A diverse range of bacterial and eukaryotic chitinases hydrolyzes the LacNAc (Gal1-4GlcNAc) and LacdiNAc (GalNAc1-4GlcNAc) motifs found on vertebrate and insect cells

range of bacterial and eukaryotic chitinases hydrolyzes the LacNAc (

still unknown, especially in organisms that rarely come into contact with chitin.
One eukaryotic example is chitinases encoded by cellular slime molds, such as Dictyostelium spp. and Polysphondylium spp. These organisms utilize mainly bacteria as a food source, and consumption of chitin-containing yeast supports growth only poorly. There are also examples of plant chitinases that have defensive physiological roles that are attributed to ribosome inactivating activity (6,7). Humans produce two active chitinases, chitotriosidase and acidic mammalian chitinase (AMCase), that can be linked to a variety of disorders including lysosomal storage disease, asthma, and neurological diseases (8 -11). Finally, it is now emerging that for several bacterial pathogens and symbionts, chitinases may be important as virulence or symbiotic factors as much as, or rather than, being important for chitin metabolism. Examples include enzymes produced by Legionella pneumophila, Listeria monocytogenes, and possibly Salmonella Typhimurium (12)(13)(14). Understanding the exact role of chitinases in infectious processes is, however, limited by the fact that their target substrates in the host remain unidentified.
The gene expression of Salmonella Typhimurium chitinase, StChiA, is up-regulated during infection of macrophage and epithelial cells (15,16). We recently found that the catalytic domain of this enzyme (StChiA⌬CBM) cleaves N-acetyllactosamine (LacNAc, 5 (Gal␤1-4GlcNAc)) from the tetramethylrhodamine (TMR, see Fig. 1) fluorescently labeled substrate analog LacNAc-TMR in addition to hydrolyzing chitin, chitooligosaccharides, and the chitin pseudo-substrate pNP-(GlcNAc) 2 (14). Hydrolysis occurs with retention of configuration and is consistent with the TMR-linker occupying the ϩ subsites of the enzyme with the Gal and GlcNAc of LacNAc in the Ϫ2 and Ϫ1 subsites, respectively. This is analogous to the hydrolysis of pNP-(GlcNAc) 2 , where p-nitrophenyl (pNP) occupies the ϩ1 subsite and (GlcNAc) 2 is bound to the Ϫ2 and Ϫ1 subsites. Because N-acetyllactosamine is a common component of mammalian glycans, it constitutes a possible chitinase target. We have also shown that human chitotriosidase cleaves LacNAc and LacdiNAc (GalNAc␤1-4GlcNAc) from LacNAc-TMR and LacdiNAc-TMR, respectively, with the latter having a turnover comparable to that of pNP-chitotrioside (17).
In this study we have cloned and characterized chitinases, which are phylogenetically related to the Salmonella enzyme StChiA, from the bacterial insect endosymbiont Sodalis glossinidius and the eukaryotic slime mold Polysphondylium pallidum (see Table 1 and Fig. 2). These three enzymes share the common characteristic that they lack a signal peptide but are still predicted to be secretory proteins (14).
For comparison we have included the phylogenetically unrelated chitinases LmChiA and LmChiB (18) from L. monocytogenes (see Table 1 and Fig. 2) and the commercially available chitinases from the actinobacterium Streptomyces griseus and the mold Trichoderma viride. The chitinases included in the present study all belong to the CAZy family 18 glycoside hydrolases, although it should be noted that the two commercial available chitinases from T. viride and S. griseus remain uncharacterized in this regard.
We observed that the Salmonella chitinase StChiA was able to bind to potential glycan targets in hosts by use of a carbohydrate microarray. In addition, we found that all enzymes surveyed (with one exception) hydrolyzed LacdiNAc from fluorescently labeled conjugates, whereas several of the enzymes hydrolyzed LacNAc from analogous conjugates that are substrate analogs of glycans found in mammals.

EXPERIMENTAL PROCEDURES
Phylogenetic Analysis-The amino acid sequences of Salmonella Typhimurium SL1344 StChiA (SL0018), S. glossinidius str. morsitans SgChiA (SG1474), P. pallidum Ppchitinase (EFA83839), L. monocytogenes LmChiA (lmo1883), and LmChiB (lmo0105) ( Table 1) were from the GenBank TM database of the National Center for Biotechnology Information (www.ncbi.nlm.nih.gov/). Protein domain predictions including determination of the glycosyl hydrolase family 18 (GH_18) domains were obtained from the SMART website. Signal peptide predictions were performed using SignalP 4.0 from the Centre for Biological Sequence Analysis, Biocentrum, Technical University of Denmark. GH_18 domains of sequences representative of bacterial chitinases in addition to representatives of eukaryotic chitinases were included in the phylogenetic anal- ysis. The phylogenetic tree (see Fig. 2) based on the corresponding GH_18 domains was constructed by the neighbor-joining method using ClustalX Version 1.81 (19) and the Mega version 4 (20) software packages with default settings. Production and Purification of Chitinases-The nucleotide sequences of the chitinase genes SG1474 of S. glossinidius str. morsitans and EFA83839 of P. pallidum PN500 were codonoptimized for expression in Escherichia coli, and the genes were synthesized with an N-terminal His 6 tag flanked by Ncol/Xhol restriction sites (Epoch Biolabs, Missouri City, TX). The synthetic genes were cloned into pET15b expression vectors, which were transformed into chemically competent E. coli TOP10 cells (Invitrogen) for long term storage. For protein expression the vector carrying SG1474 was transformed into E. coli BL21(DE3) cells (Novagen, Darmstadt, Germany), whereas the vector carrying EFA83839 was transformed into E. coli BL21/TUNER TM (DE3) cells (Novagen). 10 ng of pET-15b-SG1474 or EFA83839 were added to 50 l of E. coli TOP10, E. coli BL21(DE3), or E. coli BL21/Tuner TM (DE3) cells. The reaction mixture was incubated on ice for 30 min, heat-shocked at 42°C for 30 s, and incubated on ice for an additional 2 min, then 200 l of Luria broth (LB; Merck) was added. The reaction mixture was incubated at 37°C for 60 min with shaking and plated on LB agar plates containing 100 g/ml carbenicillin and incubated overnight at 37°C. Transformants were re-streaked on carbenicillin-containing LB plates. The following day colonies were grown in LB and stored as glycerol stocks.
The L. monocytogenes EGDe chiB gene (lmo0105) and a truncated version of chiB (lmo1883⌬CBM) lacking the C-terminal carbohydrate binding domain (amino acids 711-755) were amplified through PCR with primers designed to exclude the respective signal peptide sequences but including a His 6 tag at the N terminus. The PCR products were cloned into the pET-46 Ek/LIC vector and introduced into E. coli BL21(DE3) according to the manufacturer's instructions (Novagen pET-46 E/LIC vector kit). The absence of mis-and nonsense mutations was confirmed by sequencing (Macrogen, Seoul, Korea). The E. coli BL21(DE3) strain used for expression of the L. monocytogenes chitinase, LmChiA (lmo1883), was from a previous study (21). An E. coli BL21(DE3) strain carrying a Salmonella chitinase gene without the chitin-binding domain (SL0018⌬CBM) and used for expression of StChiA⌬CBM was prepared in a previous study (14). Expression vectors for the production of Salmonella wild type chitinase, StChiA, and the inactive enzyme variants StChiA E223Q and StChiA E223Q⌬CBM were purchased from Epoch life sciences Inc. (Missouri City, TX). In brief, StChiA E223Q was obtained by substituting the active site glutamate with a glutamine residue by site-directed mutagenesis (E223Q, GAA3 CAA), whereas ChiA E223Q⌬CBM was obtained by a further truncation from amino acid 646. The synthetic genes were cloned into NdeI/XhoI restriction sites of the pETb15 expression vector carrying a His-tag coding sequence. The resulting vectors was transformed into E. coli BL21(DE3) as previously described.
For protein purification, up to 1 liter of E. coli BL21(DE3) cells were grown at 30°C to an A 600 of 0.4 before induction with 1 mM isopropyl 1-thio-␤-D-galactopyranoside. After induction, growth was continued for 21 h at 30°C. For improved recovery and purification of the Polysphondylium chitinase, PpChitinase, E. coli BL21/Tuner TM (DE3) cells were grown at 37°C to an A 600 of 0.6 and placed on ice to cool before induction with 0.3 mM isopropyl 1-thio-␤-D-galactopyranoside. After induction, growth was continued for 21 h at 22°C. The E. coli BL21(DE3) and/or E. coli BL21/Tuner cells were harvested by centrifugation and resuspended in 10 volumes of loading buffer, 20 mM MOPS, pH 7.2, containing 0.5 M NaCl and 5 mM imidazole. The cells were disrupted using a Constant Systems cell disruptor at 4°C with a pressure of 1.36 kilobars. The lysate was centrifuged at 4°C for 1.5 h at 48,000 ϫ g, and the filtered supernatant was applied to a 1-or 5-ml HisTrap HP column (GE Healthcare) with a flow rate of 1 or 5 ml/min, respectively. For the Listeria enzymes, 1-2-ml columns of nickel-nitrilotriacetic acid-agarose (Qiagen) were used with a flow rate of 1 ml/min. All columns were washed with 100 ml of loading buffer before being eluted with 30 ml of 100 mM MOPS, pH 7.8, 0.5 M NaCl, and 0.5 M imidazole. For the Sodalis chitinase SgChiA and the Salmonella StChiA⌬CBM, fractions containing enzyme were combined, the buffer was exchanged to loading buffer by dialysis, and the enzymes were rechromatographed on a 1-ml HisTrap column with an imidazole gradient from 5 to 250 mM imidazole to obtain ultrapure enzymes. For all enzymes, the fractions containing enzyme were dialyzed against 50 mM sodium phosphate buffer, pH 6.0, at 4°C and concentrated in a Vivaspin (10,000 Da cutoff for LmChiA and 30,000-Da cutoff for other enzymes). The protein concentration of the sample was determined by the Bradford method using a commercial kit (Bio-Rad) with bovine ␥-globulin as a protein reference standard. Alternatively, protein concentration was measured with Qubit 2.0 fluorometer (Invitrogen) according to the manufacturer's protocol. Commercially available enzymes of S. griseus (Sigma) and T. viride (a mixture of two chitinases; Sigma) were dissolved according to manufacturer's instructions and included in the study as reference enzymes.
For the enzymes from Salmonella, Sodalis, Polysphondylium, and Listeria, purified in this study, 12-73 mg of recombinant protein were obtained per liter of cell culture after chromatography on HisTrap columns. The purity of proteins after isolation assessed by SDS-PAGE was as found previously (14).
Examination of Enzyme Activity-A survey of the activity of chitinases was initially carried out with the pNP chitin pseudosubstrates: pNP-GlcNAc, pNP-(GlcNAc) 2 , pNP-(GlcNAc) 3 , and the cellulose pseudo-substrate 4-nitrophenyl-␤-D-cellobioside whose structures are shown in Fig. 1. In the survey, 5 l of enzyme (40 -2000 g/ml) diluted in 50 mM sodium phosphate buffer, pH 6.0, and 45 l of 1.8 mM substrate dissolved in 50 mM sodium phosphate buffer, pH 6.0, were incubated at 30°C. To limit the conversion of substrates to a maximum of 15-20%, the following enzyme concentrations were used: SgChiA (400 g/ml), StChiA⌬CBM (200 g/ml), PpChitinase (100 g/ml), LmChiA (40 g/ml), LmChiB (200 g/ml), LmChiB⌬CBM (200 g/ml), T. viride mixture (67 g/ml) and S. griseus (400 g/ml). 50-l samples were removed after 30 min, and the reaction was quenched by adding 250 l of 0.4 M sodium carbonate (Sigma S2127). Absorbance at 405 nm was measured in a plate reader and corrected for absorption using a control sample with added sodium phosphate buffer, pH 6.0, instead of enzyme. After the initial screening, pNP-(GlcNAc) 2 absorption values were converted into concentrations by the use of a p-nitrophenol (Sigma) standard curve. Conversion rates were normalized against protein concentration and are listed as pmol/min/mg of protein.
Kinetic parameters for pNP-(GlcNAc) 2 were determined for SgChiA (23 g/ml) and PpChitinase (20 g/ml) by measuring initial rates of reaction at eight different substrate concentrations ranging from 0.007 to 1.8 mM during 25-120 min of incubation at 30°C. K m and V max were calculated using the software GraphPad Prism 4.0 (GraphPad Software), and k cat was estimated by dividing V max by enzyme concentration.
The activity of chitinases toward hyaluronic acid (Sigma, H5388) was tested in a standard enzymatic assay of hyaluronidase (Sigma) using hyaluronidase (bovine testes, Sigma H3506) as a positive control. Samples were prepared according to the manufacturer's instructions and incubated at 37°C for up to 40 h, and absorbance was read as % transmittance (%T) at 600 nm.
Kinetic parameters for LacNAc-TMR and LacdiNAc-TMR were determined for Sodalis, Salmonella, and Polysphondylium enzymes by monitoring product formation with 8 different concentrations of substrate ranging from 0.080 to 10.5 mM at 30°C by capillary electrophoresis. The reaction volume was 10 l in 50 mM sodium phosphate buffer, pH 6.0, and enzyme concentrations were 0.6 -2.5 mg/ml. Aliquots (1 l) were removed at 12-120 min and quenched with 50 l of CE running buffer as described above. Analyses were performed as described above. Separations were carried out in an uncoated fused-silica capillary of 75-m inner diameter with CE running buffer. TMR-labeled compounds were detected and quantitated using an Argos 250B fluorescence detector (Flux Instruments, Basel, Switzerland) equipped with an excitation filter of 546.1/10 nm and an emission filter of 570 nm. All experiments were carried out at a normal polarity, i.e. inlet anodic. Data were processed with the PrinCE 7.0 software. Ultra-HPLC-mass spectrometry analysis was employed to screen extended substrates terminating in LacNAc including LacNAc␤1-6LacNAc␤-TMR, LacNAc␤1-6(Gal␤1-3)GalNAc␣-TMR (extended core 2), and LacNAc␤1-2-Man-TMR substrates ( Fig. 1), which are only available in limited amounts (26,27). Assays were carried out in 4 l containing 50 mM sodium phosphate buffer, pH 6.0, 1 g of each enzyme, and 5 pmol of TMR-substrate. After overnight incubation at room temperature, a 1-l aliquot was removed, mixed with 300 l of buffer (10 mM ammonium formate, pH 4.5 (22%), acetonitrile (88%) (v/v)) for analysis by ultra-HPLC-MS as previously described (17).
Nuclear Magnetic Resonance-Substrates, either pNP-(GlcNAc) 2 , (GlcNAc) 3 , LacNAc-TMR, or LacdiNAc-TMR (0.4 -1 mg), were dissolved in 0.7 ml of buffer, transferred to 5-ml nuclear magnetic resonance (NMR) tubes, and standard one-dimensional 1 H NMR spectra of substrates were acquired before and after the addition of 5-50 l of undiluted SgChiA and StChiA⌬CBM (LacdiNAc-TMR only). NMR spectra were recorded at 15 or 25°C on a Bruker Avance 800 instrument and analyzed as previously described (14). Deuterated buffer was prepared by lyophilizing 5 ml of 50 mM sodium phosphate buffer, pH 6.0, followed by the addition of 2 ml of 2 H 2 O to the dried solids, re-lyophilization, and suspension in 5 ml of 2 H 2 O. Spectra were recorded at 799.3 MHz using tetramethylsilane (␦ ϭ 0 ppm) as the internal standard with a 32 scan composite presaturation.

RESULTS
Phylogenetic Analyses-A phylogenetic analysis of selected family 18 glycosyl hydrolase catalytic domains revealed that they cluster roughly into three distinct groups (colored in red, green, and blue, respectively, in Fig. 2). The groups colored in green and blue fit into Cluster A according to the analysis of Karlsson and Stenlid (28). Catalytic domains in these two groups in general carry a chitin-insertion domain conferring a deep substrate binding cleft, which suggests exo-activity (29). The group colored in blue contains StChiA, SgChiA, and PpChitinase from Salmonella, Sodalis, and Polysphondylium, respectively, and is characterized by a lack of signal peptides, yet all are predicted to be secretory proteins. On the other hand, the green group containing LmChiB from Listeria and also human chitinases in general carry signal peptides. The group colored in red (Fig. 2) containing LmChiA fit into group IV in cluster B according to Karlsson and Stenlid (28). The red group is characterized by the lack of a chitin-insertion domain, which confers a shallow substrate binding cleft suggesting endo-activity (29).
Glycan Array Scanning-The glycan array scanning was performed to screen for ligand binding by the full-length active Salmonella StChiA enzyme as well as inactive variants with or without the chitin binding domain. Average background signal was below 1000 relative fluorescence units (RFU), and average signals above 3000 RFU were considered as positives. The glycan array scanning showed binding of the inactive Salmonella chitinases (StChiA E223Q and StChiA E223Q⌬CBM) to GlcNAc-containing glycans including the chitin oligosaccha-rides chitopentaose and chitohexaose (glycan no. 493 and 503; Table 2, supplemental Table 1). The active enzyme (StChiA) did not bind to the latter two compounds, unlike the inactive variants, showing that hydrolytic activity interfered with glycan binding. The active enzyme also bound to a range of non-chi-tinous substrates with binding to five LacNAc-containing compounds (compounds No. 384, 385, 420, 498, 229; Table 2,  supplemental Table 1) as well as TF (Thomsen-Friedenreich antigen Gal␤1-3GalNAc) antigen and TF ␤ antigen structures (glycan number 88 and 89, respectively; Table 2 and supple-  Ϫ, average signal is below 3000 relative fluorescence units (RFU); ϩ, average signal is Ͼ3000 RFU; ϩϩ, average signal is Ͼ10,000 RFU; ϩϩϩ, average signal is Ͼ15,000 RFU. Table 1). The inactive enzymes only bound to some of these substrates (compound numbers 384, 385, 420, and 498) suggesting that changes in the active site also affected binding.
Substrate Specificity and Kinetic Parameters-The activity spectrum of chitinases was initially screened by the use of the chitin pseudo-substrates pNP-GlcNAc, pNP-(GlcNAc) 2 , pNP-(GlcNAc) 3 , and pNP-cellobioside, the mammalian and insect model substrates LacNAc-TMR and LacdiNAc-TMR, Type I-TMR (Gal␤1-3GlcNAc-TMR isomer of LacNAc-TMR), and hyaluronic acid ( Table 3). All of the tested chitinases demonstrated activity toward the chitin analogs pNP-(GlcNAc) 2 and pNP-(GlcNAc) 3 . None of the chitinases tested was active against pNP-cellobioside, and only the reference enzyme T. viride chitinase mixture demonstrated activity toward pNP-GlcNAc and hyaluronic acid. All chitinases except the S. griseus chitinase were to varying degrees active toward the substrate LacdiNAc-TMR. The StChiA⌬CBM and SgChiA and to some extent also the LmChiB chitinase were active against LacNAc-TMR. The latter results were confirmed by use of capillary electrophoresis except for LmChiB (Tables 4 and 5). None of the chitinases were active against the isomeric Type I-TMR (Table 3).
Conversion rates in pmol/min/mg of protein were determined for all chitinases toward pNP-(GlcNAc) 2 , LacNAc-TMR, and LacdiNAc-TMR, and clear differences among the chitinases were observed. For pNP-(GlcNAc) 2 , the LmChiA was the fastest converter (after the reference T. viride chitinase) followed by StChiA⌬CBM. The Salmonella enzyme StChiA⌬CBM was the fastest converter of LacdiNAc-TMR (Table 4).
We further studied the kinetics of the three phylogenetically related enzymes. Both StChiA⌬CBM and SgChiA hydrolyzed LacNAc-TMR and LacdiNAc-TMR with greater activity toward LacdiNAc-TMR for both enzymes as seen by the higher k cat value (Table 5, Fig. 3). StChiA⌬CBM had a k cat for Lacdi-NAc-TMR of 0.18 s Ϫ1 , which is 25% that of the k cat for the chitinase substrate pNP-(GlcNAc) 2 , whereas k cat /K m was 20% that of pNP-(GlcNAc) 2 . The k cat for LacNAc-TMR hydrolysis was 0.046 s Ϫ1 about 7% of pNP-(GlcNAc) 2 . For SgChiA, the k cat value for LacdiNAc-TMR was 0.035 s Ϫ1 compared with a k cat value of 0.015 s Ϫ1 for LacNAc-TMR hydrolysis. The PpChitinase also showed some activity against LacdiNAc-TMR, although it was slower than the StChiA⌬CBM and SgChiA as seen from the low k cat value of 0.007 s Ϫ1 . On the other hand, the PpChitinase was more active against pNP-(GlcNAc) 2 than StChiA⌬CBM and SgChiA as seen from the high k cat value of 10.3 s Ϫ1 compared with 0.7 and 3.1 s Ϫ1 for StChiA⌬CBM and SgChiA, respectively ( Table 5).
Hydrolysis of larger substrates with terminal LacNAc (Fig. 1) was monitored by ultra-HPLC-MS due to the limited amounts of material available. Only LacNAc␤1-6LacNAc␤-TMR was a substrate. It was hydrolyzed by stepwise removal of terminal LacNAc forming LacNAc-TMR disaccharide. Subsequent hydrolysis of LacNAc-TMR can yield the TMR linker (Fig. 4B). The LacNAc disaccharides that are cleaved are not detected by the fluorescence detector used in our analysis. The Sodalis SgChiA enzyme was the most efficient with respect to hydrolysis with essentially complete conversion to a mixture of LacNAc-TMR (65%) and the TMR-linker (33%). LmChiB was the least efficient enzyme with conversion to 3% each of Lac-NAc-TMR and the TMR-linker. LmChiA showed no activity with this extended substrate, whereas the StChiA⌬CBM enzyme only removed the terminal LacNAc giving 62% Lac-NAc-TMR (Fig. 4, A and C).
Nuclear Magnetic Resonance Studies of Sodalis and Salmonella Chitinases with pNP-(GlcNAc) 2 , pNP-(GlcNAc) 3 , LacNAc-TMR, and LacdiNAc-TMR-LacdiNAc-TMR and LacNAc-TMR were shown to be substrates of SgChiA and StChiA⌬CBM with release of LacdiNAc and LacNAc, respectively, as moni-   tored by 1 H NMR. LacdiNAc consists of a GalNAc and a GlcNAc unit, and these two units gave distinct anomeric signals in 1 H NMR as can be seen in Fig. 5, which shows the hydrolysis of LacdiNAc-TMR by SgChiA. From the bottom, the first spectrum (Fig. 5A) shows the substrate before adding enzyme with the peak at 4.18 ppm from H1 of GlcNAc and the peak at 4.33 ppm from H1 of GalNAc. The spectra B, C, and D show the formation of the product, LacdiNAc, at 1, 3.5, and 9 h, respectively, after the addition of the enzyme. The peak at 4.37 ppm is the signal from H1 of ␤-GalNAc, when the reducing end is ␤-GlcNAc, and the peak at 4.38 ppm is the signal from H1 of ␤-GalNAc when the reducing end is ␣-GlcNAc. The signal at 4.55 ppm is the H1 of ␤-GlcNAc reducing end, and the signal at 5.19 ppm is the H1 of ␣-GlcNAc reducing end. Note that the ␣-anomer is formed later than the reducing end ␤-anomer thus confirming that the reaction occurs with retention of configu-ration as the reducing end ␤-anomer is the only product of enzyme-catalyzed steps.

DISCUSSION
We have investigated the affinity of a Salmonella chitinase toward a large range of substrates by use of a glycan microarray technique followed by investigations of conversion rates of   selected substrates by several chitinases and, finally, by screening a small selection of potential biological substrates. The glycan array scanning highlighted a specific affinity of the full-length Salmonella chitinase, StChiA, toward various GlcNAc-containing glycans. As expected, chitin-like substrates such as chitopentaose and chitohexaose were targeted by this chitinase. Some GlcNAc-containing glycans of mammalian-like LacNAc structures were also targets ( Table 2). LacNAc structures have been identified as terminating glycans of glycoproteins and glycolipids on vertebrate cells and in the human glycome (30,31). The inclusion of the inactive Salmonella variants StChiA E223Q and StChiA E223Q⌬CBM in the study ensured that a reliable overview of binding affinities was obtained.
Detailed analyses of hydrolysis of the most promising of these mammalian and/or insect-like substrates were examined by use of a wider range of bacterial and eukaryotic chitinases. The chitinases were selected from a broad phylogenetic range and included the related (Fig. 2) enzymes from the Enterobacteriaceae Salmonella Typhimurium and S. glossinidius as well as the slime mold P. pallidum and the unrelated chitinases from the Firmicute L. monocytogenes. Some of the enzymes (the Listeria LmChiB and Sodalis SgChiA) encode additional domains annotated as chitin binding domains, whereas others contain only the catalytic domain (LmChiA, StChiA⌬CBM, Polysphondylium PpChitinase; Fig. 2). Additional studies are needed to clearly distinguish the respective roles of the family 18 glycosyl hydrolase domain and additional domains. The observation of marginal differences in hydrolytic activities for LmChiB and LmChiB⌬CBM without the chitin binding domain (Tables 3 and 4) suggest that there is a minimal effect of the chitin binding domain on the hydrolytic activities under the conditions tested.
The enzymatic analyses were performed using the fluorescently labeled model substrates LacdiNAc-TMR and LacNAc-TMR. A general activity toward LacdiNAc-TMR was observed FIGURE 4. Ultra-HPLC-mass spectrometry analyses of LacNAc␤1,6-LacNAc-TMR before and after reaction with four different chitinases. A, chromatograms of the reaction mixtures were monitored with a fluorescence detector (( ex ϭ 540 nm; em ϭ 580 nm), and each peak was identified by mass spectrometry. EU, emission units. B, structures of the substrate (a) and the products (LacNAc-TMR (b) and Linker-TMR (c)), corresponding to labels on the chromatograms. The yellow circle and the blue square indicate Gal and GlcNAc, respectively. C, hydrolytic reaction yields based on peak areas of the TMR products formed. For a presentation of the exact structures, please see Fig. 1 and supplemental Fig. 1. for all of the chitinases examined, except S. griseus chitinase, with the release of LacdiNAc from LacdiNAc-TMR. This is consistent with LacdiNAc binding to the Ϫ2 and Ϫ1 subsites and the hydrophobic linker with TMR in the enzymes ϩ subsites. The hydrolytic activity toward LacNAc-TMR was much lower, and only the SgChiA and StChiA⌬CBM released LacNAc from LacNAc-TMR to any extent within the 2-h incubation time ( Table 4). The increased activity toward LacdiNAc-TMR is attributed to the presence of two N-acetyl groups, on the reducing end for substrate-assisted catalysis and on the non-reducing end for additional recognition by the respective chitinases. It is notable that the T. viride enzyme mixture also exhibited a high degree of hydrolytic activity toward LacdiNAc-TMR. Further analyses are needed to elucidate which T. viride chitinase(s) is responsible for this activity and to evaluate the potential biological role. The general activity of chitinases toward a LacdiNAc-terminating substrate is a novel observation only reported previously for the human chitotriosidase (17).
Hydrolysis of LacNAc model substrates is also not commonly reported. In addition to the previous studies of a Salmonella chitinase and the human chitotrisidase by the authors of the present paper (14,17) only two studies have demonstrated such activity expressed by the chitinases in the bacterial species Amycolatopsis orientalis (32) and Bacillus circulans (33). The Amycolatopsis chitinase not only hydrolyzed chitobiose from pNP-(GlcNAc) 2 but also released free oligosaccharides and pnitrophenol from Gal␤1,4GlcNAc␤1,3Gal␤1,4GlcNAc␤-pNP GlcNAc␤1,3Gal␤1,4GlcNAc␤-pNP, and LacNAc-pNP (Gal␤1,4GlcNAc␤-pNP). LacNAc and LacNAc-pNP were not detected in the course of tetrasaccharide hydrolysis, consistent with pNP occupying the ϩ1 subsite and the tetrasaccharide occupying the enzymes Ϫ4 to Ϫ1 subsites.
To demonstrate that the activity toward LacdiNAc-TMR and LacNAc-TMR was not a minor side activity of the chitinases, we examined the substrate specificities and Michaelis-Menten kinetic parameters for LacdiNAc-TMR, LacNAc-TMR, and pNP-(GlcNAc) 2 for StChiA⌬CBM, SgChiA, and PpChitinase.
StChiA⌬CBM and SgChiA showed similar trends in hydrolytic activity toward LacNAc-TMR and LacdiNAc-TMR, with an increased activity for LacdiNAc-TMR for both enzymes as seen by the 2-4 times higher k cat values (Table 5). When comparing k cat and k cat /K m values for LacNAc-and LacdiNAc-TMR to the higher k cat and k cat /K m values for the chitin pseudosubstrate pNP-(GlcNAc) 2 , the k cat value for the StChiA⌬CBM enzyme with LacdiNAc-TMR was still 25% that of the value observed with pNP-(GlcNAc) 2 (Table 5). Even higher k cat values toward LacdiNAc-TMR relative to pNP-(GlcNAc) 2 have been observed for the human chitotriosidase (17). The rather low level of activity indicated by the low k cat values observed for the PpChitinase toward LacdiNAc-TMR (Table 5)  The presence of chitinases in P. pallidum may provide an activity of general importance, such as degradation of yeast prey, as they are also found in another major cellular slime mold genus, Dictyostelium (Fig. 2); these two genera mainly belong to different groups within the Dictyostelia (40, 41).
To investigate whether the observed activities toward LacNAc-TMR and LacdiNAc-TMR are biologically relevant, we screened the Listeria, Salmonella, and Sodalis chitinases (LmChiA and LmChiB, StChiA⌬CBM, and SgChiA, respectively) for activities toward extended substrates including LacNAc␤1-6LacNAc␤-TMR, LacNAc␤1-6(Gal␤1-3)GalNAc␣-TMR (extended core 2), and LacNAc␤1-2-Man-TMR (Fig. 1). Only LacNAc␤1-6LacNAc␤-TMR was a substrate. It is of interest to note that the two enzymes (StChiA⌬CBM and SgChiA) that showed a high catalytic activity toward this substrate both were associated with the phylogenetic group colored in blue and belonging to cluster A ( Fig. 2; Ref. 28). Further studies are, however, necessary to demonstrate that this group of chitinases indeed has an increased preference for this target.
The stepwise hydrolysis pattern for LacNAc␤1-6LacNAc␤-TMR is consistent with the terminal LacNAc occupying the Ϫ2 and Ϫ1 subsites and the inner LacNAc-TMR occupying the ϩ1 and ϩ2 subsites. The hydrolysis of LacNAc␤1-6LacNAc␤-TMR can be attributed to the flexibility of the ␤1,6 linkage. The product of hydrolysis is LacNAc␤-TMR, which in turn was further hydrolyzed by SgChiA and LmChiB enzymes. In this case the linker occupies the ϩ subsites. The product distribution profiles indicate that the tetrasaccharide is hydrolyzed twice as rapidly as LacNAc-TMR by the SgChiA, whereas they are equally good substrates for LmChiB. The inability of StChiA⌬CBM to further hydrolyze LacNAc␤-TMR might be due to the shorter linker to TMR in this compound ((CH 2 ) 2 ) compared with the linker to LacNAc-TMR used for kinetics ((CH 2 ) 8 ) or to a greater affinity for tetrasaccharide such that it precludes binding and hydrolysis of LacNAc␤-TMR. This highlights caution in the interpretation of kinetic results when nonnatural leaving groups such as fluorescent linkers or the widely used pNP are used. The hydrolysis patterns were distinct from those seen for Amycolatopsis chitinase, which bound all oligosaccharide substrates with pNP in the ϩ1 subsite. These results are in agreement with what has recently been observed for the human chitotriosidase (17). It can be concluded that terminal LacNAc or LacdiNAc structure-linked ␤1,4 or ␤1,6 in oligosaccharides, glycoproteins, or glycolipids are potential substrates for several chitinases. Especially for the Salmonella StChiA⌬CBM and Sodalis SgChiA, this activity was significant, whereas this was not for the case for the Listeria chitinases (Fig.  4). Interestingly, it has been proposed that GH_18 enzymes from the human gut bacterial symbiont Bacteroides thetaiotaomicron may target LacNAc repeats of O-glycan chains (42).
It can be suggested that for Salmonella StChiA and Sodalis SgChiA chitinases, glycans that contain either a LacNAc or LacdiNAc motif constitute part of targets that contribute to a biological response, most likely in association with infection of mammalian hosts (Salmonella) or symbiosis with insects (Sodalis). This hypothesis is in agreement with the observation that the gene expression of the Salmonella chitinase is markedly up-regulated during infection of macrophage and epithe-lial cells (15,16). We are currently conducting studies aimed at identifying the potential biological substrates for a selection of these enzymes including the Listeria and Salmonella chitinases.