The active site of cellobiohydrolase Cel6A from Trichoderma reesei: the roles of aspartic acids D221 and D175.

Article abstract of DOI:10.1021/ja012659q

Trichoderma reesei cellobiohydrolase Cel6A is an inverting glycosidase. Structural studies have established that the tunnel-shaped active site of Cel6A contains two aspartic acids, D221 and D175, that are close to the glycosidic oxygen of the scissile bon

Full text of DOI:10.1021/ja012659q

Published on Web 08/02/2002
The Active Site of Cellobiohydrolase Cel6A from Trichoderma
reesei: The Roles of Aspartic Acids D221 and D175
Anu Koivula,† Laura Ruohonen,† Gerd Wohlfahrt,†,# Tapani Reinikainen,†
Tuula T. Teeri,*,†,+ Kathleen Piens,‡ Marc Claeyssens,‡ Martin Weber,§
Andrea Vasella,§ Dieter Becker,| Michael L. Sinnott,*,| Jin-yu Zou,
,¥
Gerard J. Kleywegt, Michael Szardenings, Jerry Ståhlberg, and
T. Alwyn Jones*,
Contribution from VTT Biotechnology, P.O. Box 1500, FIN-02044 VTT, Espoo, Finland,
Laboratory of Biochemistry, State UniVersity of Ghent, K.L. Ledeganckstraat 35, B-9000 Ghent,
Belgium, Laboratory for Organic Chemistry, Swiss Federal Institute of Technology,
ETH-Zentrum, UniVersita¨tsstrasse 16, CH-8092 Switzerland, Department of Paper Science,
UMIST, P.O. Box 88, SackVille Street, Manchester M60 1QD, England, and Department of Cell
and Molecular Biology, Uppsala UniVersity, Biomedical Center Box 596,
S-75124, Uppsala, Sweden
Received December 4, 2001
Abstract: Trichoderma reesei cellobiohydrolase Cel6A is an inverting glycosidase. Structural studies have
established that the tunnel-shaped active site of Cel6A contains two aspartic acids, D221 and D175, that
are close to the glycosidic oxygen of the scissile bond and at hydrogen-bonding distance from each other.
Here, site-directed mutagenesis, X-ray crystallography, and enzyme kinetic studies have been used to
confirm the role of residue D221 as the catalytic acid. D175 is shown to affect protonation of D221 and to
contribute to the electrostatic stabilization of the partial positive charge in the transition state. Structural
and modeling studies suggest that the single-displacement mechanism of Cel6A may not directly involve
a catalytic base. The value of D2O(V) of 1.16 ( 0.14 for hydrolysis of cellotriose suggests that the large
direct effect expected for proton transfer from the nucleophilic water through a water chain (Grotthus
mechanism) is offset by an inverse effect arising from reversibly breaking the short, tight hydrogen bond
between D221 and D175 before catalysis.
Introduction
efficient on highly crystalline cellulose. The two cellobiohy-
drolases have active sites enclosed in tunnels formed by long
Cellulose is an insoluble material composed of long chains
of â-1,4 linked glucosyl units. Native cellulose can be com-
pletely degraded to soluble sugars by synergistic mixtures of
many different types of cellulases produced by a wide range of
microbes.1,2 The soft rot fungus Trichoderma reesei produces
an effective mixture of cellulases. The most abundant enzymes
are two cellobiohydrolases, Cel7A and Cel6A, formerly called
CBHI and CBHII,3 and these are also the enzymes that are most
loops on the surface of the catalytic domains, and they therefore
act primarily at the cellulose chain ends.4-9 Cel7A removes
cellobiose units from the reducing end, while Cel6A prefers
the nonreducing end.4,7-9 In keeping with their extended
substrate-binding sites, Cel7A and Cel6A are processive en-
zymes catalyzing several bond cleavages before the dissociation
of the enzyme-substrate complex.2,4
The reaction mechanisms for all cellulases appear to involve
partial proton donation to the leaving group by a carboxylic
acid group on the enzyme. Hydrolysis occurs with retention or
inversion of the anomeric configuration depending on the nature
of the nucleophile. If the nucleophile is a carboxylate group of
the enzyme, then a double displacement mechanism occurs.10
* To whom correspondence should be addressed. T.T.T. (phone) +46-
8-55378381, (fax) +46-8-55378483, (e-mail) Tuula@biochem.kth.se;
M.L.S. (phone) +44-161-200-3897, (fax) +44-161-200-3858, (e-mail)
michael.sinnott@umist.ac.uk; T.A.J. (phone) +46-18-4714982, (fax) +46-
18-536971, (e-mail) alwyn@xray.bmc.uu.se.
† VTT Biotechnology.
‡ State University of Ghent.
§ Swiss Federal Institute of Technology.
(4) Rouvinen, J.; Bergfors, T.; Teeri, T.; Knowles, J. K.; Jones, T. A. Science
1990, 249, 380-6.
| UMIST.
(5) Srisodsuk, M.; Kleman-Leyer, K.; Keranen, S.; Kirk, T. K.; Teeri, T. T.
Eur. J. Biochem. 1998, 251, 885-92.
Uppsala University.
# Present address: Orion Corp, P.O. Box 65, FIN-02101 Espoo, Finland.
+ Present address: Department of Biotechnology, Royal Institute of
Technology, SCFAB, S-10691 Stockholm, Sweden.
¥ Present address: Cosmix Molecular Biologicals GMBH, Mascheroder
Weg 1B, D-38124 Braunschweig, Germany.
(6) Kleman-Leyer, K.; Siika-aho, M.; Teeri, T. T.; Kirk, T. K. Appl. EnViron.
Microbiol. 1996, 62, 2883-2887.
(7) Divne, C.; Stahlberg, J.; Reinikainen, T.; Ruohonen, L.; Pettersson, G.;
Knowles, J. K.; Teeri, T. T.; Jones, T. A. Science 1994, 265, 524-8.
(8) Divne, C.; Stahlberg, J.; Teeri, T. T.; Jones, T. A. J. Mol. Biol. 1998, 275,
309-25.
(1) Wood, T. M.; Garcia-Campayo, V. Biodegradation 1990, 1, 147-161.
(2) Teeri, T. T. Trends Biotechnol. 1997, 15, 160-167.
(3) Henrissat, B.; Teeri, T. T.; Warren, R. A. FEBS Lett. 1998, 425, 352-4.
(9) Barr, B. K.; Wolfgang, D. E.; Piens, K.; Claeyssens, M.; Wilson, D. B.
Biochemistry 1998, 37, 9220-9.
9
10.1021/ja012659q CCC: $22.00 © 2002 American Chemical Society
J. AM. CHEM. SOC. 2002, 124, 10015-10024
10015

A R T I C L E S
Koivula et al.
A covalent glycosyl-enzyme intermediate is formed and then
hydrolyzed with the acid catalyst now acting as a general base
for the deprotonation of an incoming water molecule. Cel7A is
such a retaining glycosidase.11,12 If, on the other hand, the
nucleophile is a water molecule, a single displacement occurs,
in which the water molecule is deprotonated by an enzyme
carboxylate group, acting as a general base. Because a water
molecule is placed between the reaction center and the enzyme
carboxylate group of inverting enzymes, the two catalytic
carboxyl and carboxylate groups tend to be further apart in
inverting enzymes.13 Cel6A and the other enzymes in family 6
are inverting glycosidases.11,12
The structure determination of the catalytic domains of both
T. reesei cellobiohydrolases has paved the way for detailed
studies of their catalytic machineries.4,7,8,14-18 In Cel6A, the
extended active site can accommodate six glucosyl units.17,18
The binding of four glucosyl units within the tunnel, corre-
sponding to the -2 to +2 sites, contributes to the transition-
state stabilization,4,19,20 while binding to the +4 site is apparently
only required for the action of Cel6A on crystalline substrates.15
The early structural studies of Cel6A revealed a network of
interacting residues apparently contributing to the formation of
the active site.4 Two aspartic acids, D221 and D175, were
identified as the residues closest to the scissile bond. The two
carboxyl/carboxylate groups of these residues are within hydrogen-
bonding distance of each other, and the carboxylate of D221
points to the glycosidic oxygen between sites -1 and +1. D175
also interacts with the hydroxyl group of Y169 and the
guanidino group of R174. Because of these interactions, it was
suggested that D175 is probably ionized, whereas D221 is
protonated, and thus able to act as the proton donor. D175 was
thought to ensure the correct protonation state of the catalytic
acid and also, possibly, to stabilize the charged reaction
transition state.4 Mutagenesis experiments and crystallographic
studies of complexes with non-hydrolyzable cello-oligosaccha-
rides indicate that steric clashes contribute to sugar ring
distortion from a chair conformation at the -1 subsite.18,20
Structural studies of these complexes and the closely related
Humicola insolens enzyme also indicate that one of the loops
defining the tunnel is capable of conformational changes.17,18
As a consequence of this conformational change, the interactions
of D175/D221 with Y169/R174 are broken, and the side chain
of S181 is pushed into the -1 site. A chain of two water
molecules then links the hydroxyl of S181 to the carboxylate
of D175, Figure 1. They are positioned 3.0 and 4.2 Å,
Figure 1. Binding of the substrate in the active-site tunnel of Cel6A. Some
of the key residues are labeled, and hydrogen bonds are indicated by dotted
lines. The figure shows the complex of the Y169F mutant with a non-
hydrolyzable cellotetraose derivative (methyl 4-S-â-cellobiosyl-4-thio-â-
cellobioside). Protein side chains are colored with carbons gold, nitrogens
blue, and oxygens red. Ligand carbons are colored gold, oxygens pink, and
the linking sulfur atom is yellow. Water molecules are colored green.
Portions of the main chain are also shown, color-ramped from red at the
N-terminus to blue at the C-terminus. The figure was created in O59 and
rendered with MOLRAY.59
respectively, from the anomeric carbon in the -1 site. Ring
distortion in the -1 site is not a consequence of this confor-
mational change.18 In the present study, we have used a
combination of site-directed mutagenesis, X-ray crystallography,
molecular mechanics calculations, and enzyme mechanistic
studies to elucidate the roles of the two key aspartyl residues,
D221 and D175, in the active site of Cel6A.
Methods
Enzyme Kinetics. Values of kcat and Km for hydrolysis of cellotriose,
cellotetraose, cellopentaose, and cellohexaose (Glc3 to Glc6) (Merck)
in 10 mM sodium acetate buffer, pH 5.0 at 27 °C, were determined by
HPLC (Waters Millipore) with RI detection as described earlier.14,21
The substrate concentrations were 400 µM for Glc3 and 2-250 µM
for Glc4-Glc6 for both mutants. The kinetic constants for the wt Cel6A
enzyme have been published earlier.20,21 The pH dependence of the
D175A mutant was determined at 27 °C using cellotetraose as substrate.
The cellotetraose concentration in the reaction mixture was 200 µM
(D175A), and the final enzyme concentration was 1.0-4.0 µM. The
following buffer systems were used: glycine-HCl (pH range 1.4-3.8),
sodium acetate (pH range 4.0-5.6), sodium phosphate (pH range 6.0-
8.0), glycine-NaOH (pH range 9.0-10.6); the concentrations of the
buffer solutions were 10 mM. In all of the measurements, samples were
taken at 8-10 different time points to obtain reliable values for the
initial rates. The kinetic constants were obtained from the initial
velocities of the reaction curves by a nonlinear regression data analysis
with the program Enzfit.22
(10) Koshland, D. E. Biol. ReV. 1953, 28, 416-436.
(11) Knowles, J. K. C.; Lehtovaara, P.; Murray, M.; Sinnott, M. L. J. Chem.
Soc., Chem. Commun. 1988, 1401-1402.
(12) Claeyssens, M.; Tomme, P.; Brewer, C. F.; Hehre, E. J. FEBS Lett. 1990,
263, 89-92.
(13) McCarter, J. D.; Withers, S. G. Curr. Opin. Struct. Biol. 1994, 4, 885-92.
(14) Koivula, A.; Lappalainen, A.; Virtanen, S.; Ma¨ntyla¨, A. L.; Suominen, P.;
Teeri, T. T. Protein Expression Purif. 1996, 8, 391-400.
(15) Koivula, A.; Kinnari, T.; Harjunpaa, V.; Ruohonen, L.; Teleman, A.;
Drakenberg, T.; Rouvinen, J.; Jones, T. A.; Teeri, T. T. FEBS Lett. 1998,
429, 341-6.
Kinetics of Fluoride Ion Release. Fluoride ion liberation was
followed essentially as described in ref 23. The temperature (25°C)
was kept constant during measurements using a Haake temperature bath.
Data were collected via a serial interface with an IBM-compatible PC
using the software Hyperterminal for recording. Data were analyzed
using the PC software FigP. Before initial rates of hydrolysis of
R-cellobiosyl fluoride were measured, the contaminating â-anomer was
(16) Ståhlberg, J.; Divne, C.; Koivula, A.; Piens, K.; Claeyssens, M.; Teeri, T.
T.; Jones, T. A. J. Mol. Biol. 1996, 264, 337-49.
(17) Varrot, A.; Schulein, M.; Davies, G. J. Biochemistry 1999, 38, 8884-91.
(18) Zou, J.-y.; Kleywegt, G. J.; Ståhlberg, J.; Driguez, H.; Nerinckx, W.;
Claeyssens, M.; Koivula, A.; Teeri, T. T.; Jones, T. A. Structure 1999, 7,
1035-1045.
(19) Harjunpaa, V.; Teleman, A.; Koivula, A.; Ruohonen, L.; Teeri, T. T.;
Teleman, O.; Drakenberg, T. Eur. J. Biochem. 1996, 240, 584-91.
(20) Koivula, A.; Reinikainen, T.; Ruohonen, L.; Valkeajarvi, A.; Claeyssens,
M.; Teleman, O.; Kleywegt, G. J.; Szardenings, M.; Rouvinen, J.; Jones,
T. A.; Teeri, T. T. Protein Eng. 1996, 9, 691-9.
(21) Teleman, A.; Koivula, A.; Reinikainen, T.; Valkeajarvi, A.; Teeri, T. T.;
Drakenberg, T.; Teleman, O. Eur. J. Biochem. 1995, 231, 250-8.
(22) Marquart, D. W. J. Soc. Ind. Appl. Math. 1963, 11, 431-441.
(23) Konstantinidis, A.; Sinnott, M. L. Biochem. J. 1991, 279, 587-93.
9
10016 J. AM. CHEM. SOC. VOL. 124, NO. 34, 2002

Catalytic Mechanism of T. reesei Cel6A
A R T I C L E S
Table 1. Kinetic Parameters for the wt Cel6A and Mutant Enzymes D221A and D175A on Cello-oligosaccharides (Glcn, n ) 4-6)
Measured in 10 mM Sodium Acetate Buffer, pH 5.0 at 27 °Ca
wt Cel6A
D221A
D175A
substrate
kcat (s-1
)
Km (µM)
kcat/Km (s-1 µM-1
)
kcat (s-1
)
kcat (s-1
)
Km (µM)
kcat/Km (s-1 µM-1
)
Glc3
Glc4
Glc5
Glc6
0.06 ( 0.01
4.1 ( 0.5
1.1 ( 0.2
14 ( 2
17 ( 5
0.004 ( 0.002
2.0 ( 0.5
e0.0001
e0.0012
e0.0010
e0.0007
e0.0003
nd
2.6 ( 0.5
1.3 ( 0.4
14 ( 6
0.0013 ( 0.0010
0.025 ( 0.001
0.020 ( 0.001
4 ( 1
1.4 ( 0.7
nd
0.0003 ( 0.0004
0.02 ( 0.01
0.8 ( 0.5
1.0 ( 0.7
aOwing to low activity, Km values could not be measured for the D221A mutant. Kinetic constants were calculated by a nonlinear regression data analysis
program (Enzfit), which also gives the standard deviation. nd ) not determined.
completely hydrolyzed using Cel7A.24 In the case of â-cellobiosyl
fluoride, the level of substrate degradation after deacetylation was tested
with the fluoride ion probe by comparing fully hydrolyzed substrate
(using T. reesei Cel6A for degradation) with the starting material.
Batches with more than 6% degradation were not used for kinetic
experiments. Enzyme concentrations for R-cellobiosyl fluoride mea-
surements were 3.85 (D175A) and 7.4 µM (D221A), and for â-cello-
biosyl fluoride measurements they were 3.85 and 0.96 µM. Hydrolysis
of 1-fluorocellobiosyl fluoride (1 mM) by wt Cel6A (8 µΜ) was
monitored for 5 days, with the electrode being recalibrated for each
reading.
for saccharides 29 and additional parameters for the reaction intermedi-
ates were used in these calculations. These additional parameters were
fitted to reproduce data from ab initio calculations with Gaussian 9430
at the B3LYP/6-31G* level. A two-stage restrained electrostatic
potential fit method was used to obtain atomic point charges.31 All
simulations were based on a number of experimental structures
determined by X-ray crystallography for the wild-type Cel6A and two
mutants.4,18 All histidines, except H266, were assumed to be protonated
at both nitrogens in accordance with the pH value of 5.0 used in most
experiments. H266 was deemed to be protonated only at Nꢀ, after taking
into account its structural environment. All carboxylic acid side chains
were negatively charged in the enzyme-ligand complex simulations,
except E107, D170, D419, and D221, which were kept protonated.
Initial ligand position and chain directionality for the force field
calculations were derived from the X-ray structure of the Y169F/(Glc)2-
S-(Glc)2 complex.18 The force field calculations were carried out with
explicit TIP3P water molecules32 at a dielectric constant ꢀ )1 and a
cutoff distance of 9 Å for the nonbonded interactions. During the
molecular dynamics simulations, the SHAKE algorithm33 was used to
constrain the covalent bonds to their equilibrium bond length. Only
the atoms within a 16 Å radius around the glycosidic oxygen between
-1 and +1 site were allowed to move during these simulations. The
structures were energy-minimized and then equilibrated by molecular
dynamics for 50 ps at 298 K with an integration interval of 1 fs. After
equilibration, a trajectory over 500 ps was saved every 1 ps. The
resulting structural ensembles were compared to X-ray structures, and
trajectories were analyzed for interactions of mechanistic importance
using the CARNAL module of the AMBER package.27
Solvent Isotope Effect on the Hydrolysis of Cellotriose by Wild-
type Cel6A. The solvent isotope effect on kcat was measured with 0.300
mM substrate (20 × Km) at 37.0 °C. Initial rates of three separate
reactions in H2O and D2O were obtained by analysis of liberated glucose
with high-performance anion exchange chromatography, using pulsed
amperometric detection, under the conditions specified by the manu-
facturer of the equipment (Dionex, Sunnyvale, CA). To eliminate
complexities of interpretation arising from solvent isotope effects on
essential ionizations, the rate in H2O was measured at the pH optimum
of the enzyme (4.5) in 10 mM sodium acetate buffer. The rate in D2O
was measured at the same buffer ratio as in H2O, where it gave a pH
of 4.50. Because the likely critical ionizations arise from carboxylic
acid residues, we make the usual assumption that the solvent isotope
effect on ionizations of the same chemical type is constant.
Ligand-Binding Studies. Binding of 4-methylumbelliferyl-â-glu-
coside, -cellobioside, and -cellotrioside [MeUmb(Glc), MeUmb(Glc)2,
and MeUmb(Glc)3] in 50 mM sodium acetate buffer, pH 5.0 at 16 °C,
was studied by direct fluorescence quenching titrations as described
before.25 Fluorescence was measured with an Aminco SPF-500 spec-
trofluorometer equipped with a thermostated cuvette holder. Excitation
was at 318 nm (spectral bandwith 2 nm), and the emission was
measured at 360 nm with a bandwidth of 15, 20, or 40 nm. Cellobiose
and cellotriose (Merck) were used as competitive ligands for MeUmb-
(Glc)2 (indicator) in the displacement titrations. The dissociation
constants for different oligosaccharides were obtained by nonlinear
least-squares fitting to the fluorescence data as described in ref 20.
Crystallographic Analysis. Data collection and refinement statistics
are listed in Tables S1 and S2 in the Supporting Information.
Coordinates and structure factors have been deposited at the PDB with
codes 1HGY and 1HGW.
Results
Production and Purification of the Mutant Proteins. The
mutant proteins were produced in T. reesei strains devoid of
the wt Cel6A gene for the D221A mutant and both the Cel6A
and the Cel5A genes for the D175A mutant. Analysis of their
enzymatic properties was carried out using protein preparations
with no or minute activity on substrates diagnostic for the
endoglucanases.14
Catalytic Effects of the Mutations. The catalytic constants
for the action of wt Cel6A and the two mutants on soluble cello-
oligosaccharides are presented in Table 1. The values of kcat of
Modeling of Enzyme-Ligand Complexes. The protein-modeling
package BRAGI26 and the molecular-mechanics program AMBER 4.027
were used for simulations of the enzyme-ligand complexes. The
AMBER all-atom force field28 together with the GLYCAM force field
(29) Woods, R. J.; Dwek, R. A.; Fraser-Reid, B. J. Phys. Chem. 1995, 99, 3832-
3840.
(30) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Gill, P. M. W.; Johnson, B.
G.; Robb, M. A.; Cheeseman, J. R.; Keith, T.; Petersson, G. A.;
Montgomery, J. A.; Raghavachari, K.; Al-Laham, M. A.; Zakrzewski, V.
G.; Ortiz, J. V.; Foresman, J. B.; Cioslowski, J.; Stefanov, B. B.;
Nanayakkara, A.; Challacombe, M.; Peng, C. Y.; Ayala, P. Y.; Chen, W.;
Wong, M. W.; Andres, J. L.; Replogle, E. S.; Gomperts, R.; Martin, R. L.;
Fox, D. J.; Binkley, J. S.; Defrees, D. J.; Baker, J.; Stewart, J. P.; Head-
Gordon, M.; Gonzalez, C.; A., P. J. Gaussian 94, revision B.3; Gaussian,
Inc: Pittsburgh, PA, 1995.
(24) Becker, D.; Johnson, K. S.; Koivula, A.; Schulein, M.; Sinnott, M. L.
Biochem. J. 2000, 345 Pt 2, 315-9.
(25) van Tilbeurgh, H.; Pettersson, G.; Bhikabhai, R.; De Boeck, H.; Claeyssens,
M. Eur. J. Biochem. 1985, 148, 329-34.
(26) Schomburg, D.; Reichelt, J. J. Mol. Graphics 1988, 6, 161-165.
(27) Pearlman, D. A.; Case, D. A.; Caldwell, J. C.; Seibel, G. L.; Singh, U. C.;
Weiner, P.; K.P.A. AMBER, 4.0 ed.; University of California: San
Francisco, 1991.
(31) Bayly, C. I.; Cieplak, P.; Cornell, W. D.; Kollman, P. A. J. Phys. Chem.
1993, 97, 10296-10280.
(32) Jorgensen, W. L.; Chandrasekhar, J.; Madura, J. D.; Imey, R.; Klein, M. J.
Chem. Phys. 1983, 79, 926-935.
(28) Cornell, W. D.; Cieplak, P.; Bayly, C. I.; Gould, I. R.; Merz, K. M.;
Ferguson, D. M. S.; D.C.F., T.; Caldwell, J. W.; Kollman, P. A. J. Am.
Chem. Soc. 1995, 117, 5179-5197.
(33) Ryckaert, J. P.; Cicotti, G.; Berendsen, H. J. C. J. Comput. Phys. 1977,
23, 327-341.
9
J. AM. CHEM. SOC. VOL. 124, NO. 34, 2002 10017

A R T I C L E S
Koivula et al.
Table 3. Dissociation Constants for the Binding of Carbohydrates
to Wild-type Cel6A, D221A, and D175A Mutants, in 50 mM
Sodium Acetate Buffer, pH 5.0 at 16 °Ca
K (µM)
d
ligand
wt Cel6A
D221A
D175A
MeUmb(Glc)1
MeUmb(Glc)2
MeUmb(Glc)3
Glc2
18b
2.6
100
5
0.1
330
2.9
nd
0.8
0.2
1100
4.5
0.1c
1400
15
Glc3
a nd ) not determined. b The uncertainty in all measured Kd values is
estimated to be 15% on the basis of four repeated experiments. c The
measurement with wt Cel6A on MeUmb(Glc)3 was performed at 8 °C.
Figure 2. pH-activity profiles of wt Cel6A (- - 0 - -), Y169F (- - / - -) (data
from ref 20), and D175A (- - 2 - -) mutant enzymes represented as relative
activities. The activities were measured at 27 °C at constant cellotetraose
concentration, which was approximated as saturating concentration. The
absolute activity value of the D175 mutant at pH 5.0 corresponds to about
0.03% of the Cel6A wild-type activity.
the two mutants are given in Table 3. Neither of the mutations
introduced to the active site of Cel6A led to any major changes
in its binding behavior.
Structures of the D221A and D175A Mutants. The two
Cel6A mutants crystallize in different space groups, but both
contain two molecules per asymmetric unit. The structures that
we report, therefore, provide four views of the protein. The
overall structures of the mutants are very similar, except for
one of the tunnel-defining loops, corresponding to residues 172-
182. The mutants have four different conformations that affect
the active site and the -1 binding site, in particular (Figures 3
and 4). They range from molecule A of D221A, where the tunnel
is squeezed at the -1 site and is essentially identical to the
Y169F/(Glc)2-S-(Glc)2 complex structure described in ref 18,
Figures 1 and 3b, to an almost open active site in molecule A
of D175A, Figure 3c. Molecule B of D175A has the conforma-
tion first seen in the wt apo-enzyme, Figure 3b, whereas D221A
molecule B is somewhat more open as previously observed in
the wild-type/m-iodobenzyl â-D-glucopyranosyl-â(1,4)-D-xy-
lopyranoside complex structure.18 In the squeezed tunnel of
D221A molecule A, the conformational change breaks the
interaction of D175 with R174, as previously observed in the
Y169F/(Glc)2-S-(Glc)2 structure.18 However, the side chain of
D175 has a different conformation, probably as a result of the
D221A mutation. The remaining residues in the 172-182 loop
adopt a conformation very similar to the one observed in the
Y169F/(Glc)2-S-(Glc)2 structure. In the B molecule, the D175/
R174 interaction is also broken, but the loop is more open, and
its density is less clearly defined. This, in turn, results in higher
average temperature factors (see Table S2 in the Supporting
Information).
Table 2. Kinetic Parameters for the Hydrolysis of Cellobiosyl
Fluorides, in 0.100 M Sodium Acetate Buffer, pH 5.0 at 25 °C
enzyme
R-cellobiosyl fluoridea
â-cellobiosyl fluoride
wild-type
kcat ) 0.54 s-1
kcat ) 4.0 s-1
Km ) 2.35 (mM)2
kcat < 1.3 × 10-5 s-1
Km: not determined
kcat ) ∼3 × 10-4 s-1
Km: not determined
Km ) 0.15 mM
kcat ) 0.103 s-1
Km ) 1.3 mM
kcat ) 0.013 s-1
Km ) 0.18 mM
D221A
D175A
a Kinetic parameters calculated for a bimolecular reaction.24
the D221A mutant indicate almost complete loss of activity,
whereas the D175A mutant retains some activity on longer
oligosaccharides. Owing to its low activity, the Km values could
not be accurately calculated for the D221A mutant. Essentially
no changes were observed in the Km values for the D175 mutant
on cellotetraose or cellopentaose (Table 1). The specificity
constant (kcat/Km) of D175A is over 6000-fold lower than that
of wt Cel6A on cellotetraose, whereas only a 40-fold reduction
is observed on a longer substrate, Table 1.
Figure 2 shows the pH dependence of the reaction rate for
cellotetraose degradation for wt Cel6A, the D175A mutant, and
the Y169F mutant that was earlier shown to influence the pH
behavior of Cel6A.20 The data were measured at saturating
substrate concentrations based on our earlier kinetic data on wt
Cel6A.20 The activity of the D175A mutant was found to be
lower than the activity of wt Cel6A over the whole pH range
studied, but the effect is more pronounced for the basic limb.
Determination of the activities of the mutant enzymes on the
R- and â-cellobiosyl fluorides revealed a 40-fold reduction in
kcat for the hydrolysis of the â-anomer by the D221A mutant,
but a >40 000-fold reduction in kcat for the hydrolysis of the
R-anomer (Table 2). In contrast, the D175A mutant shows a
large, similar reduction in kcat for hydrolysis of both anomers
(1800-fold R, 300-fold â).
In the D175A crystal structure, one of the molecules has a
much more open active site as a result of the largest confor-
mational change in the 172-182 loop that we have so far
observed in any Cel6A structure. In particular, the -1, +1, and
+2 sites are more exposed to the solvent, whereas the side
chains forming the -2 site remain essentially unchanged,
Figures 3c and 4.
Molecular Modeling Based on the Experimental 3-D
Structures of Cel6A. To obtain a more detailed picture of the
reaction pathway of Cel6A, a series of possible reaction
intermediates was modeled starting with different experimental
structures of the wild-type Cel6A or its mutants.
Even with 8 µΜ wt enzyme, fluoride ion liberation from
1-fluorocellobiosyl fluoride was only ∼6 times faster than the
blank rate; this corresponds to a kcat of ∼2 × 10-6 s-1, if the
enzyme is saturated .
The solvent kinetic isotope effect for the hydrolysis of
cellotriose by wt Cel6A was only slightly above 1 [D2O(V) )
1.16 ( 0.14].
A model for the ground-state binding situation with cellotet-
raose was derived, starting from the Y169F/(Glc)2-S-(Glc)2
complex structure,18 that is, the most closed loop structure, with
the -1 site sugar modeled in different skew-boat and chair
Binding Properties of the Mutant Proteins. The binding
constants for a number of different ligands to wt Cel6A and
9
10018 J. AM. CHEM. SOC. VOL. 124, NO. 34, 2002

Catalytic Mechanism of T. reesei Cel6A
A R T I C L E S
Figure 4. Conformational changes that occur in one of the tunnel-defining
loops in Cel6A. D175A molecule A is colored yellow, the apo wild-type
molecule is gold, and the Y169F in complex with (Glc)2-S-(Glc)2 is magenta.
The (Glc)2-S-(Glc)2 molecule from the Y169F mutant complex is also
included. The surface is derived from the most open active site structure
D175A molecule A.
experimentally observed 2S0 conformation, which indicates that
it is strongly stabilized by the environment. In the starting
structure, D175 was defined as charged, and the proposed
catalytic acid D221 was protonated, acting as a hydrogen bond
donor to D175. During the simulation, the side chain of D221
remained hydrogen bonded to the side chain of D175, and its
closest oxygen had a distance of 4.0 ( 0.3 Å to the glycosidic
oxygen to be protonated during the reaction. The closest
approach of the carboxyl proton to the glycosidic oxygen was
3.07 Å (Figure 5a).
In the next MD simulation, the bridging proton between D175
and D221 was shifted to the other oxygen of D221. Upon giving
up its hydrogen bond to D175, the side chain of D221 rotated
through its ø1 angle, thus breaking the contact with D175. This
brought the protonated oxygen of D221 closer to the glycosidic
oxygen, 3.3 ( 0.3 Å, over the first 300 ps, after which the
hydrogen bond between D175 and D221 was formed again. The
approach of the carboxyl hydrogen of D221 to the glycosidic
oxygen was much closer than 2.0 Å. This close approach, and
a good positioning of the carboxyl group with the nonbonding,
doubly occupied orbital of the glycosidic atom of the -1 site,
is consistent with a facile proton transfer, Figure 5b. The side-
chain conformation of D221 observed in this simulation is
practically identical to that observed for the corresponding
residue in the structure of the family 6 endoglucanase, E2 from
Thermomonospora fusca.34 In the E2 structure, however, the
loop carrying the residue corresponding to our D175 is turned
away from the active site, and no contact is formed between
the two aspartyl residues.
Figure 3. Molecular surfaces of the Cel6A tunnel region: (a) wt Apo. (b)
Y169F in complex with (Glc)2-S-(Glc)2. (c) D175A molecule A. The (Glc)2-
S-(Glc)2 molecule has been modeled on the basis of the Y169F mutant
complex structure. The figures were made with GRASP.60
Independently of whether the reaction passes through an
oxycarbenium cation as the reactive intermediate, or via an SN2-
type transition state, the C1 atom will become trigonal. The
main difference between an SN1 reactive intermediate and an
SN2 transition state is reflected in the distance of the two ligands
to the C1 atom. We have chosen to model the transition state
using an oxycarbenium cation. A model of this intermediate
conformations. Here, the sulfur linkage of the ligand in the
experimentally determined structure was changed to an oxygen
linkage, and the mutation Y169F was reverted to the wt tyrosyl-
residue prior to the simulation. During the course of the 500 ps
MD simulation, the movable atoms showed an rmsd of 1.09 (
0.04 Å to the X-ray structure with a maximal deviation of 1.19
Å. The sugar in the -1 site remained in the skew-boat
conformation (2S0). MD simulations carried out with different
initial conformations of the -1 sugar all converged to the
(34) Spezio, M.; Wilson, D. B.; Karplus, P. A. Biochemistry 1993, 32, 9906-
16.
9
J. AM. CHEM. SOC. VOL. 124, NO. 34, 2002 10019

A R T I C L E S
Koivula et al.
Figure 5. Snapshots of the molecular dynamics simulations of the proposed catalytic states. Only the residues in the -1 and +1 sites are shown. All
hydrogens and the backbone atoms of Y169, D175, and D221 are omitted for clarity. Important hydrogen bonds are shown as green dots, and the interaction
between water and C1, important for the nucleophilic attack, is shown in orange. The models of different reaction states show (a) the ground state with the
-1 site sugar in a skew-boat conformation (2S0), (b) the ground state with the D221 side chain rotated toward the glycosidic oxygen, (c) an oxycarbenium
ion intermediate modeled in a boat conformation (2,5B) as an analogue to the transition state, and (d) two sugars of the reaction products (R-cellobiose in the
-1 and -2 sites and â-cellobiose in the +1 and +2 sites).
was built starting from the ground-state structure, and the
geometry of the oxycarbenium cation was derived by ab initio
calculations with Gaussian 94.30 To “scan” the active site for a
basic residue, which could activate a water molecule for the
nucleophilic attack at the C1 carbon, different initial conforma-
tions of the glycosidic bond between the -2 and -1 sites and
of the ring in the -1 site were used for MD simulations. Most
of these simulations converged to structures very similar to the
ground-state model described above and the X-ray structure of
the Y169F/(Glc)2-S-(Glc)2 complex (rmsd < 0.5 Å). None of
the simulations suggested an amino acid residue positioned to
act as the catalytic base. However, some of the models revealed
water molecules that could attack the C1 of the oxycarbenium
cation. Although in some of these models the water is positioned
on the side that would lead to a â-product, the energetically
most favorable model establishes geometry suitable to form the
expected R-product. This model shows the cation in a 2,5B
conformation. The distance from C1 to the water atom during
the 500 ps MD is 3.1 ( 0.2 Å as compared to 3.6 ( 0.2 Å in
the ground-state simulation. This water is fixed by the S181
side chain and the D401 backbone carbonyl and is connected
to the D175 side chain via a second water molecule (Figure
5c). The modeling experiment is, therefore, in good agreement
with the local water structure observed in the Y169/(Glc)2-S-
(Glc)2 complex structure of ref 18, Figure 1. After the bond
cleavage in the simulation, the sugar ring relaxed back to the
chair conformation, and the product was pushed slightly out of
the active site (Figure 5d).
Discussion
Active Site Conformational Changes. The key catalytic
residues in all inverting and most retaining glycoside hydrolases
are a pair of acidic residues from the protein.35 In inverting
enzymes, the carboxylate group of one residue acts as a base,
the carboxyl group of the other as an acid, to promote direct
displacement of the leaving aglycone by water.13 Recently, it
has become increasingly clear that another crucial element in
the catalytic mechanism of cellulases is the ring distortion at
the -1 site.8,9,18,20,36 The crystal structure of T. reesei Cel6A
revealed a pair of aspartyl side chains in its active site, D175
and D221, that were at hydrogen-bonding distance from each
other and on the same side of the scissile bond.4 The environ-
ment suggested that D221 is protonated, and it was proposed
to act as the acid catalyst. Subsequent crystallographic and
mutagenesis studies of wt and Y169F enzyme revealed a more
(35) Davies, G.; Sinnott, M. L.; Withers, S. G. In ComprehensiVe biological
catalysis; Sinnott, M. L., Ed.; Academic Press: London, 1998; Vol. I, pp
119-208.
(36) Sulzenbacher, G.; Driguez, H.; Henrissat, B.; Schulein, M.; Davies, G. J.
Biochemistry 1996, 35, 15280-7.
9
10020 J. AM. CHEM. SOC. VOL. 124, NO. 34, 2002

Catalytic Mechanism of T. reesei Cel6A
A R T I C L E S
complex active site. The complementary structural, mutagenesis,
and substrate binding studies were interpreted to suggest that
Y169 helps to distort the glucose ring into a more reactive
conformation.20 Further wt and Y169F structures in complex
with a non-hydrolyzable cellotetraose-like ligand, methyl 4-S-
â-cellobiosyl-4-thio-â-cellobioside, showed that the glucosyl
unit in the -1 site is indeed distorted into a 2SO conformation.18
The ligand conformations, however, are identical in both
enzymes and, hence, not a result of the tyrosyl-residue acting
alone. The mutant structure, though, undergoes a conformational
change in one of the principal tunnel-forming loops. As a result,
the interaction of D175 with R174 is broken, whereas the D175/
D221 interaction is maintained. Also, a string of water molecules
becomes visible, linking the carboxylate of D175 to the anomeric
carbon C1 at the -1 site, Figure 1. One water molecule, forming
a hydrogen bond with D175, is positioned 4.2 Å from C1. The
other is at a distance of 3.0 Å and correctly positioned to produce
the R-glucosyl residue upon catalysis. The latter water molecule
also interacts with the hydroxyl group of S181 (a residue that
moves ∼6 Å as a result of the conformational change) and the
main-chain carbonyl of residue 401. However, there is no
interaction of this potentially catalytic water with an enzyme
residue, which serves as a base. The proposed acid-catalyst
D221, on the other hand, is closely positioned to the sulfur in
the glycosidic linkage between the -1 and +1 sites. However,
the interaction geometry suggests that a hydrogen bond is
maintained between the carboxylate groups of D175 and D221
in the Y169F/(Glc)2-S-(Glc)2 complex (Figure 1). This may be
a consequence of the increased van der Waals radius of the
sulfur in the glycosidic linkage. The distortion of the glucosyl
unit in the -1 site is not a result of the conformational change,
because essentially the same ligand structure was present in the
complex with the wt enzyme, which did not show the loop
adjustment.18
The T. reesei cellobiohydrolases Cel6A and Cel7A are known
to act on crystalline cellulose. To do this, they must first remove
a cellulose chain from the crystal, which requires energy. Some
time ago, Sinnott and co-workers37 proposed that the cellobio-
hydrolases are molecular machines that coupled the energy
required for crystallite disruption to the energy released on
glycoside hydrolysis. In support of this idea, it was later shown
that the energetic coupling of crystallite disruption to glucan
hydrolysis was reversible; radiolabeled cellobiose was incor-
porated into Acetobacter xylinum cellulose by Cel6A under
conditions where the hydrolysis in free solution was essentially
irreversible.38 A key feature of known molecular machines39 is
that the respective molecules exist in two (or more) significantly
different conformational states. The different conformations
observed around the active site of Cel6A are directly coupled
to the catalytic events and may contribute to the movement of
the glucan chains in the active site tunnel of the enzyme. Even
these observations are in agreement with the molecular machine
hypothesis.
catalytically inactive on oligosaccharides of different length,
although its binding properties are identical to those of the wt
enzyme (Tables 1, 3). Data supporting the role of D221 as the
proton donor were also obtained by using R- and â-cellobiosyl
fluorides (Table 2). The fluoride ion does not require complete
protonation for a C-F bond to be cleaved in water, because
the pKa of H-F is 3.2 and F- is unprotonated in aqueous
solution, even though it is strongly hydrogen bonded (vide infra).
Consequently, mutation of the proton donor should have a
modest effect on the hydrolysis of â-cellobiosyl fluoride. In the
case of the D221A mutation, a 40-fold effect was nevertheless
observed. It is possible that this is a consequence of interactions
between the catalytic residues. For example, because of their
proximity, the D221A mutation may have a major effect on
the properties of D175, which in turn may modify the activity.
There is a requirement for the departing fluoride ion to be
solvated by hydrogen bonding, even though it does not require
full protonation. In the case of oligosaccharides and polysac-
charides, however, the leaving group absolutely requires pro-
tonation during departure, because its pKa is in the region of
13-14. Therefore, the much greater effect of the D221A
mutation on the hydrolysis of these substrates confirms the
assignment of D221 as the acid catalyst. Furthermore, the
D221A mutant was completely inactive toward R-cellobiosyl
fluoride. In general, inverting glycosidases hydrolyze glycosyl
fluorides of the “wrong” anomeric configuration by a resynthesis-
hydrolysis mechanism40 illustrated in Figure 6. In the first step,
two molecules of the fluoride are condensed by the enzyme in
a reversed-protonated state (acid catalyst deprotonated and base
catalyst protonated) to yield a new glycosidic link, which is
then hydrolyzed by the normal pathway. After a recent reevalu-
ation of earlier data, T. reesei Cel6A seems to hydrolyze
R-cellobiosyl fluoride by the Hehre mechanism.24,37,41 A key
feature of the first, transglycosylation, step of this mechanism
is that the acid-catalytic group, as its conjugate base, deproto-
nates the hydroxyl group of the second saccharide molecule. If
this catalytic group is removed, as in the mutant D221A, the
transglycosylation should cease, and this is indeed observed.
The experimental structures and the simulations presented
here, as well as data obtained with other family 6 cellulases
from Cellulomonas fimi,41,42 Thermomonospora fusca,43 and
Humicola insolens,17,44,45 are also consistent with the role of
D221 as the catalytic acid. The molecular dynamics simulations
suggest that breaking the hydrogen bond contact between D175
and D221 may induce a new, catalytically relevant side-chain
conformation of D221. This conformation, experimentally
observed in both the T. fusca E234 and the H. insolens Cel6A
structures,17,44 brings the protonated carboxyl-oxygen of D221
closer to the glycosidic oxygen.
The Role(s) of D175. Some residual activity (2-3%) on
barley â-glucan was measured for the D175A mutant enzyme
produced in Saccharomyces cereVisiae, while the D221A mutant
(40) Hehre, E. J. In Enzymatic degradation of insoluble carbohydrates; Saddler,
J. N., Penner, M. H., Eds.; ACS Symposium Series 618; American Chemical
Society: Washington, DC, 1995; pp 68-78.
D221 Is the Catalytic Acid. Hitherto, all of the data obtained
on T. reesei Cel6A and the other enzymes in this family support
the role of D221 as the catalytic acid. The D221A mutant was
(41) Damude, H. G.; Ferro, V.; Withers, S. G.; Warren, R. A. Biochem. J. 1996,
315, 467-72.
(42) Damude, H. G.; Withers, S. G.; Kilburn, D. G.; Miller, R. C., Jr.; Warren,
R. A. Biochemistry 1995, 34, 2220-4.
(37) Konstantinidis, A. K.; Marsden, I.; Sinnott, M. L. Biochem. J. 1993, 291,
883-8.
(38) Sweilem, N. S.; Sinnott, M. L. In Carbohydrases from Trichoderma and
other microorganisms; Claeyssens, M., Nierinckx, W., Piens, K., Eds.;
Royal Society of Chemistry: London, 1998; pp 13-20.
(39) Banting, G.; Higgins, S. J. Tetrahedron Lett. 2000, 28, 35.
(43) Wolfgang, D. E.; Wilson, D. B. Biochemistry 1999, 38, 9746-9751.
(44) Varrot, A.; Hastrup, S.; Schulein, M.; Davies, G. J. Biochem. J. 1999, 337,
297-304.
(45) Davies, G. J.; Brzozowski, A. M.; Dauter, M.; Varrot, A.; Schulein, M.
Biochem. J. 2000, 348 Pt 1, 201-7.
9
J. AM. CHEM. SOC. VOL. 124, NO. 34, 2002 10021

A R T I C L E S
Koivula et al.
significant. However, the residual activities observed would
seem to argue against a role as the catalytic base. The activity
of the mutant on the cellobiosyl fluorides is also informative.
If D175 were a simple catalytic base, the mutation should have
a large effect on the hydrolysis of the â-fluoride. In the case of
the R-fluoride, the first step of its hydrolysis is a transglyco-
sylation, which involves the active site residues in their reverse-
protonated state. Here, the departure of the fluoride ion does
not require a relatively strong acid; water as a H-bond donor
ought to suffice to allow for solvolysis. Therefore, eliminating
the group, now acting as the catalytic acid, should have only a
modest effect. The D175A mutation, however, had similar, large,
effects on the hydrolysis of both fluorides (Table 2), and,
therefore, D175 cannot be the catalytic base in the normal sense.
Its removal clearly decreases the pKa of the catalytic acid, as
shown in Figure 2 by the shift in the pH-rate profile.
The effect of substitution of the anomeric hydrogen of
â-cellobiosyl fluoride substrate with a second fluorine suggests
that the transition state is exceptionally electron-deficient, so
that D175 may stabilize electrostatically. Fluoroglycosyl fluo-
rides, in general, are comparatively efficiently hydrolyzed by
many glycosidases.46 They are slower substrates than glycosyl
fluorides, but by factors rarely approaching the 1000 (R) or
30 000 (â) seen in spontaneous hydrolysis. We have suggested
that the (V/K) ratio, converted to a free energy (∆∆Gq ) -RT
ln{(kcat/Km)F/-(kcat/Km)F2}), can be used as a measure of charge
development at the first transition state.23,37,47 The 1-fluorocel-
lobiosyl fluoride was synthesized via the action of xenon
difluoride on the glycosyl diazirine (see Supporting Information)
but proved to be only a barely detectable substrate. The Km
values observed with other inverting glycosidases acting on
glycosyl fluoride and 1-fluorosyl fluoride are very similar to
each other. Assuming that this is also the case with Cel6A
permits us to extract a kcat value from the hydrolysis experiment
(conducted at [S] ≈ 7Km), and then to calculate a ∆∆Gq from
the kcat values alone. The ∆∆Gq value obtained in this way was
∼36 kJ mol-1, which is higher than the equivalent value (-RT
ln 3000 ) 25 kJ mol-1) for spontaneous hydrolysis. This
indicates a higher degree of charge buildup in the transition
state for Cel6A catalysis than that for spontaneous hydrolysis.
Provided ∆∆Gq values are based on kcat/Km, this conclusion
is independent of the assumption made in our early work that
the effect of the second fluorine is to destabilize the oxocar-
benium ion. AM1 calculations (Bernet, B.; Vasella, A., unpub-
lished) now indicate that R-fluorine substitution, in fact,
stabilizes oxocarbenium ions. Thermochemical measurements
of Ru¨chardt’s group48,49 suggest rather that ground-state stabi-
lization is the source of the inertness of the fluoroglycosyl
fluorides. Additional F-F and O-F anomeric interactions could
account for ∼65 kJ mol-1 stabilization in the difluorides.
Nonetheless, because in comparisons between different enzyme
systems based on kcat/Km the ground state is the same, we can
still take ∆∆Gq as a measure of progress toward the oxocar-
benium ion, whether the effect used to measure such progress
Figure 6. (a) General mechanism for an inverting â-glucanase. (b) Hehre
mechanism for hydrolysis of the “wrong” glycosyl fluoride applied to the
reaction with R-cellobiosyl fluoride. “Exploded” transition states will be
involved in both cases, but the degree of bonding to leaving group and
nucleophile will not necessarily be the same, nor need the leaving group,
nucleophilic, and anomeric carbon to be strictly collinear. Additionally, the
development of partial double bond character in O5-C1 may not be
synchronous with the development of partial positive charge on C1. In this
particular case, the reactive conformation of the pyranose ring in the subsite
2
4
-1 is S0, not the C1 conformation shown here.
(46) Sinnott, M. L. In Recent adVances in carbohydrate bioengineering; Gilbert,
H. J., Davies, G. J., Henrissat, B., Svensson, B., Eds.; Royal Society of
Chemistry: London, 1999; pp 45-61.
was totally inactive on this substrate (our unpublished data).
When produced in T. reesei, the D175A mutant showed very
little activity on oligosaccharides, particularly on cellotetraose,
without major changes in its substrate binding properties (Tables
1 and 3). These data indicate that D175 is catalytically
(47) Konstantinidis, A. K.; Sinnott, M. L.; Hall, B. G. Biochem. J. 1993, 291,
15-17.
(48) Rakus, K.; Verevkin, S. P.; Peng, W. H.; Beckhaus, H.-D.; Ruchardt, C.
Liebings Ann. 1995, 2059-2067.
(49) Schaffer, F.; Verevkin, S. P.; Rieger, H. J.; Beckhaus, H. D.; Ruchardt, C.
Liebigs Ann. 1997, 1333-1334.
9
10022 J. AM. CHEM. SOC. VOL. 124, NO. 34, 2002

Catalytic Mechanism of T. reesei Cel6A
A R T I C L E S
arises from loss of stabilizing ground-state geminal interactions
or oxocarbenium ion destabilizing interactions.
the transition state of T. reesei Cel6A, there is little need for
efficient base catalysis. One possibility, suggested by the
experimental structures, is that a proton is transferred from the
nucleophilic water, via the second water, to D175 in a Grotthus-
type mechanism. To investigate this, we measured the solvent
isotope effect on kcat for the hydrolysis of the poor substrate,
cellotriose. With such a poor substrate, noncovalent events such
as product release are unlikely to limit kcat, and the effect is
probably intrinsic. The value obtained, 1.16 ( 0.14, was
surprisingly small considering literature precedents; one would
expect direct contributions both from the proton transfer to the
poor leaving group and from the nucleophilic water. In model
systems, general acid catalysis of acetal hydrolysis is associated
with values of kH2O/kD2O of 1.5-2.5.52-54 Solvent isotope effects
(kH2O/kD2O) on nonenzymic reactions, in which water is a
nucleophile and the leaving group does not require protonation,
are in the region 1.2-2.1,55,56 increasing with the nucleophilic
involvement of water. A proton inventory on glucoamylase-
catalyzed hydrolysis of a substrate with a good leaving group
suggested that the effect of 1.8 arose solely from the proton
being transferred from the nucleophilic water to the catalytic
base.57 The contributions to the solvent kinetic isotope effect
therefore cannot be just changes in fractionation factors of the
acid catalyst and nucleophilic water, which would give an effect
of 2-3. A large inverse contribution must also be involved.
This could, for example, be the rupture of a short, tight hydrogen
bond in a preequilibrium step. Such hydrogen bonds are
characterized by low fractionation factors; a value of 0.4 has
been reported for the hydrogen bond between the catalytic
The exceptionally highly charged transition state of this
enzyme may require additional electrostatic stabilization. The
electrostatic interaction of D175 with the ring oxygen in the
-1 site at the transition state could provide such stabilization.
The distance between the -1 sugar ring oxygen and the closest
carboxylate oxygen of D175 in the structure of the Y169F/
(Glc)2-S-(Glc)2 complex is only 4.5 Å, and there are no shielding
atoms in between. Residue D175 is, therefore, suitably posi-
tioned to provide such electrostatic stabilization. The large effect
of the D175A mutation on the hydrolysis of both fluorides then
becomes easier to understand. Similar interactions have been
proposed on theoretical grounds,50 and some experimental
support for them has been found in model systems.51
Role of D401? On the basis of its strict conservation and its
position relative to the catalytic acid, it has been suggested that
the residue corresponding to D401 in T. reesei Cel6A might
act as the catalytic base in family 6 glycosyl hydrolases.13,41,42
However, probing into the function of this residue in other
family 6 enzymes by mutagenesis studies has led to contradic-
tory results. The data initially obtained with CenA from C. fimi
supported its role as the base,41,42 while recent experiments with
T. fusca E2 reveal a dramatic decrease in substrate binding.43
In the structure of T. reesei Cel6A, D401 has numerous
hydrogen bond acceptor interactions to the nearby side chains
of R353, Y306, K395 and the main chain of residue 369 that
would ensure a low pKa for this carboxylate. Furthermore, in
the structures that we have determined to date, the carboxylate
group of D401 is seen to interact with the O3 hydroxyl, at the
edge of the glucosyl unit in the -1 site,18 Figure 1. Loss of this
interaction may account for the decreased binding observed in
the related T. fusca E2 enzyme.43 The data obtained in ref 43
and the structural data discussed above thus argue against the
role of D401 as a base. The structural data indicate that D175,
together with S181, participates in the coordination of the two
water molecules that are most closely positioned to attack the
anomeric carbon, Figure 1. The crystallographic structures
cannot resolve their charge states, but one water, coordinated
by the S181 side chain and the D401 backbone carbonyl oxygen,
is only 3.3 Å from the anomeric carbon. The molecular
dynamics simulation of the carbocationic intermediate suggests
that this water is stably coordinated and is the most probable
candidate for the catalytic water. This remains connected to
D175 via the second water molecule observed in the crystal
structure of the Y169F/(Glc)2-S-(Glc)2 complex. The candidate
catalytic water is 6.0 Å from the carboxylate group of D401,
and this residue undergoes no conformational changes in any
of our crystallographic studies. It seems possible that, rather
than removing the catalytic base, mutating the residue equivalent
to D401 in other family 6 enzymes introduces a charge
imbalance that may destabilize an oxycarbenium-ion-like transi-
tion state. Its removal would leave two positively charged side
chains, R353 and K395, without compensating acidic groups
close to the -1 site.
aspartate and protonated histidine in chymotrypsinogen.58
A
short hydrogen bond exists between D221 and D175 in the
unliganded enzyme (D221 OD2, D175 OD1 separation 2.7 Å)
and persists in the Y169F/(Glc)2-S-(Glc)2 complex (D221 OD2,
D175 OD1 separation 2.5 Å), which is our closest analogue of
the ES complex with cellotetraose.18 This hydrogen bond must
be broken before D221 can act as an acid, and, if rapid and
reversible, its breaking will make a large inverse contribution
to D2O(V). A solvent isotope effect close to unity is thus
anticipated for the active-site architecture revealed by crystal-
lography.
Conclusions
Our model for the roles of the active-site residues D221 and
D175 of the T.r. Cel6A is the following: (1) D221 acts as the
catalytic acid, and (2) D175 acts to stabilize a uniquely electron-
deficient transition state through electrostatic interactions. It also
modulates the pKa of D221 and may act as a proton acceptor
through a water chain. The degree of proton transfer at the
transition state, however, is very slight.
Acknowledgment. At VTT, we are grateful to Arja Lappa-
lainen, Tarmo Pellikka, and Susanna Virtanen for the help in
(52) Capon, B.; Smith, M. C.; Anderson, E.; Dahm, R. H.; Sankey, G. H. J.
Chem. Soc. 1969, B, 1038-1047.
(53) Anderson, E.; Fife, T. H. J. Am. Chem. Soc. 1969, 91, 7163-7166.
(54) Fife, T. H.; Brod, L. H. J. Am. Chem. Soc. 1970, 92, 1681-1684.
(55) Lee, I.; Koh, J.; Park, Y. S.; Lee, H. W. J. Chem. Soc., Perkin Trans.
1993, 2, 1575-1582.
Proton Transfer from the Nucleophilic Water. Because
there is very little bond formation to the nucleophilic water in
(56) Zhu, J.; Bennet, A. J. J. Am. Chem. Soc. 1998, 120, 3887-3893.
(57) Huang, X.; Surry, C.; Hiebert, T.; Bennet, A. J. J. Am. Chem. Soc. 1995,
117, 10614-10621.
(50) Bakthavachalam, V.; Lin, L. G.; Cherian, X. M.; Czarnik, A. W. Carbohydr.
Res. 1987, 170, 124-35.
(51) Cherian, X. M.; Van Arman, S. A.; Czarnik, A. W. J. Am. Chem. Soc.
1988, 110, 6566-6568.
(58) Markley, J. L.; Westler, W. M. Biochemistry 1996, 35, 11092-7.
(59) Harris, M.; Jones, T. A. J. Acta Crystallogr. 2001, D57, 1201-1203.
(60) Nicholls, A.; Sharp, K.; Honig, B. Proteins: Struct., Funct., Genet. 1991,
11, 281.
9
J. AM. CHEM. SOC. VOL. 124, NO. 34, 2002 10023

A R T I C L E S
Koivula et al.
protein purification, Anne Valkeaja¨rvi for help in HPLC
analysis, and Tuula Kuurila, Riitta Suihkonen, and Ulla Lahtinen
for skillful technical assistance. Ro¨hm Enzyme Finland is
thanked for the fungal expression strains, and David Wilson is
thanked for helpful comments on the manuscript. This work
has been supported by grants from Nordisk Industrifond, the
European Commission (Projects BIO2-CT94-3030 and BIO4-
CT96-0580), the Academy of Finland, and the Swedish Natural
Science Research Council, NFR (T.A.J.).
Supporting Information Available: Molecular biology pro-
tocols, details of the chemical syntheses, procedures for protein
crystallization, and structure determination (PDF). This material
is available free of charge via the Internet at http://pubs.acs.org.
JA012659Q
9
10024 J. AM. CHEM. SOC. VOL. 124, NO. 34, 2002