Enzymatic Mechanism for Arabinan Degradation and Transport in the Thermophilic Bacterium Caldanaerobius polysaccharolyticus

ABSTRACT

The plant cell wall polysaccharide arabinan provides an important supply of arabinose, and unraveling arabinan-degrading strategies by microbes is important for understanding its use as a source of energy. Here, we explored the arabinan-degrading enzymes in the thermophilic bacterium Caldanaerobius polysaccharolyticus and identified a gene cluster encoding two glycoside hydrolase (GH) family 51 α-l-arabinofuranosidases (CpAbf51A, CpAbf51B), a GH43 endoarabinanase (CpAbn43A), a GH27 β-l-arabinopyranosidase (CpAbp27A), and two GH127 β-l-arabinofuranosidases (CpAbf127A, CpAbf127B). The genes were expressed as recombinant proteins, and the functions of the purified proteins were determined with para-nitrophenyl (pNP)-linked sugars and naturally occurring pectin structural elements as the substrates. The results demonstrated that CpAbn43A is an endoarabinanase while CpAbf51A and CpAbf51B are α-l-arabinofuranosidases that exhibit diverse substrate specificities, cleaving α-1,2, α-1,3, and α-1,5 linkages of purified arabinan-oligosaccharides. Furthermore, both CpAbf127A and CpAbf127B cleaved β-arabinofuranose residues in complex arabinan side chains, thus providing evidence of the function of this family of enzymes on such polysaccharides. The optimal temperatures of the enzymes ranged between 60°C and 75°C, and CpAbf43A and CpAbf51A worked synergistically to release arabinose from branched and debranched arabinan. Furthermore, the hydrolytic activity on branched arabinan oligosaccharides and degradation of pectic substrates by the endoarabinanase and l-arabinofuranosidases suggested a microbe equipped with diverse activities to degrade complex arabinan in the environment. Based on our functional analyses of the genes in the arabinan degradation cluster and the substrate-binding studies on a component of the cognate transporter system, we propose a model for arabinan degradation and transport by C. polysaccharolyticus.

IMPORTANCE Genomic DNA sequencing and bioinformatic analysis allowed the identification of a gene cluster encoding several proteins predicted to function in arabinan degradation and transport in C. polysaccharolyticus. The analysis of the recombinant proteins yielded detailed insights into the putative arabinan metabolism of this thermophilic bacterium. The use of various branched arabinan oligosaccharides provided a detailed understanding of the substrate specificities of the enzymes and allowed assignment of two new GH127 polypeptides as β-l-arabinofuranosidases able to degrade pectic substrates, thus expanding our knowledge of this rare group of glycoside hydrolases. In addition, the enzymes showed synergistic effects for the degradation of arabinans at elevated temperatures. The enzymes characterized from the gene cluster are, therefore, of utility for arabinose production in both the biofuel and food industries.

INTRODUCTION

Arabinose is the second most abundant pentose sugar found in nature and is of considerable importance in both microbial metabolism and the biotechnology industry (1–4). Arabinose can be found in various plant cell wall polysaccharides, with arabinans being the most important polymers. Arabinans are neutral side chains of the pectic polysaccharide rhamnogalacturonan I, and depending on the plant source and the ripening stage, they can be a quantitatively important part of the cell wall of dicotyledonous plants. The arabinan backbone is composed of α-1,5-linked arabinofuranose residues, which can be further branched with arabinofuranose residues at position O-3 or O-2 or at both positions (5). In addition, dimeric side chains containing a terminal β-arabinofuranose were recently established as pectic structural elements (6, 7). Therefore, the degradation of these plant cell wall polysaccharides requires diverse enzymes that can cleave a variety of glycosidic bonds in substrates with different degrees of polymerization. Furthermore, the hydrolysis of arabinans is necessary for the complete and efficient depolymerization of the root crop sugar beet, which contains a large amount of complex arabinans (8).

The degradation of l-arabinan involves the synergistic activity of two major enzymes: α-l-arabinofuranosidases (ABFs; EC 3.2.1.55) and α-1,5-arabinanases (ABNs; EC 3.2.1.99). ABF enzymes cleave the terminal nonreducing α-1,2-, α-1,3-, and α-1,5-linked arabinofuranosyl residues. They attack and remove the side chains, allowing the ABN enzymes to hydrolyze the α-1,5-l-arabinofuranoside glycosidic linkages of the arabinan backbone and therefore producing shorter oligosaccharide chains, which are further attacked by ABF enzymes to release arabinose monomers (9–11). Based on the carbohydrate active enzyme (CAZy) classification (http://www.cazy.org/) (12), the two enzymes belong to different glycoside hydrolase (GH) groups. ABF enzymes are members of GH families 3, 43, 51, 54, and 62, whereas ABN enzymes belong solely to GH family 43. Recently, a recombinant β-arabinofuranosidase from Bifidobacterium longum was characterized, which led to the definition of GH family 127 (13). The characterized protein was able to hydrolyze glycoprotein-derived β-arabinofuranose residues. However, this GH family might also be important in cleaving complex β-arabinofuranose containing arabinan side chains.

There is a growing interest in arabinan/arabinose-degrading enzymes due to their potential use in many industrial and biotechnological applications. l-Arabinose can be used as a natural noncaloric sweetener or for the production of the artificial food sweetener d-tagatose (1). Other applications of arabinan/arabinose-degrading enzymes in the food and beverage industries include the enhancement of wine flavor (14), clarification of fruit juice (11), and increasing bread quality and shelf life (15). For biofuel production, arabinose-cleaving enzymes together with hemicellulose-degrading enzymes are essential for the efficient hydrolysis of hemicellulose in lignocellulosic biomass to fermentable sugars (16, 17). Therefore, the identification of arabinan-degrading enzymes that can convert renewable biomass to arabinose could have a measurable impact on these different biotechnological applications.

Caldanaerobius polysaccharolyticus, previously known as Thermoanaerobacterium polysaccharolyticum, is a rod-shaped, Gram-positive, non-spore-forming, anaerobic thermophile isolated from the leachate of an organic waste pile from a canning factory in Hoopeston, IL, USA (18, 19). This bacterium grows optimally at 65 to 68°C (19), and many of the enzymes characterized from C. polysaccharolyticus have thermostability profiles favorable for biotechnological applications (20–22).

In the present report, a gene cluster from C. polysaccharolyticus predicted to encode an endoarabinanase (CpAbn43A), two α-l-arabinofuranosidases (CpAbf51A and CpAbf51B), two β-arabinofuranosidases (CpAbf127A and CpAbf127B), and a β-arabinopyranosidase (CpAbp27A) was studied. The genes were cloned, expressed in Escherichia coli, and the arabinan-degrading activities of the purified recombinant proteins were biochemically characterized. By using purified, naturally occurring oligosaccharides, we obtained insights into the substrate specificities of the enzymes. In addition, the results presented here allowed inferences to be drawn on the hydrolysis and transport strategy of arabinan by C. polysaccharolyticus, a bacterium isolated from a polysaccharide-rich environment. Moreover, the capacity of the enzymes to function at high temperatures, together with their thermostabilities and synergistic activities in degrading arabinan, make them potential candidates for applications in the biofuel and food industries.

RESULTS

Expression and purification of the arabinan-degrading enzymes.

Bioinformatics analysis of a partial genome sequence of C. polysaccharolyticus revealed two genes that were autoannotated by the RAST server (23–25) as encoding an α-l-arabinofuranosidase (CpAbf51A) and an endoarabinanase (CpAbn43A). This observation agreed with the discovery that C. polysaccharolyticus is equipped for the degradation of different polysaccharides (19–21, 26). The genes for polysaccharide degradation in this bacterium are, however, often in a cluster, including the xylan degradation cluster (21) as well as the cluster for mannan degradation (26). To determine whether there is missing genetic information that may further enhance our understanding of the metabolism of arabinose-configured nutrients, we designed primers to close the gap between the genes encoding CpAbf51A and CpAbn43A (see Fig. S1 in the supplemental material). Cloning, sequencing, and translation of the nucleotide sequence of the PCR amplicon demonstrated that the region carries a set of genes that may facilitate arabinan degradation and presumably also transport of arabino-oligosaccharides (Fig. 1A). Analysis of the polypeptide sequences of the proteins using the dbCAN tool revealed the presence of two putative α-l-arabinofuranosidases composed of a GH family 51 catalytic domain, a putative endoarabinanase composed of a GH43 catalytic domain, two putative β-arabinofuranosidases composed of respective GH127 catalytic domains, and a putative β-arabinopyranosidase composed of a GH27 catalytic domain (Fig. 1B). Besides the described GH domains, only CpAbn43A contained a signal peptidase II recognition site. Carbohydrate-binding modules (CBMs) were not detected in the polypeptide sequences of any of the enzymes. The gene cluster also contained genes encoding a putative carbohydrate ABC transporter system composed of an ATP binding component (CALPO_RS0104525), a permease component (CALPO_RS0104530), and a substrate binding component (CALPO_RS0104535 or CpSBC4535). Consistent with its predicted periplasmic location, CpSBC4535 contained a signal peptidase II recognition site. However, from the bioinformatics data, it was not possible to draw conclusions about possible regulators or operons. The genes encoding the putative enzymes and the predicted substrate binding component (CpSBC4535) of the putative arabino-oligosaccharide ABC transporter system were cloned into an expression vector, overexpressed as hexa-histidine fusion proteins, and purified by heat denaturation, immobilized metal affinity chromatography, and, where necessary, anion-exchange chromatography and gel filtration chromatography. The SDS-PAGE analysis of the purified proteins revealed single bands for all proteins with migrations consistent with the calculated molecular masses of the hexa-histidine fusion proteins, based on their amino acid sequences (Fig. 1C).

FIG 1.

Characterization of the arabinan-degrading enzymes from C. polysaccharolyticus. (A) Identification of a gene cluster that contains genes responsible for arabinan degradation and transport. (B) Domain architectures of the arabinan-degrading enzymes from C. polysaccharolyticus. (C) Twelve percent SDS-PAGE of purified proteins encoded by the arabinan degradation gene cluster. (D) Specific activities of CpAbn43A, CpAbf51A, CpAbf51B, and CpAbp27A on arabinan (0.5%, wt/vol) and pNP-linked sugars (1 mM) in citrate buffer (pH 6.0 and 5.5) at 65°C. Enzyme domains are indicated by ovals, and the signal peptide is indicated by a rectangle. DB, debranched; pNP, para-nitrophenyl; Ara, arabinose; f, furanose; p, pyranose.

Substrates and biochemical analyses of the recombinant proteins.

The products of the genes in the putative arabinan degradation cluster were screened for activity with a library of para-nitrophenyl (pNP)-linked sugars as described in our previous publications (21, 26). In addition, different plant cell wall polysaccharides were used to evaluate the activities of the enzymes. Branched arabinan from sugar beet was used, as it has been demonstrated to contain arabinofuranose residues attached to position O-3 and/or O-2 of the α-1,5-linked arabinan backbone (8). Furthermore, according to the manufacturer and previously published data, approximately 60% of the backbone arabinose units are replaced at various positions. Thus, the branched arabinan is a good substrate for α-arabinofuranosidases. Debranched arabinan, which results from digestion of arabinan with an arabinofuranosidase, exhibits a backbone with a very small amount of branching. Thus, this polysaccharide would be preferably hydrolyzed by endo-1,5-arabinanases.

Both C. polysaccharolyticus CpAbf51A and CpAbf51B showed activity against pNP-α-l-arabinofuranoside and arabinan (Fig. 1D). However, for CpAbf51A, a very low activity on debranched arabinan was detected, whereas CpAbf51B showed high activity against this substrate under the tested conditions. The products of hydrolysis from the polysaccharides consisted of arabinose (data not shown), suggesting that CpAbf51A and CpAbf51B are α-l-arabinofuranosidases. CpAbn43A showed detectable activity only for the polymeric substrates arabinan and debranched arabinan (Fig. 1D). The end products of hydrolysis of CpAbn43A on these substrates consisted primarily of arabinobiose (data not shown), indicating that CpAbn43A is an endoarabinanase. The higher activity for CpAbn43A with debranched arabinan suggests that substitutions decrease the accessibility of the enzyme to the α-1,5-linked arabinan backbone. CpAbp27A was the only enzyme that showed appreciable activity on pNP-β-l-arabinopyranoside. The enzyme also showed only low activity on the polymeric substrates and no activity on pNP-α-l-arabinofuranoside (Fig. 1D). These results suggest that CpAbp27A is a β-arabinopyranosidase, which can cleave terminal β-arabinopyranose residues. C. polysaccharolyticus CpAbf127A and CpAbf127B did not show activity on any of the pNP-linked sugars or the polysaccharides tested. However, pNP-β-l-arabinofuranoside, which would be one of the preferred substrates for a GH127 protein, could not be tested because it is not commercially available.

To test whether the substrate binding component (CpSBC4535) of the putative arabino-oligosaccharide ABC transporter can bind to arabino-oligosaccharides, isothermal titration calorimetry (ITC) was carried out with α-1,5-linked arabino-oligosaccharides, with various degrees of polymerization, as the substrates. It was demonstrated that the gene product (or the putative substrate binding component, CpSBC4535) bound to all the substrates to some extent (see Fig. S2 in the supplemental material). However, the binding did not saturate. This might be due to insufficient folding of the protein or the lack of the other transporter components or needed cofactors. Nevertheless, these results suggest that the ABC transporter is responsible for the transport of arabinose-configured oligosaccharides.

Determination of the temperature and pH optima and kinetic parameters of the enzymes.

The optimum pH values and the optimum temperatures of the enzymes were determined by using pNP-α-l-arabinofuranoside (CpAbf51A and CpAbf51B), debranched arabinan (CpAbn43A), and pNP-β-l-arabinopyranoside (CpAbp27A), as the substrates. Because no appropriate substrate was available for CpAbf127A and CpAbf127B, the optimum pH and temperature could not be determined. The optimum pH values of the characterized enzymes all ranged between 5.5 and 6.5, while the optimum temperatures ranged between 60°C and 75°C. (Table 1).

TABLE 1.

Optimum pH and temperature values of CpAbn43A, CpAbf51A, CpAbf51B, and CpAbp27A determined with debranched arabinan and pNP-linked sugars^a

Enzyme	Substrate	Optimum temp (°C)	Optimum pH
CpAbn43A	Debranched arabinan	75	6.0
CpAbf51A	pNP-α-l-arabinofuranoside	70	6.5
CpAbf51B	pNP-α-l-arabinofuranoside	65	5.5
CpAbp27A	pNP-β-l-arabinopyranoside	60	5.5

^aThe experiments were performed in triplicate, and data are reported as means ± standard deviations.

To characterize the kinetic properties of the exo-acting enzymes, Michaelis-Menten plots were generated with pNP-α-l-arabinofuranoside (CpAbf51A and CpAbf51B) and pNP-β-l-arabinopyranoside (CpAbp27A) as the substrates (see Fig. S3 in the supplemental material). The k_cat and K_m values were determined by nonlinear curve fits, and the data are presented in Table 2. CpAbf51A showed a lower k_cat (87.4 s⁻¹) than CpAbf51B (128.8 s⁻¹). The K_m value of CpAbf51B was about twice (1.9 mM) that of CpAbf51A (1.1 mM), thus yielding a slightly higher catalytic efficiency (k_cat/K_m) for CpAbf51A on this artificial substrate. With the specific activity measurement, however, CpAbf51A showed a more appreciable higher specific activity than CpAbf51B (Fig. 1D). CpAbp27A exhibited a K_m of 3.3 mM and a k_cat of 11.8 s⁻¹ and thus a catalytic efficiency of only 3.6 mM⁻¹ s⁻¹.

TABLE 2.

Kinetic properties of CpAbf51A, CpAbf51B, and CpAbp27A with pNP-linked sugars^a

Enzyme	Substrate	kcat (s−1)	Km (mM)	kcat/Km (mM−1 s−1)
CpAbf51A	pNP-α-l-arabinofuranoside	87.4 ± 3.3	1.1 ± 0.2	79.5 ± 15
CpAbf51B	pNP-α-l-arabinofuranoside	128.8 ± 3.4	1.9 ± 0.1	69.4 ± 3.2
CpAbp27A	pNP-β-l-arabinopyranoside	11.8 ± 0.2	3.3 ± 0.03	3.6 ± 0.07

^aThe experiments were performed in triplicate, and data are reported as means ± standard deviations.

Thermostability of CpAbn43A, CpAbf51A, CpAbf51B, and CpAbp27A.

A factor critical to the applicability of enzymes in biotechnological processes is their degree of thermostability. To evaluate this property, CpAbf51A, CpAbf51B, CpAbn43A, and CpAbp27A were incubated in their optimum pH buffer at different temperatures over a period of up to 24 h. The residual activity was then determined by using their cognate substrates. CpAbp27A retained most of its activity at 65°C and 70°C for 1 h, but longer incubation times at all temperatures led to drastic decreases of activity (Fig. 2D). In contrast, the other three enzymes retained more than 60% of their activity at 65°C and 70°C for up to 24 h (Fig. 2A, B, and C). CpAbf51A and CpAbf51B were more stable at higher temperatures, as they retained 50% of their activity after 8 h of incubation at 75°C (Fig. 2B and C), than CpAbn43A, which retained only 20% of its activity at the same temperature (Fig. 2A). At 80°C, CpAbf51A and CpAbn43A lost more than 60% of their activities after 10 min of incubation, while CpAbf51B and CpAbp27A retained at least 75% of their activities. These results showed that the C. polysaccharolyticus arabinan-degrading enzymes are thermostable and therefore may be useful for biocatalytic processes that take place in the 65 to 70°C range.

FIG 2.

Thermostability of CpAbn43A (A), CpAbf51A (B), CpAbf51B (C), and CpAbp27A (D) at different temperatures (65, 70, 75, 80°C). The enzymes (CpAbn43A, 2.0 μM; CpAbf51A, 0.5 μM; CpAbf51B, 0.05 μM; and CpAbp27A, 0.5 μM) were incubated in 50 mM citrate buffer (CpAbn43A, pH 6.0; CpAbf51A, pH 6.5; CpAbf51A, pH 5.5; CpAbp27A, pH 5.5) in the absence of the substrate at the indicated temperatures, and the residual activity was assayed at different time intervals on the appropriate substrate (see Materials and Methods).

Functional characterization of linkage specificity of the arabinan-degrading enzymes.

The enzymes CpAbn43A, CpAbf51A, and CpAbf51B released reducing ends from branched and debranched arabinan polysaccharides (Fig. 1D), which contain arabinofuranose residues attached to positions O-5, O-2, and O-3 of arabinose. In addition, thin-layer chromatography analysis of the degradation of linear, α-1,5-linked arabino-oligosaccharides with a degree of polymerization of up to 6 demonstrated that CpAbf51A and CpAbf51B can completely hydrolyze these substrates to arabinose (see Fig. S4A and B in the supplemental material). CpAbn43A preferably hydrolyzed the oligosaccharides with a higher degree of polymerization, yielding arabinobiose as the main degradation end product (Fig. S4C). Furthermore, it was demonstrated that arabinotriose was only partially hydrolyzed by CpAbn43A, while arabinobiose could not be degraded (Fig. S4C and D).

The foregoing results allow only the conclusion that CpAbn43A, CpAbf51A, and CpAbf51B can hydrolyze α-1,5-linked arabinofuranose residues and give no detailed information about the hydrolysis of other linkage types. Furthermore, very little knowledge was gained on the activity of CpAbf127A and CpAbf127B from the hydrolytic assays with the polymeric substrates or linear, α-1,5-linked arabino-oligosaccharides. Therefore, various branched arabino-oligosaccharides, previously isolated from different plant materials (6, 7, 27), were incubated with the purified proteins. These oligosaccharides did not allow further characterization of CpAbp27A, since the available oligosaccharides contain only arabinofuranose units. Accordingly, CpAbp27A did not show activity on any of the natural substrates used, which is in good agreement with its sole activity on pNP-β-l-arabinopyranose.

To gain more insight into the linkage specificities of CpAbn43A, CpAbf51A, and CpAbf51B, their activities on oligosaccharides with various branching patterns were analyzed (Fig. 3). The oligosaccharides contained a trisaccharide fragment of the α-1,5-linked arabinan backbone, which has an arabinofuranose substitution at position O-3 (Fig. 3A, A-4a), O-2 (Fig. 3B, A-4b), or O-3 and O-2 (Fig. 3C, A-5a). The incubation of these oligosaccharides with the two α-arabinofuranosidases CpAbf51A and CpAbf51B led to an almost complete hydrolysis to arabinose, indicating a wide specificity of these enzymes regarding their linkage recognition. Only the incubation of the doubly substituted trisaccharide A-5a with CpAbf51B showed some minor residues of an unknown product. The products from the hydrolysis with CpAbn43A showed more-complex product patterns. In agreement with the experiments described above on the hydrolysis of linear α-1,5-linked arabino-oligosaccharides, CpAbn43A incubation with the substrates did not result in complete cleavage of the substrates to arabinose. The hydrolysis of A-4a with CpAbn43A yielded only arabinose, arabinobiose, and arabinotriose as the main products, whereas the incubation mixtures of A-4b and A-5a showed some remnants of the substrate and other unknown products in addition to arabinose, arabinobiose, and arabinotriose. Because A-4a contains only O-3-bound side chains while A-4b and A-5a contain side chains linked to position O-2, these results indicate that O-5 and O-3 linkages are preferably hydrolyzed by CpAbn43A. The same trends were observed when oligosaccharides that contain O-3 substitutions of neighboring backbone arabinofuranose residues were hydrolyzed (see Fig. S5 in the supplemental material). CpAbn43A released arabinose, arabinobiose, and arabinotriose as the main end products, while incubation with CpAbf51A and CpAbf51B resulted in complete hydrolysis to arabinose. These results demonstrate that CpAbn43A, CpAbf51A, and CpAbf51B show a high degree of flexibility in terms of the linkage type and the degree of polymerization of the arabino-oligosaccharides hydrolyzed.

FIG 3.

HPAEC-PAD analysis and structures of the hydrolysis products released from the arabino-oligosaccharides A-4a (A), A-4b (B), and A-5a (C) by CpAbn43A, CpAbf51A, and CpAbf51B. Incubations were carried out for 16 h at 60°C with CpAbn43A, CpAbf51A, and CpAbf51B (final concentration, 1 μM). CpAbp27A, CpAbf127A, and CpAbf127B did not show any activity on the tested substrates.

CpAbf127A and CpAbf127B did not show any activity on the exclusively α-linked arabino-oligosaccharides investigated above. However, by using an arabino-oligosaccharide composed of an α-1,5-linked trisaccharide replaced with a β-1,3-linked arabinobiose side chain at position O-3 (Fig. 4A, A-5b), it was possible to obtain information about the activity of CpAbf127A and CpAbf127B. Due to the terminal β-arabinofuranose unit in the side chain, CpAbn43A, CpAbf51A, and CpAbf51B were not able to cleave the side chain, and an unknown tetrasaccharide and arabinose were obtained as the main hydrolysis products. Most likely, this oligosaccharide (A-4c) results from the cleavage of the nonreducing, O-5-linked α-arabinofuranose unit. In contrast, the incubation of A-5b with CpAbf127A and CpAbf127B results in the formation of arabinose and A-4a. Thus, the two enzymes cleave the terminal β-arabinofuranose residue. These two enzymes can, therefore, be confirmed as β-arabinofuranosidases.

FIG 4.

(A) HPAEC-PAD analysis and structures of the hydrolysis products released from the arabino-oligosaccharide A-5b by CpAbn43A, CpAbf51A, CpAbf51B, CpAbf127A, and CpAbf127B. Incubations were carried out for 16 h at 60°C with CpAbn43A, CpAbf51A, CpAbf51B, CpAbf127A, and CpAbf127B (final concentration, 1 μM). CpAbp27A did not show any activity on A-5b. The presence of A-4a and the molecular weight of A-4c were confirmed by LC-porous graphitic carbon (PGC)-MS. (B) HPAEC-PAD analysis and structures of the hydrolysis products released from the arabino-oligosaccharide A-5b by the wild type (WT) and the mutated (E336A) forms of CpAbf127B. Incubations were carried out for 16 h at 60°C at a final enzyme concentration of 1 μM.

By using site-directed mutagenesis, Fujita et al. (13) were able to demonstrate that the Glu338 residue of the β-arabinofuranosidase (HypBA1) from Bifidobacterium longum is critical for catalytic activity. By comparing the polypeptide sequences of CpAbf127A and CpAbf127B with the amino acid sequence of the β-arabinofuranosidase from Bifidobacterium longum, it was possible to identify the corresponding amino acid in CpAbf127B as Glu336. To evaluate if the two proteins have the same mechanism of hydrolysis, the putative critical residue was mutated by site-directed mutagenesis to an Abf127B-E336A derivative, and the activity on A-5b was compared to that of the unmutated (wild-type [WT]) enzyme. The mutated form of CpAbf127B (Abf127B-E336A) did not show activity on A-5b, confirming the importance of Glu336 for the hydrolysis of β-arabinofuranose residues by CpAbf127B (Fig. 4B).

Synergism of CpAbn43A and arabinofuranosidases in arabinan hydrolysis.

CpAbn43A cannot adequately convert short oligosaccharides to arabinose, whereas the arabinofuranosidases CpAbf51A and CpAbf51B efficiently hydrolyzed all α-linked arabino-oligosaccharides tested. We hypothesized that an arabinofuranosidase would work synergistically with CpAbn43A to degrade arabinan. To evaluate this possibility, arabinan and debranched arabinan were incubated with CpAbn43A alone or in combination with CpAbf51A. The hydrolysis products were analyzed by quantifying reducing sugar equivalents and resolving the sugars by high-performance anion-exchange chromatography with pulsed amperometric detection (HPAEC-PAD). The individual activity of CpAbf51A was higher with arabinan (Fig. 5A and B), whereas the individual activity of CpAbn43A was higher with debranched arabinan (Fig. 5C and D). Incubation of the two enzymes together led to an increase in arabinose production from arabinan and debranched arabinan. These results emphasize the importance of a side chain-cleaving enzyme such as CpAbf51A in the coordinated degradation of arabinan by CpAbn43A.

FIG 5.

Synergism of CpAbn43A and CpAbf51A in the hydrolysis of arabinan and debranched arabinan. CpAbn43A and CpAbf51A (final concentration, 25 nM each) were incubated separately or together with arabinan or debranched arabinan (0.5%, wt/vol) in citrate buffer (50 mM, pH 6.0). After hydrolysis for 15 h at 65°C, samples were analyzed by the reducing sugar assay (A and C) or HPAEC-PAD (B and D).

To obtain detailed insights into the synergism of CpAbn43A and CpAbf51A, two-dimensional nuclear magnetic resonance (NMR) spectroscopy was applied to analyze the hydrolysates of debranched arabinan. The signals of the C-1–H-1 and the C-5–H-5 correlation (Fig. 6A) were used to follow the extent of the enzymatic degradation. After incubation with CpAbn43A or CpAbf51A alone, these signals were still clearly detected besides the signals for monomeric arabinose (Fig. 6B), whereas the combination of the two enzymes led to a considerable decrease of their signal intensity (Fig. 6C). These results are in good agreement with the results from the reducing sugar assay and underline the synergism of the two enzymes.

FIG 6.

NMR investigation of the synergism of CpAbn43A and CpAbf51A in the hydrolysis of debranched arabinan. Debranched arabinan (5%, wt/vol) was incubated alone (A) or with CpAbf51A and CpAbn43A separately (B) or together (C) (final enzyme concentration, 2.5 μM each) in phosphate buffer (50 mM, pH 6.0). After hydrolysis for 16 h at 60°C and heat inactivation, samples were analyzed by NMR spectroscopy. The C-1–H-1 and C-5–H-5 correlations of 1,5-linked arabinose were evaluated by using standard compounds and used to follow the enzymatic degradation. The intense signals in panel C arise from monomeric arabinose, the end product of the degradation.

Phylogenetic analysis of C. polysaccharolyticus GH51 and GH43 proteins.

The bioinformatics analysis of the polypeptide sequences of CpAbf51A and CpAbf51B showed that they belong to the GH51 family. This family of polypeptides contains a large number of sequenced representatives, and several enzymatic activities have been ascribed to the members. Using the polypeptide sequences present in the CAZy database and those of CpABf51A and CpAbf51B, a phylogenetic tree was constructed to examine the phylogenetic placement of the two GH51 proteins in relation to other biochemically characterized and uncharacterized homologs (Fig. 7). From the root, three radiations that may be further divided into subfamilies were observed on the phylogenetic tree. Superimposition of enzymatic activities on the three groups demonstrated that two clusters harbor arabinofuranosidases (cluster II and cluster III) whereas the third (cluster I) contains mixed GH activities. Thus, the polypeptides that have been functionally characterized in the GH51 family include endoglucanases (EC 3.2.1.4) and bifunctional enzymes with either endoglucanase/endoxylanase (EC 3.2.1.8) activities or endoglucanase/xyloglucanase (EC 3.2.1.55) activities. Cluster II has a large number of members. However, so far, one member each of only two of its subclusters has been characterized as a mesophilic arabinofuranosidase. In contrast, cluster III, which we have subdivided into four subclusters (A, B, C, and D), is a mixture of characterized and uncharacterized groups. The subclusters III-A and III-B, located very close to the root of cluster III, have no functionally characterized member, whereas subclusters III-C and III-D have several biochemically characterized members. It appears that subcluster III-C started as mesophilic arabinofuranosidases that then evolved into a mixture of thermophilic and mesophilic enzymes. Within subcluster III-C is CpAbf51A. Thermophilic and mesophilic enzymes abound in the large subcluster III-D; this group has the most biochemically characterized members, and within this group is CpAbf51B. To obtain information about the phylogenetic properties of CpAbn43A, the phylogeny of this GH family, reported by Mewis et al. (28), was reconstructed; however, in this case, the polypeptide sequence of CpAbn43A was included in the analysis. By using this method, CpAbn43A was identified as a member of the proposed subfamily GH43_4. Similar to many of its related polypeptides (69%) examined by Mewis and coworkers (28), CpAbn43A has a signal peptide, which suggests that this protein is secreted outside the cytoplasm.

FIG 7.

A phylogenetic tree of the GH51 family, showing 3 main radiations. Members of cluster I (dark green) have been identified to exhibit endoglucanase activity (EC 3.2.1.4) and in some cases bifunctional endoglucanase/endoxylanase (EC 3.2.1.8) activities or endoglucanase/xyloglucanase activities (EC 3.2.1.151). All biochemically characterized enzymes in the remaining clusters exhibited arabinofuranosidase activity (EC 3.2.1.55). Two enzymes from cluster II (blue) were reported as mesophilic enzymes. To date, no biochemically characterized enzymes have been described for clusters III-A (purple) and III-B (red). Clusters III-C (orange) and III-D (light green) contained several biochemically characterized enzymes and include both mesophilic and thermophilic enzymes.

DISCUSSION

Arabinan, a highly complex side chain of the pectic polysaccharide rhamnogalacturonan I and one of the most important sources of arabinose, is an important microbial source of energy in the environment. This observation is manifested in the large number of microbes reported to harbor enzymes for its hydrolysis (29–35), including Bacteroides spp., Bacillus subtilis, and Bacillus stearothermophilus. In addition, arabinan is an important product for the food and biotechnology industries (4, 5). This polysaccharide is composed of a backbone of α-1,5-linked arabinose units and is further branched with side chains of arabinose units at position O-3 or O-2 or at both positions (5). More-complex structural elements such as dimeric side chains or terminal β-arabinofuranoses might also be present (6, 7). The degradation of arabinan requires mainly the synergistic action of two enzymes: endo-1,5-arabinanases (EC 3.2.1.99) and α-l-arabinofuranosidases (EC 3.2.1.55) (9–11). Recently, a recombinant GH127 β-arabinofuranosidase (EC 3.2.1.185), which might also be involved in the degradation of complex arabinan side chains, was characterized (13). Arabinan-degrading enzyme activities have been reported in both fungi and bacteria. Among the best-studied microorganisms that produce arabinan-degrading enzymes are Bacillus subtilis and Aspergillus niger (35, 36). In the present study, we identified a gene cluster containing genes predicted to encode a GH43 family α-l-arabinanase, two GH51 family α-l-arabinofuranosidases, two GH127 β-l-arabinofuranosidases, and one GH27 β-l-arabinopyranosidase in the thermophilic bacterium C. polysaccharolyticus. To gain insight into the role of these polypeptides in the degradation of plant cell wall polysaccharides, the proteins were recombinantly expressed in E. coli and purified close to homogeneity, and their biochemical properties were studied to determine the linkages that they target for cleavage.

Consistent with their predicted function as α-l-arabinofuranosidases, CpAbf51A and CpAbf51B exhibited activity on pNP-α-l-arabinofuranoside and polymeric arabinans. The two GH51 proteins also hydrolyzed arabino-oligosaccharides with various linkages and various degrees of polymerization, which allowed a detailed characterization of the substrate specificities. Thus, these enzymes can cleave arabinofuranose residues, which are bound to the arabinan backbone at positions O-5, O-3, and O-2. In contrast, the associated enzyme CpAbn43A cleaved debranched arabinan at a much higher rate (5-fold) than arabinan. This finding agrees with the predicted function of CpAbn43A as an endoarabinanase that preferably cleaves the exposed arabinan backbone, producing oligosaccharides or shorter chains. Although CpAbn43A cleaved most of the tested arabino-oligosaccharides to some extent, this enzyme exhibited lower activity on the oligosaccharides than did CpAbf51A and CpAbf51B.

By using the β-arabinofuranose containing oligosaccharide A-5b as the substrate, we were able to identify CpAbf127A and CpAbf127B as β-l-arabinofuranosidases. To date, two GH127 proteins have been characterized in detail. A GH127 protein from Bacteroides thetaiotaomicron was demonstrated to be an acetic acid hydrolase (37), while a GH127 protein from Bifidobacterium longum was determined to be a β-l-arabinofuranosidase by using arabino-oligosaccharides abundant in glycosylated proteins (13). However, the earlier studies did not allow unambiguous conclusions on the degradation of arabinans because no arabinan-derived oligosaccharides were used. Thus, the results obtained in the present study provide important information about the role of GH127 enzymes in arabinan degradation. The enzyme designated CpAbp27A showed activity on pNP-β-l-arabinopyranose. Although β-l-arabinopyranoses are not reported as structural elements of arabinans, this enzyme might assist in the deconstruction of complex pectins containing β-l-arabinopyranoses.

The synergistic action of CpAbn43A with an α-arabinofuranosidase, such as CpAbf51A, is essential for the complete depolymerization of branched and debranched arabinan to arabinose. Compared to the total amount of arabinose released individually by CpAbn43A and CpAbf51A, a 2.5-fold increase in arabinose was detected when arabinan was treated with a mixture of the two enzymes. The increased hydrolysis most likely originates from the debranching of the backbone due to the action of CpAbf51A, and thus removing steric hindrance during cleavage of the backbone by CpAbn43A. The oligosaccharides released from the dual actions of the two enzymes are then subsequently cleaved by CpAbf51A. In support of this interpretation, the NMR spectroscopic analysis of debranched arabinan hydrolyzed with both enzymes yielded a very low intensity for the signals derived from 1,5-linked arabinofuranoses. These results suggest that the debranched arabinan backbone can be extensively degraded through a concerted action of the two enzymes. Similar synergistic action was reported for enzymes from Bacillus subtilis, which produces two arabinofuranosidases; α-l-arabinofuranosidase-2, which specializes in removing the side chains, and α-l-arabinofuranosidase-1, which specializes in the degradation of the linear fragments generated by the arabinanases (38).

The maximum activities of CpAbn43A, CpAbf51A, CpAbf51B, and CpAbp27A were obtained between pH 5.5 and pH 6.5. In addition, the enzymes function with maximum activity at high temperatures, up to 75°C for CpAbf51A. Comparable data (with pH 5.0 to 7.0 and temperatures of 60°C to 75°C) were also reported for arabinofuranosidases and arabinanases from other thermophilic bacteria (10, 39–50). However, higher optimum temperatures from 80°C to 90°C in the same pH range were reported for enzymes from Caldicellulosiruptor saccharolyticus (46, 51), Sulfolobus solfataricus (52), Clostridium thermocellum (53), and Thermotoga maritima (54). CpAbn43A, CpAbf51A, and CpAbf51B were thermostable at high temperatures, retaining 60% of their activity after 24 h at temperatures of up to 70°C. These properties are in agreement with those of other enzymes characterized from C. polysaccharolyticus, including the mannan-degrading enzymes Man5A and Man5B (26) and also the xylan-degrading enzymes from this bacterium (21). CpAbf51B was more stable at 80°C, retaining 75% of its activity after 10 min, while CpAbn43A and CpAbf51A retained only a smaller amount of their activity under the same conditions. Compared to other arabinofuranosidases and arabinanases from thermophilic bacteria, the arabinan-degrading enzymes of C. polysaccharolyticus showed high thermostability, although some of the previously characterized enzymes were more stable at higher temperatures for longer time periods (10, 39–52, 54). Furthermore, the two arabinofuranosidases from C. polysaccharolyticus showed kinetic properties comparable to those of hitherto reported thermophilic arabinofuranosidases from various thermophilic bacteria (39–44, 46, 47, 51, 53, 54).

CpAbf51A and CpAbf51B, as well as CpAbf127A and CpAbf127B, show a similar domain architecture (Fig. 1B), but the alignment of their polypeptide sequences yielded only 30% and 22% identity, respectively. The biochemical characterization demonstrated some differences, as CpAbf51A showed a higher specific activity on pNP-α-l-arabinofuranoside whereas CpAbf51B showed a higher specific activity on debranched arabinan (Fig. 1D). In addition, the two enzymes showed slight differences in their optimum pH and temperature values (Table. 1). However, no differences between CpAbf51A and CpAbf51B or CpAbf127A and CpAbf127B could be derived from the hydrolysis of variously branched arabino-oligosaccharides. In the phylogenetic analysis (Fig. 7), CpAbf51A and CpAbf51B shared a subgroup with a number of other biochemically characterized α-l-arabinofuranosidases (EC 3.2.1.55). CpAbf51B was in the same subgroup as an α-l-arabinofuranosidase from Bacillus subtilis (AbfA), which exhibited a higher activity on debranched arabinan than branched arabinan and, therefore, likely prefers O-5 linkages (38). This is in good agreement with the higher activity of CpAbf51B for debranched arabinan. CpAbf51A was in the same subgroup as another α-arabinofuranosidase from Bacillus subtilis (Abf2), which was characterized in the same study as B. subtilis AbfA. This enzyme exhibited significantly lower activity on linear substrates but a higher activity on branched arabinans, which is also in good agreement with the data obtained for CpAbf51A. Therefore, the two GH51 enzymes from C. polysaccharolyticus are potentially expressed to hydrolyze different arabinan structural elements. However, under the conditions used in this study, both CpAbf51A and CpAbf51B were able to hydrolyze arabino-oligosaccharides containing O-2, O-3, and O-5 linkages.

Our phylogenetic analysis of the GH43 family showed that CpAbn43 belongs to the GH43_4 subfamily. This subfamily and subfamilies GH43_5 and GH43_6 are the only ones predicted to exhibit α-l-arabinanase activity (EC 3.2.1.99), which is in agreement with the results from this study. A thermostable arabinanase from Thermotoga petrophila (49) and an arabinanase from B. subtilis (55) were also assigned to the GH43_4 subfamily. Similar to CpAbn43A, the T. petrophila and B. subtilis enzymes cleaved debranched arabinan preferably. In addition, the T. petrophila arabinanase released mainly arabinobiose and arabinotriose from linear arabino-oligosaccharides, which is not different from the data obtained for CpAbn43A. However, the data on the substrate specificities presented in this study provide additional insights to the function of the GH43_4 subfamily.

The occurrence of genes encoding various arabinan-degrading enzymes suggested that C. polysaccharolyticus is able to metabolize naturally occurring arabinan polysaccharides. To examine this hypothesis, C. polysaccharolyticus cells were grown on arabinose or arabinan as the sole carbon source (see Fig. S7 in the supplemental material). Both energy sources supported good growth to a comparable final optical density at 600 nm (OD₆₀₀) within 24 h. Although the growth on arabinan was slower, the results still demonstrated that C. polysaccharolyticus is a bacterium that can derive energy from the complex polysaccharide arabinan. Based on the growth experiment and the polypeptides characterized from this bacterium, we present a model for arabinan degradation in C. polysaccharolyticus. CpAbn43A, a protein with a signal peptidase II recognition site, is most likely secreted and covalently anchored to a lipid moiety on the outside of the cell surface. The other arabinan-degrading enzymes are most likely intracellularly located, which leads us to propose the following model for arabinan metabolism in C. polysaccharolyticus (Fig. 8). In our model, secreted enzymes, such as CpAbn43A, cleave arabinan polysaccharides into shorter chain products (arabino-oligosaccharides). These products are transported intracellularly by the ABC transporter embedded in the cell membrane. The intracellularly located enzymes, such as CpAbf51A, further hydrolyze the arabino-oligosaccharides into arabinose, which is then converted into ribulose. The ribulose enters the pentose phosphate pathway for generation of energy and other important precursors for cellular metabolism. Our observations are consistent with the versatility previously reported for this bacterium (21, 26). The release of arabino-oligosaccharides by C. polysaccharolyticus through its CpAbn43A into the extracellular environment, coupled with its release of xylo-oligosaccharides and manno-oligosaccharides, suggests that these gene clusters aid C. polysaccharolyticus in the metabolism of hemicellulose and pectins in the environment. Furthermore, the thermostability and the synergistic action of the enzymes from the gene cluster, functionally analyzed in the present study, make them candidates for enzymatic hydrolysis of arabinan-containing polysaccharides for use in both the food and biotechnology industries.

FIG 8.

A schematic model of the proposed arabinan metabolism by C. polysaccharolyticus. Branched arabinan and debranched arabinan are degraded into arabino-oligosaccharides by CpAbn43A (orange). The arabino-oligosaccharides are transported into the cell through an ABC transporter (gray) and further hydrolyzed into arabinose by enzymes such as CpAbf51A (green). Arabinose isomerase converts arabinose into ribulose, which is then converted to ribulose-5-phosphate to enter the pentose phosphate pathway.

MATERIALS AND METHODS

Materials.

Caldanaerobius polysaccharolyticus (ATCC BAA-17) was isolated from a waste pile at a canning factory in Hoopeston, IL, USA (19). The pET-28a vector, pET-46b EK/LIC cloning kit, and Perfect Protein marker were purchased from Novagen (San Diego, CA). The pET-28a vector was modified by replacing the gene for kanamycin resistance with that for ampicillin resistance (56). The PicoMaxx high fidelity PCR system, Pfu DNA polymerase, and E. coli JM109 and BL21-CodonPlus (DE3) RIL competent cells were obtained from Stratagene (La Jolla, CA).

The restriction enzymes NdeI, XhoI, and DpnI and the 1-kb DNA ladder were purchased from New England BioLabs (Ipswich, MA). The DNeasy blood and tissue kit and the QIAprep Spin Miniprep kit were obtained from Qiagen, Inc. (Valencia, CA). The Talon metal affinity resin was purchased from Clontech Laboratories, Inc. (Mountain View, CA). Amicon Ultra-15 centrifugal filter units with a 50,000-Da molecular mass cutoff were purchased from Millipore (Billerica, MA). Isopropyl β-d-thiogalactopyranoside, antibiotics, agarose, and sodium citrate were obtained from Fisher Scientific (Pittsburgh, PA).

Linear arabino-oligosaccharides (degree of polymerization, 2 to 6) were obtained from Megazyme (Bray, Ireland). Arabinose was purchased from Sigma-Aldrich (St. Louis, MO). Arabinan and debranched arabinan were purchased from Megazyme (Bray, Ireland). Branched arabino-oligosaccharides were isolated from various plant materials and characterized by HPAEC-PAD, liquid chromatography-mass spectrometry (LC-MS), and NMR spectroscopy (6, 7, 27). Other chemicals and sources of materials are described in the relevant paragraphs below.

Gene cloning and expression.

The bacterium C. polysaccharolyticus was cultured in TYG medium to mid-log phase, and genomic DNA was extracted and purified using the Qiagen DNeasy blood and tissue kit with an integrated RNase treatment step. The partial genome sequence of C. polysaccharolyticus was generated by the W. M. Keck Center for Comparative and Functional Genomics at the University of Illinois at Urbana-Champaign. Autoannotation of the partial genome sequence of C. polysaccharolyticus revealed CALPO_RS0104520 (gene encoding CpAbf51A) and CALPO_RS0104500 (gene encoding CpAbn43A) as encoding putative α-l-arabinofuranosidase and endoarabinanase enzymes, respectively. To determine if there was additional genetic information on arabinan degradation near the two genes, the surrounding genomic DNA was sequenced. Sequencing of C. polysaccharolyticus genomic DNA was performed by a combination of shotgun sequencing and paired-end sequencing methods. Briefly, one-half plate of shotgun 454 GS-FLX data and one-half plate of 8-kb paired-end GS FLX Titanium data were obtained using the 454 Life Sciences Genome Sequencer (Branford, CT). The partial genome sequence was assembled by Newbler version 2.0 (Roche, 454 Life Sciences) into 93 scaffolds comprised of 332 contigs. The draft genome had multiple sequence gaps; however, the scaffolds provided some general insight into the arrangement of the contigs. For example, CpAbf51A occurred ∼1.6 kb from the end of contig 15, and the scaffold generated from the paired-end reads suggested that CpAbn43A was located on contig 17, nearby on the genome. The scaffold generated from the Newbler assembly predicted an ∼900-bp gap between the two contigs. We therefore designed a primer set (P1 and P2) (Table 3) to amplify the DNA spanning the intervening region (Fig. S1) for nucleotide sequencing. An approximately 900-bp fragment was amplified by PCR, and the product was purified from an agarose gel and cloned into pGEM-T by TA cloning. The cloned fragment was sequenced by the W. M. Keck Center using the M13 forward and reverse primers. The sequences were then assembled into a contiguous fragment, and the new DNA fragment was uploaded to the RAST annotation server (23–25) to identify the translated sequences. In the following, the locus tags and accession numbers from the C. polysaccharolyticus whole-genome shotgun sequence NZ_KE386494 are used. For clarity, the protein names shown in Fig. 1 are used throughout the manuscript.

TABLE 3.

Primer sequences used for gene cloning and gap closure

Primer name	Orientation	Sequencea
P1	Forward	5′-AATTTGTGCAGTGGGCAAAGCG-3′
P2	Reverse	5′-CGGCACATCACTAAAATCCTTGCTG-3′
CpAbn43A	Forward	5′-CCCATTATACCTATTCTCTTCACTTTGCTCTCCTCTGCG-3′
	Reverse	5′-TGCTCAATTGAGCGATTTTTTCATCCAAGGTCATCCTGG-3′
CpAbf51A	Forward	5′-GACGACGACAAGATGAAGGGCAACAGTAAAGAAAG-3′
	Reverse	5′-GAGGAGAAGCCCGGTTAAACAACTTCCACTTTTG-3′
CpAbf51B	Forward	5′-GACGACGACAAGATGAGCAAGAAAGCAAAAATGATTTTAGAC-3′
	Reverse	5′-GAGGAGAAGCCCGGTTAATTCTTTTTTGCAAGGCG-3′
CpAbp27A	Forward	5′-GACGACGACAAGATGCGATAGATAATTGCGGTAGAG-3′
	Reverse	5′-GAGGAGAATAGGTCATCCCATCTATCCCAAAAA-3′
CpAbf127A	Forward	5′-GACGACGACAAGGCACGAGCATATGAGGTAAAGAAAGA-3′
	Reverse	5′-GAGGAGAAGCCCGTTAGCCAATACTTGCCATCTAGAG-3′
CpAbf127B	Forward	5′-GACGACGACAAGATGAATAATGCTGCAAAATTTCAAGCAAAGCC-3′
	Reverse	5′-GAGGAGAAGCCCGGTTACTTCTCTCTCACCCAC-3′
CpAbf127B-SDM		3′-ACGGCCTATGCTGCGACCTGTGCAGCG-5′
CpSBC4535	Forward	5′-ATCGATCGCATATGAAAAATACCAGTGCGACTTCC-3′
	Reverse	5′-ATCGATCGCTCGAGTTAGTTATTTGCTGCTTTCCATTTATC-3′

^aOligonucleotide primers were synthesized by Integrated DNA Technologies (Coralville, IA).

The genes encoding the glycoside hydrolases of the gene cluster were amplified using the C. polysaccharolyticus genomic DNA as a template with the primers listed in Table 3 and the PicoMaxx high fidelity PCR kit. To facilitate subsequent ligation into the pET46 Ek/LIC vector, the forward primers were engineered to incorporate a 5′-GACGACGACAAG extension and the reverse primers were designed to include a 5′-GAGGAGAAGCCCGGT extension. The resultant amplicons of the genes were treated with the exonuclease activity of T4 DNA polymerase and annealed with the pET46 Ek/LIC vector using the Ek/LIC cloning kit (Novagen). The LIC mixture was introduced into E. coli JM109 competent cells by electroporation using a Gene Pulser Xcell from Bio-Rad (Hercules, CA). The recombinant plasmids were then purified using the QIAprep Spin Miniprep kit and sequenced to confirm the correctness of the nucleotide sequence of the cloned genes (W. M. Keck Center for Comparative and Functional Genomics at the University of Illinois at Urbana-Champaign). The primers for CpSBC4535 (WP_026486275) were engineered to incorporate NdeI and XhoI restriction sites, and the resulting amplicons were cloned into pGEM-T by TA cloning. Subsequently, the gene was excised with NdeI and XhoI and subcloned into pET-28a.

The genes in the recombinant plasmids were cloned in frame with an N-terminal hexahistidine peptide to facilitate protein purification. The plasmids were used in transforming E. coli BL-21 CodonPlus (DE3) RIL cells by heat shock transformation and plated on LB agar medium supplemented with ampicillin (100 μg/ml) and chloramphenicol (50 μg/ml). The plates were incubated at 37°C overnight. A single transformant for each recombinant plasmid was selected and precultured in LB liquid medium (10 ml) supplemented with ampicillin (100 μg/ml) and chloramphenicol (50 μg/ml) at 37°C for 8 h with aeration (225 rpm). The precultures were then transferred into 2.8-liter Fernbach flasks containing fresh LB medium (1 liter) supplemented with the two antibiotics and cultured at 37°C with vigorous shaking (225 rpm/min) to an optical density of 0.3 at 600 nm. IPTG (isopropyl β-d-thiogalactopyranoside; final concentration, 0.1 mM) was then added, and growth was continued for another 16 h at 16°C.

Purification of the recombinant proteins.

After 16 h of induction of gene expression, the cells were harvested by centrifugation (4,000 × g, 4°C) for 15 min and resuspended in lysis buffer (30 ml; 50 mM Tris-HCl, 300 mM NaCl [pH 7.0]). The cells were then lysed by two passages through an EmulsiFlex C-3 cell homogenizer from Avestin (Ottawa, Canada). The cell debris was removed by centrifugation (20,000 × g, 4°C) for 20 min. To remove the heat-labile E. coli proteins from the clarified lysate, the supernatant was heated at 65°C for 30 min and centrifuged (20,000 × g, 4°C) for 15 min. The resulting fusion proteins contained N-terminal hexahistidine tags; therefore, immobilized metal affinity chromatography was used for the protein purification. The cell extract was loaded onto Talon metal affinity resin (Clontech, Mountain View, CA) preequilibrated with binding buffer (50 mM Tris-HCl, 300 mM NaCl [pH 7.5]) and incubated for 1 h at 4°C. The resin slurry was transferred to a gravity column, and after washing with 50 column volumes of binding buffer, the bound protein was eluted with 10 column volumes of elution buffer (50 mM Tris-HCl, 300 mM NaCl, 250 mM imidazole [pH 7.5]). One-milliliter fractions were collected, and the elution process was monitored by mixing 10 μl of each fraction with 90 μl of protein assay dye reagent from Bio-Rad (Hercules, CA).

If necessary, the elution fractions of each of the proteins were pooled separately and exchanged into an anion-exchange column binding buffer (20 mM Tris-HCl, pH 7.6) using Amicon Ultra-15 centrifugal filter units (50,000-Da molecular mass cutoff). The proteins were then loaded onto a 5-ml HiTrap Q anion-exchange column (GE Healthcare, Piscataway, NJ), and eluted with a linear gradient of elution buffer (20 mM; 1 M NaCl, pH 7.6) from 0 to 50% over 20 column volumes and then from 50 to 100% over 5 column volumes. The fractions containing the purified protein were pooled and further purified with a Superdex 200 Hiload 16/60 size exclusion column with protein storage buffer (50 mM Tris-HCl, 150 mM NaCl [pH 7.5]) as the mobile phase. The anion-exchange and gel filtration chromatography assays were performed using an AKTAxpress fast protein liquid chromatograph equipment (GE Healthcare, Piscataway, NJ).

The collected fractions were pooled, concentrated, and analyzed with 12% SDS-PAGE with Laemmli's method (57). The protein concentrations were determined using a NanoDrop 1000 from Thermo Scientific (Waltham, MA) with the following calculated extinction coefficients: CpAbn43A, 146,220 M⁻¹ cm⁻¹; CpAbf51A, 98,780 M⁻¹ cm⁻¹; CpAbf51B, 103,750 M⁻¹ cm⁻¹; CpAbp27A, 124,135 M⁻¹ cm⁻¹; CpAbf127A, 171,145 M⁻¹ cm⁻¹; CpAbf127B, 138,505 M⁻¹ cm⁻¹; putative substrate-binding component, 134,885 M⁻¹ cm⁻¹. The molecular masses were as follows: CpAbn43A, 52,070 g mol⁻¹; CpAbf51A, 56,020 g mol⁻¹; CpAbf51B, 56,956 g mol⁻¹; CpAbp27A, 48,948 g mol⁻¹; CpAbf127A, 71,669 g mol⁻¹; CpAbf127B, 74,296 g mol⁻¹; CpSBC4535, 57,755 g mol⁻¹.

Hydrolysis of pNP-linked sugars.

The hydrolytic activities of the enzymes against para-nitrophenyl (pNP)-linked substrates were determined by a continuous colorimetric assay. For this reaction, pNP-linked substrates (1.0 mM) were incubated with the enzymes (CpAbn43A and CpAbf51A, 20 nM; CpAbf51B, 25 nM; CpAbp27A, 100 nM) in citrate buffer (50 mM, pH 5.5) at 65°C for 30 min. The rate of pNP released during the reactions was determined by monitoring the absorbance at 400 nm continuously in a thermostated Cary 300 UV-Vis spectrophotometer from Varian Inc. (Palo Alto, CA). Control reactions without enzyme were separately performed, and the spontaneously hydrolyzed substrate was subtracted from the enzyme-catalyzed hydrolysis. The initial velocities were then calculated using the Beer-Lambert law with the extinction coefficient of pNP (1,636 M⁻¹ cm⁻¹) at pH 5.5.

Specific activity on polysaccharide substrates.

The hydrolytic activity of the enzymes against polysaccharides (arabinan and debranched arabinan) was assayed by incubating each polysaccharide (final concentration, 0.5%, wt/vol) with the enzymes (CpAbn43A and CpAbf51A, 5 nM; CpAbf51B, 25 nM; CpAbp27A, 100 nM) in citrate buffer (50 mM sodium citrate, pH 6.0) at 65°C. Initial experiments were performed at different enzyme concentrations to ensure that the 30-min time point was within the linear region of the reaction progress curve. The control reaction was performed under the same conditions without adding the enzyme. After incubation for 30 min, the reaction was terminated by heating for 10 min at 100°C, and the mixture was centrifuged for 10 min at 12,000 × g, after which the reducing sugar concentration in the supernatant was determined using the para-hydroxybenzoic acid hydrazide method (58).

Substrate binding assay with isothermal titration calorimetry.

The interaction between linear 1,5-linked arabino oligosaccharides (DP 3 to 6) and the putative substrate binding component CpSBC4535 (CALPO_RS0104535) was investigated by using a VP-ITC microcalorimeter (Microcal Inc., MA) at 25°C. The oligosaccharides (0.5 mM; 20 mM HEPES, 300 mM NaCl) were injected into the protein solution (50 μM; 20 mM HEPES, 300 mM NaCl) in 28 successive 10-μl aliquots at 300-s intervals.

Determination of optimum pH and temperature with pNP-linked sugars.

The optimum temperature and pH were determined with pNP-α-l-arabinofuranose (CpAbf51A and CpAbf51B) or pNP-β-l-arabinopyranose (CpAbp27A) as the substrate. For pH optimum assays, the enzymes (CpAbf51A, 20 nM; CpAbf51B, 25 nM; CpAbp27A, 500 nM) were incubated with pNP-α-l-arabinofuranose or pNP-β-l-arabinopyranose (1.0 mM) at 65°C in different buffers: pH 4.0 to 6.5 (50 mM citrate buffer, 150 mM NaCl) and pH 7.0 to 8.0 (50 mM phosphate buffer, 150 mM NaCl). For temperature optimum assays, pNP-α-l-arabinofuranose or pNP-β-l-arabinopyranose (1.0 mM) was incubated with the enzymes (CpAbf51A, 20 nM; CpAbf51B, 25 nM; CpAbp27A, 500 nM) in citrate buffer (50 mM, 150 mM NaCl, and pH 6.5 for CpAbf51A and pH 5.5 for CpAbf51B and CpAbp27A) at different temperatures (40, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95°C). The release of pNP was measured continuously by monitoring the absorbance at 400 nm, extinction coefficients of pNP in different buffers were determined individually, and the resulting values were used to calculate the initial velocities in the different reactions.

Optimum temperature and pH for CpAbn43A with debranched arabinan as a substrate.

In the pH profile assays, CpAbn43A (5 nM) was incubated with a debranched arabinan solution (0.5%, wt/vol) at 75°C in a Thermomixer R from Eppendorf (Hauppauge, NY) with a shaking speed of 400 rpm for 10 min. The same buffers listed above were used for the pH analyses; however, the temperature optimum study was carried out with citrate buffer at pH 6. Reactions were terminated by heating at 100°C for 10 min, after which the mixtures were cooled down on ice and then centrifuged at 12,000 × g for 10 min, and the reducing sugar ends were quantified using the para-hydroxybenzoic acid hydrazide assay. A standard curve was constructed using known concentrations of arabinose.

Thermostability assays.

In the thermostability assays, CpAbn43A (500 nM), CpAbf51A (2.0 μM), CpAbf51B (50 nM), and CpAbp27A (500 nM) were incubated in citrate buffer (50 mM; pH 6.5 for CpAbn43A and CpAbf51A and pH 5.5 for CpAbf51B and CpAbp27A) at different temperatures (65, 70, 75, 80°C) in the absence of substrate. Samples were taken at different time intervals (0, 10, 30, 60, 180, 240, 480, 720, 1,440 min) and stored at −20°C until activity assays were performed. The activity assays were performed with the corresponding pNP-linked sugars (CpAbf51A, CpAbf51B, CpAbp27A) and arabinan (CpAbn43A), as described above.

Determination of the kinetic parameters.

The kinetic properties of CpAbf51A were assayed with pNP-α-l-arabinofuranose (CpAbf51A and CpAbf51B) or pNP-β-l-arabinopyranose (CpAbp27A) as the substrate at 75°C (CpAbf51A) or 65°C (CpAbf51B and CpAbp27A) in citrate buffer (50 mM and pH 6.5 for CpAbn43A and CpAbf51A; 50 mM and pH 5.5 for CpAbf51B and CpAbp27A). The concentrations of pNP-α-l-arabinofuranose or pNP-β-l-arabinopyranose were varied from 0.08 to 10 mM, and initial velocities were determined for each substrate concentration. The initial rates were plotted against the substrate concentrations, and the Michaelis-Menten constant (K_m) and the maximum velocity (V_max) were estimated with a nonlinear curve fit using GraphPad Prism v5.0 (GraphPad, San Diego, CA). The k_cat was calculated as the quotient of the V_max and the concentration of enzyme used in the reaction.

Hydrolysis of arabino-oligosaccharides.

The hydrolytic activity of the enzymes (0.5 μM) against linear arabino-oligosaccharides (10 mg/ml; degree of polymerization, 2 to 6) in citrate buffer (pH 5.5, 50 mM; 50 mM NaCl; 15 h of incubation) was analyzed by thin-layer chromatography and HPAEC-PAD as described previously (59–62). Branched arabino-oligosaccharides (100 μM) were incubated with the enzymes (final concentration, 1 μM) for 16 h at 60°C. After inactivation of the enzymes at 100°C, the hydrolysates were diluted and analyzed by HPAEC-PAD on an ICS-5000 system (Thermo Scientific Dionex, CA) equipped with a CarboPac PA-100 column (250 by 2 mm; Thermo Scientific Dionex). A flow rate of 0.25 ml/min and a gradient composed of the following eluents were used at 25°C: (A) bidistilled water, (B) 0.1 M sodium hydroxide, (C) 0.1 M sodium hydroxide plus 0.2 M sodium acetate, (D) 0.25 M sodium hydroxide plus 1 M sodium acetate. Before every run, the column was washed with 100% D for 15 min, linear gradients to 100% C (10 min) and subsequently to 90% A and 10% B were applied, and the column was equilibrated with 90% A and 10% B for 20 min. After injection, the following gradient was applied: 0 to 10 min, isocratic 90% A and 10% B; 10 to 20 min, linear to 100% B; 20 to 65 min, linear to 50% B and 50% C; 65 to 80 min, linear to 100% C; 80 to 90 min, linear to 100% D.

Site-directed mutation of CpAbf127B was performed using the PCR-based QuikChange Lightning Multi site-directed mutagenesis kit (Agilent Technologies, CA). The putative critical amino acid of CpAbf127B was identified by aligning its sequence to the sequence of the B. longum β-arabinofuranosidase characterized by Fujita et al. (13). The mutagenic primer is shown in Table 3. The CpAbf127B plasmid was mutated and isolated according to the manufacturer's instructions, and the inserts in the mutated plasmids were sequenced to confirm the introduced mutation. The mutated recombinant protein was expressed and purified as described above.

Synergistic activity of CpAbf51A and CpAbn43A.

Synergism of CpAbn43A and CpAbf51A in the hydrolysis of arabinan and debranched arabinan was investigated by incubating CpAbn43A or CpAbf51A or both (final concentration, 25 nM each) with arabinan or debranched arabinan (0.5%, wt/vol) in citrate buffer (50 mM, pH 6.0). After hydrolysis for 15 h at 65°C, samples were analyzed by the reducing sugar assay and HPAEC-PAD as described above.

For the NMR spectroscopic investigation of the synergistic effects of the two enzymes, debranched arabinan (50 mg/ml, 0.05 M sodium phosphate buffer, pH 6.0) was incubated with CpAbf51A or CpAbn43A or both (final concentration, 2.5 μM) for 16 h at 60°C. After heat inactivation (100°C, 10 min), 540 μl of the hydrolysate was mixed with 60 μl of deuterium oxide. Acetone (0.5 μl) was used for spectrum calibration (2.22 ppm/30.89 ppm, according to Gottlieb et al. [63]). Proton (pulse sequence, WATERSUP) and phase-sensitive heteronuclear single quantum coherence (HSQC) (pulse sequence, HSQCAD) spectra were acquired on an Agilent 600-MHz NMR spectrometer equipped with a 5-mm probe.

Phylogenetic analyses.

All 1,831 complete GH51 domain sequences were extracted from the CAZy database on 25 May 2017. Multiple-sequence alignment was performed by MUSCLE (64). To generate high-quality and relevant alignments, MAFFT (65) was used to iteratively remove highly dissimilar sequences, as described previously (28). Following these quality control measures, 1,414 sequences remained. With these sequences, FASTTree (66) was used to generate a phylogenetic tree based on the midpoint root method. The tree was then visualized with iTOL v3 (67). The phylogenetic classification of CpAbn43A was determined by reconstruction of the GH43 phylogenetic tree as reported by Mewis et al. (28), except for incorporation of CpAbn43 in the analyses.

Growth of C. polysaccharolyticus on arabinan or arabinose as a sole carbon source.

C. polysaccharolyticus was cultured anaerobically at 65°C in butyl rubber-stoppered Balch tubes using a previously reported defined medium (21) in an atmosphere consisting of 100% CO₂. As the sole carbohydrate source, arabinose or arabinan was added at 0.5% (wt/vol). Inoculation passages were done three times, prior to growth curve measurements, to adapt the cells to the respective culture media. Subsequently, 0.1 ml of precultured cells was inoculated into 10 ml of fresh medium, and the absorbance readings at OD₆₀₀ were monitored with time, using a Spectronic 21D spectrophotometer (Milton Roy, Ivyland, PA).

Accession number(s).

The accession numbers of the polypeptides analyzed in this study are as follows: CpAbn43A, WP_026486268; CpAbf51A, WP_026486272; CpAbf51B, WP_026486270; CpAbp27A, WP_035172167; CpAbf127A, WP_026486276; CpAbf127B, WP_026486269; and CpSBC4535 or CALPO_RS0104535, WP_026486275.