Endo/exo mechanism and processivity of family 18 chitinases produced by Serratia marcescens


V.G.H. Eijsink, Department of Chemistry, Biotechnology and Food Science, The Norwegian University of Life Sciences, PO Box 5040, 1432 Ås, Norway
Tel: +47 64965892
Fax: +47 64965901
E-mail: vincent.eijsink@umb.no
Website: http://www.umb.no


We present a comparative study of ChiA, ChiB, and ChiC, the three family 18 chitinases produced by Serratia marcescens. All three enzymes eventually converted chitin to N-acetylglucosamine dimers (GlcNAc2) and a minor fraction of monomers. ChiC differed from ChiA and ChiB in that it initially produced longer oligosaccharides from chitin and had lower activity towards an oligomeric substrate, GlcNAc6. ChiA and ChiB could convert GlcNAc6 directly to three dimers, whereas ChiC produced equal amounts of tetramers and dimers, suggesting that the former two enzymes can act processively. Further insight was obtained by studying degradation of the soluble, partly deacetylated chitin-derivative chitosan. Because there exist nonproductive binding modes for this substrate, it was possible to discriminate between independent binding events and processive binding events. In reactions with ChiA and ChiB the polymer disappeared very slowly, while the initially produced oligomers almost exclusively had even-numbered chain lengths in the 2–12 range. This demonstrates a processive mode of action in which the substrate chain moves by two sugar units at a time, regardless of whether complexes formed along the way are productive. In contrast, reactions with ChiC showed rapid disappearance of the polymer and production of a continuum of odd- and even-numbered oligomers. These results are discussed in the light of recent literature data on directionality and synergistic effects of ChiA, ChiB and ChiC, leading to the conclusion that ChiA and ChiB are processive chitinases that degrade chitin chains in opposite directions, while ChiC is a nonprocessive endochitinase.


carbohydrate-binding module


degree of polymerization

GlcNAc or A


GlcN or D


Chitin is a linear insoluble polymer of β-1,4 linked N-acetylglucosamine (GlcNAc or A), which is synthesized by crustaceans, molluscs, algae, insects, fungi and yeasts. Second only to cellulose, chitin is an abundant biopolymer with an annual production of 100 billon tons [1]. In nature, two major types of chitin occur which are characterized by an antiparallel (α-chitin) or a parallel (β-chitin) arrangement of the N-acetylglucosamine chains [2,3]. One industrial exploitation route for chitin involves its conversion to chitosan, a water-soluble copolymer of GlcNAc and d-glucosamine (GlcN or D), which may be obtained by partial de-N-acetylation of chitin, and for which several applications exist [4]. The term chitosan refers collectively to water-soluble copolymers which are N-acetylated to different extents. Chitosans prepared by homogeneous de-N-acetylation of chitin have been found to have a random distribution of A and D units [5–7]. The successive sugar units in the chitin/chitosan chain are rotated 180° relative to each other. Thus, the functional and structural unit in these polymers is a disaccharide.

Despite its robust nature, its insolubility and its abundant production, chitin does not accumulate in most ecosystems, indicating that nature has developed effective processes for chitin degradation. Bacteria capable of degrading chitin usually produce a battery of chitinases [8–10]. consisting of a catalytic domain and, often, one or more smaller domains involved in substrate binding [11–14]. While several individual chitinases have been characterized in detail [15–18], little is known about important issues such as exo- vs. endo-action and processivity. This precludes full understanding of nature's chitinolytic machineries.

Serratia marcescens is one of the most intensively studied chitinolytic bacteria. When grown on chitin, S. marcescens produces three chitinases (ChiA, ChiB, and ChiC), a chitin-binding protein (Cbp21) lacking chitinase activity, and a hexosaminidase which further degrades the major end product of the chitinases, GlcNAc2[8,16,19–22]. The chitinase genes have been cloned from several S. marcescens strains by several research groups [17,20,23]. All three chitinases belong to the family 18 of glycosyl hydrolases [24], which possess a (β/α)8 barrel catalytic domain with approximately six sugar subsites [11,13,25–27]. Since chitin hydrolysis by family 18 chitinases directly involves the N-acetyl group of the sugar in the −1 subsite [28,29], productive substrate binding requires an N-acetylglucosamine to be bound in this subsite. Other subsites show less stringency in this respect, and may productively bind to, e.g. GlcN [30]. Indeed it has been shown that ChiB can degrade chitosans, even those with low degrees of acetylation [31].

The crystal structures of ChiA [11] and ChiB [13] revealed deep substrate-binding clefts, in part due to the presence of a 70–90-residue insertion in the catalytic domain (‘α + β domain’[11]), which is characteristic for these chitinases but which is absent in other chitinases (such as in hevamine and ChiC [25]; see below). Both ChiA and ChiB contain an additional substrate-binding domain which extents the substrate-binding cleft on the side where the nonreducing end of the substrate binds in ChiA [11] and on the side where the reducing end of the substrate binds in ChiB [13]. In ChiA the deep substrate-binding cleft seems rather accessible, whereas the substrate-binding cleft of ChiB is more ‘tunnel-like’[29,32]. On the basis of structural characteristics [11,13] and enzymological work [14,16,17,33] it has been suggested that ChiA and ChiB are exochitinases which degrade chitin chains from opposite ends, ChiA from the reducing end and ChiB from the nonreducing end (see Fig. 1).

Figure 1.

Schematic drawing of subsites, chitin binding domains and proposed orientation of polymer substrates in ChiA and ChiB. Fn3, Fibronectin type 3 domain; CBM5, chitin binding module. Dotted lines indicate that the polymer substrates are much longer than shown in the figure. Reducing sugars are shown in grey.

The structure of ChiC is not known, but its amino acid sequence shows that it consists of a catalytic domain and two putative chitin-binding domains, which are located C-terminally in the sequence [20]. The catalytic domain of ChiC lacks the α + β domain which makes up a wall in the substrate binding grooves of ChiA and ChiB. This suggests that ChiC has a much more open substrate-binding groove, as observed in the crystal structure of the endochitinase hevamine [25]. ChiC often occurs in two forms in cultures of S. marcescens: the complete protein, sometimes called ChiC1, and a proteolytically truncated variant, called ChiC2, which lacks the two putative chitin-binding domains [20,34].

Some enzymatic properties of ChiA, ChiB and ChiC have previously been studied and compared [16,17,33], showing, among other things, that these chitinases may act synergistically when present in certain combinations. Nevertheless, there still is limited insight into important properties of these enzymes (as concluded in recent studies by Suzuki et al. [16] and Hult et al. [33]). Remaining important issues concern the endo- or exo-character of the enzymes [17,33] and the possible occurrence of multiple attack mechanisms (processivity) [35,36]. Processive enzyme action has been studied extensively for cellulases [37–39] and has been suggested to occur in ChiA on the basis of the observation (by microscopy) that ChiA sharpened one end of β-chitin microfibrils [14]. In the present study, we have addressed these remaining issues, primarily by studies of the degradation of the water-soluble GlcNAc/GlcN heteropolymer chitosan, which provide insight into the exo- or endo- character of the enzymes as well as into the occurrence of processivity. In addition, we have compared the action of ChiA, ChiB and ChiC towards two types of chitin and GlcNAc6.


Degradation of hexamer (A6)

Figure 2 shows the initial time course of the degradation of A6 by ChiA, B and C. ChiA and ChiB degraded the hexamer at approximately the same speed, whereas ChiC reacted more slowly (approximately threefold; Table 1). For all enzymes the end products obtained upon prolonged incubation were A and A2 (results not shown; transglycosylation was never observed). Since the monomer is not expected nor has been observed to be produced directly from a tetramer, pentamer or hexamer ([15–17]; Horn and Eijsink, unpublished observations; Fig. 2) it must be derived from trimers produced during the reaction. Thus, productive binding of the hexamer to ChiA, ChiB or ChiC has three possible outcomes: A2+A4, A3+A3, or, if the enzymes operate processively, A2+A2+A2. The degradation patterns of ChiA and ChiB were very similar with A2 as the dominant product, whereas ChiC produced similar amounts of A2 and A4. The fact that A2 dominates very early in the reaction (i.e. when the concentration of A6, is much higher than that of the intermediate product A4), suggests that part of the A6 substrate is degraded processively, producing A2 + A2 + A2 as initial products.

Figure 2.

Product formation during degradation of GlcNAc6 with ChiA, ChiB and ChiC. The reaction mixtures contained identical amounts of substrate (100 µm) and enzyme (3.2 nm). The concentrations of the various products in the reaction mixtures were calculated from HPLC chromatograms, as described in Experimental procedures. ○, dimers; □, tetramers; •, hexamers; bsl00066, trimers; ▪, monomers.

Table 1.  Degradation of chitin with different chitinases.
EnzymeFinal A2/A3
Final A2/A3 ratio
Initial rate
(β-chitin)b (s−1)
Initial rate
(α-chitin)b (s−1)
Initial rate
[(GlcNAc)6] (s−1)
  1. a Molar A2/A3 ratio calculated from observed A and A2 concentrations as (A2-A)/A; see text. bFormation of chitobiose in the initial linear phase of degradation.


The percentages of A6 that were converted to A3 + A3 were calculated by dividing half the molar amount of A3 by the decrease in the molar amount of A6 concentration multiplied by 100. For ChiA and ChiB, after 5 min, these percentages were 21.2% and 21.7%, respectively. For ChiC the percentage was lower, amounting to 16.3% after 10 minutes. Thus, all three enzymes have a preference for producing dimers.

Degradation of chitin

As shown in Fig. 3A, degradation of β-chitin with ChiA and ChiB initially yielded A2 and small amounts of A and A3. ChiC also yielded A2 as the dominant product, but the relative amount of A3 was larger than for the other two chitinases. In addition, ChiC initially also yielded small amounts of A and A4 and very minor amounts of A5 (hardly detectable in Fig. 3A). As the degradation proceeded some of the oligosaccharides initially produced were degraded along with chitin, and the end products of all three enzymes were A and A2 (Fig. 3B).

Figure 3.

HPLC analysis of the degradation of β-chitin by ChiA, ChiB and ChiC. The reaction mixtures contained 360 nm of enzyme and 1 mg·mL−1β-chitin. (A) Initial products after 10 min. (B) Final products after 48 h; only monomers and dimers are observed. Note that each oligosaccharide produces two peaks, representing the α anomer and the β anomer; the α anomer elutes first.

Considering what is known about chitinases in general and previous experimental observations for Serratia chitinases [15–17], it is very unlikely that monomers are produced directly from chitin. Instead, they are produced when A3 is degraded to A2 and A. Thus, the molar amount of monomer seen after complete degradation of the substrate (Fig. 3B) may be taken to represent the cumulative molar amount of trimer formed during the degradation reaction. The molar amount of A2 directly produced from chitin (i.e. not from degradation of A3) equals the observed molar amount of A2 minus the observed molar amount of A. The (A2 − A)/A ratio (= A2/A3 ratio) is interesting because it can give information about substrate-binding modes and/or processivity [37,40], as discussed further below. Interestingly, the three enzymes show clear differences in their A2/A3 ratios, ChiB producing the highest ratio (Table 1).

Product formation curves for β-chitin degradation (Fig. 4A) had a short linear part, which permitted determination of initial degradation rates (Table 1). After prolonged incubation, no remaining β-chitin could be observed in the reaction mixtures of ChiA and ChiC. However, in the case of ChiB, which is only about twofold less effective than the other two enzymes in terms of initial rate (Table 1), complete conversion of the substrate was never obtained, not even after several weeks of incubation with repeated addition of excessive amounts of enzyme.

Figure 4.

Production of chitobiose during degradation of β-chitin (A) or α-chitin (B) by ChiA, B and C. The reaction mixtures contained 50 nm (A) or 360 nm (B) of ChiA (squares), ChiB (circles) or ChiC (triangles) and 0.1 mg·mL−1β-chitin (A) or 1 mg·mL−1α-chitin (B). The level of chitobiose obtained with ChiB on β-chitin after 215 h (A) is about 10% lower than the maximum level obtained after several weeks of incubation.

Degradation of α-chitin yielded similar mixtures of chito-oligosaccharides, both initially and at the end of the reaction, and yielded similar A2/A3 ratios as degradation of β-chitin (Table 1). Degradation of α-chitin occurred with slightly lower initial rates than degradation of β-chitin and none of the enzymes was capable of completely degrading the substrate. The fraction of the substrate degraded by ChiA was approximately three times higher than the fraction converted by ChiB or ChiC (Fig. 4B).

Degradation of chitosan

Reactions with chitosan were run for periods between 15 min and 1 week, as described in Experimental procedures. To verify the occurrence of enzyme depletion and/or product inhibition, reaction mixtures that had been incubated for several days and in which the reaction had proceeded to a supposedly final stage, were supplied with either new enzyme or new substrate, incubated, and analysed for product formation. These control reactions showed, for all three enzymes, that neither enzyme depletion nor product inhibition occurred under the conditions used in this study (data not shown).

The size distributions of oligomers obtained at various stages of degradation of a highly acetylated, high molecular weight and water-soluble chitosan (FA =0.65; Mn = 160 000) are shown in Fig. 5 (ChiA), Fig. 6 (ChiB) and Fig. 7 (ChiC). The results for ChiB have been described and discussed previously [31]. The stage of the reaction is characterized by the α value, which is the fraction of hydrolysed glycosidic bonds (α would be 0.5 if a long polymer were to be converted to dimers only).

Figure 5.

Size-distribution of oligomers during the degradation of chitosan (FA = 0.65) by ChiA. The peaks are marked by numbers which indicate the lengths (DP) of the oligomers they contain, or, in the case of peaks containing only one known compound, by the sequence of the oligomer. The annotation of the peaks is based on the use of standard samples, as well as NMR analyses (see Experimental procedures and [31]). The α-values denote the degree of scission; the lower panel represents the maximum obtainable α-value. Undegraded chitosan elutes in the void volume of the column, as do all chitosan fragments with a DP > 40. See Experimental procedures section for more details.

Figure 6.

Size distribution of oligomers during the degradation of chitosan (FA = 0.65) by ChiB. See legend to Fig. 4 for details.

Figure 7.

Size distribution of oligomers during the degradation of chitosan (FA = 0.65) by ChiC. See legend to Fig. 4 for details.

Figures 5–7 reveal clear differences between ChiC and the two other enzymes. The degradation by ChiC resulted in the disappearance of the void peak already at an α-value as low as 0.05, while the produced oligomers had lengths spanning the complete detectable range (2 to approximately 40; Fig. 7). This clearly indicates endo-activity. In contrast, in the reactions with ChiA and ChiB, the void peak did not disappear completely until the α-value was higher than 0.20 (Figs 5 and 6). This would be expected for enzymes that act in an exo-fashion and/or have a highly processive mode of action (see below). Most importantly, the initial degradation products obtained with ChiA and ChiB consisted almost exclusively of an even number of sugar units. This prooves that ChiA and ChiB act processively, as discussed below. Later in the degradation process, concomitantly with the depletion of chitosan, the longer oligomers initially released during processive action (see below) were degraded further, resulting in the production of odd-numbered oligomers. The final product mixtures obtained with ChiA and ChiB contained a continuum of odd- and even-numbered oligomers (see α = 0.35 and 0.38 in Figs 5 and 6, respectively), with the AA dimer as dominant product.

Comparison of Figs 5 and 6 reveals several differences between ChiA and ChiB. For example, in the initial phase of the degradation reaction (up to α =0.20) the oligomer fractions obtained with ChiB became larger with decreasing oligomer length (Fig. 6). Degradation with ChiA yielded a slightly different pattern: initially, the hexamer and octamer peaks had similar sizes, while the tetramer peak was much smaller. Also, ChiA produced less odd-numbered oligomers than ChiB in the initial phase of the reaction.

Using NMR, the composition and (partial) sequences of the oligomers present in the dimer to tetramer fractions obtained with the various enzymes were determined (Table 2; see Experimental procedures and [31] for a description of the methodology). The results show that in all cases the reducing ends of the oligomers consisted exclusively of A units, as expected for family 18 chitinases. In the case of ChiC, the sugar preceding the reducing end was almost exclusively an A, but a small amount DA dimers was formed towards the end of the reaction. In the case of ChiA and ChiB, D was observed more frequently, approaching 35% (i.e. as in the substrate) in the dimer fraction towards the end of the reaction. Thus, ChiA and ChiB have a preference for A in their −2 subsites, but this preference is less strong than in the case of ChiC. Principally, all three enzymes are capable of cleaving after –DA– if –AA– containing substrates become depleted. In the initial phases of the reactions, newly formed nonreducing ends had an A/D ratio of 65/35 (determined by carbon NMR, data not shown), which is the same ratio as in the substrate. Thus, none of the enzymes have preferences for A or D in the +1 subsite that are strong enough to be noticeable in these experiments.

Table 2.  Composition of dimer, trimer and tetramer fractions at different α during degradation of chitosan (FA = 0.65) by ChiA, B and C.
ChiA0.1581% AA81% DAA100% -AA
19% DA19% ADA 
 0.3564% AA51% DAA56% -AA
36% DA28% ADA
21% DDA
44% -DA
ChiB0.1186% AA71% DDA100% -AA
14% DA29% AAA 
 0.3866% AA95% DAA75% -AA
34% DA3% DDA
2% ADA
25% -DA
ChiC0.20100% AA66% DAA
34% AAA
100% -AA
 0.3881% AA100% DAA100% -AA
19% DA  


In the present study, we used natural substrates and chitosan to characterize the family 18 chitinases produced by S. marcescens. Studies with insoluble, resilient polymer substrates such as chitin (or cellulose) are intrinsically difficult because it is hard to analyse the substrate fraction and because intermediately formed soluble oligomers are much better substrates than the insoluble polymer. In the case of chitinases, these problems can be partly avoided by studying the degradation of chitosan.

Generally, enzymatic degradation of polysaccharides occurs from one of the chain ends (exo-mechanism) or from a random point along the polymer chain (endo-mechanism). Each of these two mechanisms can occur in combination with a processive mode of action, meaning that the substrate is not released after successful cleavage but slides through the active site for the next cleavage event to occur. Processivity reduces the search space for enzymes from 3D to 1D and is thought to be especially important when degrading insoluble substrates [32,41,42]. For example, processive cellobiohydrolases have deep, ‘tunnel-like’ substrate-binding clefts that are thought to embrace and processively hydrolyse a single polymer chain detached from the insoluble substrate [32,41,42]. It has been suggested that aromatic residues lining these substrate-binding clefts are important for the ‘sliding’ of the substrate through the cleft [41,43]. Like cellobiohydrolases, ChiA and ChiB have deep substrate-binding clefts lined with aromatic residues [11,13], suggesting that the two enzymes act processively. The experimental analysis of processivity is not straightforward, as discussed below and, e.g. in [44]. One approach is to study the shape of the substrate during enzymatic degradation by microscopy [14,33,38,45]; here, sharpening of the fibril tips is considered a sign of processive exo-action.

Because of the 180° rotation between consecutive sugar units, processive action on chitin will yield dimers, while trimers can only be produced by an exo-chitinase in the first hydrolytic step. Therefore, processivity may be assessed by studying the A2/A3 ratios in product mixtures [37,40]. One enzyme–substrate association event followed by processive degradation has two potential outcomes, depending on whether the initial product is a dimer or trimer: (1) production of X A2 or (2) production of 1 A3 and [X−1] A2, respectively. If these events have equal frequencies, the A2/A3 ratio will be X + [X−1] = 2X−1, giving X-values of 4.2, 6.8 and 2.6 for ChiA, ChiB and ChiC, respectively (derived from data in Table 1). However, it must be emphasized that A2/A3 ratios may also simply reflect different preferences for two initial binding modes, which release dimers or trimers, respectively. The experiments with A6 indicated that ChiA and ChiB have a preference for initial cleavage of a dimer. If this same preference would apply to the situation in which a polymer is degraded, the actual degree of processivity would be lower than what is suggested by the A2/A3 ratios in Table 1. In other words: a nonprocessive exo-enzyme with a strong preference for cleaving off dimers would also yield high A2/A3 ratios. Even a nonprocessive endo-enzyme could in principle yield high A2/A3 ratios if it would have a strong preference for releasing dimers from oligosaccharide intermediates. The latter seems to apply to ChiC: this enzyme produces a majority of dimers in chitin degradation experiments, while the chitosan experiments clearly show that this enzyme is not processive (see below); the experiments with A6 show that, indeed, ChiC preferably cleaves off dimers from short substrates. In conclusion, while the results obtained with natural substrates do indicate processivity, they do not lead to unequivocal conclusions.

Chitosan is an interesting substrate for gaining more insight into chitinase action, because the soluble substrate is easier to analyse than chitin. In addition, and most importantly, chitinases can bind nonproductively to chitosan (e.g. complexes that place a deacetylated sugar in the −1 subsite), which allows discrimination between independent binding events and processive binding events, as explained below.

Product mixtures obtained very early in reactions of the putative exoenzymes ChiA and ChiB with chitosan contained significant amounts of longer oligosaccharides with predominantly even-numbered chain lengths (note that only dimers and trimers are observed during degradation of chitin). Thus, both ChiA and ChiB are capable of productive binding events in which parts of the substrate extend from both sides of the active site cleft. For ChiB this is somewhat unexpected because association with substrate was originally thought to be sterically blocked in front of the −3 subsite [13]. Structural inspections and modelling studies (data not shown) using the available structure of a complex of ChiA with A8[46] superimposed on the ChiB structure indicate though that the sterical barrier in ChiB is not very strong and it is well conceivable that the chitosan chain can bend to the extent that it's association with the enzyme is not hampered by this putative barrier.

The observation that ChiA and ChiB almost exclusively produce even-numbered oligomers in the initial phase of the reaction with chitosan (a substrate having a random distribution of A and D units [5–7]), is of crucial importance since it provides unequivocal evidence for processivity. If upon substrate binding the sugar in subsite −1 is GlcNAc (A), ChiA and ChiB are in principle capable of hydrolysis and will cleave off an odd- or en even-numbered oligomer (for an exo-enzyme this would be a dimer or a trimer). However, if the −1 subsite contains a deacetylated unit (D), the enzyme cannot hydrolyse the substrate. If the enzyme would release its substrate after each productive or nonproductive binding event, the ratio between odd- and even-numbered longer oligomers would be close to 1 : 1, since the putative product sites consist of only two (ChiA, +1 and +2) or three (ChiB, −3 to −1) subsites (Fig. 1). Thus, the enzymes cannot discriminate between, e.g. a heptamer and an octamer in their product sites. If, however, the enzymes act processively, productive or nonproductive initial binding events would be followed by sliding of the substrate through the active site cleft by two sugar units at the time, until a new productive complex emerges and hydrolysis occurs. In such a mechanism the first product will be odd- or even-numbered, whereas all other products resulting from the same enzyme–substrate association event would be even-numbered. In the case of chitin, all these even-numbered products are dimers, whereas in the case of chitosan, these products are longer because part of the complexes formed during the processive movement are nonproductive. This also explains the somewhat counterintuitive observation that the putative exoenzymes ChiA and ChiB produce longer oligomers such as octa- and decamers in the very beginning of the reaction. In these oligomers, the sequence of A and D units apparently is such that productive complexes were only formed after, e.g. four (octamer) or five (decamer) processive ‘moves’ through the active site cleft. Most of these longer oligomers will still be cleavable (after rebinding) but only through binding modes that have not previously been explored during processive movement, i.e. binding modes that cleave the even-numbered oligomers into two odd-numbered products. This is exactly what is observed in the later phases of the reactions. Interestingly, the observation of longer even-numbered oligomers in the beginning of the reaction implies that the chitinases traverse stretches of unreactive polymer while moving from one productive complex to the other, a process sometimes referred to as ‘sliding’ or facilitated diffusion [36,47].

The A2/A3 ratios of Table 1 indicate that ChiB has the highest processivity when degrading the most natural substrate tested, chitin. In contrast, ChiA shows the strongest predominance of even-numbered oligomers in the reaction with chitosan (compare Figs 5 and 6), suggesting that this enzyme is the most processive one. It is possible that the suggested sterical hindrance beyond the −3 subsite in ChiB [13] limits processivity in the case of a chitosan substrate, since processive degradation of this substrate requires that complexes are formed in which more than two or three sugars are bound on the glycon side of the catalytic centre. Since the substrate-binding cleft in ChiB is more ‘tunnel’-like than in ChiA [13] one would a priori expect that ChiB is the more processive enzyme [32,39,41,48].

The degradation of β-chitin with ChiC revealed the presence of longer oligomer products (A4 and A5) in the initial phase of the reaction (such products have not been detected previously; see [16]). ChiC showed low activity towards A6 and no direct conversion of A6 to three A2 molecules. Degradation of β-chitin resulted in a low A2/A3 ratios (Table 1). Taken together, these observations suggest that ChiC is a nonprocessive endo-acting enzyme. The most compelling evidence for the ChiC reaction mechanism comes from the studies with chitosan. Figure 6 shows that ChiC converts chitosan to a continuum of oligomers of different sizes and that the polymer peak disappears early in the degradation reaction. Also, there is initially no accumulation of dimers or other even numbered oligomers.

The slow disappearance of the void peak in the reactions of ChiA and ChiB with chitosan seems to confirm previous suggestions that these two chitinases act in an exo-fashion. It should be noted though, that the chromatographic analyses of product patterns shown in Figs 5 and 6 cannot discriminate between processive exoenzymes and highly processive enzymes that initially attack the substrate in an endo-fashion. The latter type of enzyme would only produce even numbered oligomers directly from chitosan, but could produce odd-numbered oligomers (which are observed, see Figs 5 and 6) by reprocessing even numbered intermediate oligomeric products. It is known that some processive cellobiohydrolases occasionally bind the substrate in an endo-fashion, showing that the loops that form the ‘roof’ of the substrate-binding cleft may open occasionally [49–51]. Since the ‘roofs’ of ChiA and ChiB are rather open [11,13], it is conceivable that these enzymes also show occasional endo-binding [48].

Hult et al. [33] have recently used microscopy to study the degradation of β-chitin fibrils by ChiA and ChiB, and their results support the idea that these two chitinases are exo-acting enzymes, at least when hydrolysing insoluble chitin. Interestingly, by using microscopy to study the degradation of an end-labelled substrate, these authors were capable of showing that ChiA and ChiB degrade the chitin chains from the reducing and nonreducing ends, respectively, as previously suggested on the basis of the enzyme crystal structures [13].

While an A bound to subsite −1 is an absolute requirement for hydrolysis to occur, all three family 18 chitinases also showed a preference for an A in the −2 subsite (seen as a preference for producing oligomers with an AA at the reducing end; Table 2). This preference was strongest for ChiC where DA only appeared in the form of a dimer at very high α. We could not detect any A/D preference in the +1 subsites of the three enzymes.

Synergetic effects of the three S. marcescens chitinases have been shown with colloidal chitin [17], α-chitin [16] and β-chitin [33,52]. Taken together, the present and previous studies [16,33] show that the three chitinases have different and complementary activities (endo- vs. exo-) and directionalities, which can explain synergism. However, several observations remain unexplained, for example the observation that the exoenzyme ChiB and the endoenzyme ChiC show little, if any, synergy [16,52]. The recent finding that the chitin-binding protein (Cbp21) produced by S. marcescens potentiates chitinase action by disrupting the structure of the β-chitin substrate [52] points to another possible explanation for synergistic effects. The three S. marcescens chitinases have different chitin-binding domains (also called carbohydrate-binding modules, CBMs): ChiA contains a fibronectin type III (FnIII)-like CBM; ChiB contains a family 5 CBM and ChiC contains a family 12 and an FnIII-like CBM (see http://afmb.cnrs-mrs.fr/CAZY/for family classification). It is conceivable that the primary role of these CBMs is to potentiate catalytic activity by disrupting the substrate, rather than simply to promote enzyme–substrate binding. Synergistic effects could thus be due to one enzyme increasing substrate accessibility for other enzymes by hitherto unknown disruptive mechanisms involving the CBMs. A similar role has occasionally been proposed for cellulase-binding domains in cellulases [53,54]. Interestingly, Watanabe et al. have shown that deletion of the two Fn III domains of ChiA1 from Bacillus circulans did not affect chitin-binding, but strongly reduced chitin hydrolysing activity [12]. Differences in the ability to disrupt the substrate could also explain why the three chitinases have different abilities to fully convert α- and β-chitin (Fig. 4, see above), while displaying rather similar initial degradation rates (Table 1; these rates are likely to reflect degradation of amorphous, easily accessible regions in the substrate).

In conclusion, the chitinolytic machinery of S. marcescens consists of two processive exo-enzymes with different directionalities on chitin, ChiA and ChiB, and a nonprocessive endo enzyme, ChiC. Thus, when Serratia applies its battery of chitinases to degrade chitin, ChiC is likely to supply the exo-enzymes with new reducing and nonreducing ends, which are substrates for ChiA and ChiB, respectively. The disruptive effect of Cbp21 [52], and presumably of the noncatalytic domains of the chitinases, makes the crystalline regions of chitin more accessible for hydrolysis.

Experimental procedures


Squid pen β-chitin (3 µm in size) was from Seikagaku (Tokyo, Japan; product number 400627; average molecular weight 2 × 105 Da). Chitosan, with a degree of N-acetylation of 65% (FA = 0.65), was prepared by homogeneous N-deacetylation of milled (1.0 mm sieve) shrimp shell chitin [55]. This procedure results in a chitosan with a random sequence of acetylated and de-acetylated units [31]. Chito-oligosaccharides, α-chitin (product number C-3641) and all other chemicals were purchased from Sigma (St Louis, MO, USA).


The chitinase genes chia, chib from S. marcescens strain BJL200 were expressed in Escherichia coli DH5α (Life Technologies, Rockville, MD, USA) under control of their own promoters [17]. ChiA and ChiB were purified from periplasmatic extracts of early stationary phase cultures, essentially as described previously [13,17]. The extracts, in 0.65 mm MgCl2, 0.1 mm phenlymethylsulphonyl fluoride, 1 mm EDTA, were diluted 1.4-fold and adjusted to 20 mm Tris/HCl pH 8.0 and 0.4 m ammonium sulphate. Two millilitres of this dilution was loaded onto a phenyl-sepharose HR 5/5 column (5 × 50 mm) in an FPLC system, equilibrated in a buffer (20 mm Tris/HCl pH 8.0, 1 mm EDTA, 0.1 mm phenlymethylsulphonyl fluoride) containing 0.4 m ammonium sulphate. After loading the sample, the column was washed with the starting buffer followed by a 5 mL linear gradient of 0.4–0 m ammonium sulphate. Subsequently, a linear gradient of 0–6% (v/v) 2-propanol was applied to elute the enzyme.

The chic gene (EMBL database, accession no. AJ630582) was amplified from S. marcescens strain BL200 chromosomal DNA and cloned into the T7 promoter expression vector pRSETB (Invitrogen), using primer 5′-CGGGAATTCCATATGAGCACACAAATAACAC-3′ (NdeIChiC) to create an NdeI site for translational fusion, and a primer located downstream of the putative terminator in the chic gene [20]. The resulting plasmid (pRSETB-chic) was transformed into the E. coli strain BL21 (DE3) star (Invitrogen). For production of ChiC in E. coliBL21 (DE3) expression of chic was induced by 0.4 mm isopropyl-β-D-thiogalactoside when the OD600 was between 0.5 and 0.7. After growth in 150 min the cultures were harvested. ChiC was purified using the same protocol as described above for ChiA and ChiB[17; B. Synstad and V.G.H. Eijsink, unpublished results].

Enzyme purity was verified by SDS/PAGE and estimated to be > 95% in all cases (see [17] for an example). Protein concentrations were determined using the Bio-Rad Protein Assay (Bio-Rad Laboratories, Hercules, CA, USA), with BSA as standard.

Enzymatic degradation of chitin, GlcNAc6 and chitosan

For qualitative analysis of product formation, 1 mg·mL−1α- or β-chitin was hydrolysed at 37 °C in 50 mm sodium acetate buffer (pH 6.1) with 360 nm of ChiA, B or C. Reactions were stopped by adding 3 µL m HCl to 100 µL samples. The same conditions were used to estimate intitial rates for α-chitin degradation. Quantitative analysis of β-chitin degradation was conducted as described in [53]. Initial rates were estimated using the first linear part of the product formation curve.

Hydrolysis of GlcNAc6 was carried out in 50 mm sodium acetate with 50 µg·mL−1 BSA at pH 6.1 and 37 °C. The enzyme concentration was 3.2 nm for all three enzymes. All reactions were stopped by adding 3 µL 2 m HCl to 100 µL samples.

To analyse degradation of chitosan, 10 mg of the polymer (FA = 0.65), were dissolved in 1.0 mL H2O. After adding 1.0 mL buffer (0.08 m NaAc, 0.2 m NaCl, pH 5.5) and 0.2 mg BSA, the samples were immersed in a shaking water bath at 37.0 °C. The reactions were started by adding the enzyme (5 µg ChiA or ChiB, or 3 µg ChiC) and the reactions were allowed to proceed for 15 min to 1 week. Samples were taken at regular time intervals and reactions were stopped by lowering the pH to 2.5 by addition of 1.0 m HCl, and immersing the samples in boiling water for 2 min.

Chromatography of oligosaccharides

Mixtures of chito-oligosaccharides were analysed by HPLC using a Tosoh TSK Amide 80 column (0.46 × 25 cm) with an Amide 80 guard-column (Tosoh, Tokyo, Japan). A 10-µL sample was injected on the column and the chitin fragments were eluted isocratically at 0.7 mL·min−1 with 70% acetonitrile at room temperature. The chito-oligosaccharides were monitored by measuring absorbance at 210 nm and the amounts were quantified by measuring peak areas. Peak areas were compared to peak areas obtained with standard samples with known concentrations of chito-oligosaccharides. Using these standard samples, it was established that there was a linear correlation between peak area and oligosaccharide concentration within the concentration range used in this study, for each of the oligomers that were analysed.

Oligomers produced by enzymatic depolymerization of chitosan were separated on three XK 26 columns, packed with SuperdexTM 30, from Pharmacia Biotech (Uppsala, Sweden), with an overall dimension of 2.60 × 180 cm. The mobile phase was 0.15 m ammonium acetate, pH 4.50 and the flow rate was 0.80 mL·min−1. The relative amounts of oligomers were monitored with an online refractive index detector (Shimadzu RID 6 A), and the data were logged with a CR 510 Basic Data logger, from Campbell Scientific Inc (Logan, UH, USA). Fractions of 3.2 mL were collected and, where appropriate, pooled for analysis of the oligomers. This method and its performance have been described in detail previously [31]. It has been shown that this method allows the separation of mixtures of partially N-acetylated oligomers according to size (degree of polymerization, DPn), regardless of chemical composition, in the separation range between DP = 4 and a DP of approximately 20. Within the monomer–trimer range, some sequence specific separation was observed, as indicated in the figures presented. Studies with standard samples have shown that there is a linear relationship between peak areas and the amount (mass) of injected oligomer, irrespective of DP and degree of acetylation [31].

1H-NMR Spectroscopy

Proton NMR was used to partially sequence shorter oligomers and to calculate the DPn in the reaction mixtures as described previously [31]. The chitosan degradation is given as the degree of scission, α (= 1/DPn, where DPn is the number-average of the degree of polymerization), which represents the fraction of glycosidic linkages that has been cleaved. Complete conversion of the polymer to dimers would yield an α of 0.50.


This work was supported by grants from the Norwegian Research Council (140497/420 and 134674/ 110). We thank Xiaohong Jia for technical assistance.