We have recently isolated and heterologously expressed BcPeh28A, an endopolygalacturonase from the phytopathogenic Gram-negative bacterium Burkholderia cepacia. Endopolygalacturonases belong to glycoside hydrolase family 28 and are responsible for the hydrolysis of the non-esterified regions of pectins. The mode of action of BcPeh28A on different substrates has been investigated and its enzymatic mechanism elucidated. The hydrolysis of polygalacturonate indicates that BcPeh28A is a non-processive enzyme that releases oligomers with chain lengths ranging from two to eight. By inspection of product progression curves, a kinetic model has been generated and extensively tested. It has been used to derive the kinetic parameters that describe the time course of the formation of six predominant products. Moreover, an investigation of the enzymatic activity on shorter substrates that differ in their overall length and methylation patterns sheds light on the architecture of the BcPeh28A active site. Specifically the tolerance of individual sites towards methylated saccharide units was rationalized on the basis of the hydrolysis of hexagalacturonides with different methylation patterns.
- Burkholderia cepacia
- capillary electrophoresis (CE)
- glycoside hydrolase
- high-performance anion-exchange chromatography with pulsed amperometric detection (HPAEC-PAD)
- kinetic model
- pectin degradation
Primary cell walls are important structures of plant cells that are involved in a variety of physiological processes . These structures are composed mainly of polysaccharides such as cellulose, hemicellulose and pectin. Cellulose microfibrils, composed of 30–36 chains of β-1,4-linked glucose, together with hemicelluloses, form a cohesive network that is embedded in a soluble matrix of polysaccharides, glycoproteins and low-molecular-mass compounds. Pectic substances are the most abundant components within this matrix. On the basis of backbone structures, pectic substances can be divided into three classes: HGA (homogalacturonan), RG-I (rhamnogalacturonan I) and RG-II (rhamnogalacturonan II). HGA is a linear homopolymer composed of 1,4-linked GalA (α-D-galacturonic acid) residues, where several GalA residues can be methyl-esterified or acetylated depending on the plant source. RG-I has a backbone of alternating rhamnose and GalA residues, where several rhamnose residues can be substituted at C-4 with neutral and acidic oligosaccharide side chains. RG-II is composed of at least seven 1,4-linked α-D-GalA residues, to which branched side chains are attached .
The modification of such a complex substrate is required in the course of plant physiological processes, such as cell wall turnover and remodelling , but it also occurs as a consequence of a pathogen attack [4,5]. Cell-wall-degrading enzymes isolated from pathogenic organisms include exopolygalacturonases and endoPGs (endopolygalacturonases), pectin and pectate lyases, rhamnogalacturonases and rhamnogalacturonan lyases. Moreover, other accessory enzymes are needed to degrade the branched side groups, such as acetyl and methyl esterases . Interestingly, many pathogenic organisms produce multiple isoenzymes, suggesting that each isoform harbours a specific catalytic, and thus physiological, function .
Among the pectin-degrading enzymes, a large number of endoPGs have been isolated from pathogenic fungi such as Aspergillus niger  and Botris cinerea , and their specific hydrolytic activities have been investigated. EndoPGs are secretory enzymes responsible for the hydrolysis of the α-1,4 glycosidic bond in HGA. On the basis of their amino acid sequence similarities, these enzymes have been grouped into the glycoside hydrolase family 28, which also includes exopolygalacturonases, rhamnogalacturonases and xylogalacturonan hydrolases. The catalytic action of endoPGs proceeds according to an acid–base mechanism: a conserved aspartate residue (Asp201 in A. niger endoPGII; see Figure 1) acts as proton donor, while a water molecule, structurally conserved in two of the known endoPGs crystal structures [10,11], is the nucleophile. The activation of this water molecule is ascribed to two aspartic acid residues (Asp180 and Asp202 in A. niger endoPGII; see Figure 1) that act as general bases. These three aspartate residues are strictly conserved among the endoPGs .
In order to define the mode of action of endoPGs, the most commonly used experimental approach consists of the detection of the products that are released upon hydrolysis of a substrate chain [13–18]. On the basis of the hydrolysis product progression profile, endoPGs have been classified as single-attack or multiple-attack enzymes. Accordingly, in the case of singleattack enzymes, substrate molecules are cleaved only once after the formation of the ES (enzyme–substrate) complex, resulting in the release of oligomers of different chain lengths. The multiple-attack enzymes, also termed ‘processive’ enzymes, instead cleave the same substrate molecule several times, and small oligomers, or even monomers, are released during the initial period of the reaction. The available information on the mode of action of several fungal endoPGs is reported in Table 1. In particular, a comparative study of the modes of action of A. niger endoPGI and endoPGII has been carried out, combining the information taken from the amino acid sequence alignment, and from biochemical and structural studies. These two enzymes, despite their high degree of amino acid sequence identity (71.4%) and similarity (74.2%), exhibit a different mode of action on polygalacturonate. The two structures are very similar, with a relative rmsd (root-mean-square deviation) of 0.8 Å (1 Å=0.1 nm) on 335 corresponding Cα atoms [11,19]. Both enzymes adopt a right-handed β-helix fold that has been observed in all the known structures of endoPGs [11,12,20–22]. The structure of A. niger endoPGII, which includes ten ‘rungs’, each consisting of three β-strands separated by turn regions, is depicted in Figure 1. Site-directed mutagenesis studies on A. niger endoPGI and endoPGII demonstrated that the switch between a processive and a non-processive mode of action can be ascribed to the presence of an arginine or a serine residue respectively in a conserved segment belonging to rung 2 of the β-helix .
Little information is available on the mode of action of bacterial endoPGs, and our current understanding has been gathered from data on enzymes isolated from Agrobacterium vitis, Ralstonia solanacearum and Erwinia carotovora that have been previously classified as endo-acting enzymes. By analysing their product progression profile, the former enzyme has been classified as a dimer-releasing enzyme and the latter ones as trimer-releasing enzymes . The-three dimensional crystal structure of E. carotovora endoPG has been elucidated , but no structure–function relationship studies have been carried out so far.
Other studies aimed at characterizing the active site of endoPGs have been carried out by analysing the products released following the hydrolysis of pure OGAs (oligogalacturonides) that differ in their DP (degree of polymerization) and in their methyl-esterification patterns [13–18,25].
The active site of depolymerizing enzymes includes several saccharide-unit-binding sites, referred to as subsites, that have been labelled, by convention, from −n to +n, where −n represents the non-reducing end and +n the reducing end of the sugar moieties respectively . The cleavage occurs between the subsites −1 and +1. The latter subsites have been most extensively studied, since they include strictly conserved amino acid residues that are involved in catalysis (Figure 1) . The other subsites are mainly involved in substrate recognition and have been less well characterized, since no significant conservation in the amino acid residues is detectable at these positions. Structure-based site-directed -mutagenesis studies of the A. niger endoPGII unveiled a total of seven subsites, from −4 to+3, allowing for several amino acid residues to be assigned to each of them (Figure 1) [23,28].
To the best of our knowledge, no characterization of the structural determinants responsible for the catalytic activity of bacterial endoPGs has been reported.
We have recently heterologously expressed and purified to homogeneity BcPeh28A, an endoPG secreted by the plant-pathogenic Gram-negative bacterium Burkholderia cepacia, strain A.T.C.C. 25416, referred to also as ‘pectic hydrolase enzyme A’ . BcPeh28A does not exhibit a significant amino acid sequence identity with respect to fungal and other bacterial endoPGs: A. niger endoPGI [ID% (percentage identity) 21.5], A. niger endoPGII (ID% 21.3) and E. carotovora (ID% 23.8), R. solanacearum (ID% 22.8) and A. vitis (ID% 22.9) endoPGs.
With the aim of characterizing the active site of BcPeh28A, the evident lack of sequence similarity notwithstanding, we now, for the first time, report on the mode of action of a bacterial endoPG on polygalacturonate and on shorter substrate molecules that differ in their overall length and in their methylation patterns.
Production and purification of BcPeh28A
BcPeh28A was heterologously expressed and purified as reported elsewhere .
Amino acid sequence alignments
The pairwise alignments were performed, with default parameters, by algorithms available within the EMBOSS bio-informatics suite accessible at the following URL: http://www.ebi.ac.uk/emboss/align/.
PGA sodium salt from citrus fruit was purchased from Sigma–Aldrich. According to the manufacturer, the raw material was de-esterified with NaOH and the final level of esters was approx. 5%. The product was purified from high-grade pectin with a molecular-mass distribution of 50–150 kDa.
A sample containing a mixture of OGAs with DPs ranging from 1 to 8 was kindly provided by Dr Giovanni Salvi, Dipartimento di Biologia Vegetale, Università ‘La Sapienza’, Rome, Italy. It was used as the standard in the analysis by CE (capillary electrophoresis) and HPAEC-PAD (high-performance anion-exchange chromatography with pulsed amperometric detection). The mixture was obtained as described in .
A fully unmethylated decagalacturonide, purified from a pectin digest as described in , was kindly given by Professor Knud J. Jensen, Department of Natural Sciences, University of Copenhagen, Copenhagen, Denmark.
BcPeh28A (0.5 μg/ml) was incubated at 20 °C in 2.0 ml of 50 mM sodium acetate buffer, pH 3.5, in the presence of 0.25% PGA as substrate. Aliquots of 50 μl were taken at regular intervals, and the reaction was stopped by adding an equal volume of derivatization solution (described below). The samples were then boiled at 95 °C for 20 min. CE, as described below, was used for identifying and assessing the molar number distribution of the OGAs present in the mixture, following the enzymatic reaction.
The detailed kinetic characterization of BcPeh28A, using PGA as substrate, was done in several steps. At first, direct information on the reaction path was gathered by inspecting the product progression curves. Then a family of independent and likely reaction schemes was constructed, and, for each of them, the corresponding sets of differential equations were derived. Each reaction scheme was then tested by simultaneous fitting to the progress curves. A non-linear regression analysis was performed by a computer program that fitted the parameters of a set of stiff differential equations to multiple experimental curves . In a system of differential equations that describes the time course of six products, more than 12 parameters should be evaluated. In most, if not all cases, it is necessary to make rational assumptions in order to reduce the number of unknowns, due to the limitations in the experimental data. Hence, the bimolecular rate constants for enzyme–substrate binding were assumed to be close to the diffusion rate limit either for the PGA or for the oligomers (k1=k3=k5=k7=5×108 M−1·s−1); the catalytic constants for OGAs with DPs of 4–6 were assumed to be equal (k4=k6=k8); the dissociation constant of the E–PGA (enzyme–PGA) complex, k−1, always converging to very small values, was set equal to zero; the initial enzyme concentration was set to 2.0×10−8 M (see under ‘Time-course experiment’ above). Table 2 reports the imposed and derived kinetic constants in detail. The basic criteria for the goodness-of–fit were obeyed throughout the analysis .
Mode of action on OGAs
Hydrolysis of decagalacturonide
BcPeh28A (0.075 μg/ml) was incubated at 37 °C with decagalacturonide (50 μM) in 50 mM sodium acetate buffer, pH 3.5. Aliquots (100 μl) of the reaction mixture were taken at regular time intervals and the reaction was stopped by the addition of an equal volume of 2.0 mM Tris/HCl and 50 mM NaOH, which resulted in a final pH of 8.3–8.5. In order to avoid the spontaneous breakdown of the obtained OGAs, reported to occur at high pH , the samples were immediately analysed by HPAEC-PAD, as described below. The identification of the hydrolysis products was done in accordance to the retention time of a standard containing a mixture of OGAs with DPs in the range 1–8.
Hydrolysis of hexagalacturonides
BcPeh28A (1.0 μg/ml) was incubated at 37 °C with 25 μM hexagalacturonides (1–4) in 50 mM sodium acetate buffer, pH 3.5, for different incubation times. The enzymatic digest was analysed by ESI-TOF-MS (electrospray ionization MS with a time-of-flight analyser), as described below.
CE experiments were carried out using a Hewlett–Packard HP3D CE system, equipped with a diode-array UV detector. Fused capillaries were from Agilent Technologies (total length: 104 cm; effective length: 95.5 cm; internal diameter 50 μm; extended light path). Before sample injection, a 4 min conditioning of the capillary with the running buffer was necessary, preceded by a 2 min wash with 0.1 M NaOH [pressure equal to 950 mbar; 1 bar=105 Pa]. Signals were acquired by continuously monitoring UV absorbance at 195 and 285 nm. All analyses were done at 25 °C. The separation potential was equal to 27 kV. Samples were injected at 50 mbar for 6 s. The monomer, dimer and trimer of GalA (Sigma–Aldrich), as well as a sample containing a mixture of OGAs with DPs of 1–8, were used as standards. To permit sensitive UV detection, the saccharides were derivatized by reaction with 4-ABN (4-aminobenzonitrile) in the presence of NaCNBH3 (sodium cyanoborohydride) . The derivatization solution was 0.16 M NaCNBH3 and 0.5 M 4-ABN in methanol/acetic acid (95:5, v/v). Only one molecule of the 4-ABN chromophore is attached to each oligomer, leading to a UV response that is independent of the length of the OGAs. This represents an advantage over chromatographic techniques such as HPAEC-PAD, where the electrochemical response must be determined for each oligomer . The derivatization solution was diluted 1:1 with an aqueous solution of the uronic acid(s), and the derivatization mixture was heated for 20 min at 95 °C. Prior to injection into the CE system, samples were spun down for 1 min at 13000 g. The analysis was performed by MEKC (micellar electrokinetic capillary chromatography) as described in . The buffer used was 600 mM boric acid, containing 75 mM SDS (pH 8.0). The peak areas divided by migration time (A/t) were used for quantitative purposes . To obtain a calibration curve, electropherograms for different amounts of GalA were recorded, and their corresponding A/t values were plotted as a function of nmol of GalA. Calibration was linear over the range 1.0–14.0 nmol.
The analysis of oligomers released upon hydrolysis of decagalacturonide by BcPeh28A was carried out with a HPAEC-PAD system (Dionex Corporation) equipped with a GP50 gradient pump, an ED50 electrochemical detector and an LC25 chromatography oven. The separation was performed using a CarboPac PA-200 Analytical anion-exchange column (3 mm×250 mm) with Guard column (3 mm×50 mm). The volume of the injection loop was 25 μl, and the column temperature was 35 °C. The flow-through cell of the pulsed amperometric detector consisted of an gold working electrode (1.0 mm diameter) and a pH–silver/silver chloride combination reference electrode; the titanium body of the cell served as the counter-electrode. The sequence of potentials (E) applied to the electrode was set as recommended by the manufacturer for carbohydrate detection, as follows:
The software Chromeleon, version 6.60 (Dionex Corporation) was used for the acquisition and processing of the chromatograms. The column was equilibrated with 0.1 M NaOH/0.2 M sodium acetate. The samples were eluted by applying a linear gradient of 0.2–0.8 M sodium acetate in 0.1 M NaOH at 0.5 ml/min for 60 min. The identification of the hydrolysis products was done by correlating with the retention times of a standard containing a mixture of OGAs with DPs of 1–8.
ESI-TOF-MS was performed on a microTOF Focus system (Bruker Daltonics) equipped with a nitrogen generator N2LCMS1 (Claind). The samples were diluted 5-fold with methanol/water/formic acid (49:49:2, by vol.) and injected with a syringe pump in the mass spectrometer. The injection flow rate was 180 μl/h; the capillary voltage was 4500 V and the end-plate offset was 500 V (positive mode); the dry temperature was 180 °C, the dry gas flow was 4 l/min and the nebulizer pressure was 0.4 bar.
RESULTS AND DISCUSSION
The mode of action of BcPeh28A on a polymeric substrate was investigated in a time-course experiment that monitored the products released upon PGA hydrolysis by collecting electropherograms. The suitability of MEKC with UV detection for the separation of uronic acids  was tested on a standard sample containing a mixture of OGAs with DPs of 1–8, as shown in Figure 2(A).
The product progression profile (Figure 3A) indicates that, during the first few minutes of the reaction, OGAs with DPs ranging from 3 to 8 were produced. This confirms that BcPeh28A is an endo-acting enzyme. The electropherogram of the reaction mixture shown in Figure 2(B) shows the molar product distribution present in the enzymatic digest. The occurrence of OGAs with different chain lengths clearly demonstrates that BcPeh28A can be classified as single-attack enzyme. The minimum length of the released OGAs varies among pectin-degrading enzymes and provides further information on the active-site architecture of the enzyme. For the PGA hydrolysis catalysed by BcPeh28A, the end products of the reaction are monomers, dimers and trimers, with a clear excess of trimers (Figure 3B). To verify whether BcPeh28A is capable of further degrading the trimers, a commercial sample of tri-GalA was incubated with the enzyme. No cleaved products were observed (results not shown). Hence the substrate-binding cleft of BcPeh28A is likely to span at least four susbsites. This finding allows us to classify BcPeh28A as a trimer-releasing enzyme, in keeping with other bacterial endoPGs .
In order to gain a more detailed insights into the catalytic site of BcPeh28A, its amino acid sequence was pairwise and structure-based aligned with that of the well-studied non-processive enzyme A. niger endoPGII. It appears (Figures 1 and 4) that the catalytic amino acids forming subsites −1 and +1, belonging to rungs 5, 6, 7 and 8, are very well conserved, whereas the amino acid residues associated with the distal subsites are more variable (Met150/Glu192, Asn186/Gly234, Glu252/Asp310, Asp282/Gln342, Gln288/Thr347, Tyr326/Glu417). It is noteworthy that a serine residue, Ser104, is present in putative rung 2 of BcPeh28A, thus supporting the previously established criteria according to which a single amino acid residue can determine the processive or non-processive behaviour of endoPGs .
A kinetic model that describes the BcPeh28A mode of action on PGA has been constructed (see ‘Kinetic model’ in the Experimental section). The reaction scheme has been rationalized, as illustrated in Figure 5(A). By inspecting the product progress curves (Figure 3), it is possible to assume that OGAs with DPs of 3–6 are produced rapidly after mixing PGA with the enzyme, whereas monomers and dimers emerge only later as second-generation products. Additionally, as previously mentioned, monomers, dimers and trimers accumulate over time, whereas OGAs with DPs of 4–6 appear at the beginning of the reaction and are cleaved later to yield shorter ones. Thus trimer formation is the result of a rapid release from either PGA or DP 4–6 reprocessing. The percentage of uncleaved PGA at the end of the reaction has been evaluated to be 99%, confirming that the time-course experiment was performed in a large excess of substrate. The excellent agreement between the experimental progression curves shown in Figure 3(B) and those obtained as the result of the kinetic model based on a numerical integration method, is shown in Figure 5(B).
A number of conclusions can be drawn from the kinetic constants derived from the model (see Table 2). It has been assumed that the dissociation rate constant of the E–PGA complex, k−1, can be approximated to zero. As a consequence, the specificity constant (kcat/Km) value for PGA approaches the diffusion-controlled rate limits, i.e. the specificity constant for BcPeh28A for PGA approaches its maximum. This means that the catalytic velocity on PGA is only restricted by the rate at which BcPeh28A encounters the polymer in solution. Hence, every encounter between the enzyme and the substrate is productive. Moreover, the kcat value for PGA is higher if compared with the kcat values for oligomers, suggesting that BcPeh28A degrades long polymeric chains faster than the short OGAs.
With regard to the hydrolysis of OGAs, it has been assumed that, once BcPeh28A binds to an oligomer with a DP of 4–6, catalysis takes place at the same rate, independently of the length of the substrate bound (k4=k6=k8). This implies that the rate-determining step of the enzymatic reaction on OGAs is the formation of a stable and productive ES complex. At this point, it is noteworthy that the only kinetic parameters that enable differentiation of oligomers from one another are the dissociation constants, whose values decrease with increasing chain length of the substrates (k−3>k−5>k−7). In order to explain this trend, it is convenient to introduce the concept of binding modes. They can be referred to as the multiple putative modes of interaction that may occur between the enzyme subsites and the saccharide units of the substrate chain. From a catalytic point of view, only the binding modes that involve binding at the subsites −1 and +1 can lead to a product release, thereby permitting discrimination between productive and non-productive binding modes.
Longer substrate molecules, such as those with a DP of 6, that can interact with a large number of enzyme subsites, exhibit a higher overall binding affinity when compared with those of shorter OGAs such as those with a DP of 4 or 5. Binding of shorter OGAs is unfavourable, which is confirmed by their higher dissociation constants. Such a trend of dissociation constants consequently determines the decreasing trend of Km values with the increase in the substrate chain length. It should be emphasized that, since the binding of oligomers such as those with a DP of 4 or 5 can occur in more than one way (Figure 5C), one would expect different affinities for each binding mode. The kinetic model indirectly allowed the identification of the favoured binding modes by introducing a partition coefficient for the distribution of the end products. The resulting distribution led us to conclude that the subsites at positions −3, −2, −1 and +1 display the highest binding affinities, being those that are most likely to be involved in the formation of the favoured productive ES complexes (Figure 5C). Similar conclusions have been reported by studying the active-site architectures of endo-(1→5)-α-L-arabinanases from A. niger and A. aculeatus  and of the endo-β-(1→4)-xylanase from Penicillium simplicissimum . Crystallographic studies on the P. simplicissimum xylanase and biochemical data on the hydrolysis of xylo-oligosaccharides with DPs of 2–6 revealed that 4 is the minimum substrate length for cleavage. This suggests that the two catalytic residues divide the binding cleft into a ‘substrate-recognition area’, from the active site towards the non-reducing end of a bound xylan chain, with strong and specific binding for the first three xylose sugar units, and a ‘product-release area’ with considerably weaker and less specific binding.
Mode of action on OGAs
Hydrolysis of decagalacturonide
To investigate whether BcPeh28A displays a preferential cleavage mode on shorter substrate chains, its mode of action on a pure decagalacturonide was studied. The hydrolysis products were analysed by the highly sensitive HPAEC-PAD technique, since the available amount of the initial substrate was under the sensitivity limit for OGA detection by MEKC-UV. The purity of the decagalacturonide, prior to the enzymatic incubation, was checked by HPAEC-PAD. Small amounts of octa- and nona-galacturonides were detected (Figure 6A). Following the incubation with the enzyme (Figure 6B), the decagalacturonide sample was hydrolysed, and the formation of OGAs with DPs of 3–7 was detected. The reaction was stopped as previously described, by adding 2.0 mM Tris/HCl and 50 mM NaOH. This buffer produced a large peak that masked the signal of monomers and dimers, making their detection impossible. HPAEC-PAD analysis of the decagalacturonide digestion products indicated that BcPeh28A is an endo-acting enzyme that displays a non-processive behaviour also on shorter substrates. As previously observed for the PGA hydrolysis, the main end product of decagalacturonide hydrolysis was identified to be the trimer.
Hydrolysis of hexagalacturonides with different methylation patterns
In order to obtain a more thorough understanding of the enzyme–substrate topology interaction, the mode of action of BcPeh28A on hexagalacturonides (compounds 1–4) that differ in their methylation patterns (Figures 7A and 7B) was investigated. The hydrolysis of compounds 1 and 2, which exhibited an asymmetric methylation pattern, allowed us to define the tolerance towards the methylated saccharide unit of the individual subsites. The definition of the substrate location has been previously established by biochemical studies for the endoPGII of A. niger  and further confirmed by crystallographic studies of the unbound endoPG from Stereum purpureum (a fungus causing silver-leaf disease in apple trees) and of its binary complex with one D-GalA (a β-D-galactopyranouronate was found in subsite +1) and of its ternary complex with two D-GalAs (an additional β-D-galactofuranuronate was found at subsite −1) . Thus the non-reducing end of the substrate points towards the N-terminus of the enzyme, as illustrated in Figure 1.
The end products of the hexagalacturonides (compounds 1–4) hydrolysis were analysed by HPAEC-PAD and ESI-TOF-MS. As reported elsewhere [40,41], the use of the HPAEC-PAD technique is not suitable for characterizing the hydrolytic products, since the methylated OGAs undergo spontaneous demethylation at the pH values reached during the analysis. Consequently, although typical chromatograms showed peaks attributable to both methylated and demethylated oligomers, it was not possible to quantify the amounts of the individual hydrolysis products. Nevertheless, HPAEC-PAD was used to determine the optimal relative concentrations of enzyme and substrate at which the hydrolysis could occur. Indeed, since the methylated hexagalacturonides (compounds 1–4) are not good substrates for BcPeh28A, higher enzyme/substrate ratios were chosen with respect to previous experiments.
HPAEC-PAD revealed that BcPeh28A does not hydrolyse compounds 2 and 4 (results not shown). It is possible to hypothesize that BcPeh28A cannot bind modified substrate molecules such as extensively methyl- or acetyl-esterified OGAs. In fact, such chemical modifications affect the net charge distribution and the steric hindrance of the OGAs and interfere with the ionic interactions, hydrogen-bonding network and hydrophobic interactions that are the driving forces in the formation and stabilization of the ES complexes. By contrast, compounds 1 and 3 are substrates of BcPeh28A, since methylated hydrolysis products could be identified in the chromatograms (results not shown). The identity of the hydrolysis products was assessed by ESI-TOF-MS. Figures 8(A) and 8(B) show the mass spectra of the hydrolysis products originating from the enzymatic degradation of compound 1 after 1 h of incubation at 37 °C. The identification of a peak corresponding to compound 1 (peak 1: m/z 1139.25, [M (the molecular ion)+Na]+) suggested that it was not completely hydrolysed. The monomethylated dimer (peak 1a: m/z 407.08, [M+Na]+) and the dimethylated tetramer (peak 1b: m/z 773.16, [M+Na]+) were identified as the hydrolysis products.
Hence the identification of these two hydrolysis products strongly indicates that the only productive ES complex would involve binding of the hexagalacturonate 1 with the unmethylated saccharide units at the subsites −2, −1 and +1 respectively (Figure 9, panel 1). The structural studies of the ternary complex of S. purpureum endoPG with two D-GalAs  revealed that, in fact, the carboxylate of a GalA residue located at position −1 is engaged in electrostatic interactions with several conserved positively charged amino acids [e.g. the arginine and lysine residues belonging to the RIK (Arg-Ile-Lys) segment present in rung 8; see Figures 1 and 4]. These interactions significantly contribute to the stability of the ES complex. In addition, as observed in the crystal structure of the ternary complex of S. purpureum endoPG , the binding at subsite +1 may provide adequate free energy to permit distortion of the conformation of the sugar residue at subsite −1. The distorted half-chair conformation of the sugar ring bound at site −1 is thought to be involved in the stabilization of the oxocarbenium cation-like transition state . Therefore the binding of a methylated saccharide unit at subsite −1 prevents the formation of a productive ES complex.
The mass spectra of the products released upon the enzymatic hydrolysis of compound 3 are shown in Figures 8(C) and 8(D). No peak attributable to compound 3 was found after an incubation time of 20 min at 37 °C. The monomethylated dimer (peak 3a: m/z 407.08, [M+Na]+) and monomethylated trimer (peak 3b: m/z 583.11, [M+Na]+) were identified as end products. The identification of these two hydrolysis products strongly suggests that two productive ES complexes can occur (Figure 9, panels 2 and 3). In either case, BcPeh28A will accommodate three consecutive unmethylated saccharide units at subsites −2, −1 and +1.
The ESI-TOF-MS analyses showed that compound 1 was not fully hydrolysed by BcPeh28A after 1 h of incubation at 37 °C (Figure 8B), in contrast with compound 3, which, under the same conditions, was completely degraded after 20 min. This finding suggests that compound 3 is a substrate preferred over compound 1. Their different susceptibilities to the enzymatic hydrolysis can be ascribed to the two binding modes in contrast with the one productive binding mode that probably exist for compounds 3 and 1 respectively (Figure 9). However, it should be emphasized that these findings are of a qualitative nature. The ionization efficiency of methylated OGAs is influenced by their chain length and methylation pattern. Namely, the degree of methylation of hexagalacturonides 1 and 3 may affect changes in their volatility and hydrophobicity . Therefore ESI-TOF-MS was reliable only for a qualitative assessment of the mixtures analysed.
Overall, the observed action patterns of BcPeh28A on compounds 1 and 3 and its inability to hydrolyse compounds 2 and 4 indicate that, in order to form a productive ES complex, the substrates should at least contain three consecutive unmethylated saccharide units. In addition, among the six subsites identified, those referred to as −2, −1 and +1 can be supposed to be responsible for crucial ES contacts. Indeed, methylated saccharide units cannot be accommodated at these subsites, thus excluding the formation of the productive ES complex observed for compounds 2 and 4. The subsites −4,−3, +2, +3 and +4, if all of them exist, are likely to be less critical for the formation of a productive ES complex, since they can accommodate a methylated saccharide unit (Figure 9). This finding is in agreement with results obtained from the digestion of pectins with a different extents of methyl-esterification by A. niger and E. carotovora endoPGs .
In perspective, further studies are needed to fully elucidate the structural determinants that are responsible for the binding affinities of each subsite by using pure unmethylated and ad hoc methyl-esterified OGAs with DPs of 5–8, in concert with a site-directed mutagenesis approach aimed at determining the exact topology of the amino acid residues at each subsite.
We gratefully acknowledge Professor William G. T. Willats (Department of Molecular Biology, University of Copenhagen, Copenhagen, Denmark) and Professor K. J. Jensen (Department of Natural Sciences, University of Copenhagen, Copenhagen, Denmark) for their suggestions and for critically reading the manuscript before its submission. We thank Mr D. Massa for his skilful assistance in the preparation of the Figures.
Abbreviations: 4-ABN, 4-aminobenzonitrile; BcPeh28A, an endopolygalacturonase from the phytopathogenic Gram-negative bacterium Burkholderia cepacia; CE, capillary electrophoresis; DP, degree of polymerization; endoPG, endopolygalacturonase; ESI-TOF-MS, electrospray ionization MS with a time-of-flight analyser; GalA, α-D-galacturonic acid; HGA, homogalacturonan; HPAEC-PAD, high-performance anion-exchange chromatography with pulsed amperometric detection; ID%, percentage identity; MEKC, micellar electrokinetic capillary chromatography; OGA, oligogalacturonides; PGA, polyGalA; RG-I, rhamnogalacturonan I; RG-II, rhamnogalacturonan II; rmsd, root-mean-square deviation
- © 2007 Biochemical Society