Introduction

Hemicellulose is one of the most relevant renewable resources of the biosphere [1], accounting for one third of the available organic carbon [2]. The hemicellulosic fraction comprises a wide variety of polysaccharides including xylans —the most abundant—, which are present in green plants and some algae [3, 4]. Xylans have aroused great interest for the production of biofuels (second-generation bioethanol) and high value-added compounds [5,6,7,8].

Xylans are made up of a β(1 → 4)-xylopyranose skeleton that can be highly decorated, e.g. with other monosaccharides ‒mostly arabinose‒, 4-O-methyl-α-D-glucuronic acid, and ferulic or acetic acids [9]. The percentage and composition of these ramifications is strongly determined by the xylan source [10]. Xylans can be hydrolyzed by chemical (autohydrolysis) [11] or enzymatic methods [12,13,14]. The latter are more convenient since they do not generate undesired by-products [15, 16]. However, due to intrinsic xylan complexity, a synergistic action of several enzymes (endoxylanases, β-xylosidases, arabinofuranosidases, α-glucuronidases, etc.) is required to achieve the complete breakdown of the polymer [17, 18].

Endo-β-1,4-xylanases (EC 3.2.1.8) belong to families GH10, GH11 and GH30 of glycosyl hydrolases (GH) [19], and hydrolyze the xylan backbone into oligomers of different size called xylooligosaccharides (XOS) [20,21,22,23]. XOS are soluble carbohydrates whose degree of polymerization (DP) is typically between 2 and 10. The composition of XOS depends on the source of the endoxylanase, the auxiliary enzymes employed and the origin of xylan [15, 16, 24,25,26]. There is an increasing interest in XOS due to their potential as second-generation prebiotics [27,28,29,30], as well as other possible health benefits related with their antioxidant, antitumorigenic, antiallergic, anti-inflammatory, and antimicrobial properties [31,32,33,34,35,36,37]. Moreover, XOS do not cause undesirable organoleptic effects, which makes them promising food ingredients for functional foods [38, 39].

Xylans containing glucuronic acid have a great potential as they inhibit the growth of several tumors [40]. Acidic xylooligosaccharides (or aldouronics) are hetero-oligomers of xylose with uronic acid residues (glucuronic or methyl-glucuronic acids) randomly linked by α-1,2 bonds, which are obtained by hydrolysis of glucurono-xylans (hardwood xylans). Several endoxylanases have been assayed for the synthesis of aldouronics [41,42,43]. These molecules possess several bioactivities including an antibacterial effect [44], antioxidant activity [33], inhibitory effect on gastric inflammation [45], and preventive effects on atopic dermatitis [45, 46]. Some of these studies have compared neutral and acidic XOS and conclude that the presence of uronic ramifications is an important determinant of bioactivity [33, 46]. Therefore, it is of great importance to develop methodologies for the synthesis and characterization of acidic xylooligosaccharides, to deepen the study of structure–activity relationships of these carbohydrates.

The goal of this work was to develop a simple and scalable method for the enzymatic production and isolation of acidic xylooligosaccharides by hydrolysis of a glucurono-xylan (from beechwood). Additionally, the major compound was structurally analyzed and its antioxidant activity assayed.

Materials and methods

Enzymes and reagents

Pectinex Ultra SP-L, Shearzyme 500L, Ultraflo L, and Vinoflow were kindly donated by Novozymes. Depol 670L, Depol 692L and Depol 740L were acquired from Biocatalysts Ltd. Rapidase TF was kindly supplied by DSM. Beechwood xylan was purchased from Megazyme. D-xylose was from Merck. ABTS (2,2′-Azino-bis(3-ethylbenzothiazoline-6-sulfonic acid) diammonium salt) and R-Trolox were from Sigma-Aldrich. Standards of xylooligosaccharides from DP1 to DP6 were acquired in Megazyme (Wicklow, Ireland). A mixture of linear XOS from DP1 to DP6 was from Wako Pure Chemical Industries Ltd. Dextran molecular weight standards kit (180 to 43,500 Da) was from Phenomenex. Sep-Pak Accell Plus QMA Plus strong anion exchange cartridges (10 g, 35 mL, 300 Å) were from Waters. Dialysis tubing with MWCO 0.1–0.5 kDa was from VWR.

Enzyme activity

The xylanolytic activity was determined by the 3,5-dinitrosalicylic acid (DNS) method [47] adapted to 96-well microplates. For the assay, 45 μL of 1% (w/v) beechwood xylan (previously dissolved at 80 °C in 10 mM sodium acetate buffer pH 4.6) was mixed with the enzyme solution (5 μL), conveniently diluted to fit into the D-xylose calibration curve. The mixture was incubated at 60 °C for 30 min and the concentration of reducing sugars determined by DNS. One unit of activity (U) corresponded to the release of one µmol of xylose reducing equivalents per minute.

Small-scale production of XOS by enzymatic hydrolysis of beechwood xylan

Hydrolysis reactions were performed with 2% (w/v) beechwood xylan and 3 U/mL of enzyme (referred to total reaction volume) in distilled water. Reaction mixtures (5 mL) were incubated at 60 °C under orbital stirring and aliquots were withdrawn at different time points. Aliquots were mixed with absolute ethanol to 70% (v/v) final solvent concentration in order to inactivate the enzyme and precipitate the unreacted xylan. Samples were centrifuged at 3000 rpm for 10 min and the supernatant was analyzed by high-performance anion-exchange chromatography coupled with pulsed amperometric detection (HPAEC-PAD).

Analysis of the xylan hydrolysis by size exclusion chromatography

The evolution of xylan hydrolysis was analyzed by size exclusion high-performance liquid chromatography (HPLC-SEC). The equipment consisted of a Varian 9012 pump, an autosampler L-2200 (VWR), and an evaporative light scattering detector (ELSD 2000-ES, Alltech). A 300 × 7.6 mm PolySep 4000 column (Phenomenex) was used at 25 °C. The analysis was performed at isocratic conditions using 0.25 M ammonium acetate (pH 4.7) as mobile phase. The flow rate was 0.6 mL/min and 20 µL of samples were injected. The ELSD detector conditions were set at 118 °C and 3.2 L/min of N2. Prior to the analysis, the samples were diluted 1:2 (v/v) with water and filtrated through 0.45 µm nylon filters. Analyses were performed in duplicate.

Analysis of XOS by HPAEC-PAD

Reaction samples were diluted with water (1:10), filtered through 0.45-µm nylon filters (Cosela), and analyzed by HPAEC-PAD on a Dionex ICS3000 system (Dionex, Thermo Fischer Scientific Inc., Waltham, MA) consisting of an SP gradient pump, an electrochemical detector with a gold working electrode and Ag/AgCl as reference electrode, and an autosampler (model AS-HV). All eluents were degassed by flushing with helium. A pellicular anion-exchange 4 × 250 mm Carbo-Pack PA-200 column (Dionex) connected to a 4 × 50 mm CarboPac PA-200 guard column was used at 30 °C. Eluent preparation was performed with Milli-Q water, NaOH and sodium acetate (NaOAc).

Two methods were developed for XOS analysis, depending on the origin of the sample. In method A, the initial mobile phase was 15 mM NaOH at 0.5 mL/min for 12 min. A gradient from 15 to 75 mM NaOH and from 0 to 80 mM NaOAc was performed in 8 min at the same flow rate. Then, the column was washed for 10 min with a solution containing 320 mM NaOAc and 100 mM NaOH, and further equilibrated with 15 mM NaOH. Method B was designed for a better separation of the acidic pool, and was also performed at a flow rate of 0.5 mL/min. The mobile phase contained 10 mM NaOH from start to end, whilst two gradients were performed with NaOAc. The first was an 80 min-gradient from 0 to 60 mM NaOAc, and the second was from 60 to 160 mM in 20 min. Finally, the column was equilibrated back to 10 mM NaOH. The identification of the different carbohydrates was done based on commercially standards when available (glucose, arabinose and xylooligosaccharides from DP1 to DP6). Peaks were integrated using the Chromeleon software. All analyses were performed in duplicate.

Scale-up of synthesis and fractionation of acidic XOS

The reaction mixture (100 mL) contained 2% (w/v) beechwood xylan and 3 U/mL of enzyme (referred to total reaction volume) in distilled water. It was incubated at 60 °C under orbital stirring during 72 h to favor the formation of acidic XOS. Then, 3 volumes of absolute ethanol were added to inactivate the enzyme and precipitate the remaining xylan. The reaction mixture was concentrated by rotary evaporation in a R-210 rotavapor (Buchi) until the reaction volume was approximately 30 mL. The solution was freeze-dried (Freezone 2.5 PLUS, Labconco) and the resulting XOS desiccated with P2O5 overnight.

Neutral and acidic XOS were fractionated by solid phase extraction (SPE) employing an extraction manifold (Waters) connected to a vacuum pump. A silica-based, hydrophlilic, strong anion exchange cartridge (Sep-Pak Accell Plus QMA 35 cc Vac Cartridge, 10 g, Waters) with large pore size (300 Å) was used for the isolation of the XOS. The cartridge was conditioned and equilibrated with distilled water. The solid XOS were dissolved in 10 mL of water and the sample was loaded into the cartridge. Neutral XOS were eluted with 70 mL of distilled water as mobile phase. It was important to adjust the vacuum pump so that a flow rate of 1–2 drops/s was obtained. Afterwards, 70 mL of 10 mM ammonium acetate buffer (pH 7.4) were used to elute the acidic XOS.

Both fractions were analyzed by HPAEC-PAD. The acidic pool was concentrated by rotary evaporation and further dialyzed using MWCO 0.1-–0.5-kDa tubing (VWR) to eliminate the remaining salts. Finally, samples were freeze-dried (model Freezone 2.5 PLUS, Labconco) and analyzed by HPAEC-PAD, mass spectrometry and NMR analysis.

Mass spectrometry

The molecular weight of neutral and acidic XOS was assessed by MALDI-TOF mass spectrometry using an equipment with Ultraflex III TOF/TOF (Bruker, Billerica, MA, USA) and a Nd-YAG laser. Registers were taken in positive reflector mode within the mass interval 40–5000 Da, with external calibration and with 20 mg/mL 2,5-dihydroxybenzoic acid (DHB) in acetonitrile (3:7) (v/v) as matrix. Samples were mixed with the matrix in a 4:1 proportion and 0.5 µL were analyzed.

NMR analysis

The structure of the major acidic XOS was assessed by combining one-dimensional [1D, 1H and 1D-selective TOCSY experiments] and two-dimensional [2D, COSY, TOCSY, NOESY, HSQC, and HMBC] NMR experiments. The compounds were solubilized in deuterated water (ca. 10 mM). The spectra were recorded on a Bruker AVIII-600 spectrometer equipped with a TXI probe with gradients in the z axis, at 298 K. Chemical shifts were expressed in parts per million (ppm) with respect to the 0 ppm point TSP-d4, employed as internal standard. The pulse sequences were provided by Bruker. For the HSQC experiment, values of 5 ppm and 1024 points for the 1H dimension and 70 ppm and 512 points for the 13C dimension were used. For the HMBC experiment, values of 5 ppm and 1024 points for the 1H dimension and 140 ppm and 384 points for the 13C dimension were used. For the homonuclear COSY and NOESY experiments, 4 ppm windows were used with a 1024 × 256–512 point matrix. For the NOESY experiments, the mixing time was 500 ms.

Elemental analysis

Elemental analysis was performed with an analyzer LECO CHNS-932 to estimate the remaining salt (ammonium acetate) in the final product, based on the nitrogen content.

Antioxidant activity of acidic XOS

Scavenging of ABTS radical cation was assayed as described [48], adapted to 96-well plates. The radical cation was produced dissolving 7 mM ABTS in water and mixing with a 2.45 mM (final concentration) potassium persulfate solution. The reaction was kept in the dark at room temperature overnight. Then, the solution was diluted with water until the absorbance at 734 nm was 0.7. Neutral and acidic XOS were tested at different concentrations by adding 20 µL of these solutions to 230 µL of the ABTS·+ solution previously prepared. The absorbance at 734 nm was monitored during 10 min using different concentrations of the compounds. The decrease of absorbance (expressed in percentage referred to the initial value) was calculated. These values were represented vs. the concentration of products, and the SC50 (concentration at which each substance decreases to half the absorbance of radical cation ABTS·+ in 10 min) was calculated. (R)-Trolox was used as reference antioxidant. Samples were measured in triplicate and data was expressed as the media ± standard deviation.

Results and discussion

Xylanolytic activity screening

The hydrolytic activity of several commercial fungal enzymes towards 1% (w/v) beechwood xylan was assayed under standard conditions: pH 4.6 and 60 °C. The enzymes were selected based on their activity towards polysaccharides. The activity was quantified by the dinitrosalicylic (DNS) acid assay, which is based on the detection of the released reducing sugars. The calibration curve was obtained with xylose.

Table 1 summarizes the enzymes that displayed a significant activity. It is noteworthy that several enzyme preparations whose declared activity is not xylanolytic (e.g., Ultraflo L, a β-glucanase from Humicola insolens) displayed a notable activity vs. beechwood xylan.

Table 1 Screening of xylanolytic activity in commercial enzyme preparations

The most active preparation was Depol 670L (70.3 U/mL, expressed in reducing xylose equivalents), which was selected for further experiments. Depol 670L is a multi-enzymatic preparation from Trichoderma reesei developed for the hydrolysis of lignocellulosic materials [49].

Production of XOS by enzymatic hydrolysis of beechwood xylan

The hydrolysis of beechwood xylan by Depol 670L was analyzed in detail. The reaction was carried out using 2% (w/v) beechwood xylan as substrate. The amount of enzyme was adjusted to 3 U/mL (measured by the DNS method, see Experimental Section), referred to the final reaction volume. We observed that the optimum temperature for this cellulolytic preparation from Trichoderma reesei was 60 °C (data not shown). Regarding the optimum pH, although the highest activity was observed in the pH range 4.5–5.0, the reaction was performed in water to facilitate the further separation of neutral and charged XOS by ion-exchange cartridges.

The evolution of xylan hydrolysis was followed by size-exclusion chromatography with ELSD detection (Fig. 1). The column was calibrated with dextrans from 180 to 43,500 Da. As shown, the molecular weight of the substrate decreased significantly with time, and it tended to stabilize after 22 h. The peak on the right (marked with an asterisk) corresponds to salts present in the enzyme and xylan, and increased with time due to the formation of free xylose. The SEC chromatograms show that Depol 670L is very efficient in the hydrolysis of xylan.

Fig. 1
figure 1

Size exclusion chromatograms (SEC-ELSD) of a hydrolysis of 2% (w/v) beechwood xylan with Depol 670L at different reaction times (the peak marked with asterisk corresponds to salts and free xylose); b dextran standards of molecular mass from 180 to 43,500 Da

The progress of XOS formation was also followed by anion-exchange chromatography with pulsed amperometric detection (HPAEC-PAD), which offers the highest resolution for this kind of mixtures [17, 23]. The chromatograms at 3 h and 48 h are shown in Fig. 2. We observed two groups of peaks: the less retained, which correlated with neutral XOS standards; and the more retained, which should correspond to acidic XOS (glucuronylated). The production of neutral XOS reached its maximum yield at 3 h and was significantly lower at 48 h, with xylose and xylobiose as the main reaction products. Acidic XOS appeared to follow the same progress, since the number of peaks corresponding to acidic compounds decreased substantially after 3 h. However, two main peaks remained stable throughout the reaction. We observed that one of these two peaks diminished considerably between 48–72 h (data not shown). Our data indicated that the stability of the xylanase from Trichoderma reesei was significant under reaction conditions. As the goal of this work was to obtain acidic XOS with a defined composition, 72 h was chosen as the optimum reaction time.

Fig. 2
figure 2

HPAEC-PAD chromatograms at 3 and 24 h. Ara: arabinose; Glc: glucose; Xyl: Xylose; Xyl2: xylobiose; Xyl3: xylotriose; Xyl4: xylotetraose; Xyl5: xylopentaose; Xyl6: xylohexaose

Large-scale purification of acidic XOS

The scheme of the process followed to isolate acidic XOS is represented in Fig. 3. The reaction was scaled to 100 mL using 2% (w/v) beechwood xylan in water and 3 U/mL of Depol 670L (step 1). After incubation at 60 °C during 72 h, 3 volumes of ethanol were added to inactivate the enzyme and precipitate the remaining xylan (step 2). After centrifugation, the reaction mixture was concentrated by rotary evaporation and the resulting product was lyophilized and desiccated (step 3). Approximately 1.5 g of total carbohydrates (75% yield) was obtained. The percentage of hydrolyzed beechwood xylan is higher than the reported with other xylanolytic enzymes, which is typically lower than 20% [33, 44]. This can be the consequence of the presence of a cocktail of hydrolytic enzymes in Depol 670L preparation, as the synergistic action of xylanolytic enzymes is demonstrated to increase the yield of XOS [50].

Fig. 3
figure 3

Scheme for the scale-up of synthesis and fractionation of acidic XOS

The separation of neutral and acidic XOS was achieved by anion-exchange chromatography using solid-phase extraction cartridges (step 4). Neutral XOS were eluted with water and the acidic XOS with 10 mM ammonium acetate pH 7.4. The acidic XOS fraction was concentrated, dialyzed with 0.1-–0.5-kDa tubing to eliminate the remaining ammonium acetate, and finally lyophilized (step 5).

After lyophilization, elemental analysis was used to approximately calculate the amount of ammonium acetate in the sample based on the nitrogen percentage. We estimated that ammonium acetate comprised about 20% (w/w) of the final product. The amount of obtained product was 225 mg, which corresponded to a yield of 11.2%. The methodology for production and fractionation of acidic XOS was easily reproducible.

Characterization of acidic XOS

Both fractions (neutral and acidic) were analyzed by HPAEC-PAD (Fig. 4). As shown, the separation of both fractions was very effective. The neutral fraction was formed mainly by xylose and xylobiose, whereas the acidic one was more heterogeneous. However, this fraction contained a major product that was characterized by mass spectrometry and NMR.

Fig. 4
figure 4

HPAEC-PAD analysis of a neutral pool; b acidic pool. Xyl: xylose; Xyl2: xylobiose. The main peak of the acidic fraction is marked with an asterisk

MALDI-TOF analysis (Fig. 5) showed the presence of molecular ions of substituted oligosaccharides. The major peak showed a mass of 495.2 Da, which corresponds to a product containing two molecules of xylose and one of 4-O-methyl-α-D-glucuronic acid (X2_MeGlcA + Na +) and an additional peak of 511.1 Da that corresponded to the potassium adduct. In addition, some minor peaks were observed, which corresponded to acidic XOS containing 3–6 xylose residues and a single 4-O-methyl-α-D-glucuronic acid residue (X3_MeGlcA to X6_MeGlcA).

Fig. 5
figure 5

MALDI-TOF analysis of the acidic fraction of XOS. Xn: number of xyloses bonded by β(1 → 4) linkages; MeGlcA: 4-O-methyl-α-D-glucuronic acid

The structure of the main compound present in the acidic fraction was determined by NMR techniques. Four anomeric signals were found in the HSQC spectra (Fig. 6): two for the α and β anomeric forms of the non-reducing end and two for the other residues present in the major compound. With the aid of 1D-selective TOCSY experiments (Figure S1, Supplementary Material) the signals of every individual residue were identified. The glycosidation positions were deduced from the NOESY and HMBC experiments (Fig. 6) that correlate protons close in space or in the number of separating bonds, respectively. The major product was 2′-O-α-(4-O-methyl-α-D-glucuronosyl)-xylobiose (Fig. 7). Table 2 summarizes the 1H/13C chemical shifts and other relevant information as J couplings, and the key correlations of HMBC and NOESY spectra that led to determine the structure.

Fig. 6
figure 6

a HSQC, b HMBC, and c NOESY spectra of the sample where the trisaccharide MeGlcAα-(1 → 2)-Xylβ-(1 → 4)-Xyl is the major product

Fig. 7
figure 7

Structure of the trisaccharide MeGlcAα-(1 → 2)-Xylβ-(1 → 4)-Xyl

Table 2 1H/13C chemical shifts of the trisaccharide 4-O-Me-α-D-GlcpA-(1 → 2)-β-D-Xylp-(1 → 4)-D-Xylp. For some key positions, J coupling values are provided to show anomeric conformation. Glycosylation or methylation 13C positions are indicated in bold. N and H columns indicate relevant NOESY or HMBC correlations for each position, respectively. The numeration of units is indicated in Fig. 7

Antioxidant activity of acidic XOS

The antioxidant activity of XOS produced by hydrolysis of xylans, mostly containing ferulic acid substituents, has been widely reported [51,52,53,54]. However, Valls et al. were the first in describing the antioxidant activity of XOS derived from glucuronoxylans [33]. In particular, the authors analyzed mixtures containing acidic XOS from X3_MeGlcA to X8_MeGlcA.

In the present work, we have analyzed the antioxidant activity of acidic XOS containing majorly one single product (X2_MeGlcA). The selected method was the ABTS·+ radical discoloration assay adapted to 96-well plates [55]. Results were compared with those obtained with beechwood xylan and a commercially available mixture of neutral XOS. Trolox, a well-known and powerful antioxidant commonly employed as reference in antioxidant activity assays, was also analyzed to determine the Trolox Equivalent Antioxidant Activity (TEAC). Since TEAC is used to compare molar concentrations of pure compounds, and the synthesized trisaccharide is slightly contaminated with larger acidic XOS (mainly X3_MeGlcA and X4_MeGlcA), the molar concentration could not be determined. For that reason, we defined the half maximal scavenging concentration (SC50) as the concentration (in mg/mL) at which the compound is able to decrease 50% the absorbance of radical cation ABTS·+ in 10 min (Fig. 8).

Fig. 8
figure 8

ABTS·+ discoloration (measured at 734 nm) in the presence of beechwood xylan, acidic XOS (X2_MeGlcA) and neutral XOS

SC50 was calculated for the synthesized acidic product (X2_MeGlcA), neutral XOS, xylan and Trolox (Table 3). As shown, neutral XOS displayed no antioxidant activity, whereas the acidic XOS exhibited higher activity than xylan. This can be a consequence of the enrichment in acidic groups compared with the polysaccharide. In this context, Valls et al. reported that the presence of MeGlcA ramifications determined the antioxidant activity of this kind of compounds [33]. Such activity could be related to the presence of a carbonyl group (C = O) linked to the sugar in the form of uronyl or acetyl substitutions [56]. However, other factors such as the length of the xylooligosaccharide exert an influence on antioxidant activity [33].

Table 3 Concentration (SC50) at which xylan, neutral and acidic XOS decrease to half the absorbance of radical cation ABTS·+ in 10 min

In conclusion, we describe in this work a simple and reproducible methodology for the production of a safe and natural antioxidant by enzymatic hydrolysis of a renewable resource, the glucurono-xylan from beechwood. The product is enriched in the trisaccharide X2_MeGlcA and free of neutral XOS. Although its antioxidant activity is significantly lower than the obtained with Trolox, the toxicity and side effects of synthetic antioxidants favors the implementation of natural substances in the food industry. In addition, other bioactivities for X2_MeGlcA can be explored, in particular its anti-inflammatory, antimicrobial and prebiotic properties.