C. B. Faulds, Institute of Food Research, Norwich Research Park, Colney, Norwich NR4 7UA, UK Fax: +44 160 350 7723 Tel: +44 160 325 5152 E-mail: email@example.com
Feruloyl esterases hydrolyse phenolic groups involved in the cross-linking of arabinoxylan to other polymeric structures. This is important for opening the cell wall structure making material more accessible to glycoside hydrolases. Here we describe the crystal structure of inactive S133A mutant of type-A feruloyl esterase from Aspergillus niger (AnFaeA) in complex with a feruloylated trisaccharide substrate. Only the ferulic acid moiety of the substrate is visible in the electron density map, showing interactions through its OH and OCH3 groups with the hydroxyl groups of Tyr80. The importance of aromatic and polar residues in the activity of AnFaeA was also evaluated using site-directed mutagenesis. Four mutant proteins were heterologously expressed in Pichia pastoris, and their kinetic properties determined against methyl esters of ferulic, sinapic, caffeic and p-coumaric acid. The kcat of Y80S, Y80V, W260S and W260V was drastically reduced compared to that of the wild-type enzyme. However, the replacement of Tyr80 and Trp260 with smaller residues broadened the substrate specificity of the enzyme, allowing the hydrolysis of methyl caffeate. The role of Tyr80 and Trp260 in AnFaeA are discussed in light of the three-dimensional structure.
The plant cell wall is a complex mixture of polysaccharides, proteins, phenolics and lipids. The polysaccharides form the skeleton of the plant cell wall, and are composed of cellulose microfibrils embedded within a matrix of hemicellulose or pectin, depending on the plant tissue. Hemicelluloses are the most abundant renewable polymers after to cellulose and they are the key components in the degradation of plant biomass. However, this degradative process is often inefficient because most polymers of cellulose and hemicellulose are either insoluble or simply too closely associated with the insoluble matrix. In cereals, the main hemicellulosic polymer is arabinoxylan, which is composed of a β-(1,4) glycosidic-linked d-xylopyranosyl units, substituted at positions O-2 or O-3 with arabinose. To deconstruct or modify arabinoxylans, plants or microorganisms require a battery of glycoside hydrolases (xylanases, α-arabinofuranosidases, β-xylosidases, glucuronidases) and esterases (feruloyl, acetyl). Feruloyl esterases (EC 18.104.22.168) release ferulic acid (FA) (Fig. 1) from arabinose-substituted xylans and rhamnogalacturonans . While most of the feruloyl esterases to date have been grouped into the carbohydrate esterase family 1  (for more information, see http://www.afmb.cnrs-mrs.fr/CAZY/), a complementary classification based on amino acid sequence similarities and substrate specificity has putatively grouped feruloyl esterases into four types, A–D . AnFaeA is a type-A feruloyl esterase isolated from Aspergillus niger.
From protein sequence homology, feruloyl esterases belong to the same family as the serine proteases, esterases and lipases, with a serine residue acting as the nucleophile in a catalytic triad comprising the hydroxyl group of the active serine, the imidazole side chain of histidine and a buried carboxylic acid chain . Although the mechanism of deferuloylation has not been reported, it is probable, based on the general hydrolytic mechanism of esterases , that the basic His248 (AnFaeA numbering) removes a proton from the hydroxyl group of Ser133 and that the nucleophilic oxygen attacks the carbonyl carbon of the feruloyl group to form a tetrahedral intermediate.
Amino acids with aromatic side chains play a prominent role in binding carbohydrates . The hydrophobic patch of a sugar moiety, resulting from the disposition of the equatorial and axial hydroxyls to one side of the pyranose ring of a sugar monomer, aligns itself upon binding with the aromatic ring face of the amino acid to contribute to selectivity of fit of the substrate to the binding site of the enzyme . Tryptophan has been shown to be essential for substrate binding in most of the glycoside hydrolases studied to date, such as cellulases , xylanases [8,10,11] and α-amylases . The protein sequence of AnFaeA, showed the presence of four tryptophan residues in the molecule  and chemical modification of the mature protein with N-bromosuccinimide (NBS) demonstrated that one tryptophan essential for activity was exposed on the surface of the enzyme .
The structure of AnFaeA has recently been solved (PDB accession numbers 1USW, 1UZA, 1UWC) [15,16]. The enzyme displays an α/β hydrolase fold  similar to that found in fungal lipases, such as those from Thermomyces lanuginosa and Rhizomucor miehei. This structure is different from that reported for the feruloyl esterases from Clostridium thermocellum[20,21], although the catalytic triads can be superimposed allowing direct extrapolation of the position of the oxyanion pocket. Crystallography and point replacement of the nucleophilic serine of AnFaeA, Ser133, allowed the identification of the active site, confined by a lid (residues 68–90) and a loop (residues 226–244) which confers plasticity to the substrate binding site . While structurally resembling lipases, AnFaeA does not exhibit lipase activity . From these studies we postulated that Tyr80 could play an essential role in substrate binding and specificity. In addition, Trp260, located at the C terminus and near the surface is the closest tryptophan to the active centre.
In this study, we used site-directed mutagenesis and X-ray crystallography to give insights into the specificity and affinity of AnFaeA for methyl hydroxycinnamic acid substrates.
Results and Discussion
Crystal structure of the S133A AnFaeA–FAXX complex and design of AnFaeA mutants
To determine which residues are important for the interaction of ferulate and ester-linked carbohydrates with AnFaeA, the crystal structure of the inactive S133A nucleophilic mutant of AnFaeA complexed to the feruloylated trisaccharide O-[5-O-[(E)-feruloyl]-α-l-arabinofuranosyl}-(1(ρ)3)-O-β-d-xylopyranosyl-(1→4)-d-xylopyranose (FAXX) was solved at 2.5-Å resolution. AnFaeA requires both the hydroxycinnamate as well as a carbohydrate grouping as part of the substrate for optimal activity, and a feruloylated trisaccharide consisting of the linkage of sugars found in FAXX has been shown to be the optimal size of substrate . Three molecules of S133A AnFaeA mutant are present in the asymmetric unit and show no significant differences between the native and mutant structures, as revealed by the low r.m.s.d. deviation of their backbones (0.47 Å) after superimposition of both structures. The electron density maps revealed the presence of a FA moiety bounded at the active site in the three molecules of the asymmetric unit (not shown). However, the remaining groups of the substrate (i.e. the arabinose and the two xylose units) were not visible in the 2Fo–Fc map, neither difference map indicated the presence of carbohydrate groups. As the S133A mutant is inactive , the substrate should be complete and therefore the carbohydrate moiety is probably disordered. The substrate is placed in a long and narrow cavity (Fig. 2A). The active site cavity is mainly confined by the flap (residues 68–80) and the 226–244 loop (Fig. 2B). As shown in Fig. 3, the arrangement of the FA in the active site is essentially the same to that observed in the high-resolution structure of the AnFaeA–FA complex determined by McAuley et al. (PDB code 1UWC) . The FA interacts (Fig. 2B,C) through the OH group at C4 with the hydroxyl group of Tyr80 in the enzyme substrate complex. Despite the apparently long distance found in the present structure (3.8 Å) the hydroxyl group of Tyr80 probably interacts with the OCH3 group at C-3 as it also occurs in the high resolution structure of 1UWC . Tyr80 is one of the residues that takes part in the formation of the substrate cavity and its arrangement delimitates the long substrate cavity where the aromatic ring of the FA is placed (Fig. 4A). Moreover, the global arrangement of residues in the substrate cavity provides a molecular surface in which OCH3 group fits perfectly (Fig. 4A). The carboxylate moiety is located at the oxyanion hole defined by the Leu134-N main-chain and both the backbone N atom and the OH group from Thr68. Leu199, Val243 and Ile196 provide the hydrophobic environment to stabilize the aromatic and the hydrocarbon chain of the FA. The importance of the interaction between an aromatic tyrosine and the phenolic ring of the substrate is consistent with the biochemical specificity of this enzyme [4,24]. The role of Tyr100 was previously probed by site-directed mutagenesis; mutating the Phe100 (AnFaeA numbering) of the lipase from Thermomyces lanuginose to Tyr was essential to confer ferulate ester-hydrolysing activity . Of the four tryptophan residues in the sequence of AnFaeA, only one is located near the surface, as demonstrated by chemical modification, and is essential for activity . Trp260 is the terminal residue, located on a flexible loop  and although far from the active site, this residue is the closest Trp in the vicinity of the active site, and is thus a probable candidate for substrate specificity (Fig. 4B). This residue is buried in a hydrophobic cavity surrounded by Met253, Thr19 and Ala23 side chains. In the present work, site-directed mutagenesis is used to probe the role of polarity and/or hydrophobicity in the environment of Tyr80 and Trp260.
The active site of AnFaeA is placed in a long and narrow cavity that connects two crevices at the molecular surface , displaying hydrophobic residues that stabilizes the aromatic moiety of the substrate. As with the structure of the C. thermocellum feruloyl esterase in complex with FAXX, FAE_XynZ-FAXX (1JT2) , the FA moiety is clearly visible in the active site but the carbohydrate parts of the substrate are not visible, suggesting that tight binding of the carbohydrate is not required for catalysis.
Production and characterization of AnFaeA variants
All four mutants (Y80S, Y80V, W260S, W260V) were efficiently produced in Pichia pastoris as confirmed by SDS/PAGE and western blot analysis with anti-FaeA polyclonal antibodies. Purified recombinant variants were obtained in yields ranging from 163 mg·L−1 (Y80V) to 628 mg·L−1 (W260V) using a single chromatographic step (hydrophobic interaction: HIC). While wild-type AnFaeA was purified using the phenyl sepharose HIC column , the mutants were retained on the column, even by reducing the hydrophobicity of the buffer. Due to this, these four variants of AnFaeA were then purified using a butyl sepharose column.
To evaluate the consequence of altering the hydrophobicity or the bulking effect in the active site of AnFaeA around the Tyr80 mutation, and the effect of altering the only surface exposed tryptophan residue, Trp260, on activity, the four variants were tested on the methyl esters of hydroxycinnamic acids: methyl ferulate (MFA) and compared to wild-type AnFaeA. All of the variants displayed feruloyl esterase activity albeit at a reduced value compared to the wild-type enzyme. The effects of these mutations on the secondary structure of the Y80V, Y80S, W260V, W260S were tested by CD. All mutants show an increase in α-helix content, reflecting possible small local structural rearrangements (Table 1). However, as the kinetic values for Y80S and Y80V and for W260S and W260V were similar, such changes in structure did not duly affect the catalytic arrangement of the enzyme.
Table 1. Secondary structure of wild type and mutant AnFaeA, from circular dichroism and SELCON analysis.
Kinetic analyses and substrate specificity of AnFaeA variants
The kinetic parameters (kcat and Km) of the Y80V, Y80S, W260V, W260S mutants were determined against MFA, methyl sinapate (MSA), methyl caffeate (MCA) and methyl p-coumarate (MpCA) (Fig. 1). All variants showed a significant decrease in the hydrolytic rate compared to the wild-type enzyme in addition to a slight increase in Km for all the substrates, except for MCA (Table 2). The decrease in the hydrolytic rate was between 1.5- and 4-fold, with the largest changes occurring with MFA and MSA as substrates. Both Tyr80 and Trp260 variants were able to hydrolyse MCA. The type of substitution on the phenolic ring of the substrate is important for defining the type of feruloyl esterase [3,24]. Previous inhibition studies showed that AnFaeA binds MCA but does not hydrolyse it, suggesting that the enzyme possesses a fairly nonspecific binding site . In the present study, replacement of Tyr80 or Trp260 by a nonaromatic amino acid resulted in the reduction in the activity and broadened the specificity of AnFaeA for phenolic acids, in particular for MCA.
Table 2. Kinetic parameters of the wild-type and mutated AnFaeA determined against the methyl esters (1 mm) of ferulate (MFA), sinapate (MSA), caffeate (MCA) and p-coumarate (MpCA). nd, Activity not detected.
70.74 (± 1.44)
1.56 (± 0.04)
2.56 (± 0.08)
20.06 (± 0.67)
18.33 (± 0.44)
0.78 (± 0.05)
1.22 (± 0.07)
1.17 (± 0.08)
0.88 (± 0.07)
1.01 (± 0.06)
84.95 (± 2.26
3.48 ± 0.11)
7.85 (± 0.28)
27.76 (± 0.62)
28.28 (± 0.49)
0.24 (± 0.02)
0.84 (± 0.07)
1.23 (± 0.09)
0.35 (± 0.02)
0.76 (± 0.03)
0.01 (± 0.002)
0.02 (± 0.002)
0.32 (± 0.02)
0.26 (± 0.02)
3.84 (± 0.82)
3.02 (± 0.52)
4.10 (± 0.34)
4.72 (± 0.45)
0.73 (± 0.05)
0.10 (± 0.003)
0.26 (± 0.01)
0.49 (± 0.02)
0.29 (± 0.02)
4.26 (± 0.45)
2.07 (± 0.11)
2.88 (± 0.14)
3.26 (± 0.31)
3.99 (± 0.33)
From the close up view of the phenolic binding pocket (Fig. 2) it is clear that two tyrosine residues, Tyr80 and Tyr100, are closely located near the substituent groups around the phenolic ring, in agreement with the results from the mutagenesis study. It is possible that the removal of the bulky tyrosine from the pocket in the mutant variants must result in a local realignment allowing the accommodation of a hydroxyl group at O-3 of FA instead of the methoxyl group. In comparison, the structures of the two feruloyl esterases from C. thermocellum, FAE_XynY (PDB accession code 1GKK)  and FAE_XynZ (1JJF)  show that ferulate binds in a small blunt-ended surface depression, with the hydroxyl group interacting with an Asp residue and the methoxyl group with a Trp, instead of Tyr as in AnFaeA. The tryptophan did not form a direct stacking interaction with the phenolic ring of FA, instead contributing to the hydrophobic environment by forming a small cavity with a leucine residue on one side of the binding depression .
The structural implication of Trp260 in binding of the substrate is less clear. Although relatively far from the active site (≈ 14 Å) (Fig. 2a), biochemical evidence demonstrated that Trp260 interacted with the active site pocket, as modification of AnFaeA with 4500-fold excess of N-bromosuccinimide (a chemical oxidizer of Trp residues) resulted in an 80% loss of activity against MSA . This is not due to this residue having a role in enzyme stability, as joining a bacterial dockerin domain to the C-terminal end of AnFaeA through Trp260 did not significantly affect the activity of the feruloyl esterase . One hypothesis is that Trp260 may be in a position to interact with the carbohydrate moiety of a feruloylated polysaccharide. In FAE_XynZ, the C-terminal tryptophan, Trp265, is located in a hydrophobic pocket of primarily aromatic residues adjacent to the binding pocket  whereas it is absent in FAE_XynY . However, the interactions between Trp260 and the sugar moieties of the substrate could not be directly demonstrated due to both the lack of resolved sugar interactions in the AnFaeA–FAXX complex, and the nature of the methyl hydroxycinnamates used as substrates. Alternatively, from the 3D structure and the measured effects on the kinetic parameters, it is possible to hypothesize that this tryptophan affects the mobility of the catalytic histidine . Such a shift was reported in the side chain position of His260 and His247 in the FAE_XynZ–FAXX  and the AnFaeA–FA complexes , respectively. While in the free enzyme His247 is present in a single conformation corresponding to the active orientation for a catalytic histidine residue, in the case of the enzyme–product complex His247 can move providing an inactive form. As it was described in the high resolution AnFaeA–FA complex , in the complex structure His247 can present two histidine conformations which easily interconvert from an active to an inactive form. However in our case, the S133A AnFaeA–FA electron density maps did not reveal any difference between the arrangement of His247 in the complexed and free enzyme structures, as His247 is always in the active conformation (Fig. 3). In FAE_XynZ, Trp265 is only 4 Å from the catalytic histidine (His247) , allowing direct interaction, which is not the case in AnFaeA. In other carbohydrate-active esterases, the exposed catalytic His187 residue of the acetylxylan esterase, AXE-II, of P. purporogenum forms a hydrogen bond with a sulphate ion forcing the histidine to adopt an altered conformation . This has also been observed with a cutinase from Fusarium solani. With AXE-II, transition of histidine from a resting state to an active state necessitated the rearrangement of other residues of the active site, most notably the movement of Tyr177 which moved 2 Å away to accommodate the catalytic histidine in the active state. While no change in the position of His247 was determined when the free and complexed structures were compared, Trp260 still can influence both catalytic rate and specificity. The role of Trp260 in the catalytic mechanism of AnFaeA requires further examination. The above differences in reported structures and activities of feruloyl esterases are reflected in the cladogram for carbohydrate-active esterases with known 3D structures (Fig. 5). While AnFaeA closely resembles the lipases of Rhizomucor meihei and Thermomyces lanuginosa, of the feruloyl esterases, FAE_XynZ from C. thermocellum shows the closest homology. AnFaeA releases 5,5′ diFA from cereal-derived material  and the 3D structure shows how the dimer can be accommodated within the active site . The structure of FAE_XynZ suggested that the open and solvent-exposed FA binding site can interact with diFA , while FAE_XynY could not accommodate such a substrate. This is in agreement with the closeness demonstrated in the phylogenic analysis (Fig. 3). FAE_XynY, on the other hand, is further removed from AnFaeA and may resemble more the acetylxylan esterases of Penicillium purpogenum (1BS9)  and Trichoderma reesei (1QOZ) . Further biochemical characterization of these enzymes is required to test these hypotheses.
In vitro site-directed mutagenesis of the faeA gene on plasmid pFAE-W was performed by using the QuickChange™ Site-Directed Mutagenesis Kit from Stratagene (La Jolla, CA, USA) following the manufacturer's instructions with two exceptions: DH5αEscherichia coli cells where used instead of Epicurian Coli XL1-Blue, and the elongation step in each thermal cycle was extended from the recommended 18 s (2 s per kb) to 25 s. Alanine, serine and valine replacement codons were chosen taking into account codon usage in yeast. Two complementary oligonucleotides were used for replacement of S133A, W260V or W260S, however, following the observations of Makarova et al. , a single primer was successfully used for Y80V or Y80S replacements. The plasmids carrying the resulting mutant faeA alleles were called pFAE-S133A, pFAE-W260V, pFAE-W260S, pFAE-Y80V and pFAE-Y80S, respectively. Table 3shows the sequence of all the oligonucleotides used in this work. In all cases, the plasmid region containing the faeA gene, as well as the AOX1 promoter and terminator, was sequenced completely to rule out the presence of any additional mutation. Spheroplasts from Pichia pastoris strain GS115 were transformed with these plasmids, by using the Pichia expression kit from Invitrogen (Carlsbad, CA, USA) and His+MutS strains were selected for the expression of the mutated versions of AnFaeA.
Table 3. Sequences of the oligonucleotides used for site directed mutagenesis.
Purification of the AnFaeA mutants
AnFaeA mutants were purified from the P. pastoris cultures by HIC based on previously described protocols . Apart from the wild-type and S133A mutant, the other mutants were retained on a phenyl-sepharose column (Amersham Biotech, Little Chalfont, Bucks, UK), even after a water elution. A butyl-sepharose column (Amersham Biotech) was used to purify these mutants.
Crystallization, data collection and processing
Co-crystallization of S133A AnFaeA–FAXX complex was performed by the hanging-drop, vapour-diffusion method at 291 K, testing the conditions obtained for the native protein (1.8 m ammonium sulphate, 0.1 m Hepes, pH 7.5). FAXX was purified from Driselase-hydrolysed de-starched wheat bran, as described previously . After preliminary trials, crystals suitable for X-ray studies were obtained by mixing 4 µL of a well solution (1.7 m ammonium sulphate), 1 µL of FAXX substrate (10 mm) and 2 µL of the mutant enzyme solution at 12 mg·mL−1. The crystals were tested on an in-house MAR Research IP area detector with CuKα X-rays (λ = 1.5418 Å) generated by an Enraf-Nonius rotating anode generator, but diffraction data were of low resolution. Consequently, synchrotron radiation was used. Data sets were collected at ESRF (ID14-4 beamline), with λ = 0.9184 Å. All data were processed and scaled using mosflm and scala from CCP4 package software . Data processing statistics are given in Table 4. The crystals belong to space group P21, with unit cell dimensions a = 46.74 Å, b = 130.75 Å, c = 76.51 Å and β = 98.14°Å. Specific volume calculations yielded three molecules of S133A AnFaeA in the asymmetric unit, with a solvent content of 55.3% (v/v) (VM = 2.75 Å3·Da−1).
Table 4. Data collection and refinement statistics for the S133A AnFAEA mutant in complex with FAXX. Values in parentheses correspond to the highest resolution shell. Rfactor = ∑h||Fobs| – |Fcalc||/∑|Fobs|, εEh||Fobs| – |Fcalc||/Ó|Fobs|, where Fobs and Fcalc are observed and calculated structure factor amplitudes, respectively. Rfree calculated for 7% of data excluded from the refinement.
Unit cell parameters
Resolution limit (Å)
Total no. of reflections
Unique reflections (n)
Resolution range (Å)
Water molecules (n)
Ferulic acid molecules (n)
N-acetyl glucosamine molecules (n)
r.m.s.d. from ideal
Bond lengths (Å)
Bond angles (°)
The structure of the S133A AnFaeA–FAXX complex was determined by the molecular replacement method using the program amore[38,39]. The atomic coordinates of the AnFaeA (PDB code 1USW) were used as the search model for a rotational and translational search in the 49–3.5 Å resolution range. We obtained a good solution for three molecules in the asymmetric unit and the values of the final correlation coefficient and Rfactor were 0.70 and 21.6%, respectively. The structure was refined with cns up to 2.5 Å resolution using strict ncsrefinement, and restrained ncs refinement in the last stages. Refinement statistics are given in Table 4.
Model quality and accuracy
The final model consists of three molecules of S133A AnFaeA (A, B, C), three FA molecules (one per S133A AnFaeA molecule) and 375 water molecules. As the native structure, the S133A AnFaeA is glycosylated at Asp79 and two molecules of N-acetyl glucosamine residue were built at each glycosylation site. In the complex, the electron density maps in this region reveal a carbohydrate structure larger than only two units of N-acetyl glucosamine but this could not be modelled because of the poor electron density definition. The stereochemical quality of the model was checked with the program procheck. The figures were generated with molscript, raster 3d and grasp. The atomic coordinates and structure factors for S133A AnFaeA–FA complex have been deposited in the Protein Data Bank, with accession number 2BJH.
Gel electrophoresis and immunoblotting
SDS/PAGE was carried out on a 10% Bis/Tris precast NuPAGE gel (Invitrogen) with wild-type AnFaeA as a marker. Proteins were transferred to nitrocellulose membranes by semidry blotting (Bio-Rad, Hercules, CA, USA). The blotted membranes were probed with a 1000-fold dilution of polyclonal antiserum raised in rabbits against AnFaeA . Immunoreactive proteins were visualized using alkaline phosphatase-conjugated anti-rabbit secondary antibody (Sigma, St Louis, MO, USA; 1 : 2000).
Circular dichroism spectra were collected using a JASCO 710 spectropolarimeter (Great Dunmow, Cambs, UK). Far UV CD spectra were recorded at 0.5 mg·mL−1 with a 0.2-mm path length cell. The spectra shown are an average of four accumulations, with a scan speed of 100 nm·min−1, band width 1 nm, response 1 s, data pitch 0.2 nm and range 260–190 nm. Analysis of the spectra was estimated using selcon.
Feruloyl esterase activity, assayed with hydroxycinnamate methyl esters, was determined by HPLC for all the AnFaeA mutants as described previously . All measurements were carried out in 100 mm Mops pH 6.0 at 37 °C. In all measurements, the free acid present in samples pretreated with glacial acetic acid was subtracted from that in the test assays. One unit of esterase activity was defined as the amount of enzyme required to release 1 µmol hydroxycinnamic acid·min−1·mg protein−1 at 37 °C, pH 6.0. The kinetic results obtained from the hydrolysis of a range of 0.2–2 mm methyl hydroxycinnamates was interpreted using the Michaelis–Menten kinetic model, using grafit. For each variant and each substrate, at least 10 substrate concentrations were measured in duplicate.
Multiples alignment of sequences encoding feruloyl esterases and related enzymes such as lipases, and construction of neighbor-joining cladogram , were performed with clustal w (http://www.ebi.ac.uk/clustalw/) .
This work was funded by the Biotechnology and Biological Research Council (BBSRC), UK through an ISIS travel grant to CBF and by the BBSRC and the Department of Trade and Industry (DTI), UK, through the award of an Applied Biocatalysts Link award (grant number ABC11741).