Báo cáo khoa học: Crystal structure and enzymatic properties of a bacterial family 19 chitinase reveal differences from plant enzymes pdf

The catalytic domains of family 19 chitinases have Keywords ChiG; chitinase; family 19; Streptomyces coelicolor A32; subsite structure Correspondence V.. In accordance with these structu

family 19 chitinase reveal differences from plant enzymes Ingunn A Hoell1, Bjørn Dalhus2, Ellinor B Heggset1, Stein I Aspmo1and Vincent G H Eijsink1

1 Department of Chemistry, Biotechnology and Food Science, Norwegian University of Life Sciences, A ˚ s, Norway

2 Institute of Medical Microbiology, Section for Molecular Biology, National University Hospital, Oslo, Norway

Chitinases (EC 3.2.1.14) are glycoside hydrolases that

catalyze the hydrolysis of chitin, a carbohydrate

poly-mer of 1,4-b-linked GlcNAc Chitin is found in the

cuticle of insect shells, in shells of crustaceans, and in

the cell walls of many fungi, making chitin the second

most abundant polysaccharide in nature after cellulose

[1,2] Chitinases are present in a wide variety of

organ-isms, such as bacteria, viruses, higher plants and

animals [1–4] The hydrolysis products of chitin,

chitooligosaccharides, are of interest in several

biologi-cal and biotechnologibiologi-cal processes [1,2]

Glycoside hydrolases are divided into different fam-ilies based on primary sequence, three-dimensional structure, and catalytic mechanism [5,6] Family 18 and family 19 glycoside hydrolases both contain chitin-ases Members of the two families have very different three-dimensional structures and use different catalytic mechanisms The catalytic domains of family 18 chitin-ases have a (b⁄ a)8 fold [6] and use a substrate-assisted double-displacement mechanism, which leads to retent-ion of the configuratretent-ion of the anomeric carbon [7,8] The catalytic domains of family 19 chitinases have

Keywords

ChiG; chitinase; family 19; Streptomyces

coelicolor A3(2); subsite structure

Correspondence

V G H Eijsink, Department of Chemistry,

Biotechnology and Food Science,

Norwegian University of Life Sciences,

PO Box 5003, 1432 A ˚ s, Norway

Fax: +47 64965901

Tel: +47 64965892

E-mail: vincent.eijsink@umb.no

(Received 6 June 2006, revised 1 September

2006, accepted 4 September 2006)

doi:10.1111/j.1742-4658.2006.05487.x

We describe the cloning, overexpression, purification, characterization and crystal structure of chitinase G, a single-domain family 19 chitinase from the Gram-positive bacterium Streptomyces coelicolor A3(2) Although chi-tinase G was not capable of releasing 4-methylumbelliferyl from artificial chitooligosaccharide substrates, it was capable of degrading longer chito-oligosaccharides at rates similar to those observed for other chitinases The enzyme was also capable of degrading a colored colloidal chitin substrate (carboxymethyl-chitin–remazol–brilliant violet) and a small, presumably amorphous, subfraction of a-chitin and b-chitin, but was not capable of degrading crystalline chitin completely The crystal structures of chitinase

G and a related Streptomyces chitinase, chitinase C [Kezuka Y, Ohishi M, Itoh Y, Watanabe J, Mitsutomi M, Watanabe T & Nonaka T (2006)

J Mol Biol 358, 472–484], showed that these bacterial family 19 chitinases lack several loops that extend the substrate-binding grooves in family 19 chitinases from plants In accordance with these structural features, detailed analysis of the degradation of chitooligosaccharides by chitinase G showed that the enzyme has only four subsites () 2 to + 2), as opposed to six () 3 to + 3) for plant enzymes The most prominent structural differ-ence leading to reduced size of the substrate-binding groove is the deletion

of a 13-residue loop between the two putatively catalytic glutamates The importance of these two residues for catalysis was confirmed by a site-directed mutagenesis study

Abbreviations

ChiC, chitinase C from Streptomyces griseus HUT6037; ChiF, chitinase F from Streptomyces coelicolor A3(2); ChiG, chitinase G from Streptomyces coelicolor A3(2); CM-chitin RBV, carboxymethyl-chitin–remazol–brilliant violet; 4-MU, 4-methylumbelliferyl; TEV-protease, tobacco etch virus NIa protease.

high a-helical contents and share some structural

simi-larity with chitosanases and lysozyme [9,10] Family 19

chitinases use a single-displacement mechanism, which

leads to inversion of the configuration of the anomeric

carbon [6,11] The catalytic mechanism of family 19

chitinases has been studied in detail by modeling [12],

but experimental studies that underpin the proposed

mechanism are remarkably scarce [13] In fact, in the

CAZY database of glycoside hydrolases [5] (http://

afmb.cnrs-mrs.fr/CAZY/), the catalytic proton donor

and the catalytic base are not annotated

Family 19 chitinases are commonly found in many

plants, but were only recently discovered in bacteria

The first bacterial family 19 chitinase, chitinase C

(ChiC), was found in Streptomyces griseus HUT6037

in 1996 [14] Subsequently, several bacterial family 19

chitinases have been identified, including chitinases

from Burkholderia gladioli, Vibrio cholerae,

Haemophi-lus influenzae, and Pseudomonas aeruginosa Although

many plant family 19 chitinases are known, crystal

structures are available for only two of these [9,15–17]

The first structure of a bacterial family 19 chitinase

has just recently been solved (Streptomyces griseus

HUT6037 ChiC; PDB accession code 1WVU) [18]

Streptomyces coelicolor A3(2) is a spore-forming

soil-borne Gram-positive bacterium that grows via a

branching mycelium, mainly by tip growth [19] The

genome was fully sequenced in 2002, and this revealed

that the ability of S coelicolor A3(2) to exploit

nutri-ents in the soil is associated with the ability to produce

many different hydrolases, including 13 chitinases [20]

Among these chitinases we find two putative family 19

enzymes, ChiF and ChiG [21,22], which share 84%

identity to each other, and 80% and 75% identity to

the catalytic domain of ChiC from Streptomyces

gri-seus, respectively [21] ChiF has a similar domain

structure to ChiC, consisting of a catalytic domain

and an N-terminal chitin-binding domain ChiG, on

the other hand, lacks this chitin-binding domain, and

consists only of a catalytic domain [22] The chiG gene

encodes a 244 amino acid chitinase, including a 29

amino acid leader peptide sequence The closest

relat-ive of ChiG among the two plant family 19 chitinases

with known crystal structures comes from Canavalia

ensiformis (Jack beans; 37% sequence identity)

Inter-estingly, ChiG and most other bacterial chitinases

seem to have different catalytic centers from the plant

enzymes, as there is a 13-residue deletion between the

putative catalytic residues, thus making them closer in

sequence in the former (Glu68 and Glu77 in ChiG;

Fig 1) Inspection of the structures of the two plant

enzymes shows that this deletion is located on a loop

near the (putative) + 2 subsite [9,12] Sequence

align-ments (Fig 1) show several other deletions in the bac-terial enzymes that potentially could affect interactions with the substrate In order to provide more insight into family 19 chitinases in general and into the differ-ences between plant and bacterial enzymes in partic-ular, we have overexpressed ChiG from S coelicolor A3 (2) in Escherichia coli and characterized the enzyme with respect to its catalytic properties and crystal structure The crystal structures of ChiG and ChiC [18] permitted structural comparison between bacterial and plant enzymes, which provided a structural explanation for observed differences in enzymatic properties The role of the putative catalytic residues was confirmed by site-directed mutagenesis

Results

Enzymology Overexpression of ChiG in E coli BL21Star (DE3) yielded soluble and active enzyme, which could easily

be purified by Ni-affinity chromatography (supple-mentary Fig S1; typical yields were in the range 2–7 mg of ChiG per liter of culture) Tests with several substrates showed that removal of the His tag by tobacco etch virus NIa protease (TEV-protease) did not affect the catalytic properties of the enzyme ChiG did not show any activity against 4-methylumbelliferyl (4-MU)-(GlcNAc)2or 4-MU-(GlcNAc)3, as also earlier demonstrated by Saito et al [21] On the other hand, ChiG showed activity against a-chitin and b-chitin (supplementary Fig S2), chitooligosaccharides (Figs 2 and 3), carboxymethyl-chitin–remazol–brilliant violet (CM-chitin RBV) and also against chitosan (E Hegg-set, I A Hoell & K M Va˚rum, unpublished results) Figure 2 shows the kinetics of product formation during oligosaccharide degradation Specific activities (derived from initial substrate disappearance rates, Fig 2) were 1.13 ± 0.07 lmolÆs)1Æmg)1, 0.63 ± 0.03 lmolÆs)1Æmg)1, 0.49 ± 0.05 lmolÆs)1Æmg)1, and 0.08 ± 0.01 lmolÆs)1Æmg)1 for (GlcNAc)6, (GlcNAc)5, (Glc-NAc)4 and (GlcNAc)3, respectively Degradation of both a-chitin and b-chitin yielded (GlcNAc)2and (Glc-NAc)3, whereas after long incubation times (24 h) significant amounts of GlcNAc were observed, due

to further degradation of (GlcNAc)3 to (GlcNAc)2 and GlcNAc (results not shown) The specific initial activities towards a-chitin and b-chitin (judged from

a short initial linear phase in product formation) were 0.09 ± 0.01 lmolÆs)1Æmg)1 and 0.11 ± 0.03 lmolÆs)1Æmg)1, respectively However, only a minor fraction of the chitin was degraded at these speeds; the reactions rapidly slowed down and a larger part of

the substrates remained undegraded, even after

pro-longed incubation Thus, ChiG is much less effective

than e.g the two-domain family 18 chitinase ChiA

from S marcescens which can degrade b-chitin

com-pletely (supplementary material Fig S2) Yields were

lowest for a-chitin, whereas ChiA is capable of almost

completely degrading this substrate too, albeit at a

slower rate [23]

Figure 2 shows that the tetramer is exclusively

con-verted to two dimers, meaning that there is only one

binding mode for this substrate The longer oligomeric

substrates have several potentially productive binding

modes Preferred binding modes can be analyzed by

determining anomeric ratios of products formed early

during the reaction [24–26] The results (Fig 3,

Table 1) show that during degradation substrates

stayed close to the expected equilibrium ratio of 60%

a-anomer and 40% b-anomer, whereas all products

had anomer ratios that were close to 80 : 20 One

would expect an 80 : 20 ratio: (a) if the enzyme is inverting (that is, each new reducing end has an a-ano-meric configuration, as would be expected for a family

19 enzyme [11]); and (2) if each product contains a

50 : 50 mixture of newly formed (100% a) and existing (60% a) reducing ends Figure 2A shows that (Glc-NAc)6 was hydrolyzed to (GlcNAc)4+ (GlcNAc)2 and (GlcNAc)3+ (GlcNAc)3 The efficiency of the first reaction was about double that of the second reac-tion (see legend to Fig 2) The 80 : 20 anomeric distri-bution in the tetramer and dimer fractions (Fig 3, Table 1) shows that the first reaction equally often results from cleavage between sugars 4 and 5 (new reducing end on the tetramer product) as from clea-vage between sugars 2 and 3 (new reducing end on the dimer product) Thus, the hexamer is degraded through three types of productive binding modes with approximately similar frequencies, leading to cleavage after sugars 2, 3 or 4 Hydrolysis of (GlcNAc)5initially

Fig 1 Structure-based multiple sequence alignment of all family 19 chitinases with known structure The figure shows two plant enzymes, from Hordeum vulgare (barley [9]) and Canavalia ensiformis (Jack bean [17]), and two bacterial enzymes, chitinase C (ChiC) (PDB accession code 1WVU) from Streptomyces griseus HUT 6037 (catalytic domain only [18]) and chitinase G (ChiG) The alignment was made using the protein structure comparison service SSM at the European Bioinformatics Institute (http://www.ebi.ac.uk/msd-srv/ssm [45]) Four major dele-tions in the bacterial enzymes are indicated by A, B, C and C-term, and the 161–166 loop (numbering of the barley enzyme; see text) is indi-cated by dots above the sequence Residues involved in disulfide bridge formation are marked with closed or open bullets, for conserved and nonconserved bridges, respectively The closed triangles indicate two conserved glutamate residues involved in catalysis Fully con-served secondary structure assignments are indicated with h for a-helix and s for b-strand The consensus helix comprising residues 169–

177 in ChiG is extended towards Cys183 in the other three enzymes.

yielded equal amounts of (GlcNAc)3 and (GlcNAc)2;

the 80 : 20 anomeric ratio of the products indicates

that cleavage after sugar 2 or sugar 3 occurs

approxi-mately equally often

Structure

The overall structure of ChiG (Fig 4, supplementary

Fig S3) is similar to that of the family 19 chitinase

from barley, the best studied of the plant chitinases

[9,12,13,27,28] and essentially identical to that of the

catalytic domain of ChiC from S griseus HUT6037

([18]; rms 0.84 A˚) The only notable difference between

the two bacterial enzymes occurs in the 178–183

region: the consensus helix comprising residues 169–

177 in ChiG (Fig 1) is extended towards Cys183 in

ChiC (and in the plant enzymes)

Compared to the barley enzyme, ChiC and ChiG

lack three loops (A, B and C) and a C-terminal

exten-sion (Figs 1 and 4B) In addition, one other loop,

comprising residues 161–166 in the barley enzyme, has shifted its position by up to 9 A˚ (Fig 4B) Two of the three disulfide bridges found in the plant enzymes are conserved (in ChiG: Cys87–Cys95 and Cys183– Cys215), whereas the third bridge is lacking, due to the deletion of the A-loop (Fig 1) The enzyme has a deep groove that is likely to bind the substrate [6,12] and that contains the putative catalytic residues Glu68 and Glu77 (Figs 1 and 4) For an inverting enzyme, one would expect the distance between the carboxyl oxygens of these two glutamates to be about 10 A˚ [6]

In ChiG, this distance is 9.5 A˚ for the closest pair of oxygens

The four major deletions in ChiG (loops A, B and

C and the C-terminus) as well as the one major struc-tural difference (161–166 loop) compared to the plant enzymes can be divided into two subsets of interrelated changes, with each subset affecting one side of the sub-strate-binding groove of the enzyme (Fig 4B,C) On the side where the nonreducing end of the substrate

Fig 2 Time course of the degradation of chitooligosaccharides by chitinase G (ChiG) (A) Hexamer (B) Pentamer (C) Tetramer (D) Trimer The concentrations of the various oligosaccharides are indicated by (hexamer), } (pentamer), h (tetramer), n (trimer), X (dimer) and s (monomer) All reactions were run under identical conditions, except for the reaction with trimer, in which the enzyme concentration was increased 50-fold In (A), note that a single cleavage can produce two trimers or one dimer + one tetramer; the graph thus shows that the reaction producing tetramer + dimer happens about twice as often as the trimer-producing reaction In the reactions depicted in (A) and (B), monomers were only detected after prolonged incubation, i.e after depletion of the original substrate With the tetramer (C), monomers were never detected.

binds, loop C and the C-terminal extension affect the

position of the 161–166 loop, which contains two polar

side chains (Gln162 and Lys165) thought to be

import-ant for sugar binding in subsites –3 and –4 (in the

barley enzyme [12,27]) In ChiG, in the absence of the

C-loop and the C-terminal extension, the 161–166 loop

has moved toward the catalytic center In addition, the

Gln and Lys residues have been replaced by Thr

(Thr153 and Thr157) Thus, ChiG seems to have a

reduced ability to bind sugars in the ) 3 and ) 4 posi-tions

On the side where the reducing end of the substrate binds, the interacting loops A and B in the barley enzyme extend the substrate-binding surface beyond subsite + 2, primarily through the exposed Trp72 The importance of tryptophans in positions such as Trp72 for the efficiency of chitinolytic enzymes is well estab-lished [29] There is another Trp at position 82 in loop

B, which is shielded from solvent in the barley enzyme (Fig 4C), but which is more exposed in the absence of Trp72, as in the jack bean enzyme ChiG lacks loops

A and B, and thus seems to have reduced ability to bind sugars beyond subsite + 2 In addition, Thr69, thought to be important for sugar binding in subsite + 2 of the barley enzyme, is not conserved and is replaced by Gly in ChiG

No structural data were obtained for the 11 N-ter-minal residues of ChiG Compared to the barley and the jack bean enzymes, ChiG contains an N-terminal extension of eight and seven residues respectively (Fig 1) The N-termini of the plant enzymes and the first residue in the ChiG structure (Phe12) (Fig 4A), are located on the opposite side of the enzyme to the catalytic center and the substrate-binding groove The same applies to the structurally observed N-terminus

of the catalytic domain of ChiC, which corresponds to residue 8 in ChiG [18] In ChiC, this N-terminus is part of a linker (with unknown structure) that connects

an N-terminal chitin-binding domain to the catalytic domain All in all, it is highly unlikely that the N-ter-minal extensions in ChiG directly affect catalytic archi-tecture

Mutagenesis of the catalytic center Figure 1 shows that two glutamates thought to make

up the catalytic center in the barley family 19 chitinase [9,13] are conserved in ChiG, despite the large deletion

in between these two residues The role of these gluta-mates was demonstrated by site-directed mutagenesis Table 2 shows that all mutants had greatly reduced catalytic activity Some detectable activity was still left upon mutation of Glu77 (2000–6000-fold reduction in activity), whereas mutation of Glu68 reduced activity

to below the level that could be detected with our assays (> 24 000-fold reduction in activity)

Discussion

Whereas family 19 chitinases are widespread in higher plants, their occurrence in prokaryotes has only recently been discovered [14,22,26,30] Judged by

avail-Fig 3 HPLC analysis of reaction mixtures under conditions

pre-venting anomeric equilibrium The top panel represents a standard

mixture of GlcNAc oligomers showing the standard 60 : 40 ratio

between the a-anomer and the b-anomer at equilibrium The other

panels show the results of partial hydrolysis of (GlcNAc) 6 and

(Glc-NAc)5by chitinase G (ChiG) In these panels, substrates display the

60 : 40 ratio, whereas the ratios for the products are close to

80 : 20 See text and Table 1 for details The small peaks close to

the tetramer position in the chromatogram for (GlcNAc)5and close

to the pentamer position in the chromatogram for (GlcNAc)6were

also present in control samples and are not due to enzyme action.

Table 1 Anomeric configuration in the reaction mixtures depicted

in Fig 3.

Hydrolysis of (GlcNAc) 6 Hydrolysis of (GlcNAc) 5

a-Anomer

(%)

b-Anomer (%)

a-Anomer (%)

b-Anomer (%) Products

able sequences, some bacterial family 19 chitinases

have catalytic domains that are at least as large as

those of the plant enzymes and that may contain at

least six subsites [26,30] However, the catalytic

domains of ChiG, ChiC and most other known bacter-ial family 19 chitinases are smaller than those of the plant enzymes Unfortunately, there is no direct struc-tural information concerning the interaction between

C

Fig 4 Structure of chitinase G (ChiG) and comparison with the barley chitinase (A) Cartoon showing the overall fold of ChiG with trans-parent surface The side chains of the catalytic residues, Glu68 and Glu77 are shown in red (B) Structural superposition of ChiG and the barley enzyme The picture shows a cartoon of the barley enzyme (PDB accession code 2bba; cyan) and the surface of ChiG, with the view being rotated 90 relative to (A) (the view is into the substrate-binding groove) Important structural elements are labeled (see text for details), and the catalytic glutamates are shown in red (C) Differences between the barley enzyme (cyan, left) and ChiG (blue, right)

in the substrate-binding cleft The side chains of the catalytic residues are shown in green The side chains of residues that are deleted (Trp72, Trp82), mutated (Thr69 ⁄ Gly70) or mutated and relocated (Lys165 ⁄ Thr157 and Gln162 ⁄ Thr153) in ChiG are shown in red The side chains of four fully conserved residues in subsites ) 2 (right side) to + 2 (left side) are shown in purple Note that the ) 1 and ) 2 sub-sites partly consist of backbone atoms [12]; these are structurally well conserved, but not shown in the picture The pictures were made with PYMOL [47].

family 19 chitinases and their substrates Soaking

experiments were not successful and nor were

cocrys-tallization experiments with the inactive mutant E68Q

[9] (B Dalhus, S I Aspmo & I A Hoell, unpublished

results) However, the interaction between the barley

chitinase and (GlcNAc)6 has been studied in great

detail by computational techniques, exploiting the

(lim-ited) structural similarity between family 19 chitinases

and lysozyme [9,12,27,31]; (the crystal structure of a

lysozyme–(GlcNAc)3 complex was used for modeling

purposes) By analogy to lysozyme, these studies

assumed the presence of six subsites, running from) 4

to + 2 [subsites are numbered according to standard

nomenclature; cleavage occurs between the sugar units

bound in subsites ) 1 and + 1 [32]; (note that in the

older literature, these subsites are referred to as A

() 4) to F (+ 2)] Judging by the structure of the

bar-ley enzyme, one would assume that there is affinity

for the substrate beyond the + 2 subsite, primarily

because of the prominent Trp residue at position 72,

approximately 15 A˚ from the catalytic center This Trp

would be able to interact with sugars bound at

posi-tions + 3 and + 4 Indeed, analysis of the hydrolysis

of (GlcNAc)6 by barley chitinase [28] and by a highly

similar chitinase from rice [25] led to the conclusion

that these enzymes do have a + 3 subsite with

consid-erable affinity for a sugar moiety

All residues thought to be involved in sugar binding

at the ) 2 to + 2 subsites in the barley enzyme are

fully or, at least functionally, conserved in ChiG

(Fig 4C), except for Thr69 in the + 2 subsite, which is

replaced by a glycine Beyond this central region,

ChiG clearly differs from the barley enzyme, as a

consequence of the loop deletions and the resulting

conformational change in the 161–166 loop The

dele-tion of the Trp72-containing loop (loop B in Fig 4B)

removes putative subsites + 3 and + 4, whereas the

conformational change of and the mutations in the

161–166 loop remove putative subsites ) 4 and ) 3

[12,27] Thus, in ChiG, the substrate-binding groove⁄ surface is less extended and does not seem to contain more than the four central subsites The pres-ence of only four subsites was confirmed by studies on the degradation of pentamers and hexamers (Fig 2, Table 1) For example, productive binding of the hex-amer by ChiG occurs in three different binding modes (in ‘subsites’ ) 4 to +2, ) 3 to +3 and ) 2 to +4) with almost identical frequencies This shows that there is little binding affinity in subsites beyond ) 2 and + 2 The barley and rice enzymes show clearly dif-ferent product profiles, primarily due to the presence

of a + 3 subsite [25,28] Most interestingly, whereas ChiG hydrolyzed tetrameric, pentameric and

hexamer-ic substrates with rather similar rates (varying less than 2.5-fold), the efficiency of the barley enzyme is strongly dependent on substrate length Studies by Hollis et al [27] showed that the barley enzyme degrades the hex-amer about 200 times faster than the tetrhex-amer This confirms that the barley enzyme and ChiG have dif-ferent catalytic properties, in accordance with the observed structural differences Using structural infor-mation only, Kezuka et al [18] have hypothesized that ChiC has six subsites, namely subsites) 4 to + 2, as in hen egg-white lysozyme This hypothesis is not con-firmed by the present analysis of enzymatic properties

of ChiG, or by our analysis of the ChiG and ChiC structures

Despite the deletion of the B-loop, the two putative catalytic residues Glu68 and Glu77 are structurally well conserved between ChiG and the plant enzymes Andersen et al [13] have previously shown that the corresponding residues in the barley enzyme, Glu67 and Glu89, are essential for catalysis The mutagenesis studies presented here show that Glu68 and Glu77 are essential for catalysis by ChiG Mutation of Glu68 to Gln resulted in total inactivation, whereas mutation of Glu77 did not This is in accordance with the notion that Glu68 is the catalytic acid, whereas Glu77 is the catalytic base [6,13]

The activity of ChiG towards chitooligosaccharides was found to be comparable to that of other

chitinas-es, including, for example, well-known family 18 exochitinases and endochitinases from Serratia marces-cens [23] ChiG showed relatively high initial activity towards chitin, but the overall ability to degrade the polymer was limited, as compared to, for example multidomain family 18 chitinases from S marcescens [33] (supplementary Fig S2) Thus, while ChiG is rather active towards soluble substrates and some (amorphous) regions of insoluble chitin, the enzyme is not very efficient in degrading crystalline polymeric substrates Other bacterial family 19 chitinases such as

Table 2 Specific activity of site-directed mutants of ChiG towards

(GlcNAc) 6

Enzyme

Specific activity towards (GlcNAc)6 (lmolÆmg)1Æs)1)

Relative specific activity

a Estimated on the basis of the approximate detection limit of the

assay.

ChiC and ChiF contain an additional

substrate-bind-ing domain, which could make these enzymes more

efficient with crystalline substrates However, deletion

of this domain had only a modest effect on enzyme

efficiency with crystalline chitin [34] Like ChiG, both

ChiC and ChiF display relatively high activities

towards noncrystalline chitin forms, and low activities

towards crystalline chitin [22,35] It has previously

been shown that chitinases with high activity towards

crystalline chitin, such as Bacillus circulans chitinase

A1 and S marcescens chitinase A, have extended

sub-strate-binding grooves (at least six subsites) Notably,

these grooves contain a stretch of linearly aligned

aro-matic residues that play an important role in guiding a

chitin chain from the crystalline chitin surface to the

catalytic center [29] Our finding that the bacterial

enzymes have only four subsites and the absence of

aromatic residues in these subsites may explain why

ChiG and related enzymes have low activity towards

crystalline chitin The open active site of ChiG

sug-gests that ChiG binds polymers in an endo-fashion

This was confirmed by the observation that hydrolysis

of chitin led to significant production of trimers during

degradation of both a-chitin and b-chitin (exoenzymes

tend to almost exclusively produce dimers) Studies on

the degradation of colloidal chitin by ChiC led to a

similar conclusion [14]

Most probably, family 19 chitinases were transferred

from plants to bacteria by horizontal gene transfer [22]

In plants, family 19 chitinases are thought to form part

of a defense mechanism against chitin-containing

fun-gal pathogens [36] The family 19 chitinases are thought

to attack the hyphal tips, which are believed to consist

of newly synthesized chitin that is not firmly

crystal-lized [35] This is in accordance with the observation

that family 19 chitinases generally have relatively low

activities towards the more crystalline forms of chitin

Only a few chitinolytic bacteria possess family 19

chitinases, and these also display antifungal activity

[22,35] In bacteria, chitinases, primarily belonging to

family 18, are usually thought to be produced for the

exploitation of chitinous substrates as a source of

nutri-tion Production of multiple enzymes with varying

properties (endo-action or exo-action, processive or

not, presence of additional substrate-binding domains,

preference for soluble or insoluble substrates) is

benefi-cial, because this enables the bacterium to use parallel

and potentially synergistic strategies during chitin

breakdown Chitin occurs in a variety of forms and

co-polymeric structures [37], which may require different

chitinases for effective degradation It remains to be

seen whether the two family 19 chitinases of S

coelicol-or simply add to the bacterium’s enzymatic repertoire

for effective chitin turnover, or play a specific role in some form of interaction with fungi

Experimental procedures

DNA techniques The chiG gene was amplified from genomic DNA (ATCC BAA-471D) from S coelicolor A3(2) with: primer Chi-Gul_S.coeli-F, 5¢-GCATCGTCTCACATGGAGAAGTCC GACACCCGGA-3¢ (BsmBI restriction site is in bold type); and primer ChiG_S.coeli-R, 5¢-GCATGGTACCCTAAC AGCTCAGGTT-3¢ (KpnI restriction site is in bold type) PCR reactions were conducted with Phusion DNA polym-erase (Finnzymes, Espoo, Finland) in a PTC-100 Program-mable Thermal Cycler (MJ Research, Inc., Waltham, MA, USA) The amplification protocol consisted of an initial denaturation cycle of 30 s at 98C, followed by 30 cycles

of 10 s at 98C, 30 s at 58 C, and 30 s at 72 C, followed

by a final step of 10 min at 72C Amplified fragments were ligated into vector pCR4Blunt-TOPOZero Blunt TOPO (Invitrogen, Carlsbad, CA, USA) The gene frag-ments were excised from the TOPO vectors for insertion in

an expression vector, using BsmBI and KpnI for cloning into NcoI–KpnI-digested pETM11 vector (Gu¨nter Stier, EMBL, Heidelberg, Germany) The pETM11 vector con-tains an N-terminal His6 tag followed by a TEV-protease cleavage site The final constructs were transformed into

E coliBL21Star (DE) (Invitrogen)

ChiG mutants (E68Q, E68A, E77Q and E77A) were made with the QuickChange Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA, USA), essentially as described

by the manufacturer DNA sequencing was performed using a BigDye Terminator v3.1 Cycle Sequencing Kit (Perkin Elmer⁄ Applied Biosystems, Foster City, CA, USA) and an ABI PRISM 3100 Genetic Analyser (Perkin Elmer⁄ Applied Biosystems)

Production and purification of recombinant protein

One hundred and fifty milliliters of E coli BL21Star (DE3) transformants containing the pETM11–chiG con-struct were grown at 37C in LB medium with

100 lgÆmL)1 kanamycin at 225 r.p.m., to a cell density of 0.6 at 600 nm Isopropyl-b-d-thiogalactopyranoside was added to a final concentration of 0.4 mm, and the cells were further incubated for 4 h at 30C, and harvested

by centrifugation (9 820 g, 8 min at 4C, Beckman Coul-ter Avanti J-25, Rotor JA14) The cell pellet was lysed

by making a periplasmatic extract First, the cell pellet was resuspended in 15 mL of ice-cold spheroplast buffer (10 mL of 1 m Tris⁄ HCl, pH 8.0, 17.1 g of sucrose,

100 lL of 0.5 m EDTA, pH 8.0, and 200 lL of

phenyl-methanesulfonyl fluoride and incubated on ice for 5 min.

The cells were then harvested by centrifugation (7 741 g,

8 min at 4C, Beckman Coulter Avanti JA25-5), the

supernatant was removed, and the pellet was incubated

for 10 min at room temperature The pellet was then

re-suspended in 12.5 mL of ice-cold, sterile water, and

incu-bated on ice for 45 s before supplementing with 625 lL

of 20 mm MgCl2 After centrifugation (7741 g, 8 min at

4C, Beckman Coulter Avanti JA25-5), the supernatant

was pressed through a 0.20 lm sterile filter, supplied with

20 lL of 50 mm phenylmethanesulfonyl fluoride per

10 mL of extract, and stored at 4C It has previously

been shown that these extracts are good, relatively ‘clean’

starting points for purification of intracellularly produced

chitinases [38]

ChiG was purified on an Ni-NTA column (Qiagen,

Venlo, The Netherlands) using a flow rate of 2 mLÆmin)1

The column was equilibrated in 100 mm Tris⁄ HCl buffer

(pH 8.0), containing 20 mm imidazole After the protein

sample was loaded, the column was washed with the

start-ing buffer The His-tagged protein was then eluted with

100 mm Tris⁄ HCl buffer (pH 8.0), containing 100 mm

imi-dazole The purified protein was dialyzed against 20 mm

Tris⁄ HCl (pH 8.0) and stored at 4 C

Removal of the (His)6 tag was preformed by mixing

0.1 mg of (His)6–ChiG with 75 lL of 10· TEV-protease

buffer (0.5 m Tris⁄ HCl, pH 8.0, and 5 mm EDTA), 1 mm

dithiothreitol, 0.005 mg of TEV-protease and dH2O up to

750 lL This mixture was incubated at 37C for 3 h After

incubation, the mixture was dialyzed against 100 mm

Tris⁄ HCl (pH 8.0) and 20 mm imidazole overnight The

dialyzed mixture was then applied onto an Ni-NTA

col-umn, as described above The flow-through fraction, now

containing the ChiG protein with no (His)6 tag, was

dia-lysed against 20 mm Tris⁄ HCl (pH 8.0) and stored at 4 C

The protein produced via this procedure contains a

three-residue N-terminal extension (Gly-Ala-Met) compared to

the mature wild-type enzyme

Structure determination and bioinformatics

High-quality diffracting crystals of ChiG were obtained

by the vapor diffusion method in hanging drops Prior

to crystallization, ChiG was concentrated by using a

Centricon Plus-20 Centrifugal Filter Device as described

by the manufacturer (Millipore, Billerica, MA, USA) in

20 mm Tris⁄ HCl (pH 8.0) to a final concentration of

10 mgÆmL)1 Equal volumes of the protein solution were

mixed with the reservoir solution containing 13% (w⁄ v)

PEG8000 and 110 mm zinc acetate in 80 mm sodium

cacodylate buffer (pH 6.5), and equilibrated against the

reservoir solution at room temperature Crystals, in the

shape of thin plates, grew to a final size of about

0.2 mm within a week Crystals were mounted in nylon

loops and flash-frozen in liquid nitrogen following a

short (< 10 s) soak in mother liquor containing addi-tional PEG400 to a final concentration of 15%

Complete X-ray data were collected at beamline ID14-EH3 at the ESRF in Grenoble, equipped with an ADSC Q4R detector Diffraction images were processed with mos-flm [39] and scaled and merged with scala in CCP4 [40,41] The crystals belong to space group P21 with cell dimensions a¼ 48.67 A˚, b ¼ 74.38 A˚, c ¼ 64.18 A˚ and

b¼ 108.6, and diffracted to at least 1.5 A˚ resolution Crystal data and data collection statistics are summarized

in Table 3

Calculation of the Matthews coefficient suggested two molecules in the asymmetric unit The structure was solved

by molecular replacement using cns [42] The search model was a polyalanine chain comprising residues 90–294 of the ChiC structure (1WVU) Two solutions in the cross-rota-tion funccross-rota-tion were readily identified, and a subsequent translation search gave the positions in the unit cell Side chains were progressively added, guided by information

Table 3 Crystal parameters, data collection and refinement statis-tics for Streptomyces coelicolor chitinase G (ChiG).

Crystal parameters Crystal dimensions (mm) 0.2 · 0.2 · 0.05

Unit cell dimensions a ¼ 48.67 A˚, b ¼ 74.38 A˚,

c ¼ 64.18 A˚, b ¼ 108.6 Data collection

R merge (%)a 7.6 (48.1)

Refinement Resolution range (A ˚ ) 74.3–1.50

rms deviation from ideal geometry

Ramachandran distribution (%)d

a Values in parentheses refer to the outermost shell of data (1.58– 1.50 A ˚ ) b Values in parentheses refer to the outermost shell of data (1.54–1.50 A ˚ ) c Five per cent of data, randomly distributed over the full resolution range, were flagged as belonging to the

R free cross-validation set, not used in the refinement.dAccording

to COOT [43].

from both the rA-weighted 2Fo–Fcand Fo–Fcmaps, during

several cycles of modeling using coot [43], following

refine-ment with refmac5 [44] Water molecules were appended

using the ‘add water’ function of coot Four peaks close to

His67⁄ Glu182 and Asp184 in molecule A and His67 ⁄

Glu182 and Asp137 in chain B, originally modeled as

waters, were replaced by zinc ions, based on the refined

B-values and residual peak heights in the Fo–Fcmap These

zinc ions originate from the crystallization buffer The main

chain was readily traced from residue Phe12 all the way to

the C-terminal Cys215 A few side chains at the protein

sur-face are flexible, with no distinct conformation Refinement

of ChiG with zero occupancy for residues 170–183

con-firmed the (slightly) different conformation for ChiG in the

178–183 region, as judged by inspection of the difference

Fourier electron density map The final model comprises

408 residues in two chains, four zinc ions and 324 water

molecules

Coordinates and structure factors have been deposited in

the Protein Data Bank, accession code 2CJL

To create the alignment of Fig 1, structural superposition

with other family 19 chitinases was performed using the

pro-tein structure comparison service SSM at the European

Bioinformatics Institute (http://www.ebi.ac.uk/msd-srv/ssm)

[45]

Enzymology

Protein concentrations were determined according to

Brad-ford with the Bio-Rad Protein Assay Kit (Bio-Rad,

Hercu-les, CA, USA) with BSA as a standard

Analyses of the specific activity against

chitooligosaccha-rides were performed in 100 lL reaction mixtures

con-taining 200 lm (GlcNAc)3, (GlcNAc)4, (GlcNAc)5, or

(GlcNAc)6(Sigma, St Louis, MO, USA), 0.1 mgÆmL)1BSA

and 0.25 nm purified ChiG in 50 mm sodium acetate buffer

(pH 6.0) In the case of the (GlcNAc)3 substrate, the

enzyme concentration was 12.5 nm In the case of ChiG

mutants, the enzyme concentration was varied between

250 nm and 500 nm (see below for details) All the reaction

mixtures were incubated at 37C for several hours, with

regular sampling Sixty microliter samples of the reaction

mixture were transferred to new tubes containing 60 lL of

70% acetonitrile, to stop the reaction, and stored at

) 20 C until they were analyzed by HPLC at room

tem-perature All reactions were analyzed in triplicate

HPLC analysis of 20 lL portions of the stored reaction

mixtures was performed on a Gilson HPLC System

(Gil-son, Inc., Middleton, WI, USA), equipped with a Tosoh

TSK-Gel amide-80 column (0.46 internal diameter· 25 cm)

(Tosoh Bioscience, Montgomeryville, PA, USA), and a 234

autoinjector (Gilson) The liquid phase consisted of 70%

(v⁄ v) acetonitrile, the flow rate was 0.70 mLÆmin)1, and

eluted oligosaccharides were monitored by recording

absorption at 210 nm

In cases where analysis of the anomeric configuration of the newly formed degradation products was desirable, reac-tions were performed with higher enzyme concentrareac-tions (20 nm) and very short incubation times (approximately

15 s) To stabilize the anomeric ratio as fast as possible and

to avoid reaching the anomeric equilibrium, reactions were stopped by freezing on liquid nitrogen and samples were stored at ) 80 C until analyzed Ten microliter samples of the reaction mixtures were injected with a Gilson 234 auto-injector immediately after thawing (that is, samples were not ‘stored’ in the autoinjector)

Analyses of the degradation of a-chitin and b-chitin were conducted by incubating 100 lL solutions containing

1 mgÆmL)1of b-chitin (squid pen b-chitin, 3 lm in size; Sei-kagaku, Tokyo, Japan) or 1 mgÆmL)1 a-chitin (crab-shell a-chitin, Sigma) and 20 nm purified ChiG in 50 mm sodium acetate buffer (pH 6.0) at 37C and 230 r.p.m for periods varying from 0 min (just after addition of enzyme) to 24 h Reactions were stopped by adding one volume of 70% (v⁄ v) acetonitrile, and samples were stored at) 80 C until they were injected with a 234 autoinjector (Gilson)

Activity tests with a colloidal chitin substrate, CM-chitin RBV (LOEWE Biochemica GmbH, Mu¨nchen), and with 4-MU-(GlcNAc)2 (Sigma) or 4-MU-(GlcNAc)3 (Sigma) were performed as described earlier [46]

Acknowledgements

We thank Gu¨nter Stier, EMBL Heidelberg, for provi-ding us with vector pETM11, Gustav Vaaje-Kolstad for help with the chitin degradation experiments and helpful discussions, and May Bente Brurberg and Svein Horn for helpful discussions The authors acknowledge the beamline staff at ID14-EH3 for tech-nical assistance, and the ESRF and the Norwegian Research Council (project Sygor) for financial support This work was funded by a grant from the Norwegian Research Council (no 140⁄ 140497)

References

1 Synowiecki J & Al-Khateeb NA (2003) Production, properties, and some new applications of chitin and its derivatives Crit Rev Food Sci Nutr 43, 145–171

2 Peter MG (2002) In Biopolymers, Vol 6: Polysaccha-rides II(Steinbu¨chel A, ed.), pp 481–574 Wiley-VCH, Weinheim

3 Zhu Z, Zheng T, Homer RJ, Kim YK, Chen NY, Cohn L, Hamid Q & Elias JA (2004) Acidic mammalian chitinase in asthmatic Th2 inflammation and IL-13 pathway activation Science 304, 1678– 1682

4 Kasprzewska A (2003) Plant chitinases) regulation and function Cell Mol Biol Lett 8, 809–824