was verified by DNA sequencing.
Preparation of TvuCMBP
To produce TvuCMBP, E. coli BL21(DE3) harboring
pETCBP was cultured in Luria-Bertani medium containing
ampicillin (50 lgÆmL
)1
)toA
600
¼ 0.6–0.9, induced with iso-
propyl b-d-thiogalactopyranoside to a final concentration

using a protocol similar to that of the native protein.
Crystallization and data collection
Crystals were grown by the hanging drop vapor diffusion
method at 20 °C. Crystals of TvuCMBP complexed with
c-CD were obtained by mixing 1 lL of well solution (25%
polyethylene glycol 6000, 0.1 m Mes pH 6.25, 5 mm c-CD)
and 1 lL of protein solution (10 mgÆmL
)1
TvuCMBP).
Crystals of SeMet-substituted TvuCMBP were obtained
with the same procedure. The crystals were transferred to a
solution consisting of 30% polyethylene glycol 6000, 0.1 m
Mes pH 6.25, 5 mm c-CD, and frozen in a 100 K nitrogen
stream. A native diffraction data set was collected at the
PF-AR NW-12 beamline (Tsukuba, Japan). Data were
processed with the program hkl2000 [34]. An attempt to
solve the structure by molecular replacement, using various
sugar-binding proteins, such as TliTMBP [20], EcoMBP
Fig. 7. Surface models of the sugar-binding sites of TvuCMBP,
EcoMBP and TliTMBP. Sugars are drawn in red sticks. The figures
were generated using
PYMOL. (A) TvuCMBP–c-CD complex. Leu59,
Lys229 and Glu361 are indicated in orange or magenta. (B)
EcoMBP–maltotetraose complex. Trp62, Asp209 and Arg344 are
indicated in cyan or yellow. (C) TliTMBP–trehalose complex.
Table 2. K
d
values of TvuCMBP for various sugars by measuring
changes in fluorescence.
Ligand K

adequate structures, the model was manually built with the
program xfit in the xtalview package [38]. The refinement
was performed with the program cns [39]. Although the
model of the SeMet-substituted TvuCMBP was initially built,
a high R
free
value (32%) was yielded after placing all of the
residues, water molecules and c-CD, probably because of the
high mosaicity of the MAD data set. Thus, the native data
set was used for further refinement. Molecular replacement
was carried out using the program molrep of CCP4 [23]
with the rough model of the SeMet derivative as a probe
model. Four TvuCMBP molecules were in an asymmetric
unit and further refined with CNS. Figures were prepared
using xtalview, pymol ( />rastop (einfinity.org/rastop/), swiss-pdb
viewer [26], molscript [40], ligplot [41] and raster3d [42].
The atomic coordinates and structural factors (code 2DFZ)
have been deposited in the Protein Data Bank (http://
www.rcsb.org/).
Fluorescence measurements
To remove c-CD, which was derived from the purification
procedure, the purified TvuCMBP was denatured at a con-
centration of 0.1 mgÆmL
)1
in 2.5 m guanidine hydrochloride,
20 mm sodium phosphate buffer (pH 6.0) at 37 °C. The
denatured TvuCMBP was then dialyzed against 20 mm
sodium phosphate buffer (pH 6.0). To confirm that the
renaturation was completed, the circular dichroism spectra
of each step were monitored using a Jasco J-720WI spectro-

References
1 Davis ME & Brewster ME (2004) Cyclodextrin-based
pharmaceutics: past, present and future. Nat Rev Drug
Discov 3, 1023–1035.
2 Tonozuka T, Ohtsuka M, Mogi S, Sakai H, Ohta T &
Sakano Y (1993) A neopullulanase-type a-amylase from
Thermoactinomyces vulgaris R-47. Biosci Biotechnol
Biochem 57, 395–401.
3 Tonozuka T, Mogi S, Shimura Y, Ibuka A, Sakai H,
Matsuzawa H, Sakano Y & Ohta T (1995) Comparison
of primary structures and substrate specificities of two
pullulan-hydrolyzing a-amylases, TVA I and TVA II,
from Thermoactinomyces vulgaris R-47. Biochim Biophys
Acta 1252, 35–42.
4 Kamitori S, Kondo S, Okuyama K, Yokota T, Shimura
Y, Tonozuka T & Sakano Y (1999) Crystal structure of
Thermoactinomyces vulgaris R-47 a-amylase II (TVA II)
hydrolyzing cyclodextrins and pullulan at 2.6 A
˚
resolu-
tion. J Mol Biol 287, 907–921.
5 Kamitori S, Abe A, Ohtaki A, Kaji A, Tonozuka T &
Sakano Y (2002) Crystal structures and structural com-
parison of Thermoactinomyces vulgaris R-47 a-amylase I
(TVAI) at 1.6 A
˚
resolution and a-amylase II (TVAII) at
2.3 A
˚
resolution. J Mol Biol 26, 443–453.

of Escherichia coli: functional significance in active
transport. J Biol Chem 272, 17615–17622.
14 Fiedler G, Pajatsch M & Bo
¨
ck A (1996) Genetics of a
novel starch utilisation pathway present in Klebsiella
oxytoca. J Mol Biol 256, 279–291.
15 Spurlino JC, Lu GY & Quiocho FA (1991) The 2.3-A
˚
resolution structure of the maltose- or maltodextrin-
binding protein, a primary receptor of bacterial active
transport and chemotaxis. J Biol Chem 266, 5202–5219.
16 Sharff AJ, Rodseth LE, Spurlino JC & Quiocho FA
(1992) Crystallographic evidence of a large ligand-
induced hinge-twist motion between the two domains
of the maltodextrin binding protein involved in active
transport and chemotaxis. Biochemistry 31,
10657–10663.
17 Quiocho FA, Spurlino JC & Rodseth LE (1997) Exten-
sive features of tight oligosaccharide binding revealed in
high-resolution structures of the maltodextrin
transport ⁄ chemosensory receptor. Structure 5, 997–1015.
18 Sharff AJ, Rodseth LE & Quiocho FA (1993) Refined
1.8-A
˚
structure reveals the mode of binding of b-cyclo-
dextrin to the maltodextrin binding protein. Biochemis-
try 32, 10553–10559.
19 Duan X, Hall JA, Nikaido H & Quiocho FA (2001)
Crystal structures of the maltodextrin ⁄ maltose-binding

25 Holm L & Sander C (1993) Protein structure compari-
son by alignment of distance matrices. J Mol Biol 233,
123–138.
26 Guex N & Peitsch MC (1997) SWISS-MODEL and the
Swiss-PdbViewer: an environment for comparative pro-
tein modeling. Electrophoresis 18, 2714–2723.
27 Pajatsch M, Gerhart M, Peist R, Horlacher R, Boos W
&Bo
¨
ck A (1998) The periplasmic cyclodextrin binding
protein CymE from Klebsiella oxytoca and its role in
maltodextrin and cyclodextrin transport. J Bacteriol
180, 2630–2635.
28 Miller DM, Olson JS, Pflugrath JW & Quiocho FA
(1983) Rates of ligand binding to periplasmic proteins
involved in bacterial transport and chemotaxis. J Biol
Chem 258, 13665–13672.
29 Uitdehaag JC, Kalk KH, van Der Veen BA, Dijkhuizen
L & Dijkstra BW (1999) The cyclization mechanism of
cyclodextrin glycosyltransferase (CGTase) as revealed
by a gamma-cyclodextrin-CGTase complex at 1.8-A
˚
resolution. J Biol Chem 274, 34868–34876.
30 van der Veen BA, Uitdehaag JC, Dijkstra BW &
Dijkhuizen L (2000) Engineering of cyclodextrin glyco-
syltransferase reaction and product specificity. Biochim
Biophys Acta 1543, 336–360.
31 Ohtaki A, Kondo S, Shimura Y, Tonozuka T,
Sakano Y & Kamitori S (2001) Role of Phe286 in
the recognition mechanism of cyclomaltooligosacchar-

¨
nger AT, Adams PD, Clore GM, DeLano WL, Gros
P, Grosse-Kunstleve RW, Jiang JS, Kuszewski J, Nilges
M, Pannu NS et al. (1998) crystallography &; nmr
system: a new software suite for macromolecular struc-
ture determination. Acta Crystallogr D 54, 905–921.
40 Kraulis PJ (1991) molscript: a program to produce
both detailed and schematic plots of protein structures.
J Appl Crystallogr 24, 946–950.
41 Wallace AC, Laskowski RA & Thornton JM (1995)
ligplot: a program to generate schematic diagrams of
protein–ligand interactions. Protein Eng 8, 127–134.
42 Merritt EA & Murphy MEP (1994) raster3d, Version
2.0. A program for photorealistic molecular graphics.
Acta Crystallogr D 50, 869–873.
43 Hiromi K, Tanaka A & Ohnishi M (1982) Fluorometric
studies on the binding of gluconolactone, glucose, and
glucosides to the subsites of glucoamylase. Biochemistry
21, 102–107.
Structure of cyclo ⁄ maltodextrin-binding protein T. Tonozuka et al.
2120 FEBS Journal 274 (2007) 2109–2120 ª 2007 The Authors Journal compilation ª 2007 FEBS