α-Galacturonidase(s): a new class of Family 4 glycoside hydrolases with strict specificity and a unique CHEV active site motif

Abstract

The catalytic activity of the Family 4 glycosidase LplD protein, whose active site motif is CHEV, is unknown despite its crystal structure having been determined in 2008. Here we identify that activity as being an α-galacturonidase whose natural substrate is probably α–1, 4-di-galacturonate (GalUA₂). Phylogenetic analysis shows that LplD belongs to a monophyletic clade of CHEV Family 4 enzymes, of which four other members are also shown to be galacturonidases. Family GH 4 enzymes catalyze the cleavage of the glycosidic bond, via a non-canonical redox-assisted mechanism that contrasts with Koshland’s double-displacement mechanism.

1. Introduction

Enzymes assigned to glycoside hydrolase Family 4 (GH4) of the 131 member glycoside hydrolase superfamily [1, 2] and http://www.cazy.org/, are unique in their requirements for an essential active-site cysteine (Cys) residue, a nucleotide (NAD⁺), a divalent metal ion (Mn²⁺) and a reducing agent (DTT) for activity [3–7]. Significantly, and in contrast with Koshland’s double-displacement mechanisms [8–11], members of GH4 catalyze the hydrolysis of the glycosidic linkage via a novel sequence of oxidation-elimination-addition and reduction reactions [11–17]. The GH4 Family is also unique in that it includes five groups of enzymes [18], whose substrate specificities correlate remarkably with the presence of a four amino acid motif including the active-site Cysteine residue: 6-phospho-β-glucosidases, CN(V/I)P [5, 9, 11, 13], 6-phospho-α-glucosidases, CDMP [4, 12, 19, 20], α–glucosidases, CHEI [6, 7, 21], α-galactosidases, CH(S/G)V [3, 22–24] and α-glucuronidases, CHGx [25]. In an earlier phylogenetic analysis of 201 GH4 enzymes, we noted a group of proteins of unknown catalytic activity with the motif CHEV (see Fig. 1A, ref. [18]). The structure of one of those proteins, LplD from Bacillus subtilis strain 168, was reported in 2008 (Stephen Almo et. al; unpublished data, Protein Data Bank, ID: 3FEF), but the enzymatic activity of that protein has remained unknown.

Fig. 1.

Maximum Likelihood phylogenetic tree. Red branches: Enzymes with CHEV active site motif; Blue branches: Enzymes with CH(S/G)V motif; Black branches: enzymes from Archaea used to root the tree. Enzymes from species labeled in red have been purified and characterized experimentally.

Because that clade was a sister clade to the α-galactosidases, we suspected these proteins might also be α-galactosidases. Accordingly, the lplD gene was cloned, expressed, and purified (Fig. S1A). Under various experimental conditions, LplD_his6 failed to hydrolyze any of a wide variety of potential α-or β-galactosidase substrates. Significantly, there was also no discernible hydrolysis of any of the p-nitrophenyl glycoside substrate analogs cleaved by other classes in the GH4 Family. This CHEV-containing protein appeared to be devoid of enzymatic function, but for two reasons we hypothesized that LplD_his6 might exhibit galacturonidase activity: (1) except for the substitution of the C6 primary CH₂OH of galactosides by a COOH moiety in galacturonides, the structures of galactosides and galacturonides are similar; (2) B. subtilis and other species in the group, reside in environments containing decaying arboreal and plant materials rich in pectin. The latter, comprises a mixture of heterogeneous acidic polysaccharides including polygalacturonan (PGA). The simplest products of pectin degradation are α-(1, 4)-linked di- and polygalacturonic acid subunits [26, 27]. Testing our hypothesis required the synthesis of commercially unavailable p-NP-α-D- and p-NP-β-D-galactopyranosiduronic acids (pNPα/βGalUA), for use as potential chromogenic substrates for the assay of enzyme activity. Here we report that the LplD protein is indeed an α–galacturonidase, as are four other CHEV proteins from four different sub-clades (Fig. 1). We present phylogenetic, enzymatic, and mechanistic analyses of the CHEV proteins, and discuss the functional significance of the Cys-motifs and their plausible contribution(s) to enzyme specificity.

2. Materials and methods

2.1 Synthesis of chromogenic substrates

Chromogenic galacturonides used in this study were prepared from α- and β-linked p nitrophenyl galactopyranosides as described in Supplementary Detailed Methods.

2.2 Colorimetric assay of α-galacturonidase activity

α-Galacturonidase activity was measured by a discontinuous colorimetric assay using pNPαGalUA as substrate. (see, Supplementary Detailed Methods).

2.3 Expression of his₆-tagged LplD protein

The modified pET26b vector (Novagen) encoding the his-tagged LplD gene fusion (pSGX3) was constructed by Dr. S. Almo and colleagues at the New York Research Center for Structural Genomics (NYSGXRC). The E. coli DH5α clone containing pSGX3 was obtained from the PlasmID repository maintained by the Harvard Institute of Proteomics ID: NYSGXRC-11137. (The PSI-MR CloneID is designated BsCD00 292418). For protein expression pSGX3 was transformed into E. coli BL21(DE3) cells.

2.4 Purification of LplD and its homologs

The purification of LplD, and the cloning and purification of its homologous gene products, are described in Supplementary Detailed Methods.

2.5 Phylogenetic analysis

Identification of homologs of the B. subtilis lplD gene, alignment of the gene sequences, and phylogenetic analyses are described in Supplementary Detailed Methods.

3. Results and Discussion

3.1 Attempts to determine the substrate specificity of LplD from B. subtilis

The CHEV-containing LplD_his6 protein of B. subtilis strain 168 was purified as described in Supplementary Detailed Methods (Fig. S1A). When incubated with the requisite cofactors for activity of Family GH4 enzymes (NAD⁺, Mn²⁺ and DTT), LplD_his6 failed to hydrolyze any of a wide variety of potential α- or β-galactosidase substrates, including melibiose, raffinose, lactose and the chromogenic analogs, pNPαGal and pNPβGal. Other potential substrate tested (without success) included: pNP-N-acetyl-α-and β-D-galactosaminide; pNP-N-acetyl-α-and β-D-glucosaminide; pNP-α-L and β-L-arabinopyranoside; pNP-α- and β-D-glucuronide; pNP-α-and β-D-glucopyranoside; pNP-α- and β-D-glucopyranoside 6-phosphate; pNP-α-D-galactopyranoside 6-phosphate and oNP-β-D-galactopyranoside 6-phosphate.

3.2 LplD from B. subtilis is an NAD⁺/Mn²⁺/DTT-dependent α-galacturonidase

Failure to detect any catalytic activity suggested to us that the natural substrate(s) for LplD and its homologs in other clades might be unusual galactosides, or perhaps galactose-containing polymers. A common characteristic of the microbial species containing the CHEV motif, including B. subtilis [28], Clostridium phytofermentans [29], Caldicellulosiruptor saccharolyticus [30], and Thermoanaerobacter italicus [31], is that they have the capacity to degrade pectin, to yield simpler products, including α-(1, 4)-linked di- and polygalacturonic acid subunits. Accordingly, we hypothesized that proteins assigned to the five clades shown in Fig. 1, might exhibit galacturonidase activity. The availability of the chromogenic glycosides, pNPαGalUA and pNPβGalUA, enabled us to test this hypothesis. When incubated in a reaction mixture containing optimum concentrations of requisite cofactors, LplD_his6 rapidly hydrolyzed pNPαGalUA (Fig. S1B). However, neither pNPβGalUA, nor the stereoisomeric pNPαGlcUA were hydrolyzed by the enzyme. These latter findings show that both the α-configuration of the anomeric center and the axial orientation of the -OH group at C4, are crucial determinants of substrate selectivity. Under optimum conditions of temperature (~ 35–38 °C) and cofactor concentration (25 mM Tris-HCl pH 7.5, 0.1 mM NAD⁺, 1 mM MnCl₂ and 5 mM DTT), LplD_his6 exhibited the following kinetic parameters with pNPαGalUA as substrate: V_(max) = 4.33 ± 0.51 μmol/min/mg prot; K_m = 0.62 ± 0.11 μM and k_cat = 3.6 sec ⁻¹. Addition of 5 mM EDTA to the reaction mixture abolished all enzymatic activity. The results of a time course analysis by TLC (Fig. 2) provided further validation for the α-galacturonidase designation of LplD. As illustrated in this figure, incubation of the GH 4 enzyme with α-(1–>4) linked di-galacturonate (GalUA₂) resulted in a slow, but significant formation of (mono) galacturonic acid (GalUA).

Fig. 2.

Thin layer chromatography of degradation of di-galacturonate (GalUA₂) to galacturonic acid (GalUA). Standards: 1=GalUA₂, 2= GalUA, 3= mixture of GalUA and GalUA₂. Standard 1 is also the zero time sample for this reaction. * indicates a reactive contaminant in the reaction mixture

3.3 Phylogenetic analysis

In our earlier phylogenetic analysis (see Fig. 1A, ref. [18]) the CHEV clade included only six members. A current BLAST [32] search of GenBank’s non-redundant protein database reveals that since 2009 an additional 24 Family GH4 proteins have been found to contain this unusual CHEV motif. To determine whether those proteins constituted a monophyletic clade the 30 CHEV sequences were together aligned with 17 α-galactosidases (CHSV motif) and four Archaea sequences of unknown function to use as an outgroup for rooting the tree. Consideration of this alignment revealed that all (except five) of the CHEV sequences were preceded by another Cys residue. The Cys in the four amino acid motif is present in all members of Family GH4, and is coordinately bound to the active-site Mn²⁺ ion. The presence of an additional cysteine residue CCHEV was surprising, and led us to consider the possibility that in this group of proteins, our presumed four amino acid motif, might actually comprise five residues in which the second Cys residue was the catalytically conserved cysteine. A maximum likelihood phylogenetic tree based on this alignment showed that the CHEV proteins did indeed constitute a monophyletic clade (Fig. 1, red lines), separate from the α–galactosidase clade (blue). In all of the α-galactosidases and the Archaeal proteins, the active site C was preceded by L, giving further support to the notion of a five-member motif (Fig. 1 and Table S1). The ancestral state of the five-member motif was estimated for each of the internal nodes (Fig. 1), and this analysis also indicated that the common ancestor of both the α–galactosidases and the CHEV-containing proteins (Node 1) initially possessed the motif LCHGV. The change to LCHEV occurred early, and distinguished the CHEV from the α-galactosidase clade. Four proteins retained the LCHEV motif (and constitute the monophyletic clade 5), while the other branch descending from the LCHEV node mutated from LCHEV → CCHEV, a change that was retained in all but one of the descendants of that branch. The phylogenetic analysis suggested two hypotheses: (1) that the change from LCHGV → LCHEV resulted in a changed substrate specificity that might be common to all those proteins shown in red in Figure 1, and as stated earlier, (2) that the functionally specific motif comprises five, rather than four residues, as previously assumed. The organisms in clade 5 are all assigned to the Archaea family Halobacteriaceae, suggesting that a Eubacterial gene had been horizontally transferred to the Halobacteriaceae early in the evolution of the CHEV lineage. One additional enzyme, from Oceanicola sp., contained the motif ECHEV. We speculated that the LCHEV enzymes might have a function different from the majority of CCHEV enzymes, and that the ECHEV protein might have yet another function or might be devoid of enzymatic activity. To test our two hypotheses, it was necessary to determine the catalytic function(s) of selected enzymes from each of the five (L/C)CHEV clades.

3.4 Does the Cys-motif of GH 4 α-galacturonidase(s) contain four or five residues ?

To examine the roles of the first residue in the five-residue Cys-motif in determination of enzyme specificity, the genes encoding the designated proteins from Thermoanaerobacter italicus (clade1), Thermoanaerobacterium saccharolyticum (clade 2), Clostridium phytofermentans Cphy3396 (clade 3), Oceanicola sp (ECHEV motif), and Halorhabdus utahensis (clade 5, LCHEV) were synthesized with his6-tags, cloned, expressed and purified (see, Supplementary Detailed Methods). Together with the LplD enzyme from Bacillus subtilis (clade 4), the six proteins provided a representative sample across the entire (L/C)CHEV clade. All of the purified proteins (except that from Halorhabdus utahensis) displayed only α-galacturonidase activity, and for reasons of simplicity pNPαGalUA served as substrate in all assays of protein activity. The Halorhabdus utahensis protein was inactive toward all substrates tested and its function (if any) is unknown. Collectively, our enzymatic data supported the hypothesis that the change of residue G → E in the motif at Node 1 (Fig. 1), was instrumental in the acquisition of α-galacturonidase specificity. Importantly, the finding that the Ocianicola sp. enzyme is a functional α-galacturonidase, provided evidence that the first of the cysteine residues in the CCHEV motif of clades 1–4, is not essential for enzyme specificity. Conceivably, the initial leucine residue in LCHEV is incompatible with this activity, which raises the question as to the catalytic nature of the enzyme when the G → E substitution first occurred. Alternatively, it may be that the first residue is not really important to activity after all, and that the proteins in clade 5 do not hydrolyse pNPαGalUA for reasons unrelated to the active site motif. In an attempt to resolve these issues, site-directed mutagenesis was used to change the Halorhabdus utahensis protein motif from LCHEV → CCHEV. This substitution failed to elicit detectable α-galacturonidase activity in the mutant protein. Likewise, change in the motif of LplD_his6 from CCHEV → LCHEV, did not significantly affect α-galacturonidase activity. Finally, change in the Oceanicola sp. motif from ECHEV → CCHEV, was also without a discernable effect on enzymatic activity. From these site- directed mutagenesis data we conclude: (i) the lack of detectable activity in the clade 5 enzyme is unrelated to the sequence of the active site motif, and probably represents a physiological change or environmental adaptation of the Archaea family Halobacteriaceae, and (ii) that the identity of the first residue in the five-residue motif is not a determinant for enzyme specificity; i.e. the functional Cys-motif in α-galacturonidases consists of only four residues. The second conclusion warrants some attention, because the phylogenetic evidence for a five-residue motif was fairly persuasive. In light of the time (and effort) required for the biochemical characterization of enzymes, we were initially tempted to present the five-residue hypothesis as a justifiable conclusion. The results of the site-directed mutagenesis studies illustrate the risks of accepting an attractive concept in the absence of supportive experimental evidence. At the same time, our rejection of the five-residue motif hypothesis is based solely on analysis with a synthetic substrate. We cannot rule out the possibility that the fifth residue might contribute to specificity for some other unknown substrate.

3.5 Two mechanisms for the conversion of digalacturonate (diGalUA/GalUA₂)

The GH Family 4 presently includes five groups of enzymes with different substrate specificities ([1], and http://www.cazy.org/). From our investigation a new class of enzymes, the α-galacturonidases, can now be assigned to this unique family. Although the members of this sixth class readily hydrolyze the chromogenic analog pNPαGalUA (Fig. S1B) it is probable that diGalUA/GalUA₂ (Fig. 3B) is the naturally occurring substrate for these enzymes.

Fig. 3.

Catalytic mechanisms of GalUA₂ degradation. A: OGalUA lyase mechanism. B: Family GH 4: LplD mechanism

In this context, it should be noted that oligogalacturonate lyase (OGalUA lyase), assigned to the pectate lyase family (CAZy database; Family PL 22, ref. 2), also cleaves GalUA₂, but produces monogalacturonate and 5-keto-4-deoxyuronate (Fig. 3A). However, the reaction mechanisms of GH 4 α-galacturonidase and OGalUA lyase are markedly different (Fig. 3). Polysaccharide lyases such as OGalUA lyase, degrade their substrates through a non-hydrolytic, elimination mechanism resulting in the release of an unsaturated oligosaccharide product that contains a double bond between C4 and C5 of the non-reducing end sugar. This mechanism relies upon the relative ease of deprotonation at C5, α to the substrate carboxylic acid moiety, followed by trans-elimination of the 4-linked glycoside. By contrast LplD is a typical GH4 glycosidase, that cleaves the glycosidic bond, via a non-canonical redox-assisted mechanism that had been unsuspected for the past 60 years of study of glycoside hydrolysis[11–15, 17]. The GH Family 4 enzymes contain a tightly bound NAD⁺ cofactor that, in the first step, oxidizes the bound substrate at the 3-position to generate a ketone. The C-2 proton, α to the carbonyl is thereby rendered acidic and is removed by an active site base to generate an anion that is stabilized, in part, by a Cys -coordinated active site Mn²⁺ ion. Glycosidic bond cleavage then ensues to generate an α,β-unsaturated keto sugar bound in the active site. Addition of water to the anomeric centre is then followed by re-protonation at C2, and finally reduction of the C3 ketone by the “on-board” NADH. This last step releases the net hydrolysis products with retention of stereochemical configuration at C1, and returns the enzyme to its resting state.

4. Summary and conclusions

Family GH4 now comprises six groups of enzymes with quite different substrate specificities: phospho-α-glucosidase (EC 3.2.1.12); α-glucosidase (EC 3.2.1.20); α-galactosidase (EC 3.2.1.22); phospho-β-glucosidase (EC 3.2.1.86); α-glucuronidase (EC 3.2.1.139) and α-galacturonidases (EC 3.2.1.67). The presence of a characteristic four- residue Cys- motif is strongly predictive of substrate specificity for proteins assigned phylogenetically to a particular group or clade. Although intriguing, the basis for this striking correlation is not immediately apparent. The individual subunits of these (oligomeric) proteins comprise ~ 450–480 residues, and it is axiomatic that the Cys-motifs per se cannot be the sole determinants for substrate selectivity. In all cases, cysteine and its adjacent residues reside within the active-site, and it is conceivable that the four amino acids provide a nucleus for architectural assembly of the requisite folds or structural domains, that ultimately dictate catalytic specificity. To date, the crystal structures of representatives from four of the six GH 4 enzyme groups have been deposited in the Protein Data Bank: phospho-α-glucosidase (1U8X); α-galacturonidase (3FEF); α-glucosidase (1OBB) and phospho-β-glucosidase (1UP4). Unfortunately, in no case has the structure of an enzyme been determined in complex with its intact substrate and requisite cofactors (NAD⁺ and Mn²⁺) at the active site. Future structural resolution of such complexes, may provide insight to the role(s) of the Cys-motifs in substrate discrimination by the six groups of enzymes in the GH4 Family.

Highlights.

• Family 4 glycosidases with a CHEV active site motif form a monophyletic clade.

• The crystal structure of one CHEV glycosidase (LplD) was determined in 2008.

• The catalytic activity of these enzymes has until now remained unknown.

• These enzymes are α-galacturonidases.

• They employ a non-canonical redox-assisted catalytic mechanism.