Rate constants, processivity, and productive binding ratio of chitinase A revealed by single-molecule analysis

Akihiko Nakamura ab, Tomoyuki Tasaki c, Yasuko Okuni a, Chihong Song d, Kazuyoshi Murata d, Toshiya Kozai e, Mayu Hara c, Hayuki Sugimoto f, Kazushi Suzuki f, Takeshi Watanabe f, Takayuki Uchihashi g, Hiroyuki Noji c and Ryota Iino *abh
aOkazaki Institute for Integrative Bioscience, Institute for Molecular Science, National Institutes of Natural Sciences, Aichi 444-8787, Japan. E-mail: iino@ims.ac.jp; Tel: +81-564-59-5230
bDepartment of Functional Molecular Science, School of Physical Sciences, SOKENDAI, Kanagawa 240-0193, Japan
cDepartment of Applied Chemistry, Graduate School of Engineering, University of Tokyo, Tokyo 113-8656, Japan
dNational Institute for Physiological Sciences, National Institutes of Natural Sciences, Aichi 444-8787, Japan
eDepartment of Physics, Kanazawa University, Kanazawa 920-1192, Japan
fDepartment of Applied Biological Chemistry, Faculty of Agriculture, Niigata University, 8050 Ikarashi-2, Nishi-ku, Niigata 950-2181, Japan
gDepartment of Physics, Nagoya University, Aichi 464-8602, Japan
hInstitute for Molecular Science, National Institutes of Natural Sciences, Aichi 444-8787, Japan

Received 9th July 2017 , Accepted 24th October 2017

First published on 24th October 2017

Serratia marcescens chitinase A is a linear molecular motor that hydrolyses crystalline chitin in a processive manner. Here, we quantitatively determined the rate constants of elementary reaction steps, including binding (kon), translational movement (ktr), and dissociation (koff) with single-molecule fluorescence imaging. The kon for a single chitin microfibril was 2.1 × 109 M−1 μm−1 s−1. The koff showed two components, kfastoff (3.2 s−1, 78%) and kslowoff (0.38 s−1, 22%), corresponding to bindings to different crystal surfaces. From the kon, kfastoff, kslowoff and ratio of fast and slow dissociations, dissociation constants for low and high affinity sites were estimated as 2.0 × 10−9 M μm and 8.1 × 10−10 M μm, respectively. The ktr was 52.5 nm s−1, and processivity was estimated as 60.4. The apparent inconsistency between high turnover (52.5 s−1) calculated from ktr and biochemically determined low kcat (2.6 s−1) is explained by a low ratio (4.8%) of productive enzymes on the chitin surface (52.5 s−1 × 0.048 = 2.5 s−1). Our results highlight the importance of single-molecule analysis in understanding the mechanism of enzymes acting on a solid–liquid interface.


Chitin is the most abundant biomacromolecule on earth, next to cellulose. It is the main outer shell component of crabs, shrimp, and insects.1 The molecular chain of chitin is an N-acetyl glucosamine homo-polymer, which is a good source of nitrogen, making it useful for the chemical and pharmaceutical industries.2 However, modification and degradation of chitin requires high temperatures, or the use of strong acids and alkalis owing to its rigid crystalline packing.3 In nature, the bacterium Serratia marcescens degrades chitin under moderate conditions by secreting a set of chitin hydrolases (chitinases) and assisting enzymes.4 For example, S. marcescens chitinase A (SmChiA) and chitinase B (SmChiB) hydrolyse chitin from its reducing and non-reducing ends, respectively,5 and chitinase C (SmChiC) attacks the middle of the chain in the amorphous region.6 Furthermore, lytic polysaccharide monooxygenase (LPMO) generates chain ends on the crystalline surface by oxidative cleavage of glycosidic bonds.7

All three chitinases (SmChiA, B, and C) are classified into glycoside hydrolase family 18 (GH18) in the Carbohydrate-active enZYmes database (CAZY: http://www.cazy.org/), according to their amino acid sequences.8 SmChiA consists of a catalytic domain (CD) and a carbohydrate-binding domain (CBD), and X-ray crystal structures have been solved (Fig. 1).9 Aromatic amino acid residues (Trp33-Trp69-Phe232-Trp245) of the CBD and CD align and form a flat surface, important for binding to the crystalline chitin surface and threading of a single chitin chain into the catalytic site.10 The CD has a cleft-like substrate binding site, and amino acid residues important for catalysis (Glu315 and Asp313) are located in the deep inside. The X-ray crystal structures of SmChiB11 and SmChiC12 have been also solved. Their CDs show similar TIM-barrel folds to SmChiA, while the catalytic clefts of SmChiA and SmChiB are deeper than that of SmChiC.12 Furthermore, the substrate binding sites of SmChiA and SmChiB are extended from opposite sides of each other by their CBDs, which are composed of a line of aromatic residues.5 Therefore, SmChiA and SmChiB were expected to be processive enzymes, which hydrolyse the polymer chain of chitin continuously without dissociation, while SmChiC was expected to be a non-processive enzyme.

image file: c7cp04606e-f1.tif
Fig. 1 Crystal structure of SmChiA (pdb ID: 1CTN) and amino acid residues mutated in this study.

Recently, the processive movements of SmChiA and SmChiB on crystalline chitin were directly observed by high-speed atomic force microscopy (HS-AFM), and their opposite directionalities were proven.13 The average translational rate constant (ktr) of SmChiA and SmChiB was 70.5 nm s−1 and 46.9 nm s−1, respectively, and the values of processivity (how many times the enzyme hydrolyses the substrate before dissociation) were calculated as 29.8 and 18.1 from the rate constant and moving time (0.42 s and 0.39 s for SmChiA and SmChiB, respectively). Although the value of processivity estimated by HS-AFM for SmChiA was similar to the value expected from a biochemical assay,14,15 the ktr of a single SmChiA molecule was much faster than that expected from a biochemical assay.6,14 One of the reasons for this gap is reaction heterogeneity on the crystalline chitin surface. Chitinase reacts on a solid–liquid interface, because chitin is a water-insoluble substrate. The chitinase reaction cycle includes chitin binding, processive polymer chain hydrolysis, and dissociation from the surface. Furthermore, the number of available chain ends on the crystalline chitin surface is limited, and a significant fraction of binding events can be non-productive. Therefore, to fully understand the chitinase reaction cycle, we should consider not only the kinetic parameters for elementary reaction steps but also the productive binding ratio.

In the present study, we directly observed binding, processive movement, and dissociation of SmChiA on the crystalline chitin surface through single-molecule fluorescence imaging. We successfully determined all kinetic parameters, including binding rate constant (kon), ktr, and dissociation rate constant (koff). For further understanding of binding modes of SmChiA to the crystalline chitin, we also analysed the number of microfibrils in a bundle of crystalline chitin, the shape of a single microfibril, and the kon and koff of an inactive mutant SmChiA(D313A) and a less-binding mutant SmChiA(W33A-W69A-F232A-W245A) (Fig. 1). Furthermore, we found that the productive binding ratio accompanying the translational movement is actually very low, explaining the apparent low hydrolysis activity of SmChiA in the biochemical assay.

Experimental procedures

Preparation of enzymes and crystalline chitin

The SmChiA expression plasmid (pNCA112) was essentially the same as described in a previous report,4 but a C-terminal histidine-6 tag (His6) was attached to SmChiA for Ni-NTA affinity purification using PCR and fragment swapping with the restriction enzymes AsiSI and HpaI. The free cysteine residue at D415C for fluorescent dye labelling was also introduced by the same procedure. SmChiA(D415C) production was performed as described in a previous report,16 using E. coli (BL21) as a host. Ten grams of harvested cells were suspended in 100 ml of 100 mM Tris–HCl (pH 8.0) and sonicated with a tablet of Complete Mini, EDTA-free Protease Inhibitor Cocktail (Roche). Supernatant (100 ml) was collected after centrifuging at 8000g and 30[thin space (1/6-em)]000g, and 2.5 ml of 4 M NaCl and 5 ml of Ni-NTA super flow (QIAGEN) were added to the solution and incubated for 5 min at 25 ± 2 °C. The Ni-NTA resin was packed into a reservoir and washed with 50 mM sodium phosphate buffer (pH 7.0) containing 100 mM NaCl with 0 and 50 mM imidazole. The enzyme was eluted with 100 and 200 mM imidazole, and DTT was added to fractions at a final concentration of 10 mM to prevent disulphide bond formation between free cysteines. The eluted enzyme solution was concentrated to 500 μl by a 10 kDa cut VIVASPIN Turbo 15 (Sartorius). Then, the sample was loaded into a Superdex 200 10/300GL (GE Healthcare) and eluted with 50 mM sodium phosphate buffer (pH 7.0) containing 100 mM NaCl. Eluted fractions were mixed and the SmChiA concentration was calculated from the absorbance at 280 nm; the enzyme extinction coefficient was ε280 = 107[thin space (1/6-em)]050 M−1 cm−1. Cy3-maleimide (GE Healthcare) dissolved in DMSO was added to the enzyme at the same molar concentration as the enzyme and incubated for 1 h at RT. After the labelling reaction, unreacted Cy3-maleimide was removed with an NAP-10 column (GE Healthcare). The enzyme labelling ratio was calculated from the absorbance at 280 nm and 550 nm; the extinction coefficient of the enzyme described above, and Cy3-maleimide (ε280 = 12[thin space (1/6-em)]000 M−1 cm−1 and ε550 = 150[thin space (1/6-em)]000 M−1 cm−1). SmChiA(D313A-D415C) and SmChiA(W33A-W69A-F232A-W245A-D415C) were purified and labelled by the same methods, but ε280 = 88[thin space (1/6-em)]430 M−1 cm−1 was used for SmChiA(W33A-W69A-F232A-W245A-D415C). Wild-type SmChiA-His6 was also purified by the same method and a portion of enzyme was reacted with Cy3-maleimide as a negative control. Crystalline β-chitin was purified from tubes of L. satsuma using the same procedure described in a previous report.13

Biochemical assay of hydrolysis activity

Thirty-seven and a half microliters of 100 nM chitinase in 50 mM sodium phosphate buffer (pH 6.0) and 75 μl of 0.5 mg ml−1 crystalline chitin suspension were mixed and incubated at 37 °C for 30 min, after a 3 min pre-incubation at 37 °C. To stop the reaction, 150 μl of alkaline ferricyanide solution [500 mM Na2CO3, 1.5 mM K3(Fe(CN)6)] was added to the mixture, and centrifuged at 10[thin space (1/6-em)]000g for 15 min at 4 °C. The supernatant was boiled at 98 °C for 15 min and the absorbance at 420 nm was measured.17 Product concentration and chitinase turnover were calculated using the N-acetyl glucosamine standard curve. Measurements were performed in triplicate. For hydrolysis activity measurement against a water-soluble substrate p-nitrophenol chitobioside (pNP-chitobioside) (Sigma-Aldrich), 70 μM of pNP-chitobioside was reacted with 500 nM enzyme in 50 mM sodium phosphate buffer (pH 6.0) at 37 °C for 10 min. Twenty microliters of 2 M Na2CO3 solution was added to 200 μl of reaction mixture and the absorbance at 405 nm was measured. The product concentration and SmChiA turnover were calculated using p-nitrophenol as a standard. Measurements were performed in triplicate.

SmChiA turnover (Fig. 6) was determined by the same procedure, but hydrolysis activity at various crystalline chitin concentrations (final concentration: 0.10 to 2.0 mg ml−1) was measured at 25 °C. Plots were fitted by the Michaelis–Menten equation v/E0 = kcat·[S]/(Km + [S]), where kcat is the turnover (s−1), E0 is the enzyme concentration (μM), [S] is the concentration of crystalline chitin (mg ml−1) and Km is the Michaelis constant (mg ml−1).

HS-AFM observation of crystalline chitin

Images of crystalline chitin were obtained by laboratory-built HS-AFM13 with a 5 μm × 5 μm field of view resolved in 500 × 500 pixels at 0.5 fps. Two microliters of 0.7% crystalline chitin suspension was placed on a mica stage hydrophobized with Fluoro Surf (FS-1010Z-0.5, Fluoro Technology, Japan), and unbound chitin was washed with 200 μl of 50 mM sodium phosphate (pH 6.0). The heights of a bundle of chitin microfibrils were measured and averaged along the long axis. The distribution of the averaged height was fitted by Gaussian functions. Largely aggregated bundles were not used for analysis.

Cryo-electron microscopic (cryo-EM) tomography of crystalline chitin

Two microliters of 0.7% crystalline chitin suspension was applied to an R1.2/1.3 Quantifoil grid (Quantifoil Micro Tools, Germany) pretreated with glow-discharge beforehand. Colloidal gold (15 nm) was mixed with the sample as a fiducial marker. The grid was blotted for 10 s and quick frozen in liquid ethane using a Vitrobot Mark IV (FEI Company, USA) set at 95% humidity and a temperature of 4 °C. The frozen grid was mounted on the side-entry Gatan 914 cryo-specimen holder (Gatan Inc., USA) and examined with a JEM2200FS electron microscope (JEOL Inc., Japan) equipped with a field-emission gun operated at 200 kV accelerating voltage and an in-column energy filter operated in a zero-energy loss mode with a slit width of 15 eV. A tilt series of crystalline chitin was collected in the range of −62° and +62° with a 2° increment step using a low dose mode, where the total electron dose was less than 100 e−2 on the specimen. Images were recorded on a DE20 direct-detector CMOS camera (Direct Electron LP. USA) with a detector magnification of 17[thin space (1/6-em)]877 corresponding to 3.66 Å per pixel. Image alignment and tomographic reconstruction were performed by IMOD software18 by using fiducial markers. The final tomograms were calculated by the SIRT algorithm using two binning images of 7.32 Å per pixel. Volume rendering was performed in the 3dmod viewer of IMOD software.

Single-molecule fluorescence imaging analysis

Coverslips used for single-molecule imaging were cleaned with ethanol and 10 M potassium hydroxide to remove contaminants on the glass surface.19 To prevent non-specific enzyme binding, 20 μl of 5 mg ml−1 BSA solution filtered through a 0.22 μm mesh was spin-coated on the coverslip. Then, 20 μl of 0.1 mg ml−1 crystalline chitin suspension was spin-coated on the coverslip, and placed on the microscope stage. Twenty microliters of 12.5 pM SmChiA(D415C-Cy3), SmChiA(D313A-D415C-Cy3) or 125 pM SmChiA(W33A-W69A-F232A-W245A-D415C-Cy3) in 50 mM sodium phosphate (pH 6.0) was dropped on the coverslip and fluorescence images of single molecules were recorded at 4 fps for kon and koff analysis, and at 3 fps for ktr analysis with a laser power of 0.14 μW μm−2. After observation, 10 μl of 10 nM SmChiA(D415C-Cy3) was dropped on the coverslip to visualise crystalline chitin.

To minimize stage drift during single-molecule observation, an ultra-stable microscopic stage (Ikeda-rika, Japan) was used. Furthermore, before the detailed analysis of image sequences, we checked the trajectory of non-moving molecules (an example is shown in Fig. 4) to confirm no apparent drift. If image sequences showed drift, they were not used for analysis. The photo-bleaching time for Cy3 was 20 s, and localization precisions in the x and y directions were 8.6 nm and 8.4 nm, respectively, at 3 fps with a laser power of 0.14 μW μm−2. Therefore, the moving distance threshold for productive molecules was set to 34 nm. Additionally, the moving time threshold was set to 1.3 s (4 frames) to extract the unidirectional movement of the same molecule. The values of kon on a single microfibril of crystalline chitin were calculated as the number of bound molecules normalized by enzyme concentration, length of microfibril, and observation time. Binding events were counted for 40 s after focusing. The lengths of the microfibrils were measured with fluorescence images of crystalline chitin stained with high concentrations of SmChiA(D415C-Cy3). Images were analysed using ImageJ software. The distributions of kon were fitted by triple or double Gaussian functions. The binding time distributions were fitted by the double exponential decay functions: y = a·exp(−bt) + c·exp(−dt), in which a, b, c and d are fitting parameters. The ktr of the moving molecule was calculated from the moving distance and moving time, and the distribution was fitted by a Gaussian function. The processivity was estimated from the obtained moving distance constant on the assumption that the SmChiA step size is 1 nm. The productive binding ratio was estimated from the ratio of the moving time sum and total molecule binding time in the initial 40 s movies. Highly aggregated chitins were not analysed to reduce the effect of steric hindrance between substrates.

Results and discussion

For single-molecule imaging, a free cysteine residue was introduced into wild-type SmChiA(D313A) and SmChiA(W33A-W69A-F232A-W245A) mutants for conjugation with a fluorescent dye, Cy3. We selected Asp415 because this residue is located at the protein surface and does not interact with the chitin surface (Fig. 1). After reacting with Cy3-maleimide, the labelling ratios for SmChiA(D415C), SmChiA(D313A-D415C), and SmChiA(W33A-W69A-F232A-W245A-D415C) were 108%, 90.7%, and 99.2%, respectively, and much higher than that for wild-type (2.3%) (Table 1). These results indicate that Cy3 was conjugated almost entirely at D415C, and other residues did not react. Against water-soluble substrate pNP-chitobioside, wild-type, SmChiA(D415C-Cy3) and SmChiA(W33A-W69A-F232A-W245A-D415C-Cy3) showed similar turnovers (in pH 6.0 at 37 °C), but that of SmChiA(D313A-D415C-Cy3) was almost one hundred times smaller than the others (Table 1). Furthermore, the hydrolysis activities of wild-type and SmChiA(D415C-Cy3) against crystalline chitin were almost identical, indicating that Cy3 labelling had no adverse effect on SmChiA function. On the other hand, the hydrolysis activity of SmChiA(D313A-D415C-Cy3) was almost zero, and that of SmChiA(W33A-W69A-F232A-W245A-D415C-Cy3) was much lower than that of wild-type as reported previously.10 Thus, SmChiA(D313A-D415C-Cy3) is actually inactive, and SmChiA(W33A-W69A-F232A-W245A-D415C-Cy3) largely decreases hydrolysis activity only against crystalline chitin. Hereafter, for simplicity, we refer to SmChiA(D415C-Cy3), SmChiA(D313A-D415C-Cy3), and SmChiA(W33A-W69A-F232A-W245A-D415C-Cy3) as wild-type, inactive, and less-binding, respectively.
Table 1 Labelling ratio and hydrolysis activity of enzymes against soluble substrate and crystalline chitin
Enzyme Labelling ratio (%) Turnovera (s−1)
pNP-Chitobioside Crystalline chitin
a Turnovers were measured at pH 6.0 at 37 °C.


2.3 1.8 × 10−2 ± 0.03 × 10−2 5.0 ± 0.03



108 1.9 × 10−2 ± 0.04 × 10−2 5.3 ± 0.03




90.7 2.4 × 10−4 ± 0.1 × 10−4 0.10 ± 0.02




99.2 2.0 × 10−2 ± 0.09 × 10−2 0.24 ± 0.01

Next, using HS-AFM and cryo-EM tomography, we analysed the size, shape, and number of microfibrils in a bundle of crystalline chitin used in this study prepared from tubeworm (Lamellibrachia satsuma). We analysed our sample without drying, because the crystal structures of the β-chitin from tubeworm were different between hydrated and dehydrated samples.20 In HS-AFM observation, single and bundled forms of chitin microfibrils were observed (Fig. 2A), and we estimated the number of microfibrils from the heights of the images. The distribution of heights averaged on the long axis of the microfibril showed quantized peaks (Fig. 2B). The height of a single microfibril was estimated as 21.7 nm. By cryo-EM tomography, microfibrils with slightly different widths were observed (Fig. 2C). To determine the shape, the cross sections of the microfibrils were analysed (Fig. 2D). Thin crystals showed the cross section of a single microfibril, and thick crystals were resolved as a bundled form of single microfibrils. The shape of a single microfibril is shown as a parallelogram, of which one sharper edge was rounded. From the lengths of the sides and angles of the edges of tomography images, a possible structure of a single chitin microfibril was constructed from the unit cell structure of hydrated crystalline chitin20 (Fig. 2E). The longer sides at the top and bottom were hydrophobic planes formed by the rings of N-acetyl glucosamines, and the other three are hydrophilic surfaces with hydroxyl and acetyl groups. The distance between two hydrophobic planes is 20.3 nm (= sin[thin space (1/6-em)]84° × 27.7 nm) and agrees with the height of a single microfibril observed by HS-AFM.

image file: c7cp04606e-f2.tif
Fig. 2 Observation of crystalline chitin with HS-AFM and cryo-EM tomography. (A) Typical images and heights of crystalline chitin in HS-AFM observation. (B) Distribution of height of microfibril bundles. (C) Slice of image by cryo-EM tomography. (D) Expanded cryo-EM tomography images. Cross sections are shown at the bottom. (E) Structural model of a single microfibril of crystalline chitin.

Then, we determined kon of wild-type, inactive and less-binding mutants on crystalline chitin (Fig. 3A and Table 2). The kon was defined as the number of chitinase molecules bound on the chitin in a unit of chitinase concentration, chitin length, and time (M−1 μm−1 s−1), as previously reported.19 The measured kon distributions showed multiple peaks, as did our previous measurements for processive cellulases TrCel7A19 and TrCel6A21 hydrolysing crystalline cellulose. We attributed multiple peaks of kon to multiple chitin microfibrils in a bundle as shown in Fig. 2B, and the distributions of kon were fitted by multiple Gaussian functions. The peak positions of each Gaussian function for wild-type were 2.1 ± 0.13 × 109, 6.8 ± 0.23 × 109, and 9.1 ± 0.96 × 109 M−1 μm−1 s−1. The minimum peak at 2.1 × 109 M−1 μm−1 s−1 was considered to be the kon for a single chitin microfibril. This value is similar to that of TrCel7A19 (2.4 × 109 M−1 μm−1 s−1), and thrice as high as that of TrCel6A21 (7.5 × 108 M−1 μm−1 s−1) on crystalline cellulose. The peak positions of Gaussian functions for inactive were 2.7 ± 0.075 × 109 and 7.5 ± 0.49 × 109 M−1 μm−1 s−1, and those for less-binding were 0.48 ± 0.018 × 109 and 0.92 ± 0.024 × 109 M−1 μm−1 s−1. Therefore, the kon for inactive was similar to that of wild-type, but that for less-binding was less than one fourth of that of wild-type. These results indicate that the four aromatic residues (Trp33, Trp69, Phe232 and Trp245) are important for initial recognition of the chitin surface, but Asp313 has less contribution.

image file: c7cp04606e-f3.tif
Fig. 3 Distributions of kon and binding time for wild-type, inactive and less-binding on crystalline chitin. Binding and dissociation events were observed with 12.5 pM wild-type and inactive, or 125 pM less-binding in 50 mM Na-phosphate buffer (pH 6.0) at 25 °C.
Table 2 Summary of kon, koff and Kd for wild-type, inactive, and less-binding to crystalline chitin
Sample k on

(M−1 μm−1 s−1)

Binding mode


k fastoff or kslowoff


k fast-offon or kslow-offon[thin space (1/6-em)]a,b

(M−1 μm−1 s−1)

K d

(M μm)

a The unit of μm represents the length of a crystalline chitin microfibril. b The konfast-off and konslow-off were calculated from the ratio of fast and slow dissociations, respectively.
Wild-type 2.1 × 109 ± 1.3 × 108 Fast dissociation


3.2 ± 0.15 1.6 × 109 2.0 × 10−9
Slow dissociation


0.38 ± 0.11 4.7 × 108 8.1 × 10−10
Inactive 2.7 × 109 ± 7.5 × 107 Fast dissociation


4.4 ± 0.20 2.3 × 109 1.9 × 10−9
Slow dissociation


1.1 ± 0.31 3.9 × 108 2.9 × 10−9
Less-binding 4.8 × 108 ± 1.8 × 107 Fast dissociation


4.1 ± 0.11 3.9 × 108 1.0 × 10−8
Slow dissociation


0.53 ± 0.087 8.5 × 107 6.2 × 10−9

Next, we determined koff from the distribution of binding time on chitin (Fig. 3B and Table 2). The distribution for wild-type was fitted by double exponential decay functions with rate constants of 3.2 ± 0.15 s−1 (kfastoff) and 0.38 ± 0.11 s−1 (kslowoff). The ratio of the two binding modes (77.5% for fast and 22.5% for slow dissociations) was calculated from the area of the fitting curves. This result indicates that SmChiA binds to crystalline chitin through at least two different modes. Similar to wild-type, binding time distributions for inactive and less-binding were also fitted by double exponential decay functions. The kfastoff and kslowoff for inactive were 4.4 ± 0.20 s−1 (85.5%) and 1.1 ± 0.31 s−1 (14.5%), and those for less-binding were 4.1 ± 0.11 s−1 (88.2%) and 0.53 ± 0.087 s−1 (17.8%), respectively. These two binding modes can be attributed to binding to different crystalline chitin surfaces shown in Fig. 2E. In the previous TrCel6A study,21 CBD-Cy3 similarly showed two binding modes to the intact enzyme, indicating that two modes can be caused even in a single domain structure. In the present study, two binding modes were observed in wild-type, inactive, and less-binding, consistent with the notion that two binding modes are caused by the substrate, not by the properties of the enzyme. The less-binding mutant showed slightly higher kslowoff than that for wild-type. On the other hand, kslowoff for inactive was 2.9 times higher than that for wild-type. Thus, Asp313 contributes to slow dissociation. In contrast, the kfastoff values for wild-type, inactive, and less-binding were similar. This result suggests that the binding mode accompanying fast dissociation is less specific, without interactions between chitin and catalytic residues/aromatic residues on the enzyme surface; like binding to the hydrophilic surface of chitin with charged residues on the enzyme surface.

Then, we defined the two components of kon as binding events accompanying the fast and slow dissociations. The values of kon for each binding mode (konfast-off and konslow-off) were estimated from the ratio of fast and slow dissociation (Table 2). From these values, the dissociation constants Kd (= koff/kon) for the two binding modes were calculated. The Kd of low and high affinity sites for wild-type were 2.0 × 10−9 and 8.1 × 10−10 M μm, respectively (Table 2). The values of Kd for less-binding were largely increased (1.0 × 10−8 and 6.2 × 10−9 M μm). Interestingly, in inactive, Kd corresponding to slow dissociation was slightly higher (2.9 × 10−9 M μm) than that of fast dissociation (1.9 × 10−9 M μm) due to the large increase of kslowoff. SmChiA has a flat binding surface with aromatic residues (Trp33, Trp69, Phe232 and Trp245), which are important for initial binding to the crystal surface and threading of a single chitin chain into the catalytic site.10 This flat and hydrophobic surface is also common to SmChiB10 and other chitin and cellulose binding proteins.22 The Kd values of slow dissociation for less-binding and inactive were 7.7 and 3.6 times larger than that of wild-type. These results imply that the affinity of SmChiA for the hydrophobic plane of the chitin crystal (Fig. 2E) is increased not only by hydrophobic residues but also by Asp313 in the catalytic site. The catalytic site of SmChiA has a cleft-like structure and is more opened than that of TrCel6A which has a tunnel-like structure covered by a loop region. Therefore, the binding mode of SmChiA can be more affected by catalytic residues than TrCel6A. On the other hand, most of the wild-type molecules which showed slow dissociation did not undergo unidirectional translational movements accompanying processive hydrolysis (see below and Table 3). One possible explanation of slow dissociation without translational movement is that the end of the chain interacts with Asp313 but does not form the Michaelis-complex. To form a productive Michaelis-complex, at least seven successive N-acetyl glucosamine residues of the chitin chain should be decrystallized from the crystalline surface and threaded into the catalytic site. On the highly crystalline chitin, the frequency of the formation of the productive Michaelis-complex can be very low.

Table 3 Ratio of productive binding for wild-type on crystalline chitin
Image sequence # of chitin Total non-moving timea (s) Total moving timea (s) Ratio (%)
a Total binding times of non-moving and moving molecules in image sequences for 40 s were analysed.
#1 6 43.7 2.7 6.2
#2 11 102.7 2.3 2.2
#3 16 130 7.7 5.8
Average ± S.D. (%) 4.8 ± 2.2

In addition to binding and dissociation, some wild-type molecules showed unidirectional translational movements on the chitin surface. On the other hand, inactive and less-active mutants did not, presumably because hydrolysis activity and stable binding on the chitin surface are required for unidirectional translational movements. An example of the movement trajectory of wild-type is shown in Fig. 4. The wild-type moves in one direction along the long axis of crystalline chitin. The distribution of ktr for unidirectionally-moved wild-type is shown in Fig. 5A, and fitted by a Gaussian function with a peak at 52.7 ± 2.0 nm s−1. The peak value of ktr is 1.3 times lower than that measured by HS-AFM (70.5 nm s−1),13 but there is no significant difference due to the broad distribution. The distribution of moving distance was then fitted by the single exponential decay function with a constant of 60.4 ± 11.1 nm (Fig. 5B). The first bin was removed from fitting because the processive movement with a short moving distance (<34 nm) was difficult to measure reliably. The value of the processivity was calculated as 60.4, when the step size was assumed to be 1 nm, as determined by the size of the reaction product, chitobiose (1.0 nm). The value of processivity as estimated from the HS-AFM was 29, almost half of that in our study.13 In other words, the moving time obtained by our single-molecule fluorescence imaging (1.2 ± 0.28 s, Fig. 5C) was 2.8 times longer than that by HS-AFM (0.41 s). In our single-molecule fluorescence imaging, we could not distinguish between the moving and non-moving molecules reliably when the moving distance was shorter than the threshold value of 34 nm (4 times the localization precision, see Experimental procedures). This is one possible reason for the longer processivity determined in our study. Another possible reason is the effect of the HS-AFM tapping force. In the case of cellulase TrCel6A, the processive movement was observed only with single-molecule fluorescence imaging,21 and not observed with HS-AFM.23 These results imply that the tapping force can affect the enzyme processivity, resulting in shorter moving distance. Single-molecule fluorescence imaging with localization precision better than the present study, and/or HS-AFM observation with lower tapping force are required to distinguish between these two possibilities.

image file: c7cp04606e-f4.tif
Fig. 4 Example of the trajectory of wild-type. A moving molecule was observed at 3 fps. The centroid of the fluorescent spot was analysed for each frame.

image file: c7cp04606e-f5.tif
Fig. 5 Distributions of ktr, moving distance and moving time for wild-type. The ktr was calculated from the moving distance and moving time. First bins of the moving distance and moving time were not included for fitting.

Finally, the turnover (kcat) of wild-type against crystalline chitin was biochemically measured by changing the substrate concentration under the same conditions as those for single-molecule fluorescence imaging analysis (Fig. 6). The kcat was 2.6 s−1, and the Michaelis constant was 0.13 mg ml−1. At saturated substrate concentration, binding between wild-type and substrate is not a rate-limiting step of the hydrolysis reaction, and virtually all enzymes were bound on the crystalline chitin. If all bound enzymes undergo translational movement accompanying hydrolysis, the kcat would be 52.7 s−1, as estimated from the ktr (52.7 nm s−1) in single-molecule analysis (Fig. 5A). Therefore, there is an apparent inconsistency between the single-molecule and biochemical analyses. Possible reasons for low kcat in the biochemical assay are decrease of ktr of individual molecules due to surface crowding (traffic jam),23 and/or low ratio of productive binding. However, the former possibility is not plausible because the concentration of the substrate is much higher than that of the enzyme, and the density of the enzyme on the chitin surface will be very low. Therefore, this apparent inconsistency can be explained by the low ratio of productive binding. To confirm this contention, we then quantitatively estimated the ratio of productive (active) molecules in all chitin-bound wild-type (Table 3). The ratio of total moving time was calculated from the number of moving and non-moving wild-type molecules and the binding time of each wild-type molecule. The average ratio value and standard deviation, as calculated from three different image sequences, was 4.8 ± 2.2% (Table 3). If all enzymes were bound on chitin and the ratio of active molecule is 4.8%, the apparent turnover will be 52.7 s−1 × 0.048 = 2.5 s−1. This value is almost identical to the kcat (2.6 s−1) estimated biochemically (Fig. 6). In a previous biochemical study, Kuusk et al. reported that the productive enzyme turnover was 105 ± 10 s−1, by estimating the productive enzyme ratio using 4-methyl-umbelliferyl-β-D-N,N′-diacetylchitobioside to indicate enzymes with a free active-site.24 This value is consistent with the ktr for SmChiA estimated by HS-AFM13 and single-molecule fluorescence imaging. They also showed that active enzyme in the reaction mixture is 11% at the reaction start. Thus, the ratios of productive molecules are very low in both highly crystalline chitin from Lamellibrachia satsuma (this study) and chitin from crab shell.24

image file: c7cp04606e-f6.tif
Fig. 6 Michaelis–Menten curve for hydrolysis of crystalline chitin by wild-type. Turnovers were measured in 50 mM Na-phosphate buffer (pH 6.0) at 25 °C, the same conditions used for single-molecule imaging analysis.

The results of this study explain why S. marcescens produces not only processive SmChiA and SmChiB, but also SmChiC and LPMO which generates the chain ends on the chitin surface. To achieve efficient crystalline chitin hydrolysis, an enzyme cocktail with different exo- and endo-type catalytic modes will be required. It is well-known that a mixture of these enzymes shows a synergistic effect with SmChiA.25 Understanding the mechanism underlying this synergistic effect is a target for our next single-molecule analysis.


The elementary rate constants of SmChiA, the kon, ktr, and koff, and processivity were successfully determined through single-molecule fluorescence imaging. The low kcat estimated by biochemical assay can be explained by the low proportion of productive SmChiA binding on crystalline chitin. The results in this study directly revealed the complex nature of the heterogeneous reaction catalysed by SmChiA at the solid–liquid interface. For SmChiA, cooperative actions with endo-type SmChiC and LPMO will be required to efficiently bind to the chain end and to increase the hydrolytic activity.

Author contributions

RI conceived and supervised the project. TT and AN carried out single-molecule fluorescence imaging and analysis. HN contributed to instrumentation of single-molecule fluorescence microscopy. AN, YO, and MH prepared samples. CS and KM carried out cryo-EM tomography and analysis. AN, TK, and TU carried out HS-AFM imaging and analysis. HS, KS, and TW provided genes encoding SmChiA. RI and AN wrote the manuscript.

Conflicts of interest

There are no conflicts to declare.


This study was supported by Grants-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science, and Technology, Japan (JP15H04366, JP16H00789, JP16H00858, JP17K19213 to R. I., JP15H06898, JP17K18429, and JP17H05899 to A. N.), Advanced Bioimaging Support (ABiS) from Grant-in-Aid for Scientific Research on Innovative Areas (JP16H06280 to A. N.), Advanced Technology Institute Research Grants 2015 (RG2709 to A. N.), and the ORION project (10341630611) of Okazaki Institute for Integrative Bioscience, National Institutes of Natural Sciences, Japan (to R. I.). The tubes of L. satsuma were kindly gifted by S. Kimura.

Notes and references

  1. M. Rhazi, J. Desbrieres, A. Tolaimate, A. Alagui and P. Vottero, Polym. Int., 2000, 49, 337–344 CrossRef CAS .
  2. N. Yan and X. Chen, Nature, 2015, 524, 155–157 CrossRef CAS PubMed .
  3. A. Einbu and K. M. Vårum, Biomacromolecules, 2008, 9, 1870–1875 CrossRef CAS PubMed .
  4. K. Suzuki, N. Sugawara, M. Suzuki, T. Uchiyama, F. Katouno, N. Nikaidou and T. Watanabe, Biosci., Biotechnol., Biochem., 2002, 66, 1075–1083 CrossRef CAS PubMed .
  5. E.-L. Hult, F. Katouno, T. Uchiyama, T. Watanabe and J. Sugiyama, Biochem. J., 2005, 388, 851–856 CrossRef CAS PubMed .
  6. S. J. Horn, M. Sørlie, G. Vaaje-Kolstad, A. L. Norberg, B. Synstad, K. M. Vårum and V. G. H. Eijsink, Biocatal. Biotransform., 2009, 24, 39–53 CrossRef .
  7. G. Vaaje-Kolstad, B. Westereng, S. J. Horn, Z. Liu, H. Zhai, M. Sørlie and V. G. H. Eijsink, Science, 2010, 330, 219–222 CrossRef CAS PubMed .
  8. V. Lombard, H. Golaconda Ramulu, E. Drula, P. M. Coutinho and B. Henrissat, Nucleic Acids Res., 2013, 42, D490–D495 CrossRef PubMed .
  9. A. Perrakis, I. Tews, Z. Dauter, A. B. Oppenheim, I. Chet, K. S. Wilson and C. E. Vorgias, Structure, 1994, 2, 1169–1180 CrossRef CAS PubMed .
  10. T. Uchiyama, F. Katouno, N. Nikaidou, T. Nonaka, J. Sugiyama and T. Watanabe, J. Biol. Chem., 2001, 276, 41343–41349 CrossRef CAS PubMed .
  11. D. M. F. van Aalten, B. Synstad, M. B. Brurberg, E. Hough, B. W. Riise, V. G. H. Eijsink and R. K. Wierenga, Proc. Natl. Acad. Sci. U. S. A., 2000, 249, 5842–5847 CrossRef .
  12. C. M. Payne, J. Baban, S. J. Horn, P. H. Backe, A. S. Arvai, B. Dalhus, M. Bjoras, V. G. H. Eijsink, M. Sørlie, G. T. Beckham and G. Vaaje-Kolstad, J. Biol. Chem., 2012, 287, 36322–36330 CrossRef CAS PubMed .
  13. K. Igarashi, T. Uchihashi, T. Uchiyama, H. Sugimoto, M. Wada, K. Suzuki, S. Sakuda, T. Ando, T. Watanabe and M. Samejima, Nat. Commun., 2014, 5, 1–7 Search PubMed .
  14. K. Mihhail, K. Silja, K. Piret, S. Morten and V. Priit, J. Biol. Chem., 2015, 290, 29074–29085 CrossRef PubMed .
  15. A. G. Hamre, S. Jana, N. K. Reppert, C. M. Payne and M. Sørlie, Biochemistry, 2015, 54, 7292–7306 CrossRef CAS PubMed .
  16. K. Suzuki, M. Suzuki, M. Taiyoji, N. Nikaidou and T. Watanabe, Biosci., Biotechnol., Biochem., 1998, 62, 128–135 CrossRef CAS PubMed .
  17. J. F. Robyt, R. J. Ackerman and J. G. Keng, Anal. Biochem., 1972, 45, 517–524 CrossRef CAS PubMed .
  18. J. R. Kremer, D. N. Mastronarde and J. R. McIntosh, J. Struct. Biol., 1996, 116, 71–76 CrossRef CAS PubMed .
  19. Y. Shibafuji, A. Nakamura, T. Uchihashi, N. Sugimoto, S. Fukuda, H. Watanabe, M. Samejima, T. Ando, H. Noji, A. Koivula, K. Igarashi and R. Iino, J. Biol. Chem., 2014, 289, 14056–14065 CrossRef CAS PubMed .
  20. D. Sawada, Y. Nishiyama, P. Langan, V. T. Forsyth, S. Kimura and M. Wada, PLoS One, 2012, 7, e39376 CAS .
  21. A. Nakamura, T. Tasaki, D. Ishiwata, M. Yamamoto, Y. Okuni, A. Visootsat, M. Maximilien, H. Noji, T. Uchiyama, M. Samejima, K. Igarashi and R. Iino, J. Biol. Chem., 2016, 291, 22404–22413 CrossRef CAS PubMed .
  22. A. B. Boraston, D. N. Bolam, H. J. Gilbert and G. J. Davies, Biochem. J., 2004, 382, 769 CrossRef CAS PubMed .
  23. K. Igarashi, T. Uchihashi, A. Koivula, M. Wada, S. Kimura, T. Okamoto, M. Penttila, T. Ando and M. Samejima, Science, 2011, 333, 1279–1282 CrossRef CAS PubMed .
  24. K. Silja, S. Morten and V. Priit, J. Biol. Chem., 2015, 290, 11678 CrossRef PubMed .
  25. G. Vaaje-Kolstad, S. J. Horn, D. M. F. van Aalten, B. Synstad and V. G. H. Eijsink, J. Biol. Chem., 2005, 280, 28492–28497 CrossRef CAS PubMed .

This journal is © the Owner Societies 2018