THE JOURNAL OF BIOLOGICAL CHEMISTRY
9875–9884, 2003
© 2003 by The American Society for Biochemistry and Molecular Biology, Inc.
Differential Regulation of a Hyperthermophilic ␣-Amylase with a
Novel (Ca,Zn) Two-metal Center by Zinc*
Received for publication, November 6, 2002, and in revised form, December 11, 2002
Published, JBC Papers in Press, December 12, 2002, DOI 10.1074/jbc.M211339200
Anni Linden‡, Olga Mayans‡§, Wolfram Meyer-Klaucke‡, Garabed Antranikian¶,
and Matthias Wilmanns‡储
From the ‡European Molecular Biology Laboratory, Notkestrasse 85, D-22603 Hamburg, Germany and the ¶Department
of Technical Microbiology, Technical University Hamburg-Harburg, Kasernenstrasse 12, D-21073 Hamburg, Germany
The crystal structure of the ␣-amylase from the hyper-
thermophilic archaeon Pyrococcus woesei was solved in
the presence of three inhibitors: acarbose, Tris, and
zinc. In the absence of exogenous metals, this ␣-amylase
bound 1 and 4 molar eq of zinc and calcium, respec-
tively. The structure reveals a novel, activating, two-
metal (Ca,Zn)-binding site and a second inhibitory zinc-
binding site that is found in the ⴚ1 sugar-binding pocket
within the active site. The data resolve the apparent
paradox between the zinc requirement for catalytic ac-
tivity and its strong inhibitory effect when added in
molar excess. They provide a rationale as to why this
␣-amylase, in contrast to commercially available ␣-amy-
lases, does not require the addition of metal ions for full
catalytic activity, suggesting it as an ideal target to max-
imize the efficiency of industrial processes like liquefac-
tion of starch.
␣-Amylases (␣-1,4-D-glucan 4-glucanohydrolase, EC 3.2.1.1)
are endo-acting hydrolases that randomly cleave the ␣-1,4-
glycosidic linkages of branched and linear carbohydrates such
as amylopectin and amylose in starch and glycogen (1–3). Their
widespread occurrence in various organisms and the consump-
tion of their substrates for food reserves and energy sources
have led to intense interest in their biomedical properties and
to major biotechnological applications in industry (4 – 6). Their
value in specific industrial processes depends critically on their
pH and optimal temperature, which vary depending on the
organism of origin. Nearly all ␣-amylases belong to class-13 of
glycosylhydrolases by virtue of their characteristic sequence
motifs. However, in general, their overall sequence similarity is
low, which, for some members, falls below the 10% level (1–3,
7). A number of crystal structures of class-13 members have
been solved, comprising ␣-amylases from organisms inhabiting
environments that span a large temperature range, from psy-
chrophilic to hyperthermophilic (3). These ␣-amylases share a
common three-domain fold with the catalytic activity located on
the C-terminal face of a central (␤␣)8-barrel, domain A. The
* The costs of publication of this article were defrayed in part by the
payment of page charges. This article must therefore be hereby marked
“advertisement” in accordance with 18 U.S.C. Section 1734 solely to
indicate this fact.
The atomic coordinates and structure factors (code 1MWO, 1MXD,
and 1MXG) have been deposited in the Protein Data Bank, Research
Collaboratory for Structural Bioinformatics, Rutgers University, New
Brunswick, NJ (http://www.rcsb.org/).
§ Present address: Biozentrum, University of Basel, Klingelbergstr.
70, CH-4056 Basel, Switzerland.
储 To whom correspondence should be addressed: EMBL, Hamburg
Outstation, c/o DESY, Notkestr. 85, D-22603 Hamburg, Germany.
Tel.: 49-40-89902-126; Fax: 49-40-89902-149; E-mail: wilmanns@
embl-hamburg.de.
This paper is available on line at http://www.jbc.org
Vol. 278, No. 11, Issue of March 14, pp.
Printed in U.S.A.
sequences of the other two domains, B and C, are void of any
conserved motifs and are not involved directly in substrate
catalysis.
Besides the recently solved three-dimensional structure of
the glycosyltrehalose trehalosidase from Sulfolobus solfatari-
cus Km1 that exhibits ␣-amylase activity (8), no other struc-
tures of archael class-13 ␣-amylases are known to date. The
␣-amylase from the hyperthermophilic archaeon Pyrococcus
Downloaded from www.jbc.org by guest, on November 14, 2010
woesei (PWA),1 which is identical to that from Pyrococcus fu-
riosus (9), was cloned and classified as a class-13 glycosylhy-
drolase (10). Throughout this study, both the ␣-amylases of
P. woesei and P. furiosus will be referred to as PWA. Biochem-
ical characterization of recombinant PWA revealed a high ther-
mal stability and maximum catalytic activity at ⬃100 °C (10 –
12). In a recent study using coupled plasma-atomic emission
spectroscopy, it was shown that PWA binds calcium and zinc
stoichiometrically and with high affinity (13). Thus, in contrast
to many other ␣-amylases, PWA activity does not require the
addition of metal ions. Due to its superior properties, exceeding
those of ␣-amylases currently employed in commercial prepa-
rations (5, 14), PWA has become a prime candidate for maxi-
mizing the efficiency of applications in the starch industry. To
reveal the molecular basis of its properties, we solved the x-ray
structure of PWA in the presence of three different active site
ligands. The structure reveals a novel (Ca,Zn)-binding site in
close proximity to the active site cleft, which is not found in
␣-amylases of any bacteria, plants, and most other Archaea.
The presence of this site indicates adaptive evolution of PWA to
the specific living conditions of P. woesei. The three complex
structures also show how competitive binding of organic com-
pounds and zinc provides a direct molecular explanation as to
why a molar excess of zinc or some chemically related metals
inhibits PWA activity.
EXPERIMENTAL PROCEDURES
Gene Amplification and Expression—The wild-type gene encoding
PWA was amplified by PCR using primers deduced from the open
reading frame of the P. furiosus wild-type ␣-amylase gene (10). An NcoI
recognition site (underlined) was fused to the 5⬘-end of the sense primer
(5⬘-CAT GCC ATG GAC ATA AAG AAA TTA ACA CCC CTC-3⬘), as was
an XhoI recognition site (underlined) to the antisense primer (5⬘-GGC
CTC GAG TCA CCC AAC ACC ACA ATA ACT CCA-3⬘). Additionally,
the natural GTG start codon was replaced with the Escherichia coli
start codon ATG (boldface). The PCR product was digested twice with
NcoI/XhoI and cloned into the NcoI/XhoI cloning site of the pET-15b
vector (Novagen, Madison, WI) to generate the plasmid pPWA. pPWA
was cloned and expressed in E. coli strain BL21(DE3) (15) as described
previously (12). The deduced amino acid sequence is identical to the
1
The abbreviations used are: PWA, P. woesei ␣-amylase; PIXE, pro-
ton-induced x-ray emission; MES, 4-morpholineethanesulfonic acid; Ac,
acarbose; BLA, B. licheniformis ␣-amylase.
9875
9876 Structure of a (Ca,Zn) Metal Center-containing ␣-Amylase
␣-amylase gene from P. furiosus (10) and to the P. woesei amylase
sequence released to the NCBI Protein Database by Lu et al.2
Purification and Concentration Determination—The recombinant
␣-amylase gene product was purified from catalytically active inclusion
bodies (12). Inclusion bodies were solubilized in 0.12 M Britton and
Robinson buffer (pH 12) and purified by hydrophobic interaction chro-
matography using a phenyl-Superose HR 5/5 column (Amersham Bio-
sciences, Freiburg, Germany). The protein was eluted in 0.02 M trieth-
anolamine containing 30% (v/v) isopropyl alcohol (pH 7.2). Fractions
containing amylolytic activity at 99 °C were pooled and further purified
by size-exclusion chromatography using a Superdex 200 16/60 prep-
grade column. The amylase was eluted in 0.02 M triethanolamine (pH
7.2) and concentrated to a final concentration of 5.5 mg/ml in an Amicon
ultrafiltration device (Millipore Corp., Eschborn, Germany). The con-
centration of the enzyme preparation was determined by estimating the
extinction coefficient according to the method of Gill and von Hippel
(16).
Metal Analysis—The metal content of isolated PWA was determined
by detection of the proton-induced x-ray emission (PIXE) at the Leipzig
2 MeV proton microprobe (17). Using this technique, all elements be-
sides those lighter than sodium can be detected in a single scan at a
minimum weight limit of ⬃1–10 ppm. The measurements were cali-
brated using the nine sulfur atoms of the PWA sequence (five cysteines
and four methionines) as an internal standard. Protein samples were
dialyzed extensively against Chelex 100-treated buffer composed of 10
mM sodium phosphate (pH 12) to remove chloride and sulfur compounds
that disturb the protein sulfur signal. The samples (⬃2 ␮l, 1 mg/ml)
were dropped onto sample holders covered with a 0.9 ␮m polyethylene
terephthalate foil. A sample area of ⬃1 mm2 was scanned with the
proton microbeam, and the characteristic x-rays were detected using a
germanium detector. A total charge of ⬃0.17 coulomb was applied
within 20 min, ensuring that no elemental loss (sulfur or metals) due to
thermal stress occurred, which would otherwise have changed the x-ray
signal intensity. The number of metals per molecule was calculated
from their calibrated signals with reference to the calibrated sulfur
signal. The overall accuracy was estimated by taking into account the
statistical errors of the x-ray yield and the uncertainty in the correction
for x-ray absorption within the sample.
Crystallization—Purified recombinant PWA was crystallized at a
concentration of 5.5 mg/ml by the hanging drop vapor diffusion method.
PWA䡠Zn was grown from 1 ␮l of protein solution and 2 ␮l of 0.1 M MES
(pH 6.5) containing 0.01 M Zn2(SO4)䡠7H2O and 25% (v/v) polyethylene
glycol monomethyl ether 550 at 19 °C. The PWA䡠Ac/Zn complex was
grown from 1 ␮l of acarbose solution (11 mg/ml), 1 ␮l of protein solution,
and 1 ␮l of 0.1 M sodium cacodylate (pH 6.5) containing 0.05 M Zn(OAc)2
and 35% (v/v) 2-methyl-2,4-pentanediol at 25 °C. The PWA䡠Tris com-
plex was crystallized from 1 ␮l of acarbose solution (11 mg/ml), 1 ␮l of
protein solution, and 1 ␮l of 0.1 M Tris (pH 8.5) containing 0.05 M MgCl2
and 40% (v/v) ethanol. A heavy atom derivative suitable for crystallo-
graphic phase determination was obtained by soaking PWA䡠Zn in a
solution containing 29% (v/v) polyethylene glycol monomethyl ether 550
and 1.0 mM CH3HgCl for 1.5 h at 19 °C prior to x-ray data collection.
X-ray Data Collection, Processing, and Reduction—Each x-ray data
collection was performed with one single crystal using 27% (v/v) poly-
ethylene glycol monomethyl ether 550 as a cryoprotectant for PWA䡠Zn.
For PWA䡠Tris, 40% (v/v) ethanol and 31.5% (v/v) polyethylene glycol 400
were used as a cryoprotectant. For PWA䡠Ac/Zn, no transfer into a
cryoprotectant was carried out. All crystals were mounted onto nylon
cryo-loops (Hampton Research, Riverside, CA) and shock-frozen in a
nitrogen stream at 100 K. X-ray data sets for the PWA䡠Zn, PWA䡠Ac/Zn,
and PWA䡠Tris complexes were recorded up to 2.2-, 2.0-, and 1.5-Å
resolution, respectively. An anomalous diffraction data set to 2.5-Å
resolution was collected from a PWA mercury derivative isoform at a
wavelength above the L-III adsorption edge of mercury. Further data
statistics are presented in Table I. The data were processed, merged,
and scaled using the HKL program suite (18).
Structure Determination, Refinement, and Model Evaluation—Pro-
grams used for the subsequent calculations were from the CCP4 pro-
gram suite (19), unless stated otherwise. Initial phases for PWA䡠Zn
were obtained using single isomorphous replacement including the
one-wavelength anomalous scattering contribution (SIRAS). Heavy
atom positions were determined from mercury derivative difference
Patterson maps using the RSPS program. From three major binding
sites, initial phases were calculated and refined using SHARP (20) up to
2.5-Å resolution. Phases were improved by solvent flattening with
2
C. Lu, J. Weizheng, and Y. Yunyan, unpublished data.
SHARP. An initial map calculated in CNS (21) was used to build the
model of the molecule using the interactive graphics program O (22).
The model was traced with the help of Bacillus licheniformis ␣-amylase
(BLA) as a template. The final model was built in a consecutive cycle of
crystallographic refinement using CNS and manual rebuilding. A total
number of 25,967 reflections in the resolution range of 20 –2.22 Å were
included in the refinement, with 3.8% (995 reflections) set aside for
cross-validation (23). The initial R-value was 44.2%, and the free R-
factor was 42.1%. After several cycles of positional and restrained
individual B-factor refinement, solvent building was performed using
the solvent 0 mode of ArpWArp (24) and CNS with waters placed in
local maxima of difference electron density maps above 3 ␴. The struc-
tures of the two PWA䡠inhibitor complexes (PWA䡠Tris and PWA䡠Ac/Zn)
were solved by molecular replacement using the structure of PWA䡠Zn as
a search model. Rotation and translation functions were calculated with
AMoRe (25) using x-ray data from 10 to 2.5 Å, resulting in a single
solution with correlation coefficients of 68.6 and 80.1% and R-factors of
39.4 and 33.5% for PWA䡠Tris and PWA䡠Ac/Zn, respectively. A test set of
1686 reflections (4.7%, PWA䡠Tris) and 1196 reflections (1.7%, PWA䡠Ac/
Zn) for the calculation of Rfree was excluded during refinement in CNS.
Initial phases of the PWA䡠Tris model were applied to automated model
building using ArpWArp for model completion. The models were veri-
fied and corrected, and acarbose and Tris molecules were built manu-
ally using the graphics software program O. Solvent building and
subsequent refinement were performed as described above.
Downloaded from www.jbc.org by guest, on November 14, 2010
Metal atoms were identified from high peaks in the difference Fou-
rier maps and in part from additional high peaks in the anomalous
difference Fourier maps. Metals were refined as calcium when no
higher peak level appeared in the anomalous electron density map and
the coordination geometry agreed with that of known protein䡠calcium
complexes. They were refined as magnesium sites when an excess of
magnesium was present during crystallization and a negative Fo ⫺ Fc
difference Fourier electron density peak appeared when the metal po-
sition was refined as calcium. Metals were refined as zinc when the
crystal growth conditions contained zinc ions and if the metals sites
showed a high anomalous peak emerging from data set collection at
12.4 keV (␭ ⫽ 0.91 Å), which is above the k-adsorption edge of zinc of
9.59 keV (␭ ⫽ 1.28 Å).
RESULTS
Overall Structure and Ligand Composition—We have solved
the structure of PWA in three different forms using experimen-
tal phases from a mercury derivative. The structures are in the
presence of three different inhibitors: 1) zinc (PWA䡠Zn); 2) the
specific substrate analog acarbose and zinc (PWA䡠Ac/Zn); and
3) Tris, which was used as a buffer for crystallization trials in
the absence of zinc (PWA䡠Tris). They have been refined to 2.2-,
2.0-, and 1.6-Å resolution, respectively. PWA displays the ca-
nonical glycosylhydrolase class-13 fold that is composed of
three domains, A–C (Fig. 1). Comparison of the overall struc-
ture of PWA with that of other known ␣-amylases using the
program DALI (26) revealed the highest structural similarity
to the homologous enzymes from the hyperthermophile B. li-
cheniformis (BLA; 29% sequence identity; root mean square
deviation of 2.5 Å based on comparison of the C␣ backbone
trace) and from barley (28% sequence identity; root mean
square deviation of 1.8 Å) (Fig. 2).
As in other ␣-amylases, the central domain A, covering res-
idues 1–109 and 170 –340, is folded as a (␤␣)8-barrel and con-
tains the active site at its C-terminal face. Domain B (residues
111–169) inserts between ␤-strand 3 and ␣-helix 3a of domain
A, thus forming part of the active site cleft. Its secondary
structure is limited to two short ␤-strands, forming a small
antiparallel ␤-sheet and a short 310-helix (Figs. 1 and 2). At
their interface, domains A and B comprise a novel (Ca,Zn)
metal center, which is described further below (see Fig. 4). A
disulfide bridge is formed by two consecutive cysteine residues
(Cys153 and Cys154) in close vicinity to the zinc-binding site of
this two-metal center. Covering a range of 58 residues only,
domain B of PWA is one of the smallest ␣-amylase B domains,
whereas other members of the family span ⬎100 residues (2, 3).
Among those with a known structure, domain B of BLA is most
 Structure of a (Ca,Zn) Metal Center-containing ␣-Amylase 9877

FIG. 1. Overall crystal structure of

PWA complexed with different li-
gands. The cylinders and arrows repre-
sent helices and ␤-strands, respectively.
Domains A–C are colored cyan, magenta,
and brown, respectively. For clarity, the
␤-strands composing the ␤-barrel of do-
main A are colored gray. Bound metals
(zinc, green; and magnesium and calcium,
orange) and acarbose molecules are indi-
cated and numbered according to Table
II. The backbone of the acarbose mole-
cules is yellow; oxygen atoms are red; and
nitrogen atoms are blue. A, top view of the

Downloaded from www.jbc.org by guest, on November 14, 2010

PWA䡠Ac/Zn complex. The active site cleft
at the front face of the PWA molecule
contains two inhibitors with partial occu-
pancies, the first of which is acarbose and
the second of which is a coordinated zinc
ion, which is virtually identical to the
nitrogen position of the 4-amino-4,6-
dideoxy-␣-D-glucose ring of acarbose. The
two inhibitors are superimposed onto
each other. B, PWA䡠Ac/Zn rotated 90°
along a horizontal axis with respect to the
orientation in A. C, PWA䡠Tris shown from
the same orientation as in A. Three zinc-
binding sites (Zn3, Zn5, and Zn6) of
PWA䡠Ac/Zn are replaced by magnesium
ions in PWA䡠Tris (Mg3, Mg5, and Mg6). In
PWA䡠Tris, no metal is found in site 4 of
the PWA䡠Ac/Zn structure.

similar (root mean square deviation of 2.3 Å based on the C␣
backbone) (27) to the corresponding domain of PWA. In con-
trast, domain B of BLA contains 43 additional residues that
form a second ␤-sheet (27). The C-terminal domain C (residues
341– 435) is arranged in an eight-stranded antiparallel ␤-sheet
containing a Greek key motif. The function of this domain still
remains unclear.
We initially identified the nature of the bound metal ions by
PIXE. For these experiments, PWA was expressed heterolo-
gously in E. coli and purified in the absence of exogenous
metals except for sodium. The PIXE data revealed the presence
of 1.1 ⫾ 0.4 molar eq of zinc and 4 ⫾ 2 molar eq of calcium
bound to PWA. In the PWA crystal structures, the type of the
metal sites was characterized by the analysis of anomalous
x-ray data and positive peaks in Fo ⫺ Fc difference Fourier
electron density maps (Table I). Based on these data, the PWA
structures show excessive metal binding; all three structures
have five metal-binding sites in common (Table II). The two
crystal forms grown in the presence of zinc (PWA䡠Zn and
PWA䡠Ac/Zn) reveal two additional metal sites, bringing the
total in these structures to seven. The presence of an anoma-
lous signal under the experimental conditions of x-ray data
collection was used to indicate the presence of zinc (Tables I
and II). In the PWA䡠Tris structure, solved from crystals grown
in the absence of zinc, but in the presence of magnesium, only
one site showed a significant anomalous signal and therefore
was refined as a zinc site (Table II). The remaining four sites,
without an anomalous signal, but retaining strong positive
difference electron density, when refined as an ordered solvent
(⬎10 ␴), were interpreted to be calcium or magnesium sites.
Thus, the stoichiometry of calcium (or magnesium) and zinc
sites observed in the PWA䡠Tris structure is in good agreement
with the PIXE data. In contrast, in the PWA䡠Zn and
PWA䡠Ac/Zn structures, all sites except one showed a significant
anomalous signal and were therefore refined as zinc sites. The
remaining site was interpreted as a calcium site. These metal
9878 Structure of a (Ca,Zn) Metal Center-containing ␣-Amylase

FIG. 2. Structure-based alignment of PWA sequences. The secondary structure assignments for PWA are shown as colored cylinders Downloaded from www.jbc.org by guest, on November 14, 2010
(310-helices and ␣-helices) and arrows (␤-strands) above the aligned sequences. AVA, Hordeum vulgare ␣-amylase (AMY2) chain A (Protein Data
Bank code 1AVA) (46). The Protein Data Bank code for BLA is 1BLI (27). The positions of amino acid residues coordinating the zinc ion in domain
B (site 1; cf. Table II) are highlighted in red; those coordinating calcium (site 2; cf. Table II) in domain B are colored yellow; and the cysteines
forming disulfide bonds are colored green. The seven invariant amino acid residues are marked in black. The alignment was carried out with DALI
using the main chain positions of each coordinate set.

sites will be referred to as activating sites (Table II, sites 1 and
2), an inhibiting site (site 7), and other sites (sites 3– 6).
Chemically Unrelated Inhibitors Competitively Bind to the
Active Site—We quantified the inhibitory effects of a number of
established ␣-amylase active site ligands, including the transi-
tion state analog acarbose and the buffer Tris (Table III). To
link the observed tight binding of calcium and zinc to a poten-
tial role in activity regulation, we also assessed the regulatory
properties of a number of metal ions (Table III). Except for
copper (full inhibition in the presence of 3 mM Cu2⫹), zinc had
the strongest inhibitory effect. In the presence of ⱖ3 mM zinc,
residual PWA activity was ⬍10%. In contrast to many other
characterized ␣-amylases, including commercially available
BLA (10, 13), PWA was not activated significantly by an excess
of calcium (Table III). Based on the measured inhibition data,
we selected three ligands (acarbose, Tris, and zinc) for crystal-
lographic characterization of the PWA active site (Fig. 3).
Acarbose is a pseudotetrasaccharide inhibitor consisting of a
valienamine unit at the nonreducing end linked to 4-amino-
4,6-dideoxy-␣-D-glucose, which is fused to maltose. In the
PWA䡠Ac/Zn structure, resulting from co-crystallization of PWA
in the presence of acarbose and zinc, one acarbose molecule is
 Structure of a (Ca,Zn) Metal Center-containing ␣-Amylase 9879
TABLE I
X-ray data and structure refinement statistics
PWA 䡠 Zn PWA 䡠 Zn (Hg soak) PWA 䡠 Tris PWA 䡠 Ac/Zn

Data collection
Beamlinea X11 BW7A BW7B BW7B
Wavelength (Å)/energy (keV)b 0.9102/13.62 1.0008/12.39 0.8453/14.67 0.8453/14.67
Resolution range (Å) 20.0–2.22 20.0–2.45 99.0–1.52 99.0–1.97
Highest resolution shell (Å) 2.28–2.22 2.51–2.45 1.55–1.52 2.02–1.97
Unit cell dimensions (Å),
space group P212121
a 62.9 62.9 51.5 62.8
b 78.2 78.2 76.5 77.2
c 106.3 106.3 136.5 106.6
No. of unique reflections 26,004 19,410 75,292 35,302
Multiplicity 3.4 3.5 8.5 6.5
具I典/具␴ (I)典b 10.2 (3.0) 14.7 (4.9) 39.9 (6.1) 18.8 (3.8)
Rsym (%)b,c 6.0 (17.8) 6.5 (21.4) 4.0 (20.9) 7.5 (32.0)
Completenessb 97.7 (93.1) 96.3 (87.4) 89.8 (69.0) 94.4 (88.1)
Refinement
Refinement range (Å) 20–2.2 20–1.6 20–2.0
R factor (%)d 19.0 15.9 19.4
Rfree (%) 22.4 17.2 21.9
No. of non-hydrogen
atoms/asymmetric unit
Total 3975 4252 4007
Protein 3570 3574 3574

Downloaded from www.jbc.org by guest, on November 14, 2010

Solvent 398 509 289
Other ligands 7 169 144
r.m.s.d.e from target values
Bonds (Å) 0.006 0.005 0.007
Angles 1.32° 1.27° 1.34°
Average B-factors (Å2)
Main chain 23.9 15.4 29.8
Side chain 25.4 16.3 30.8
All atoms 26.4 18.3 31.2
a
All beamlines are at the DORIS storage ring at EMBL/DESY (Hamburg, Germany).
b
Numbers in parentheses refer to the highest resolution shell of observed data.
c
Rsym ⫽ ⌺h⌺j 兩Ihj ⫺ 具Ih典兩/⌺h⌺j 兩Ihj兩, where h represents a unique reflection and j is symmetry equivalent indices. I is the observed intensity, and
具I典 is the mean value of I.
d
Rfactor ⫽ ⌺h 储Fo(h)兩 ⫺ 兩Fc(h)储/⌺h 储Fo(h)兩, where Fo and Fc are the observed and calculated structure factors, respectively.
e
Root mean square deviation.

bound to the active site at the C-terminal face of the (␤␣)8-
barrel of domain A. It interacts with a number of highly con-
served residues that reside in loops connecting ␤-strands 4, 5,
and 7 of the (␤␣)8-barrel with the subsequent helices (Figs. 2
and 3, A and B) (2, 3, 28). Three of its four sugar rings (A–C),
comprising the acarviosine residue and a linked glucose ring,
are visible; but the forth unit accounting for the last glucose
unit at the reducing end of the molecule is not. The location and
orientation of the acarbose inhibitor within the active site of
PWA resemble previous data from several ␣-amylase䡠acarbose
complexes. However, in contrast to previous observations (29 –
33), PWA does not display acarbose transglycosylation activity,
indicating that PWA is not catalytically active under crystalli-
zation conditions. We reasoned that the low temperature used
for crystal growth of this hyperthermophilic ␣-amylase has
been sufficient to inhibit any transglycolytic catalysis within
the crystal.
We noticed that residual anomalous difference electron den-
sity remained at the nitrogen position of the 4-amino-4,6-
dideoxy-␣-D-glucose ring within the active site of the refined
PWA䡠Ac/Zn complex (Fig. 4C). Therefore, a zinc ion was placed
into this position as an alternative inhibitor (Table II, site 7).
Zinc, like the amino group of the 4-amino-4,6-dideoxy-␣-D-glu-
cose ring of acarbose, is bound by the carboxylate group of
Glu222. The occupancies of the two inhibitors, acarbose and
zinc, were refined to final values of 0.6 and 0.4, respectively.
The structural overlay of acarbose and zinc within the active
site of the PWA䡠Ac/Zn structure displays directly the competi-
tive nature of these two chemically unrelated PWA inhibitors
(Figs. 3, A and B; and 4C). On the other hand, if acarbose was
added in 100 mM Tris buffer without zinc, a Tris molecule
entirely replaced acarbose in the active site of PWA
(PWA䡠Tris). Tris is bound by residues that interact with the
valienamine (Arg196, His288, and Asp289) and 4-amino-4,6-
dideoxy-␣-D-glucose (Glu222 and Asp289) groups of acarbose in
the PWA䡠Ac/Zn (Fig. 3, C and D), confirming previous struc-
tural data from a number of ␣-amylase䡠Tris complexes (32,
34 –36) and demonstrating its function as a potent competitive
inhibitor (34, 37, 38).
In the third crystal form (PWA䡠Zn), zinc binds to the carbox-
ylate group of the same residue (Glu222) (Fig. 3, E and F; and
Table II, site 7) that binds to the 4-amino-4,6-dideoxy-␣-D-
glucose group of acarbose (Fig. 3, A and B) and Tris (C and D).
The carboxylate group of Asp289 interacts with the zinc ion by
a solvent-mediated hydrogen bond. Comparison of the PWA䡠Zn
and PWA䡠Ac/Zn structures reveals that the position of the zinc
ion in PWA䡠Zn is nearly identical to that of the acarbose nitro-
gen and the partially occupied zinc ion in the PWA䡠Ac/Zn com-
plex. However, the PWA䡠Zn structure permits a more detailed
description of the coordination geometry of the zinc ion. Two
solvent molecules are located at positions equivalent to C-7 and
O-2 of the valienamine residue in the PWA䡠Ac/Zn complex (Fig.
3, A and B). Two additional solvent molecules are located at O-4
and O-6 of the acarbose complex and one near O-3, thus sub-
stituting the oxygen atoms of the glucose ring of the natural
substrate bound to the ⫺1 subsite. Soaking of the PWA䡠Zn
crystal form in a solution containing 15 mM acarbose at pH 6.5
confirmed partial replacement of zinc by the inhibitor (data not
shown). Thus, the PWA䡠Zn structure provides, for the first
time, a molecular rationale for previous observations (10) and
our data (Table III) showing how zinc, at concentrations ⬎5
mM, entirely inhibits the amylolytic activity of PWA. We antic-
9880 Structure of a (Ca,Zn) Metal Center-containing ␣-Amylase
TABLE II observations that magnesium does not inhibit enzyme activity
Metal-binding sites in PWA of different crystal forms (10, 11).
The values for the ␴ levels of the metal sites were obtained after
Two Different Metals Bind Near the Active Site—All three
determination of the positive difference peaks (diff) in the Fo ⫺ Fc
difference Fourier electron density maps as described under “Experi- PWA structures contain a two-metal center in close proximity
mental Procedures.” The ␴ values for the anomalous peaks (ano) were to the active site cleft, irrespective of specific crystallization
obtained from the anomalous difference electron density maps calcu- conditions and ligand binding to the active site. Therefore, it is
lated using fast Fourier transformation.
most likely that the observed metal ions originate from the
PWA 䡠 Zn PWA 䡠 Ac/Zn PWA 䡠 Tris medium that was used for heterologous expression of PWA in
Site 1 E. coli. The two metal sites are separated by 7.3–7.4 Å (Fig. 4A
Metal Ca Ca Ca and Table II, sites 1 and 2) and are located within the interface
B-factor (Å2) 27.1 24.5 12.4 of domains A and B. Only one of the two sites showed an
␴ (diff/ano) 12.7/– 15.5/– 25.9/–
anomalous signal under the energy conditions used for x-ray
Protein ligands: Asn110, Asp155, Gly157␣, Asp164, Gly202a
Site 2 data collection (Table I), thus indicating a hetero-population of
Metal Zn Zn Zn metals at this site.
B-factor (Å2) 26.8 27.9 12.7 Generally, class-13 glucosidases contain a common calcium-
␴ (diff/ano) 13.0/9.2 24.4/14.1 41.0/59.0
binding site (40), which is essential for their catalytic activity
Protein ligands: His147, His152, Cys166
Site 3 (2, 41). In the PWA structure, this site is matched by a peak
Metal Zn Zn Mg without an anomalous signal, confirming it as common calcium
B-factor (Å2) 27.9 34.9 18.0 site. The calcium ion is coordinated by seven protein ligands in
␴ (diff/ano) 10.2/11.2 16.8/– 11.8/–
a distorted octahedral geometry involving the conserved resi-
Protein ligands: Asp347, Asp349, Glu350
Site 4 due Asn110 from loop ␤3 (domain A)–␤1 (domain B), three
residues from the loop preceding ␣-helix 3a of domain A (Asp155

Downloaded from www.jbc.org by guest, on November 14, 2010

Metal Zn Zn No metal
B-factor (Å2) 24.8 40.2 as bidendate ligand, Gly157, and Asp164), Gly202 from a 310-
␴ (diff/ano) 14.1/9.3 12.2/9.7
helix between ␤-strand 4 and ␣-helix 4 of domain A, and one
Protein ligands: Lys33, b Glu36, Glu318, Lys323
Site 5 ordered solvent molecule (Fig. 4A). Thus, the number of protein
Metal Zn Zn Mg ligands in this calcium-binding site exceeds those found for the
B-factor (Å2) 19.9 27.9 11.9 same site in other class-13 ␣-amylases (2).
␴ (diff/ano) 16.2/11.7 16.0/15.2 15.4/–
In contrast to the first metal site (Table II, site 1), there is a
Protein ligands: Glu87, c Asp252, Asp256, Ile292
Site 6 strong anomalous contribution at the second site (site 2). This
Metal Zn Zn Mg has been identified as a zinc site because no other metal with a
B-factor (Å2) 24.9 35.1 12.7 measurable anomalous signal under the conditions of x-ray
␴ (diff/ano) 14.8/9.0 14.1/10.4 15.2/–
Protein ligands: Gly24,a,d, Glu80, Glu88, Glu253 c data collection was found in the PIXE analysis, confirming
Site 7 previous data (13). The zinc ion is coordinated by ligands in a
Metal Zn Zn No metal distorted tetrahedral geometry, including the imidazole groups
B-factor (Å2) 37.5 36.5 of His147 and His152, the sulfhydryl group of Cys166, and an
␴ (diff/ano) 12.9/8.9 9.7/5.2
Protein ligands: Glu222
ordered solvent molecule. This coordination geometry is typical
a
for protein䡠zinc complexes (42, 43). If the cysteine ligand
Main chain oxygen involved in metal coordination.
(Cys166) is replaced, thereby abolishing the zinc site of this
b
Side chain residue coordinating zinc observed only in PWA 䡠 Zn.
c
Symmetry-related intermolecular contact. two-metal center, the catalytic activity of PWA at high temper-
d
Main chain carbonyl oxygen involved in metal coordination ob- atures is dramatically reduced, indicating loss of thermostabil-
served only in PWA 䡠 Tris. ity (13). Thus, both sites of this two-metal center have a com-
mon role, to stabilize a catalytically active conformation in
TABLE III
Effect of metal ions and chemical reagents on PWA activity PWA at high temperatures.
The relative activity refers to the enzyme activity determined without Additional Ligand-binding Sites—The high resolution data
any additive (100 %). The measurements were performed according to of the three PWA structures have allowed the identification of
Bernfeld (51) in 0.05 M sodium acetate and 1% (w/v) starch (pH 5.5). additional ligand-binding sites (Table II). However, at present,
Each reaction was carried out with 11.5 ng of PWA for 4 min at 94 °C.
it remains largely unknown as to whether and to what extent
Relative activity these additional sites are critical for thermal stability and
Additives
1 mM 3 mM 5 mM catalytic function of the enzyme. Therefore, only a brief account
% of these sites is given.
Mg2⫹ 106 106 99 Apart from the acarbose site within the PWA active site that
Ni2⫹ 60 51 86 is observed only in the PWA䡠Ac/Zn structure, three other acar-
Ca2⫹ 105 94 83 bose sites have been identified in the two structures in which
Fe3⫹ 95 82 63
acarbose was present in the crystallization medium
Co2⫹ 92 66 55
Fe2⫹ 85 55 38 (PWA䡠Ac/Zn and PWA䡠Tris) (Fig. 1). The second acarbose-bind-
Mn2⫹ 58 44 33 ing site, with three acarbose rings visible (Ac-II), is located in a
Zn2⫹ 25 7 0 slight depression formed by the 310-helix preceding ␤-strand 1
Cu2⫹ 5 0 0
of domain A, the loop connecting ␣-helix 6b and ␤-strand 7 of
EDTA 95 91 86
Tris 91 87 83 domain A, and the loop connecting ␣-helix 8b of domain A and
Acarbose 74 54 20 the first ␤-strand of domain C (Fig. 1). A similar carbohydrate-
binding site was reported for a chimeric Bacillus amylolique-
faciens/licheniformis ␣-amylase (32). The third acarbose-bind-
ipate that the strong inhibition of many ␣-amylases by zinc (39) ing site (Ac-III) is located at the surface of domain B at a
is via the same binding to the ⫺1 site as observed in PWA䡠Zn. distance of ⬃30 Å from the active site (Fig. 1). This site corre-
In contrast, the PWA structures do not indicate that magne- sponds to the so-called accessory carbohydrate site reported for
sium binds to the active site, confirming previous biochemical the pig pancreas ␣-amylase structure (44). The fourth acar-
 Structure of a (Ca,Zn) Metal Center-containing ␣-Amylase 9881

Downloaded from www.jbc.org by guest, on November 14, 2010

FIG. 3. Active site of PWA with bound acarbose and zinc (PWA䡠Ac/Zn; A and B), Tris (PWA䡠Tris; C and D), and zinc (PWA䡠Zn; E and
F). A, C, and E, structures of active site residues in the presence of bound ligands. Each Fo ⫺ Fc difference electron density map (green) in the
absence of ligands (A, acarbose; C, Tris; and E, zinc) is contoured at 2.0, 2.0, and 4.5 ␴, respectively. In E, the anomalous difference peak (red) is
contoured at 3.7 ␴. B, D, and F, schematic representations of the ligands bound to the active site. Hydrogen bonds are shown by dashed lines. Zinc
ions and solvent molecules are shown as green and gray spheres, respectively. Solvent molecules mediating protein-inhibitor interactions are
indicated in B, D, and F; for clarity, Tyr62, Phe159, and Tyr199 are not shown.

bose-binding site (Ac-IV) is located at the surface of the domain
A/C interface (Fig. 1), again with three sugar rings visible.
These three rings interact with residues from ␣-helix 8b of
domain A and ␤-strand 5, the following loop of domain C, and
the C terminus of the protein molecule.
The presence of an anomalous difference at the remaining
metal-binding sites (Table II, sites 3– 6) depends on the pres-
ence of zinc during crystallization, suggesting that metal
binding at these sites is weaker and less selective than at the
(Ca,Zn) metal center, where the presence/absence of the
anomalous difference does not depend on crystallization con-
ditions. All three PWA structures display an additional
metal-binding site (Table II, site 3) at the loop connecting
␤-strands 1 and 2 of domain C (Fig. 4B, left panel). In the
PWA䡠Ac/Zn and PWA䡠Zn structures, a zinc ion was modeled
in this site, which is coordinated by the three carboxylate
9882 Structure of a (Ca,Zn) Metal Center-containing ␣-Amylase

FIG. 4. Unique metal-binding sites

in PWA. A, structure of the two-metal
(Ca,Zn)-binding site in PWA (cf. Table II,
sites 1 and 2). The zinc ion (gray sphere) is
coordinated by two histidines (His147 and
His152), one cysteine (Cys166), and an or-
dered solvent molecule (red). The calcium
ion is coordinated by the conserved resi-
dues Asn110, Asp155, Gly157, Asp164, and
Gly202 (cf. Fig. 2) and a solvent molecule.
The Fo ⫺ Fc difference and anomalous
difference maps are colored as described
in the legend to Fig. 3 and contoured at
4.5 and 33 ␴, respectively. B, structure of

Downloaded from www.jbc.org by guest, on November 14, 2010

the variable PWA metal-binding site in
domain C (cf. Table II, site 3). Left panel,
zinc (green sphere) is coordinated by three
negatively charged amino acid side chains
(Asp347, Asp349, and Glu350) and a solvent
molecule (red sphere). Right panel, in the
PWA䡠Tris structure, a magnesium ion is
coordinated by the same side chain resi-
dues as zinc in the left panel. However,
two additional solvent molecules lead to a
different overall coordination geometry.
C, shown is the structure of the partially
occupied inhibitory zinc-binding site in
the ⫺1 active site pocket of PWA, which is
superimposed on the bound acarbose li-
gand. The residual anomalous difference
electron density at the zinc position is
shown at a contour level of 3.2 ␴.

groups of Asp347, Asp349, and Glu350 in a trigonal planar
geometry. In the PWA䡠Ac/Zn structure, this site is also ligan-
ded by an ordered solvent molecule. In the PWA䡠Tris struc-
ture, in contrast, the site bears a magnesium (or calcium) ion
that is coordinated by the same residue ligands and three
ordered solvent molecules in a distorted octahedral geometry
(Fig. 4B, right panel). The structure suggests that this metal-
binding site generally serves a stabilizing role, which may be
further enhanced by a nearby disulfide bridge connecting
Cys388 and Cys432. Two other metal-binding sites within the
interfaces of symmetry-related molecules have been found in
the PWA structures (Fig. 1 and Table II, sites 5 and 6). One
additional metal-binding site in domain A (Table II, site 4) is
found only in the presence of zinc, involving at least one
lysine residue as ligand, which is a rare feature in known
protein crystal structures (43). Overall, the total number of
metals coordinated by a single PWA ␣-amylase molecule ex-
ceeds the number of bound metals previously reported for any
other class-13 glycosylhydrolase. The observed high number
of bound metals on the protein surface may enhance the
hyperthermostability of PWA and may reflect one strategy of
evolutionary adaptation to a high temperature environment.
DISCUSSION
Zinc Is a Competitive Active Site Inhibitor That Binds to the
Conserved ⫺1 Active Site Pocket—Acarbose and related sugar
units containing ␣-amylase inhibitors bind to at least three
specific active site pockets in class-13 ␣-amylases. The three
PWA structures demonstrate that the ⫺1 pocket of the enzyme
is the key site for competitive binding by the unrelated inhib-
itors used in this study. In PWA, this site not only binds
organic compounds like acarbose and Tris, but also serves as an
inhibitory metal-binding site of limited specificity. The two
chemically related metals Cu2⫹ and Zn2⫹ show comparable
inhibition of PWA catalysis, suggesting that they indeed bind
into the same pocket and that metal inhibition may correlate
generally with binding into the ⫺1 active site pocket (Table
III). Although the protonation state of the “metal” site in the
organic inhibitors (the ternary amino group in Tris and the
secondary amino group of the 4-amino-4,6-dideoxy-␣-D-glucose
 Structure of a (Ca,Zn) Metal Center-containing ␣-Amylase
ring in acarbose) is not directly visible in the PWA structures,
we assume that these groups are protonated and thus are kept
positively charged by the nearby carboxylate groups of Asp289
and Glu222, thereby conferring comparable specific PWA
binding.
The ⫺1 pocket residue ligands are highly conserved among
class-13 ␣-amylase sequences (Fig. 2), indicating that compet-
itive metal inhibition could be a general feature of members of
this family. To date, no systematic structural and functional
studies are available in which the ability of zinc to act as a
competitive inhibitor of ␣-amylases from different organisms
was investigated. Interestingly, a molar excess of calcium al-
most completely inhibits the catalytic activity of the ␣-amylase
from Aspergillus niger (40), whereas it has little effect on the
PWA catalytic activity. These data are reflected by the pres-
ence of calcium in the ⫺1 active site pocket in the structure of
the ␣-amylase from A. niger (40), whereas no metal ion is found
in the same site in the PWA䡠Tris structure, which was crystal-
lized in the absence of zinc. Thus, despite the conserved nature
of the ⫺1 active site pocket in ␣-amylases, its affinity for
different metal ions may vary.
The PWA (Ca,Zn)-binding Site Is Essential for PWA Activity,
but Is Not Conserved—We have solved the first structure of an
␣-amylase that comprises a mixed (Ca,Zn) two-metal center in
close proximity to the active site cleft. The three structures in
the presence of different inhibitors (zinc, acarbose/zinc, and
Tris) provide a molecular rationale for previous biochemical
analyses of PWA (13) and our current data (Table III) indicat-
ing specific and tight binding of zinc and calcium. If the zinc
site of this two-metal center is abolished by replacing its cys-
teine ligand (Cys166), the catalytic activity of PWA at high
temperatures is dramatically reduced, indicating loss of ther-
mostability (13). Comparison of ␣-amylase sequences (data not
shown) and use of Cys166 as an indicator denote that the zinc
site of the two-metal center is present only in P. woesei and its
close homolog Thermococcus hydrothermalis (84% sequence
identity). Even in the more closely related Archaea sequences
from Pyrococcus kodakaraensis and Thermococcus sp., this cys-
teine is replaced by an alanine, indicating that this site is lost.
To date, only two other structures of ␣-amylases with a
two-metal center are available. One belongs to the hyperther-
mophile B. licheniformis, which binds two calcium ions that are
probably bridged by a sodium ion (27). This site superimposes
well with the (Ca,Zn) two-metal center of PWA (Fig. 4A). Both
two-metal centers share the conserved calcium-binding site
(site I in BLA) (27), which is common to many class-13 ␣-amy-
lases, whereas the coordination geometry of the second differs.
These data indicate divergent evolutionary paths for the adap-
tation of ␣-amylases in Archaea (P. woesei) and bacteria (B.
licheniformis; Topt ⫽ 90 °C) to high temperature environments.
As such, these ␣-amylases have evolved as either a homo-
(Ca,Ca)- or a hetero-(Ca,Zn) two-metal center, respectively.
The other known ␣-amylase structure with a two-metal center
is the meso-stable ␣-amylase from barley, which shares the
highly conserved calcium-binding site and displays an addi-
tional calcium-binding site at a distance of ⬃7 Å (45, 46).
However, in contrast to the two calcium sites in BLA, the
second calcium site of the barley ␣-amylase structure does not
superimpose with the second site of the (Ca,Zn) two-metal
center in PWA.
Another specific feature of the PWA structure is the presence
of a disulfide bond in domain B that is formed by two adjacent
cysteines, Cys153 and Cys154. This sequence motif is confined to
a limited number of Archaea ␣-amylase sequences, including
those of P. kodakaraensis, T. hydrothermalis, and Thermococ-
cus sp. One loosely related plant ␣-amylase sequence from
9883
Oryza sativa (24% sequence identity) also contains the same
molecular arrangement. This sequence motif may also be in-
dicative of the previously identified close relations between
Archaea and plant ␣-amylases (47). The close proximity of the
zinc site of the two-metal center in PWA and the Cys153–Cys154
disulfide bridge suggests a possible joint requirement for ␣-am-
ylase activity under the physiological conditions of P. woesei.
However, available cysteine mutations do not influence ther-
mostability and catalytic activity significantly under the con-
ditions of the present in vitro measurements (13). Such an
atypical disulfide bridge, connecting residues that are adjacent
in sequence, is rare in available protein structures. Two of
these proteins are members of the alcohol dehydrogenase fam-
ily, specifically methanol dehydrogenase from Methylobacte-
rium extorquens (48, 49) and ethanol dehydrogenase from
Pseudomonas aeruginosa (50). The atypical disulfide bridge
may stabilize the non-planar semiquinone form of the enzyme’s
prosthetic group pyrroloquinoline quinone (49). We speculate
that, under the specific living conditions of P. woesei at high
temperatures, rigidification of the active site area by additional
conformational constraints imposed by the presence of such a
disulfide bridge may be required for in vivo catalytic activity.
Downloaded from www.jbc.org by guest, on November 14, 2010
However, the precise molecular role of this disulfide bridge in
PWA substrate catalysis with respect to thermostability still
remains to be defined experimentally.
Implications for Biotechnological Processes—Amylases,
along with other starch-hydrolyzing enzymes like pullulanases
and glucoamylases, have widespread applications in the food,
chemical, and pharmaceutical industries (4, 6) and compose
⬃30% of the current industrial enzyme production. Some of
these processes, such as the liquefaction of starch, require high
temperatures of up to 100 °C. At present, mostly the ␣-amy-
lases from B. licheniformis and B. stearothermophilus are used
commercially in the liquefaction process of starch due to their
high thermostability (5). However, the ␣-amylases from these
organisms display full catalytic activity and stability only if
calcium is added. Unfortunately, the addition of calcium inhib-
its glucoamylases and destabilizes glucose isomerases, which
are used in subsequent starch-processing reactions, thus stim-
ulating investigations into alternative ␣-amylases that do not
require addition of metal ions during enzymatic processes at
the industrial scale.
Not only is PWA superior to other ␣-amylases with respect to
thermostability, but it also lacks an exogenous calcium require-
ment for full catalytic activity (Table III) (10, 13). Therefore,
PWA has been proposed as an alternative to BLA to further
improve the efficiency of industrial processes in starch lique-
faction (5, 6). In this work, we have solved the PWA crystal
structures and revealed the molecular basis for tight calcium
and zinc binding by the identification of a nonconserved and
PWA-specific (Ca,Zn) two-metal center. Although in all struc-
tures of glycosylhydrolase class-13 amylases known so far, the
calcium in domain B is coordinated by not more than six pro-
tein ligands, in PWA, it is coordinated by seven protein ligands,
thus providing a structural rationale for the high binding af-
finity of the latter enzyme. In addition, the PWA structure
reveals a novel zinc-binding site in close proximity to the cal-
cium-binding site previously established to be essential for
catalytic activity and stability (13). Activation of PWA by traces
of zinc is, however, superseded by the competitive active site
inhibitory effects of this metal. P. woesei has evolved this struc-
tural property as a unique evolutionary adaptation most prob-
ably to retain full PWA activity in its extremophilic living
condition, utilizing zinc as a positive and negative regulator in
a concentration-dependent manner. The PWA structures offer
a strong base upon which to further engineer properties of
9884 Structure of a (Ca,Zn) Metal Center-containing ␣-Amylase
PWA that are more conducive to potential applications in in-
dustrial processes. In particular, they reveal two metal-binding
sites (zinc and calcium) with different functions, which are well
suited to optimize the properties of PWA for biotechnological
applications.
Acknowledgments—We thank E. Möller (Bayer AG, Leverkusen,
Germany) for kindly providing acarbose and D. S. Auld for helpful
discussions on the interpretation of metal-binding sites.
REFERENCES
1. Davies, G., and Henrissat, B. (1995) Structure 3, 853– 859
2. Janecek, S. (1997) Prog. Biophys. Mol. Biol. 67, 67–97
3. Pujadas, G., and Palau, J. (2001) Mol. Biol. Evol. 18, 38 –54
4. Eichler, J. (2001) Biotechnol. Adv. 19, 261–278
5. Van der Maarel, M. J., van der Veen, B., Uitdehaag, J. C., Leemhuis, H., and
Dijkhuizen, L. (2002) J. Biotechnol. 94, 137–155
6. Bertoldo, C., and Antranikian, G. (2002) Curr. Opin. Chem. Biol. 6, 151–160
7. Henrissat, B. (1991) Biochem. J. 280, 309 –316
8. Feese, M. D., Kato, Y., Tamada, T., Kato, M., Komeda, T., Miura, Y., Hirose,
M., Hondo, K., Kobayashi, K., and Kuroki, R. (2000) J. Mol. Biol. 301,
451– 464
9. Vieille, C., and Zeikus, J. G. (2001) Microbiol. Mol. Biol. Rev. 65, 1– 43
10. Jorgensen, S., Vorgias, C. E., and Antranikian, G. (1997) J. Biol. Chem. 272,
16335–16342
11. Dong, G., Vieille, C., Savchenko, A., and Zeikus, J. G. (1997) Appl. Environ.
Microbiol. 63, 3569 –3576
12. Linden, A., Niehaus, F., and Antranikian, G. (2000) J. Chromatogr. 737,
Crystallogr. Sect. D Biol. Crystallogr. 53, 448 – 455
25. Navaza, J. (1994) Acta Crystallogr. Sect. A 50, 157–163
26. Holm, L., and Sander, C. (1993) J. Mol. Biol. 233, 123–138
27. Machius, M., Declerck, N., Huber, R., and Wiegand, G. (1998) Structure 6,
281–292
28. Jespersen, H. M., MacGregor, E. A., Henrissat, B., Sierks, M. R., and Svens-
son, B. (1993) J. Protein Chem. 12, 791– 805
29. Gilles, C., Astier, J. P., Marchis-Mouren, G., Cambillau, C., and Payan, F.
(1996) Eur. J. Biochem. 238, 561–569
30. Dauter, Z., Dauter, M., Brzozowski, A. M., Christensen, S., Borchert, T. V.,
Beier, L., Wilson, K. S., and Davies, G. J. (1999) Biochemistry 38,
8385– 8392
31. Brayer, G. D., Sidhu, G., Maurus, R., Rydberg, E. H., Braun, C., Wang, Y.,
Nguyen, N. T., Overall, C. M., and Withers, S. G. (2000) Biochemistry 39,
4778 – 4791
32. Brzozowski, A. M., Lawson, D. M., Turkenburg, J. P., Bisgaard-Frantzen, H.,
Svendsen, A., Borchert, T. V., Dauter, Z., Wilson, K. S., and Davies, G. J.
(2000) Biochemistry 39, 9099 –9107
33. Aghajari, N., Roth, M., and Haser, R. (2002) Biochemistry 41, 4273– 4280
34. Aghajari, N., Feller, G., Gerday, C., and Haser, R. (1998) Protein Sci. 7,
564 –572
35. Sevcik, J., Solovicova, A., Hostinova, E., Gasperik, J., Wilson, K. S., and
Dauter, Z. (1998) Acta Crystallogr. Sect. D Biol. Crystallogr. 54, 854 – 866
36. Skov, L. K., Mirza, O., Henriksen, A., De Montalk, G. P., Remaud-Simeon, M.,
Sarcabal, P., Willemot, R. M., Monsan, P., and Gajhede, M. (2001) J. Biol.
Chem. 276, 25273–25278
37. Wang, L. H., and Hartman, P. A. (1976) Appl. Environ. Microbiol. 31, 108 –118
38. Honda, S., Ohmura, T., Segawa, Y., Minematsu, T., and Kakehi, K. (1987)
J. Chromatogr. 419, 51– 60

Downloaded from www.jbc.org by guest, on November 14, 2010

39. Irshad, M., Goel, M., and Sharma, C. B. (1981) Phytochemistry 20, 2123–2126
253–259 40. Boel, E., Brady, L., Brzozowski, A. M., Derewenda, Z., Dodson, G. G., Jensen,
13. Savchenko, A., Vieille, C., Kang, S., and Zeikus, J. G. (2002) Biochemistry 41, V. J., Petersen, S. B., Swift, H., Thim, L., and Woldike, H. F. (1990)
6193– 6201 Biochemistry 29, 6244 – 6249
14. Niehaus, F., Bertoldo, C., Kähler, M., and Antranikian, G. (1999) Appl. Mi- 41. Vallee, B., Stein, E., Sumerwell, W., and Fischer, E. (1959) J. Biol. Chem. 234,
crobiol. Biotechnol. 51, 711–729 2901–2905
15. Studier, F., and Moffat, B. (1986) J. Mol. Biol. 189, 113–130 42. Auld, D. S. (2001) BioMetals 14, 271–313
16. Gill, S. C., and von Hippel, P. H. (1989) Anal. Biochem. 182, 319 –326
43. Harding, M. M. (2001) Acta Crystallogr. Sect. D Biol. Crystallogr. 57, 401– 411
17. Butz, T., Flagmeyer, R. H., Heitmann, J., Jamieson, D. N., Legge, G. J. F.,
44. Larson, S. B., Greenwood, A., Cascio, D., Day, J., and McPherson, A. (1994) J.
Lehmann, D., Reibetanz, U., Reinert, T., Saint, A., Spemann, D.,
Mol. Biol. 235, 1560 –1584
Szymanski, R., Troger, W., Vogt, J., and Zhu, J. (2000) Nucl. Instrum.
45. Kadziola, A., Abe, J., Svensson, B., and Haser, R. (1994) J. Mol. Biol. 239,
Methods B 161, 323–327
18. Otwinowski, Z., and Minor, W. (1997) Methods Enzymol. 276, 307–326 104 –121
19. Collaborative Computational Project Number 4 (1994) Acta Crystallogr. Sect. 46. Vallee, F., Kadziola, A., Bourne, Y., Juy, M., Rodenburg, K. W., Svensson, B.,
D Biol. Crystallogr. 50, 760 –763 and Haser, R. (1998) Structure 6, 649 – 659
20. De la Fortelle, E., and Bricogne, G. (1997) Methods Enzymol. 276, 472– 494 47. Janecek, S., Leveque, E., Belarbi, A., and Haye, B. (1999) J. Mol. Evol. 48,
21. Brunger, A. T., Adams, P. D., Clore, G. M., DeLano, W. L., Gros, P., Grosse- 421– 426
Kunstleve, R. W., Jiang, J. S., Kuszewski, J., Nilges, M., Pannu, N. S., Read, 48. Blake, C. C., Ghosh, M., Harlos, K., Avezoux, A., and Anthony, C. (1994) Nat.
R. J., Rice, L. M., Simonson, T., and Warren, G. L. (1998) Acta Crystallogr. Struct. Biol. 1, 102–105
Sect. D Biol. Crystallogr. 54, 905–921 49. Ghosh, M., Anthony, C., Harlos, K., Goodwin, M. G., and Blake, C. (1995)
22. Jones, T. A., Zou, J.-Y., Cowan, S. W., and Kjeldgaard, M. (1991) Acta Crys- Structure 3, 177–187
tallogr. Sect. A 47, 110 –119 50. Keitel, T., Diehl, A., Knaute, T., Stezowski, J. J., Hohne, W., and Gorisch, H.
23. Brünger, A. T. (1992) Nature 355, 472– 475 (2000) J. Mol. Biol. 297, 961–974
24. Perrakis, A., Sixma, T. K., Wilson, K. S., and Lamzin, V. S. (1997) Acta 51. Bernfeld, P. (1955) Methods Enzymol. 1, 149 –158