OUP user menu

Galactomannan hydrolysis and mannose metabolism in Cellvibrio mixtus

Maria S.J. Centeno, Catarina I.P.D. Guerreiro, Fernando M.V. Dias, Carl Morland, Louise E. Tailford, Arun Goyal, José A.M. Prates, Luís M.A. Ferreira, Rui M.H. Caldeira, Emmanuel F. Mongodin, Karen E. Nelson, Harry J. Gilbert, Carlos M.G.A. Fontes
DOI: http://dx.doi.org/10.1111/j.1574-6968.2006.00342.x 123-132 First published online: 1 August 2006

Abstract

Galactomannan hydrolysis results from the concerted action of microbial endo-mannanases, manosidases and α-galactosidases and is a mechanism of intrinsic biological importance. Here we report the identification of a gene cluster in the aerobic soil bacterium Cellvibrio mixtus encoding enzymes involved in the degradation of this polymeric substrate. The family 27 α-galactosidase, termed CmAga27A, preferentially hydrolyse galactose containing polysaccharides. In addition, we have characterized an enzyme with epimerase activity, which might be responsible for the conversion of mannose into glucose. The role of the identified enzymes in the hydrolysis of galactomannan by aerobic bacteria is discussed.

Keywords
  • α-galactosidase
  • epimerase
  • galactomannan
  • glycoside hydrolase
  • Cellvibrio

Introduction

The plant cell wall represents the most abundant reservoir of renewable energy and carbon within the biosphere (Gilbertet alet al, 2002). Mannan consists of a backbone of β-1,4-linked mannose residues while in glucomannans the backbone comprises both glucose and mannose residues randomly distributed (Brett & Waldren, 1996). Galactose monomers may decorate the mannose residues of both hemicelluloses through α-1,6 linkages and therefore these polysaccharides are usually referred to as galactomannans and galactoglucomannans, respectively. Galactomannans are also storage carbohydrates in some plant seeds such as carob (Brett & Waldren, 1996). The hydrolysis of mannose containing polysaccharides into its monomeric components requires the action of endo-β-1,4-mannanases, exo-β-1,4-mannosidases and β-1,4-glucosidases (McCleary, 1988; Puls & Schuseil, 1996).

In the sequence-based classification of glycoside hydrolases (GH) α-galactosidases are located in families 4, 27 and 36 (Coutinho & Henrissat, 1999). Family 4 is exclusively prokaryotic, family 36 contains both bacterial and eukaryotic enzymes, while family 27 comprises predominantly eukaryotic GHs. While most eukaryotic α-galactosidases are in family 27, the majority of prokaryotic enzymes have been classified into families 4 or 36. Ultimately, the substrate specificity of these enzymes is not restricted to the de-branching activity of hemicelluloses, as α-galactosidases have been shown to be crucial in other biological systems. For example, removal of galactose from short chain galacto-oligosaccharides, such as melibiose, raffinose and stachyose that are carbohydrate reserves in leguminous plants, is catalysed by plant and microbial α-galactosidases usually located in families 4 and 36 (Fridjonssonet alet al, 1999). In eukaryotes, lysosomal α-galactosidases are involved in the catabolism of large macromolecules, such as glycoproteins and glycolipids, and these enzymes belong to family 27 (Garman & Garboczi, 2004). Interestingly, most α-galactosidases acting on branched hemicelluloses were shown to have broad substrate specificities, also hydrolysing low molecular weight substrates (Deyet alet al, 1993). There is, however, a paucity of information relating to the molecular determinants of substrate specificity in enzymes acting on the removal of side chains from hetero-polymeric mannans.

The metabolism of the hydrolysis products of galactose containing mannans and glucomannans, by aerobic soil bacteria such as Cellvibrio mixtus, is poorly understood. This is surprising, considering the detailed information that is available on the hydrolysis of plant cell wall polysaccharides. It is generally established that mannose residues are phosphorylated by a transmembrane permease to mannose 6-phosphate and then converted to fructose 6-phosphate by an intracellular phosphomannose isomerase (Binetet alet al, 1998). In addition, galactose might be transformed into galactose 1-phosphate by an intracellular galactokinase and then converted into glucose 1-phosphate through the Leloir pathway (Holdenet alet al, 2003). These possibilities remain, however, to be confirmed in Cellvibrio.

Here we report the cloning and the biochemical characterization of two enzymes involved in the degradation and metabolism of galactomannan, from the prolific plant cell wall degrader Cellvibrio mixtus. It is shown that the family 27 α-galactosidase of Cellvibrio mixtus is a polymer specific enzyme with very limited activity on galacto-oligosaccharides of the raffinose series. In addition, CmEpiA is responsible for the interconversion of mannose into glucose by Cellvibrio mixtus.

Materials and methods

Bacterial strains, plasmids and culture conditions

Escherichia coli strains used in this study were BL21(DE3), Origami (Novagen) and XL1-Blue (Stratagene). The bacteriophages and plasmids used in this work were λZAPII (Stratagene), pBluescript SK (Stratagene), pGEM-T (Promega) and pET21a (Novagen) and pET28a (Novagen). Cellvibrio mixtus (NCIMB 8633) was cultured aerobically at 20°C in Dubos mineral salts medium or on Dubos agar plates overlaid with filter paper (Millward-Sadleret alet al, 1995). Culture conditions were maintained as described previously (Millward-Sadleret alet al, 1995). The recombinant proteins, encoded in pET21a and pET28a derivatives, were expressed in E. coli BL21 and Origami strains, as described below.

General recombinant DNA procedures

Transformation of E. coli, digestion of DNA with restriction endonucleases, ligation of DNA with T4 DNA ligase, agarose gel electrophoresis, Southern hybridization, slot blot and plaque hybridizations were carried out as described previously (Berns & Thomas, 1965; Sambrooket alet al, 1989). The genomic library was constructed in λZAPII using the approach described by Clarke (1991). The library was plated on NZYM top agar at a density of three plaques per cm2 and screened for α-galactosidase activity using the methylumbelliferyl α-d-galactosidase overlay technique. DNA hybridizations were performed using the fluorescein system from Amersham according to the manufacturer's protocol. Two inverse PCR amplifications were performed to clone the full-length sequence of the genes located upstream from aga27 (encodes CmAga27A). Briefly, for the first inverse PCR experiment, Cellvibrio mixtus genomic DNA was digested with the enzyme EcoRI and resulting DNA fragments ranging from 4.5 to 5.5kb were eluted from agarose gels and religated. In the second experiment, the enzyme used was PstI and the fragments ranging from 1.5 and 2.5kb were ligated. The primer pairs used in the first and second amplifications were as follows: first amplification (EcoRI), 5′-GCAGGTCCATAAACATACGC-3′ and 5′-GGTTGGCCAATATTTTGCGC-3′; second amplification (PstI), 5′-CAATGACAGATGAAGGAAGC-3′ and 5′-CTTAGCGCTATTGCTGATTG-3′. PCRs were performed using 1U of the thermostable DNA polymerase pFU turbo (Stratagene) and following the manufacturer's instructions. PCR products were cloned into pGEM-T (Promega) and the resulting plasmids were named pZC3 and pZC4, respectively. The nucleotide sequence of DNA was determined with an ABI Prism Ready Reaction DyeDeoxy terminator cycle sequencing kit and an Applied Biosystems 377A sequencing system.

Expression and purification of recombinant proteins

To express CmAga27A, CmEpiA and CmUnkA in E. coli, the DNA encoding the mature enzymes were amplified by PCR from Cellvibrio mixtus genomic DNA using the thermostable DNA polymerase pFU Turbo (Stratagene). The primers used for the amplifications incorporated designed restriction sites (Table 1). The PCR products were cloned into pGEM-T and sequenced to ensure that no mutations had occurred during PCR. The recombinant derivatives of pGEM-T containing the Cellvibrio mixtus genes were digested with the required restriction enzymes (Table 1) and the excised DNA fragments were cloned into the similarly digested vector. The three recombinant proteins contain N- or C-terminal His6-tag. The resulting plasmids encoding recombinant CmAga27A, CmEpiA and CmUnkA, were named pZC5, pZC6 and pZC7, respectively.

View this table:
Table 1

Primers, vectors and restriction sites used to hyperexpress CmAga27A, CmEpiA and CmUnkA in Escherichia coli

GeneRestriction sitesVectorPrimers
aga27ANdeIpET21a5′CTCCATATGCAAAAATTTGAGCATCTC
XhoI5′CACCTCGAGCTGCGGTGTTAAACGCAA
epiANheIpET28a5′CTCGCTAGCAACGCCGCCATGCTTG
XhoI5′CACCTCGAGTTAAGTTGTGTGTAC
unkANheIpET21a5′CTCGCTAGCTTTAAAGAAAAAGCAAAA
XhoI5′CACCTCGAGAACCAAATCCTTCATTAC
  • Bold type represents restriction sites designed in primers.

Escherichia coli BL21 harbouring pET derivatives encoding CmEpiA and CmUnkA was cultured in LB containing 100μgmL−1 ampicillin at 37°C to mid-exponential phase (A550nm 0.6) at which point isopropyl-β-d-thiogalactopyranoside (IPTG) was added to a final concentration of 1mM and the cultures were incubated for a further 5h. Recombinant CmAga27A formed inclusion bodies when expressed under those conditions using E. coli BL21. Therefore, the pET21a derivative, pZC5, encoding the Cellvibrio mixtusα-galactosidase was transformed into E. coli Origami. Recombinant E. coli was cultured as described above but after induction the cells were maintained at 16°C for 16h. Cells were pelleted by these cultures by centrifugation and the bacterial pellets were resuspended in 50mM sodium HEPES buffer, pH 7.5, containing 1M NaCl and 10mM imidazole. The buffer of CmAga27 also contained 5mM dithiothreitol. The three recombinant proteins were purified by immobilized nickel affinity chromatography as described previously (Carvalhoet alet al, 2004) and exchanged into the buffers described below.

Enzyme assays

The activity of CmAga27A was routinely determined with 10mM p-nitrophenyl α-d-galactopyranoside (pNPαGAL), at 37°C, in 25mM Tris-HCl buffer, pH 8.5 (pH optimum of the enzyme), containing 5mM dithiothreitol. To screen for activity against 4-nitrophenyl-glycosides, substrate and enzyme concentrations were 10mM and 1μM, respectively, and assays were carried out for up to 1h. One unit of enzyme for pNPαGAL was defined as the amount of enzyme that liberates 1μmol of p-nitrophenol per min from pNPαGAL. The pH profile of the α-galactosidase was determined with 10mM pNPαGAL using the following buffers: pH 6–8.5, 50mM phosphate; pH 8.5–9.5, 50mM Tris-HCl. The resistance of the enzyme to proteinase and thermostability were determined as described previously (Fonteset alet al, 1995). Activity of CmAga27A against oligosaccharides of the raffinose series and polysaccharides was determined by incubating 5μg of enzyme, in 25mM Tris-HCl buffer, pH 8.5, containing 5mM dithiothreitol, with the appropriate substrate [1% (w/v)] in a volume of 100μL and released reducing sugar was measured with the Somogyi–Nelson assay (Nelson, 1944; Somogyi, 1952). To determine the kinetic parameters of the enzyme against its target substrates, an appropriate dilution of the enzyme was incubated with concentrations of substrate ranging from 0.05 to 10gL−1 for polysaccharides and 0.01–10mM for pNPαGAL. Activities were measured in the linear phase of the enzyme reactions. All reported results are the mean of three replicates.

Phosphomannose isomearse activity (converts mannose 6-phosphate to fructose 6-phosphate) was measured using a coupled enzymatic assay utilizing phosphoglucose isomerase (converts fructose 6-phosphate to glucose 6-phosphate) and glucose 6-phosphate dehydrogenase (oxidizes glucose 6-phosphate and reduces NADP + to NADPH). Mannose 6-phosphate was used as the substrate and the progress of the reaction was followed spectrophotometrically at 340nm by measuring the production of NADPH (Todd & Tague, 2001). To confirm the validity of the assay, the increase in absorbance at 340nm was monitored in a reaction containing 0.6 units of a commercial phosphomannose isomerase (Sigma p-5153). The reaction mixture consisted of 27mM Tris-HCl, pH 7.5, containing 5mM MgCl2, 1mM β-NADP, 1UmL−1 glucose 6-phosphate dehydrogenase (Sigma G-8289) and 0.5UmL−1 of phosphoglucose isomerase (Sigma P-5381). The reaction was started by adding 3mM of mannose 6-phosphate (Sigma M-6876) and was monitored at 37°C in an Ultrospec III spectophotometer (Pharmacia) using the enzyme kinetic module of Biochrom Ltd software package v. 1.01 (Pharmacia). Mannokinase activity was measured using, essentially, the assay system described above but using mannose as the substrate and Sigma phosphomannose isomerase to convert mannose 6-phosphate into fructose 6-phosphate. To measure the kinetic parameters of CmEpiA, 25μg of enzyme were incubated with 5–500mM mannose in a volume of 100μL at 37°C in the buffer conditions described above. Glucose production was monitored at various time points (up to 30min) with the d-Glucose-HK kit from Megazyme, using a final ATP concentration of 20mM. Optimum pH and temperatures were determined as described above.

Results

Cloning and sequencing of Cellvibrio genes

A Cellvibrio mixtus genomic library, prepared in λZAPIII, was screened for α-galactosidase activity. Four α-galactosidase positive clones were isolated from the library and purified to homogeneity by replating on the host strain. The corresponding recombinant plasmids were excised from the isolated clones and rescued into E. coli cells which displayed activity against pNPαGAL (not shown). Restriction mapping and DNA hybridization studies revealed that all isolated DNA fragments comprise the same 4.7-kb Cellvibrio mixtus genomic region (not shown) and this plasmid was named pZC2. The isolated DNA fragment was sequenced revealing the presence of an ORF of 1215bp, which was designated aga27A. The gene encodes a polypeptide of 405 residues, named CmAga27A, with a predicted Mr of 45819. The region upstream aga27, which revealed the presence of an incomplete ORF, was cloned using the inverse PCR approach described previously (Fig. 1a). The derived sequence comprised a cluster of three genes of 1179, 1230 and 1368bp, which were designated unkA, epiA and man5A, respectively (Fig. 1b). The intercistronic regions separating unk-epiA and epiA-man5A were of 3 and 12bp, respectively. The proteins encoded by these three genes were designated CmUnkA, CmEpiA and CmMan5A, respectively, and displayed predicted Mrs of 44188, 46524 and 51620. The nucleotide sequence of the 7-kb sequence characterized in this report appears in GenBank with the accession number AY526725. Southern hybridization using aga27A as the probe showed that a single copy of the gene is present in the bacterial genome (data not shown). Two putative promoters were identified upstream of unkA, epiA, man5A and aga27A, with −10 and −35 well conserved regions (Fig. 1c). In addition, two regions of dyad symmetry were identified 18 and 22 nucleotides downstream of the putative translational stop codon of man5A and aga27A, respectively (Fig. 1b). The two 26-bp regions have the potential to form a substantial stem–loop structure, characteristic of a ρ-independent transcription terminator sequences.

Figure 1

Organization of the Cellvibrio gene cluster encoding galactomannan degrading enzymes. Plasmids containing DNA fragments that allowed the assembly of the 7-kb sequence containing the four genes identified in this work are displayed in (a). Putative terminator and promoter sequences identified in the sequence and the position and orientation of the ORFs are displayed in (b). (c) depicts the sequences of the putative promoters. P, promoter; T, terminator.

Comparison of the primary sequence of CmAga27A with biological databanks, searched by blast (http://www.ncbi.nlm.nih.gov/BLAST), revealed that the protein display considerable sequence similarity with GH27 enzymes. Sequence alignments (Fig. 2a), show that CmAga27A displays the highest identities (>40%) with GH27 putative enzymes from Microbulbifer degradans (ZP_066516), Saccharopolyspora erythraea (AAC99325), Streptomyces avermitilis (NP_822650) and α-galactosidases CjAga27A and Aga27A from Cellvibrio japonicus and Clostridium josui, respectively (BAB83765). Interestingly, GH27 enzymes presenting lower identity scores with CmAga27A (below 47%) are all from eukaryotic origin. It has been shown that in GH27 substrate specificity is mediated by the nature of the β5–α5 loop, which was recently called the substrate recognition loop (Fujimotoet alet al, 2003). In the six GH27 bacterial enzymes the presence of cysteine and tryptophan residues in the loop, which coordinate the 2-hydroxyl of galactose in plant and fungi α-galactosidases, suggest that all the bacterial enzymes are α-galactose specific (Fig. 2a). A phylogeny tree of the 36 GH27 enzymes with demonstrated α-galactosidase activity (see CAZy database), reveals that prokaryotic and eukaryotic proteins have evolved from a common precursor, although mammalian, plant, fungi and bacterial sequences segregate into distinct clusters (Fig. 2b). In common with most α-galactosidases, CmAga27A displays no evidence of modular architecture or delineating linker sequences. In addition, CmAga27A contains a typical signal peptide in which two N-terminal basic residues are followed by 20 small hydrophobic amino acids capable of forming a α-helical structure, with cleavage predicted to occur between Ala-23 and Gln-24. In a previous report, it was demonstrated that CmMan5A belongs to GH5 and displays exo-β-mannosidase activity, releasing mannose from the nonreducing end of manno-oligosaccharides and polysaccharides (Diaset alet al, 2004). Analysis of the primary structures of CmEpiA and CmUnkA, suggests that both proteins are nonmodular and intracellular. Comparison of the sequence of the two deduced polypeptides with biological databanks showed that the protein encoded by unkA has various protein homologues with no demonstrated enzymatic function, while CmEpiA exhibits 55% sequence identity with a protein from M. degradans (ZP_0315883) and lower identity scores (below 25%) with bacterial N-acyl-d-glucosamine-2-epimerases and phosphomannose isomerases. In this work we were unable to assign a biochemical role for CmUnkA.

Figure 2

Alignment of bacterial GH27 enzymes (a) and unrooted phylogenetic tree of functional α-galactosidases from GH27 (b). In (a), GH27 bacterial α-galactosidases are aligned with the Homo sapiens homologue from which the 3D structure is known (Garman & Garboczi, 2004). Catalytic residues (D152 and D207) and the tryptophan that stacks galactose at the active site (W40) are depicted in gray. The substrate recognition loop is boxed (see text). Secondary structure elements are depicted using arrows (β-strands) and cylinders (α-helix). In (b), the scale bar indicates the number of substitutions per position following alignment with clustalw (Thompsonet alet al, 1994) and bootstrap analysis using the same software. The tree was displayed with TreeView (Page, 1996). Branches not circled contain sequences from fungi. Two letters that indicate the genus and species, respectively, of the endogenous host precede the protein accession numbers (GenBankTM or Swiss-Prot).

Biochemical properties of CmAga27A

To investigate the biochemical properties of CmAga27A, the protein was expressed in E. coli Origami and purified to homogeneity by metal affinity chromatography (Fig. 3a). The pH and temperature profiles of CmAga27A were determined with pNPαGAL (Figs 4a and b). The enzyme was found to display maximum catalytic activity at 50°C although it was rapidly inactivated at temperatures above 55°C (not shown). The pH optimum of CmAga27A was found to be of 8.5 and the enzyme, in common with the majority of the extracellular plant cell wall hydrolases, was completely resistant to proteolytic inactivation. Purified CmAga27A display high levels of activity against pNPαGAL although substrate inhibition was detected at substrate concentrations above 20mM. Enzyme kinetics were observed over the range 0.01–10mM pNPαGAL, allowing for the estimation of the apparent kinetic parameters that gave values of 184.9±4.8min−1 and 0.23±0.02mM for kcat and Km, respectively. In contrast, the enzyme does not hydrolyse aryl-β-mannosides, 4-nitrophenyl-β-glucopyranoside or aryl-β-arabinosides or the polysaccharides xylan, hydroxyethyl and carboxymethyl cellulose, mannan, undecorated glucomannan, lichenan and laminarin. Interestingly, CmAga27A exhibited activity against carob, locust bean and guar gum galactomannans. The kinetic constants displayed by the enzyme against these substrates are presented in Table 2. The capacity of CmAga27A to hydrolyse the α-1,6-galactoside linkage in oligosaccharides of the raffinose family was evaluated. The data demonstrated that the Cellvibrio mixtusα-galactosidase display a very marginal capacity to hydrolyse raffinose (0.01Umg−1) and estaquiose (0.006Umg−1). Taken together the results suggest that CmAga27A is an enzyme that Cellvibrio mixtus uses to remove galactose side chains from galactomannans.

Figure 3

Hyperexpression and purification of recombinant CmAga27A (a) and CmEpiA (b). BL21 cells were transformed with the respective recombinant plasmids and induced with isopropyl-β-d-thiogalactopyranoside as described in Materials and methods. Soluble extracts from uninduced (lane 1) and induced (lane 2) cells were separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis and the proteins were purified through affinity chromatography (lane 3).

Figure 4

Activity of CmAga27A (▴) and CmEpiA (▪) at different pH values (a), temperatures (b) and kinetics of CmEpiA (c). Enzyme activity in the Y-axis of (c) is expressed in molecules of glucose produced per molecule of CmEpiA per minute.

View this table:
Table 2

Kinetic parameters of CmAga27A during the hydrolysis of pNPαGAL and various galactomannans

Substratekcat (min−1)Km (gL−1)kcat/Km
pNPαGAL184.9 ± 4.80.23 ± 0.02803.91
Guar gum galactomannan5.26 ± 0.50.82 ± 0.066.42
Locust bean galactomannan8.27 ± 0.81.10 ± 0.077.52
Carob galactomannan (hv)8.81 ± 0.40.62 ± 0.0314.21
Carob galactomannan (lv)31.3 ± 0.92.21 ± 0.0414.16
  • * k cat are in molecules of product per molecule of enzyme per minute.

  • kcat/Km values are in (gL)−1min−1.

  • Substrate concentration in mM.

  • pNPαGAL, p-nitrophenyl α-d-galactopyranoside; hv, high viscosity; lv, low viscosity.

Biochemical properties of CmEpiA

Initially, the possibility that CmEpiA was the phosphomannose isomerase responsible for the interconversion of mannose 6-phosphate into fructose 6-phosphate in Cellvibrio mixtus, was evaluated. In the genetic locus identified in this report, the gene encoding CmEpiA is located 5′ to the gene encoding CmMan5A which releases mannose from oligo- and polysaccharides (Diaset alet al, 2004). This observation and the considerable homology established between the enzyme and other phosphomannose isomerases (see above) supports the proposed function of CmEpiA. The assay consisted of a coupled reaction with two enzymes and used mannose as the substrate. The data revealed that purified CmEpiA (Fig. 3b) expresses no phosphomannose activity. Under the same conditions the assay correctly determined the activity of a commercial phosphmannose isomerase, demonstrating that the enzyme system used was valid (results not shown). In addition, CmEpiA was unable to convert mannose into mannose 6-phosphate, suggesting that the enzyme also does not exhibit mannokinase activity. Finally, the ability of CmEpiA to function as an epimerase was evaluated. The data revealed that CmEpiA displays epimerase activity, converting mannose into glucose. No epimerase activity was detected on N-acyl-glucosamine or mannose 6-phosphate (not shown). The optimum pH of the enzyme is 7.5 and the temperature at which it displays maximum catalytic activity is 40°C (Figs 4a and b). The kinetic parameters of CmEpiA were determined at 37°C, revealing values of 11.2±0.3min−1 and 53.5±4.9 mM for kcat and Km, respectively, during the conversion of mannose to glucose (Fig. 4c). The activity of the epimerase is not affected by the presence of EDTA and various cations, such as Ca2+, Zn2+, Na+, Mg2+, K+ and Mn2+, suggesting that the enzyme does not have any metal ion requirement.

Discussion

This report describes the cloning and characterization of a Cellvibrio mixtus gene cluster encoding enzymes involved in the degradation of galactomannan and in the metabolism of its hydrolysis products. It is interesting to note that three of the four genes identified are very tightly linked and encode, respectively, a protein of unknown function, a mannose epimerase (CmEpiA) and an exo-β-mannosidase (CmMan5A). Recently we showed that CmMan5A is a cell bound family 5 GH that releases mannose from the nonreducing end of both oligo- and polysaccharides (Diaset alet al, 2004). The fourth gene identified here encodes a family 27 α-galactosidase, termed CmAga27A, that is located at the 3′ end of, and in the same orientation to, the other three genes.

Based on substrate specificities, α-galactosidases have been classified into two groups (Deyet alet al, 1993). One group primarily removes galactose from oligosaccharides such as melibiose, raffinose and staquiose. These enzymes have potential biotechnological application in reducing the detrimental effects associated with the ingestion of galactose containing oligosaccharides by monogastric animals (Jindouet alet al, 2002). These oligosaccharide specific enzymes are located in GH families 4 and 36. The second group of α-galactosidases acts preferentially on large macromolecules, such as galactomannans, glycoproteins and glycolipids, while retaining the capacity to hydrolyse low-Mr substrates. These enzymes are exclusive to GH family 27 (Garman & Garboczi, 2004). The biochemical properties of CmAga27A, a family 27 GH highly homologous to the eukaryotic enzymes, is unusual; the enzyme attacks galactomannan but not oligosaccharides containing terminal α-d-galactose residues. Similarly, the homologous α-galactosidase from Clostridium josui, Aga27A, was also shown to prefer galactomannan rather than small substrates (Jindouet alet al, 2002). It is unknown if this property is shared by the other enzymes of the bacterial cluster of GH27, although Cellvibrio japonicusα-galactosidase CjAga27A displayed some activity against raffinose and stachyose (Halsteadet alet al, 2000).

The substrate specificity of CmAga27A might reflect the scarcity of oligosaccharides of the raffinose family on the milieu of Cellvibrio mixtus. The presence of a typical signal peptide in the primary sequence of CmAga27A suggests that this enzyme is extracellular. This is consistent with the complete resistance of the enzyme to proteolytic inactivation, a property typical of most extra-cellular plant cell wall polysaccharidases that must retain stability within an highly proteolytic environment (Fonteset alet al, 1995). Taken together the data suggest that CmAga27A acts cooperatively with the extracellular repertoire of endo-mannanses releasing galactose and manno-oligosaccharides from galactomannans, which are then absorbed by Cellvibrio mixtus. Interestingly, CmAga27A does not possess an associated carbohydrate binding module typical of GH acting on complex polysaccharidases. Analysis of the molecular architectures of family 27 enzymes demonstrates that none of the enzymes display a modular architecture, suggesting that removal of galactose from complex macromolecules does not require the presence of carbohydrate binding modules. Ultimately, this might reflect the high solubility presented by the decorated galactomannans and/or the relative accessibility of the galactose side chains in the complex polysaccharide.

In contrast to CmAga27A, the lack of a signal peptide in the primary sequence of CmEpiA might reflect its intracellular location. It is generally assumed that mannose metabolism in bacteria follows its phosphorylation and transport into the cytoplasm by a transmembrane sugar kinase/transporter with the subsequent isomerization to fructose 6-phosphate through the action of a phosphomanose isomerase. The data presented here suggest that an alternative metabolic pathway for the utilization of mannose might exist in Cellvibrio mixtus. Intracellular nonphosphorylated mannose molecules can be directly converted into glucose through the action of CmEpiA. It remains to be established whether both metabolic pathways coexist in Cellvibrio mixtus and, if so, which one predominates. As genes encoding CmEpiA and CmMan5A are tightly linked, it is tempting to speculate that the product generated by CmMan5A, mannose, serves as substrate for CmEpiA. This observation suggests that CmEpiA might be a fundamental enzyme in the metabolism of mannose in aerobic soil bacteria. Indeed, support for this view is provided by the genome sequence of Cellvibrio japonicus, which reveals an identical gene arrangement to Cellvibrio mixtus within the manA locus. Cellvibrio japonicus contains a similar tight cluster of three genes encoding CjUnkA, CjEpiA and CjMan5A, which display 86%, 73% and 54% sequence identity with their Cellvibrio mixtus counterparts. Downstream of this locus is the gene encoding CjAga27A, which exhibits 83% sequence identity with the Cellvibrio mixtusα-galactosidase. Although it might be tempting to speculate that CjUnkA and CmUnkA mediate transport of mannose across the inner membrane, this appears unlikely as the proteins display no membrane-spanning motifs. Interestingly, however, immediately upstream of the Cellvibrio japonicus man5A-containing genetic cluster, is a gene whose product displays significant sequence identity to sodium-dependent sugar transporters, and thus may mediate the trafficking of mannose across the inner membrane. Cellvibrio japonicus also contains genes encoding the three key enzymes of the Leloir pathway, galactokinase (galK), galactose 1-phosphate uridylyltransferase (galT) and UDP-galactose 4-epimerase (galE), which catalyse the conversion of galactose to the more metabolically useful sugar glucose 1-phosphate. Galactokinase and galactose 1-phosphate uridylyltransferase are encoded by a two-gene operon, while galE is not linked to galK or galT. The genes encoding the Leloir pathway are not linked to the man5A locus. As Cellvibrio mixtus and Cellvibrio japonicus are able to release galactose and mannose from galactomannan through the same complement of enzyme activities, it is highly likely that both Cellvibrio strains will also contain the genes encoding the Leloir pathway.

Collectively, the data presented here and in other reports suggest that galactomannan hydrolysis by Cellvibrio results from the synergistic interaction of a large repertoire of enzymes. Family 5 mannanases and CmAga27A are secreted into the extracellular milieu where the mannose-containing hemicelluloses are located (Hogget alet al, 2003). The action of CmAga27A exposes the mannan backbone to modular family 5 mannanases. Membrane bound family 26 mannanses and CmMan5A further degrade the generated manno-oligosaccharides and soluble galactomannans into mannose and mannobiose (Halsteadet alet al, 2000). These hydrolysis products, which are generated at the surface of the organism, can be readily taken up by Cellvibrio mixtus. The absorbed mannose can be converted to glucose in the cytoplasm through the action of CmEpiA.

Acknowledgements

This work was supported by grants SFRH/BD/16731/2004 (to C.I.P.D.G.), PraxisXXI/BD/21250/1999 (to F.M.V.D), and POCI/CVT/61162/2004 from the Fundação para a Ciência e a Tecnologia, Portugal.

References

View Abstract