Characterization of a Fructan Exohydrolase from the Gut Bacterium Anaerostipes butyraticus

Abstract

Glycoside hydrolase family 32 (GH32) enzymes play key roles in fructooligosaccharide metabolism in gut bacteria. In this study, a GH32 enzyme (GenBank code, GFO85652) containing carbohydrate binding module 66 (CBM66) from the gut bacterium Anaerostipes butyraticus (AbFEH) was heterologously expressed in Escherichia coli. We constructed an expression plasmid that does not contain sequences for the N-terminal signal peptide and the C-terminal region potentially involving cell-wall binding. The enzyme obtained (AbFEH∆C) was purified and characterized. Thin-layer chromatography and high-performance liquid chromatography analyses revealed that AbFEH∆C produced fructose from all the substrates, sucrose, 1-kestose, inulin, and levan, and intermediate oligosaccharide products were not observed. The ratio of activities towards sucrose, 1-kestose, nystose, inulin, and levan was 6:100:83:8:95 under the conditions of this study. A region containing M and CBM66 domains was further removed from AbFEH∆C, and the activities for both 1-kestose and levan of this mutant enzyme were about 400-fold lower than those of AbFEH∆C. Kinetic analysis indicated a low K_m value for levan, while requiring higher substrate concentrations for 1-kestose and sucrose. Comparison of the predicted structure of AbFEH with crystal structures of some GH32 enzymes indicated that residues at subsite −1 were almost completely conserved, while some key residues found in GH32 enzymes were not present at subsites +1 and +2 in AbFEH. These observations suggest that AbFEH functions as fructan exohydrolase that exhibits low sucrose-hydrolyzing activity.

Abbreviations

AbFEH, Anaerostipes butyraticus fructan exohydrolase; AbFEH∆C, C region truncated AbFEH; AbFEH∆M-CBM66-C, AbFEH lacking M domain, CBM66 domain, and C region; AsInuAMN8, Arthrobacter sp. MN8 exo-inulinase; BsSacC, Bacillus subtilis levanase; CBM, carbohydrate binding module; GH, glycoside hydrolase family; HPLC, high-performance liquid chromatography; TmINV, Thermotoga maritima invertase; TLC, thin layer chromatography.

INTRODUCTION

Anaerostipes species, commonly found in the gut microbiota, have been reported to produce butyrate, a short-chain fatty acid that supports gut barrier integrity. Thus, these bacteria play a crucial role in maintaining intestinal health [1]. Multiple genes for enzymes belonging to glycoside hydrolase family (GH) 32, which potentially act on prebiotic fructans, are present in their genomes according to the Carbohydrate-Active enZymes (CAZy) database [2], making it intriguing to investigate the relationship between these enzymes and the metabolism of fructose-containing saccharides.

There are several types of saccharides containing fructose units, fructooligosaccharide, inulin, and levan, which exhibit prebiotic effects by selectively promoting the growth of beneficial gut bacteria. Fructooligosaccharide is a mixture of mostly the trisaccharide 1-kestose [GF₂, 1-kestotriose, β-D-Fruf-(2→1)-β-D-Fruf-(2↔1)-α-D-Glcp], the tetrasaccharide nystose [GF₃, 1,1-kestotetraose, β-D-Fruf-(2→1)-β-D-Fruf-(2→1)-β-D-Fruf-(2↔1)-α-D-Glcp], and the pentasaccharide 1^F-fructosyl nystose [GF₄, 1,1,1-kestopentaose, β-D-Fruf-(2→1)-β-D-Fruf-(2→1)-β-D-Fruf-(2→1)-β-D-Fruf-(2↔1)-α-D-Glcp)] [3]. Inulin is a linear β-(2→1)-linked fructose polymer, and levan is a polymer mainly linked by β-(2→6)-glycosidic linkages. These fructose-containing saccharides are mainly hydrolyzed by enzymes belonging to GH32 in gut bacteria. GH32 enzymes adopt a conserved catalytic domain with a five-bladed β-propeller fold, with the active site located in the central cavity of the β-propeller structure [4]. The enzymatic properties of many GH32 enzymes have been documented. Some GH32 enzymes also possess a carbohydrate-binding module (CBM); for example, a GH32 fructan β-fructosidase from Streptococcus mutans has two CBM66 domains [5]. However, only a few GH32 enzymes possessing CBM66 have been characterized to date.

In the analysis of GH32 enzymes from A. butyraticus JCM17466 [6], two enzymes that hydrolyze sucrose have been identified and described as GH32-1 and GH32-3 (GenBank codes, GFO84605 and GFO85652, respectively) [7]. The GH32-1 enzyme comprises 488 amino acid residues, which is a common length for GH32 enzymes, while the GH32-3 enzyme comprises 1,179 residues and possesses a CBM66 domain. The ratio of activities of the GH32-3 enzyme for sucrose, 1-kestose, and nystose has been determined to be 100:434:768, indicating that the activity for sucrose is low while those for fructooligosaccharide are high [7]. Here, we designated the GH32-3 enzyme as AbFEH. The enzyme was heterologously expressed in Escherichia coli, purified, and its properties were characterized. Our findings indicate that the enzyme functions as a fructan exohydrolase that exhibits low sucrose-hydrolyzing activity.

MATERIALS AND METHODS

Construction of the expression plasmid

Genomic DNA was isolated from A. butyraticus JCM17466 as described previously [7]. The target DNA sequence encoding AbFEH (GenBank code, GFO85652) was amplified using PCR based on DNA polymerase KOD-Plus-Neo instruction manual (Toyobo Co., Ltd., Osaka, Japan) with the following primers: 5’-ATG TCT TCT ATT ACA GAG GGA AAC ACT-3’ and 5’-CAT ATC TTC CAG GCT GCT GTT GAT TGC-3’, and the amplified fragment DNA was named full AbFEH fragment. Given that the C region of AbFEH is presumed to lack a direct role in catalytic functions (see Fig. S1A; see J. Appl. Glycosci. Web site, and the Results section for more details), the corresponding sequence encoding residues 885-1,179 was truncated to obtain the enzyme designated AbFEH∆C (Figs. S1B and C; see J. Appl. Glycosci. Web site). The nucleotide sequence encoding the N-terminal signal peptide (residues 1-28) was also deleted. PCR was performed with full AbFEH fragment as a template and the primers 5’-CAT ATG GCT AGC AAC ACT CCG CAG TTA TCT GGA CTG ACG-3’ and 5’-GTG GTG CTC GAG TTA TGT GTT AAT ATA GAA GTT CTG ATA CTC-3’ containing NheI and XhoI sites (underlined), respectively. To investigate the function of M and CBM66 domains, the sequence encoding M and CBM66 domains (residues 631-884) was further removed from the plasmid pET28a_AbFEH∆C (Fig. S1D; see J. Appl. Glycosci. Web site). PCR was performed with pET28a_AbFEH∆C as a template and the primers 5’-CAT ATG GCT AGC AAC ACT CCG CAG TTA TCT GGA CTG ACG-3’ and 5’-GTG GTG CTC GAG TTA TGT CTG CAG AGG ATA TAC CGT AAT ATC-3’ containing NheI and XhoI sites (underlined), respectively. The amplified fragment was ligated in pET28a (+) vector creating pET28a_AbFEH∆M-CBM66-C. The constructed plasmids were transformed into E. coli strain JM109. The plasmid constructs were extracted, and the sequences were confirmed by DNA sequencing.

Protein expression and purification

The recombinant plasmid (pET28a_AbFEH∆C or pET28a_AbFEH∆M-CBM66-C) was transformed into E. coli BL21(DE3) for heterologous protein expression. Transformed E. coli BL21(DE3) was cultured in 1 L of LB broth Miller (Nacalai Tesque Inc., Kyoto, Japan) supplemented with 50 µg/mL kanamycin until the absorbance at 600 nm reached 0.6. Enzyme expression was induced by adding isopropyl-β-D-thiogalactopyranoside at a final concentration of 0.1 M under shaking conditions at 130 rpm, 18 °C for 18 h. The cells were harvested by centrifugation at 6,000 × G for 5 min at 4 °C, resuspended in 30 mL of 20 mM Tris-HCl buffer pH 8.0, and disrupted using a sonicator. The supernatant was collected by centrifugation at 12,000 × G for 20 min at 4 °C and purified using nickel nitrilotriacetic acid (Ni-NTA) affinity chromatography using 10 mL Ni-NTA agarose resin (Fujifilm Wako Pure Chemical Corporation, Osaka, Japan). Recombinant protein was eluted with 20 mM Tris-HCl buffer with 20-50 mM imidazole. The purity and molecular mass of the target enzyme were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), while the concentrations were determined by measuring absorbance at 280 nm with the calculated coefficient values (AbFEH∆C, 1.573; AbFEH∆M-CBM66-C, 1.730) using Expasy Protparam (https://web.expasy.org/protparam/).

Production of the polysaccharide levan

Levan producing enzyme from Beijirinckia indica (BiBftA) from previous study was used [8]. Escherichia coli BL21(DE3) carrying the expression plasmid [9] was cultured overnight in LB agar plate with ampicillin. A single colony was inoculated in 3 mL LB broth with 50 µg mL⁻¹ ampicillin and incubated at 37 °C for 18 h. To scale up the culture volume, it was inoculated in 1 L LB broth and was incubated until A₆₀₀ = 0.6. IPTG (0.1 M) was added, and the culture was incubated in a shaker at 18 °C for 18 h. To remove the cultivation broth, it was centrifuged at 4 °C, 6,000 × G for 5 min, and the bacterial cells were suspended in 20 mM Tris-HCl (pH 8.0). Crude protein was obtained by centrifugation after sonication. Crude enzyme was used to react with sucrose (180 g/L) in 20 mM sodium phosphate buffer pH 6.0 at 30 °C for 30 min. The reaction was stopped by heating at 98 °C for 5 min and was added with 4-fold volumes of ethanol to precipitate levan for 18 h at 4 °C as described previously [8]. The precipitate was collected by centrifugation at 4,000 × G, 4 °C for 20 min and dried using a rotary evaporator. The precipitate was subsequently dissolved in deionized water for purification through dialysis. The levan solution was dialyzed against water, with the water replaced twice over a 48-h period, using a cellulose membrane (MWCO 12,000-14,000 Da; Viskase Corporation, Lombard, IL, USA) to remove low molecular weight impurities. The total sugar content of levan was determined using the phenol-sulfuric acid method by measuring the fructose concentration at 490 nm as described [10].

Enzyme activity measurement

The activity of AbFEH∆C was assessed against sucrose (60 mM), 1-kestose (60 mM; obtained from B Food Science Co., Ltd., Chita, Japan), nystose (60 mM; Fujifilm Wako Pure Chemical Corporation), inulin (12 mg mL⁻¹; produced by enzymatic synthesis, Tokyo Chemical Industry Co., Ltd., Tokyo, Japan), and levan (4.4 mg mL⁻¹) in a 20 mM sodium phosphate buffer (pH 6.0) at 30 °C for 30 min. The reaction was terminated by adding 150 μL Somogyi reagent. The reducing sugar produced during the reaction was quantified using the Somogyi-Nelson method by measuring the absorbance of liberated fructose at 500 nm as described [11]. The activity of AbFEH∆M-CBM66-C was also determined against 1-kestose (60 mM) and levan (4.1 mg mL⁻¹) using the same experimental conditions. The optimal pH of the AbFEH∆C was determined for 60 mM 1-kestose using 20 mM sodium acetate buffer (pH 4.0-5.5), 20 mM sodium phosphate buffer (pH 6.0-6.5), and 20 mM Tris-HCl buffer (pH 7.0-8.0) at 30 °C for 30 min. To evaluate thermostability, AbFEH∆C was incubated at temperatures ranging from 30 to 60 °C for 30 min, followed by cooling in an ice bath. The residual enzyme activity was tested for 60 mM 1-kestose at 30 °C for 30 min in 20 mM sodium phosphate buffer (pH 6.0), and the reducing power was measured using the Somogyi-Nelson method. For kinetic analysis, various concentrations of sucrose, 1-kestose, levan, and inulin were dissolved in 20 mM sodium phosphate buffer (pH 6.0) and incubated at 37 °C for several time points. The reducing sugar produced during the reaction was quantified using the Somogyi-Nelson method to determine the reaction rate using Michaelis-Menten kinetics. The data was analyzed through non-linear least-squares fitting using the program Fit-o-mat [12].

Thin layer chromatography and high-performance liquid chromatography

The enzyme assay using thin layer chromatography (TLC) was carried out on a silica gel 60 plate (Merck-Millipore Corporation, Burlington, MA, USA). The sample solutions were prepared by reacting 0.196 mg mL⁻¹ enzyme in 10 mM 2-(N-morpholino) ethanesulfonic acid buffer (pH 6.0) on different substrates (sucrose, 1-kestose, inulin, and levan) at 30 °C for the following time points: 0, 0.5, 1, 3, 18 h. Each of the reaction solutions was heated at 98 °C for 5 min to stop the reaction. The TLC plate was allowed to stand in eluent consisted of butanol, ethanol, and water in a 5:5:3 (v/v/v) ratio. Solutions of glucose, fructose, sucrose, and 1-kestose were used as standards. After eluent migration, the spots were detected by charring with 5 % sulfuric acid in methanol. High-performance liquid chromatography (HPLC) was also used to determine enzymes hydrolyzing pattern. Each reaction mixture was analyzed by HPLC with refractive index detection using a Shodex SUGAR KS-802 column (8 mm × 300 mm; Resonac Corporation, Tokyo, Japan) at 80 °C with a mobile phase of water at a flow rate of 0.5 mL min⁻¹ as described [13].

RESULTS

Expression plasmid construction and purification

AbFEH consists of 1,179 amino acid residues. We predicted the structure of AbFEH using ColabFold [14]. The structure was predicted (Fig. S1A; see J. Appl. Glycosci. Web site) to be composed of an N-terminal signal sequence (1-28), and five regions designated here as N domain (29-129), GH32 common region (130-630), M domain (631-721), CBM66 domain (722-884), and C region (885-1,179) (Fig. S1B; see J. Appl. Glycosci. Web site). The GH32 common region is found in all the GH32 enzymes, which are composed of five bladed β-propeller domain and super β-sandwich domain [4]. The program Clustal Omega (https://www.ebi.ac.uk/jdispatcher/msa/clustalo) with characterized enzymes in the CAZy database showed that GH32 common region of AbFEH was homologous to S. mutans exo β-fructofuranosidase (57 % identity) [5], Bacillus subtilis levanase SacC (BsSacC) (35 %) [15], and Bacteroides thetaiotaomicron exo β-fructofuranosidase (34 %) [16]. CBM66 domain of AbFEH was homologous to CBM66 domain of BsSacC (24 %) among the characterized enzymes [17].

The predicted structures of both N domain and M domain were composed of β-sandwich, and the domains were identified as immunoglobulin-like fold and cadherin-like fold, respectively, using the InterPro database search (https://www.ebi.ac.uk/interpro/). The model of C region was divided into four small domains, C1 (885-965), C2 (966-1,029), C3 (1,030-1,103), and C4 (1,104-1,179). The InterPro search indicated that C2 was shown to be a FIVAR domain [18], and both C3 and C4 domains were identified as invasin/intimin cell-adhesion fragments [19]. These structures have often been found in membrane-bound or cell wall-associated proteins in bacteria, suggesting that C region is not likely to be directly involved in the enzymatic activity. Here we constructed an expression vector pET28a_AbFEH∆C encoding His-tag and residues 29-884 (Fig. S1C; see J. Appl. Glycosci. Web site). To investigate whether CBM66 is involved in the substrate binding, another expression vector pET28a_AbFEH∆M-CBM66-C which does not carry the nucleotide sequence encoding M domain, CBM66 domain, and C region was constructed (Fig. S1D; see J. Appl. Glycosci. Web site). The enzymes, AbFEH∆C and AbFEH∆M-CBM66-C, were expressed in soluble form in E. coli, and purified using Ni-NTA affinity chromatography. The purified proteins yielded a single band on SDS-PAGE (Fig. 1A).

Fig. 1. Purification and characterization of AbFEH∆C.

　(A) SDS-PAGE showing the purification of AbFEH∆C and AbFEH∆M-CBM66-C. Lanes: M, molecular mass marker, the molecular sizes in kDa are shown on the left; 1, purified AbFEH∆C; 2, purified AbFEH∆M-CBM66-C. (B-E) TLC analysis of the reactions for sucrose (B), 1-kestose (C), inulin (D), and levan (E). Lanes: G, glucose; FSK, a mixture of fructose, sucrose, and 1-kestose. Numbers indicate reaction times (h). Note that glucose, fructose, and sucrose are detected at almost the same position. (F) Optimal pH measured in 20 mM sodium acetate buffer (pH 4.0-5.5; ●), sodium phosphate buffer (6.0-6.5; ▲), or Tris-HCl buffer (7.0-8.0; ■). (G) Thermostability after an incubation for 30 min in 20 mM sodium phosphate buffer (pH 6.0).

Enzymatic activities of AbFEHΔC and AbFEHΔM-CBM66-C

The actions of AbFEH∆C for sucrose, 1-kestose, inulin, and levan were tested by TLC (Figs. 1B-E). In the TLC analysis, glucose, fructose, and sucrose were detected at almost the same position. The patterns of TLC spots of the reaction for sucrose appeared not to be changed (Fig. 1B). The enzyme hydrolyzed 1-kestose, inulin, and levan, and the main product appeared to be fructose (Figs. 1C-E). During the hydrolysis of inulin and levan, products larger than disaccharides were not observed. The enzymatic activities of sucrose (60 mM), 1-kestose (60 mM), nystose (60 mM), inulin (12 mg mL⁻¹), and levan (4.4 mg mL⁻¹) were measured (Table 1). The activity for 1-kestose (2.33 μmol min⁻¹ mg⁻¹) was 17-fold higher than that for sucrose (0.141). Also, the activity for levan (2.23) was 12-fold higher than that for inulin (0.191). The ratio of activities towards sucrose, 1-kestose, nystose, inulin, and levan was 6:100:83:8:95, when the activity for 1-kestose was defined as 100 %.

Table 1.　Activities of AbFEH∆C and AbFEH∆M-CBM66-C for some saccharides.

Saccharide (concentration)	Activity (μmol min−1 mg−1) a
AbFEH∆C
Sucrose (60 mM)	0.141 ± 0.004
1-Kestose (60 mM)	2.33 ± 0.16
Nystose (60 mM)	1.95 ± 0.06
Inulin (12 mg mL−1)	0.191 ± 0.058
Levan (4.4 mg mL−1)	2.23 ± 0.18
AbFEH∆M-CBM66-C
1-Kestose (60 mM)	(5.70 ± 1.00) × 10−3
Levan (4.1 mg mL−1)	(5.45 ± 0.14) × 10−3

^a Mean ± standard deviation.

The effect of pH and temperature on the enzymatic activity was measured using 1-kestose as a substrate. A bell-shaped profile of the activity was observed, and the optimal pH was identified as 5.5-6.0 (Fig. 1F). The enzyme retained about 60 % of its activity after 30-min incubation at 45 ºC (Fig. 1G).

The enzymatic activity of AbFEH∆M-CBM66-C was measured in the same manner as for AbFEH∆C. The activities towards both 1-kestose and levan were almost diminished (Table 1), and both of these were about 400-fold lower than those of AbFEH∆C. The enzymatic activities of AbFEH∆M-CBM66-C for sucrose (60 mM), nystose (60 mM), and inulin (12 mg mL⁻¹) were significantly low, and their values could not be determined.

HPLC analysis and kinetic analysis of the hydrolysis of AbFEHΔC

The reaction products of AbFEH∆C were analyzed by HPLC (Fig. 2). The reaction for 60 mM sucrose revealed its hydrolysis into glucose and fructose (Fig. 2B). Despite the reaction proceeding for 18 h, a substantial portion of sucrose remained unhydrolyzed. When 60 mM 1-kestose was used as the substrate, the enzyme initially (0.5-3 h) cleaved it into sucrose and fructose (Fig. 2C). Subsequently, sucrose was further hydrolyzed into glucose and fructose, as evidenced by a small glucose peak after 18 h, indicating a slow reaction. The peak for 1-kestose was not observed after 18 h, suggesting a more efficient hydrolysis of 1-kestose compared to sucrose. AbFEH∆C hydrolysis of inulin (1.2 %) yielded fructose, along with detectable peaks for sucrose and fructose after 18 h (Fig. 2D). This is probably because majority of the inulin used in this study has a degree of polymerization ranging from 12-15 [20], suggesting that the inulin polysaccharide contains glucose and fructose units in an approximate ratio of 1:11-14. The hydrolysis of levan produced only fructose (Fig. 2E). Among all the substrates used, AbFEH∆C demonstrated the highest efficiency and fastest hydrolysis with levan. After 18 h, levan was completely hydrolyzed, as evidenced by the absence of any peak at approximately 8 min.

Fig. 2. HPLC analysis of the enzymatic hydrolysis of AbFEH∆C.

　(A) Standard sugars containing levan (L), inulin (I), 1-kestose (K), sucrose (S), glucose (G), and fructose (F). (B-E) Reaction time course of AbFEH∆C for sucrose (B), 1-kestose (C), inulin (D), and levan (E). Numbers (0, 0.5, 1, 3, and 18) indicate the reaction time (h).

The kinetic parameters of AbFEH∆C for sucrose, 1-kestose, inulin, and levan were determined. Although the activity for 1-kestose was higher than that of sucrose (Table 1), achieving saturation required very high substrate concentrations, a behavior similarly observed for sucrose (Figs. 3A and B). As a result, the plot of substrate concentration [S] against initial velocity v₀ for both sucrose and 1-kestose exhibited a linear relationship, suggesting that the reactions were measured under conditions where [S] ≪ K_m. Under the condition, the relationship can be expressed as v₀ = k_cat/K_m [E]_total [S] and the k_cat/K_m value was estimated (see for example [21]), while the individual K_m and k_cat values could not be accurately determined (Table 2). In contrast, the reaction with inulin followed Michaelis-Menten kinetics, and the K_m value was relatively low (K_m = 13.5 mg mL⁻¹) (Fig. 3C) compared to 1-kestose and sucrose. The enzyme achieved saturation with levan at a lower concentration (K_m = 0.98 mg mL⁻¹) (Fig. 3D).

Fig. 3. Kinetic analyses of AbFEH∆C for sucrose (A), 1-kestose (B), inulin (C), and levan (D).

Table 2.　Kinetic parameters of AbFEH∆C.

Saccharide	Km	kcat (s−1)	kcat/Km
AbFEH∆C
Sucrose	> 1 (M)	-	3.2×10−3 (s−1 mM−1)
1-Kestose	> 1 (M)	-	3.5×10−2 (s−1 mM−1)
Inulin	13.5 ± 2.1 (mg mL−1)	174 ± 9	12.9 (s−1 mg−1 mL)
Levan	0.98 ± 0.15 (mg mL−1)	358 ± 19	365 (s−1 mg−1 mL)
BsSacC a
Sucrose	63.6 (mM)	686	11 (s−1 mM−1)
Inulin	33.5 (mg mL−1) b	827	24.7 (s−1 mg−1 mL) b
Levan	12 (mg mL−1) b	370	30.8 (s−1 mg−1 mL) b

-, Unable to calculate.

^a Values are cited from Wanker et al. [15].

^b Values are calculated from the original data.

DISCUSSION

There have been a few reports on characterization of GH32 enzymes possessing CBM66. In this study, AbFEH∆C was expressed in E. coli, and some properties of the enzyme have been elucidated. A convenient method to prepare the polysaccharide levan using the enzyme BiBftA from B. indica has previously been established [8], which allows us to compare the enzymatic activities of AbFEH∆C for oligosaccharides and levan. The results of TLC (Fig. 1) and HPLC (Fig. 2) exhibited that AbFEH∆C produced fructose from all the substrates, sucrose, 1-kestose, inulin, and levan, and intermediate oligosaccharide products were not observed. These results suggest that the enzyme functions as fructan exohydrolase. AbFEH∆M-CBM66-C, which does not possess M and CBM66 domains, exhibited low activities against both 1-kestose and levan (Table 1), suggesting that the region of M and CBM66 domains plays a critical role in the recognition of both fructooligosaccharide and polysaccharide.

The activity of AbFEH∆C for sucrose (60 mM) was low (Table 1), and the enzyme did not completely hydrolyze sucrose after 18 h reaction in this study (Fig. 2B). The ratio of activities towards sucrose, 1-kestose, nystose, inulin, and levan was 6:100:83:8:95. Although AbFEH∆C hydrolyzed 1-kestose (60 mM), the activity (2.33 μmol min⁻¹ mg⁻¹) was lower than those of typical GH32 enzymes, such as FperFFase from Frischella perrara (492 μmol min⁻¹ mg⁻¹ for sucrose and 42.3 μmol min⁻¹ mg⁻¹ for 1-kestose) [13] and AkFFase from Aspergillus kawachii (613 μmol min⁻¹ mg⁻¹ for sucrose) [22]. The K_m values of AbFEH∆C for sucrose and 1-kestose were markedly high, which did not allow us to calculate the exact values. We estimated the k_cat/K_m values of AbFEH∆C based on the linear relationship between initial velocity v₀ and substrate concentration [S] under low substrate conditions ([S] ≪ K_m). The values were 3.2×10⁻³ s⁻¹ mM⁻¹ for sucrose and 3.5×10⁻² s⁻¹ mM⁻¹ for 1-kestose, which were about 3,000-fold lower and 300-fold lower, respectively, than that of BsSacC for sucrose (Table 2). In contrast, the kinetic parameters for inulin and levan could be determined. The K_m value of levan (0.98 mg mL⁻¹) was significantly low coupled with a high k_cat value (358 s⁻¹). The results suggest that AbFEH prefers polysaccharides.

Enzymatic activities of some levan-hydrolyzing enzymes have been reported. Chen et al. have investigated various microbial levanases that produce levan oligosaccharides. The activities for 1 % levan of the 10 levanases possessing high activities are approximately 40-300 µmol mg⁻¹ min⁻¹ [23]. Notably, none of the 10 levanases contain the C-terminal CBM66 domain. Fructan exohydrolases from plants have extensively been investigated, and the enzymes are classified into fructan 1-exohydrolase (1-FEH) and fructan 6-exohydrolase (6-FEH) depending on the linkage type attacked [24]. The levan-hydrolyzing activity of sugar beet 6-FEH (22 % sequence identity to AbFEH) is 12.5 µmol mg⁻¹ min⁻¹ [25]. There are no CBM66 domains in the plant fructan exohydrolases. The results indicate that the activities of AbFEH for fructooligosaccharide, inulin, and levan (Table 1) were lower than those of microbial and plant levanases that do not possess CBM66. The kinetic parameters of the exo-acting enzyme BsSacC [15], which possesses the C-terminal CBM66 domain, were compared (Table 2). In BsSacC, all the reactions with sucrose, inulin, and levan followed Michaelis-Menten kinetics. On the other hand, the k_cat/K_m value for levan of AbFEH∆C was markedly higher than those for sucrose, 1-kestose, and inulin.

We assessed the aforementioned substrate preference of AbFEH from the structural viewpoint. Amino acid sequence of AbFEH is most homologous to those of Thermotoga maritima invertase (TmINV, 30 % identity) [26] and Arthrobacter sp. MN8 exo-inulinase (AsInuAMN8, 29 % identity) [27] among proteins deposited in the PDB database. The structure of TmINV in complex with raffinose (α-D-Galp-(1→6)-α-D-Glcp-(1↔2)-β-D-Fruf) has been reported. To probe the amino acid residues involved in interaction with the substrate, residues within 4 Å from raffinose were searched using the program PyMOL (https://pymol.org/), and the corresponding residues in the structure of AsInuAMN8, the predicted structure of AbFEH, and the predicted structure of Anaerostipes hadrus GH32 enzyme AhDO83_10515 (56 % identity) were listed (Table 3). The enzyme AhDO83_10515 has been described as A. hadrus GH32-4 in the previous study, which exhibits similar properties to AbFEH (i.e., A. butyraticus GH32-3) [7]. Based on comparisons with other GH32 enzymes, catalytic residues of AbFEH were identified as Asp152, Asp280, and Glu322, which function as the catalytic nucleophile, transition state stabilizer, and general acid/base, respectively [28]. The amino acid residues interacting with fructose unit at subsite −1 were found to be 9 residues, which were almost completely conserved among the four enzymes despite their differences in substrate specificity.

Table 3.　Comparison of amino acid residues in the predicted structures of AbFEH and AhDO83_10515 with the crystal structures of TmInv and AsInuAMN8.

	AbFEH	AhDO83_10515	TmInv	AsInuAMN8
Organism	Anaerostipes butyraticus	Anaerostipes hadrus	Thermotoga maritima	Arthrobacter sp. MN8
PDB ID	a	a	1W2T	8I12
Identity (%)	100	56	30	29
Catalytic residue
	D152	D354	D17	D33
	D280	D479	D138	D161
	E322	E521	E190	E215
Residue interacting with fructose at subsite −1
	N151	N353	N16	N32
	Q168	Q370	Q33	Q49
	W175	W377	W41	W57
	Y207	F409	F74	F69
	S208	S410	S75	S90
	R279	R478	R137	R160
	C323	C522	C191	C216
	Y397	Y579	Y240	Y292
	W427	W609	W260	W314
Residue interacting with saccharide units at subsites +1 and +2
	D237	D439	Y92	A118
	b	b	E101	G127
	b	b	E188	V213
	b	b	T208	N240

^a Structure predicted by ColabFold.

^b No corresponding residue.

Four residues, Tyr92, Glu101, Glu188, and Tyr208 were located at subsites +1 and +2 in TmINV. The corresponding residue of Tyr92 in TmINV was Asp237 in AbFEH but those of Glu101, Glu188, and Tyr208 were not found (Table 3). In the predicted structure of AbFEH, the closest regions of the Glu101, Glu188, and Tyr208 in TmINV were identified as Ser-Glu-Gly (residues 240-242), Leu-Asn-Thr (319-321), and Gly-Gly-Arg (349-351), all of which were distantly located from the three residues in TmINV (Fig. 4A). The sequence alignment also showed that the corresponding regions of the Glu101, Glu188, and Tyr208 in TmINV were represented by gaps in AbFEH (Figs. 4B and C). The observation suggests that the substrate-binding affinity at subsites +1 and +2 in AbFEH is likely weak, resulting in the high K_m values for sucrose and 1-kestose.

Fig. 4. Comparison of the structure of AbFEH and some GH32 enzymes.

　(A) Stereo view of the residues located at subsites +1 and +2 in AbFEH predicted structure (magenta), TmINV crystal structure (PDB 1W2T, cyan), and AsInuAMN8 crystal structure (8I12, green). The trisaccharide raffinose bound to TmINV is drawn in black wireframe, and the three monosaccharide units, Fruf, Glcp, and Galp are indicated by −1, +1, and +2. In the structures of residues 240-242, 319-321, and 349-351 in AbFEH, only Cα backbones are drawn. (B, C) Amino acid residues of 240-242 (B) and 309-358 (C) in AbFEH, which were aligned to the corresponding regions in AhDO83_10515, TmInv, and AsInuAMN8. The alignment was initially performed using Clustal Omega server (https://www.ebi.ac.uk/jdispatcher/msa/clustalo) and was manually adjusted further based on the comparison among the predicted structures (AbFEH and AhDO83_10515) and the crystal structures (TmInv and AsInuAMN8). Colors: red, catalytic residue; yellow, residue at subsite −1; cyan, residue at subsites +1 and +2; magenta, residues 240-242, 319-321, and 349-351 in AbFEH. Numbering in magenta, blue, and green are for AbFEH, TmInv, and AsInuAMN8, respectively.

Despite the minimal enzymatic activity of AbFEH towards sucrose, it was nonetheless evident that it prefers the polysaccharide levan. This preference is likely due to the presence of the CBM66 domain. The function of CBM66 appears to be similar to that of CBM66 found in BsSacC [17]. In BsSacC, the CBM66 domain has been found to enhance the levan binding by interacting with terminal fructose residues, as its removal results in the decrease of enzymatic activity for levan by approximately 100-fold. Similarly, the removal of M and CBM66 domains in AbFEH∆C led to decrease the activity toward 1-kestose and levan, suggesting these domains enhance the substrate binding. Unlike BsSacC with truncated CBM66, which displayed similar activity as the full-length BsSacC against sucrose, and levan- and inulin-derived oligosaccharides [17], AbFEH∆M-CBM66-C exhibited significant low activity for both levan and 1-kestose (Table 1). BsSacC is composed of GH32 common region and CBM66 but does not possess domain M (Fig. S2; see J. Appl. Glycosci. Web site), suggesting that domain M in AbFEH likely contributes to the binding of the substrates.

To elucidate the precise substrate preference of AbFEH, activities for some other saccharides such as 6-kestose, neokestose, levanbiose, and sucrose-6-phosphate are likely necessary to determine. In fact the sugar beet 6-FEH shows the highest activity for neokestose [25]. These saccharides are, however, not available at a reasonable price, and a method using molecular docking has been employed to investigate the preference for sucrose-6-phosphate [29]. The study of sucrose-6-phosphate hydrolase SacAP1 from Priestia megaterium indicates that His68 and Lys71 are important for the recognition of sucrose-6-phosphate. The corresponding residues in AbFEH were identified as Trp175 and Met178, respectively, suggesting that AbFEH is likely to have low activity against sucrose-6-phosphate.

Anaerostipes butyraticus, a major butyrate-producing bacterium, was originally isolated from the chicken caecum [6]. The genome of a single strain of A. butyraticus has been analyzed, revealing that an open reading frame (GenBank code, GFO85653) of 4,140 base pairs is located immediately upstream of the gene for AbFEH (GenBank code, GFO85652). It has been proposed that the GFO85653 gene and the GFO85652 gene originally form a single continuous gene, and a stop codon has divided the gene into two open reading frames, one encoding 199 amino acid residues and the other encoding 1,179 amino acid residues AbFEH [7, 30]. It is unclear whether this mutation accidentally occurred, but AbFEH∆C clearly exhibited enzymatic activity in this study. The enzymatic feature of AbFEH is likely to be similar to those from Anaerostipes species such as AhDO83_10515 because of the high conservation of amino acid residues involved in their substrate binding sites (Table 3).

CONFLICT OF INTERESTS

The authors declare no conflict of interest.

ACKNOWLEDGEMENTS

We thank Yoshiko Kawabata for help with educational instruction for M.A.B. Baula, and Ding Li for help with PCR amplification. We also thank B Food Science Co., Ltd. for providing 1-kestose. This work was supported in part by a Grant-in-Aid for Scientific Research (24K08698 to TT) from the Japan Society for the Promotion of Science.

REFERENCES

• [1]  Louis P, Flint HJ. Formation of propionate and butyrate by the human colonic microbiota. Environ Microbiol. 2017; 19: 29-41.
• [2]  Drula E, Garron ML, Dogan S, Lombard V, Henrissat B, Terrapon N. The carbohydrate-active enzyme database: functions and literature. Nucleic Acids Res. 2022; 50: D571-7.
• [3]  Bali V, Panesar PS, Bera MB, Panesar R. Fructo-oligosaccharides: production, purification and potential applications. Crit Rev Food Sci Nutr. 2015; 55: 1475-90.
• [4]  Lammens W, Le Roy K, Schroeven L, Van Laere A, Rabijns A, Van den Ende W. Structural insights into glycoside hydrolase family 32 and 68 enzymes: functional implications. J Exp Bot. 2009; 60: 727-40.
• [5]  Burne RA, Schilling K, Bowen WH, Yasbin RE. Expression, purification, and characterization of an exo-β-D-fructosidase of Streptococcus mutans. J Bacteriol. 1987; 169: 4507-17.
• [6]  Eeckhaut V, Van Immerseel F, Pasmans F, De Brandt E, Haesebrouck F, Ducatelle R, et al. Anaerostipes butyraticus sp. nov., an anaerobic, butyrate-producing bacterium from Clostridium cluster XIVa isolated from broiler chicken caecal content, and emended description of the genus Anaerostipes. Int J Syst Evol Microbiol. 2010; 60: 1108-12.
• [7]  Tanno H, Fujii T, Hirano K, Maeno S, Tonozuka T, Sakamoto M, et al. Characterization of fructooligosaccharide metabolism and fructooligosaccharide-degrading enzymes in human commensal butyrate producers. Gut Microbes. 2021; 13: 1-20.
• [8]  Li D, Miyasaka Y, Kubota A, Kozono T, Kitano Y, Sasaki N, et al. Characterization and alteration of product specificity of Beijerinckia indica subsp. indica β-fructosyl­transferase. Biosci Biotechnol Biochem. 2023; 87: 981-90.
• [9]  Tonozuka T, Kitamura J, Nagaya M, Kawai R, Nishikawa A, Hirano K, et al. Crystal structure of a glycoside hydrolase family 68 β-fructosyltransferase from Beijerinckia indica subsp. indica in complex with fructose. Biosci Biotechnol Biochem. 2020; 84: 2508-20.
• [10]  Nielsen SS. Phenol-sulfuric acid method for total carbohydrates. In: Nielsen SS, editor. Food Analysis Laboratory Manual, Food Science Texts Series. Boston: Springer; 2010. p. 47-53.
• [11]  Miyasaka Y, Yokoyama K, Kozono T, Kitano Y, Miyazaki T, Sakaguchi M, et al. Structural basis for the recognition of α-1,6-branched α-glucan by GH13_47 α-amylase from Rhodothermus marinus. Proteins. 2024; 92: 984-97.
• [12]  Möglich A. An open-source, cross-platform resource for nonlinear least-squares curve fitting. J Chem Educ. 2018; 95: 2273-78.
• [13]  Kubota A, Kawai R, Li D, Kozono T, Sasaki N, Nishikawa A, et al. Enzymatic and structural characterization of β-fructofuranosidase from the honeybee gut bacterium Frischella perrara. Appl Microbiol Biotechnol. 2022; 106: 2455-70.
• [14]  Mirdita M, Schutze K, Moriwaki Y, Heo L, Ovchinnikov S, Steinegger M. ColabFold: Making protein folding accessible to all. Nat Methods. 2022; 19: 679-82.
• [15]  Wanker E, Huber A, Schwab H. Purification and characterization of the Bacillus subtilis levanase produced in Escherichia coli. Appl Environ Microbiol. 1995; 61: 1953-8.
• [16]  Sonnenburg ED, Zheng H, Joglekar P, Higginbottom SK, Firbank SJ, Bolam DN, et al. Specificity of polysaccharide use in intestinal Bacteroides species determines diet-induced microbiota alterations. Cell. 2010; 141: 1241-52.
• [17]  Cuskin F, Flint JE, Gloster TM, Morland C, Basle A, Henrissat B, et al. How nature can exploit nonspecific catalytic and carbohydrate binding modules to create enzymatic specificity. Proc Natl Acad Sci USA. 2012; 109: 20889-94.
• [18]  Buttner H, Perbandt M, Kohler T, Kikhney A, Wolters M, Christner M, et al. A giant extracellular matrix binding protein of Staphylococcus epidermidis binds surface-immobilized fibronectin via a novel mechanism. mBio. 2020; 11: e01612-20.
• [19]  Kumar P, Vyas P, Faisal SM, Chang YF, Akif M. Crystal structure of a variable region segment of Leptospira host-interacting outer surface protein, LigA, reveals the orientation of Ig-like domains. Int J Biol Macromol. 2023; 244: 125445.
• [20]  Wada T, Sugatani J, Terada E, Ohguchi M, Miwa M. Physicochemical characterization and biological effects of inulin enzymatically synthesized from sucrose. J Agric Food Chem. 2005; 53: 124653.
• [21]  Kashima T, Okumura K, Ishiwata A, Kaieda M, Terada T, Arakawa T, et al. Identification of difructose dianhydride I synthase/hydrolase from an oral bacterium establishes a novel glycoside hydrolase family. J Biol Chem. 2021; 297: 101324.
• [22]  Nagaya M, Kimura M, Gozu Y, Sato S, Hirano K, Tochio T, et al. Crystal structure of a β-fructofuranosidase with high transfructosylation activity from Aspergillus kawachii. Biosci Biotechnol Biochem. 2017; 81: 1786-95.
• [23]  Chen L, Hill A, Petit JL, Mariage A, de Berardinis V, Karboune S. Discovery and enzymatic screening of genome-mined microbial levanases to produce second-generation β-(2,6)-fructooligosaccharides: catalytic properties. ACS Chem Biol. 2023; 18: 465-75.
• [24]  Van den Ende W, De Coninck B, Van Laere A. Plant fructan exohydrolases: a role in signaling and defense? Trends Plant Sci. 2004; 9: 523-8.
• [25]  Van den Ende W, De Coninck B, Clerens S, Vergauwen R, Van Laere A. Unexpected presence of fructan 6-exohydrolases (6-FEHs) in non-fructan plants: characterization, cloning, mass mapping and functional analysis of a novel ‘cell-wall invertase-like’ specific 6-FEH from sugar beet (Beta vulgaris L.). Plant J. 2003; 36: 697-710.
• [26]  Alberto F, Jordi E, Henrissat B, Czjzek M. Crystal structure of inactivated Thermotoga maritima invertase in complex with the trisaccharide substrate raffinose. Biochem J. 2006; 395: 457-62.
• [27]  Zhou J, Lu Q, Peng M, Zhang R, Mo M, Tang X, et al. Cold-active and NaCl-tolerant exo-inulinase from a cold-adapted Arthrobacter sp. MN8 and its potential for use in the production of fructose at low temperatures. J Biosci Bioeng. 2015; 119: 267-74.
• [28]  Reddy A, Maley F. Studies on identifying the catalytic role of Glu-204 in the active site of yeast invertase. J Biol Chem. 1996; 271: 13953-7.
• [29]  Dobrange E, Porras-Domínguez JR, Van den Ende W. The complex GH32 enzyme orchestra from Priestia megaterium holds the key to better discriminate sucrose-6-phosphate hydrolases from other β-fructofuranosidases in bacteria. J Agric Food Chem. 2024; 72: 1302-20.
• [30]  Endo A, Tanno H, Kadowaki R, Fujii T, Tochio T. Extracellular fructooligosaccharide degradation in Anaerostipes hadrus for co-metabolism with non-fructooligosaccharide utilizers. Biochem Biophys Res Commun. 2022; 613: 81-6.