structre of mutant cellulose

8/13/2019 Structre of Mutant Cellulose

1/12

Structures of Mutants of Cellulase Cel48F ofClostridium cellulolyticumin Complex with LongHemithiocellooligosaccharides Give Rise to a New Viewof the Substrate Pathway during Processive Action

Goetz Parsiegla1, Corinne Reverbel2, Chantal Tardif2,Hugues Driguez3 and Richard Haser4

An efficient breakdown of lignocellulosic biomass is a prerequisite for theproduction of second-generation biofuels. Cellulases are key enzymes inthis process. We crystallized complexes between hemithio-cello-deca anddodecaoses and the inactive mutants E44Q and E55Q of the endo-processivecellulase Cel48F, one of the most abundant cellulases in cellulosomes fromClostridium cellulolyticum, to elucidate its processive mechanism. In bothcomplexes, the cellooligosaccharides occupy similar positions in the tunnelpart of the active site but are more or less buried into the cleft, which hoststhe active site. In the E44Q complex, it proceeds along the upper part of thecavity, while it occupies in the E55Q complex the same productive bindingsubsites in the lower part of the cavity that have previously been reported inCel48F/cellooligosaccharide complexes. In both cases, the sugar moieties

are stabilized by stacking interactions with aromatic side chains and Hbonds. The upper pathway is gated by Tyr403, which blocks its access in theE55Q complex and offers a new stacking interaction in the E44Q complex.The new structural data give rise to the hypothesis of a two-step mechanismin which processive action and chain disruption occupy different subsites atthe end of their trajectory. In the first part of the mechanism, the chain maysmoothly slide up to the leaving group site along the upper pathway, whilein the second part, the chain is cleaved in the already described productive

binding position located in the lower pathway. The solved native structureof Cel48F without any bound sugar in the active site confirms the two side-chain orientations of the proton donor Glu55 as observed in the complexstructures.

2007 Elsevier Ltd. All rights reserved.

1Laboratoire de l'Architectureet Fonction des MacromoleculesBiologiques, UMR 6098 CNRSand University of Aix-Marseille,Parc Scientifique etTechnologique de Luminy,Case 932, 163 Avenue deLuminy, 13288 MarseilleCedex 09, France2Unit de Bionergtiqueet Ingnierie des Protines,Institut de Biologie Structurale

et Microbiologie, CentreNational de la RechercheScientifique, 31 CheminJoseph-Aiguier, 13402 MarseilleCedex 20 and Universit deProvence, Place Victor Hugo,13331 Marseille Cedex 03,France3Centre de Recherches sur lesMacromolcules Vgtales,Centre National de la RechercheScientifique and UniversitJ. Fourier de Grenoble, BP53,

38041 Grenoble Cedex 9, France4Institut de Biologie et Chimiedes Protines UMR 5086,Laboratoire deBioCristallographie, CentreNational de la RechercheScientifique and UniversitLyon IFR 128 BioSciencesGerlandLyon Sud,7 Passage du Vercors,69367 Lyon Cedex 07, France

*Corresponding author.E-mail address:[email protected] used: GH, glycoside hydrolase; CNS, Computation and Neural Systems.

doi:10.1016/j.jmb.2007.10.039 J. Mol. Biol. (2008) 375, 499510

Available online at www.sciencedirect.com

0022-2836/$ - see front matter 2007 Elsevier Ltd. All rights reserved.
mailto:[email protected]://dx.doi.org/10.1016/j.jmb.2007.10.039http://dx.doi.org/10.1016/j.jmb.2007.10.039http://dx.doi.org/10.1016/j.jmb.2007.10.039mailto:[email protected]


2/12

Received 20 July 2007;received in revised form11 October 2007;accepted 15 October 2007Available online22 October 2007

Edited by G. SchulzKeywords:cellulase; GH family 48; hemithiocellooligosaccharide; complexstructure; processive action

Introduction

In times of growing public awareness of thedangers related to the increasing amount of carbondioxide in the atmosphere, alternative sources of

energy that do not participate in the global green-house effect become more and more attractive.One of the alternatives to the combustion of non-renewable carbon-based fuels is the consumption offuel obtained from renewable carbon sources suchas biomass. The most important liquid fuel pro-duced from biomass worldwide is bioethanol. In thefirst generation of the so-called biofuels,only moreeasily accessible sugars from crops such as corn,sugarcane, or sugar beets are extracted, and theresulting glucose is fermented to obtain bioethanol.Unfortunately, the impact of this first-generation

bioethanol on the emission of greenhouse gas is onlymoderate.1 Second-generation biofuels will use the

agricultural waste part of crops or low-input high-diversity grassland biomass, with a much higherimpact on the greenhouse gas emission.2,3 Thesugars in this kind of biomass are located in theplant's cell walls, which are composed of lignin,hemicelluloses, and cellulose. Mannose, galactose,and the C5 sugars of the lignocellulosic biomass arelocated in hemicelluloses, while the most importantstock of the C6 sugar glucose is found in cellulose.Both C5 and C6 sugars may be fermented separatelyor simultaneously to obtain bioethanol.46

Second-generation bioethanol production recruitsits enzymatic arsenal from bacteria and fungi, where

cellulose fibers are degraded by specialized endo-processive and processive cellulases, optimizedfor plant cell wall digestion and acting in a syner-gistic manner. Their first digestion producescellobiose, which is further transformed to glucose

by -glucosidases. Depending on the organism,these specialized glycoside hydrolases (GHs) aresecreted independently or assembled on scaffoldingproteins to form multienzyme complexes calledcellulosomes.7,8 Independent-acting cellulases orcellulosomal components are highly modular pro-teins, which may be composed of numerousmodules of various functions: catalytic modules,protein/protein interaction modules, cellulose-bind-

ing modules, and others. Catalytic modules ofGHs have been classified according to their catalyticmechanism and their structural resemblance andare 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 abundantproteins in cellulosomes whenpurifiedfromcelluloseis a cellulase containing a catalytic module fromfamily GH48.12,13 All GH48 cellulases characterizeduntil now are processive cellulases, liberating cello-

biose moieties from the reducing end of the cellulosechain by an inverting cleavage mechanism. An initialendo cut preceding the processive action has beenpostulated as well.13 Crystal structures of native andmutated inactive GH48 catalytic modules from two

bacteria, Cel48A (previously called CelS) fromClos-tridium thermocellum and Cel48F (previously calledCelF) from Clostridium cellulolyticum, have beensolved in complex with either cellobiose and cello-hexaose (both species)or cellotetraose, cellobiitol, andshortthiooligosaccharide inhibitors (C. cellulolyticumonly).1416 These complexes made it possible to

identify 10 possible sugar subsites: 7 in the tunnelbefore the cleavage site (no continuous sugar chainspanning the1 site could be observed) and 3 othersin the following shorter cleft part of the active center.In consequence, a pathway of the substrate in theactive site was traced, but the mechanism of thepropagation of the sugar chain leading to theprocessive action remains unclear. In this report, wepresent structures of Cel48F mutants in complex withlong hemithiocellooligosaccharides that reveal twodifferent pathways of sugar propagation around thecleavage site, leading to a new hypothesis of sugartranslocation throughout the active site of GH48catalytic modules. This information may be key inelucidating the still unknown pathway during theprocessive action of these cellulases.

Results and Discussion

In order to complete the substrate interactionscheme all along the active site of cellulase Cel48F ofC. cellulolyticum and to better understand its pro-cessive mechanism, we also decided to analyzeCel48F in complex with long thiocellooligosachhar-ide inhibitors instead of complexes with cellooligo-saccharides. The presence of the thioglycosidic bond

prevents the formation of an intramolecular hydro-gen bond between the GlcC2-OH and the followingGlcC6-OH, which contributes to the increase in thesolubility of the corresponding sugar compound in

500 Processive Action in Cel48F


3/12

aqueous media. Within the thiooligosaccharidesused for preparing the complexes, each secondglycosidic bond is replaced by a thioglycosidic

bond, resulting in an alternating thioglycosidic/glycosidic linkage pattern. Their improved solubilityallowed us to perform soaking experiments withhemithiocellooligosaccharide chains of 10 and 1214sugar moieties, therefore representing up to twotimes the length of cellohexaose, the longest naturalcellooligosaccharide sufficiently soluble under simi-lar aqueous conditions. Although the geometry ofthe thioglycosidic bond differs slightly from thatof the glycosidic one (Fig. 1), the physicochemicalresemblance of glycosidic chains to chains contain-ing alternating thioglycosidic/glycosidic linkage issufficient to allow the latter to occupy the same

positions in the active site of Cel48F (shown later inthis article). We soaked crystals of the mutants E55Qand E44Q of Cel48F with hemithiocellooligosacchar-ide inhibitors and solved their structures at 2.0 and1.9 resolution, respectively. All crystallographicdata are summarized inTable 1.

All attempts to crystallize Cel48F without a ligandfailed, but, finally, we succeeded in solving thestructure of native Cel48F in the ligand-free form at1.85- resolution by introducing a small seedingcrystal of Cel48F/cellobiose complex into a dropletthat contained the crystallization buffer plus D-glu-cose. The electron density of a single glucose mole-cule could not be detected in this structure, but theD-glucose seems to favor crystal growth of Cel48F inits observed conformation, which could not be ob-tained in dropletswithout anysugar supplement.14,16

The ligand-free structure obtained from the co-crystallization experiment with D-glucose revealedthe flexibility of some active-site side chains, whichappeared more static in the structures of ligandcomplexes and therefore provides insight into some

dynamic aspects that may be involved in its catalyticmechanism. Surprisingly, the hemithiocellooligosac-charides in the two complex structures follow twodifferent trajectories in the active site, which will becalled the upper and the lower pathway.

Description of theE55Q/hemithiocellooligosaccharide complex

To obtain information on the conformation of thesugar in subsite 1 during the enzymatic reaction,we solved the structure of the inactive mutant E55Qin complex with a hemithiocellooligosaccharide

composed of 10 glucose moieties. The electron den-sity of nine sugar subsites continuously spanningpositions +2 to 7 along the lower pathwayappeared in the unrefined Fourier difference elec-tron density map (Fig. 2a). In contrast to previouslypublished structures, subsite 1 was not empty, ascontinuous electron density between subsites 2and +1 was observed. The observed density in the1 position was less well defined than that in theother subsites but allowed us to fit a glucose moietyin a slightly disturbed 4C1chair conformation. Onlythe O3 hydroxy group of this sugar subunit isstabilized by H bonding, one with OD2 of Asp230and a second with the hydroxy group of Tyr323. No

sugar moiety occupies subsite +3 on the leavinggroup side asobserved in the published cellohex-aose complex;15 subsite +2 was the last one occu-pied in the hemithiocellooligosaccharide complex.

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

Table 1.Statistics of data collection and refinement

CelF/D-glucose E55Q/IG10 E44Q/IG12

Resolution () 401.85 302.0 501.90Cell:a,b,c () 61.53, 84.77,

122.1261.24, 84.76,

121.7461.48, 84.72,

121.81

Completeness (%) 98.2 (88.9) 95.2 (87.7) 99.7 (99.0)Multiplicity 3.8 (2.3) 3.9 (3.3) 3.7 (3.2)I/I 6.8 (1.9) 6.7 (3.0) 8.1 (1.7)Rsym(%) 9.1 (35.4) 7.8 (19.9) 7.7 (39.9)Number of water

molecules397 414 374

AverageB-factorprotein (2)

19.4 15.8 19.2

AverageB-factorinhibitor (2)

20.5 35.1

Rfree(%) 19.5 18.2 20.7rms bonds () 0.005 0.013 0.005rms angles () 1.3 1.3 1.2R(%) 16.5 14.0 18.2Protein Data

Bank ID1g9g 2qno 1g9j

Each data set was collected with a single crystal. The values inparentheses correspond to the highest shell of resolution.Rsym=hkli|Ihkl Ihkl,i|/hkli|Ihkl|, where i is the numberof reflectionshkl.

501Processive Action in Cel48F


4/12

The subsite positions of all sugar moieties wereequivalent to those previously published for thecellohexaose complex (rmsd of 0.67 for allmatching atoms), an observation that stresses thestereochemical resemblance between the inhibitorand 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 entrancebeyond subsite7, was not detectable in the electrondensity map. Although the overall orientation of the

sugar moieties is in good agreement with thepreviously published complexes, the hemithiocel-looligosaccharide chain is forced to an unusualorientation around the sugar moiety in subsite 1.Here, the orientation of its C6-OH group does notfollow the normally observed 180 alternating rota-tion along the cellooligosaccharide molecule but is

extremely twisted outside the ideal rotation. Itsrotation is only halfway,about 90, between sub-sites 1 and 2 and returns to a normal twist ofnearly 180 between subsites 1 and +1. The same

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 DataBank 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 areimportant in the interaction with the hemithiocellooligosaccharide are indicated. The electron density around the sugarchain is drawn at 1level.



5/12

orientation in subsite 1 has been observed pre-viously in a cellotetraose complex with mutation

E55Q, but in this case, the connectivity of the sugarchain was not continuous and subsite 1 wasoccupied by the nonreducing end of a cellotetraosemolecule intruding from the leaving group side.15

Description of theE44Q/hemithiocellooligosaccharide complex

On the basis of the location of sugars in the cata-lytic site, three residues were candidates for provid-ing the corresponding acid and base in the cleavagereaction, namely, Glu55, Glu44, and Asp230. Glu55was 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 glutamineto test the importance of the charge at this position.The Cel48FE44Q mutated enzyme was found toretain around 1% of the activity of the wild-type

protein on phosphoric acid-swollen cellulose, whileCel48FE55Q was completely inactive. The wild-

type Cel48F weakly degradesthe substituted solu-ble carboxymethyl cellulose.13 This activity wassuggested to be related to the initial endo mode ofaction. The E44Q enzyme retained this weakactivity, suggesting that the initial endo cuttingactivity was not greatly affected by the charge beingremoved. We cocrystallized this E44Q mutant inthe presence of a hemithiocellooligosaccharidecomposed of 12 sugar moieties and solved thestructure of the complex. Only 8 of the 12 sugarmoieties were visible in the electron density map(Fig. 2b). To our surprise, the sugar moieties didnot occupy the same subsites as observed in thecomplex of the inactive form E55Q. Taking into

account that the repetitive unit in the cellooligosac-charide chain is a cellobiose, the sugar positions ofthe chain in this complex were shifted for nearly halfa repetitive unit backward compared to the E55Q/

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



6/12

cellohexaose complex (Fig. 3b). In the E44Q/hemi-thiocellooligosaccharide complex, the sugar chainfollows the same pathway along the tunnel region

but does not descend into the depression thatfollows the tunnel part and contains the catalyticresidue Glu55 as observed in the case of the E55Qcomplex. Instead, it continues to follow a linearpathway located about 6 above the depressionand ending between subsites +1 and +2, where itforms a triple hydrogen bond with the mutatedresidue E44Q (Fig. 4). This newly observed upperpathway is only accessible after a rotation of the sidechain of Tyr403 for about 120, which blocks theaccess to the upper part of the depression in thecellohexaose complex. The rotation of Tyr403 hastwo effects: firstly, it opens this pathway, andsecondly, it offers an additional stabilization forthe sugar chain via a hydrophobic stacking interac-

tion. In total, two new stacking interactions invol-ving the sugar chain can be observed in the upperpathway: one additional with Tyr403 and anotherwith Phe180 located on the other side of the sugarchain replacing the stacking interaction provided byTrp154 at the equivalent subsite 2 of the lowerpathway. The spacing between Tyr403 and Phe180 isin the range of half a sugar residue, which followsthe pattern of stacking residues in the tunnel part.That kind of architecture is thought to reduce thesliding energy of the sugar chain along the pathwayand has been observed for processive cellulases orother sugar transportation channels.15,17,18 The twonew stacking interactions prolong, therefore, thegreasy slideof the tunnel up to the positions of thecellobiose leaving group site, located in the opencleft region. All interactions of the sugar chainprovided by these new subsites are summarized inFig. 4.

In addition to the rotation of Tyr403, the reorienta-tion of five other residues in the active site, namely,Glu55, Gln44, Met42, Trp411, and Trp417, could beobserved in the E44Q complex. In contrast topreviously described complexes, Glu55 is not incontact with a sugar residue. It is rotated about 120around1,replacingwithitsOE1andOE2atomsthepositions of twowater molecules present in the E55Q

complex. In this new orientation, it forms a hydro-gen-bond network with Asp230 OD1 and OD2atoms 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, whichhydrolyzes the -glycosidic bond with a retainingmechanism, or xylanases and cellulases from GHfamily 8, which cleaves the-glycosidic bond by aninverting mechanism.1922 In all three cases, even iftheir mechanisms are different, the associatedaspartic acid is thought to activate the protondonor by elevating its pKa and to stabilize thedistorted reaction intermediate in the 1 position.

The replacement of this residue in Cel48F byasparagine leads to an inactivation of the enzyme, asit has been observed in CGT from Bacillus circulansstrain 251, -amylase from Aspergillus oryzae, or

CGT from B. circulans strain 8 (here, mutation toalanine).2325

The replacement of Glu44 by a glutamine inducesa significant reorientation of the corresponding sidechain. A rotation of about 60 around 1swings theamide deeper in the active-site cleft and breaks its H

bond with His36. The resulting liberated space isoccupied by a chloride ion, located closely to theformer position of the Glu44 OE2 atom (at about1.5 away). The chloride ion is coordinated byHis36 and Arg609 and may compensate the negativecharge eliminated by the mutation.

Both aromatic residues Trp417 and Trp411 are indouble conformations and may be grouped in twopairs of related orientations. In the first pair,the orientations do not differ from those alreadyobserved in the other complex structures, whereasin the second pair, Trp411 and Trp417 are rotated

around 190 and 20, respectively. With this newlyobserved orientation, Trp411 intrudes into subsite+2 of the lower pathway and is in hydrophobiccontact with the reducing end of the sugar chain inthe upper pathway. One is tempted to suggest thatalternation between the two pairs of differentorientations of these aromatic side chains mayparticipate in the stabilization or liberation of thesugar moieties in the leaving group side.

The ligand-free Cel48F structure

In the ligand-free native structure, the orientationsof the active-site residues are closer to those

observed in the E55Q structure than to those seenin the E44Q model. All side chains in the active siteare pointing in the same direction as in the E55Qcomplex, with the exception of three residues,namely, Glu55, Trp417, and Lys224. The first,Glu55, adopts in the ligand-free native structure adouble conformation, occupying both orientationsobserved in the E44Q and E55Q complexes. Thesecond, Trp417, is located between its positionobserved in the E55Q complex and the location ofthe more conservative tryptophan position in theE44Q complex. The third, Lys224, is in double con-formation: one side-chain orientation is the same as

already reported for the E44Q and E55Q complexesand the other is a new orientation in which the sidechain intrudes deeply in the active-site tunnel. Thismakes in total three different side-chain orientationsfor Lys224, which stresses its role as a flexiblecontact in the sugar transportation process.

Implications for the processive mechanism

Cellobiose, composed of two glucose moieties, isthe repetitive unit of the cellulose chain and is themain product of the digestion process. Only one ofthe two sugar orientations in subsite1 is stabilizedand permits the cleavage reaction. An interesting

question about the processive action of cellobiohy-drolases is now how a cellobiose moiety passessmoothly across the cleavage site. If it slides by onlyone glucose unit across the active site, it occupies



7/12

Fig. 4. Hydrogen-bonding network of the hemithiocellooligosaccharide with the mutated enzyme E44Q. Distances are indicated in aninteractions with the sugar chain are drawn in thicker lines.


8/12

two positions, which have been named productiveand nonproductive binding. In Cel6A, which is aprocessive cellulase fromHumicola insolens, nonpro-ductive binding seems to be less stable, and itimposes reorientations of side chains and smallrigid-body movements around the cleavage site.18The present data appear to be consistent with analternative solution of the problem for GHs of family48, supporting a mechanism that evades nonpro-ductive binding around the cleavage site.

Our two new structures of active-site mutantswith hemithiocellooligosaccharides represent twokey steps in the processive transportation and

digestion process along the active site of Cel48F.The mutation E44Q has obviously interrupted thelow-energy sliding process along the tunnel part,

before the last two residues could integrate in theirfinal location in the leaving group site, whereas themutation E55Q has prevented the cleavage of theglycosidic bond and the release of the cellobiose unitfrom the leaving group site. The other publishedcomplexed structures so far additionally support thepresence of multiple low-energy positions in thetunnel part and show the compatibility of thehemithiocellooligosaccharide inhibitor withbindingsites occupied by the natural substrate.15,16 Finally,the ligand-free native structure shows the confor-

mations of all active-site residues without a boundsubstrate.The two observed pathways within the active site

have their specific features optimized for their

different functions. The upper, linear pathway offersstacking interactions with aromatic side chains,which build a low-energy sliding pathway up tothe leaving group site, optimized for sugar trans-port. The lower, descending pathway disturbs theinternal sugar chain stabilization around the clea-vage site and brings the scissile glycosidic bond inrange of the proton donor. In contrast to theobserved multiple binding sites in the tunnel part,only one sugar conformation was observed in eachleaving group subsite.15 This may suggest thatsubsites 1 and +1 (and maybe subsite +2 as well)are each specifically tailored to stabilize the transi-

tion state of only one of the two differently orientedglucose moieties. This specificity would disfavorlow-energy sliding, which might explain the pre-sence of the two observed pathways across thecleavage site. Recently, the comparison of Cel48Afrom C. thermocellum with the active site of Cel8Afrom the same organism proposed that Tyr323,Asp230, and Glu55 may be catalytic residuesresponsible for thedisruption of the sugar chain inthe GH48 cellulases.14 These residues do not interactwith the sugar chain during its sliding process inthe upper pathway but, in contrast, are important

binding partners in subsites1 and +1 in the lowerpathway.

During the transition from the upper to the lowerpathway, several conformational changes in theenzyme and the substrate chain must occur. Tyr403seems to be critical for the gating between both

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 arewater molecules; sugar moieties from the hemithiocellooligosaccharide are colored orange.



9/12

pathways. It swings inside the tunnel and seems topush the sugar chain in the subsites around thecleavage site. This residue is involved in both path-ways: as a stacking partner in the upper one and asan H-bond donor in the lower one. The centralstacking partner in the leaving group sites +1 and+2, Trp411, adopts two distinct conformations in theE44Q complex, like Tyr403, and may be a secondgatingresidue.Itsfunctioncouldbetolimitthesugarsliding in the upper pathway and to temporallystabilize the leaving group in the lower one.

The orientation of the last three sugar moieties atthe reducing end of the sugar chain has to be rotatedfor nearly 180 during their descent from the upperto the lower pathway to fit in subsites +2 to 1 inthe conformation observed in the E55Q complex.Because the fourth glucose moiety is alreadycorrectly oriented to fit in subsite 2, the rotation

will impose a twist of the glycosidic bond in subsite1. This subsite is the proposed transition statebinding site, tailored to stabilize a high-energy con-formation. In the E55Q complex, the sugar occupy-ing subsite 1 has a less well defined electrondensity and was fitted in a slightly disturbed 4C1conformation. We suppose that it has a potentialto be flexible and may adopt a more distortedgeometry during catalysis, as it has been observed inthe active site of the structurally related family 8glucoside hydrolases.26 The E44Q complex shedslight on the question why there is so much spaceabove the cleavage site: to permit the rotation of thelast three sugar residues.

Observations at the cleavage site

Obviously, the proton donor of the cleavagereaction Glu55 adopts two conformations duringcatalysis: If the substrate chain occupies the lowerpathway, it is H bonded to the scissile glycosidic

bond, or if the sugar chain follows the upper path-way, Glu55 is H bonded to Asp230 and Arg234(Fig. 5). In the ligand-free native structure, bothconformations of Glu55 could be observed. Asp230,which is normally H-bonded to Glu55 or to watermolecules, adopts a double conformation in the

E55Q/hemithiocellooligosaccharide complex. It iseither H-bonded to the OH-2of the sugar in subsite1 by conserving its usually observed conformationor bound to Arg421 after a rotation of 110 around1. The double conformation of Asp230 mayindicateconformational changes of this residue when subsite1 is occupied during catalysis. The identity of thecatalytic base in the cleavage reaction is still notobvious. Guimaraes et al. propose that Tyr323 isstabilizing a water molecule that may attackthe C4carbon of the sugar moiety in subsite 1.14 Thishypothesis is based on a structural comparison ofCel48Awith GH8 enzymes. Our data do not confirmsuch a hypothesis but do not reject it as well. We also

observe a water molecule that is well positioned4.2 above the C1 of the scissile bond and makesH bonds to the O2 of the sugar moiety in the 1subsite and to the O6 of the glucose moiety in sub-

site +1, but it is 4.9 away from the OH of Tyr323.Although the new structures of mutated Cel48F with

bound hemithiocellooligosaccharides shed somelight on the processive action of GH48 catalyticmodules, the details of its cleavage reaction stillremain unclear.

Conclusion

The complexes of the Cel48F mutants E44Q andE55Q with hemithiocellooligosaccharides furtherstrengthen our understanding of the enzyme'sprocessive action along the active site. The E55Qcomplex confirms the possibility of accommodatingthe hemithiocellooligosaccharides in the activesite despite their geometrical differences with thenatural substrates and gives a first view of theconformation of the sugar in subsite 1. The native

structure of Cel48F without any bound sugar inthe active site confirms the two orientations of thecatalytic proton donor Glu55 observed in the twomutant structures and adds information aboutside-chain orientations of residues that interactwith the substrate chain during catalysis withoutits presence. Finally, the E44Q complex seems to bea probable missing linkstructure that proposes analternative sugar propagation pathway from thetunnel into the active site. It reveals new subsitepositions, stabilized by stacking interactions andhydrogen bonds, which indicate a new intermedi-ate pathway during the sliding process. Thepresence of two pathways that are occupied during

either the processive action or the cleavage reactionto maintain low-energy sliding into the leavinggroup subsites across the cleavage site is a newlyobserved concept in processive enzymes. There-fore, the presented data give us new clues tounderstanding the structure/function relationshipof the complex architecture of GH family 48 activesites.

Materials and Methods

Cel48FE44Q construction

Mutation E55Q was obtained as reportedelsewhere.15

The same double PCR mutagenesis technique27 was usedto generate the point mutation E44Q in the cel48Fgene carried by the pETFp plasmid.28 The site-directedmutagenesis was performed using the complementarymutagenic oligonucleotides Fm44f (5-ACACTGATGGT-CCAAGCTCCTGACTACGGA-3) and Fm44r (5-TCC-GTAGTCAGGAGCTTGGACCATCAGTGT-3 ) tointroduce the CAA Gln codon in place of the GAA Glu44codon (mutagenic nucleotides in italic type). Expand HighFidelity PCR System (Roche) was used in the PCRexperiments. The mutated 5part of the cassette was syn-thesized using the forward primer F18 (5-CCCTATA-CATATGGCTTCAAGTCCTGCAAAC-3 ) a n d t h emutagenic Fm44f reverse primer. F18 carrying an NdeIsite (underlined) was located at the beginning of themature protein-encoding sequence (hybridizing region in

boldface type). The mutated 3 part of the cassette was



10/12

synthesized using the mutagenic Fm44f forward primerand the reverse primer F22 (5-GCCTTGAAGTTTCA-TCTCC-3) located downstream from the unique KpnI sitein the cel48Fsequence. These two purified PCR productswere mixed and amplified using F22 and F18 to generate

the whole mutated cassette. The amplified fragment waspurified (Qiaquick, Qiagen), digested by NdeI and KpnI,and then cloned between the NdeI and KpnI sites intopETFc to replace the wild-type sequence. The resultingplasmid carrying the mutation was called pETF44. Thesuccessful insertion of the fragment was verified bysequencing.

The mutated proteins E44Q and E55Q were producedfrom BL21(DE3)[pETF44] and BL21(DE3)[pETF55] cul-tures, respectively. They were purified, and their enzy-matic activity was assayed using phosphoric acid-swollencellulose as substrate following the procedures previouslyreported for Cel48F wild type.13

Crystallization and 3D structure solution

Crystals suitable for x-ray diffraction were grown usingvapor diffusion and microseeding techniques with micro-crystals of the Cel48F/IG4 complex, which has beendescribed previously.16 Crystals were grown under con-ditions as described for the Cel48F/IG4 except for E44Qwhere the crystallization buffer contained 20 mM CaCl2instead of MgCl2. Crystals of native Cel48F were grown inthe presence of 8 mM D-glucose, those of the mutant E44Qwere grown in the presence of 8 mM hemithiocellodode-caose, and those of the mutantE55Q were grown in thepresence of 2 mM cellobiose.29 The observed E55Q/hemithiocellooligosaccharide complex was obtained bysoaking the crystal with a hemithiocelloligosaccharide

composed of 10 sugar moieties, 14 h prior to mounting thecrystal in the capillary. This soak wasperformed by adding1 l of a solution of 20 mM hemithiocellooligosaccharide incrystallization buffer to the crystal-containing dropletwhose volume was approximately 4 l. The diffractiondata sets of all crystals were collected using glasscapillaries at 15 C with a MAR 345 image plate scanner.In the case of the E44Q/hemithiocellooligosaccharidecomplex, we used CuK radiation from the in-houseNonius FU 581 rotating anode generator equipped with agraphite monochromator, whereas the data on the nativeCel48F 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 setwas collected at the EMBL Outstation of the DeutschesElektronen-Synchrotron, Hamburg, at the X11 beamline atawavelengthof0.91.Thecrystalsweresufficientlystableto allow the collection of an entire data set from one singlecrystal. All crystals were isomorphous to the previouslyreported Cel48F/IG4 complex16 and belonged to the spacegroup P212121. The crystal data of all complexes aresummarized inTable 1.

The data were processed and reduced using theDENZO program30 and the CCP4 package.31 The native,ligand-free Cel48F structure and the Cel48FE44Q com-plex were solved using the Crystallography and NMRSystems (CNS) program.32 The E55Q complex wasinitially solved using the CNS program and was addi-tionally postrefined using the refmac program of the CCP4suite. The refinement procedure began with 20 cycles ofrigid-body refinement, using the protein chain and thecalcium-ion position of Cel48F/IG4complex as the searchmodel. The bulk-solvent correction33 and the weighting

factor WA were automatically recalculated after each stepof the refinement. The rigid-body refinement wasfollowed by a simulated annealing procedure at 3000 K.Manual fitting was performed for residues that were stillout of density and for the mutated residues. The watpick

procedure of the CNS package was then used to insertwater molecules at 3 difference density level. Watermolecules were kept if their corresponding electrondensity reached a 1 level in the (2FoFc) electron densitymap. 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 leavinggroup side (+1, +2) and followed by a positionalrefinement. Accurate bond parameters for the refinementof thiooligosaccharides were taken from the Engh andHuber parameters35 of the CNS package and thioglycosi-dic bond parameters from the structure of 2,3,4,6-tetra-O-acetyl-1-S-benzhydroximoyl--D-glucopyranose.36 Theprogram O was used for all model-building stages.37

After completion of the model, a restrained individualB-factor refinement was performed, followed by a finalcorrection of the positions of the water molecules andpositional refinement of all atoms. The E55Q complex wasfurther refined with the refmac program of the CCP4package.38 Five cycles of restrained maximum likelihoodrefinement followed by 1 cycle of ARP/wARP to com-plete the solvent model and, again, 20 cycles of maximumlikelihood refinement were used to recalculate thestructure.39 The hemithiocellooligosaccharide was refinedusing the same parameters for the thioglycosidic link as

before.The refined structures of the complexes were super-

imposed, with the least-squares function of the programO, by comparing the atoms of the protein backbone with

the Cel48F structure, which served as the reference, toelucidate differences in the positions of the substrate/inhibitor molecules and in the side-chain orientations.Figures2and3 were calculated using theprogram PyMOL.

Acknowledgements

We thank the staff of the BM30A beamline at theEuropean Synchrotron Radiation Facility and thestaff of the X11 beamline at the Deutsches Elektro-nen-Synchrotron for their assistance during the

collection of the E55Q complex and the ligand-freestructure, respectively. The work was supported bythe Centre National de la Recherche Scientifique andfinanced by the European Community (ContractBIO2-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 cancontribute to energy and environmental goals.Science,311, 506508.

2. Kim, S. & Dale, B. E. (2004). Global potential bio-ethanol production from wasted crops and cropresidues.Biomass Bioenergy, 26, 361375.

3. Tilman, D., Hill, J. & Lehman, C. (2006). Carbon-negative biofuels from low-input high-diversity grass-land biomass.Science, 314 , 15981600.



11/12

4. Sedlak, M. & Ho, N. W. (2004). Production of ethanolfrom cellulosic biomass hydrolysates using geneticallyengineered Saccharomyces yeast capable of cofer-menting 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 geneti-cally engineered Saccharomyces yeasts for effectivecofermentation 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-ethanolthe 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 ofClostridium celluloly-ticum.J. Biotechnol. 57, 314.

8. Schwarz, W. H. (2001). The cellulosome and cellulosedegradation by anaerobic bacteria. Appl. Microbiol.

Biotechnol. 56, 634

649.9. Henrissat, B. & Davies, G. (1997). Structural andsequence-based classification of glycoside hydrolases.Curr. Opin. Struct. Biol. 7, 637644.

10. Coutinho, P. M. & Henrissat, B. (1999). Carbohydrate-active enzymes: an integrated database approach. InRecent Advances in Carbohydrate Bioengineering (Gilbert,H.J.,Davies,G.,Henrissat,B.&Svensson,B.,eds),TheRoyal Society of Chemistry, Cambridge, UK.

11. Henrissat, B. & Bairoch, A. (1996). Updating thesequence-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 Piromycesis a family48 glycoside hydrolase.DNA Sequence, 13, 313320.13. Reverbel-Leroy, C., Pags, S., Belaich, A., Belaich, J. P.

& Tardiff, C. (1997). The processive endocellulaseCelF, a major component of the Clostridium celluloly-ticum cellulosome: purification and characterization ofthe 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 andcatalytic mechanism of cellobiohydrolase CelS, themajor enzymatic component of theClostridium thermo-cellumcellulosome.J. Mol. Biol. 320, 587596.

15. Parsiegla, G., Reverbel-Leroy, C., Tardif, C., Belaich,J. P., Driguez, H. & Haser, R. (2000). Crystal structuresof 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 crystalstructure of the processive endocellulase CelF ofClostri-dium cellulolyticumin complex with a thiooligosacchar-ide inhibitor at 2.0 resolution. EMBO J.17,55515562.

17. Meyer, J. E. W. & Schulz, G. E. (1997). Energy profile ofmaltooligosaccharide permeation through maltoporinas derived from the structure and from a statisticalanalysis 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 forligand binding and processivity in cellobiohydrolase

Cel6A fromHumicola insolens.Structure, 11, 855864.19. Kadziola, A., Abe, J., Svensson, B. & Haser, R. (1994).Crystal and molecular structure of barley alpha-amylase.J. Mol. Biol. 239, 104121.

20. Klein, C. & Schulz, G. E. (1991). Structure of cyclo-dextrin 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 structureof a cold-adapted family 8 xylanase at 1.3 resolu-tion. Structural adaptations to cold and investigationof the active site.J. Biol. Chem. 278, 75317539.

23. Klein, C., Hollender, J., Bender, H. & Schulz, G. E.(1992). Catalytic center of cyclodextrin glycosyltrans-ferase derived from X-ray structure analysis com-

bined with site-directed mutagenesis.Biochemistry, 31,87408746.

24. Nagashima, T., Tada, S., Kitamoto, K., Gomi, K.,Kumagai, C. & Toda, H. (1992). Site-directed muta-genesis of catalytic active-site residues of Taka-

amylase 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). Crystal-lographic studies of the interaction of cyclodextringlycosyltransferase from Bacillus circulans strain 251with 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 in-verting glycosidase in complex with substrate.J. Mol.Biol. 316, 10611069.

27. Higuchi, R., Krummel, B. & Saiki, R. K. (1988). Ageneral method of in vitro preparation and specificmutagenesis of DNA fragments: study of protein and

DNA interactions.Nucleic Acids Res. 16, 7351

7367.28. Reverbel-Leroy, C., Belaich, A., Bernadac, A., Gaudin,C., Belaich, J. P. & Tardiff, C. (1996). Molecular studyand overexpression of the Clostridium cellulolyticumcelF cellulase gene inEscherichia coli.Microbiology, 142,10131023.

29. Moreau, V. & Driguez, H. (1996). Enzymic synthesis ofhemithiocellodextrins.J. Chem. Soc., Perkin Trans. 1, 16, 525527.

30. Otwinowski, Z. & Minor, W. (1997). Processing ofX-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 crystal-

lography. 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 suitefor macromolecular structure determination. ActaCrystallogr., Sect. D: Biol. Crystallogr. 54, 905921.

33. Jiang, J. S. & Brunger, A. T. (1994). Protein hydrationobserved by x-ray diffraction: solvation properties ofpenecillopepsion and neuraminidase crystal struc-tures.J. Mol. Biol. 243, 100115.

34. Brnger, A. T. (1992). The free R value: a novelstatistical quantity for assessing the accuracy of crystalstructures.Nature, 355, 472474.

35. Engh, R. A. & Huber, R. (1991). Accurate bond and

angle parameters for X-ray protein structure refine-ment.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-



12/12

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 proteinmodels in electron density maps and the location of

errors in these models.Acta Crystallogr., Sect. A: Found.Crystallogr. 47, 110119.

38. Murshudov, G. N., Vagin, A. A. & Dodson, E. J. (1997).Refinement of macromolecular structures by themaximum-likelihood method. Acta Crystallogr., Sect.D: Biol. Crystallogr. 53, 240255.

39. Perrakis, A., Morris, R. & Lamzin, V. S. (1999). Auto-

mated protein model building combined with itera-tive structure refinement.Nat. Struct. Biol. 6, 458463.


structre of mutant cellulose

Documents