Novel Function of CtXyn5A from Acetivibrio thermocellus: Dual Arabinoxylanase and Feruloyl Esterase Activity in the Same Active Site

Abstract

Enzymes’ uncharacterised side activities can have significant effects on reaction products and yields. Hence, their identification and characterisation are crucial for the development of successful reaction systems. Here, we report the presence of feruloyl esterase activity in CtXyn5A from Acetivibrio thermocellus, besides its well‐known arabinoxylanase activity, for the first time. Activity analysis of enzyme variants mutated in the catalytic nucleophile, Glu279, confirmed removal of all activity for E279A and E279L, and increased esterase activity while removing xylanase activity for E279S, thus allowing the proposal that both reaction types are catalysed in the same active site in two subsequential steps. The ferulic acid substituent is cleaved off first, followed by hydrolysis of the xylan backbone. The esterase activity on complex carbohydrates was found to be higher than that of a designated ferulic acid esterase (E‐FAERU). Therefore, we conclude that the enzyme exhibits a dual function rather than an esterase side activity.

Same site, different outcome: A novel feruloyl esterase functionality of arabinoxylanase CtXyn5A from A. thermocellus has been discovered. The enzyme catalyses both types of reaction in a sequential fashion in the same active site. As its esterase activity is higher on carbohydrate polymers than that of a designated ferulic acid esterase, we conclude that the enzyme exhibits a dual function rather than an esterase side activity.

Introduction

Enzymes have become widely popular in industry due to their outstanding substrate selectivity and ability to catalyse very specific reactions, thereby minimising the generation of unwanted by‐products.[ ¹ , ² ] However, they are not perfect in this regard, and in the context of carbohydrate metabolism, evolution of enzymes that catalyse more than one reaction in the degradation of complex plant materials can be advantageous. Hence, enzymes are known that catalyse reactions other than their main activity as well as accept different substrates. This feature is suspected to be a result of evolutionary divergence of one multifunction enzyme into a variety of catalytically different enzymes.[ ¹ , ³ , ⁴ ] Due to the complexity of natural substrates, enzymes with dual functions originating from the presence of two catalytic domains can also be found. ^[5]

The additional or side activities of these enzymes are often disregarded in industrial applications. However, they can have significant impacts on the manufacturing process or final product. In the case of dough development, it has been shown that the commonly applied xylanase, Pentopan Mono BG, impacts the dough's rheological properties through a transglutaminase side activity. ^[6] Hence, characterising enzymatic side reactions and multi‐functions does not only provide us with a greater understanding of natural biochemical processes, but is also crucial for the development of successful industrial operations and products involving the use of enzymes. ^[7]

Carbohydrate active enzymes are widespread in the food, feed and pharma industries. ^[2] Many interesting candidates can be found in the glycoside hydrolase family 5 (GH5), which is characterised by a wide range of members all hydrolysing β‐linked oligosaccharides, polysaccharides and glycoconjugates. The recently established subfamily 34 (GH5_34) comprises arabinoxylanases, which are active on the β‐1,4 glycosidic linkage between two xylopyranose (Xylp) units of which one is α‐1,3‐linked to an arabinofuranosyl unit (Araf). This substitution is a requirement that must be present for hydrolysis to occur[ ⁸ , ⁹ ] (Figure 1). Neighbouring the ‐1 subsite (harbouring the catalytic acid base [Glu171] and nucleophile [Glu279]) is a pocket labelled ‐2* which accommodates and binds to the critical Araf substituent. The three amino acids responsible for this specificity of the ‐2* sub site are Glu68, Tyr92 and Asn139. These residues interact with Araf in the ‐2*subsite and are conserved in all members of the family. ^[10]

Figure 1.

Illustration of the reaction catalysed by the arabinoxylanase from GH5_34. The figure shows hydrolysis of a xylan fragment substituted with α‐l‐arabinose on the second xylose unit from the left to result in a product with an arabinosylated reducing‐end xylose.

The first member of this family to be characterised was CtXyn5A, which is a two‐domain variant (GH5‐CBM6) of the multimodular enzyme from Acetivibrio thermocellus ATCC 27405 (previously Clostridium thermocellum), a bacterium found in hot springs and self‐rotting biomass.[ ⁹ , ¹¹ , ¹² ] CtXyn5A is especially accommodating of Araf substitutions, allowing them in all its subsites ranging from −2 to +2. Therefore, it can act on more highly Araf‐substituted substrates than classical endo‐xylanases found in the glycoside hydrolase families 10 and 11, generating interesting new products.[ ¹⁰ , ¹¹ , ¹² ] Hydrolysis of rye arabinoxylan revealed that all reaction products contained an α‐1 Araf substitution linked to the O3 of the Xylp unit at the reducing end. About 30 % of the products contained an additional α‐1 Araf substitution. The size of the products ranged between degrees of polymerisation (DP) of 2 to 10 with an average DP of 2.8 in the backbone and 4.0 overall.[ ⁹ , ¹¹ , ¹³ ]

In commelinid plants, to which important food crops such as cereals belong, the arabinoxylan often contains a hydroxycinnamic acid substitution at the C5 hydroxy group of its Araf units. ^[14] Most commonly, comprising 90 % of the cases, this substitution is an esterified ferulic acid (FA). ^[15] As self‐rotting biomass is the natural habitat of A. thermocellus, ^[12] the CtXyn5A producer, the enzyme has most likely developed mechanisms to cope with feruloylated arabinoxylans. However, so far there are no reports describing any interactions of CtXyn5A with feruloylated arabinoxylan.[ ¹⁵ , ¹⁶ ] Increasing interest in the production of health‐promoting arabinoxylo‐oligosaccharides (AXOS) as well as feruloylated arabinoxylo‐oligosaccharides (FAXOs) demands for a better understanding of the action of GH5_34 members on feruloylated arabinoxylan (Figure 2).

Figure 2.

CtXyn5A esterase activity releasing the ferulic acid linked to O‐5 of arabinose in the arabinoxylan chain.

Here, we report the interaction of CtXyn5A, a GH5_34, with FA substitutions on arabinoxylan for the first time. CtXyn5A was found to exhibit esterase activity, cleaving the FA substitution from the Araf units in complex substrates, hence, making the substrate more accessible for the enzyme's catalytic glycosyl activity. This esterase activity was further characterised on model substrates and explanations of structural mechanisms for the observed behaviours are given.

Results and Discussion

During our previous study on synergistic enzymatic degradation of pretreated complex carbohydrates originating from the agricultural side streams oat hull, oat bran and corn bran, the arabinoxylanase CtXyn5A in combination with laccase (a copper enzyme that oxidises various types of phenols and similar lignin‐related aromatic compounds with the reduction of molecular oxygen to water), resulted in a similar modification of the lignin fraction as the combination of a ferulic acid esterase (FAE) with the laccase. ^[17] This observation was only made for these two enzyme combinations, leading us to suspect an esterase activity of CtXyn5A.

Therefore, the release of ferulic acid (FA) from two complex carbohydrate polymers (enriched in feruloylated arabino‐substituted xylans) was analysed after incubations with CtXyn5A. The two arabinoxylan containing samples were prepared by ultrasound assisted alkali pre‐treatment of corn bran and oat hull and both polymers contained a significant amount of FA (Table 1). CtXyn5A was found to release more FA from the arabinoxylan backbone than the FAE used (Figure 3). These data suggest that CtXyn5A exhibits a rather high esterase activity. Additionally, this esterase activity may increase the arabinoxylanase activity as the removal of the feruloyl group makes the arabinoxylan backbone more accessible. Due to both of these observations, the CtXyn5A‐enzyme could be classified as multifunctional rather than exhibiting a FAE side activity. The term multifunctionality generally describes an enzyme catalysing more than one biologically relevant reaction, ^[1] which in this case corresponds to an enzyme with significant feruloyl esterase and arabinoxylanase activities.

Table 1.

Specific activities of the enzymes ferulic acid esterase (FAE) E‐FAERU and CtXyn5A on the model substrates at 1 mM concentration.

	Specific activity [μmol min−1 mg−1]
Enzyme	pNP‐butyrate	pNP‐laurate	pNP‐ferulate
E‐FAERU	0.081	0.042	0.040
CtXyn5A	0.213	0	0
CtXyn5A (imidazole corrected)	0.036	0	0

Figure 3.

Amount of free ferulic acid as a percentage of the total amount of ferulic acid present in the ultrasound‐assisted alkali pretreated substrates oat hull and corn bran prior to any reaction (control) and after reaction with the FAE E‐FAERU, CtXyn5A and imidazole (n=3).

The commercially available CtXyn5A formulation from NZYTech was used in the trials. This formulation contains imidazole from the purification procedure (used in the elution step in the affinity chromatography), which is necessary for the storage of the enzyme in soluble form. However, imidazole is known to chemically cleave ester bonds.[ ¹⁸ , ¹⁹ ] In order to exclude that the presence of imidazole is responsible for the increased ferulic acid release in the reactions with CtXyn5A, control reactions containing only imidazole in the same concentration as present in the CtXyn5A formulation were performed. After the imidazole reaction on both oat hull and corn bran, the detected amount of free FA was higher than in the control experiment, where no enzyme or imidazole was added, but lower than the amount of free FA present after reaction with imidazole containing CtXyn5A (Figure 3). The difference in free FA released can thus be attributed to the action of the CtXyn5A enzyme itself. Contradicting previous suggestions, ^[15] CtXyn5A hence seems to be rather unsuitable for the production of feruloylated arabinoxylo‐oligosaccharides (FAXOs).

To better characterise the ability of CtXyn5A to cleave ester bonds, activity assays on the model substrates pNP‐butyrate (four carbon chain), pNP‐laurate (12‐carbon chain) and pNP‐ferulate (most similar to the bond in the complex carbohydrates) were performed, and compared to the activity of the FAE, E‐FAERU. The results are presented in Figure 4. Both enzymes, as well as imidazole, showed activity on pNP butyrate (Figure 4A). The specific activities presented in Table 1 suggest that CtXyn5A is the more active enzyme on this substrate. However, as shown in Figure 4A, presence of imidazole also results in release of pNP from pNP‐butyrate. Therefore, this contribution was subtracted from the slope of CtXyn5A and a specific activity lower than that of E‐FAERU was found (0.036 vs 0.081 μmol min⁻¹ mg⁻¹). In contrast, only E‐FAERU was active on pNP‐laurate (Figure 4B) and pNP ferulate (Figure 4C) with low, but very similar specific activities of 0.042 and 0.040 μmol min⁻¹ mg⁻¹, respectively. The E‐FAERU reaction on pNP‐ferulate was surprisingly characterised by a rather lengthy lag phase of 40 minutes, after which the first pNP cleavages could be detected. Repeated trials, showed that the long lag phase could be due to poor substrate solubility, making incubation for a period of time necessary to solubilise enough substrate for the reaction to begin.

Figure 4.

Enzyme activity curves describing the amount of pNP released from the substrates A) pNP‐butyrate, B) pNP‐laurate and C) pNP‐ferulate by ferulic acid esterase (FAE), CtXyn5A or imidazole over time (n=3). The equations in the boxes only describe the linear part of each curve. The legend for all graphs can be found in (C).

The lack of activity on pNP‐laurate and ‐ferulate suggests that the esterase activity of CtXyn5A on pNP‐linked substrates is limited to ester bonds linked to rather short‐chain substrates with high reactivity, which stands in contrast to the observations on complex carbohydrates. These latter substrates tend to have a high carbohydrate chain length, ^[20] but in these cases CtXyn5A cleaved the ferulate ester bound to the complex carbohydrate polymer rather efficiently. Interactions with the carbohydrate chain further away from the feruloyl group might thus facilitate activity.

Based on the ambiguous results obtained from the pNP‐substrate analysis using the wild‐type enzyme, a point mutation was introduced in which the nucleophile Glu279 in the active site was replaced by serine. The E279S exchange gives the active site a residue composition more similar to the active sites of typical esterase families. Ferulic acid esterases are characterised by a Ser‐His‐Asp/Glu catalytic triad in the active site. The serine nucleophile attacks the carbonyl carbon of the substrate, releasing FA in a multistep mechanism involving two covalent tetrahedral intermediates. ^[21] By replacing the stronger nucleophile Glu279 in the active site of CtXyn5A with the weaker serine, it is hypothesised that the glycosidase activity is eliminated, while the esterase activity is enhanced. The removal of glycosidase activity by this approach has been confirmed in structural studies of CtXyn5A (PDB ID: 5LA2). ^[10]

To further investigate if this active site is exclusively responsible for either the glycosidase or the esterase activity, two more mutants were designed: E279L and E279A, in which Glu279 was replaced with leucine and alanine, respectively. It was hypothesised that, with these mutations, both activities will be eliminated. Moreover, this would allow us to conclude whether both activities originate from the CtXyn5A active site, which in that case would display a dual function.

The esterase activities of CtXyn5A‐E279S, CtXyn5A‐E279A and CtXyn5A‐E279L were first screened using the pNP model substrates (Figure 5). As observed for CtXyn5A (Table 1), a very low activity on pNP‐butyrate was found for CtXyn5A‐E279S (0.032 μmol min⁻¹ mg⁻¹ after imidazole correction). No activity was detected for CtXyn5A‐E279A or CtXyn5A‐E279L after subtracting the imidazole reaction. This showed that most of the activity can be attributed to the chemical reaction of pNP‐butyrate with imidazole. In addition, no activity was detected on the longer and bulkier substrates pNP‐laurate and pNP‐ferulate, which was unexpected for CtXyn5A‐E279S since its active site resembled the one in E‐FAERU that showed activity on all three substrates (Figure 4).

Figure 5.

Enzyme activity curves describing the amount of pNP released from the substrates pNP‐butyrate, pNP‐laurate and pNP‐ferulate by A) CtXyn5A E279S, B) CtXyn5A‐E279A and C) CtXyn5A‐E279L over time (n=3). The equations in the box only describe the linear part of the pNP‐butyrate curve.

Incubation of CtXyn5A‐E279S with the complex xylans from oat hull and corn bran, however resulted in a much higher release of FA, compared to corresponding incubations with either CtXyn5A or E‐FAERU (Figure 6). In contrast, the released FA from reactions including either CtXyn5A‐E279A and CtXyn5A‐E279L only corresponded to the effect of imidazole (Figure 6). The increased esterase activity with the serine nucleophile, and elimination of the same when the nucleophile was replaced with alanine or leucine, confirmed the hypothesis that catalysis occurs in the same active site and that the enzyme exhibits a dual function, as discussed above. With those variants, reactions were also performed containing a fivefold higher enzyme concentration in the same volume to preserve comparable amounts of imidazole in all samples.

Figure 6.

Amount of free ferulic acid as a percentage of the total amount of ferulic acid present in the ultrasound‐assisted alkali pretreated substrates oat hull and corn bran (control) and after reaction with CtXyn5A, FAE, imidazole, CtXyn5A‐E279A, CtXyn5A‐E279A(x5), CtXyn5A‐E279L, CtXyn5A‐E279L(x5) and the CtXyn5A‐E279S (n=3). The designation x5, means that a fivefold higher concentration of enzyme was used in the reaction.

The point mutation of the Glu279 nucleophile to Ser, Ala, or Leu was also expected to eliminate the ability of the enzyme to hydrolyse glycosidic bonds. Analysis of reactions with all enzyme variants on xylans from oat hulls and corn bran showed that the chromatograms after reactions with the three mutated variants were identical to those of the substrates alone, while hydrolysis products were detected after the reaction with CtXyn5A (Figure 7). Hence, it was confirmed that all the three introduced mutations eliminated xylanase activity, and only the CtXyn5A‐E279S variant displayed enhanced ferulic acid esterase activity on long carbohydrate substrates.

Figure 7.

HPAEC‐PAD chromatograms showing the presence of oligosaccharides in the substrates A) oat hull xylan and B) corn bran xylan prior to enzymatic treatment (pink), after treatment with CtXyn5A (black) and after treatment with CtXyn5A‐E279S (blue). C) Oat hull before (black) and after treatment with CtXyn5A‐E279A (blue), CtXyn5A‐E279A(x5) (pink), CtXyn5A‐E279L (brown) and CtXyn5A‐E279L(x5) (cyan) and D) Corn bran xylan before (blue) and after treatment with CtXyn5A‐E279A (black), CtXyn5A(x5) (pink), CtXyn5A‐E279L (brown) and CtXyn5A‐E279L(x5) (cyan).The displayed chromatograms are cropped to enhance the visibility of the peaks. The full chromatograms are visible in the upper right corners.

Enzymes displaying both xylanase and feruloyl esterase activity from A. thermocellus (previously C. thermocellum) have been identified before. However, in those cases, the reactions were catalysed in different domains of a multiprotein complex, the cellulosome.[ ²¹ , ²² ] To our knowledge, here we have identified an arabinoxylanase and feruloyl esterase dual function enzyme from A. thermocellus, where both reactions are catalysed in the same active site for the first time. This finding suggests an interesting and so far unexplored reaction mechanism.

Ligand docking studies revealed two potential binding sites for feruloyl‐5‐arabino‐α‐1,3‐xyloside in CtXyn5A. One was located in the carbohydrate binding module (CBM6), without any potential catalytic amino acid, and another one in the GH5 catalytic module in the same active site as for the glycoside hydrolysis. Thus, this docking study suggests that the same catalytic site can catalyse both glycoside and ester hydrolysis (Figure 8A and B). The catalytic nucleophile for the glycoside hydrolysis was previously identified as Glu279, ^[8] which can also act as a nucleophile for the ester bond hydrolysis of the feruloylated arabinoxylans. As stated above, esterases typically share a conserved catalytic triad: Ser, His, and Asp/Glu, where serine is the nucleophile. Based on this fundamental feature, Glu279 was mutated to Ser (CtXyn5A‐E279S), constituting the catalytic triad‐Ser279, His253, and Glu171‐comparable to that of the typical esterases. Indeed, this triad shares similar distances between interacting atoms when compared with feruloyl esterases from fungi (Aspergillus niger) or bacteria (A. thermocellus; Figure 8C and D), which explains the esterase activity of CtXyn5A‐E279S. On the other hand, serine is not a suitable nucleophile for catalysing glycosidic bond hydrolysis, consistent with the functional analysis performed here (Figure 7), and its previous use as an inactive variant for crystallographic studies. ^[10] The oxyanion stabilisation in typical esterases, including lipases, is through two hydrogen bonds from the amide groups of peptide bonds or via hydrogen (γ‐type) of tyrosine. ^[23] In CtXyn5A or the variant CtXyn5A‐E279S, Tyr255 and Glu171 are in suitable positions to contribute to the oxyanion stabilisation (Figure 8F). The distances from the hydroxy group of Tyr255 and the carboxylic group of Glu171 to the ester carbonyl oxygen are 2.7 and 3.1 Å, respectively, based on the enzyme ligand docking (Figure 8F).

Figure 8.

Structure analysis of CtXyn5A‐E279S in comparison with feruloyl esterases. A) Feruloyl domain of the cellulosomal xylanase from A. thermocellus (PDB ID: 1JT2) in complex with ferulic acid. B) CtXyn5A‐E279S in complex (obtained by docking) with feruloyl‐5‐arabino‐α‐1,3‐xyloside. C) Overlapping positions of the catalytic triads of CtXyn5A‐E279S (brown) and the feruloyl esterase from A. niger (PDB ID: 1UWC, blue). D) Overlapping positions of the catalytic triads of CtXyn5A‐E279S (brown) and the triad of the feruloyl domain of A. thermocellus (PDB ID: 1JT2, pink). E) Residues in the feruloyl domain of A. thermocellus (PDB ID: 1JT2) surrounding ferulic acid (yellow). F) Residues in the catalytic site of CtXyn5A‐E279S surrounding the ligand feruloyl‐5‐arabino‐α‐1,3‐xyloside (yellow). G) Hydrophobicity surface representation (Kyte‐Dolittle scale ^[30] ) of the active site of the feruloyl domain of A. thermocellus (PDB ID: 1JT2, ferulic acid in pink). H) CtXyn5A‐E279S in complex with ferulic acid (blue) and with feruloyl‐5‐arabino‐α‐1,3‐xyloside (brown).

To the best of our knowledge, there are available crystal structures of feruloyl esterases in complex only with ferulic acid (Figure 8E and G), but not with feruloyl‐sugars. Different from other feruloyl esterases, CtXyn5A or CtXyn5A‐E279 have activity only with a complex feruloylated sugar‐substrate. Therefore, the sugar moiety seems necessary for the enzyme‐substrate interaction. The docking with feruloyl‐5‐arabino‐α‐1,3‐xyloside, shows that the aromatic ring of the feruloyl group fits well in the subsite −2*, and is stabilised by hydrogen bonds (Glu68 and NH backbone of Gly136) as well as by several hydrophobic interactions (Figure 8F). The ethylene moiety is in subsite −1, with the ester bond close to the catalytic triad, while the sugar moieties bind to the aglycone subsites +1 and +2 (Figure 8F and H). Previous work has suggested that aglycone subsites are less specific than glycone subsites,[ ¹¹ , ¹³ ] which is consistent with the possibility of binding arabinose in the +1 subsite. The xylose unit in the +2 subsite has the hydroxy groups 4 and 1 exposed to the solvent, opening the possibility to extend the xylose backbone in both directions.

Based on the experimental and computational analysis presented here, we propose that in the presence of a polymeric feruloylated polysaccharide substrate, CtXyn5A first removes the feruloyl group and then hydrolyses the xylan backbone. For the esterase activity, residues in the −2* subsite interact with the feruloyl group, residues in the −1 subsite interact with the arabinofuranoside, and residues in the aglycone subsite interact with the xyloside units of xylan while, for the glycoside hydrolase activity, the subsite −2* residues bind to the arabinofuranoside, and the glycone and aglycone subsite residues create interactions with the xylan backbone as previously described.[ ¹¹ , ¹³ ] Smaller ligands, such as para‐nitrophenyl ferulate, are unable to make such interactions with the aglycone subsites, as reflected by a significantly lower binding energy (data not shown) compared with that of the complex polysaccharide substrates. Therefore, the sugar moiety is concluded necessary for a suitable interaction with the enzyme.

Conclusions

The CtXyn5A enzyme from A. thermocellus, previously classified as an arabinoxylanase, was found to express a dual function combining arabinoxylanase with feruloyl esterase activity on complex carbohydrate substrates. As its feruloyl esterase activity is higher on carbohydrate polymers, which are expected substrates for the enzyme, compared to a designated ferulic acid esterase, it is concluded that the enzyme exhibits a dual function rather than an esterase side activity. Interestingly, both types of reactions appear to be catalysed in the same active site. Furthermore, by introducing the E279S mutation, a highly promising new ferulic acid esterase was produced that is mainly active on carbohydrate polymers, but has an increased activity on complex substrates compared to the commercially available E‐FAERU.

Experimental Section

Substrates, enzymes and chemicals: Two different complex carbohydrate substrates as well as three different model substrates were used in this study. The complex carbohydrates were enriched in highly feruloylated arabinoxylans originating from the agricultural side streams corn bran and oat hulls as described in Schmitz et al.. ^[17] Details on their origin and pre‐treatment are described in Table 2. In brief, both the corn bran and the oat hulls were diluted to a 1 : 5 ratio (w/v) in Milli‐Q water and sonicated in a Labassco Soronex RK100H ultrasonic bath (Bandelin) for 10 min at 35 kHz. The mixture was centrifuged at 3893 g for 10 min and the solid phase was incubated with 5 M sodium hydroxide at 63 °C for 9 h with shaking. The same centrifugation step was repeated, and the liquid fraction was collected and neutralised to a pH between 5 to 6 with 37 % hydrochloric acid. The containing arabinoxylan was precipitated with the addition of four volumes 99.5 % ethanol and incubated overnight at 4 °C. The ethanol was removed with a final centrifugation at 3893 g for 5 min followed by evaporation, and afterwards the material was freeze dried in a LyphLock 12 lyophilizer (Labconco). The oat hull was first resuspended in Milli‐Q water to gain higher solubilisation in later treatments.

Table 2.

Description of the origin and prior chemical and enzymatic pre‐treatment of the complex carbohydrate substrates used in this study.

	Corn bran	Oat hull
variety seed origin	unknown	Kerstin and Galant SW‐Seed, Sweden
growth location	USA	Sweden
harvest year	2019	2019
supplier	Bunge Ltd	Lantmännen ek. för.
chemical and enzymatic pre‐treatment	destarching and ultrasound assisted alkali according to refs. [13, 29]	ultrasound assisted alkali according to ref. [29]
arabinoxylan content [%]	28.8±1.8	20.9±3.3
ferulic acid content [mg g−1]	227±69	90±3

The model substrates p‐nitrophenyl (pNP) butyrate and pNP‐laurate were purchased from Sigma‐Aldrich; pNP‐ferulate was obtained from Carbosynth. All had a purity of≥98 %. The commercially available ferulic acid esterase (FAE) E‐FAERU from Megazyme (Bray, Ireland, 30 U/mg at pH 6.5 and 40 °C) and the truncated arabinoxylanase of glycoside hydrolase family 5 subfamily 34, termed CtXyn5A (GH5 CBM6) from NZYTech (Lisbon, Portugal, 1 mg/mL, ≥90 %) were used in the ester hydrolysis reactions. The imidazole and buffer salts were purchased from Sigma‐Aldrich, while the organic solvents were obtained from VWR.

Production and purification of CtXyn5A‐E279S, CtXyn5A‐E279A and CtXyn5‐E279L: The truncated genes encoding the two‐domain construct (GH5 CBM6) corresponding to CtXyn5A from A. thermocellus, with the catalytic nucleophile Glu279 replaced by Ser (E279S), Ala (E279A) and Leu (E279L) were chemically synthesised (GenScript USA Inc., Piscataway, NJ, USA) with native codons and cloned into the expression vectors pET21b (+) with a His‐tag introduced in the C terminus. The resulting plasmids (pET28b:CtXyn5A‐E279S) were transformed into Escherichia coli BL21(DE3) competent cells (Sigma) by heat‐shock transformation. A 1 % pre‐inoculum of the respective recombinant strain (pre‐cultivated in LB medium over night at 37 °C with 100 μg/mL of ampicillin as a selection factor) was added to 300 mL LB medium. The cultures were incubated with shaking at 37 °C until an optical density at 600 nm (OD_600 nm) of 0.6–0.8 was obtained. Then they were induced with 1 mM isopropyl β‐d‐1‐thiogalactopyranoside (IPTG) and the incubation was continued at 20 °C overnight. The cells were harvested, lysed by sonication and centrifuged at 26,000 g to separate soluble proteins from the cell debris. CtXyn5A‐E279S, CtXyn5A‐E279A and CtXyn5‐E279L were purified by immobilised metal ion affinity chromatography using an ÄKTA start protein purification system (GE Healthcare Bio‐Sciences) with a HisTrap™ High Performance column (1 mL, Cytiva). The system was equilibrated, and unbound proteins were washed out using a binding buffer consisting of 100 mM Tris‐HCl, 500 mM NaCl, pH 7.4. The His tagged CtXyn5A‐E279S was eluted isocratically with a buffer containing 100 mM Tris ⋅ HCl, 500 mM NaCl, 365 mM imidazole, pH 7.4. CtXyn5A‐E279A and CtXyn5‐E279L were eluted with 20 column volume gradients. The protein concentration was determined spectrophotometrically by measuring the absorbance at 280 nm and the purity was estimated with sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS‐PAGE; Figure 9). All proteins were maintained in elution buffer after purification, and the concentration of imidazole was the same as in the commercial wild‐type in order for all protein to be in a comparable buffering system in their soluble form.

Figure 9.

SDS‐PAGE gel showing commercial CtXyn5A wild‐type (WT), and mutated variants: CtXyn5A‐E279S, CtXyn5A‐E279A and CtXyn5A‐E279L after purification in elution buffer with 100 mM Tris ⋅ HCl, 500 mM NaCl and 365 mM imidazole at pH 7.4.

Enzyme and imidazole reactions on complex carbohydrates: For the reactions carried out on complex carbohydrates, 50 mg of each of the substrates described in Table 2 was dissolved in 1 mL of 100 mM sodium phosphate buffer (pH 6). To start the reactions, 355 μg of E‐FAERU or 10 μg of CtXyn5A or 10 μg of the respective point mutated enzyme was added (ca. 10 μL). Due to the expected inactivation of the Ala279 and Leu279 enzyme variants, reactions were set containing a fivefold higher enzyme concentration in the same volume, to preserve comparable amounts of imidazole in all samples. All reactions were performed at 40 °C under constant shaking. After 24 h, they were terminated by boiling at 110 °C for 5 min. To estimate possible non‐enzymatic hydrolysis by heating during the reaction and termination, control reactions were prepared containing only substrate and buffer. The pH was maintained at 6 throughout this process, both for reactions with oat hull and corn bran xylans, including all enzymatic reactions and controls. To avoid aggregation and precipitation of CtXyn5A as well as the mutated variants their storage solutions contain up to 2.5 % imidazole (corresponding to 365 mM), which is known to chemically cleave ester bonds. ^[20] In order to assess its influence on ester bond hydrolysis, control experiments were carried out in which 2.5 % imidazole in 100 mM sodium phosphate buffer (pH 7.2) was added in the same volume as the enzyme in the reactions.

Free ferulic acid analysis: The amount of free ferulic acid present in the enzymatic and imidazole reaction mixtures after reaction on the complex carbohydrates was quantified using the HPLC method developed by Sajib and colleagues. ^[24] In brief, separation of different free phenolic acids was achieved on a Kinetex 2.6 μm Biphenyl 100 Å column (50 mm×2.1 mm; Phenomenex) under isocratic elution at a flow rate of 0.3 mL/min for 12 min with an Ultimate 3000 Dionex HPLC system. The utilised mobile phase consisted of two mixtures of methanol, acetic acid and Milli‐Q water differing in concentration: 87 % of a 10 : 2 : 88 mixture and 13 % of a 90 : 2 : 8 mixture.

Esterase activity assay: To further assess the esterase activity and specificity of E‐FAERU, CtXyn5A, CtXyn5A‐E279S and imidazole, colorimetric activity assays on pNP‐butyrate, pNP‐laurate and pNP‐ferulate (Figure 10) were carried out following the general method first described by Huggins and Lapides. ^[25] In brief, the substrates were dissolved in acetonitrile/propan‐2‐ol (1 : 1) to yield a concentration of 10 mM. In 96‐well microtiter plates, 20 μL of these solutions were mixed with 75 μL of Milli‐Q water and 100 μL of sodium phosphate buffer (100 mM, pH 7.2). The reactions were started by adding 5 μL of the enzyme or imidazole solutions. The CtXyn5A, CtXyn5A‐E279S and imidazole solutions were used as previously prepared, but the commercial FAE formulation was first diluted 12.5 times to reach a concentration similar to that of CtXyn5A. The cleavage of pNP from the substrate was monitored through absorbance measurements at 410 nm in a Multiskan Go microplate spectrophotometer (Type 1510, Thermo Scientific) at 37 °C for up to 60 min.

Figure 10.

Structural representation of synthetic 4‐nitrophenyl substrates used for the esterase activity assay.

Oligosaccharide analysis: The oligosaccharides present in the pre‐treated corn bran and oat hull substrates before and after enzymatic treatment with CtXyn5A and CtXyn5A‐E279S were analysed by HPAEC‐PAD (ICS‐5000, Dionex, Thermo Scientific). A CarboPac PA200 analytical column (250 mm×3 mm, 5.5 μm) equipped with a respective guard column (50 mm×3 mm) was used for separation of the different products. The mobile phase consisted of 100 mM sodium hydroxide and a gradient of sodium acetate of 0–100 mM during the first 10 min after which it was kept constant at 100 mM until the end of the run at 23 min.

Molecular modelling: The molecular structures, available in the Protein Data Bank, CtXyn5A (PDB ID: 2Y8K), CtXyn5A‐E279S (PDB ID: 5LA2), and ferulic acid esterase (PDB ID: 5YAE) from Streptomyces cinnamoneus, were used for comparison or docking. All visual analyses and pictures were performed with UCSF Chimera v 1.15. ^[23] Ferulic acid and feruloyl‐5‐arabino‐α‐1,3‐xyloside, used as ligands, were modelled in YASARA v 21.12.19. ^[26] The ligands were energetically minimised with AMBER14 ^[27] force field and subjected to docking with CtXyn5A, and local docking into the active site of CtXyn5A‐E279S by using Autodock ^[28] incorporated in YASARA.

Author contributions

Conceptualisation: ES, SL, JALP, ENK; Formal Analysis: ES, SL, JALP; Funding Acquisition: ENK, PA; Investigation: ES, SL, JALP; Methodology: ES, SL; JALP; Project administration: ENK, PA; Resources: ENK, PA; Supervision: ENK, PA; Validation: ES, SL, JALP; Visualisation: ES, SL, JALP; Writing – original draft: ES; SL; JALP; Writing – review and editing: All co‐authors.

Conflict of interest

The authors declare no conflict of interest.

Schmitz E., Leontakianakou S., Adlercreutz P., Nordberg Karlsson E., Linares-Pastén J. A., ChemBioChem 2023, 24, e202200667.

Data Availability Statement

Data sharing is not applicable to this article as no new data were created or analyzed in this study.