Novel Dextranase Catalyzing Cycloisomaltooligosaccharide Formation and Identification of Catalytic Amino Acids and Their Functions Using Chemical Rescue Approach

Background: Catalytic residues and molecular mechanism of GH-66 enzymes were hitherto unknown.

Results: Novel dextranase produced isomaltotetraose and cyclo-isomaltosaccharides. Its nucleophile (Asp³⁴⁰) and acid/base catalyst (Glu⁴¹²) were identified by a chemical rescue approach.

Conclusion: Three GH-66 enzyme types were newly classified for the first time.

Significance: This work elucidates production of isomaltotetraose and cycloisomaltosaccharides, classification of GH-66, identification of catalytic residues, and novel dextran-forming type chemical rescue.

Abstract

A novel endodextranase from Paenibacillus sp. (Paenibacillus sp. dextranase; PsDex) was found to mainly produce isomaltotetraose and small amounts of cycloisomaltooligosaccharides (CIs) with a degree of polymerization of 7–14 from dextran. The 1,696-amino acid sequence belonging to the glycosyl hydrolase family 66 (GH-66) has a long insertion (632 residues; Thr⁴⁵¹–Val¹⁰⁸²), a portion of which shares identity (35% at Ala³⁹–Ser¹³⁰⁴ of PsDex) with Pro³²–Ala⁷⁵⁵ of CI glucanotransferase (CITase), a GH-66 enzyme that catalyzes the formation of CIs from dextran. This homologous sequence (Val⁸³⁷–Met⁹³² for PsDex and Tyr⁴⁰⁴–Tyr⁴⁹² for CITase), similar to carbohydrate-binding module 35, was not found in other endodextranases (Dexs) devoid of CITase activity. These results support the classification of GH-66 enzymes into three types: (i) Dex showing only dextranolytic activity, (ii) Dex catalyzing hydrolysis with low cyclization activity, and (iii) CITase showing CI-forming activity with low dextranolytic activity. The fact that a C-terminal truncated enzyme (having Ala³⁹–Ser¹³⁰⁴) has 50% wild-type PsDex activity indicates that the C-terminal 392 residues are not involved in hydrolysis. GH-66 enzymes possess four conserved acidic residues (Asp¹⁸⁹, Asp³⁴⁰, Glu⁴¹², and Asp¹²⁵⁴ of PsDex) of catalytic candidates. Their amide mutants decreased activity ($\raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$1,500$}\right.$ to $\raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$40,000$}\right.$ times), and D1254N had 36% activity. A chemical rescue approach was applied to D189A, D340G, and E412Q using α-isomaltotetraosyl fluoride with NaN₃. D340G or E412Q formed a β- or α-isomaltotetraosyl azide, respectively, strongly indicating Asp³⁴⁰ and Glu⁴¹² as a nucleophile and acid/base catalyst, respectively. Interestingly, D189A synthesized small sized dextran from α-isomaltotetraosyl fluoride in the presence of NaN₃.

Introduction

Endodextranases (EC 3.2.1.11; Dexs)² randomly hydrolyze the α-1,6-linkage of dextran (1) and are classified into the glycoside hydrolase 49 and 66 (GH-49 and GH-66) families based on amino acid sequence similarity (2). Cycloisomaltooligosaccharide glucanotransferase (CITase) is also classified into the same group of GH-66 and catalyzes the conversion of dextran to cycloisomaltooligosaccharide (CI) by intramolecular transglycosylation (3, 4). CI is an attractive cyclic sugar for the prevention of dental plaque formation due to its strong inhibition of mutansucrase-catalyzed insoluble glucan formation (5). CI with 10 glucose units (CI-10; CI-n, CI with n glucose units) exhibits stronger inclusion ability than do cyclodextrins (cyclic α-1,4-glucosidic sugars) (6).

GH-66 enzymes are composed of five regions from the N to the C terminus (7, 8) as follows: N-terminal variable region, conserved region (CR), CITase-specific region (CIT-SR), C-terminal conserved region, and C-terminal variable region (C-VR). All dextranases reported so far are devoid of CIT-SR. CIT-SR, which forms carbohydrate-binding module 35 (CBM-35), contributed to the CI formation of CITase (7). More recently, we resolved the three-dimensional structure of Dex from Streptococcus mutans (SmDex), which is truncated at the N-terminal variable region and C-VR (9, 10). Our x-ray studies demonstrated that CR forms a catalytic (β/α)₈ barrel as has been observed in the GH-13, -27, and -31 proteins, indicating that they probably share a common evolutionary origin (11). Our studies also predicted that three acidic amino acids at CR of Paenibacillus sp. Dex (PsDex) (Asp¹⁸⁹, Asp³⁴⁰, and Glu⁴¹²) are candidates for catalytic residues (9, 10, 12), although the functions of those candidates have yet to be elucidated.

Chemical rescue (ChR) is a reaction that is used to recover the activity of mutant enzyme by an exogenously added organic or inorganic compound that functions instead of the mutated residue. For example, the catalytic residue-mutated glycosylase displays its activity on a fluoride substrate by the addition of an anion (e.g. N₃⁻ or formate), which remains at the position of the altered residue. The glycosylase, mutated at its so-called nucleophile, catalyzes the formation of a product (glycosyl azide) that has an anomeric configuration opposite that of the substrate. The ChR of glycosylases, mutagenized at the acid/base catalyst, forms a product (glycosyl azide) with the same anomeric configuration as that observed in the intact enzyme-catalyzed hydrolytic reaction (13, 14). Therefore, the use of ChR is among the most convenient approaches to identifying catalytic residues functions.

We recently discovered PsDex from Paenibacillus sp. isolated from soil on the Hokkaido University campus. Interestingly, PsDex produced isomaltotetraose (IG-4) and small amounts of CIs (degree of polymerization (DP) = 7–14; CI-7 to CI-14). The amino acid sequence of PsDex harbors CBM-35 at its CIT-SR, permitting the classification of the GH-66 enzymes into three types: (i) pure Dex, (ii) Dex with low cyclization activity, and (iii) CITase with low high cyclization activity and low hydrolytic activity. As mentioned before, the three-dimensional structure analysis of SmDex (9, 10) implicated three acidic PsDex residues as candidates for catalytic residues (Asp¹⁸⁹, Asp³⁴⁰, and Glu⁴¹²). However, the function of those three residues remains unclear. To address this, we applied the ChR approach to their functional analysis. This study, for the first time, identifies the function of the catalytic residues of GH-66 enzymes.

EXPERIMENTAL PROCEDURES

Bacterial Strains and Plasmids

Paenibacillus sp. was isolated from soil on the Hokkaido University campus (Sapporo, Japan). This bacterium was aerobically incubated in 5 ml of preculture medium containing 0.1% dextran T2000 (Amersham Biosciences), 1% Bacto-peptone (BD Bioscience), 0.01% yeast extract (BD Bioscience), and 0.05% NaCl (pH 7.0) at 30 °C. For the production of PsDex, the precultured Paenibacillus sp. was added to a fermenter containing 1,000 ml of medium (3 g of Na₂HPO₄, 1.5 g of KH₂PO₄, 0.25 g of NaCl, 0.5 g of NH₄Cl, 0.5 mg of thiamine HCl, 0.5 mm MgSO₄, 0.05 mm CaCl₂, 100 mm MES-NaOH buffer (pH 7.0), and 0.5% dextran T2000 per liter) and cultivated aerobically three times at 30 °C for 20 h. Escherichia coli DH5α and E. coli BL21 (DE3)-CodonPlus-RIL (Stratagene, La Jolla, CA) were used for the construction of expression plasmids and for the production of recombinant PsDex and its C-terminally truncated PsDex (Ala³⁹–Ser¹³⁰⁴; PsDex-CT), respectively. Plasmids of pBluescript II SK(+) (Stratagene) and pET-23d (Novagen, Darmstadt, Germany) were used for the subcloning of DNA fragments amplified by polymerase chain reaction (PCR). E. coli was grown in LB medium (10 g of Bacto-tryptone (BD Bioscience), 5 g of yeast extract (BD Bioscience), and 5 g of NaCl in 1 liter of H₂O, pH 7.0) containing ampicillin (50 μg/ml).

Purification of Native PsDex

Paenibacillus sp. cells were collected by centrifugation (8,000 × g for 10 min at 4 °C), resuspended in 500 ml of 20 mm potassium phosphate buffer (pH 6.5) (buffer A), homogenized by an ice-chilled French press (Ohtake, Tokyo, Japan; 1,350 kg/cm², 3 times), and centrifuged (14,000 × g for 20 min at 4 °C) to discard the insoluble components. Supernatants were pooled as crude extract containing 183 units of PsDex with 0.0178 unit/mg. Solid ammonium sulfate was slowly added to the crude extract up to 30% saturation, and the turbid solution was maintained at 4 °C overnight. The resulting precipitant was collected by centrifugation (12,000 × g for 20 min at 4 °C), dissolved in 600 ml of buffer A, dialyzed against buffer A, and centrifuged (8,000 × g for 10 min at 4 °C) to remove insoluble materials. The supernatant (865 ml, 178 units, 0.0548 unit/mg) received a 0.02% final concentration of sodium azide. The solution was applied to a column of DEAE-TOYOPEARL 650M (3.1 × 66 cm; Tosoh, Tokyo, Japan) equilibrated with buffer A, followed by elution with a 0–1 m sodium chloride of linear gradient. The active fractions (65.2 units, 0.688 unit/mg) were dialyzed against buffer A containing 1.0 m ammonium sulfate (buffer B), loaded onto a column of Butyl-TOYOPEARL 650M (2.2 × 55 cm; Tosoh) equilibrated with buffer B, and eluted with a linear gradient of 1.0–0 m ammonium sulfate. Active fractions (40.0 units, 8.51 units/mg) were dialyzed against buffer A containing 0.05 m sodium chloride (buffer C), concentrated to 3.5 ml, and added to a gel filtration column using Sepharose 6B (2 × 110 cm; Amersham Biosciences) equilibrated with buffer C. The active fractions (17.9 units, 14.3 units/mg) were dialyzed against buffer A. All purification steps were performed at 4 °C.

Protein concentration of the crude extract and ammonium sulfate separation was measured using the Bradford method (15) with bovine serum albumin as a standard. Protein concentration at other purification steps was determined using 13.9 of A^1% at 280 nm obtained by determination of the amino acid contents of the hydrolysate (6 n HCl for 24 h at 110 °C; 500-pmol sample) using an Amino Tac JLC-500/V amino acid analyzer (JEOL, Tokyo, Japan). Dextranolytic activity was assayed using the copper bicinchoninate method (16) with glucose as the standard by estimating the reducing power from 0.4% dextran T2000 in 20 mm sodium acetate (pH 5.5) at 35 °C. One unit of hydrolytic activity is defined as the amount of enzyme that released 1 μmol of reducing power per min under the assay conditions.

Electrophoresis

SDS-PAGE was performed using Laemmli's method (17) with a Mini-Protean III cell apparatus (Bio-Rad) and a 10% gel, followed by protein staining with Rapid CBB KANTO (Kanto Kagaku, Tokyo, Japan) or by sugar detection with a periodic acid-Schiff staining (PAS) kit (Nacalai Tesque, Kyoto, Japan) (18).

Analyses of Monosaccharide Composition and Sugar Contents

Lyophilized native PsDex (50 μg) was dissolved in 100 μl of 2 m trifluoroacetic acid containing 0.1 m HCl, heated for 10 h at 100 °C, neutralized with 1 m NaOH, and lyophilized. Monosaccharides were separated using high performance anion exchange chromatography equipped (Dionex, Sunnyvale, CA) and CarboPac PA1 column (4 × 250 mm; Dionex) and eluted with 15 mm NaOH at a flow rate of 1.0 ml/min. Carbohydrate contents were measured by Robyt's microscale phenol-sulfuric acid method (19) with d-galactose as a standard.

Amino Acid Sequence Analysis

Purified native PsDex (50 pmol) was directly blotted on a polyvinylidene fluoride membrane using ProSorb (Applied Biosystems, Foster City, CA), followed by sequencing using a model 477A protein sequencer (Applied Biosystems) with a model 120A on-line phenylthiohydantoin analyzer (Applied Biosystems). For the analysis of the internal sequence, Lys-C peptidase (Wako, Tokyo, Japan)-digested peptides were prepared from 4-vinylpyridine-modified native PsDex (2 nmol), isolated using high pressure liquid chromatography (HPLC) with a C8P-50 column (4.6 × 150 mm; Asahikasei Asahipak Column, Osaka, Japan) using a gradient of 0–80% acetonitrile in 0.1% trifluoroacetic acid, and subjected to the aforementioned sequence analysis.

Cloning of PsDex Gene from Paenibacillus sp.

All PCRs were performed using KOD DNA polymerase or KOD Dash DNA polymerase (Toyobo, Osaka, Japan), and DNA sequence analyses utilized an ABI PRISM 310 Genetic Analyzer (Applied Biosystems). The first PCR was performed using two degenerate primers: psdexAs (5′-¹¹⁵GCIAATCAGGAGGAGAAGCA¹³⁴-3′; I, inosine; 115 or 134, counting nucleotide numbers from ATG of the initial Met; sense primer corresponding to N-terminal sequence of ANQEEKQ) and psdexBa (5′-⁷⁸³GTTGTTCATTTGGTAIGCCAT⁷⁶¹-3′; antisense primer corresponding to Lys-C peptidase-digested peptide of MAYQMNY) with genomic DNA as a template, followed by successive PCR using psdexCa (5′-¹⁹⁷⁶CCIGGGAAIATIGCTTCICCTTG¹⁹⁵⁴-3′; antisense primer corresponding to Lys-C peptidase-digested peptide of QGEAIFPG). Southern blot analysis was performed using a digoxigenin-labeled fragment (the first PCR product of 669 bp amplified with psdexAs and psdexBa) using DIG-High Prime (Roche Applied Science) on XbaI-digested chromosomal DNA, resulting in a positive signal with a 5-kb size. This XbaI-digested 7.7-kb fragment (1 ng/μl) was subjected to self-ligation and used as a template for inverse PCR with psdexDa (5′-²⁴³ACCGCTCCAATCTTCGTTGTCCC²²¹-3′; antisense primer) and psdexEs (5′-¹⁸³⁷GACGATTACATGACATTCCATGTC¹⁸⁶⁰-3′; sense primer), amplifying the 5.5-kb DNA fragment. For amplification of the 3′-region of the PsDex gene, a fragment (627 bp) isolated from the NdeI-SacI-digested inverse PCR product was used in the second Southern blot analysis using XhoI-digested genomic DNA. A positive 7.5-kb XhoI-fragment was circulated by self-ligation and used as a template for inverse PCR with psdexFa (5′-³⁰¹⁴TACGTAAAGATCAGGCTTTGTTCTC²⁹⁹⁰-3′; antisense primer) and psdexGs (5′-³⁶³⁶CAATGATTCCAACTGGAGAAATG³⁶⁵⁸-3′; sense primer), amplifying a 6-kb fragment involving a PsDex gene at the 3′-end. Using psdexHs (5′-GGATCCTACGGACTCGCTGTAGG⁻¹²⁷-3′; sense primer) and psdexIa (5′-³²⁸⁵GGCATCCATAAGCCGTACAG³²⁶⁶-3′; antisense primer), the final PCR amplified a DNA fragment of 5.45 kb containing an open reading frame (ORF) of 5,091 bp with a 5′-flanking region of 126 bp and a 3′-flanking region of 230 bp. The three obtained clones were completely identical, confirming the absence of a PCR error.

A computer-based sequence analysis to find homologous regions was performed using the BLAST network service (20) with the Swiss-Prot/TrEMBL database. Prediction of a protein signal peptide was done using the Signal P server (21).

Construction of Expression Vectors of PsDex and PsDex-CT

After C761 was replaced by G to remove an inner NcoI site without amino acid substitution, NcoI and NotI sites were introduced to the 5′ and 3′ termini of the ORF, respectively, and the ORF was then introduced to the NcoI-NotI site of pET23d to produce a PsDex protein that lacked the original signal sequence (Met¹–Ala³⁸). PCR was performed using psdexRs (5′-TACCATGGCCAATCAGGAAGAGAAGC¹³³-3′, where the NcoI site is underlined; sense primer) and psdexRa (5′-⁷⁷³TGATAGGCCATGCCCGCTGAGCCG⁷⁵⁰-3′, where the boldface letter corresponds to C761 replacement by G; antisense primer). The resultant DNA fragment was used as a primer for megaprimer PCR (22) with psdexRCA2 (5′-AAGCGGCCGCTTCGATCAGATCCAACAA⁵⁰⁷¹-3′, where the NotI site is underlined; antisense primer). The PCR product and pET-23d were digested by NcoI and NotI (Takara, Kyoto, Japan), followed by connection with a DNA Ligation Kit version 2 (Takara) to construct an expression vector to produce PsDex (Ala³⁹-Ala¹⁶⁹⁶, an original Glu¹⁶⁹⁶ replaced by Ala) with a His₆ tag at its C terminus.

The DNA harboring PsDex-CT with the His₆ tag at its C terminus was constructed using PCR with psdexCTs (5′-²³³⁴CATCTTCCCTAAAACACCGGG²³⁵⁴-3′; sense primer) and psdexCTa (5′-³⁹²²CAGCGGCCGCAGATCCATCCGGGAATAGC-3894–3′, where the NotI site is underlined), replacing the original Pro¹³⁰⁵ by Ala. The resultant PCR product and pET23d carrying the PsDex gene were digested by SacI and NotI (Takara) and connected using a DNA Ligation Kit version 2.

Eleven kinds of mutations at Asp¹⁸⁹, Asp³⁴⁰, Glu⁴¹², Asp¹²⁵⁴, and Cys¹¹²⁴ of PsDex-CT, having a His₆ tag at the C-terminal, was performed by megaprimer PCR (22) using the appropriate primers. The amplified DNA fragment was digested by SacI and NotI, followed by introduction to the corresponding restriction sites of PsDex-CT gene-carrying pET-23d.

Production and Purification of PsDex and PsDex-CT

PsDex, PsDex-CT, and the mutants of PsDex-CT at Asp¹⁸⁹, Asp³⁴⁰, Glu⁴¹², Asp¹²⁵⁴, and Cys¹¹²⁴ were produced in E. coli transformants carrying the respective expression plasmid. Each transformant was cultured in 1,000 ml of LB medium containing 50 μg/ml ampicillin at 37 °C until the absorbance at 600 nm reached ∼0.5. Protein production was induced with 0.2 mm isopropyl β-thiogalactoside, followed by further incubation with vigorous shaking at 18 °C for 20 h. After E. coli cells were disrupted by sonication using a Sonifier 250 (Branson, Danbury, CT), PsDex and PsDex-CT were purified to homogeneity by nickel-chelating chromatography using Chelating Sepharose Fast Flow (Amersham Biosciences). Active fractions were dialyzed against buffer A and concentrated using a CentriPrep YM-50 apparatus (Millipore, Bedford, MA). The concentration of purified protein was measured using 12.2 (recombinant PsDex) and 5.54 (PsDex-CT and its 11 mutants) of A^1% at 280 nm determined by the aforementioned analysis of their amino acid compositions.

Enzyme Reaction

The effects of pH on dextranolytic activity were examined by incubating each enzyme (26.0 nm PsDex and 30.0 nm PsDex-CT) with 0.4% dextran T2000 at 35 °C in 16 mm modified Britton-Robinson buffer (pH 2.60–11.4; pH of mixture of acetate, phosphate, and glycine was adjusted by NaOH; buffer D). For pH stability, PsDex (190 nm) and PsDex-CT (203 nm) were maintained at 4 °C for 18 h in 36 mm buffer D, followed by an assay of residual enzyme activity under the aforementioned conditions. For thermal stability, PsDex (25.0 nm) and PsDex-CT (42.2 nm) were maintained at 25–60 °C and 20 mm sodium acetate buffer (pH 5.5) (Na-AB) for 15 min, and residual activity was then measured under the assay conditions.

PsDex- or PsDex-CT-catalyzing CI-production was performed by incubating 30 nm enzyme with 3% dextran T10 at 35 °C and 20 mm Na-AB. The remaining dextran and long oligosaccharides were discarded by the addition of ethanol (2 volumes) to 25 ml of the reaction mixture, forming a precipitate at 4 °C for 1 h. The supernatant was concentrated to 2 ml by a vacuum evaporator, reacted with buckwheat α-glucosidase (7.6 units/ml) and Aspergillus niger α-glucosidase (12 units/ml) at 37 °C and 20 mm Na-AB for 15 h to digest short linear oligosaccharides, passed through a small column of Amberlite MB-4 (Organo, Tokyo, Japan), and applied to the Sep-PackC18 cartridge (Waters, Milford, MA) followed by a first elution with milli-Q water (5 ml) for washing the remaining short linear oligosaccharides and by second elution with 20% ethanol (5 ml) to recover the CIs. The ethanol fraction was concentrated to 0.4 ml and analyzed by HPLC using a model D-2000 refractive index detector (Hitachi, Tokyo, Japan), using a Shodex RS Pak DC-613 column (6 × 150 mm; Showa Denko, Tokyo, Japan), followed by elution at 60 °C with 61% acetonitrile. Molecular masses of the CIs were determined using model JMS-SX102A electrospray ionization-mass spectrometry (JEOL, Tokyo, Japan). ¹³C NMR spectra were recorded using a Bruker AMX-500 spectrometer at 125 MHz with an external standard of trimethylsilyl propionate.

Gram-level production of IG-4 was performed by reacting recombinant PsDex (1.0 mg) with 3% dextran T2000 (10 g) as a substrate at 20 mm Na-AB for 24 h. Resultant IG-n was concentrated, applied to carbon-Celite column chromatography (23), and eluted with a 0–5% isopropyl alcohol linear gradient, resulting in about 4.5 g of pure IG-4. Purified IG-4 (100 mg) was converted to α-isomaltotetraosyl fluoride (IG4F) (24) for the substrate used in the ChR reaction.

The ChR reaction was performed by the reaction of the mutated PsDex-CT at Asp¹⁸⁹, Asp³⁴⁰, or Glu⁴¹² (see Table 2 for mutants used) with 6 mm IG4F and salt (0.2–2.4 m NaN₃, sodium formate, or NaNO₃; Table 2) at 35 °C and 200 mm Na-AB. Fluoride ions liberated from IG4F were measured by estimating the formation of a lanthanum complex (25). The ChR product was analyzed using thin layer chromatography (TLC) with a silica gel plate (60F₂₅₄; Merck) and a solvent of nitromethane/1-propanol/water (4:10:3, v/v/v; two-time development), followed by visualization at 110 °C for 5 min after dipping of the TLC plate in 5% sulfuric acid in methanol containing 0.03% α-naphthol. Purification of two ChR products formed by D340A and E412Q was done using TLC. After development, each product was recovered from the TLC plate. Their structures were analyzed on a model JMS-SX102A fast atom bombardment mass spectrometer and a model AMX-500 ¹H NMR at 500 MHz.

TABLE 2.

Chemical rescue of mutants constructed from PsDex-CT

ChR was evaluated by the determination of F⁻ from 6.0 mm IG4F, where the v for PsDex-CT was 4.63 μmol/min/mg.

Enzyme	Salt added	v	Ratio (-fold of v without salts)
		μmol/min/mg	-fold
D189G	0	0.050	1.0
	0.2 m NaN3	0.16	3.2
	2.4 m sodium formate	0.55	11
D189A	0	0.17	1.0
	0.4 m NaN3	1.54	9.1
	2.4 m sodium formate	1.49	8.2
D189N	0	0.12	1.0
	0.2 m NaN3	0.19	1.6
	2.4 m sodium formate	0.18	1.5
D340G	0	0.00016	1.0
	0.4 m NaN3	0.0027	1.7
	2.4 m sodium formate	0.0065	4.0
D340A	0	0.00030	1.0
	0.4 m NaN3	0.0019	6.3
	2.4 m sodium formate	0.0038	12.6
E412G	0	0.00072	1.0
	0.4 m NaN3	0.0038	5.3
	2.4 m sodium formate	0.0032	4.4
E412A	0	0.00082	1.0
	0.4 m NaN3	0.0052	6.3
	2.4 m Sodium formate	0.0042	5.1
E412Q	0	0.00065	1.0
	0.4 m NaN3	0.0072	11
	2.4 m sodium formate	0.0064	9.8

RESULTS

Purification and Characterization of PsDex

A dextran-degrading bacterium was isolated from soil. DSMZ (Braunschweig, Germany) identified this strain as Paenibacillus sp. on the basis of the 16S rRNA sequence and cellular fatty acid composition analyses. Dextranolytic activity at 24 h after cultivation reached 100 units/ml of medium, 3 and 97% of which were found in the supernatant and cell-homogenized fraction, respectively. The cell suspension, which reacted directly with 0.4% dextran T2000 at 35 °C and 50 mm sodium acetate buffer (pH 5.5; Na-AB), displayed almost the same activity as that observed with the cell-homogenized fraction, implying that PsDex resided at the cell surface. Microscopic observation confirmed that the cells were not disrupted during incubation with dextran.

PsDex was isolated from the cell-disrupted fraction by four purification procedures; the final step entailed gel filtration and purified a 200-kDa protein with a single band on SDS-PAGE (Fig. 1A, lane 1). This protein band was positive for PAS (Fig. 1A, lane 2). Both monosaccharide composition and sugar content analyses confirmed that PsDex was a glycoprotein that had 5.1% d-galactose without glucosamine, galactosamine, or N-acetyl sugars thereof. PsDex was the most active at pH 5.5 and was stable at pH 5.0–9.0 and up to 40 °C.

FIGURE 1.

SDS-PAGE of purified native PsDex (A), recombinant PsDex (B), and PsDex-CT (C). Lanes M, size markers; lanes 1, 3, and 4, protein staining; lane 2, detection of glycoprotein using PAS.

Table 1 summarizes the kinetic parameters of the hydrolytic reaction on IG-n (IG-3 to IG-7) and dextran T2000 with their cleavage patterns. PsDex had extremely low activity on IG-4, producing low amounts of glucose and IG-3 and no activity on IG-2 and IG-3. Isomaltooligosaccharides having a DP value of >5 became favorable substrates, and IG-5 was cleaved to glucose and IG-4. IG-5 was formed from IG-6, IG-7, and dextran T2000, indicating the preference of PsDex for the production of IG-5 at an initial stage of the reaction. As shown in Fig. 2B, PsDex formed IG-4 as a final product with a 53% yield, enabling the preparation of gram quantities of IG-4 from 10 g of dextran T2000.

TABLE 1.

Kinetic parameters for IG-n and dextran T2000 by PsDex and products with their levels

Substrate	Km	k0	k0/Km	Products with levelsa
	μm	s−1	μm
IG-2, IG-3, IG-4				No or weak activity
IG-5	0.414	4.36	10.5	+++, Glc/IG-4
IG-6	0.386	7.53	19.5	+++, Glc/IG-5; ++, IG-2/IG-4
IG-7	0.288	7.93	27.5	+++, IG-2/IG-5; ++, IG-3/IG-4; +, Glc/IG-6
Dextran T2000	0.000183	44.3	242000	+++, IG-5/IG-4; +, >IG-6/<IG-3

^a +++, ++, and + indicate strong, moderate, and weak production levels, respectively, tested by TLC.

FIGURE 2.

Linear sugar products (B) or cyclic sugar products (D and F) from dextran and activity restoration of D189C/C1124Y by KI treatment (E). A and B, dextran T2000 (1%, w/v) was incubated with PsDex (14 μg/100 ml of reaction mixture) at 20 mm Na-AB and 35 °C for 24 h. Products were analyzed by HPLC using a Shodex RS pack DC-16 column with elution of 55% acetonitrile at 60 °C. A shows standards of glucose (peak 1) and IG-2 to IG-7 (peaks 2–7, respectively). C, D, and F, HPLC profiles at reaction times of 6 h (C) and 24 h (D). ¹³C NMR recorded signals (F) of C1–C6 of glucosyl units of CI formed by PsDex-CT (signals from a produced CI-8). E, D189C/C1124Y (10 μg/0.1 ml) was treated with various concentrations of KI at 35 °C and 100 mm sodium phosphate buffer (pH 7.0) for 24 h, followed by measurement of dextranolytic activity under assay conditions.

PsDex Gene and Reduced Primary Structure

The PsDex gene was cloned from genomic DNA of Paenibacillus sp. by a series of PCRs and Southern blot procedures. The isolated gene had two putative promoters 5′-⁻⁵⁶TTAAA⁻⁵²-3′ and 5′-⁻²⁹TAAAT⁻²⁵-3′, together with a putative ribosome-binding site, 5′-⁻¹⁶AGGGAGGA⁻⁹-3′. At the 3′-flanking region of 229 bp, no termination loop was found downstream of the stop codon (5′-⁵⁰⁸⁹TAA⁵⁰⁹¹-3′). The deduced amino acid sequence contained an N-terminal sequence of ANQEEKQSSQAGLRALTVS and many internal sequences (data not shown) of Lys-C peptidase-digested peptides, indicating that a signal peptide was cleaved at the typical processing site between Ala³⁸ and Ala³⁹ (26) and that the isolated gene encoded PsDex protein. A BLAST homology search revealed PsDex to be a GH-66 enzyme family member without obvious similarity to GH-49 enzymes (27–29). Multiple alignments with GH-66 enzymes predicted that PsDex is composed of five regions together with signal sequence (Met¹–Ala³⁸) as follows: N-terminal variable region (Ala³⁹–Ala⁵³), CR (Leu⁵⁴–Ala⁴⁵⁰), PsDex-specific region (Thr⁴⁵¹–Val¹⁰⁸²), C-terminal conserved region (Gly¹⁰⁸³–Val¹²⁸⁷), and C-VR (Asp¹²⁸⁸–Pro¹⁶⁹⁶). The PsDex-specific region is composed of three portions: long insertion 1 (LI-1; Thr⁴⁵¹–Glu⁸¹²), CIT-SR (Ser⁸¹³–Val⁹⁴⁶) and long insertion 2 (LI-2; Val⁹⁴⁷–Val¹⁰⁸²). The similar sequences of LI-1 and LI-2 are not found in other GH-66 enzymes, including CITase (4) and streptococcal Dexs (30–32). PsDex displays the highest identity with CITase (35%), whereas streptococcal Dexs display a range of only 25–26%.

Expression and Characterization of Recombinant PsDex and PsDex-CT

C-VR of PsDex contains a Pro/Ser-rich region (PSRR; Pro¹³⁰⁵–Pro¹³²⁹), followed by three sets of putative surface layer homology domains (SLHDs; Tyr¹⁵¹⁸–Ala¹⁶⁸⁵; see “Discussion” for details). Therefore, we constructed a PSRR/SLHD-truncated PsDex (PsDex-CT carrying Ala³⁹–Ser¹³⁰⁴) to elucidate the influence of those regions on catalytic function. Full-size PsDex or PsDex-CT was produced in Escherichia coli cells, and enzyme activities in the medium were increased to be 162 units/1,000 ml and 433 units/1,000 ml, which were larger than that of native PsDex (100 units/1,000 ml) by 1.6 and 4.3 times, respectively. Each enzyme was purified as a single band protein (Fig. 1, A and B) with a specific activity of 14.0 units/mg for PsDex and 7.20 units/mg for PsDex-CT. The N-terminal sequence started at Ala³⁷ in each protein, and electrospray ionization-MS and SDS-PAGE analyses determined the molecular sizes of 186 kDa for PsDex and 143 kDa for PsDex-CT (Fig. 1, B and C), indicating the absence of proteolytic digestion during production, which occurred in the production of SmDex (8). PsDex and PsDex-CT showed the same pH (stable at pH 5.2–10; optimum at pH 5.5) and temperature (stable at <37 °C) properties, together with the formation of mainly IG-4 from dextran T2000. All of these properties were identical to those of native PsDex, allowing us to use those expressed enzymes for further experiments.

Investigation of CI Production

Because the primary structure of PsDex included CIT-SR, which is only found in CITase, we investigated the PsDex-CT-associated CI production from dextran T10 (Fig. 2C). Structures of seven purified CIs were analyzed using electrospray ionization-MS and NMR. The mass signals of [M + Na]⁺ were 1157.29, 1319.57, 1481.67, 1644.11, 1806.28, 1968.21, 2130.62, and 2292.91, the estimated masses of which were multiples of 162.14, corresponding to a mass of 1 glucosyl unit, indicating that each carbohydrate was a non-reducing sugar with cyclic form. ¹³C NMR (Fig. 2F) also supported the PsDex-CT-catalyzed CI formation of CI-7 to CI-14 (3, 6). Prolonged incubation with the enzyme resulted in CI degradation (Fig. 2D) because PsDex-CT had endo-wise dextranolytic activity, which could cleave the CIs with almost equal efficiency. Native and recombinant PsDexs without truncation also catalyzed the same CI production, which was followed by degradation.

Candidates for Catalytic Residues

Multiple alignments indicated that four acidic amino acids (Asp¹⁸⁹, Asp³⁴⁰, Glu⁴¹², and Asp¹²⁵⁴) of PsDex are highly conserved in GH-66 enzymes. Our three-dimensional structure of SmDex (9, 10) indicated that three former residues are the most likely candidates for catalytic amino acids. We mutated those acidic amino acids of PsDex-CT, including Asp¹²⁵⁴, and then measured their enzyme activities. D1254N displayed 4.3 units/mg (36% of PsDex-CT with 7.20 units/mg), D189N showed 0.00481 unit/mg, D340N showed 0.0000689 unit/mg, and E412Q showed 0.000178 unit/mg. The latter mutants lost most of their dextranolytic activities.

ChR was applied to Asp¹⁸⁹, Asp³⁴⁰, and Glu⁴¹² mutants (Fig. 3). Sodium azide or sodium formate enhanced the fluoride ion-releasing velocity from 6.0 mm α-isomaltotetraosyl fluoride (IG4F) by 1.5–13 times (Table 2). Structures of products formed in the presence of the azide ion (Figs. 3, A and B, and 4C) were analyzed by three approaches, namely the reducing power test, fast atom bombardment MS, and ¹H NMR. D340G and E412Q synthesized the non-reducing sugars that had the same molecular mass (691 Da), whereas they exhibited different J₁₂ values of 8.5 and 2.5 Hz in ¹H NMR, respectively, indicating that D340G formed β-isomaltotetraosyl azide and E412Q formed α-isomaltotetraosyl azide. On the other hand, D189A with 0.4 m NaN₃ or 0.4 m sodium formate produced IG-4 and/or several IG-n from 6.0 mm IG4F, 20 mm IG-5, and 0.40% dextran T2000 (Fig. 3, D and E), suggesting that a ChR of D189A catalyzed the hydrolytic reaction. The same products were also observed in the presence of 0.4 m NaNO₃ (lanes 11, 16, and 21 of Fig. 3, D and E). Interestingly, the formation of IG-4 from IG4F and dextran T2000 appeared by D189A without any salt (Fig. 3, D (lane 8) and E (lane 18)), and those IG-4 productions were markedly enhanced in the presence of salts (Fig. 3, D (lanes 9–11) and E (lanes 19–21)).

FIGURE 3.

ChR catalyzed by mutated PsDex-CT. A, D340A-catalyzed ChR. Lanes 1 and 2, reaction time of 0 and 24 h, respectively. B, purified product observed in lane 2. Std, standard sugars of glucose and IG-2 to IG-7. Lane 3, product purified by HPLC. C, ChR catalyzed by Glu⁴¹² mutants. Lanes 4–6, ChRs by E412G/Q/A, respectively. D and E, D189A-catalyzed ChR. IG4F, IG-5, and dextran T2000 were used for substrates in lanes 7–11, 12–16, and 17–21, respectively; substrate with NaN₃ was spotted to lanes 7, 12, and 17; substrate with D189A was spotted to lanes 8, 13, and 18; substrate with D189A and NaN₃ was spotted to lanes 9, 14, and 19; substrate with D189A and sodium formate was spotted to lanes 10, 15, and 20; substrate with D189A and NaNO₃ was spotted to lanes 11, 16, and 21. F, polysaccharide formation by D189A-catalyzed ChR. Cont, without D189A; R, with D189A.

FIGURE 4.

Multiple alignments of SLHDs of PsDex and B. anthracis surface array protein (A), CBM-35 (B), and CBM-6 (C). A, SLHD; top and bottom three amino acid sequences, C-VR of PsDex and N terminus of B. anthracis surface array protein, respectively; white letters with black background, identical residues at more than five sequences; letters with gray background, identical residues at more than two sequences of PsDex or surface array protein; region with square, LTRAE motif of SLHD. B, CBM-35; PsDR1, this study; BcCR1 and BcCR2, two sequences of CITase from Bacillus circulans T-3040 (GenBank^TM ID AB073929) (7); BgCts, 6-glucosyltransferase CBM-35 from Bacillus globisporus (AB073929) (35); AoGac, exo-β-d-glucosaminidase from Amycolatopsis orientalis (AY962188) (36). White letters with black background, identical residues at more than four sequences; letters with gray background, identical residues at three sequences. C, CBM-6; PsDR2, this study; SdAga, β-agarase from S. degradans strain 2–40 (EMBL ID CP000282) (37); AaAga, α-agarase from A. agarilytica (GenBank number AAF26838.1) (38); white letters with black background, identical residues at all sequences; letters with gray background, identical residues at two sequences, including PsDR2.

D189A displayed hydrolysis at its ChR, suggesting that transglycosylation might also occur. D189A was reacted with 30 mm IG4F at a high concentration in the presence of 0.4 m NaN₃, and we found that the product had a large molecular size because this sugar stayed at its original position on the TLC plate. The addition of ethanol to the reaction mixture resulted in the formation of turbid material (Fig. 3F). This turbid material was identified as a dextran due to the production of IG-n by SmDex treatment. Its size was analyzed by estimating DP using reducing power and total sugar, indicating an average DP of about 100. This size can be categorized as a small dextran.

D189C/C1124Y Activity Restoration by KI Treatment

Asp¹⁸⁹ of PsDex-CT was substituted by a Cys residue, followed by KI oxidation. Prior to this mutation, an original and sole Cys¹¹²⁴ of PsDex-CT was replaced with a Tyr residue to construct C1124Y, and the further mutation at Asp¹⁸⁹ was then introduced to form D189C/C1124Y. C1124Y maintained the same dextranolytic activity (7.22 units/mg) as observed in PsDex-CT (7.20 units/mg), whereas D189C/C1124Y decreased its specific activity (0.0768 unit/mg) by $\raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$94$}\right.$ time. An SH group of D189C/C1124Y was probably converted to a sulfinate moiety by KI-treatment (26). No free Cys residue was confirmed using Ellman's titration method (33). As shown in Fig. 2E, dextranolytic activity increased and reached its greatest value at 250 mm KI (0.522 unit/mg; 6.8-fold higher than the original activity of D189C/C1124Y), followed by reduction by incubation, with KI exceeding at more than 250 mm.

DISCUSSION

Relationship between Structure and Function of PsDex

In C-VR of PsDex are three tandem 55-residue repeated SLHDs (Tyr¹⁵¹⁸–Thr¹⁵⁶⁰, Phe¹⁵⁷⁸–Ala¹⁶²⁰, and Tyr¹⁶⁴²–Ala¹⁶⁸⁵) that share great similarity to the N-terminal three-repeat SLHDs (Phe³⁴–Ala¹⁹⁷) of the Bacillus anthracis surface array protein (Fig. 4A). This array protein remains at the cell surface layer of Gram-positive bacteria by SLHD-associated non-covalent binding to a secondary cell wall carbohydrate in the bacterial cell wall (34). SLHDs of PsDex seem to be active because 97% dextranolytic activity of the native enzyme during cultivation was found on the cell surface of Paenibacillus sp. of Gram-positive bacteria. A possible linker of PSRR connects SLHR to the C-terminal conserved region. Both recombinant PsDex and PsDex-CT exhibited properties identical to those of the native enzyme, indicating that the C-terminal PSRR and three SLHRs of PsDex do not affect any catalytic functions.

The PsDex-specific region consists of three regions of LI-1 (Ile⁴⁵¹–Glu⁸¹²), CIT-SR (Ser⁸¹³–Val⁹⁴⁶), and LI-2 (Val⁹⁴⁷–Val¹⁰⁸²), and CIT-SR possesses a putative CBM-35 (PsDR1 in Fig. 4B). On the other hand, CITase carries two CBM-35-similar portions (BcCR1 and BcCR2 in Fig. 4B) (35, 36) at its CIT-SR and C-VR, respectively (7). The BcCR1-homologous region is found at CIT-SR of PsDex, although a region similar to BcCR2 is not found in PsDex. Although the BcCR2-homologous sequence is missing at PsDex, a putative CBM-6 is observed at C-VR of PsDex (PsDR2 in Fig. 4C, having about 20% homology at β-agarase from Saccharophagus degradans strain 2-40 (EMBL ID CP000282) (37) and α-agarase from Alteromonas agarilytica (GenBank number AAF26838.1) (38)), indicating that PsDex selected CBM-6 during protein evolution. Interestingly, Funane et al. (7) predicted that CBM-35 at CIT-SR (BcCR1) catalytically contributes to CI production of CITase. This prediction prompted us to study CI production using PsDex and SmDex with and without CIT-SR, respectively. Obviously, both PsDex and PsDex-CT formed CI-7 to CI-14 from dextran (Fig. 2C), whereas SmDex did not display the production of any cyclic sugars, implying that PsDex-carrying CIT-SR might contribute to the production of CIs, such as CITase (7). Because PsDR2-defective PsDex-CT produced CIs, the PsDR2 that was similar to CBM-6 did not contribute to cyclic sugar formation. These findings allow us to classify GH-66 members into three groups: (i) type 1 enzyme, only catalyzing dextranolytic reaction (e.g. SmDex); (ii) type 2 enzyme, exhibiting both hydrolytic and CI-forming activity (e.g. PsDex); and (iii) type 3 enzyme, exhibiting CI-producing activity with low dextranolytic activity (e.g. CITase). Our attempts to find more type 2 enzymes among newly categorized GH-66 members are currently under way.

Interestingly, PsDex is a glycoprotein with only galactose units. Many N-glycosylated and/or O-glycosylated proteins have been discovered among ∼25 kinds of bacterial genera (39). To our knowledge, PsDex is the first glycoprotein in the genus Paenibacillus that contains only galactose. The O-glycosylation at the Tyr residue has been reported on an S-layer protein from Paenibacillus alvei, which has a polysaccharide composed of Glc, Gal, ManNAc, and Rha (40). The galactose residue in PsDex is not considered to be involved in dextranolytic activity because the enzyme activity is available in the E. coli-expressing wild-type PsDex. This host lacks the glycosylation system.

Catalytic Residues and Their Functions

This study determined, for the first time, the catalytic residues of GH-66 enzyme and their functions. The three findings of (i) point mutation approaches at Asp¹⁸⁹, Asp³⁴⁰, and Glu⁴¹² (Table 2), (ii) ChR reaction using mutated enzymes (Table 2 and Fig. 3), and (iii) our three-dimensional structure analysis (9, 10) are evidence that Asp³⁴⁰ and Glu⁴¹² are catalytic residues. Reaction products derived from ChR (Fig. 3, A–C) clearly revealed that Asp³⁴⁰ is a so-called catalytic nucleophile, and Glu⁴¹² is an acid/base catalyst by virtue of the formation of β-isomaltotetraosyl azide and α-isomaltotetraosyl azide, respectively.

Asp¹⁸⁹ exhibited the interesting ChR phenomena involving catalysis of the hydrolytic reaction (Fig. 3, D and E), meaning that Asp¹⁸⁹ mutants display the ordinal ChR reaction simply by occupying the Asp¹⁸⁹ position with an external anion of N₃⁻, formate, or NO₃⁻. We thought that an Asp¹⁸⁹-mutated enzyme (D189A) would catalyze the transglycosylation on IG4F because D189A becomes a simple hydrolase in the presence of an anion. Surprisingly, a dextran-type polysaccharide was formed (Fig. 3F), indicating that D189A catalyzed a “sequential ChR reaction”; this, to our knowledge, is a novel discovery in the endolytic hydrolase reaction reacting on α-glucan. The reaction mechanism of sequential ChR still cannot be explained clearly, but the produced short IG-n was used for further D189A-mediated ChR. A negative charge is essential for the position of Asp¹⁸⁹ because D189C/C1124Y enhanced its dextranolytic activity by KI treatment (Fig. 2E), which oxidizes the SH-group of Cys¹¹²⁴ to form a possible anionic sulfinate (26).