Identification and Characterization of a Novel Galactofuranose-Specific β-D-Galactofuranosidase from Streptomyces Species

β-D-galactofuranose (Galf) is a component of polysaccharides and glycoconjugates and its transferase has been well analyzed. However, no β-D-galactofuranosidase (Galf-ase) gene has been identified in any organism. To search for a Galf-ase gene we screened soil samples and discovered a strain, identified as a Streptomyces species by the 16S ribosomal RNA gene analysis, that exhibits Galf-ase activity for 4-nitrophenyl β-D-galactofuranoside (pNP-β-D-Galf) in culture supernatants. By draft genome sequencing of the strain, named JHA19, we found four candidate genes encoding Galf-ases. Using recombinant proteins expressed in Escherichia coli, we found that three out of four candidates displayed the activity of not only Galf-ase but also α-L-arabinofuranosidase (Araf-ase), whereas the other one showed only the Galf-ase activity. This novel Galf-specific hydrolase is encoded by ORF1110 and has an optimum pH of 5.5 and a Km of 4.4 mM for the substrate pNP-β-D-Galf. In addition, this enzyme was able to release galactose residue from galactomannan prepared from the filamentous fungus Aspergillus fumigatus, suggesting that natural polysaccharides could be also substrates. By the BLAST search using the amino acid sequence of ORF1110 Galf-ase, we found that there are homolog genes in both prokaryotes and eukaryotes, indicating that Galf-specific Galf-ases widely exist in microorganisms.


Enzyme assay
Galf-ase and Araf-ase activity was determined using pNP-β-D-Galf or pNP-α-L-Araf as a substrate, respectively. The enzyme solution was prepared in 45 μL, which was mixed with 2.5 μL of 10 mM substrate and 2.5 μL of 1 M acetate buffer, pH 4.5. After incubation for the appropriate time at 37°C, 50 μL of 1 M sodium carbonate was added to terminate the reaction, and the liberated pNP was determined from absorbance at 405 nm. One unit (U) of enzyme activity was defined as the amount of enzyme required to liberate 1 mmol of pNP per min [24,28]. The activity of exoglycosidases was assessed using appropriate pNP-glycosides (α-D-Xyl and β-D-Xyl from Seikagaku; the others from Sigma).

Preparation of genomic DNA
Genomic DNA of strain JHA19 was extracted as described previously with certain modifications [41]. After culture in 100 mL YMG medium at 30°C for 1 week, the culture of the strain JHA19 was centrifuged at 5000 rpm for 15 min and the cell pellet was resuspended in 5 mL TE 10 (10 mM Tris-HCl, 10 mM EDTA, pH 8.0). After addition of 10 mg lysozyme (Wako) and 10 mg achromopeptidase (Wako), the cell suspension was incubated at 37°C for 20 min. The resultant sample was added with 100 μL TE 10 , 2.5 ml EDTA (0.5 M, pH 8.0), 1.25 mL 10% (w/ v) SDS and 125 μL proteinase K (20 mg/mL) (Wako) and incubated overnight at 37°C. After another incubation at 65°C for 5 min, 20 mL TE 10 was added. Ten mL of the resultant sample was taken and mixed with 20 mL TE 10 , 2 mL 3 M sodium acetate and 20 mL phenol/chloroform (1:1, v/v) by gently rotating for 30 min. After centrifugation at 4500 rpm for 20 min, the aqueous phase was divided into two tubes. Each tube was added with 2.5 volume 100% ethanol and centrifuged at 4500 rpm for 10 min. The pellet was dried and suspended in 10 mL TE. Those two genomic DNA suspension tubes were combined into one tube, which was added with 10 μL RNase (10 mg/mL) and incubated at 37°C for 30 min. The resultant sample was added with 200 μL 10% (w/v) SDS and 50 μL proteinase K (10 mg/mL) and incubated at 55°C for 1 h. After addition of 2 mL 3 M sodium acetate, the sample was mixed with 20 mL phenol/ chloroform by gently rotating for 30 min. After centrifugation at 4500 rpm for 20 min, the aqueous phase was divided into a few tubes. Each tube was added with 3 volume 100% ethanol and centrifuged at 4500 rpm for 20 min, and the pellet was dried and suspended in 1 mL TE, which was used as the genomic DNA sample.
16S ribosomal RNA gene analysis 16S rRNA gene sequence was amplified by PCR from the genomic DNA sample of strain JHA19 using universal primers listed in S1 Table. The DNA sequence of the PCR product was applied to a BLAST search, and the strain species was identified.

Whole-genome sequencing analysis
Whole-genome shotgun sequencing of the strain JHA19 was conducted using an FLX454 sequencer (Illumina). As a result, 252 Mbp was generated from 6x10 5 sequencing reads, which gave 32.7 fold-coverage. For sequence assembling, the program Newbler version 2.7 was used, and 70 contigs were generated. The genome annotation was performed with both Glimmer version 3.02b and BLAST 2.2.26. More detailed information will be presented elsewhere.

Preparation of recombinant Galf-ase proteins
To construct recombinant expression plasmids, four candidate Galf-ase genes were amplified by PCR using the DNA polymerase PrimeStarGXL (Takara), primers shown in S1 Table and genomic DNA of JHA19 as a template. An EcoRI digested pET50b vector and amplified DNA were ligated with In-Fusion HD Cloning Kit (Takara).
Escherichia coli BL21(DE3)CodonPlus strain transformed with each Galf-ase expression plasmid was precultured in LB medium (Miller, Merck) at 37°C overnight. OD 600 of cells was adjusted to 0.05 and cultured until OD 600 = 0.8, added with 100 mM IPTG and cultured overnight at 15°C. Cells were centrifuged at 7000 rpm for 7 min, resuspended in 5 mL 20 mM MOPS (pH 8.0) and lysed by ultrasonication on ice. The cell lysates were centrifuged at 15000 rpm for 10 min at 4°C and the supernatants were applied to a HisTrapTM FF 1 mL column (GE Healthcare). Recombinant protein purification was performed according to the manufacturer's instructions.

Preparation of galactomannan from Aspergillus fumigatus
Galactomannan (GM) was prepared from A. fumigatus essentially as described previously with some modifications [27]. Conidia were harvested from a plate of minimal medium (1% glucose, 0.6% NaNO 3 , 0.052% KCl, 0.052% MgSO 4 ･7H 2 O, 0.152% KH 2 PO 4 , biotin (trace) and Hunter's trace elements, pH 6.5), where the A. fumigatus A1163 (CEA10) strain was grown at 37°C for 3 days. The collected conidia were inoculated in a 500 mL Sakaguchi flask with 100 mL YNB medium supplemented with galactose (YNBG medium; 0.67% yeast nitrogen base, 0.5% (NH 4 ) 2 SO 4 , 9% galactose) and precultured at 37°C for 24 h. The preculture was transfered in a 5 L round-bottom flask with 1 L YNBG medium and cultured at 37°C for 14 days. Thereafter, cells from 4.4 L culture were added with formaldehyde at a final concentration of 1% and left for 24 h. After centrifugation, the supernatant was dialyzed with water for 3 days, then evaporated and lyophilized. The resultant sample was dissolved in 5 mL 20 mM phosphate buffer (pH 7.0), applied to TOYOPEARL DEAE-650 (TOSOH) and sequentially eluted with water, 0.5 M and 1 M NaCl solutions in 20 mM phosphate buffer (pH 7.0). The water eluate was dialyzed with 10 mM and 5 mM phosphate buffer (pH 7.0) and water overnight, for 6 h and 1 h, respectively. The resultant solution was evaporated and lyophilized, then used as the GM sample.

TLC analysis
N-terminal tags (2xHis 6 and Nus) were cleaved off the recombinant ORF1110 protein using HRV3C protease (Novagen) and removed by chromatography on a HisTrapTM FF 1 mL column. The flow-through sample was concentrated to 22.5 μL (7.6 mU/μL) and incubated with 25 μL GM (1 mg/μL) and 2.5 μL acetate buffer (1 M, pH 4.5) at 37°C for 24 h. The sample was then separated by TLC using a TLC Silica gel 60 plate (Millipore) and 1-butanol/ethanol/water (2:1:1, v/v/v) as solvent. For detection the TLC plate was sprayed with 0.2% orcinol and 10% methanol/sulfuric acid and baked at 120°C for 10 min.

ELISA
To analyze Galf-ase activity of the ORF1110 protein by ELISA, Platelia Aspergillus Ag EIA Kit (Bio-Rad) was used according to the manufacturer's instructions. Briefly, 50 μL of positive control containing GM, 0.5 μL of 7.5 mU ORF1110 Galf-ase and 1 μL of acetate buffer (1 M, pH 4.5) were mixed in a total volume of 100 μL and incubated at 37°C for 0, 1, 3 or 6 h. The resultant samples were diluted four times and their absorbance at 450 nm was measured.

Identification of a soil microorganism that exhibits Galf-ase activity
To search for a Galf-specific Galf-ase, we isolated 282 bacterial strains, mainly actinomycetes, from soil samples. Culture supernatants of three isolated strains, named JHA19, JHA26 and EMA216, exhibited Galf-ase activity using pNP-β-D-Galf as a substrate. In addition to the Galf-ase activity, we detected the activities of β-galactosidase (pyranose form), α-mannosidase, β-N-acetylgalactosaminidase and β-N-acetylglucosaminidase from the culture supernatant of JHA19 using the corresponding pNP-glycosides as substrates. Since the activity of Galf-ase was higher than that of Araf-ase, which was hardly detected in the culture supernatant of JHA19, it suggested that this strain might harbor enzyme(s) specific for Galf-ase. Therefore, we chose strain JHA19 for further enzymatic characterization.
Strain JHA19 displayed filamentous growth on a plate ( Fig 1A) and appeared like a Grampositive and bacillary bacterium (Fig 1B), suggesting that it belongs to the Streptomyces species. To further identify this strain, we performed a BLAST search based on the 16S rRNA gene sequence, and found that it shows 99% identity to Streptomyces coelicolor, S. albogriseolus, S. tendae, S. ambofaciens and S. lividans (Fig 1C). This result clearly demonstrated that strain JHA19 belongs to the Streptomyces species.

Exploration of candidate Galf-ase genes in strain JHA19
To search for genes encoding Galf-ases, we conducted a whole-genome shotgun sequencing of strain JHA19. We determined most of the genome sequence, the details of which will be reported elsewhere. We searched the sequence for ORFs that showed high sequence similarity to known furanosidase genes and found four Galf-ase candidates named ORF0232, ORF1110, ORF2125 and ORF2812 (Fig 2; Table 1). Based on a domain search using Pfam, we predicted that ORF0232, ORF2125 and ORF2812 may have Araf-ase activity because they show the highest similarity to reported Araf-ases. Indeed, the ORF0232 protein includes glycosyl hydrolases family 62 domain whose known activity is Araf-ase, the ORF2125 protein contains an Araf-ase C-terminus domain and the ORF2812 protein also has an Araf-ase B domain (AbfB), which is typically seen in GH54 Araf-ases [31,37,43]. Furthermore, a BLAST search revealed that ORF1110 has the highest similarity to a gene encoding an uncharacterized GH2 family protein which contains an AbfB domain based on the program CAT. Therefore, we further analyzed these four candidate genes, including ORF1110.

Enzymatic activities of recombinant proteins
We introduced ORF0232, ORF1110, ORF2125 and ORF2812 sequences into an E. coli expression vector lacking lacZ to circumvent a potential risk of contamination of subsequent enzymatic assays by β-galactosidase. The recombinant proteins were expressed and purified by a Ni affinity column. We first confirmed that samples from E. coli cells harboring an empty vector had no enzymatic activity for pNP-α-L-Araf nor pNP-β-D-Galf (data not shown). Recombinant proteins expressed from ORF0232, ORF2125 and ORF2812 showed Araf-ase activity for pNP-α-L-Araf as a substrate, like their homologs (Fig 3A, 3C and 3D). In addition, we measured the ratio of the activity of Araf-ase to Galf-ase, and found that ORF0232, ORF2125 and ORF2812 proteins exhibited the activity for both Araf-ase and Galf-ase. AbfA and AbfB in A. niger also showed both Araf-ase and Galf-ase activities, but the activity of Galf-ase was 10-fold less than that of Araf-ase, unlike proteins of ORF0232, ORF2125 and ORF2812 [30]. Although homologs of ORF0232, ORF2125 and ORF2812 are reported as Araf-ases, these recombinant proteins also displayed the Galf-ase activity, suggesting that enzymes reported as Araf-ases might generally exhibit the Galf-ase activity. In contrast, the recombinant protein of ORF1110 exhibited Galf-ase activity only, but not Araf-ase activity, suggesting that this GH2 family protein is a Galf-specific Galf-ase ( Fig 3B). Thus, we focused on examining chemoenzymatic characteristics of the ORF1110 protein.

Chemoenzymatic properties of ORF1110 encoded Galf-ase
To determine the substrate specificity of the recombinant ORF1110 protein, we measured hydrolytic activity using a variety of pNP-glycosides in their pyranose form (β-D-Gal, α-   No activity was observed with any of these substrates, except with pNP-β-D-Galf, confirming that this enzyme specifically hydrolyzes β-D-Galf. The optimum pH for ORF1110 Galf-ase activity was found to be 5.5 (Fig 4A). The thermal stability of the enzyme was examined by heating it at various temperatures for 10 min. The enzyme was found to be stable at temperatures up to 40°C.
The activity of the Araf-ase TtAFase belonging to the GH2 family in Thermotoga thermarum was reported to be highly inhibited by addition of either Cu 2+ or Zn 2+ [44]. Hence, we investigated the effects of metal ions (at a concentration of 5 mM) on the Galf-ase activity of ORF1110. We found that the ORF1110 Galf-ase activity was mostly inactivated by addition of Cu 2+ , Zn 2+ and EDTA to 6.5%, 49% and 55% of its original activity, respectively.
Next, we examined the effect of the substrate pNP-β-D-Galf concentration on the initial velocity of the enzyme reaction. The apparent Km and Vmax were 4.4 mM and 0.35 mM/min, respectively. Even though this protein does not have Araf-ase activity, it exhibited competitive inhibition by L-arabino-1,4-lactone, an Araf-ase inhibitor (Ki, 51 mM) (Fig 4B) [45]. This suggests that there may be different substrate recognition mechanism between Galf-ase and Arafase at the active site.

Crucial amino acid residues of ORF1110 Galf-ase
The recombinant ORF1110 protein lacking an AbfB domain exhibited almost the same Galfase activity as the full-length protein, suggesting that this domain is likely not required for the Galf-ase activity (data not shown).
Since the protein encoded by ORF1110 shows low similarity to well-known Araf-ases, it is not possible to predict which amino acid residues are crucial for its enzymatic activity by sequence comparison. Thus, we first used a sequence alignment of the GH2 family proteins that show higher sequence similarities to the ORF1110 Galf-ase (Fig 5). The sequence alignment revealed a number of conserved aspartic acid and glutamic acid residues. Using site directed mutagenesis we individually changed each conserved residues to alanine in ORF1110 and measured the effect of the mutations on Galf-ase activity. Most mutations had an effect on Galf-ase activity with D423A and E464A having the most drastic effect, suggesting that the glycosyl hydrolase 2C domain is the catalytic center of this enzyme, and that several amino acid residues are involoved ( Table 2). ORF1110 Galf-ase can hydrolyze Aspergillus fumigatus GM Lastly, we tested whether ORF1110 Galf-ase could catalyze not only the artificial substrate pNP-β-D-Galf but also a natural Galf-containing oligosaccharide. β-D-Galf exists in glycan parts at the cell surface of Aspergillus species. Thus, we extracted GM, including Galf chains, from A. fumigatus strain A1163 (CEA10) and analyzed Galf-ase activity by TLC (Fig 6A). The results indicated that Gal was released from the GM sample, suggesting that ORF1110 Galf-ase can hydrolyze a natural GM oligosaccharide from A. fumigatus (Fig 6B).

Discussion
In this study, we have isolated a strain of Streptomyces which possesses Galf-specific Galf-ase encoded by ORF1110. To our best knowledge, this is the first report about Galf-specific Galfase that does not also exhibit Araf-ase activity. Since we found that the ORF1110 Galf-ase belongs to GH2, which generally has a β-D-galactosidase activity, we examined hydrolase activity of the ORF1110 enzyme towards pNP-β-D-galactopyranoside. However, no activity was detected, suggesting that the ORF1110 enzyme activity is specific to furanose substrates.
BLAST search suggested that ORF1110 protein-like Galf-ases exist in a wide range of organisms from bacteria to eukaryotes (Fig 7). We cloned an ORF1110 homologous gene in Streptomyces griseus and confirmed that the derived recombinant protein exhibited Galf-specific Galf-ase activity (unpublished data). In Aspergillus species, there are also genes corresponding to ORF1110. Since the Galf biosynthetic pathway is important for their hyphal growth, Galfdegradation and metabolism pathways regulated by Galf-specific Galf-ase would be also crucial for fungal physiology. However, little is known about molecular mechanisms of Galf-degradation and metabolism. Therefore, it would be interesting to investigate the physiological functions of genes encoding Galf-specific Galf-ases in Aspergillus species. It was reported that Araf-ases AbfA and AbfB in A. niger, belonging to the GH51 and GH54, respectively, exhibit activities of both Araf-ase and Galf-ase [30]. Although both pNP-β-D-Galf and pNP-α-L-Araf are recognized as substrates by AbfB, affinity for pNP-β-D-Galf is The recombinant ORF1110 Galf-ase was mixed with GM prepared from A. fumigatus and incubated at 37°C for 24 h. The resultant sample was subjected to TLC analysis. As references, galactopyranose (Galp) and lactose (Lac) were spotted. As a negative control, a sample without recombinant ORF1110 Galf-ase was spotted. GM, galactomannan. (B) A schematic diagram of the predicted partial structure of A. fumigatus GM proposed previously [7]. The ORF1110 Galf-ase seems to catalyze terminal Galf residues of A. fumigatus GM. lower resulting in less Galf-ase activity compared to Araf-ase. Considering that the ORF1110 protein exhibits only Galf-ase activity, almost no Araf-ase activity, and shows the competitive inhibition by L-arabino-1,4-lactone, the C6 atom of Galf in the substrate pNP-β-D-Galf appears to be crucial in the hydrogen bonding required for the proper positioning of the substrate on the catalytic site. pNP-α-L-Araf, which structure is similar to that of pNP-β-D-Galf, would enter the active site of the ORF1110 Galf-ase, but pNP-α-L-Araf may exhibit less hydrogen bonding due to lack of the C6 atom, resulting in lower activity of Araf-ase than Galf-ase. The structural analysis of the ORF1110 Galf-ase will be required to reveal the details of the catalytic mechanism.
We confirmed the Galf-ase activity of the ORF1110 protein for A. fumigatus GM in two ways: One was by detecting non reducing terminal Galf by TLC analysis, and the other was by observing a 40% reduction in ELISA assays using EB-A2 antibody (data not shown). Although we could not find to analyze as a candidate Galf-ase, we found another putative hydrolase gene adjacent to ORF1110 in the JHA19 genome. This predicted hydrolase belongs to GH2 and contains signal peptide like the ORF1110 Galf-ase. These information suggests that this putative hydrolase might be simultaneously expressed with ORF1110 to function together with the ORF1110 Galf-ase. Using A. fumigatus GM as a substrate, it was shown that the culture supernatant of A. fumigatus, unlike ORF1110 Galf-ase, produced several bands on TLC, suggesting that there might be not only exo-Galf-ase but also endo-Galf-ase activity in the fungus [27]. In addition, a detailed structural analysis of the sugar chain on glycoproteins demonstrated that β-1,2and β-1,6-linked Galf, except for β-1,5-linked Galf, also exist [27]. Further work will be needed to determine which linkage of Galf is hydrolyzed by ORF1110 Galf-ase.
In conclusion, we have characterized a novel Galf-specific Galf-ase encoded by ORF1110 in strain JHA19. Considering that ORF1110 Galf-ase homologs are widely present and Galf residues are present on the cell surface of pathogenic microbes such as A. fumigatus, it is crucial to further understand the molecular mechanisms driving Galf-catalyzing enzymes for establishing novel pharmaceutical therapy against fungal pathogens. Supporting Information S1