Characterization of Ruminococcus albus cellodextrin phosphorylase and identification of a key phenylalanine residue for acceptor specificity and affinity to the phosphate group

Authors


Abstract

Ruminococcus albus has the ability to intracellularly degrade cello-oligosaccharides primarily via phosphorolysis. In this study, the enzymatic characteristics of R. albus cellodextrin phosphorylase (RaCDP), which is a member of glycoside hydrolase family 94, was investigated. RaCDP catalyzes the phosphorolysis of cellotriose through an ordered ‘bi bi’ mechanism in which cellotriose binds to RaCDP before inorganic phosphate, and then cellobiose and glucose 1-phosphate (Glc1P) are released in that order. Among the cello-oligosaccharides tested, RaCDP had the highest phosphorolytic and synthetic activities towards cellohexaose and cellopentaose, respectively. RaCDP successively transferred glucosyl residues from Glc1P to the growing cello-oligosaccharide chain, and insoluble cello-oligosaccharides comprising a mean of eight residues were produced. Sophorose, laminaribiose, β-1,4-xylobiose, β-1,4-mannobiose and cellobiitol served as acceptors for RaCDP. RaCDP had very low affinity for phosphate groups in both the phosphorolysis and synthesis directions. A sequence comparison revealed that RaCDP has Gln at position 646 where His is normally conserved in the phosphate binding sites of related enzymes. A Q646H mutant showed approximately twofold lower apparent Km values for inorganic phosphate and Glc1P than the wild-type. RaCDP has Phe at position 633 corresponding to Tyr and Val in the +1 subsites of cellobiose phosphorylase and N,N′-diacetylchitobiose phosphorylase, respectively. A F633Y mutant showed higher preference for cellobiose over β-1,4-mannobiose as an acceptor substrate in the synthetic reaction than the wild-type. Furthermore, the F633Y mutant showed 75- and 1100-fold lower apparent Km values for inorganic phosphate and Glc1P, respectively, in phosphorolysis and synthesis of cellotriose.

Abbreviations
CBP

cellobiose phosphorylase

CDP

cellodextrin phosphorylase

ChBP

N,N′-diacetylchitobiose phosphorylase

GH

glycoside hydrolase

Glc1P

glucose 1-phosphate

RaCDP

Ruminococcus albus NE1 CDP

Introduction

Ruminococcus albus is an important ruminal bacterium that is involved in the digestion of dietary cellulose [1]. This bacterium possesses various cellulolytic enzymes, including a cellulase (EC 3.2.1.4) [2], β-glucosidase (EC 3.2.1.21) [3] and cellobiose phosphorylase (CBP) (EC 2.4.1.20) [4]. It was shown that, R. albus intracellularly degrades cello-oligosaccharides mainly through phosphorolysis catalyzed by CBP and cellodextrin phosphorylase (CDP) (EC 2.4.1.49) rather than hydrolysis catalyzed by β-glucosidase [5]. Recently, R. albus CBP has been characterized [4], but the enzymatic properties of CDP from this bacterium remain unclear.

CDP, which was first identified in cell-free extracts of Clostridium thermocellum [6], catalyzes the reversible phosphorolysis of cello-oligosaccharides to α-glucose 1-phosphate (Glc1P) and cello-oligosaccharides with reduced chain lengths. CDPs from C. thermocellum [7-9] and Clostridium stercorarium [10] have been characterized. They have the highest phosphorolytic activities towards cellotriose and cellotetraose, respectively.

According to the sequence-based classification system of carbohydrate-active enzymes [11], CDP is a member of glycoside hydrolase (GH) family 94, together with CBP, N,N′-diacetylchitobiose phosphorylase (ChBP) (EC 2.4.1.280), the C-terminal region of cyclic β-1,2-glucan synthase, which has phosphorolytic activity towards β-1,2-glucooligosacccharides [12], and laminaribiose phosphorylase (EC 2.4.1.31). The catalytic domains of GH family 94 enzymes are formed by an (α/α)6 barrel [13-15], which has a fold similar to the catalytic domain of clan GH-L enzymes, including glucoamylase and maltose phosphorylase [16, 17]. Inorganic phosphate nucleophilically attacks the anomeric carbon of the glycosyl residue at the −1 subsite with the assistance of the conserved Asp, which acts as a general acid catalyst to donate a proton to the glycosidic oxygen [18].

In the genome of R. albus 7, two genes encoding GH family 94 proteins are found. The deduced amino acid sequence of Rumal_0187 is completely identical to that of CBP from R. albus NE1 [4]. On the other hand, it is difficult to predict the function of another gene, Rumal_2403, because its deduced amino acid sequence is not closely related to that of any known GH family 94 enzymes (Fig. 1). However, it was demonstrated that the cell-free extract of R. albus B199 grown on cellodextrin shows CDP activity [5], thus the CDP gene is presumably encoded in the genome of R. albus. We considered Rumal_2403 to be a possible candidate gene encoding CDP, although the shared homology of the deduced amino acid sequence of Rumal_2403 with CDPs from C. stercorarium and C. thermocellum is low (36% and 15%, respectively). In this study, the gene corresponding to the Rumal_2403 gene of R. albus 7 (the RaCDP gene) was cloned from R. albus NE1 [19]. The recombinant protein produced in Escherichia coli was characterized in detail, and the functions of some amino acid residues involved in the formation of acceptor and phosphate-binding sites are discussed based on the results of mutational analyses.

Figure 1.

Phylogenetic tree of the characterized GH family 94 enzymes. Multiple-sequence alignment was performed using clustalw (http://www.genome.jp/tools/clustalw/), and a phylogenetic tree was constructed using mega5 (http://www.megasoftware.net/). RaCDP, Ruminococcus albus NE1 cellodextrin phosphorylase (CDP; GenBank accession number ADU22883.1); CsCDP, Clostridium stercorarium CDP (GenBank accession number AAC45511.1); CtCDP, Clostridium thermocellum YM4 CDP (GenBank accession number BAB71818.1); CgCBP, Cellvibrio gilvus cellobiose phosphorylase (CBP) (GenBank accession number BAA28631.1); RaCBP, R. albus NE1 CBP (GenBank accession number ADU20744.1); CtCBP, C. thermocellum YM4 CBP (GenBank accession number AAL67138.1); TmCBP, Thermotoga maritima MSB8 CBP (GenBank accession number AAD36910.1); CuCBP, Cellulomonas uda CBP (GenBank accession number AAQ20920.1); AfCBGS, the C-terminal part of Agrobacterium fabrum C58 cyclic β-1,2-glucan synthetase (compared region was 1400–2832; GenBank accession number AAK73356.1); AvCBGS, the C-terminal part of Agrobacterium vitis F2/5 cyclic β-1,2-glucan synthetase (compared region was 1401–2831; GenBank accession number AAQ08605.1); PsLBP, Paenibacillus sp. YM1 laminaribiose phosphorylase (GenBank accession number BAJ10826.1); AlLBP, Acholeplasma laidlawii PG-8A laminaribiose phosphorylase (GenBank accession number ABX81345.1); VpChBP, Vibrio proteolyticus N,N′-diacetylchitobiose phosphorylase (ChBP) (GenBank accession number BAC87867.1); VfChBP, Vibrio furnissii ChBP (GenBank accession number AAG23740.1)

Results and Discussion

Production, purification and basic properties of recombinant RaCDP

The RaCDP gene of R. albus NE1 was obtained by PCR using primers designed based on the Rumal_2403 gene of R. albus 7. The sequence of the amplified DNA fragment was completely identical to the Rumal_2403 gene of R. albus 7. Cell-free extract of the E. coli transformant over-expressing the RaCDP gene showed phosphorolytic activity towards cellotriose. The recombinant enzyme was purified to homogeneity by Ni-chelating column chromatography. From 0.5 L of culture medium, 25.1 mg of the purified enzyme (20.4 U·mg−1) was obtained. The molecular masses of the purified enzyme measured by SDS/PAGE and gel filtration column chromatography were 95 and 177 kDa, respectively, indicating that RaCDP forms a homodimer in solution as observed for other GH family 94 enzymes [4, 18, 20, 21]. RaCDP showed highest activity at pH 6.0 and 50 °C, and was stable in the pH range 5.8–8.3 and below 40 °C.

Kinetic mechanism of phosphorolysis of cellotriose by RaCDP

Phosphorolytic reaction velocities towards various concentrations of cellotriose and inorganic phosphate were measured (Fig. 2). The lines obtained from double reciprocal plots for 1/[cellotriose] versus 1/v at various concentrations of inorganic phosphate crossed at a certain point, indicating that the reaction occurred through a sequential ‘bi bi’ mechanism. To determine the order of substrate binding and product release, product inhibition analysis of the phosphorolysis of cellotriose was performed (Fig. 3). Competitive inhibition by Glc1P was observed in the reaction at varying concentrations of cellotriose and a fixed concentration of inorganic phosphate. Mixed-type inhibition was observed under conditions of varying concentrations of cellotriose (where the concentration of inorganic phosphate was fixed) versus cellobiose and varying concentrations of inorganic phosphate (where the concentration of cellotriose was fixed) versus cellobiose or Glc1P. Thus, during phosphorolysis of cellotriose, cellotriose and inorganic phosphate bind to RaCDP in that order, and Glc1P is released after cellobiose. The kinetic parameters for the phosphorolysis of cellotriose were determined as follows: kcat = 91.2 ± 3.4 s−1, KmA = 5.41 ± 0.88 mm, KmB = 124 ± 12 mm and KiA = 12.6 ± 1.6 mm (A, cellotriose; B, inorganic phosphate). Generally, GH family 94 enzymes have low Km values for inorganic phosphate (0.14–1.5 mm) [4, 18, 20-24], but RaCDP had a significantly higher Km value than the known enzymes (described below in detail). In the synthetic reaction, RaCDP also had a high Km(app) value for Glc1P (166 ± 10 mm), which was determined by measuring the reaction rates towards various concentrations of Glc1P and 40 mm cellobiose. RaCDP had a much higher kcat value than the characterized CDPs [10, 25], thus this enzyme may be functional in the degradation of cello-oligosaccharides in vivo.

Figure 2.

Double-reciprocal plots of the phosphorolysis of cellotriose in the presence of various concentrations of inorganic phosphate. Phosphorolytic velocities at 4–20 mm cellotriose and a fixed concentration of sodium phosphate buffer (pH 6.0) were measured. The concentrations of inorganic phosphate were 80 mm (black circle), 160 mm (white circle), 240 mm (black triangle) and 400 mm (white triangle). Values are means ± standard deviation for three independent experiments.

Figure 3.

Product inhibition analysis for the phosphorolysis of cellotriose. Inhibition of phosphorolysis of cellotriose by Glc1P and cellobiose was analyzed. The fixed concentrations of inorganic phosphate and cellotriose were 50 mm (A, C) and 20 mm (B, D), respectively. (A) Inhibition of Glc1P against cellotriose; (B) inhibition of Glc1P against inorganic phosphate; (C) inhibition of cellobiose against cellotriose; (D) inhibition of cellobiose against inorganic phosphate. Circles, no inhibitor; black triangle, 10 mm Glc1P; black square, 20 mm Glc1P; white triangle, 10 mm cellobiose; white square, 20 mm cellobiose. Values are means ± standard deviation for three independent experiments. In (A), data were fitted to the equation for competitive inhibition; in (B–D), data were fitted to the equation for mixed inhibition.

Substrate specificity of RaCDP in the phosphorolytic and synthetic reactions

The apparent kinetic parameters for phosphorolysis of a series of cello-oligosaccharides were determined in the presence of 250 mm inorganic phosphate (Table 1). RaCDP had phosphorolytic activity towards cello-oligosaccharides longer than cellobiose. The Km(app) value decreased with an increase in substrate chain length, and the kcat(app) value for cellotetraose was the highest among the cello-oligosaccharides tested. Cellohexaose was the best substrate in terms of its kcat(app)/Km(app). The kcat(app)/Km(app) for this substrate was 4.2-fold higher than towards cellotriose. Cellotetraose and cellotriose are the best substrates for the phosphorolytic reactions catalyzed by CDPs from C. stercorarium and C. thermocellum, respectively [10, 25], thus RaCDP has greater preference for long-chain substrates than the enzymes described previously.

Table 1. Apparent kinetic parameters of phospholytic and synthetic reactions of RaCDP. Initial reaction velocities for phosphorolysis were measured at varying concentrations of cellooligosaccharides in the presence of 250 mm inorganic phosphate. Those for synthesis were measured at varying concentrations of acceptor substrates in the presence of 250 mm Glc1P. ND, not detected at 40-fold higher enzyme concentration than for other substrates. NT, not tested
SubstratePhosphorolysisSynthesis
kcat(app) (s−1)Km(app) (mm)kcat(app) /Km(app) (s−1 mm−1)kcat(app) (s−1)Km(app) (mm)kcat(app)/Km(app) (s−1 mm−1)
CellobioseNDNDND47.1 ± 2.613.2 ± 1.13.56
Cellotriose76.2 ± 0.86.04 ± 0.2212.643.7 ± 1.05.01 ± 0.268.71
Cellotetraose92.8 ± 2.64.16 ± 0.2522.337.8 ± 0.83.97 ± 0.279.50
Cellopentaose83.8 ± 1.42.41 ± 0.1234.828.9 ± 0.72.73 ± 0.1610.6
Cellohexaose55.8 ± 0.61.04 ± 0.1153.418.2 ± 0.53.22 ± 0.265.63
SophoroseNTNTNT14.2 ± 0.7343 ± 180.0414
LaminaribioseNTNTNT33.3 ± 3.5119 ± 190.281
XylobioseNTNTNT15.4 ± 0.650.9 ± 3.80.302
MannobioseNTNTNT41.3 ± 1.765.0 ± 3.70.635
CellobiitolNTNTNT3.26 ± 0.1273.7 ± 7.70.0443

In synthetic reactions in the presence of 250 mm Glc1P, all cello-oligosaccharides tested served as acceptor substrates, although synthetic activity towards d-glucose was not observed, consistent with the finding that RaCDP did not phosphorolyze cellobiose. RaCDP had the highest kcat(app)/Km(app) value toward cellopentaose. This result was in agreement with the fact that RaCDP had the highest phosphorolytic activity towards cellohexaose. The kcat(app)/Km(app) value towards cellohexaose was approximately half of that towards cellopentaose. This indicates that the acceptor binding site, which binds to the reducing-end part of the substrate from the scissile bonds, has five subsites with positive affinity. Sophorose, laminaribiose, β-1,4-xylobiose, β-1,4-mannobiose and cellobiitol are also acceptor substrates of RaCDP. The kcat(app)/Km(app) values toward β-1,4-mannobiose, β-1,4-xylobiose and laminaribiose were 18%, 8.5% and 7.9% of that towards cellobiose, respectively. Cellobiitol was a very poor acceptor for RaCDP, indicating that a closed glucose ring is recognized at the +2 subsite of the enzyme. Gentiobiose, lactose, N,N′-diacetylchitobiose and α-linked glucobioses did not serve as acceptor substrates.

Production of cello-oligosaccharides via the synthetic reaction of cellobiose and Glc1P

Production of cello-oligosaccharides from 250 mm Glc1P and 50 mm cellobiose was monitored (Fig. 4). At the initial stage of the reaction, cellotriose was produced rapidly, reaching a maximum at approximately 20 min. Cellotetraose, cellopentaose and cellohexaose were also produced, indicating that RaCDP successively transfers a glucose moiety to the newly produced oligosaccharides. The highest concentrations of cellotriose, cellotetraose, cellopentaose and cellohexaose obtained under the analytical conditions were 11.2 mm (40 min), 5.8 mm (60 min), 2.8 mm (80 min), and 1.3 mm (80 min), respectively. The yield of cello-oligosaccharides calculated from the reacted cellobiose increased during the incubation time analyzed. After a reaction of 60 min, the yield was 52%, which is comparable to that for C. thermocellum ATCC27405 CDP. The yields of cello-oligosaccharides from 200 mm cellobiose and Glc1P and from 20 mm cellobiose and 80 mm Glc1P has been reported as 48% and 54%, respectively [7, 9]. The cellobiose concentration decreased with a longer reaction time, and the concentration of cellobiose at 80 min was 13.3 mm, indicating that the yield of cello-oligosaccharides was 73%. At 100 min, insoluble material was observed in the reaction mixture, similar to the reaction catalyzed by C. thermocellum CDP [25]. To analyze the chemical structure of the insoluble material, the synthetic reaction with a fivefold increase of enzyme was performed for 3 h. The products collected by centrifugation (weight 7 mg) were dissolved in 4% NaOD, and the 1H-NMR was recorded. The spectrum of this product completely corresponded with that of the hydrolysate of cellulose [26], indicating that the insoluble product is a mixture of cello-oligosaccharides. Based on the relative peak areas of the 1-H of glucose at the reducing end and other glucose residues, the mean degree of polymerization was estimated to be 8. From this length of cello-oligosaccharides and amount of cellobiose used, the yield was estimated to be 11%.

Figure 4.

Time course of the synthesis of cello-oligosaccharides by RaCDP. The synthesis of cello-oligosaccharides from 50 mm cellobiose and 250 mm Glc1P was monitored. The black circle, white circle, black triangle and white triangle show cellotriose, cellotetraose, cellopentaose, and cellohexaose, respectively. The black square indicates the yield of cello-oligosaccharides calculated from amount of the reacted cellobiose. Values are means ± standard deviation for three independent experiments.

Comparison of amino acid sequences between RaCDP and other GH94 enzymes

The structure of the binding site of inorganic phosphate has been analyzed for several GH family 94 enzymes [13-15]. Highly conserved amino acid residues corresponding to Arg351, His666, Gln712, Thr731, Gly732 and Thr733 of Cellvibrio gilvus CBP are involved in formation of this site. RaCDP has Gln646 in the position equivalent to His666 of C. gilvus CBP, although the other amino acid residues forming the inorganic phosphate binding site are well conserved (Fig. S1).

In ChBP from Vibrio proteolyticus, Val631 situated at the +1 subsite forms a small hydrophobic pocket to accommodate the N-acetyl group of N-acetyl-d-glucosamine. Most GH family 94 enzymes have a bulky residue, Tyr, at the corresponding position, thus Val631 of V. proteolyticus ChBP is thought to be a crucial residue for binding to the N-acetyl-d-glucosamine residue at the +1 subsite [13, 15]. In the case of CBP, the Tyr residue corresponding to Val631 of V. proteolyticus ChBP participates in a hydrogen-binding interaction with the 2-OH group of the glucose residue at the +1 subsite [13]. RaCDP has Phe633 at the corresponding position (Fig. S1), implying that RaCDP has different acceptor specificity from other GH family 94 enzymes at the 2-OH position of a glycosyl residue bound to the +1 subsite. In fact, the kcat(app)/Km(app) for β-1,4-mannobiose was 18% of that of cellobiose, making this a relatively good acceptor substrate for RaCDP, although interactions with the +2 subsite may contribute to the binding of β-1,4-mannobiose to the acceptor binding site. The synthetic activity of other known CDPs towards β-1,4-mannobiose has not been investigated. For CBPs, d-mannose is a poor acceptor substrate. The apparent kcat/Km values of CBPs for the synthetic reaction of d-mannose are < 1.2% of those for d-glucose [4, 22-24, 27]. Although the OH group of the Tyr side chain is conserved in most GH family 94 enzymes other than ChBP, it may result in an unfavorable interaction with the axial OH group at the C2 position of the glycosyl residue at the +1 subsite.

Functional analysis of the inorganic phosphate binding site of RaCDP

Gln646 of RaCDP, corresponding to the conserved His involved in binding to inorganic phosphate as described above, was replaced by His by site-directed mutagenesis. The mutant enzyme was produced and purified as for the wild-type enzyme. The apparent kinetic parameters for inorganic phosphate and Glc1P in the direction of phosphorolysis and synthesis, respectively, were determined. The kcat(app) values of the mutant enzyme (Q646H) for inorganic phosphate and Glc1P were 9.5% and 3.9% of those of the wild-type, respectively, while the Km(app) values for inorganic phosphate and Glc1P were less than half of those of the wild-type (Table 2), indicating that Gln646 is partly responsible for the high Km(app) values for inorganic phosphate and Glc1P. However, the Q646H mutant still has much lower affinity toward phosphate groups than the known GH family 94 enzymes. Substitution of Gln646 by His decreased the kcat(app) of RaCDP for the synthetic reaction more than for the phosphorolysis. His666 of C. gilvus CBP is located at a position far from the acceptor binding site, thus Gln646 of RaCDP, corresponding to His666 of C. gilvus CBP, does not appear to have direct interactions with acceptor substrates. Introduction of His at position 646 of RaCDP may indirectly alter the interaction with the acceptor substrate, and decrease the synthetic activity.

Table 2. Comparison of apparent kinetic parameters of inorganic phosphate and Glc1P between the wild-type and mutated enzymes
SubstrateInorganic phosphateaGlc1P b
kcat(app) (s−1)Km(app) (mm)kcat(app) /Km(app) (s−1 mm−1)kcat(app) (s−1)Km(app) (mm)kcat(app) /Km(app) (s−1 mm−1)
  1. a

    Phosphorolytic velocities towards various concentrations of inorganic phosphate were measured in the presence of 5 mm cellotriose.

  2. b

    Synthetic velocities towards various concentrations of Glc1P and 40 mm cellobiose were measured.

Wild-type11.8 ± 0.1181 ± 50.065239.6 ± 1.5166 ± 1000.239
F633Y0.769 ± 0.0112.40 ± 0.110.3201.93 ± 0.040.155 ± 0.01512.5
Q646H1.12 ± 0.02780.2 ± 5.80.01401.53 ± 0.0775.1 ± 3.40.0204

Functional analysis of the acceptor binding site of RaCDP

Phe633 of RaCDP, corresponding to the conserved Tyr and Val of CBP and ChBP, respectively, was replaced by Tyr (F633Y). In the phosphorolysis of cellotriose, the F633Y mutant had 3.6- and 4.9-fold lower kcat(app) and Km(app) values for cellotriose, respectively, than the wild-type, resulting in a 1.4-fold higher kcat(app)/Km(app) (Table 3). Apparent kinetic parameters for the acceptor substrates, cellobiose and β-mannobiose, were measured in the presence of 250 mm Glc1P (Table 3). F633Y had 1.1- and 2.2-fold higher Km(app) values for β-mannobiose and cellobiose, respectively, than those of the wild-type. The kcat(app) values for β-mannobiose were more severely decreased than for cellobiose. The kcat(app) values for cellobiose and β-mannobiose were 16.5- and 145-fold lower, respectively, than those of wild-type, resulting in lower kcat(app)/Km(app) ratios for β-mannobiose and cellobiose, indicating that F633Y has a lower preference for β-mannobiose than the wild-type. The side-chain OH group of the introduced Tyr633 may cause steric hindrance or unfavorable interactions at the +1 subsite when binding β-mannobiose as an acceptor.

Table 3. Comparison of apparent kinetic parameters between the wild-type and F633Y. Apparent kinetic parameters for phosphorolysis of cellotriose, determined in the presence of 20 mm inorganic phosphate, are shown. The apparent kinetic parameters for synthetic reactions towards cellobiose and β-1,4-mannobiose are the values determined from the reaction rates towards various concentrations of the acceptor substrates and 250 mm Glc1P
SubstrateF633YWild-type
kcat(app) (s−1)Km(app) (mm)kcat(app) /Km(app) (s−1 mm−1)kcat(app) (s−1)Km (app) (mm)kcat(app) /Km(app) (s−1 mm−1)
Cellotriose0.821 ± 0.0190.855 ± 0.0750.9602.93 ± 0.084.18 ± 0.170.700
Cellobiose2.86 ± 0.2028.9 ± 3.00.099247.1 ± 2.613.2 ± 1.13.56
β-1,4-mannobiose0.285 ± 0.02572.4 ± 11.00.0039641.3 ± 1.765.0 ± 3.70.635

Surprisingly, F633Y had significantly higher affinity for phosphate than the wild-type. In the phosphorolysis of cellotriose, F633Y had a 75-fold lower Km(app) value for inorganic phosphate, 2.40 ± 0.11 mm, which is close to the values for known GH family 94 enzymes (Table 2). The kcat(app) for inorganic phosphate was 15-fold lower, but the kcat(app)/Km(app) value for inorganic phosphate was 4.9-fold higher, because of the very low Km(app) value. This mutant enzyme also had much higher affinity for Glc1P in the synthetic reaction than the wild-type. The Km(app) value for Glc1P was 1100-fold lower than for the wild-type. The kcat(app)/Km(app) value for Glc1P was 52-fold higher than the wild-type, although the kcat(app) for Glc1P was 20.5-fold lower than the wild-type. Phe633 of RaCDP is predicted to be far from the phosphate-binding site based on the three-dimensional structures of related enzymes, and this residue does not appear to have a direct interaction with inorganic phosphate and the phosphate group of Glc1P. In the complex of C. gilvus CBP, inorganic phosphate, d-glucose (+1 subsite) and glycerol (−1 subsite), the OH group of the side chain of Tyr653 appears to participate in a hydrogen-bonding interaction with Lys658, forming a hydrogen bond to Gln712, and thus is involved in formation of the phosphate-binding site. The orientation of Gln712, which is dependent on the interaction with Lys658, may be crucial for the observed high affinity for phosphate. In RaCDP, Lys638 and Gln692 are predicted to be located at positions equivalent to Lys658 and Gln712 of C. gilvus CBP, respectively. The side chain of Phe633 cannot form a hydrogen bond to Lys638, which is predicted to regulate the orientation of Gln692, thus the orientation of Gln692 of RaCDP might be unfavorable for binding to the phosphate group. Amino acid residues corresponding to Lys658 and Gln712 of C. gilvus CBP are conserved in most GH family 94 enzymes (Fig. S1), and the hydrogen bonding network, as discussed above, probably plays an important role in the high affinity for the phosphate group. CDP from C. thermocellum has Arg and Ser at the positions equivalent to Lys658 and Gln712 of C. gilvus CBP, respectively, but, in the synthetic direction, this enzyme has a Km(app) value for Glc1P of 4.7 mm [6], which is similar in magnitude to the values of most GH family 94 enzymes. Two characterized laminaribiose phosphorylases have a Phe residue at a position corresponding to Tyr653 of C. gilvus CBP, but these enzymes have low Km values for inorganic phosphate (0.14–0.4 mm) for phosphorolysis of laminaribiose, unlike RaCDP [21, 28]. These enzymes show very low sequence identity to other GH family 94 enzymes including RaCDP, and do not have an amino acid residue corresponding to Gln712 of C. gilvus CBP (Fig. S1). These structural differences suggest that different strategies for optimization of binding to the phosphate group may exist in these enzymes.

Possible functions of RaCDP in the metabolism of carbohydrates

RaCDP catalyzes synthetic reactions of both cello-oligosaccharides and β-1,4-mannooligosaccharides, indicating that RaCDP also phosphorolyzes complex oligosaccharides in which cello-oligosaccharides are linked to the non-reducing end of β-1,4-mannooligosaccharides (Fig. 5). These oligosaccharides may be produced by hydrolysis of glucomannan with glucosyl residues in the main chain. RaCDP produces a β-1,4-manno-oligosaccharide through phosphorolysis of such complex oligosaccharides. The β-1,4-manno-oligosaccharide produced is further metabolized by phosphorolysis and epimerization as shown previously [29]. β-1,4-manno-oligosaccharides longer than β-1,4-mannobiose are phosphorolyzed to β-1,4-mannobiose by β-1,4-manno-oligosaccharide phosphorylase, and β-1,4-mannobiose is epimerized to 4-O-β-d-mannosyl-d-glucose by cellobiose 2-epimerase (EC 5.1.3.11) for further phosphorolysis catalyzed by 4-O-β-d-mannosyl-d-glucose phosphorylase (EC 2.4.1.281). β-1,4-manno-oligosaccharide phosphorylase has broad acceptor specificity in the synthetic direction, similar to RaCDP, and cellobiose is a good acceptor substrate, indicating that complex oligosaccharides, in which a β-1,4-mannooligosaccharide is bound to the non-reducing end of a cello-oligosaccharide, are substrates for phosphorolysis catalyzed by this enzyme. β-1,4-manno-oligosaccharide phosphorylase may produce cello-oligosaccharides from this type of complex oligosaccharide for further phosphorolysis catalyzed by CDP and CBP.

Figure 5.

Possible degradation mechanism of complex oligosaccharides derived from glucomannan. Left pathway: the cello-oligosaccharide part, which is linked at the non-reducing end of an oligosaccharide, is phosphorolyzed by CDP. If longer than β-1,4-mannobiose, the resulting β-1,4-manno-oligosaccharide is further phosphorolyzed by β-1,4-manno-oligosaccharide phosphorylase (MP2). β-1,4-mannobiose is epimerized by cellobiose 2-epimerase (CE), and phosphorolyzed by 4-O-β-d-mannosyl-d-glucose phosphorylase (MP1). Right pathway: the β-1,4-manno-oligosaccharide part, which is linked at the non-reducing end of an oligosaccharide, is phosphorolyzed by MP2. The cello-oligosaccharide liberated is degraded by CDP and CBP.

Experimental procedures

Preparation of expression plasmids for wild-type and mutant RaCDPs

The RaCDP gene was amplified by PCR using genomic DNA of R. albus NE1 as the template, primers 5′-CTTGTAAAAAATGCGGATATATG-3′ and 5′-GGACTGATATGACCGAATAG-3′, designed based on the Rumal_2403 gene, and Primestar HS DNA polymerase (Takara Bio, Otsu, Japan). The amplified DNA fragment was cloned into the pBluescript II SK(+) vector (Stratagene, La Jolla, CA, USA) via the EcoRV site. The DNA sequence of the amplified region was analyzed using an ABI Prism 310 DNA sequencer (Applied Biosystems, Foster City, CA, USA). This plasmid was used as the template for PCR to construct an expression plasmid for RaCDP. The DNA fragment, amplified using primers 5′-GAGCATATGACTATGCAGTATGGATATTT-3′ and 5′-GAACTCGAGGCCCATAACTACGGTGAT-3′, harboring the NdeI and XhoI sites, was cloned into the NdeI and XhoI sites of the pET-23a vector (Novagen, Darmstadt, Germany).

The expression plasmids for Q646H and F633Y were prepared using a Primestar mutagenesis basal kit (Takara Bio). The expression plasmid for the wild-type was used as the template, and the sequences of primers were 5′-TTCAGCCACACACAGGGCTGGATAATC-3′ and 5′-CTGTGTGTGGCTGAATATACCGCCGTT-3′ for Q646H, and 5′-CACATATACAACCCCGACACCAAGGAG-3′ and 5′-GGGGTTGTATATGTGCATCAGCGCACC-3′ for F633Y.

Production and purification of the wild-type and mutant RaCDPs

Transformants of E. coli BL21 (DE3) cells harboring the expression plasmid for each RaCDP derivative were cultured in 500 mL Luria–Bertani medium supplemented with 50 μg·mL−1 ampicillin until the attenuance at 600 nm reached 0.4. Production of the recombinant protein was induced by addition of isopropyl β-d-thiogalactoside at a final concentration of 0.1 mm, and the incubation was continued with vigorous shaking at 18 °C for 24 h. The bacterial cells were harvested by centrifugation (7200 g) and suspended in 75 mL of 20 mm MES/NaOH buffer containing 0.5 m NaCl (pH 6.0, buffer A) , and disrupted by sonication using a Branson Sonifier 450 (Danbury, CT, USA). Cell debris was removed by centrifugation, and the cell-free extract obtained was applied to a Ni-chelating Sepharose column (column size, 1.6 cm I.D. × 5.0 cm, GE Healthcare, Uppsala, Sweden) equilibrated with buffer A. After thorough washing with buffer A, adsorbed protein was eluted with 0.5 m imidazole in buffer A. Highly purified fractions, confirmed by SDS/PAGE, were collected, and dialyzed against 20 mm MES/NaOH buffer (pH 6.0). The purified sample was frozen at −80 °C until analysis.

Protein assay

The protein concentrations of the crude extract and fractions from the column chromatography were measured by the Bradford method [30] and the UV method [31], respectively. BSA (Nacalai Tesque, Kyoto, Japan) was used as the standard protein for the Bradford method. The concentration of the purified enzyme was determined based on the concentration of each amino acid after complete acid hydrolysis. The purified enzyme was hydrolyzed in 6 M HCl at 110 °C for 24 h, and the resulting amino acids were measured by the ninhydrin colorimetric method using a JLC-500/V AminoTac amino acid analyzer (Jeol, Tokyo, Japan) [32].

Standard enzyme assay

Twenty microliters of a reaction mixture consisting of an appropriate concentration of enzyme, 50 mm sodium phosphate buffer (pH 6.0) and 20 mm cellotriose (Seikagaku, Tokyo, Japan) were incubated at 37 °C for 10 min. The enzyme solution was diluted using 20 mm MES/NaOH buffer (pH 6.0) containing 1 mg·mL−1 BSA. The enzyme reaction was terminated by addition of 20 μL of 4 m Tris/HCl buffer (pH 7.0), and the Glc1P produced was measured by the phosphoglucomutase/glucose-6-phosphate dehydrogenase method [33].

Effects of pH and temperature

The optimum pH and temperature were investigated by measuring phosphorolytic activities at various pH values and temperatures, respectively. For analysis of optimum pH, a reaction mixture consisting of an appropriate concentration of the enzyme, 100 mm reaction buffer, 20 mm sodium phosphate buffer (pH 6.0) and 20 mm cellotriose was incubated at 37 °C for 10 min, and liberated Glc1P was measured as described above. The reaction buffer comprised sodium acetate buffer for pH 2.5–5.5, MES/NaOH buffer for pH 5.2–6.7, and Tris/HCl buffer for pH 6.5–9.5.

Stable ranges of pH and temperature were determined based on the residual activity after pH and heat treatments, respectively. For the pH treatment, the enzyme was incubated at pH 2.0–12.0 using 100 mm Briton–Robinson buffer at 4 °C for 24 h. For the heat treatment, the enzyme was incubated in 62.5 mm sodium phosphate buffer (pH 6.0) for 15 min, and then immediately cooled on ice. The enzyme was considered to be stable in the ranges of pH and temperature over which the enzyme maintained more than 90% of its original activity.

Analysis of the kinetic mechanism of phosphorolysis of cellotriose

Phosphorolytic velocities at various concentrations of cellotriose (4–20 mm) and sodium phosphate buffer (pH 6.0, 80–400 mm) were measured using the standard enzyme assay as described above. The kinetic parameters were calculated by fitting the reaction rates to the following equation for a sequential ‘bi bi’ mechanism [34] using grafit version 7.0.2 (Erithacus Software, Horley, UK):

display math

where A is cellotriose and B is inorganic phosphate.

Product inhibition analysis was performed to determine the order of substrate binding and product release. The enzyme concentration was fixed at 100 nm. First, the phosphorolytic velocities towards 4–20 mm cellotriose and 50 mm sodium phosphate buffer (pH 6.0) in the presence of 0–20 mm Glc1P or cellobiose were measured. Then, the reaction rates towards 20 mm cellotriose and 20–100 mm sodium phosphate buffer (pH 6.0) in the presence of Glc1P or cellobiose were measured as described above. For the inhibition analysis by Glc1P, the cellobiose produced was degraded by CBP, and the resulting d-glucose was measured by the glucose oxidase/peroxidase method [35]. After stopping the reaction by heating the reaction mixture at 80 °C for 10 min, 20 μL of the reaction mixture was mixed with 100 μL of 60 mm sodium phosphate buffer (pH 6.0), 30 μL of 5.7 U·mL−1 R. albus NE1 CBP [4], and 20 μL of Glucose CII-Test Wako (Wako Pure Chemical Industries, Osaka, Japan), and incubated at 37 °C for 30 min. The absorbance at 505 nm was measured, and the cellobiose concentration was calculated based on the standard curve obtained with 0–0.5 mm cellobiose.

Phosphorolysis of cello-oligosaccharides

The reaction rates for phosphorolysis of various concentrations of cello-oligosaccharides and fixed concentration of inorganic phosphate were measured. Twenty microliters of a reaction mixture consisting of enzyme, 250 mm sodium phosphate buffer (pH 6.0) and 1–20 mm of each cello-oligosaccharide [cellobiose was purchased from Sigma (St Louis, MO, USA), and the other oligosaccharides were purchased from Seikagaku] was incubated at 37 °C for 10 min, and the liberated Glc1P was measured as described above. Concentrations of the wild-type and F633Y were 20 and 610 nm, respectively. The apparent kinetic parameters kcat(app) and Km(app) were calculated from the reaction rates obtained by fitting to the Michaelis–Menten equation.

Acceptor specificity in the synthetic reaction

The reaction velocities for the synthesis of various concentrations of oligosaccharides were measured in the presence of 250 mm Glc1P. Twenty microliters of a reaction mixture consisting of the enzyme (wild-type, 210 nm; F633Y, 490 nm for cellobiose and 3.1 μm for β-1,4-mannobiose), 250 mm Glc1P, 50 mm MES/NaOH buffer (pH 6.0) and 1–100 mm of each oligosaccharide (acceptor substrate) was incubated at 37 °C for 10 min, and the liberated inorganic phosphate was measured as described previously [36]. The apparent kinetic parameters were calculated as described above. A series of cello-oligosaccharides [sophorose (Sigma), laminaribiose (Megazyme, Wicklow, Ireland), xylobiose (Wako Pure Chemical Industries), β-1,4-mannobiose (Megazyme) and cellobiitol (Sigma)] were tested as acceptor substrates.

Apparent kinetic parameters for inorganic phosphate and Glc1P

The reaction rates for the phosphorolysis of cellotriose were measured at various concentrations of inorganic phosphate as described above to determine the apparent kinetic parameters for inorganic phosphate. The reaction mixture (20 μL) consisted of enzyme, inorganic phosphate, 5 mm cellotriose and 50 mm MES/NaOH buffer (pH 6.0). The concentrations of the wild-type, Q646H and F633Y were 63 nm, 7.7 μm and 0.61 μm, respectively. The concentrations of inorganic phosphate were 25–250 mm for the wild-type, 25–300 mm for Q646H, and 0.63–20 mm for F633Y.

To determine the apparent kinetic parameters for Glc1P, the reaction rates for the synthesis of cellotriose were measured as described above at a fixed concentration of cellobiose and various concentrations of Glc1P. The reaction mixture (20 μL) consisted of enzyme, Glc1P, 40 mm cellobiose and 50 mm MES/NaOH buffer (pH 6.0). The concentrations of the wild-type, Q646H and F633Y were 63 nm, 2.5 μm and 0.61 μm, respectively. The concentrations of Glc1P were 40–400 mm for the wild-type, 20–300 mm for Q646H, and 0.25–2.5 mm for F633Y.

Production of cello-oligosaccharides by the synthetic reaction

The synthetic reaction toward cellobiose and Glc1P was monitored. The reaction mixture (100 μL) consisting of 69 nm RaCDP, 50 mm cellobiose, 250 mm Glc1P and 100 mm MES/NaOH buffer (pH 6.0) was incubated at 37 °C for 0–60 min. The enzyme reaction was stopped by heating the reaction mixture at 80 °C for 10 min. Cello-oligosaccharides produced were measured by HPLC under the following conditions: injection volume, 10 μL; column, Asahipak NH2P-50 4E (column size, 4.6 cm I.D. × 250 mm; Shodex, Tokyo, Japan); column temperature, 40 °C; elution, descending linear gradient of 60–50% acetonitrile over 10 min; flow rate, 0.8 mL·min−1; detection, pulsed amperometry.

The structure of insoluble product was analyzed. The reaction mixture (1 mL) consisting of 343 nm RaCDP and the other components as described above was incubated at 37 °C for 3 h. After heating the mixture at 80 °C for 10 min, the insoluble material was collected by centrifugation. The insoluble product dried in vacuo was dissolved in 1 mL 4% NaOD/D2O, and the 1H-NMR was recorded using a Bruker AMX-500 spectrometer (Bruker Daltonics, Billerica, MA, USA).

Acknowledgements

We thank Eri Fukushi (GC-MS & NMR Laboratory, Faculty of Agriculture, Hokkaido University, Japan) and Tomohiro Hirose (Instrumental Analysis Division, Equipment Management Center, Creative Research Institution, Hokkaido University, Japan) for NMR analysis and amino acid analysis, respectively.

Ancillary