Characterization of an Endo-β-1,6-Galactanase from Streptomyces avermitilis NBRC14893

Abstract

The putative endo-β-1,6-galactanase gene from Streptomyces avermitilis was cloned and expressed in Escherichia coli, and the enzymatic properties of the recombinant enzyme were characterized. The gene consisted of a 1,476-bp open reading frame and encoded a 491-amino-acid protein, comprising an N-terminal secretion signal sequence and glycoside hydrolase family 5 catalytic module. The recombinant enzyme, Sa1,6Gal5A, catalyzed the hydrolysis of β-1,6-linked galactosyl linkages of oligosaccharides and polysaccharides. The enzyme produced galactose and a range of β-1,6-linked galacto-oligosaccharides, predominantly β-1,6-galactobiose, from β-1,6-galactan chains. There was a synergistic effect between the enzyme and Sa1,3Gal43A in degrading tomato arabinogalactan proteins. These results suggest that Sa1,6Gal5A is the first identified endo-β-1,6-galactanase from a prokaryote.

Actinomycetes are among the most numerous and ubiquitous soil bacteria. In addition to their broad range of metabolic abilities, they are important for carbon recycling in soil environments. These gram-positive bacteria are characterized by their complex morphological differentiation, resembling that of filamentous fungi, and the ability to produce a wide variety of secondary metabolites (3). Actinomycetes have been previously described as rhizosphere-colonizing bacteria (20). Plant root exudates stimulate rhizosphere growth of actinomycetes that are strongly antagonistic to fungal pathogens, while the actinomycetes utilize root exudates for growth and synthesis of antimicrobial substances (4, 27).

Arabinogalactan proteins (AGPs) are found in tree exudate gums and coniferous woods (1). In addition to the complexity of protein components, the AGP molecule contains complex branched glycans (15, 19, 21), which are 90 to 99% of the AGP mass. Type II arabinogalactan and short oligoarabinosides are the two types of carbohydrate attached to the AGP backbone. Type II arabinogalactans have β-1,3-linked galactosyl backbones (sugars in the present study are d series unless otherwise designated) in mono- or oligo-β-1,6-galactosyl and/or -arabinosyl side chains (6, 7). l-Arabinose and lesser amounts of other auxiliary sugars, such as glucuronic acid, l-rhamnose, and l-fucose, are attached to the side chains primarily at nonreducing termini (6). Glycoside hydrolases capable of degrading the carbohydrate moieties of AGPs would be a significant advance; however, there has been limited research on such enzymes.

Two kinds of galactanases, exo-β-1,3-galactanase from Phanerochaete chrysosporium (Pc1,3Gal43A) and endo-β-1,6-galactanase from Trichoderma viride (Tv1,6Gal5A, formerly Tv6GAL) were cloned for the first time (12, 16). Exo-β-1,3-galactanases degrade β-1,3-galactan backbones of AGPs and belong to glycoside hydrolase (GH) family 43 (GH43) (Carbohydrate Active Enzymes database, http://www.cazy.org [8-10]). This enzyme was also cloned from Clostridium thermocellum and Arabidopsis thaliana, demonstrating a wide distribution including fungi, bacteria, and plants (13). In contrast, an enzyme that hydrolyzes β-1,6-galactan side chains of AGPs (endo-β-1,6-galactanase, GH5) was obtained from T. viride (16, 22), and the amino acid sequence of Tv1,6Gal5A was classified as a new member of GH5. Tv1,6Gal5A specifically hydrolyzed the β-1,6-galactan chain and produced mainly β-1,6-galactobiose together with different degrees of polymerization (dp) of β-1,6-galacto-oligosaccharides. A similar enzyme was purified from Aspergillus niger (2).

In the present work, we focused on endo-β-1,6-galactanase because there are no reports of the enzyme produced by organisms other than fungi. The BLAST search using the amino acid sequence of Tv1,6Gal5A showed similarity with a hypothetical protein of Streptomyces avermitilis (GenBank accession number BAC72917; 54% identity and 69% similarity). We cloned the putative endo-β-1,6-galactanase gene from S. avermitilis to test the recombinant protein for endo-β-1,6-galactanase activity. In addition, the synergistic effect of endo-β-1,6-galactanase and exo-β-1,3-galactanase in AGP degradation was investigated. This is the first report of endo-β-1,6-galactanase originating from a prokaryote.

MATERIALS AND METHODS

Substrates.

β-1,3-β-1,6-Galactan from Prototheca zopfii, β-1,3-galactan, native and α-l-arabinofuranosidase-treated AGPs from radishes, β-1,3- and β-1,6-linked galacto-oligosaccharides, β-1,4-galactotriose, p-nitrophenyl (PNP)-α-l-Araf, and arabinan were prepared as described previously (12-14, 22). Methyl β-glycosides of β-1,3-linked galactotetraose and -pentaose and β-1,6-linked galactopentaose and -hexaose were a kind gift from P. Kováč of the National Institutes of Health (National Institute of Diabetes and Digestive and Kidney Diseases, Bethesda, MD). Debranched arabinan, pectic galactan from lupine, and lupine galactan were from Megazyme. Gum arabic was from Nacalai Tesque Inc. (Kyoto, Japan). Pustulan from Umbilicaria papullosa was from Calbiochem (Darmstadt, Germany). Carboxymethyl curdlan (CM-curdlan) was from Wako Pure Chemical Industries, Ltd. (Osaka, Japan). Chitosan from crab shells, lichenan, PNP-glycosides, β-1,4-galactobiose, larch arabinogalactan, laminarin, birchwood xylan, and oat spelt xylan were from Sigma (St. Louis, MO).

Purification of tomato AGPs.

Tomato juice (900 g) was purchased from a supermarket and lyophilized to decrease sample volume. The dried sample was dissolved in 20 ml of water, and proteins were precipitated by adjustment to 5% (wt/vol) trichloroacetic acid. After standing for 2 h at 4°C, the resultant precipitate was removed by diatomite filtration using Celite no. 500 (Wako Pure Chemical Industries Ltd.). The trichloroacetic acid in the filtrate was extracted three times with two volumes of diethyl ether. The aqueous phase (40 ml) was dialyzed against 10 mM ammonium formate buffer, pH 7.0, and loaded onto a DEAE-Sepharose Fast Flow (GE Healthcare UK Ltd., Little Chalfont, United Kingdom) column (5 mm by 300 mm). The bound materials were eluted with a linear gradient of ammonium formate (0.01 to 1 M). The fractions containing AGPs were detected by a single radial gel diffusion assay (26). The fractions (150 ml) were concentrated by evaporator and Ultrafree-MC (Millipore Corp., Bedford, MA). The resulting solution was then desalted on a PD-10 column (GE Healthcare UK Ltd.), and the eluate was used as purified tomato AGP.

Expression of recombinant endo-β-1,6-galactanase.

Streptomyces avermitilis NBRC14893 was obtained from the National Institute of Technology and Evaluation (Kazusa, Japan). The strain was grown on medium containing 1% gum arabic or larch arabinogalactan, 0.1% yeast extract, 0.1% peptone, 0.5% potassium dihydrogenphosphate, and 0.05% magnesium sulfate to confirm the endo-β-1,6-galactanase activity. Genomic DNA from S. avermitilis was prepared by using InstaGene Matrix (Bio-Rad Laboratories Inc., Hercules, CA) according to the manufacturer's instructions. The gene encoding a putative endo-β-1,6-galactanase (SAV_5205; GenBank accession number BAC72917) was amplified from S. avermitilis genomic DNA by PCR using Phusion DNA polymerase (Finnzymes, Espoo, Finland) and the following primers: forward, 5′-CAT ATG ACA GGT ACC GCA CGG GCC GAC-3′; reverse, 5′-AAG CTT TAC CGT CAC TCC GTC CAC CTC-3′. The amplified DNA was digested with NdeI and HindIII and cloned into pET30(+) (Novagen, Darmstadt, Germany) with a restriction enzyme site of NdeI-HindIII (underlined). Escherichia coli Rosetta (DE3) (Novagen) cells harboring the expression plasmid were cultured, and expression was induced with 0.1 mM isopropyl-β-thiogalactopyranoside for 24 h at 20°C. The recombinant enzyme (Sa1,6Gal5A) was purified with Ni-nitrilotriacetic acid-agarose (Qiagen GmbH, Hilden, Germany). The enzyme was detected by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. The relevant fractions were pooled and dialyzed against deionized water. The final preparation thus obtained was used as purified enzyme. The enzyme concentration was determined by measuring absorbance at 280 nm, assuming that an absorbance of 1.0 indicated a concentration of 1 mg/ml.

Enzyme assay.

Activity of endo-β-1,6-galactanase was determined by measuring the liberated reducing sugars, as galactose equivalents, using the Somogyi-Nelson method (24). The reaction mixture was 20 μl of McIlvaine buffer (0.2 M Na₂HPO₄, 0.1 M citric acid), pH 5.5, and 25 μl of 1% (wt/vol) β-1,3-β-1,6-galactan from P. zopfii. After 5 min of preincubation at 37°C, 5 μl of enzyme was added to the solution and the reaction mixture was incubated at 37°C for 10 min. The reaction was terminated by heating the solution at 100°C for 5 min. One unit of enzyme activity is defined as the amount of enzyme that released 1 μmol of galactose/min. The effects of pH and temperature on enzyme activity were investigated (12, 13).

Substrate specificity.

The substrate specificity was determined using various PNP-glycosides as substrates as described previously (12, 13). The substrate specificity for polysaccharides was determined at 37°C in McIlvaine buffer, pH 5.5, with 0.5% (wt/vol) polysaccharide as the substrate and 35 nM enzyme. After incubation for the appropriate reaction time, the liberated reducing sugars were measured by the Somogyi-Nelson method (24). The hydrolysis products of β-1,3-β-1,6-galactan from P. zopfii were determined by high-performance anion-exchange chromatography with a pulsed amperometric detection system (HPAEC-PAD; Dionex Corp., Sunnyvale, CA). The products were detected by comparing their retention times by HPAEC to those of standard oligosaccharides, such as β-1,6-galactobiose, β-1,6-galactotoriose, and β-1,6-galactotetraose.

The oligosaccharide substrate specificity and the catalytic efficiency were analyzed using β-1,6-, β-1,3-, and β-1,4-linked galactosyl oligosaccharides. Briefly, the enzyme (1 nM) was incubated with substrate (10 μM) in McIlvaine buffer, pH 5.5, at 37°C. At regular time intervals, the amount of degradation of each substrate was quantified by HPAEC-PAD (12). The assay was performed in triplicate.

The activity toward native AGP from tomato juice was assayed as follows. Sa1,6Gal5A (35 nM) and Sa1,3Gal43A (1.4 μM) were incubated individually with 0.5% (wt/vol) native AGP from tomato juice in McIlvaine buffer, pH 5.5, at 37°C. To evaluate possible synergistic effects between Sa1,6Gal5A and Sa1,3Gal43A, 1:1 mixtures containing the same amounts of both enzymes were incubated under the same conditions. The liberated reducing sugars, as galactose equivalents, were then measured by the Somogyi-Nelson method (24). The assay was performed in triplicate.

RESULTS

Expression and characterization of recombinant endo-β-1,6-galactanase from S. avermitilis.

The open reading frame SAV_5205 of S. avermitilis NBRC14893 (1,476 bp) encoded a putative protein (491 amino acids) (Fig. 1). In order of decreasing similarity, the sequence resembled the following: a hypothetical protein from Neurospora crassa OR74A (55% identity and 67% similarity; GenBank accession number EAA29705), a hypothetical protein from Stigmatella aurantiaca DW4/3-1 (49% identity and 62% similarity; EAU67098), and a hypothetical protein from Clostridium thermocellum ATCC 27405 (34% identity and 50% similarity; ABN53343). The sequence at the N terminus (amino acids 1 to 21) was predicted to be a signal sequence (SOSUI [http://bp.nuap.nagoya-u.ac.jp/sosui/]). The mature region of Sa1,6Gal5A was successfully expressed in E. coli, and 9.3 mg of purified recombinant enzyme was obtained from 1 liter of culture. The molecular mass of Sa1,6Gal5A by sodium dodecyl sulfate-polyacrylamide gel electrophoresis was estimated as 52 kDa (data not shown). When Sa1,6Gal5A was incubated with β-1,3-β-1,6-galactan from P. zopfii, the enzyme hydrolyzed the substrate and the specific activity for β-1,3-β-1,6-galactan was 56.5 units/mg. The enzyme achieved maximal activity at pH 5.5 and 40°C (data not shown).

FIG. 1.

Comparison of the amino acid sequences of β-1,6-galactanases. Shown is an alignment of the deduced protein sequences of T. viride Tv1,6Gal5A (GenBank accession number BAC84995), S. avermitilis Sa1,6Gal5A (BAC72917; present study), F. oxysporum FoGal1 (BAF42338), and A. niger (CAK38078) by using ClustalW. Residues identical between Tv1,6Gal5A and Sa1,6Gal5A and FoGal1 and CAK38078 are boxed. The amino acid residues indicated by asterisks are active-site residues in the GH5 family.

Substrate specificity of Sa1,6Gal5A.

The GH5 family includes exo-β-1,3-glucanase (EC 3.2.1.58), licheninase (EC 3.2.1.73), cellulase (EC 3.2.1.4), β-1,4-cellobiosidase (EC 3.2.1.91), endo-β-1,6-glucanase (EC 3.2.1.75), β-mannosidase (EC 3.2.1.25), endo-β-1,4-mannanase (EC 3.2.1.78), chitosanase (EC 3.2.1.132), endo-β-1,4-xylanase (EC 3.2.1.8), and endo-β-1,6-galactanase. Sa1,6Gal5A was tested for the activities reported for the GH5 family except for endo-β-1,6-galactanase. Sa1,6Gal5A did not hydrolyze laminarin (β-1,3-β-1,6-glucan), CM-curdlan (β-1,3-glucan), lichenan (β-1,3-β-1,4-glucan), CM-cellulose (β-1,4-glucan), pustulan (β-1,6-glucan), guar gum (galactomannan), locust bean gum (galactomannan), chitosan, soluble oat spelt xylan, soluble birchwood xylan, PNP-β-Galp, PNP-α-Galp, PNP-β-Glcp, PNP-α-Glcp, PNP-β-l-Arap, PNP-β-Xylp, PNP-β-Manp, PNP-α-l-Fucp, and PNP-β-Fucp (data not shown).

Subsequently, the activity of Sa1,6Gal5A toward the other polysaccharides containing galactan was investigated. The enzyme hydrolyzed substrates containing β-1,6-linked galactosyl residues such as β-1,3-β-1,6-galactan, native AGP from tomato juice, and native and α-l-arabinofuranosidase-treated AGPs from radishes. With the activity toward β-1,3-β-1,6-galactan from P. zopfii estimated as 100%, the relative activities for native AGP from tomato juice, native AGP from radishes, and α-l-arabinofuranosidase-treated radish AGP were 9%, 27%, and 356%, respectively. The enzyme did not hydrolyze any of the β-1,3- and β-1,4-linked galactose-containing polysaccharides such as β-1,3-galactan, pectic galactan (β-1,4-galactan), and lupine galactan (β-1,4-galactan) (data not shown). These results suggested that the enzyme cleaves only the β-1,6-linked galactan chain. This specificity is also apparent from the results of the activity toward oligosaccharides. Sa1,6Gal5A specifically hydrolyzed β-1,6-linked galacto-oligosaccharides but not β-1,3-linked and β-1,4-linked galacto-oligosaccharides (data not shown). When the enzyme hydrolyzed β-1,6-galactotriose, galactose and β-1,6-galactobiose were produced in the initial stage of the hydrolysis (data not shown). In the case of the hydrolysis of β-1,6-galactotetraose, galactose, β-1,6-galactobiose, and β-1,6-galactotriose were detected (data not shown). The relative activities for β-1,6-galactobiose, β-1,6-galactotriose, β-1.6-galactotetraose, methyl-β-1,6-galactopentaoside, and methyl-β-1,6-galactohexaoside were 0.4%, 11%, 45%, 100%, and 75%, respectively, when the highest activity was estimated as 100%. Thus, it was concluded that Sa1,6Gal5A cleaves only the β-1,6-galactosyl linkage.

Mode of action.

The products of Sa1,6Gal5A hydrolysis of β-1,3-β-1,6-galactan were analyzed by HPAEC-PAD (Fig. 2). After 10 min of incubation, the hydrolysis products generated were a series of β-1,6-galacto-oligosaccharides with different dp. With longer incubation, galactose was also detected with a range of β-1,6-galacto-oligosaccharides. The peak of galactobiose was the most prominent after both 10-min and 30-min incubations.

FIG. 2.

HPAEC-PAD analysis of hydrolysis products of β-1,3-β-1,6-galactan from P. zopfii generated by Sa1,6Gal5A. Sa1,6Gal5A was incubated with 0.5% (wt/vol) substrate in McIlvaine buffer, pH 5.5, at 37°C. After 0, 10, and 30 min of incubation, the reaction products were analyzed by HPAEC-PAD. Abbreviations: Gal, galactose; Gal₂, β-1,6-galactobiose; Gal₃, β-1,6-galactotriose; Gal₄, β-1,6-galactotetraose.

The catalytic efficiency of Sa1,6Gal5A against β-1,6-galacto-oligosaccharides with different dp was determined. The enzyme (1 nM) was incubated with 10 μM substrate in McIlvaine buffer, pH 5.5, at 37°C. The oligosaccharides were quantified periodically by HPAEC-PAD. Progress curves of oligosaccharides cleavage were used to determine k_cat/K_m. It was found that, for dp of 2, 3, 4, 5, and 6, k_cat/K_m values were 35 ± 3, 1,019 ± 17, 3,975 ± 56, 8,933 ± 307, and 6,688 ± 348 min⁻¹ mM⁻¹, respectively. The efficiency increased as dp increased from 2 to 5, with maximum activity at a dp of 5, indicating five major subsites of Sa1,6Gal5A. The results showed that the enzyme preferentially hydrolyzed galacto-oligosaccharides with dp of >3. This was also apparent in the degradation pattern of β-1,3-β-1,6-galactan (Fig. 2), with the enzyme preferentially liberating β-1,6-galactobiose. These results suggested that Sa1,6Gal5A was an endoacting enzyme.

Hydrolysis of AGPs.

Both Sa1,6Gal5A and Sa1,3Gal43A hydrolyzed native AGP from tomato juice. The amount of reducing sugars generated by Sa1,6Gal5A did not increase, even when concentration was increased 10-fold (0.35 μM). When mixtures containing both Sa1,6Gal5A (35 nM) and Sa1,3Gal43A (1.4 μM) were incubated with tomato AGP, the amount of reducing sugars was greater than the hypothetical sum of values from individual enzymes (Fig. 3). This suggested a synergistic effect of Sa1,6Gal5A and Sa1,3Gal43A. There was effective synergism in the initial hydrolysis stage.

FIG. 3.

Time course of tomato AGP degradation by Sa1,6Gal5A and Sa1,3Gal43A. Sa1,6Gal5A (35 nM) and Sa1,3Gal43A (1.4 μM) were incubated together or individually with 0.5% (wt/vol) tomato AGP in McIlvaine buffer, pH 5.5, at 37°C for up to 2 h. The reducing powers of reaction products were measured by the Somogyi-Nelson method. Symbols: ▪, Sa1,6Gal5A; •, Sa1,3Gal43A; ▴, Sa1,6Gal5A and Sa1,3Gal43A; ▵, hypothetical sum of Sa1,6Gal5A and Sa1,3Gal43A values generated by each enzyme.

DISCUSSION

Possible hydrolyzing activities to β-galactan are found in the enzyme families GH1, GH2, GH5, GH35, GH42, GH43, and GH53. Exo-β-1,3-galactanases belonging to GH43 can be distinguished from the other enzymes by their unique substrate specificity, hydrolyzing only the β-1,3 linkage of two d-galactosyl residues at the nonreducing ends of the substrates in an exoacting manner. The enzyme produces oligosaccharides together with galactose, because the enzyme can accommodate β-1,6-linked galactosyl side chains when catalyzing AGP hydrolysis. Endo-β-galactanases are classified into GH5 and GH53 and β-galactosidases into GH1, GH2, GH35, and GH42. A radish β-galactosidase belonging to GH35 can cleave both β-1,3 and β-1,6 linkages of galacto-oligosaccharides but not β-1,4 linkages of galacto-oligosaccharides (17). However, β-galactosidases of the other families prefer the β-1,4 linkage and do not hydrolyze β-1,3 and β-1,6 linkages of galacto-oligosaccharides. The galactanases of GH53 are endotype enzymes and prefer β-1,4-linked galactan. Endo-β-1,6-galactanase is distinguished from GH35 and GH53 enzymes by specifically hydrolyzing only β-1,6 linkages of galactan and galacto-oligosaccharide in an endoacting manner. An endo-β-1,6-galactanase from T. viride (Tv1,6Gal5A) that degrades β-1,6-galactan side chains of AGPs belongs to GH5 (16).

The GH5 family includes many kind of enzymes such as cellulase (EC 3.2.1.4), β-mannosidase (EC 3.2.1.25), endo-β-1,4-mannanase (EC 3.2.1.78), exo-β-1,3-glucanase (EC 3.2.1.58), endo-β-1,6-glucanase (EC 3.2.1.75), endo-β-1,4-xylanase (EC 3.2.1.8), licheninase (EC 3.2.1.73), chitosanase (EC 3.2.1.132), and β-1,4-cellobiosidase (EC 3.2.1.91). This family is classified into clan GH-A, which is the major clan of glycoside hydrolases, along with GH1, GH2, GH5, GH10, GH17, GH26, GH30, GH35, GH39, GH42, GH50, GH51, GH53, GH59, GH72, GH79, and GH86 (8-10). The clan GH-A enzymes have an inferred (β/α)₈-barrel structure. Catalysis involves a retaining mechanism, and their enzyme mechanism is known as retaining, owing to the double-displacement mechanism, which is catalyzed by two glutamic acid residues acting as an acid/base pair and a nucleophile (8-10). Structural analyses for GH5 enzymes, an exo-β-1,3-glucanase from Candida albicans (5), and a β-mannanase from Thermomonospora fusca (11), revealed that eight amino acids encompassing the catalytic site were conserved in GH5. Comparison of the deduced amino acid sequence of Sa1,6Gal5A with those of other GH5 members including Tv1,6Gal5A showed that eight amino acids were conserved: R87, H170, N216, E217 (acid/base), H292, Y294, E320 (nucleophile), and W350 (Fig. 1).

Enzymes termed endo-β-1,6-galactanases have been reported so far from A. niger, Fusarium oxysporum, and T. viride. The A. niger enzyme liberated galactose and β-1,6-galactobiose from acid-treated larch wood and Norway spruce arabinogalactans (18). When the enzyme hydrolyzed black liquor galactan I, galactose and β-1,6-galactobiose were the main products, with a small amount of arabinose. When arabinan was isolated from black liquor and used as a substrate, arabinose was the major product. The enzyme substrate specificity is quite similar to that for the enzyme from F. oxysporum (FoGal1) (23). When FoGal1 hydrolyzed larch arabinogalactan, predominantly β-1,6-galactobiose and small amounts of galactose and arabinose were produced. The hydrolysis products from larch arabinogalactan produced by FoGal1 were similar to the purified hydrolysis products of black liquor galactan produced by A. niger β-1,6-galactanase (18). In contrast, Sa1,6Gal5A cleaved only β-1,6-galactan and β-1,6-galacto-oligosaccharides and did not produce any l-arabinose, and the enzyme did not act on larch arabinogalactan by itself. The specificity was the same as that for the T. viride enzyme. Therefore, even though these enzymes are all classified in GH5, they can be distinguished by their substrate specificities. The N-terminal amino acid sequence of purified enzyme from A. niger was ISSSPLSTSGG(N)IVD (parentheses indicate uncertainty in interpretation). A BLAST search using the 15-amino-acid sequence showed similarity with an A. niger hypothetical protein (CAK38078). The resulting amino acid sequence had 63% similarity to that of FoGal1 (Fig. 1). In contrast, there was no significant similarity between these enzymes and Sa1,6Gal5A. Because the similarity between Sa1,6Gal5A and Tv1,6Gal5A was 69%, they were distinguishable by their amino acid sequences. β-1,6-Galactan-degrading activities of the enzymes from A. niger and FoGal1 were not presented, so it is unclear whether they cleaved β-1,6 linkages of arabinogalactan. Because GH5 includes both exo-β-1,3-glucanase and endo-β-1,6-glucanase, such variations may also occur for β-galactanases; this is supported by data in which exo-β-1,3-galactanases produced β-1,6-galacto-oligosaccharides from gum arabic and larch arabinogalactan (12-14, 25).

In the present study, we demonstrated the first prokaryotic endo-β-1,6-galactanase from S. avermitilis and showed that the strain possessed two kinds of galactanases, exo-β-1,3-galactanase and endo-β-1,6-galactanase. When Sa1,3Gal43A and Sa1,6Gal5A hydrolyzed native AGP from tomato juice, there was a synergistic effect of the two enzymes (Fig. 3). However, neither Sa1,3Gal43A nor Sa1,6Gal5A hydrolyzed gum arabic or larch arabinogalactan. These substrates are highly branched, and attached auxiliary sugars prevented access of the galactanases to galactan chains. When the distributions of the Streptomyces coelicolor and S. avermitilis enzymes that hydrolyze AGPs were compared, it was found that there were crucial differences in the enzymes that hydrolyze the galactan backbone. There were two kinds of enzymes in S. avermitilis, endo-β-1,6-galactanase and exo-β-1,3-galactanase; this is an advantage in the hydrolysis of AGPs. In S. coelicolor there was only β-galactosidase belonging to GH35. This was consistent with the lack of S. coelicolor growth with gum arabic as the sole carbon source, although S. avermitilis grew well (data not shown). These facts strongly suggest that endo-β-1,6-galactanase and exo-β-1,3-galactanase play significant roles in hydrolysis of the carbohydrate moieties of AGPs.