Opinion

Towards new enzymes for biofuels:lessons from chitinase researchVincent G.H. Eijsink1, Gustav Vaaje-Kolstad1, Kjell M. Varum2 and Svein J. Horn1

1 Department of Chemistry, Biotechnology and Food Science, Norwegian University of Life Sciences, P.O. Box 5003, N-1432 As,

Norway2 Norwegian Biopolymer Laboratory (NOBIPOL), Department of Biotechnology, Norwegian University of Science and Technology,

7491 Trondheim, Norway

Enzymatic conversion of structural polysaccharides inplant biomass is a key issue in the development ofsecond generation (‘lignocellulosic’) bioethanol. Theefficiency of this process depends in part on the abilityof enzymes to disrupt crystalline polysaccharides, thusgaining access to single polymer chains. Recently, newinsights into how enzymes accomplish this have beenobtained from studies on enzymatic conversion of chitin.First, chitinolytic microorganisms were shown to pro-duce non-hydrolytic accessory proteins that increaseenzyme efficiency. Second, it was shown that a proces-sive mechanism, which is generally considered favorablebecause it improves substrate accessibility, might in factslow down enzymes. These findings suggest new focalpoints for the development of enzyme technology fordepolymerizing recalcitrant polysaccharide biomass.Improving substrate accessibility should be a key issuebecause this might reduce the need for using processiveenzymes, which are intrinsically slow and abundantlypresent in current commercial enzyme preparations forbiomass conversion. Furthermore, carefully selectedsubstrate-disrupting accessory proteins or domainsmight provide novel tools to improve substrate acces-sibility and thus contribute to more efficient enzymaticprocesses.

IntroductionMany living organisms use networks of fibrous and crystal-line polysaccharides tomaintain structural integrity. Enzy-matic conversion of the most recalcitrant of thesepolysaccharides is of great biological and economic import-ance and affects processes varying from the interplay be-tween, for example, plants or insects and their pathogens tothe production of second generation (‘lignocellulosic’)bioethanol. In plants, the major structural polysaccharideis cellulose [b(1->4)linked glucose], whereas non-plantssuch as insects, crustaceans and fungi employ chitin [b(1->4)linkedN-acetylglucosamine], which occurs in two majorforms, a-chitin and b-chitin (Box 1). In nature, degradationof cellulosic or chitinous biomass is achieved by mixtures ofhydrolytic exo- and endo-acting enzymes that act in asynergistic manner [1,2]. There is currently great interestin these enzymatic machineries because they have thepotential to convert lignocellulosic biomass to fermentablesugars; however, costs for such enzymes are currently a

Corresponding author: Eijsink, V.G.H. (vincent.eijsink@umb.no).

228 0167-7799/$ – see front matter � 2008 Elsevier

major limiting factor for the commercial development oflignocellulosic bioethanol [1,3–5].

Enzymes acting on crystalline polysaccharides face sev-eral challenges. They need to be able to associate with theinsoluble substrate, disrupt the polymer packing and,importantly, guide a single polymer chain into the catalyticcenter (Figures 1 and 2). As early as 1950, Reese and co-workers suggested that hydrolysis of cellulose wouldrequire an (unknown) non-hydrolytic component thatwas able to disrupt polymer packing in the substrate,thereby increasing its accessibility for the hydrolyticenzyme [6]. However, even today there is no clear-cutexperimental evidence for the existence of such a com-ponent in cellulolytic enzyme systems. Alternatively, pro-ductive interactions with the substrate might also bepromoted by an intrinsic property of the enzymes them-selves, namely the ability to employ a processive (‘multipleattack’) mechanism [7]. Processive cellulases or chitinaseshave been shown to have long and deep active site clefts, oreven tunnels (Figure 1) [8]. Single polymer chains arethreaded through these clefts or tunnels while disacchar-ides are being cleaved off at the catalytic center (Box 2,Figures 1 and 2). This mechanism is considered to bebeneficial for enzyme efficiency because the enzymeremains closely associated with the detached single poly-mer chain in between hydrolytic steps. Furthermore, thedetached chain is prevented from re-associating with thecrystalline material, which would make it less accessible.

Recent studies on the enzymatic conversion of chitinhave provided important new insights into both mechan-isms that would allow improved substrate accessibility.First, chitin-degrading organisms were found to produce anon-hydrolytic accessory protein, showing for the first timethat accessory proteins such as those hypothesized byReese et al. [6] do indeed exist, at least in chitinolyticmachineries. Second, the unique experimental possibilitiesoffered by using the soluble polymeric chitin-derivativechitosan (Box 1) provided new insights in the benefitsand, importantly, disadvantages of processivity for enzymeefficiency. Here, we discuss these recent findings and arguethat they point toward new directions for the future de-velopment of enzymes for biomass conversion.

Accessory proteinsIn 2005, Vaaje-Kolstad et al. [9,10] showed that CBP21, anon-hydrolytic 19 kDa chitin-binding protein [11] produced

Ltd. All rights reserved. doi:10.1016/j.tibtech.2008.02.004 Available online 25 March 2008

Box 1. Cellulose and chitin

Cellulose and chitin are the most abundant biopolymers in the

terrestrial and marine environments, respectively. Cellulose is a

linear b-(1–4)-linked polymer of D-gluco-pyranose units in the 4C1

conformation (Figure Ia) and chitin is a cellulose derivative where

the 2-hydroxy group has been substituted with an acetamido group

(Figure Ib ). Different polymorphic forms of cellulose have been

described where the glucan chains are believed to be packed in a

parallel fashion in cellulose I (i.e. with the reducing ends pointing in

the same direction) and in anti-parallel fashion in cellulose II. It is

still not understood how it is possible to convert parallel cellulose

chains into an anti-parallel packing in processes that are essentially

solid-state transformations. Interestingly, chitin also exists in two

main crystalline forms, where a-chitin is similar to cellulose II and b-

chitin has a parallel chain arrangement comparable to that of

cellulose I. From crystal structure data for cellulose and chitin and

the unit cell dimensions it is possible to calculate packing densities,

which are estimated at 1.62 g/cm3 for cellulose and a significantly

lower 1.46 g/cm3 for chitin. Little is known about the effect of this

difference on the enzymatic degradation of insoluble cellulose or

chitin. Other substrate parameters, such as the degree of crystal-

linity and the degree of heterogeneity of the samples, might be

equally important.

The insolubility of cellulose and chitin in water is a major obstacle

for in-depth studies of enzymatic degradation, especially when

complex issues such as enzyme processivity are addressed (see Box

2). Therefore, several chemically modified soluble cellulose and

chitin forms have been developed. Soluble cellulose derivatives are

obtained by modification of free hydroxyl groups. The most

commonly used soluble cellulose form in research is carboxy-

methylcellulose, in which on average �0.5–0.9 hydroxyl groups per

sugar (primarily O-2) have been modified. These relatively large

extra groups pose steric limitations to the enzyme–substrate

interaction that in turn might affect processivity. Chitosans, a family

of well-characterized water-soluble chitin derivatives [35], are

derived from chitins by removing varying fractions of the N-acetyl

groups, which results in less bulky amino groups on the polymer.

These polymers are very valuable substrates for in-depth studies of

processivity in family 18 chitinases (see Box 2) [2,36,44,45].

Figure I. Chemical structures of (a) cellulose and (b) chitin.

Opinion Trends in Biotechnology Vol.26 No.5

by one of the most efficient bacterial chitin-degraders,Serratia marcescens, strongly increased the efficiency ofchitinases in degrading certain forms of crystalline chitin(Figure 2). This was the first clear example of an accessoryprotein produced to improve the efficiency of enzymatichydrolysis of an insoluble recalcitrant polysaccharide sub-strate. Interestingly, CBP21-like proteins can be found in

many chitinolytic microorganisms, as well as in insectviruses that need to perforate the chitinous heteropolymericperitrophic matrix of the insect gut before infection. It couldbe shown that insect virus infectivity was dramaticallyincreasedwhen insect larvae were fed a CBP21-like protein([12] and references therein).

The functionality of CBP21 was tested in combinationwith several individual pure chitinases [10]. The magni-tude of the synergistic effect of CBP21 depended on theenzyme used because different enzymes exhibit differentintrinsic abilities to degrade the substrate. CBP21 bindsonly to b-chitin, and hence can only contribute to thedegradation of his substrate. There are CBP21-likeproteins that are known to bind a-chitin, but their poten-tial effect on chitin degradation has not yet been explored[13]. Clues to the molecular function of CBP21 come fromits crystal structure and site-directed mutagenesis stu-dies [9,10], which showed that the binding surface ofCBP21 contains conserved polar residues that are crucialfor its synergistic effect (Figure 2c). Interestingly, singlepoint mutations that abolished the synergistic effect ofthe protein had only moderate effects on the chitin bind-ing affinity [10]. This demonstrates that the action ofCBP21-like proteins is driven by other factors in additionto mere binding activity. An interesting possibility is thatthe assembly of polar amino acid side chains on thebinding surface is needed to form a specific set of multiplehydrogen bonds with the substrate, thus disrupting thehydrogen-binding network between individual polymerchains. This could also explain why CBP-like proteinsonly bind to certain types of substrates, which havespecific packing arrangements and hydrogen-bondingnetworks (Box 1).

The example of CBP21 clearly shows that in natureaccessory proteins have evolved to improve substrateaccessibility and to act synergistically with hydrolyticenzymes. So far, a corresponding example is lacking fromcellulase research, but there are a few indications thatproteins exist that are capable of disrupting cellulose andcellulose-containing heteropolymeric complexes. Forexample, plants produce proteins, so-called expansins,which contribute to a loosening of the cell wall necessaryfor plant growth [14]. Expansins have also been detectedin nematodes that are able to degrade plant cell walls [15].It is known that expansinsmechanicallyweaken plant cellwalls and pure cellulose paper [16], and the use of expan-sins to improve cellulase efficiency has been suggested[17]. However, application of expansins has been ham-pered by the fact that these proteins are difficult to pro-duce in vitro [17,18]. Interestingly, two very recent studiesshowed that expansin-like grass pollen allergens, whichare easy to obtain [18], and a non-characterized proteinfrom corn stover [19] could be used to increase cellulaseefficiency.

Another potential CBP21 analogue that could act oncellulose is swollenin from the cellulose-degrading fungusTrichoderma reesei. Swollenin contains both a cellulose-binding domain and an expansin-like domain [20] and,furthermore, its expression is co-regulated with cellulaseexpression [21]. Swollenin showed clear disruptive effectson cellulose, but its application as an accessory protein to

229

Figure 1. Chitinases, cellulases and their substrate-binding clefts. (a) Shown is a surface representation of the processive chitinase B (ChiB) from Serratia marcescens

(Protein Data Bank (PDB) code 1E15, [48]). Catalytic domains are blue, additional substrate-binding domains are gray. Surface-exposed aromatic amino acids in the

substrate-binding groove or surface are colored in red. The approximate position of the catalytic acid residue is indicated by a yellow asterisk. (b) Different view of ChiB

highlighting the deep, almost ‘tunnel-like’ substrate-binding cleft. (c) Details of the molecular interactions between ChiB and a chitin pentamer (purple) [46]. The sugars in

this pentameric substrate are numbered by the subsite to which they bind (Box 2). The catalytic amino acid (Glu144) is shown in yellow. The orientation of the enzyme is

approximately as in panel (a). (d) Shown is the structure of the processive cellulase Cel9A from T. Fusca (PDB-code 1JS4, [49]). Colors are as in (a) and (b) and the catalytic

amino acid is shown in yellow.

Opinion Trends in Biotechnology Vol.26 No.5

cellulases has so far not been described. Available infor-mation for fungal genomes suggests that expansin-likedomains occur regularly in fungal proteins that act onplant polysaccharides, suggesting that these domainsare indeed used for substrate disruption, either as func-tional parts of swollenin-like proteins or as accessorydomains in cellulases and related enzymes.

Based on sequence comparisons, carbohydrate-bindingdomains and proteins have been classified in�50 differentfamilies of carbohydrate-binding modules (CBMs) [22,23].CBP21 belongs to CBM family 33. Most CBMs do not existas free proteins but rather as substrate-binding domains inmulti-domain hydrolytic enzymes, such as chitinases andcellulases (Figures 1 and 2). The roles of these domains incellulases and chitinases have received much attention,and it is well established that they contribute to enzymeefficiency by increasing affinity for the substrate [23–28].Importantly, it has been shown that some cellulose-bind-ing domains can change the structure of the crystallinesubstrate [29–34] and that this ‘destructuring’ effect alonemight have a positive effect on cellulase efficiency [29–32].Thus, cellulose-binding domains in cellulases might havesimilar effects on cellulolytic activity as CBP21-likeproteins have on chitinolytic activity.

230

ProcessivityOnce a hydrolytic enzyme is bound to an individual poly-mer chain, it seems beneficial for it to act processively, thatis, the enzyme remains associated with the chain in be-tween hydrolytic steps, thus preventing its substrate fromre-associating with the insoluble material (Figure 2a).However, it is not well established how processivity con-tributes to overall enzyme efficiency. Processivity cannotbe easily measured because the insoluble substrate is notamenable to straightforward biochemical analysis. More-over, methods that are able to determine processivity havelimitations (Box 2).

In the case of chitin, it is possible to study processivityusing soluble chitin variants, so-called chitosans [35],which resemble chitin and therefore are able to interactwith chitinases in a natural manner (see Boxes 1 and 2).Using chitosans, we could recently show that the proces-sivity of the chitobiohydrolase chitinase B (ChiB) fromSerratia marcescens was almost completely abolished byone single point mutation of an aromatic residue close tothe catalytic center (Figure 1c, Figure 3) [36]. As expected,this non-processive mutant was less effective in degradinginsoluble chitin. However, the degradation rates forsoluble and thus more accessible substrates, such as a

Figure 2. Schematic illustration of chitin degradation. (a) Chitin degradation by a processive chitinase from S. marcescens [50]. The substrate shown is b-chitin, in which the

reducing ends of the parallel polysaccharides point to the right. For clarity, the sugars are oversized compared to the enzymes. The chitinase consists of a catalytic (blue)

and an accessory chitin-binding (gray) domain. Accessibility of the substrate is low, resulting in a low extend of chitin degradation. (b) Chitin degradation in the presence of

CBP21 (red). CBP interferes with chitin crystal packing, thus increasing substrate accessibility and enzyme efficiency and hence chitin degradation. These two panels also

illustrate enzyme processivity, that is, the enzyme tends to remain associated with the polymer chain while cleaving off disaccharides. (c) Surface representation of CBP21.

The side chains of conserved residues in the (rather flat) binding surface are colored yellow [9]. (d) The effect of CBP21 on chitinase efficiency is illustrated (adapted from

[10]). In the presence of CBP21, complete conversion of the substrate is achieved �6 times faster than in the absence of CBP21.

Opinion Trends in Biotechnology Vol.26 No.5

chitin hexamer oligosaccharide and polymeric chitosan,were increased by fourfold and, spectacularly, 29-fold,respectively. This demonstrates that the beneficial effectof processivity on substrate accessibility, in the case ofactivity towards chitin, comes at a cost of enzyme speedwhenmore accessible substrates such as chitosan are used.In other words, the ‘stickyness’ that keeps processiveenzymes tightly associated with their substrate in betweenconsecutive hydrolytic steps is unfavorable for solublesubstrates whose efficient degradation does not dependon keeping the single polymer chains detached from thecrystalline matrix. From a technological point of view, thisleads to the important conclusion that processive enzymescan and should be avoided if the crystalline structure of thesubstrate could be pre-disrupted by other means so thatsubstrate accessibility is no longer the rate-limiting step.

The substrate-binding grooves of processive cellulasesand chitinases are lined with aromatic residues,suggesting that these enzymes employ similar strategiesfor substrate-binding and processivity (Figure 1). Detailedcrystallographic studies of cellulase enzyme–substratecomplexes support the hypothesis that these aromaticresidues play important roles in the enzyme–substrateassociation during the processive action [37]. It has beenshown in a few cases that mutation of aromatic residues inthe substrate-binding clefts of cellulases led to reducedprocessivity [38–41]. The observed effects were generallyless drastic than those reported for chitinases, but thiscould also reflect technical limitations in measuring pro-cessivity (Box 2). Most importantly, a careful re-evaluationof the available cellulase literature [39–41] confirms anegative correlation between processivity and enzyme effi-

231

Box 2. Processivity and how it can be measured

Figure I shows initial endo-binding of a chitin polymer to a processive

glycoside hydrolase, which contains six subsites numbered �3 to +3

and an additional carbohydrate-binding domain (‘CBM’) (pictured is

the arrangement in S. marcescens ChiB [46]). The reducing end sugar

is colored gray and the direction of processivity is indicated by an

arrow. A correctly positioned N-acetyl group (symbolized by small

black balls on sticks) in the –1 subsite is essential for catalysis to occur

(catalysis is ‘substrate-assisted’) [47]. Note that consecutive sugars in

cellulose or chitin are rotated by 1808 (see Box 1), with the

consequence that only every second sugar will have a productive

configuration when the polymer is threaded through the active site of

an enzyme. In the scheme shown, the first hydrolysis (indicated by (i))

will produce a pentasaccharide, whereas all further cleavages (ii–iv)

that result from the same initial enzyme–substrate association will

produce disaccharides. The saccharide pentamer will eventually be

degraded to a disaccharide and a trisaccharide (some enzymes might

convert the trisaccharide to a monosaccharide and a disaccharide).

So, although initial cleavages might yield odd-numbered saccharide

oligomers that eventually yield trisaccharides, subsequent processive

cleavages yield disaccharides only. Thus, the ratio between the

amount of disaccharides and the amount of trisaccharides (or

monomers) will increase with the processivity of the enzyme.

However, it should be noted that this ratio is affected by the way

the enzyme binds to and degrades intermediate products. For

example, a hexameric intermediate product might be converted to

two trimers or to three dimers, depending on processivity-indepen-

dent binding preferences in the respective subsites. Another method

for measuring processivity is a comparison of the amount of soluble

and non-soluble reducing ends produced during a reaction. Proces-

sive enzymes yield more soluble reducing ends because they cleave

many times within one polymer chain (as opposed to cleaving once in

several polymer chains; this latter would increase the number of

reducing ends in the insoluble fraction). A problem with this method

is that non-processive enzymes that preferably bind in an exo-mode

(i.e. at chain ends) will yield similar results to processive enzymes.

Chitosan presents a unique opportunity to observe processivity in

chitinases that belong to the family 18 of glycoside hydrolases [22]

because these enzymes have an absolute requirement for an

acetylated sugar in the –1 subsite in order for substrate binding to

be productive [47]. Thus, complexes formed during processive action

on chitosan (partially deacetylated chitin) might be non-productive

[2,44] because the sugar bound in the –1 subsite might lack the acetyl

group. For processive enzymes, this translates into a product profile

that is dominated by longer even-numbered saccharide oligomers

during the initial phases of the degradation reaction (see also

Figure 3a in main text). When a non-productive complex emerges,

the enzyme does not dissociate from the substrate (which would have

yielded a more random distribution of odd- and even-numbered

products: see Figure 3b,c in main text). Instead, processive ‘move-

ment’ continues until the next productive complex emerges after the

enzyme has moved along an additional 2, 4 or 6 (or more) sugars. So,

in a processive enzyme, all products resulting from the same initial

enzyme–chitosan binding event will be even-numbered, except for

the very first product (as explained above). In the case of chitin, all the

even-numbered products would be disaccharides. See Ref. [45] for

more details.

Figure I. Processive degradation of chitin. The numbers (i-iv) indicate the initial

four consecutive cuts in the polymer.

Opinion Trends in Biotechnology Vol.26 No.5

ciency toward soluble substrates. Recent evidence comesfrom a study of the processive cellulase Cel9A from Ther-mobifida fusca (Figure 1d) [41]. Li et al. demonstrated thatmutations of aromatic residues close to the catalytic centerof Cel9A reduced processivity and activity towards crystal-line cellulose, while at the same time improving activitytowards (soluble and more accessible) carboxymethyl cel-lulose [41].

Implications for bioethanol-related enzyme researchEnzymatic hydrolysis of cellulose to fermentable sugars isa rather difficult task and currently requires 40–100 timesmore enzyme per gallon of ethanol than hydrolysis of cornstarch [1]. Consequently, and despite considerableresearch efforts in the past two decades, the costs ofenzymes remain a major factor that limits the profitabilityof producing cellulose-based bioethanol. Technologies thatproduce more efficient enzymes will have beneficial effectsfor the biofuel industry, not only because they lowerenzyme cost per gallon of ethanol but also because theymight result in faster processing times, which, in turn,reduce capital investment per gallon of productioncapacity. Major efforts are underway to discover or toengineer novel enzymes with the aim to develop cheaperand more-effective enzyme mixtures that can be applied inthe bioethanol industry [1,42].

232

Although there is currently no direct experimentalevidence that would allow us to quantitatively extrapolatethe results obtained from chitinase research to cellulases,the significant results from chitinases neverthelessindicate that focusing on the roles of accessory proteinsor protein domains and on the role of processivity is apromising approach for the future development of cellulo-lytic enzyme systems. For example, as shown in Figure 2d,full conversion of a chitin substrate was obtained about sixtimes faster when the accessory protein CBP21 was added(see [10] for more examples). The non-processive variant ofChiB acted almost 30 times faster than the processive wild-type on soluble polymeric substrate, which is a consider-able rate enhancement. Similar effects have not yet beendescribed for cellulolytic enzymemachineries but might bevalidated once the roles of potential accessory proteins ordomains and of processivity on their function have beenelucidated in more detail.

Several common commercial enzyme preparations forconversion of cellulose, such as CelluclastTM from Novo-zymes, contain a mix of cellulases produced by the fungusTrichoderma reesei, and the overall enzyme activity isdominated by the processive cellobiohydrolases CBHI(Cel7A) and CBHII (Cel6A). It is well established thatthese enzymes are crucial for the efficiency of the enzymepreparations towards cellulose [1,7]. Clearly, as explained

Figure 3. Degradation of water-soluble chitosan with various chitinases. Shown are size exclusion chromatograms obtained for the degradation products after �14% of the

glycosidic bonds in the highly polymeric substrate had been cleaved. Oligosaccharides are labeled according to the number of sugar monomers (note that products

containing five or less monomers might show multiple peaks because species with different sequences are partly resolved). (a) Wild-type ChiB acts processively on

chitosan. As explained in detail in Box 2, this results in the retention of a large polymer peak and a degradation product profile that is dominated by even-numbered

products. (b) The W97A mutant variant of ChiB has lost most of its processivity with the consequence that no polymer peak remained. The product profile covered all

possible fragment lengths, with a diminished dominance of longer even-numbered products (e.g. hexamers and octamers). In addition, degradation of chitosan was almost

30 times faster than observed with the wild-type enzyme [36]. (c) For comparsion, the degradation profile of ChiC, a natural non-processive endo-acting chitinase from

Serratia marcescens, is shown. In the W97A mutant, the processive endo-acting wild-type ChiB (a) [45] becomes less processive which results in a product profile (b) that

looks more like that of a truly non-processive enzyme (c). Abbreviation: wt, wild-type. Panels (a) and (b) adapted from Ref. [36].

Opinion Trends in Biotechnology Vol.26 No.5

above, the processive character of the enzymes is beneficialfor interacting with insoluble substrates. However, theresults obtained from chitinases show that the processiveenzyme character might also explain why enzymatic cel-lulose conversion, despite years of research, is still a slowprocess. Other, non-processive cellulases might prove to bemore efficient if they are offered a sufficiently disrupted(i.e. accessible) substrate or if they can act in concert withan optimized substrate-disrupting accessory protein ordomain.

The remarkable results obtained with CBP21 couldstimulate the technological exploration of the promisingexpansins and swollenin-like proteins, as well as furtherstudies that are aimed at optimizing the contribution ofcellulose-binding domains with disruptive effects. Suchaccessory proteins or domains are expected to generallyaccelerate hydrolysis reactions and might, in addition,reduce the need to use processive (and hence slow)enzymes. Useful accessory proteins could be added tocellulase mixtures or the microorganisms producing thesemixtures could be genetically modified to also producethese proteins. Both strategies would lead to the gener-ation of novel enzyme mixtures that combine optimalcatalytic and accessory domains.

Enzymatic hydrolysis of lignocellulosic biomass requiresa chemical and/or physical pre-treatment step to disruptplant fiber structure and to increase substrate accessibility

[3,5]. For example, cellulaseswill release hardly any solublesugar fromawoodchip substrate,whereasthemajorityof itscellulose can be solubilized if the polymer packing in thewoodstructure isfirstdisruptedbyharshtechniques suchassteamexplosion [43]. The type of pre-treatment, its outcomeand the choice of subsequent enzyme treatments are inter-connected and depend on the type of biomass used. It isimportant to takeall theseaspects intoaccountwhentestingtechnological applications of potential accessory proteins ordomains. CBP21-like proteins are specific for certain typesof chitin, and cellulose-binding domains are specific forcertain types and structures of cellulose crystals[23,26,27]. An additional factor that further complicateshydrolysis of lignocellulosic biomass is its heterogeneouspolysaccharide composition, consisting of cellulose, hemi-cellulose and lignin co-polymers. Naturally occurring acces-sory proteins or domains might have evolved to act on suchcomplex heteropolymers. Taking these factors into account,it is to be expected that accessory proteins or domains thatwork well towards a relatively homogeneous polymericsubstrate under laboratory conditions might not work onnatural heteropolymeric substrates or pretreated biomass,or vice versa.

ConclusionAlthough it remains to be seen to what extent the recentfindings of the roles of processivity and accessory proteins

233

Opinion Trends in Biotechnology Vol.26 No.5

in chitin degradation can be extrapolated to cellulolyticenzymes, it is undisputable that they signal an importantdirection for future research on enzymes for biofuels. Theroles of cellulose-binding domains, which have alreadybeen partly unraveled, need to be explored further, andefforts to technologically exploit these domains must beincreased. A more intense search for accessory proteins inthe genomes of cellulose-degrading or -modifying organ-isms might yield new useful proteins that could be pro-duced on a large scale, such as the expansin-like proteins[18]. It is possible that potentially important, hithertounknown accessory proteinsmight be discovered by critical(re-)evaluation of gene expression data, which are avail-able for, for example, Trichoderma reesei [21]. Finally, theremarkable properties of the non-processive mutant ofChiB show that the role of processivity in cellulose degra-dation needs more attention. The question remainswhether there are other ways to obtain the benefits ofprocessivity, for example by using pre-treatment stepsand/or CBP21-like accessory proteins. The data availablefor chitinases clearly demonstrate that a better combinedunderstanding of these issues is important for the furtherimprovement of enzyme technology for biomass conver-sion, which will subsequently stimulate commercial pro-duction of lignocellulosic ethanol.

AcknowledgementsWork in the authors’ laboratories was sponsored by the NorwegianResearch Council, grants 140497, 164653 and 171991. We are grateful toprevious and current co-workers for their input. We thank Frode Mo forinformation on packing densities and Pawel Sikorski for critical readingof the manuscript.

References1 Merino, S.T. and Cherry, J. (2007) Progress and challenges in enzyme

development for biomass utilization. Adv. Biochem. Eng. Biotechnol.108, 95–120

2 Horn, S.J. et al. (2006) Endo/exo mechanism and processivity of family18 chitinases produced by Serratia marcescens. FEBS J. 273, 491–503

3 Hahn-Hagerdal, B. et al. (2006) Bio-ethanol – the fuel of tomorrow fromthe residues of today. Trends Biotechnol. 24, 549–556

4 Himmel, M.E. et al. (2007) Biomass recalcitrance: engineering plantsand enzymes for biofuels production. Science 315, 804–807

5 Wyman, C.E. (2007) What is (and is not) vital to advancing cellulosicethanol. Trends Biotechnol. 25, 153–157

6 Reese, E.T. et al. (1950) The biological degradation of soluble cellulosederivatives and its relationship to the mechanism of cellulosehydrolysis. J. Bacteriol. 59, 485–497

7 Teeri, T.T. (1997) Crystalline cellulose degradation: new insight intothe function of cellobiohydrolases. Trends Biotechnol. 15, 160–167

8 Rouvinen, J. et al. (1990) Three-dimensional structure ofcellobiohydrolase II from Trichoderma reesei. Science 249, 380–386

9 Vaaje-Kolstad, G. et al. (2005) Crystal structure and binding propertiesof the Serratia marcescens chitin-binding protein CBP21. J. Biol.Chem. 280, 11313–11319

10 Vaaje-Kolstad, G. et al. (2005) The non-catalytic chitin-binding proteinCBP21 from Serratia marcescens is essential for chitin degradation.J. Biol. Chem. 280, 28492–28497

11 Suzuki, K. et al. (1998) Chitin binding protein (CBP21) in the culturesupernatant of Serratia marcescens 2170. Biosci. Biotechnol. Biochem.62, 128–135

12 Mitsuhashi, W. et al. (2007) Spindles of an entomopoxvirus facilitate itsinfection of the host insect by disrupting the peritrophic membrane.J. Virol. 81, 4235–4243

13 Schnellmann, J. et al. (1994) The novel lectin-like protein CHB1 isencoded by a chitin-inducible Streptomyces olivaceoviridis gene andbinds specifically to crystalline a-chitin of fungi and other organisms.Mol. Microbiol. 13, 807–819

234

14 Cosgrove, D.J. (2000) Loosening of plant cell walls by expansins.Nature 407, 321–326

15 Qin, L. et al. (2004) Plant degradation: a nematode expansin acting onplants. Nature 427, 30

16 McQueen-Mason, S. and Cosgrove, D.J. (1994) Disruption of hydrogenbonding between plant cell wall polymers by proteins that induce cellwall extension. Proc. Natl. Acad. Sci. U. S. A. 91, 6574–6578

17 Cosgrove, D.J. (2001) The Penn State Research Foundation.Enhancement of accessibility of cellulose by expansins, UnitedStates Patent 20016326470

18 Cosgrove, D.J. and Tanada, T. (2007) The Penn State ResearchFoundation. Use of GR2 proteins to modify cellulosic materials andto enhance enzymatic and chemical modification of cellulose, UnitedStates Patent 20070166805

19 Han, Y. and Chen, H. (2007) Synergism between corn stover proteinand cellulase. Enzyme Microb. Technol. 41, 638–645

20 Saloheimo, M. et al. (2002) Swollenin, a Trichoderma reesei proteinwith sequence similarity to the plant expansins, exhibits disruptionactivity on cellulosic materials. Eur. J. Biochem. 269, 4202–4211

21 Foreman, P.K. et al. (2003) Transcriptional regulation of biomass-degrading enzymes in the filamentous fungus Trichoderma reesei.J. Biol. Chem. 278, 31988–31997

22 Coutinho, P.M. andHenrissat, B. (1999) Carbohydrate-active enzymes:an integrated database approach. In Recent Advances in CarbohydrateBioengineering (Gilbert, H.J. et al., eds), pp. 3–12, The Royal Society ofChemistry

23 Boraston, A.B. et al. (2004) Carbohydrate-bindingmodules: fine-tuningpolysaccharide recognition. Biochem. J. 382, 769–781

24 Gilkes, N.R. et al. (1988) Precise excision of the cellulose bindingdomain from two Cellulomonas fimi cellulases by an homologousprotease and the effect on catalysis. J. Biol. Chem. 263, 10401–10407

25 Tomme,P. et al. (1988) Studies on the cellulolytic systemofTrichodermareeseiQM9414.Analysis of domain function of two cellobiohydrolases bylimited proteolysis. Eur. J. Biochem. 170, 575–581

26 Carrard, G. et al. (2000) Cellulose-binding domains promote hydrolysisof different sites on crystalline cellulose. Proc. Natl. Acad. Sci. U. S. A.97, 10342–10347

27 Lehtio, J. et al. (2003) The binding specificity and affinity determinantsof family 1 and family 3 cellulose binding modules. Proc. Natl. Acad.Sci. U. S. A. 100, 484–489

28 Watanabe, T. et al. (1994) The roles of the C-terminal domain and typeIII domains of chitinase A1 from Bacillus circulans WL-12 in chitindegradation. J. Bacteriol. 176, 4465–4472

29 Din, N. et al. (1991) Non-hydrolytic disruption of cellulose fibres by thebinding domain of a bacterial cellulase. Biotechnology. 9, 1096–1099

30 Pages, S. et al. (1997) Role of scaffolding protein CipC of Clostridiumcellulolyticum in cellulose degradation. J. Bacteriol. 179, 2810–2816

31 Gao, P.J. et al. (2001) Non-hydrolytic disruption of crystalline structureof cellulose by cellulose binding domain and linker sequence ofcellobiohydrolase I from Penicillium janthinellum. Sheng Wu HuaXue Yu Sheng Wu Wu Li Xue Bao (Shanghai) 33, 13–18

32 Kataeva, I.A. et al. (2002) The fibronectin type 3-like repeat from theClostridium thermocellum cellobiohydrolase CbhA promoteshydrolysis of cellulose by modifying its surface. Appl. Environ.Microbiol. 68, 4292–4300

33 Xiao, Z. et al. (2001) Cellulose-binding domain of endoglucanase IIIfrom Trichoderma reesei disrupting the structure of cellulose.Biotechnol. Lett. 23, 711–715

34 Nimlos, M.R. et al. (2007) Molecular modelling suggests induced fit offamily I carbohydrate-binding modules with a broken-chain cellulosesurface. Protein Eng. Des. Sel. 20, 179–187

35 Varum, K.M. and Smidsrød, O. (2005) Structure-property relationshipin chitosans. In Polysaccharides: Structural Diversity and FunctionalVersatility (Dumitriu, S., ed.), pp. 625–642, Marcel Dekker

36 Horn, S.J. et al. (2006) Costs and benefits of processivity in enzymaticdegradation of recalcitrant polysaccharides. Proc. Natl. Acad. Sci.U. S. A. 103, 18089–18094

37 Varrot, A. et al. (2003) Structural basis for ligand binding andprocessivity in cellobiohydrolase Cel6A from Humicola insolens.Structure 11, 855–864

38 Koivula, A. et al. (1998) Tryptophan 272: an essential determinant ofcrystalline cellulose degradation by Trichoderma reeseicellobiohydrolase Cel6A. FEBS Lett. 429, 341–346

Opinion Trends in Biotechnology Vol.26 No.5

39 Von Ossowski, I. et al. (2003) Engineering the exo-loop of Trichodermareesei cellobiohydrolase, Cel7A. A comparison with Phanerochaetechrysosporium Cel7D. J. Mol. Biol. 333, 817–829

40 Zhou, W. et al. (2004) Kinetic studies of Thermobifida fusca Cel9Aactive site mutant enzymes. Biochemistry 43, 9655–9663

41 Li, Y. et al. (2007) Processivity, substrate binding, and mechanism ofcellulose hydrolysis by Thermobifida fusca Cel9A. Appl. Environ.Microbiol. 73, 3165–3172

42 Percival Zhang, Y-H. et al. (2006) Outlook for cellulose improvement:screening and selection strategies. Biotechnol. Adv. 24, 452–481

43 Chandra, R.P. et al. (2007) Substrate pretreatment: the key to effectiveenzymatic hydrolysis of lignocellulosics? Adv. Biochem. Eng.Biotechnol. 108, 67–93

44 Sørbotten, A. et al. (2005) Degradation of chitosans with chitinase Bfrom Serratia marcescens; production of chito-oligosaccharides andinsight into enzyme processivity. FEBS J. 272, 538–549

Endeav

The quarterly magazine

philosophy o

You can access Ende

ScienceDirect, where you

editorial comment and a c

illustrated articles on th

Featuri

Information revolution: William ChambeDoes history count?

Waking up to shell shock: psychiatry in the UDeserts on the sea floor: Edward Forbes and his azoic hypoth‘Higher, always higher’: technology, the military and aviation

Bully for Apatosauru

Coming

Environmentalism out of the IndusPandemic in print: the spread of influen

Earthquake theories in the early mScience in fiction - attempts to make a scie

The birth of botanical Dro

And much, mu

Endeavour is available on Science

45 Sikorski, P. et al. (2006) Serratia marcescens chitinases with tunnel-shaped substrate-binding grooves show endo activity and differentdegrees of processivity during enzymatic hydrolysis of chitosan.Biochemistry 45, 9566–9574

46 Van Aalten, D.M.F. et al. (2001) Structural insights into the catalyticmechanism of a family 18 exo-chitinase. Proc. Natl. Acad. Sci. U. S. A.98, 8979–8994

47 Tews, I. et al. (1997) Substrate-assisted catalysis unifies two families ofchitinolytic enzymes. J. Am. Chem. Soc. 119, 7954–7959

48 Van Aalten, D.M.F. et al. (2000) Structure of a two-domainchitotriosidase from Serratia marcescens at 1.9 A resolution. Proc.Natl. Acad. Sci. U. S. A. 97, 5842–5847

49 Sakon, J. et al. (1997) Structure and mechanism of endo/exocellulaseE4 from Thermomonospora fusca. Nat. Struct. Biol. 4, 810–818

50 Perrakis, A. et al. (1994) Crystal structure of a bacterial chitinase at2.3 A resolution. Structure 2, 1169–1180

our

for the history and

f science.

avour online on

’ll find book reviews,

ollection of beautifully

e history of science.

ng:

rs, the publishing pioneer by A. Fyfe

by K. Anderson

S military during World War II by H. Pols

esis for a lifeless deep ocean by T.R. Anderson and T. Rice

medicine during the age of the two world wars by C. Kehrt

s by P. Brinkman

soon:

trial Revolution by C. Macleod

za in the Fin de Siecle by J. Mussell

odern period by F. Willmoth

nce out of literary criticism by J. Adams

sophila by S. Leonelli

ch more. . .

Direct, www.sciencedirect.com

235