Heterologously Expressed Family 51 α-l-Arabinofuranosidases from Oenococcus oeni and Lactobacillus brevis

Abstract

Putative α-l-arabinofuranosidases of Oenococcus oeni and Lactobacillus brevis were heterologously expressed and characterized. We report the basic functional properties of the recombinant enzymes in comparison to those of a commercial family 51 arabinosidase of Aspergillus niger.

Recent studies (3, 7-9, 20, 21) demonstrated that several bacterial species, including wine-related lactic acid bacteria (LAB), possess the general ability to hydrolyze glycosylated plant secondary metabolites. With the increasing availability of annotated LAB genomes in public databases (GenBank and CAZy) (2, 4, 16), a reasonable approach to investigate bacterial glycoside metabolism is through experimental confirmation of putative data. The purpose of this short paper is to report the basic biochemical properties of three novel α-l-arabinofuranosidases derived from LAB (Oenococcus oeni and Lactobacillus brevis). So far, bacterial arabinosidases have been isolated from Bacillus subtilis (13), Bifidobacterium breve (23), Geobacillus stearothermophilus (12), and Clostridium thermocellum (25).

Putative α-l-arabinofuranosidase genes (abf) were amplified from O. oeni ATCC BAA-1163 (abf_{O. oeni}) and L. brevis DSM 20054 (abf1/abf2_{L. brevis}) and sequenced. Primer sequences (not shown) were based on the published genomic sequences of O. oeni ATCC BAA-1163 (NZ_AAUV00000000; OENOO_34003) and L. brevis ATCC 367 (NC_008497.1; LVIS_1750; LVIS_2221). In silico analysis using SignalP (1), TMpred (11), and PSORT (26) predicted that the enzymes would be soluble and cytoplasmic. According to the current classification of glycosyl hydrolases (4, 10), the presented arabinosidases can be classified as members of glycoside hydrolase (GH) family 51. Although the nucleotide sequence of abf_{O. oeni} exists in the genomes of O. oeni PSU-1 (NZ_AAUV00000000) and O. oeni AWRIB429 (NZ_ACSE00000000) as well, it is not annotated as such in both strains due to open reading frame interruptions. Interestingly, a transition mutation (G→A) resulting in the amber stop codon TAG occurs in both strains at the same position (nucleotide 503).

Subsequently, the abf genes were cloned into expression vectors of the pET21 series (Novagen, Madison, WI) with C-terminal His₆ fusion tags in frame. Escherichia coli T7 expression-competent cells (New England BioLabs) were used as the expression host for all constructs. Protein production was carried out in Terrific broth (24) with lactose (5 g liter⁻¹) as the inducer. The expression levels obtained per liter of broth were as follows: abf_{O. oeni}, 2.3 kU; abf1_{L. brevis}, 5 kU; and abf2_{L. brevis}, 1.6 kU.

As shown in Fig. 1 (SDS-PAGE with Coomassie staining), the recombinant arabinosidases were electrophoretically pure after immobilized metal affinity chromatography followed by anion-exchange chromatography (with Ni-charged chelating Sepharose and Resource Q, respectively; both from GE Healthcare, Uppsala, Sweden) performed according to the supplier's recommendations. The pure enzymes were stored in 0.02 M citrate-phosphate buffer (pH 7) prepared according to McIlvaine (17). Protein concentrations were determined with the Bio-Rad protein assay (Bio-Rad, Inc., Hercules, CA). Molecular masses were estimated by gel filtration and SDS-PAGE (with and without sample cross-linking); the results are shown in Fig. 1 and Table 1. Gel filtration (Superdex 200 column, GE Healthcare) was carried out with 0.1 M McIlvaine buffer plus 0.15 M NaCl, pH 7.0, at a flow rate of 0.3 cm min⁻¹. The column was calibrated with molecular mass markers ranging from 29,000 to 700,000 kDa (Sigma-Aldrich). For protein cross-linking prior to SDS-PAGE (Fig. 1B), 5 μl of 2.5% (vol/vol) glutardialdehyde was added to 100 μl of sample containing 0.1 mg protein. After 15 min at 37°C, the reactions were stopped with 10 μl of 1 M Tris-HCl (pH 8.0). As a control, β-amylase (200-kDa molecular mass marker; Sigma-Aldrich) was applied to the gel with and without glutardialdehyde treatment, confirming the tetrameric state of the protein (Fig. 1B, lanes 5 and 6). Although the data displayed in Table 1 suggest the oligomeric state of all three bacterial arabinosidases, it was not possible to infer definite molecular masses, especially because of the discrepancies observed between the gel filtration and electrophoretic experiments. The only known three-dimensional (3D) structures of family 51 arabinosidases (from G. stearothermophilus and C. thermocellum) (12, 25) indicated the hexameric organization (trimers of dimers) of these proteins. Interpreting the results presented here (Table 1), we propose that Abf_{O. oeni} and Abf1_{L. brevis} are at least dimeric and Abf2_{L. brevis} is at least tetrameric.

FIG. 1.

SDS-PAGE of the recombinant arabinosidases of O. oeni and L. brevis. (A) Ten percent separation gel (homogenous). Lanes 1 and 5, Precision Plus protein standard (Bio-Rad); lane 2, Abf_{O. oeni}; lane 3, Abf1_{L. brevis}; lane 4, Abf2_{L. brevis}. (B) PhastGel Gradient 8-25 (GE Healthcare) run after glutardialdehyde cross-linking of the samples. Lane 1, Abf_{O. oeni}; lane 2, Abf1_{L. brevis}; lane 3, Abf2_{L. brevis}; lane 4, Spectra multicolor high-range protein ladder (Fermentas); lane 5, β-amylase with cross-linking; lane 6, β-amylase without cross-linking.

TABLE 1.

Estimated molecular masses of recombinant arabinosidases obtained by SDS-PAGE and gel filtration^a

Enzyme	Theoretical subunit mass (kDa)	Molecular mass (kDa) by:
		SDS-PAGEb	SDS-PAGE with cross-linkingc	Gel filtration
abfO. oeni	59	50	130	190
abf1L. brevis	57	55	120	272
abf2L. brevis	58	60	250	315
β-Amylased	50	40c	160	202

^aThe SDS-PAGE values represent approximate estimates obtained from Fig. 1. Gel filtration (including molecular mass calibration) was performed in duplicate.

^bTen percent separation gel (homogenous) (Fig. 1A).

^cPhastGel Gradient 8-25 (GE Healthcare) (Fig. 1B).

^d200-kDa molecular mass marker for gel filtration (Sigma-Aldrich).

All enzyme assays were carried out in 0.1 M McIlvaine buffers. Standard reaction conditions were pH 5.5 and 37°C. Assays with p-nitrophenyl-glycosides were stopped after 10 min of reaction (2-fold volumetric excess of 0.5 M Na₂CO₃) and p-nitrophenol (pNP) was quantified photometrically at 400 nm. One unit of enzyme activity corresponds to 1 μmol of pNP released per min. Michaelis-Menten kinetics assays (Table 2) were performed with pNP-α-l-arabinofuranoside (0.05 to 10 mM in the assay). The reaction time was 1 min. The highest catalytic efficiency (k_cat/K_m) was observed with Abf1_{L. brevis} (461 s⁻¹ mM⁻¹), and the lowest (1.68 s⁻¹ mM⁻¹) was observed with the β-glucosidase of O. oeni (18).

TABLE 2.

Kinetic constants of bacterial arabinosidases determined with pNP-α-l-arabinofuranoside at 37°C, pH 5.5

Kinetic constant	Result for enzymea:
	BglO. oenib	AbfO. oeni	Abf1L. brevis	Abf2L. brevis
Km (mM)	5.73	0.671	0.144	0.358
Vmax (U mg−1)	7.03	12.4	69.9	46.8
kcat (s−1)c	9.61	12.2	66.4	45.2
kcat/Km (s−1 mM−1)	1.68	18.2	461	126

^aAll data represent the average of duplicate experiments.

^bO. oeni β-glucosidase (18).

^cCalculated per theoretical subunit mass (“catalytic center activity” according to reference 22).

The specific activities for various substrates are shown in Table 3. When arabinosaccharides (all from Megazyme, Wicklow, Ireland) were used as substrates, 1 U of activity is expressed as μmol of arabinose released per min. Sugars were quantified by high-performance liquid chromatograpy (HPLC) (Dionex DX500, CarboPac PA 100 column; Dionex, Sunnyvale, CA). All enzymes displayed mainly arabinofuranosidase activity and only low xylosidase activities. No activities for pNP-α-l-arabinopyranoside, pNP-α-l-rhamnopyranoside, or pNP-β-d-galactopyranoside were detected. The bacterial enzymes were able to degrade arabinobiose and arabinotriose but had only low activity on arabinan (both linear and branched). In contrast to the bacterial arabinosidases, the arabinosidase of Aspergillus niger (AnAbf) displayed high hydrolytic activity (9.6 U mg⁻¹) on branched arabinan, indicating additional activities toward α-1,2 or α-1,3 arabinofuranosidase bonds. Further, only AnAbf released small quantities of arabinobiose and arabinotriose in the assays with arabinan (linear and branched) (data not shown).

TABLE 3.

Specific activities of arabinosidases for synthetic and natural substrates determined at 37°C, pH 5.5^a

Substrate	Sp act (U mg−1)
	AnAbfb	AbfO. oeni	Abf1L. brevis	Abf2L. brevis
pNP-α-l-arabinofuranoside	19c	12	69	42
pNP-β-d-glucopyranoside	NDd	ND	ND	<0.1
pNP-β-d-xylopyranoside	ND	<0.1	0.35	<0.1
α-1,5-l-Arabinobiose	24	4.1	34	6.2
α-1,5-l-Arabinotriose	22	3.0	19	3.4
Linear α-1,5-l-arabinan	1.4	0.056	1.1	0.12
Branched arabinan	9.6	0.031	1.5	0.28

^aThe substrate concentrations in the assays were 10 mM, except for arabinan (5 mg ml⁻¹). All data represent the average of duplicate experiments.

^bGH family 51 arabinosidase of A. niger (commercial product, Megazyme, Wicklow, Ireland).

^cAccording to the supplier.

^dND, not detectable.

Table 4 lists the effect of sugars and ethanol on the enzyme activities. AnAbf and the α-l-arabinosidase side activity of the β-glucosidase of O. oeni (Bgl_{O. oeni}) (18) were included. Arabinose (37.5 g liter⁻¹) had an inhibitory effect (10 to 36% activity reduction) on all arabinosidases. The presence of ethanol (12% [vol/vol]) caused an activity loss of around 50%. Interestingly, the arabinosidase side activity of Bgl_{O. oeni} was almost completely inhibited by glucose, while the β-glucosidase activity of the enzyme was only reduced to 63% at 36 g liter⁻¹ glucose (18). In the same study (18), we found that β-glucosidase was activated (1.4-fold) by ethanol (12% [vol/vol]), but the enzyme's arabinosidase activity was reduced to 23% at the same ethanol concentration (Table 4).

TABLE 4.

Influence of sugars and ethanol in the assay on arabinosidase activities determined with pNP-α-l-arabinofuranoside^a

Sugar or ethanol	Relative enzyme activity (%)
	AnAbfb	BglO. oenic	AbfO. oeni	Abf1L. brevis	Abf2L. brevis
Control	100	100	100	100	100
Sugars
Arabinose (37.5 g liter−1)	63.7	103	75.2	66.5	90.0
Glucose (45 g liter−1)	105	2.11	93.2	89.7	88.3
Xylose (37.5 g liter−1)	106	85.2	97.4	85.6	87.2
Ethanol
6% (vol/vol)	88.8	52.3	64.2	76.5	73.7
12% (vol/vol)	51.4	23.3	42.6	63.6	51.0

^aResults show the influence of sugars (0.25 M) and ethanol in the assay (37°C, pH 5.5, reaction time of 10 min) on arabinosidase activities determined with pNP-α-l-arabinofuranoside (1 mM). All data represent the average of duplicate experiments.

^bGH family 51 arabinosidase of A. niger (commercial product; Megazyme, Wicklow, Ireland).

^cO. oeni β-glucosidase (18).

pH profiles (Fig. 2A) were determined with pNP-α-l- arabinofuranoside (1 mM) in 0.01 M McIlvaine buffers ranging from pH 3.0 to 7.5. While the bacterial arabinosidases lost activity at low pH values, the fungal enzyme showed its highest activity at pH 3.5. The influence of temperature (Fig. 2B) on the activities was determined by variation of the temperature in the standard assay (pH 5.5) from 4 to 90°C, demonstrating high temperature optima for AnAbf (67°C) and Abf1_{L. brevis} (62°C), while Abf_{O. oeni} and Abf2_{L. brevis} showed their highest activities between 40 and 60°C.

FIG. 2.

Influence of pH and temperature on arabinosidase activities determined with pNP-α-l-arabinofuranoside (1 mM). •, AnAbf; ○, Abf_{O. oeni}; ▵, Abf1_{L. brevis}; and ▴, Abf2_{L. brevis}. (A) Assays (37°C, 10-min reaction time) were performed with 0.1 M McIlvaine buffers pH 3.0 to 7.5. (B) Standard assay (pH 5.5, 10 min) performed at 4 to 90°C. All data represent the average of duplicate experiments. A relative activity of 100% refers to standard assay conditions (37°C, pH 5.5).

In conclusion, the data presented above demonstrate that both O. oeni and L. brevis are at least genetically equipped with functional arabinofuranosidases. With p-nitrophenyl-glycosides as model substrates, all three bacterial enzymes exhibited mainly specificity for the α-l-arabinofuranoside form (Table 2), albeit with considerable variation in catalytic efficiency (Table 3). The classification as (exo-) α-1,5-l-arabinofuranosidases (EC 3.2.1.55) was also confirmed by the ability to cleave arabinobiose and arabinotriose, but with relatively poor activities on arabinan (both linear and branched). Using these results as starting point, further studies should be conducted to determine the expression levels of the corresponding genes in vivo, especially to elucidate their actual metabolic function. Of particular interest in these terms is certainly the wine lactic acid bacterium O. oeni. Reviewing the current scientific literature, it seems likely that O. oeni is to become a model organism in the study of the bacterial β-glucoside metabolism: Primarily driven by the interest in the hydrolysis of glycosylated aroma precursors during the malolactic fermentation, abundant information on the activities of O. oeni toward various synthetic (7, 8) and natural (3, 6) glycosides is already available. Moreover, several intracellular glycosidases (18; this article) of O. oeni, including a phosphoglucosidase related to the phosphoenolpyruvate-dependent phosphotransferase system (5), have been identified to date. Therefore, it is crucial to investigate how the expression of these genes is concerted during the malolactic fermentation, which should help us gain a better understanding of the complexity of the (lactic acid) bacterial glycosidase metabolism.

Another important aspect in characterizing novel arabinosidases is the biotechnological interest in recruiting efficient biocatalysts suitable for the degradation of hemicelluloses (14, 19) and the release of aroma compounds from glycosylated precursors present in wine or fruit beverages (15). However, as demonstrated above, the bacterial arabinosidases presented here displayed no advantages over the fungal enzyme used as a control, especially given their poor activities at low pH and the negative effect of ethanol. We will further investigate if and under what conditions these enzymes are suitable to liberate aroma compounds during wine or fruit juice processing.

Nucleotide sequence accession numbers.

All nucleotide sequences have been deposited in GenBank under the following accession numbers: HM363021 (abf_{O. oeni}), HM363023 (abf1_{L. brevis}), and HM363024 (abf2_{L. brevis}).