doi:10.1016/j.jmb.2007.10.

039

J. Mol. Biol. (2008) 375, 499510

Available online at www.sciencedirect.com

Structures of Mutants of Cellulase Cel48F of Clostridium cellulolyticum in Complex with Long Hemithiocellooligosaccharides Give Rise to a New View of the Substrate Pathway during Processive Action
Goetz Parsiegla 1 , Corinne Reverbel 2 , Chantal Tardif 2 , Hugues Driguez 3 and Richard Haser 4
Laboratoire de l'Architecture et Fonction des Macromolecules Biologiques, UMR 6098 CNRS and University of Aix-Marseille, Parc Scientifique et Technologique de Luminy, Case 932, 163 Avenue de Luminy, 13288 Marseille Cedex 09, France Unit de Bionergtique et Ingnierie des Protines, Institut de Biologie Structurale et Microbiologie, Centre National de la Recherche Scientifique, 31 Chemin Joseph-Aiguier, 13402 Marseille Cedex 20 and Universit de Provence, Place Victor Hugo, 13331 Marseille Cedex 03, France
3 Centre de Recherches sur les Macromolcules Vgtales, Centre National de la Recherche Scientifique and Universit J. Fourier de Grenoble, BP53, 38041 Grenoble Cedex 9, France 2 1

An efficient breakdown of lignocellulosic biomass is a prerequisite for the production of second-generation biofuels. Cellulases are key enzymes in this process. We crystallized complexes between hemithio-cello-deca and dodecaoses and the inactive mutants E44Q and E55Q of the endo-processive cellulase Cel48F, one of the most abundant cellulases in cellulosomes from Clostridium cellulolyticum, to elucidate its processive mechanism. In both complexes, the cellooligosaccharides occupy similar positions in the tunnel part of the active site but are more or less buried into the cleft, which hosts the active site. In the E44Q complex, it proceeds along the upper part of the cavity, while it occupies in the E55Q complex the same productive binding subsites in the lower part of the cavity that have previously been reported in Cel48F/cellooligosaccharide complexes. In both cases, the sugar moieties are stabilized by stacking interactions with aromatic side chains and H bonds. The upper pathway is gated by Tyr403, which blocks its access in the E55Q complex and offers a new stacking interaction in the E44Q complex. The new structural data give rise to the hypothesis of a two-step mechanism in which processive action and chain disruption occupy different subsites at the end of their trajectory. In the first part of the mechanism, the chain may smoothly slide up to the leaving group site along the upper pathway, while in the second part, the chain is cleaved in the already described productive binding position located in the lower pathway. The solved native structure of Cel48F without any bound sugar in the active site confirms the two sidechain orientations of the proton donor Glu55 as observed in the complex structures.
2007 Elsevier Ltd. All rights reserved.

Institut de Biologie et Chimie des Protines UMR 5086, Laboratoire de BioCristallographie, Centre National de la Recherche Scientifique and Universit Lyon IFR 128 BioSciences GerlandLyon Sud, 7 Passage du Vercors, 69367 Lyon Cedex 07, France

*Corresponding author. E-mail address: goetz.parsiegla@afmb.univ-mrs.fr. Abbreviations used: GH, glycoside hydrolase; CNS, Computation and Neural Systems.
0022-2836/$ - see front matter 2007 Elsevier Ltd. All rights reserved.

500 Received 20 July 2007; received in revised form 11 October 2007; accepted 15 October 2007 Available online 22 October 2007

Processive Action in Cel48F

Edited by G. Schulz

Keywords: cellulase; GH family 48; hemithiocellooligosaccharide; complex structure; processive action

Introduction
In times of growing public awareness of the dangers related to the increasing amount of carbon dioxide in the atmosphere, alternative sources of energy that do not participate in the global greenhouse effect become more and more attractive. One of the alternatives to the combustion of nonrenewable carbon-based fuels is the consumption of fuel obtained from renewable carbon sources such as biomass. The most important liquid fuel produced from biomass worldwide is bioethanol. In the first generation of the so-called biofuels, only more easily accessible sugars from crops such as corn, sugarcane, or sugar beets are extracted, and the resulting glucose is fermented to obtain bioethanol. Unfortunately, the impact of this first-generation bioethanol on the emission of greenhouse gas is only moderate.1 Second-generation biofuels will use the agricultural waste part of crops or low-input highdiversity grassland biomass, with a much higher impact on the greenhouse gas emission.2,3 The sugars in this kind of biomass are located in the plant's cell walls, which are composed of lignin, hemicelluloses, and cellulose. Mannose, galactose, and the C5 sugars of the lignocellulosic biomass are located in hemicelluloses, while the most important stock of the C6 sugar glucose is found in cellulose. Both C5 and C6 sugars may be fermented separately or simultaneously to obtain bioethanol.46 Second-generation bioethanol production recruits its enzymatic arsenal from bacteria and fungi, where cellulose fibers are degraded by specialized endoprocessive and processive cellulases, optimized for plant cell wall digestion and acting in a synergistic manner. Their first digestion produces cellobiose, which is further transformed to glucose by -glucosidases. Depending on the organism, these specialized glycoside hydrolases (GHs) are secreted independently or assembled on scaffolding proteins to form multienzyme complexes called cellulosomes.7,8 Independent-acting cellulases or cellulosomal components are highly modular proteins, which may be composed of numerous modules of various functions: catalytic modules, protein/protein interaction modules, cellulose-binding modules, and others. Catalytic modules of GHs have been classified according to their catalytic mechanism and their structural resemblance and are accessible via the CAZY database.911 Cellulases

can be found in six different GH families, namely, GH5, GH6, GH7, GH8, GH9, and GH48. A key enzyme and one of the three most abundant proteins in cellulosomes when purified from cellulose is a cellulase containing a catalytic module from family GH48.12,13 All GH48 cellulases characterized until now are processive cellulases, liberating cellobiose moieties from the reducing end of the cellulose chain by an inverting cleavage mechanism. An initial endo cut preceding the processive action has been postulated as well.13 Crystal structures of native and mutated inactive GH48 catalytic modules from two bacteria, Cel48A (previously called CelS) from Clostridium thermocellum and Cel48F (previously called CelF) from Clostridium cellulolyticum, have been solved in complex with either cellobiose and cellohexaose (both species) or cellotetraose, cellobiitol, and short thiooligosaccharide inhibitors (C. cellulolyticum only).1416 These complexes made it possible to identify 10 possible sugar subsites: 7 in the tunnel before the cleavage site (no continuous sugar chain spanning the 1 site could be observed) and 3 others in the following shorter cleft part of the active center. In consequence, a pathway of the substrate in the active site was traced, but the mechanism of the propagation of the sugar chain leading to the processive action remains unclear. In this report, we present structures of Cel48F mutants in complex with long hemithiocellooligosaccharides that reveal two different pathways of sugar propagation around the cleavage site, leading to a new hypothesis of sugar translocation throughout the active site of GH48 catalytic modules. This information may be key in elucidating the still unknown pathway during the processive action of these cellulases.

Results and Discussion

In order to complete the substrate interaction scheme all along the active site of cellulase Cel48F of C. cellulolyticum and to better understand its processive mechanism, we also decided to analyze Cel48F in complex with long thiocellooligosachharide inhibitors instead of complexes with cellooligosaccharides. The presence of the thioglycosidic bond prevents the formation of an intramolecular hydrogen bond between the GlcC2-OH and the following GlcC6-OH, which contributes to the increase in the solubility of the corresponding sugar compound in

Processive Action in Cel48F

501

Fig. 1. Comparison of the geometry of a glycosidic (left panel) and a thioglycosidic (right panel) bond.

aqueous media. Within the thiooligosaccharides used for preparing the complexes, each second glycosidic bond is replaced by a thioglycosidic bond, resulting in an alternating thioglycosidic/ glycosidic linkage pattern. Their improved solubility allowed us to perform soaking experiments with hemithiocellooligosaccharide chains of 10 and 1214 sugar moieties, therefore representing up to two times the length of cellohexaose, the longest natural cellooligosaccharide sufficiently soluble under similar aqueous conditions. Although the geometry of the thioglycosidic bond differs slightly from that of the glycosidic one (Fig. 1), the physicochemical resemblance of glycosidic chains to chains containing alternating thioglycosidic/glycosidic linkage is sufficient to allow the latter to occupy the same positions in the active site of Cel48F (shown later in this article). We soaked crystals of the mutants E55Q and E44Q of Cel48F with hemithiocellooligosaccharide inhibitors and solved their structures at 2.0 and 1.9 resolution, respectively. All crystallographic data are summarized in Table 1.
Table 1. Statistics of data collection and refinement
CelF/
D-glucose

All attempts to crystallize Cel48F without a ligand failed, but, finally, we succeeded in solving the structure of native Cel48F in the ligand-free form at 1.85- resolution by introducing a small seeding crystal of Cel48F/cellobiose complex into a droplet that contained the crystallization buffer plus D-glucose. The electron density of a single glucose molecule could not be detected in this structure, but the D-glucose seems to favor crystal growth of Cel48F in its observed conformation, which could not be obtained in droplets without any sugar supplement.14,16 The ligand-free structure obtained from the cocrystallization experiment with D-glucose revealed the flexibility of some active-site side chains, which appeared more static in the structures of ligand complexes and therefore provides insight into some dynamic aspects that may be involved in its catalytic mechanism. Surprisingly, the hemithiocellooligosaccharides in the two complex structures follow two different trajectories in the active site, which will be called the upper and the lower pathway. Description of the E55Q/hemithiocellooligosaccharide complex To obtain information on the conformation of the sugar in subsite 1 during the enzymatic reaction, we solved the structure of the inactive mutant E55Q in complex with a hemithiocellooligosaccharide composed of 10 glucose moieties. The electron density of nine sugar subsites continuously spanning positions + 2 to 7 along the lower pathway appeared in the unrefined Fourier difference electron density map (Fig. 2a). In contrast to previously published structures, subsite 1 was not empty, as continuous electron density between subsites 2 and + 1 was observed. The observed density in the 1 position was less well defined than that in the other subsites but allowed us to fit a glucose moiety in a slightly disturbed 4C1 chair conformation. Only the O3 hydroxy group of this sugar subunit is stabilized by H bonding, one with OD2 of Asp230 and a second with the hydroxy group of Tyr323. No sugar moiety occupies subsite + 3 on the leaving group side as observed in the published cellohexaose complex;15 subsite + 2 was the last one occupied in the hemithiocellooligosaccharide complex.

E55Q/IG10 302.0 61.24, 84.76, 121.74 95.2 (87.7) 3.9 (3.3) 6.7 (3.0) 7.8 (19.9) 414 15.8 20.5 18.2 0.013 1.3 14.0 2qno

E44Q/IG12 501.90 61.48, 84.72, 121.81 99.7 (99.0) 3.7 (3.2) 8.1 (1.7) 7.7 (39.9) 374 19.2 35.1 20.7 0.005 1.2 18.2 1g9j

Resolution () Cell: a, b, c () Completeness (%) Multiplicity I/I Rsym (%) Number of water molecules Average B-factor protein (2) Average B-factor inhibitor (2) Rfree (%) rms bonds () rms angles () R (%) Protein Data Bank ID

401.85 61.53, 84.77, 122.12 98.2 (88.9) 3.8 (2.3) 6.8 (1.9) 9.1 (35.4) 397 19.4 19.5 0.005 1.3 16.5 1g9g

Each data set was collected with a single crystal. The values in parentheses correspond to the highest shell of resolution. Ihkl Ihkl,i|/hkli|Ihkl| , where i is the number Rsym = hkli| of reflections hkl.

502

Processive Action in Cel48F

Fig. 2. (a) Unrefined Fourier difference density map at 3 level around the sugar chain of E55Q/thiocellodecaose, calculated after rigid-body refinement (R-factor = 19.8) with the structure of E55Q/cellohexaose complex (Protein Data Bank ID: 1fbw) without the cellohexaose.15 The sugar positions of the thiocellulose of the refined structure E55Q/ thiocellodecaose are shown. (b) Refined structure of the E44Q/hemithiocellooligosaccharide complex. Residues that are important in the interaction with the hemithiocellooligosaccharide are indicated. The electron density around the sugar chain is drawn at 1 level.

The subsite positions of all sugar moieties were equivalent to those previously published for the cellohexaose complex (rmsd of 0.67 for all matching atoms), an observation that stresses the stereochemical resemblance between the inhibitor and the natural substrate (Fig. 3a). The entire inhibitor was composed of 10 sugar moieties, but the position of the 10th sugar moiety, obviously located outside the tunnel entrance beyond subsite 7, was not detectable in the electron density map. Although the overall orientation of the

sugar moieties is in good agreement with the previously published complexes, the hemithiocellooligosaccharide chain is forced to an unusual orientation around the sugar moiety in subsite 1. Here, the orientation of its C6-OH group does not follow the normally observed 180 alternating rotation along the cellooligosaccharide molecule but is extremely twisted outside the ideal rotation. Its rotation is only halfway, about 90, between subsites 1 and 2 and returns to a normal twist of nearly 180 between subsites 1 and + 1. The same

Processive Action in Cel48F

503

Fig. 3. (a) Superposition of 1.5 cellohexaose molecules as observed in the cellohexaose complex (Protein Data Bank ID:1fbw; in violet) and the continuous fragment of nine sugar moieties from the hemithiocellooligosaccharide (in green) in the active-site tunnel of mutant enzyme E55Q. (b) Superposition of the refined structures of the hemithiocellooligosaccharides in mutant E55Q (in green) and in E44Q (in magenta).

orientation in subsite 1 has been observed previously in a cellotetraose complex with mutation E55Q, but in this case, the connectivity of the sugar chain was not continuous and subsite 1 was occupied by the nonreducing end of a cellotetraose molecule intruding from the leaving group side.15 Description of the E44Q/hemithiocellooligosaccharide complex On the basis of the location of sugars in the catalytic site, three residues were candidates for providing the corresponding acid and base in the cleavage reaction, namely, Glu55, Glu44, and Asp230. Glu55 was suggested to play the role of the proton donor.16 As residue Glu44 might play a role in the cleavage reaction, we replaced this glutamate by a glutamine to test the importance of the charge at this position. The Cel48FE44Q mutated enzyme was found to retain around 1% of the activity of the wild-type

protein on phosphoric acid-swollen cellulose, while Cel48FE55Q was completely inactive. The wildtype Cel48F weakly degrades the substituted soluble carboxymethyl cellulose.13 This activity was suggested to be related to the initial endo mode of action. The E44Q enzyme retained this weak activity, suggesting that the initial endo cutting activity was not greatly affected by the charge being removed. We cocrystallized this E44Q mutant in the presence of a hemithiocellooligosaccharide composed of 12 sugar moieties and solved the structure of the complex. Only 8 of the 12 sugar moieties were visible in the electron density map (Fig. 2b). To our surprise, the sugar moieties did not occupy the same subsites as observed in the complex of the inactive form E55Q. Taking into account that the repetitive unit in the cellooligosaccharide chain is a cellobiose, the sugar positions of the chain in this complex were shifted for nearly half a repetitive unit backward compared to the E55Q/

504 cellohexaose complex (Fig. 3b). In the E44Q/hemithiocellooligosaccharide complex, the sugar chain follows the same pathway along the tunnel region but does not descend into the depression that follows the tunnel part and contains the catalytic residue Glu55 as observed in the case of the E55Q complex. Instead, it continues to follow a linear pathway located about 6 above the depression and ending between subsites + 1 and + 2, where it forms a triple hydrogen bond with the mutated residue E44Q (Fig. 4). This newly observed upper pathway is only accessible after a rotation of the side chain of Tyr403 for about 120, which blocks the access to the upper part of the depression in the cellohexaose complex. The rotation of Tyr403 has two effects: firstly, it opens this pathway, and secondly, it offers an additional stabilization for the sugar chain via a hydrophobic stacking interaction. In total, two new stacking interactions involving the sugar chain can be observed in the upper pathway: one additional with Tyr403 and another with Phe180 located on the other side of the sugar chain replacing the stacking interaction provided by Trp154 at the equivalent subsite 2 of the lower pathway. The spacing between Tyr403 and Phe180 is in the range of half a sugar residue, which follows the pattern of stacking residues in the tunnel part. That kind of architecture is thought to reduce the sliding energy of the sugar chain along the pathway and has been observed for processive cellulases or other sugar transportation channels.15,17,18 The two new stacking interactions prolong, therefore, the greasy slide of the tunnel up to the positions of the cellobiose leaving group site, located in the open cleft region. All interactions of the sugar chain provided by these new subsites are summarized in Fig. 4. In addition to the rotation of Tyr403, the reorientation of five other residues in the active site, namely, Glu55, Gln44, Met42, Trp411, and Trp417, could be observed in the E44Q complex. In contrast to previously described complexes, Glu55 is not in contact with a sugar residue. It is rotated about 120 around 1, replacing with its OE1 and OE2 atoms the positions of two water molecules present in the E55Q complex. In this new orientation, it forms a hydrogen-bond network with Asp230 OD1 and OD2 atoms and with the NH1 atom of Arg234 (Fig. 5). Similar environments around the proton donor have been observed in other GH families, for example, amylases and CGTases from GH family 13, which hydrolyzes the -glycosidic bond with a retaining mechanism, or xylanases and cellulases from GH family 8, which cleaves the -glycosidic bond by an inverting mechanism.1922 In all three cases, even if their mechanisms are different, the associated aspartic acid is thought to activate the proton donor by elevating its pKa and to stabilize the distorted reaction intermediate in the 1 position. The replacement of this residue in Cel48F by asparagine leads to an inactivation of the enzyme, as it has been observed in CGT from Bacillus circulans strain 251, -amylase from Aspergillus oryzae, or

Processive Action in Cel48F

CGT from B. circulans strain 8 (here, mutation to alanine).2325 The replacement of Glu44 by a glutamine induces a significant reorientation of the corresponding side chain. A rotation of about 60 around 1 swings the amide deeper in the active-site cleft and breaks its H bond with His36. The resulting liberated space is occupied by a chloride ion, located closely to the former position of the Glu44 OE2 atom (at about 1.5 away). The chloride ion is coordinated by His36 and Arg609 and may compensate the negative charge eliminated by the mutation. Both aromatic residues Trp417 and Trp411 are in double conformations and may be grouped in two pairs of related orientations. In the first pair, the orientations do not differ from those already observed in the other complex structures, whereas in the second pair, Trp411 and Trp417 are rotated around 1 90 and 20, respectively. With this newly observed orientation, Trp411 intrudes into subsite + 2 of the lower pathway and is in hydrophobic contact with the reducing end of the sugar chain in the upper pathway. One is tempted to suggest that alternation between the two pairs of different orientations of these aromatic side chains may participate in the stabilization or liberation of the sugar moieties in the leaving group side. The ligand-free Cel48F structure In the ligand-free native structure, the orientations of the active-site residues are closer to those observed in the E55Q structure than to those seen in the E44Q model. All side chains in the active site are pointing in the same direction as in the E55Q complex, with the exception of three residues, namely, Glu55, Trp417, and Lys224. The first, Glu55, adopts in the ligand-free native structure a double conformation, occupying both orientations observed in the E44Q and E55Q complexes. The second, Trp417, is located between its position observed in the E55Q complex and the location of the more conservative tryptophan position in the E44Q complex. The third, Lys224, is in double conformation: one side-chain orientation is the same as already reported for the E44Q and E55Q complexes and the other is a new orientation in which the side chain intrudes deeply in the active-site tunnel. This makes in total three different side-chain orientations for Lys224, which stresses its role as a flexible contact in the sugar transportation process. Implications for the processive mechanism Cellobiose, composed of two glucose moieties, is the repetitive unit of the cellulose chain and is the main product of the digestion process. Only one of the two sugar orientations in subsite 1 is stabilized and permits the cleavage reaction. An interesting question about the processive action of cellobiohydrolases is now how a cellobiose moiety passes smoothly across the cleavage site. If it slides by only one glucose unit across the active site, it occupies

Processive Action in Cel48F Fig. 4. Hydrogen-bonding network of the hemithiocellooligosaccharide with the mutated enzyme E44Q. Distances are indicated in angstroms. Residues that perform stacking interactions with the sugar chain are drawn in thicker lines.

505

506

Processive Action in Cel48F

Fig. 5. Hydrogen-bonding network around the proposed proton donor Glu55 as observed in the E44Q/ hemithiocellooligosaccharide structure (a) and in the E55Q/hemithiocellooligosaccharide complex (b). Red spheres are water molecules; sugar moieties from the hemithiocellooligosaccharide are colored orange.

two positions, which have been named productive and nonproductive binding. In Cel6A, which is a processive cellulase from Humicola insolens, nonproductive binding seems to be less stable, and it imposes reorientations of side chains and small rigid-body movements around the cleavage site.18 The present data appear to be consistent with an alternative solution of the problem for GHs of family 48, supporting a mechanism that evades nonproductive binding around the cleavage site. Our two new structures of active-site mutants with hemithiocellooligosaccharides represent two key steps in the processive transportation and digestion process along the active site of Cel48F. The mutation E44Q has obviously interrupted the low-energy sliding process along the tunnel part, before the last two residues could integrate in their final location in the leaving group site, whereas the mutation E55Q has prevented the cleavage of the glycosidic bond and the release of the cellobiose unit from the leaving group site. The other published complexed structures so far additionally support the presence of multiple low-energy positions in the tunnel part and show the compatibility of the hemithiocellooligosaccharide inhibitor with binding sites occupied by the natural substrate.15,16 Finally, the ligand-free native structure shows the conformations of all active-site residues without a bound substrate. The two observed pathways within the active site have their specific features optimized for their

different functions. The upper, linear pathway offers stacking interactions with aromatic side chains, which build a low-energy sliding pathway up to the leaving group site, optimized for sugar transport. The lower, descending pathway disturbs the internal sugar chain stabilization around the cleavage site and brings the scissile glycosidic bond in range of the proton donor. In contrast to the observed multiple binding sites in the tunnel part, only one sugar conformation was observed in each leaving group subsite.15 This may suggest that subsites 1 and + 1 (and maybe subsite + 2 as well) are each specifically tailored to stabilize the transition state of only one of the two differently oriented glucose moieties. This specificity would disfavor low-energy sliding, which might explain the presence of the two observed pathways across the cleavage site. Recently, the comparison of Cel48A from C. thermocellum with the active site of Cel8A from the same organism proposed that Tyr323, Asp230, and Glu55 may be catalytic residues responsible for the disruption of the sugar chain in the GH48 cellulases.14 These residues do not interact with the sugar chain during its sliding process in the upper pathway but, in contrast, are important binding partners in subsites 1 and + 1 in the lower pathway. During the transition from the upper to the lower pathway, several conformational changes in the enzyme and the substrate chain must occur. Tyr403 seems to be critical for the gating between both

Processive Action in Cel48F

507 site + 1, but it is 4.9 away from the OH of Tyr323. Although the new structures of mutated Cel48F with bound hemithiocellooligosaccharides shed some light on the processive action of GH48 catalytic modules, the details of its cleavage reaction still remain unclear. Conclusion The complexes of the Cel48F mutants E44Q and E55Q with hemithiocellooligosaccharides further strengthen our understanding of the enzyme's processive action along the active site. The E55Q complex confirms the possibility of accommodating the hemithiocellooligosaccharides in the active site despite their geometrical differences with the natural substrates and gives a first view of the conformation of the sugar in subsite 1. The native structure of Cel48F without any bound sugar in the active site confirms the two orientations of the catalytic proton donor Glu55 observed in the two mutant structures and adds information about side-chain orientations of residues that interact with the substrate chain during catalysis without its presence. Finally, the E44Q complex seems to be a probable missing link structure that proposes an alternative sugar propagation pathway from the tunnel into the active site. It reveals new subsite positions, stabilized by stacking interactions and hydrogen bonds, which indicate a new intermediate pathway during the sliding process. The presence of two pathways that are occupied during either the processive action or the cleavage reaction to maintain low-energy sliding into the leaving group subsites across the cleavage site is a newly observed concept in processive enzymes. Therefore, the presented data give us new clues to understanding the structure/function relationship of the complex architecture of GH family 48 active sites.

pathways. It swings inside the tunnel and seems to push the sugar chain in the subsites around the cleavage site. This residue is involved in both pathways: as a stacking partner in the upper one and as an H-bond donor in the lower one. The central stacking partner in the leaving group sites + 1 and + 2, Trp411, adopts two distinct conformations in the E44Q complex, like Tyr403, and may be a second gating residue. Its function could be to limit the sugar sliding in the upper pathway and to temporally stabilize the leaving group in the lower one. The orientation of the last three sugar moieties at the reducing end of the sugar chain has to be rotated for nearly 180 during their descent from the upper to the lower pathway to fit in subsites + 2 to 1 in the conformation observed in the E55Q complex. Because the fourth glucose moiety is already correctly oriented to fit in subsite 2, the rotation will impose a twist of the glycosidic bond in subsite 1. This subsite is the proposed transition state binding site, tailored to stabilize a high-energy conformation. In the E55Q complex, the sugar occupying subsite 1 has a less well defined electron density and was fitted in a slightly disturbed 4C1 conformation. We suppose that it has a potential to be flexible and may adopt a more distorted geometry during catalysis, as it has been observed in the active site of the structurally related family 8 glucoside hydrolases.26 The E44Q complex sheds light on the question why there is so much space above the cleavage site: to permit the rotation of the last three sugar residues. Observations at the cleavage site Obviously, the proton donor of the cleavage reaction Glu55 adopts two conformations during catalysis: If the substrate chain occupies the lower pathway, it is H bonded to the scissile glycosidic bond, or if the sugar chain follows the upper pathway, Glu55 is H bonded to Asp230 and Arg234 (Fig. 5). In the ligand-free native structure, both conformations of Glu55 could be observed. Asp230, which is normally H-bonded to Glu55 or to water molecules, adopts a double conformation in the E55Q/hemithiocellooligosaccharide complex. It is either H-bonded to the OH-2 of the sugar in subsite 1 by conserving its usually observed conformation or bound to Arg421 after a rotation of 110 around 1. The double conformation of Asp230 may indicate conformational changes of this residue when subsite 1 is occupied during catalysis. The identity of the catalytic base in the cleavage reaction is still not obvious. Guimaraes et al. propose that Tyr323 is stabilizing a water molecule that may attack the C4 carbon of the sugar moiety in subsite 1.14 This hypothesis is based on a structural comparison of Cel48A with GH8 enzymes. Our data do not confirm such a hypothesis but do not reject it as well. We also observe a water molecule that is well positioned 4.2 above the C1 of the scissile bond and makes H bonds to the O2 of the sugar moiety in the 1 subsite and to the O6 of the glucose moiety in sub-

Materials and Methods

Cel48FE44Q construction Mutation E55Q was obtained as reported elsewhere.15 The same double PCR mutagenesis technique27 was used to generate the point mutation E44Q in the cel48F gene carried by the pETFp plasmid.28 The site-directed mutagenesis was performed using the complementary mutagenic oligonucleotides Fm44f (5-ACACTGATGGTCCAAGCTCCTGACTACGGA-3) and Fm44r (5-TCCGTAGTCAGGAGCTT G GACCATCAGTGT-3 ) to introduce the CAA Gln codon in place of the GAA Glu44 codon (mutagenic nucleotides in italic type). Expand High Fidelity PCR System (Roche) was used in the PCR experiments. The mutated 5 part of the cassette was synthesized using the forward primer F18 (5-CCCTATACATATG GCTTCAAGTCCTGCAAAC -3 ) and the mutagenic Fm44f reverse primer. F18 carrying an NdeI site (underlined) was located at the beginning of the mature protein-encoding sequence (hybridizing region in boldface type). The mutated 3 part of the cassette was

508
synthesized using the mutagenic Fm44f forward primer and the reverse primer F22 (5-GCCTTGAAGTTTCATCTCC-3) located downstream from the unique KpnI site in the cel48F sequence. These two purified PCR products were mixed and amplified using F22 and F18 to generate the whole mutated cassette. The amplified fragment was purified (Qiaquick, Qiagen), digested by NdeI and KpnI, and then cloned between the NdeI and KpnI sites into pETFc to replace the wild-type sequence. The resulting plasmid carrying the mutation was called pETF44. The successful insertion of the fragment was verified by sequencing. The mutated proteins E44Q and E55Q were produced from BL21(DE3)[pETF44] and BL21(DE3)[pETF55] cultures, respectively. They were purified, and their enzymatic activity was assayed using phosphoric acid-swollen cellulose as substrate following the procedures previously reported for Cel48F wild type.13 Crystallization and 3D structure solution Crystals suitable for x-ray diffraction were grown using vapor diffusion and microseeding techniques with microcrystals of the Cel48F/IG4 complex, which has been described previously.16 Crystals were grown under conditions as described for the Cel48F/IG4 except for E44Q where the crystallization buffer contained 20 mM CaCl2 instead of MgCl2. Crystals of native Cel48F were grown in the presence of 8 mM D-glucose, those of the mutant E44Q were grown in the presence of 8 mM hemithiocellododecaose, and those of the mutant E55Q were grown in the presence of 2 mM cellobiose.29 The observed E55Q/ hemithiocellooligosaccharide complex was obtained by soaking the crystal with a hemithiocelloligosaccharide composed of 10 sugar moieties, 14 h prior to mounting the crystal in the capillary. This soak was performed by adding 1 l of a solution of 20 mM hemithiocellooligosaccharide in crystallization buffer to the crystal-containing droplet whose volume was approximately 4 l. The diffraction data sets of all crystals were collected using glass capillaries at 15 C with a MAR 345 image plate scanner. In the case of the E44Q/hemithiocellooligosaccharide complex, we used CuK radiation from the in-house Nonius FU 581 rotating anode generator equipped with a graphite monochromator, whereas the data on the native Cel48F structure were collected at the BM30A beamline (FIP) of the European Synchrotron Radiation Facility, Grenoble, France, using radiation at a wavelength of 0.98 and the E55Q/thiocelloligosaccharide complex data set was collected at the EMBL Outstation of the Deutsches Elektronen-Synchrotron, Hamburg, at the X11 beamline at a wavelength of 0.91 . The crystals were sufficiently stable to allow the collection of an entire data set from one single crystal. All crystals were isomorphous to the previously reported Cel48F/IG4 complex16 and belonged to the space group P212121. The crystal data of all complexes are summarized in Table 1. The data were processed and reduced using the DENZO program30 and the CCP4 package.31 The native, ligand-free Cel48F structure and the Cel48FE44Q complex were solved using the Crystallography and NMR Systems (CNS) program.32 The E55Q complex was initially solved using the CNS program and was additionally postrefined using the refmac program of the CCP4 suite. The refinement procedure began with 20 cycles of rigid-body refinement, using the protein chain and the calcium-ion position of Cel48F/IG4 complex as the search model. The bulk-solvent correction33 and the weighting

Processive Action in Cel48F factor WA were automatically recalculated after each step of the refinement. The rigid-body refinement was followed by a simulated annealing procedure at 3000 K. Manual fitting was performed for residues that were still out of density and for the mutated residues. The watpick procedure of the CNS package was then used to insert water molecules at 3 difference density level. Water molecules were kept if their corresponding electron density reached a 1 level in the (2Fo Fc) electron density map. This selection significantly reduced the difference between the R-factor and the Rfree-factor.34 Finally, the substrate/inhibitor molecules were inserted in the model, beginning with the most obvious positions at the leaving group side (+ 1, + 2) and followed by a positional refinement. Accurate bond parameters for the refinement of thiooligosaccharides were taken from the Engh and Huber parameters35 of the CNS package and thioglycosidic bond parameters from the structure of 2,3,4,6-tetra-Oacetyl-1-S-benzhydroximoyl--D-glucopyranose.36 The program O was used for all model-building stages.37 After completion of the model, a restrained individual B-factor refinement was performed, followed by a final correction of the positions of the water molecules and positional refinement of all atoms. The E55Q complex was further refined with the refmac program of the CCP4 package.38 Five cycles of restrained maximum likelihood refinement followed by 1 cycle of ARP/wARP to complete the solvent model and, again, 20 cycles of maximum likelihood refinement were used to recalculate the structure.39 The hemithiocellooligosaccharide was refined using the same parameters for the thioglycosidic link as before. The refined structures of the complexes were superimposed, with the least-squares function of the program O, by comparing the atoms of the protein backbone with the Cel48F structure, which served as the reference, to elucidate differences in the positions of the substrate/ inhibitor molecules and in the side-chain orientations. Figures 2 and 3 were calculated using the program PyMOL.

Acknowledgements
We thank the staff of the BM30A beamline at the European Synchrotron Radiation Facility and the staff of the X11 beamline at the Deutsches Elektronen-Synchrotron for their assistance during the collection of the E55Q complex and the ligand-free structure, respectively. The work was supported by the Centre National de la Recherche Scientifique and financed by the European Community (Contract BIO2-CT 94-3018 and Contract BIO4-CT 97-2303).

References
1. Farrell, A. E., Plevin, R. J., Turner, B. T., Jones, A. D., O'Hare, M. & Kammen, D. M. (2006). Ethanol can contribute to energy and environmental goals. Science, 311, 506508. 2. Kim, S. & Dale, B. E. (2004). Global potential bioethanol production from wasted crops and crop residues. Biomass Bioenergy, 26, 361375. 3. Tilman, D., Hill, J. & Lehman, C. (2006). Carbonnegative biofuels from low-input high-diversity grassland biomass. Science, 314, 15981600.

Processive Action in Cel48F 4. Sedlak, M. & Ho, N. W. (2004). Production of ethanol from cellulosic biomass hydrolysates using genetically engineered Saccharomyces yeast capable of cofermenting glucose and xylose. Appl. Biochem. Biotechnol. 113116, 403416. 5. Ho, N. W., Chen, Z., Brainard, A. P. & Sedlak, M. (1999). Successful design and development of genetically engineered Saccharomyces yeasts for effective cofermentation of glucose and xylose from cellulosic biomass to fuel ethanol. Adv. Biochem. Eng./Biotechnol. 65, 163192. 6. Hahn-Hagerdal, B., Galbe, M., Gorwa-Grauslund, M. F., Liden, G. & Zacchi, G. (2006). Bio-ethanol the fuel of tomorrow from the residues of today. Trends Biotechnol. 24, 549556. 7. Belaich, J. P., Tardif, C., Belaich, A. & Gaudin, C. (1997). The cellulolytic system of Clostridium cellulolyticum. J. Biotechnol. 57, 314. 8. Schwarz, W. H. (2001). The cellulosome and cellulose degradation by anaerobic bacteria. Appl. Microbiol. Biotechnol. 56, 634649. 9. Henrissat, B. & Davies, G. (1997). Structural and sequence-based classification of glycoside hydrolases. Curr. Opin. Struct. Biol. 7, 637644. 10. Coutinho, P. M. & Henrissat, B. (1999). Carbohydrateactive enzymes: an integrated database approach. In Recent Advances in Carbohydrate Bioengineering (Gilbert, H. J., Davies, G., Henrissat, B. & Svensson, B., eds), The Royal Society of Chemistry, Cambridge, UK. 11. Henrissat, B. & Bairoch, A. (1996). Updating the sequence-based classification of glycosyl hydrolases. Biochem. J. 316, 695696. 12. Steenbakkers, P. J., Freelove, A., Van Cranenbroek, B., Sweegers, B. M., Harhangi, H. R., Vogels, G. D. et al. (2002). The major component of the cellulosomes of anaerobic fungi from the genus Piromyces is a family 48 glycoside hydrolase. DNA Sequence, 13, 313320. 13. Reverbel-Leroy, C., Pags, S., Belaich, A., Belaich, J. P. & Tardiff, C. (1997). The processive endocellulase CelF, a major component of the Clostridium cellulolyticum cellulosome: purification and characterization of the recombinant form. J. Bacteriol. 179, 4652. 14. Guimaraes, B. G., Souchon, H., Lytle, B. L., David Wu, J. H. & Alzari, P. M. (2002). The crystal structure and catalytic mechanism of cellobiohydrolase CelS, the major enzymatic component of the Clostridium thermocellum cellulosome. J. Mol. Biol. 320, 587596. 15. Parsiegla, G., Reverbel-Leroy, C., Tardif, C., Belaich, J. P., Driguez, H. & Haser, R. (2000). Crystal structures of the cellulase Cel48F in complex with inhibitors and substrates give insights into its processive action. Biochemistry, 39, 1123811246. 16. Parsiegla, G., Juy, M., Reverbel-Leroy, C., Tardif, C., Belaich, J. P., Driguez, H. & Haser, R. (1998). The crystal structure of the processive endocellulase CelF of Clostridium cellulolyticum in complex with a thiooligosaccharide inhibitor at 2.0 resolution. EMBO J. 17, 55515562. 17. Meyer, J. E. W. & Schulz, G. E. (1997). Energy profile of maltooligosaccharide permeation through maltoporin as derived from the structure and from a statistical analysis of saccharideprotein interactions. Protein Sci. 6, 10841091. 18. Varrot, A., Frandsen, T. P., von Ossowski, I., Boyer, V., Cottaz, S., Driguez, H. et al. (2003). Structural basis for ligand binding and processivity in cellobiohydrolase Cel6A from Humicola insolens. Structure, 11, 855864. 19. Kadziola, A., Abe, J., Svensson, B. & Haser, R. (1994). Crystal and molecular structure of barley alphaamylase. J. Mol. Biol. 239, 104121.

509
20. Klein, C. & Schulz, G. E. (1991). Structure of cyclodextrin glycosyltransferase refined at 2.0 resolution. J. Mol. Biol. 217, 737750. 21. Alzari, P. M., Souchon, H. & Dominguez, R. (1996). The crystal structure of endoglucanase CelA, a family 8 glycosyl hydrolase from Clostridium thermocellum. Structure, 4, 265275. 22. Van Petegem, F., Collins, T., Meuwis, M. A., Gerday, C., Feller, G. & Van Beeumen, J. (2003). The structure of a cold-adapted family 8 xylanase at 1.3 resolution. Structural adaptations to cold and investigation of the active site. J. Biol. Chem. 278, 75317539. 23. Klein, C., Hollender, J., Bender, H. & Schulz, G. E. (1992). Catalytic center of cyclodextrin glycosyltransferase derived from X-ray structure analysis combined with site-directed mutagenesis. Biochemistry, 31, 87408746. 24. Nagashima, T., Tada, S., Kitamoto, K., Gomi, K., Kumagai, C. & Toda, H. (1992). Site-directed mutagenesis of catalytic active-site residues of Takaamylase A. Biosci. Biotechnol. Biochem. 56, 207210. 25. Knegtel, R. M., Strokopytov, B., Penninga, D., Faber, O. G., Rozeboom, H. J., Kalk, K. H. et al. (1995). Crystallographic studies of the interaction of cyclodextrin glycosyltransferase from Bacillus circulans strain 251 with natural substrates and products. J. Biol. Chem. 270, 2925629264. 26. Guerin, D. M., Lascombe, M. B., Costabel, M., Souchon, H., Lamzin, V., Beguin, P. & Alzari, P. M. (2002). Atomic (0.94 ) resolution structure of an inverting glycosidase in complex with substrate. J. Mol. Biol. 316, 10611069. 27. Higuchi, R., Krummel, B. & Saiki, R. K. (1988). A general method of in vitro preparation and specific mutagenesis of DNA fragments: study of protein and DNA interactions. Nucleic Acids Res. 16, 73517367. 28. Reverbel-Leroy, C., Belaich, A., Bernadac, A., Gaudin, C., Belaich, J. P. & Tardiff, C. (1996). Molecular study and overexpression of the Clostridium cellulolyticum celF cellulase gene in Escherichia coli. Microbiology, 142, 10131023. 29. Moreau, V. & Driguez, H. (1996). Enzymic synthesis of hemithiocellodextrins. J. Chem. Soc., Perkin Trans. 1, 1 6, 525527. 30. Otwinowski, Z. & Minor, W. (1997). Processing of X-ray diffraction data collected in oscillation mode. In Methods in Enzymology (Carter, C. W. & Sweet, R. M., eds), vol. 276, Academic Press, London, UK. 31. Collaborative Computational Project, Number 4 (1994). The CCP4 suite: programs for protein crystallography. Acta Crystallogr., Sect. D: Biol. Crystallogr. 50, 760763. 32. Brnger, A. T., Adams, P. D., Clore, G. M., DeLano, W. L., Gros, P., Grosse-Kunstleve, R. W. et al. (1998). Crystallography & NMR system: a new software suite for macromolecular structure determination. Acta Crystallogr., Sect. D: Biol. Crystallogr. 54, 905921. 33. Jiang, J. S. & Brunger, A. T. (1994). Protein hydration observed by x-ray diffraction: solvation properties of penecillopepsion and neuraminidase crystal structures. J. Mol. Biol. 243, 100115. 34. Brnger, A. T. (1992). The free R value: a novel statistical quantity for assessing the accuracy of crystal structures. Nature, 355, 472474. 35. Engh, R. A. & Huber, R. (1991). Accurate bond and angle parameters for X-ray protein structure refinement. Acta Crystallogr., Sect. A, 47, 392400. 36. Durier, V. & Driguez, H. (1992). Stereochemical investigation of 2,3,4,6-tetra-O-acetyl-1-S-benzhy-

510
droximoyl--D-glucopyranose. Acta Crystallogr., Sect. C: Cryst. Struct. Commun, 48, 17911794. 37. Jones, T. A., Zou, J. Y., Cowan, S. W. & Kjeldgaard, M. (1991). Improved methods for building protein models in electron density maps and the location of errors in these models. Acta Crystallogr., Sect. A: Found. Crystallogr. 47, 110119.

Processive Action in Cel48F 38. Murshudov, G. N., Vagin, A. A. & Dodson, E. J. (1997). Refinement of macromolecular structures by the maximum-likelihood method. Acta Crystallogr., Sect. D: Biol. Crystallogr. 53, 240255. 39. Perrakis, A., Morris, R. & Lamzin, V. S. (1999). Automated protein model building combined with iterative structure refinement. Nat. Struct. Biol. 6, 458463.