Introduction

Flavonoids are polyphenolic substances which are synthesised by a wide range of vascular plants (Iwashina 2000), acting as visual attractors, photoreceptors, feeding repellents, antimicrobials, and antioxidants (Pietta 2000). The flavonol quercetin (3,5,7,3′,4′-pentahydroxyflavone) is found in considerable amounts in vegetables and fruits like onions, broccoli, and apples (Duthie et al. 2003) and is an important constituent of the human diet. A number of health benefits are attributed to this plant polyphenol mainly due to its antioxidant properties (Pietta 2000; Prior 2003), and it is considered to be a potential anti-cancer agent (Lamson and Brignall 2000).

Microbial degradation of quercetin has been reported for fungi, e.g., Aspergillus spp., Pullularia sp., and Fusarium oxysporum (Westlake et al. 1961; Barz 1971) and for bacteria such as Pseudomonas putida, Rhizobium spp., and Actinoplanes missouriensis (Schultz et al. 1974; Rao and Cooper 1994; Rose and Fetzner 2006). Intestinal Clostridium species and Eubacterium ramulus (Winter et al. 1989; Schneider and Blaut 2000) as well as methanogenic consortia (Herrmann et al. 2001) anaerobically convert the flavonoid.

In the initial step of aerobic degradation, quercetinase catalyzes the 2,4-dioxygenolytic cleavage of quercetin to form 2-protocatechuoylphoroglucinol carboxylic acid and carbon monoxide. Quercetinases have so far been isolated from Aspergillus flavus (Oka et al. 1972), A. niger (Hund et al. 1999), A. japonicus (Fusetti et al. 2002), and Bacillus subtilis (Bowater et al. 2004; Barney et al. 2004), and the corresponding genes of A. japonicus and B. subtilis have been characterized (Kooter et al. 2002; Bowater et al. 2004). The crystal structures of A. japonicus and B. subtilis quercetinase revealed that both enzymes belong to the cupin superfamily (Fusetti et al. 2002; Gopal et al. 2005).

The cupin fold consists of a characteristic β-barrel domain (lat. cupa: small barrel), which comprises two conserved amino acid motifs G(X)5HXH(X)3,4E(X)6G and G(X)5PXG(X)2H(X)3N, separated by an inter-motif region that varies in length from 11 to >100 aa (amino acid) residues. Depending on the number of cupin domains present, the superfamily is subdivided into monocupins (single cupins), bicupins, and multicupins (Dunwell et al. 2000, 2001, 2004). The active site of enzymatic members is located in the centre of the β-barrel and includes four highly conserved residues (two His and Glu in motif 1, His in motif 2), which can bind different metal ions. As the homodimeric quercetinases of A. japonicus and B. subtilis exhibit two cupin domains per monomer, they are both counted among the bicupin subset. Whereas the fungal enzyme contains Cu2+ (Oka et al. 1972; Kooter et al. 2002), the Bacillus enzyme is presumed to prefer Mn2+ as cofactor (Schaab et al. 2006), despite having been purified as an iron enzyme from a recombinant E. coli clone (Bowater et al. 2004; Barney et al. 2004).

Streptomycetes are wide-spread soil bacteria which play an important role in the decomposition of biopolymers such as lignin, cellulose, hemicellulose, chitin, keratin, and pectin (Locci 1989). Lignin monomers and related compounds like trans-cinnamic acid, p-coumaric acid, ferulic acid, or vanillin are also utilized by members of this genus (Sutherland et al. 1983). Modification reactions of flavonoids, such as regiospecific hydroxylation and O-methylation, were described for several Streptomyces spp. (Hosny et al. 2001; Yoon et al. 2005; Kim et al. 2006), but knowledge on the potential of Streptomycetes to degrade flavonoids is very limited. In 1959, Westlake et al. (1959) reported that 24 out of 51 Streptomyces spp. tested were able to utilize rutin (quercetin 3-O-glycoside); however, the reactions and enzymes involved have not been investigated as yet.

Here, we report the detection of a CO-forming quercetinase in Streptomyces sp. strain FLA, the identification and heterologous expression of the corresponding gene designated queD (for quercetin dioxygenase), and the purification and characterization of recombinant QueD. In contrast to quercetinases from Bacillus subtilis and Aspergillus japonicus, QueD from this Streptomyces strain belongs to the monocupin family, which raises interesting questions about the evolution of bacterial and fungal quercetinases.

Materials and methods

Bacterial strains and plasmids, culture conditions, and preparation of Streptomyces protoplasts

Streptomyces sp. designated strain FLA (for flavonol utilization) was isolated from a soil sample collected near Stuttgart, Germany, by enrichment on quercetin agar. S. alboniger DSM 40043, S. aureofaciens ATCC 10762, S. avermitilis MA-4680 (DSM 46492), S. cinereoruber DSM 40012, S. cinnamoneus, S. coelicolor A3(2) (DSM 40783; S. violaceoruber), S. echinatus Tü 12, S. eurythermus DSM 40014, S. flaveolus Tü 55, S. fradiae T59-235, S. glaucus Tü 490, S. griseoflavus Tü 52, S. lividans TK23, S. prasinopilosus DSM 40098, S. prasinus DSM 40099, S. tendae Tü 901, S. toxytricini DSM 40178, and S. viridochromogenes DSM 40110 were also tested for quercetin consumption. Strains and plasmids used for cloning and expression of queD are listed in Table 1. Streptomyces and Escherichia coli strains were grown at 30°C in Standard I medium (Merck, Darmstadt, Germany) and at 37°C in Lysogeny Broth (LB; Sambrook et al. 1989), respectively; LB contained ampicillin (100 μg ml−1) and chloramphenicol (34 μg ml−1), if appropriate.

Table 1 Bacterial strains and plasmids used for cloning of queD from Streptomyces sp. FLA
Full size table

Streptomyces strains were inoculated onto overlay agar plates with mineral salts medium (Schlegel et al. 1961) and 2 mM quercetin in the upper layer to investigate quercetin consumption. For the preparation of cell extracts to determine quercetinase activity, Streptomyces strains were grown in Standard I medium for 2 days before 1 mM quercetin was added. After another 4 h of incubation, cells were harvested by centrifugation.

To assess growth of Streptomyces sp. FLA on quercetin as a sole source of carbon and energy, equal amounts of a washed cell suspension were used to inoculate mineral salts medium (Schlegel et al. 1961) containing 1 mM quercetin. Cultures were grown for approximately 5 weeks until decolorisation of the yellow medium indicated consumption of quercetin, and dry weight of biomass was determined. Mineral salts medium containing 1% glucose and medium lacking any carbon source, inoculated with the same amount of cells, were used as controls.

For the preparation of quercetinase, Streptomyces sp. FLA was grown in GYM medium (0.4% glucose, 0.4% yeast extract, 1% malt extract, pH 7.2) for 2 days at 30°C, then 1 mM quercetin was added and the culture was incubated for another 2 days at 25°C. Cells were harvested, washed twice with saline (0.5% NaCl, 0.012% MgSO4·7H2O), resuspended in mineral salts medium containing 2 mM quercetin, and harvested after two more days of growth at 25°C.

To examine whether the presence of glucose as an additional carbon source besides quercetin affects synthesis of quercetinase, Streptomyces sp. FLA was grown in Standard I medium for 24 h, 1 mM quercetin was added, and cells were harvested by centrifugation after another 24 h of incubation. Cell pellets were washed twice in saline, resuspended in mineral salts medium (Schlegel et al. 1961) containing 2 mM quercetin or 2 mM quercetin and 1% glucose, incubated for 24 h at 30°C, and harvested.

For preparation of protoplasts, S. lividans TK23 was cultivated in YEME-medium; R5 agar plates were used for protoplast regeneration after transformation. These media and the method used for protoplast preparation are described in Kieser et al. (2000). Media for growth of S. lividans TK23 harbouring pIJ702 or derivatives contained thiostrepton (25 μg ml−1). To identify clones with quercetinase activity, S. lividans TK23 was inoculated onto overlay agar plates with mineral salts medium (Schlegel et al. 1961) and 2 mM quercetin in the upper layer.

Preparation of crude extracts, estimation of protein concentration, and enzyme assays

Streptomyces cells were resuspended in 50 mM Tris/HCl buffer, pH 7 (pH 9 for protein purification from wild-type strain FLA) and disrupted by sonication. Cell debris was removed by centrifugation. Protein concentrations were determined according to the method of Zor and Selinger (1996), using bovine serum albumin as standard.

Quercetinase activity was determined spectrophotometrically by measuring quercetin consumption. The reaction mixture in a total volume of 1 ml contained 50 μl of 1.2 mM quercetin dissolved in dimethyl sulfoxide (DMSO) and appropriate amounts of protein in 50 mM Tris/HCl buffer, pH 7.0 (for wild-type quercetinase) or pH 8.0 (for recombinant quercetinase). One unit was defined as the amount of enzyme that converts 1 μmol of quercetin per min at 25°C in these buffers (ε 367 nm, pH8 = 14,850 M−1 cm−1; ε 367 nm, pH7 = 17,100 M−1 cm−1). All kinetic assays were carried out in air-saturated buffer. For the determination of kinetic constants for quercetin, substrate concentrations of 1.75–150 μM were used, and apparent K m and k cat values were deduced from Hanes plots (Hanes 1932). Assays were done at least in triplicate.

Enzyme-catalyzed formation of carbon monoxide from quercetin was detected as described by Waterman (1978). Protein solution (300 μl), 1.2 mM quercetin dissolved in DMSO (200 μl), and 400 μl of 50 mM Tris/HCl buffer, pH 7.0 were mixed and incubated for 3 min. After subsequent addition of 50 μl hemoglobin (30 mg ml−1 in H2O) and 50 μl sodium dithionite (50 mg ml−1 in H2O), a visible absorption spectrum was recorded (450–650 nm). CO-hemoglobin exhibits two absorption maxima at 540 and 570 nm, whereas reduced O2-hemoglobin shows a single maximum at 552 nm.

The activity of purified recombinant quercetinase towards flavonoids was determined by measuring oxygen consumption with a Clark-type oxygen electrode (Digital Model 10, Rank Brothers Ltd, Cambridge, England). The assay contained 50 μl of a flavonoid solution (1 mM in DMSO), 20 μl enzyme (0.0112–1.12 U), and 930 μl 50 mM Tris/HCl buffer, pH 7; such pH was chosen because myricetin rapidly decomposed at pH 8. To detect quercetinase-catalyzed CO production from flavonoids, the same assay was carried out in a 1.5 ml reaction tube, with a filter paper soaked with aqueous PdCl2 (1:500, w/v) placed in the lid. CO released by the enzyme-catalyzed reaction reduces Pd2+ to elemental palladium, which precipitates as a black solid (Arendt and Dörmer 1972).

Preparation of quercetinase from strain FLA and determination of its N-terminal amino acid sequence

Ammonium sulfate was added to crude extract supernatant of strain FLA at 4°C to 10% saturation, and the supernatant after centrifugation for 20 min at 48,000g was loaded onto a phenyl-Sepharose CL-4B column (20 ml) (Amersham Biosciences, Freiburg, Germany) that had been equilibrated in 50 mM Tris/HCl buffer containing 0.4 M (NH4)2SO4, pH 9.0. After two washing steps with 0.4 and 0.1 M (NH4)2SO4, proteins were eluted with a linear gradient (20 ml) from 0.1 to 0 M (NH4)2SO4 in equilibration buffer. Fractions showing quercetinase activity were pooled and chromatographed at a 1 ml Uno™-Q column (BioRad, München, Germany) that had been equilibrated in 50 mM Tris/HCl buffer, pH 9.0. Proteins were eluted with a linear gradient (25 ml) from 0 to 0.28 M NaCl in the equilibration buffer. The active fractions were combined and concentrated by ultra-filtration. The concentrate was separated in a preparative native 7.5% polyacrylamide gel (“high pH discontinuous system” according to Hames 1990). The active protein was eluted from the gel by dialysis, concentrated by ultra-filtration, and stored at −80°C.

For N-terminal sequencing, the quercetinase preparation was separated in a 12.5% SDS-polyacrylamide gel (Laemmli 1970) and transferred to a polyvinylidene fluoride membrane (Millipore, Eschborn, Germany) by semi-dry electroblotting (Towbin et al. 1979). After staining (0.1% Coomassie Brilliant Blue R-250 in 50% methanol, 7% acetic acid), destaining (50% methanol, 7% acetic acid) and drying of the membrane, the amino-terminal sequence of the 21 kDa protein was determined by automated Edman degradation by Dr. Bernhard Schmidt (Zentrum Biochemie und Molekulare Zellbiologie, Biochemie II, Georg-August-Universität Göttingen, Germany).

DNA techniques

Chromosomal DNA of Streptomyces sp. FLA was prepared as described by Pospiech and Neumann (1995). Plasmid DNA was isolated with the E.Z.N.A.® Plasmid Miniprep Kit (PeqLab, Erlangen, Germany). Gel extraction of DNA fragments for cloning was done with the E.Z.N.A.® Gel Extraction Kit (PeqLab, Erlangen, Germany). DNA restriction, dephosphorylation, ligation, and agarose gel electrophoresis were carried out using standard procedures (Sambrook et al. 1989). Preparation of competent E. coli DH5α and BL21 (DE3) pLysS cells and transformation was performed as described by Hanahan (1983).

Polymerase chain reaction (PCR) for amplification of Streptomyces DNA was performed using Pfu DNA polymerase (Fermentas, St. Leon-Roth, Germany), with glycerol added to the reaction mixture to a final concentration of 5%. For strain identification, 16S rDNA of Streptomyces sp. FLA was amplified by PCR using the primers GM3F and GM4R (Muyzer et al. 1995). DNA sequencing was performed by MWG-Biotech AG (Ebersberg, Germany).

DNA probes and hybridization

Southern and colony blotting, hybridization, and colorimetric detection with nitroblue tetrazolium salt and 5-bromo-4-chloro-3-indolyl phosphate were carried out following the DIG System User´s Guide for Filter Hybridization (Roche molecular biochemicals, 1995). The degenerated oligonucleotide probe Q1 (5′-AC(G/C)ATCGA(A/G)TACGC(G/C)AC(G/C)CG(G/C)CACCG(G/C)GC(G/C)CG-3′) was synthesized based on the N-terminal amino acid sequence of quercetinase (TIEYATRHRAR) and 3′-labeled with digoxigenin (MWG-Biotech AG, Ebersberg, Germany). To generate the specific probe Q2, a 377 bp fragment of the quercetinase gene was amplified by PCR using the primers 5′-GACCATCGAATACGCCACC-3′ and 5′-CGACCTGCGAGTGGTGGC-3′, and pUC18KQ (Table 1) as template. The PCR product was purified with the High Pure PCR Product Purification Kit and digoxigenin-labeled with the DIG-High Prime Kit (both kits from Roche, Mannheim, Germany). Prehybridization for 2 h and hybridization overnight with Q1 or Q2 were carried out at 68°C.

Construction of genomic libraries

To generate enriched gene libraries for Streptomyces sp. FLA, KpnI- or BamHI-restricted genomic DNA was separated in 1.0% agarose gels and vacuum-blotted to nylon membranes (Porablot NY plus from Macherey-Nagel, Düren, Germany). Fragments in the size of 2.5–3.5 kb (KpnI) and 3.0–4.0 kb (BamHI), showing positive hybridization signals with probes Q1 and Q2, respectively, were extracted from preparative agarose gels and ligated into linearised and dephosphorylated vector pUC18. E. coli DH5α transformants were screened by colony blotting using probes Q1 (KpnI library) and Q2 (BamHI library).

Sequence analysis

Gene-coding sequences were identified using the program FramePlot version 2.3.2 (Ishikawa and Hotta 1999), which considers the high G + C distribution at the third position of Streptomyces genes. Database searches were carried out at the NCBI with the BLAST family of programs (Altschul et al. 1990) from the NCBI website (http://www.ncbi.nlm.nih.gov). Conserved protein domain sequences were found by CD-Search (Marchler-Bauer et al. 2005). For calculating similarities and identities, and for compiling multiple alignments, the programs GAP (Needleman and Wunsch 1970) and ClustalW (Higgins et al. 1996) were used, respectively. Secondary structure and signal peptide predictions were carried out using PredictProtein (Rost et al. 2004) and both SignalP and TatP (Bendtsen et al. 2004, 2005), respectively.

Expression of queD in Streptomyces lividans TK23

For heterologous expression of queD from its own promoter, the 3.4 kb BamHI insert of pUC18BQ was ligated into the BglII digested Streptomyces clonig vector pIJ702, resulting in pIJ702HM. Polyethylene glycol-assisted protoplast transformation, carried out using the “rapid small-scale procedure” described in Kieser et al. (2000), was used to transfer the recombinant plasmid to S. lividans TK23. The transformants were screened on quercetin overlay agar plates. Quercetinase activity was indicated by formation of zones of decolorisation on the yellow agar plates.

Cloning and overexpression of queD in E. coli

The gene queD was amplified by polymerase chain reaction using pUC18BQ as a template and the primers 5′-GGAATTCCATATGACCATCGAATACGCCAC-3′ and 5′-CCCAAGCTTCCTTCCCTCGATACTCCCGGTGTGCCACTG-3′, which insert a NdeI site (underlined) at the start codon, and a HindIII site (underlined and in italics) at the stop codon. After digestion with NdeI and HindIII, the PCR product was ligated into the vector pET23a, restricted with the same enzymes, resulting in pET23aqueD. Sequencing of both strands of the insert of pET23aqueD confirmed that its sequence was identical to that of the template. E. coli BL21 (DE3) pLysS was transformed with pET23aqueD and the recombinant strain was grown to an optical density at 600 nm of 0.5, before expression of queD was induced by addition of 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG). The temperature was reduced to 25°C, and after 6 h of induction, cells were harvested by centrifugation at 10,000g for 10 min at 4°C and stored at −80°C. Gene expression from the T7 promoter of pET23a produces QueD protein with a C-terminal hexahistidine-tag, designated QueDHis6.

Purification of recombinant QueDHis6

E. coli BL21 (DE3) pLysS pET23aqueD cells were resuspended in 50 mM Tris/HCl buffer, pH 7.5, containing 1 mM MgCl2, and lysed by the pLysS-encoded lysozyme. After DNA digestion for 1.5 h at 4°C using 25 units Benzonase (Novagen) per ml, the extract was centrifuged at 39,000g for 30 min at 4°C, and imidazole and NaCl were added to the supernatant to a final concentration of 10 and 300 mM, respectively. This protein solution was loaded onto a MT5 column (BioRad), packed with Ni-NTA Agarose (Qiagen, Hilden, Germany) that had been equilibrated in 50 mM Tris/HCl buffer containing 20 mM imidazole and 300 mM NaCl (pH 7.5). Recombinant QueDHis6 was eluted from the column with a linear gradient (20 ml) of imidazole (20–500 mM) in the same buffer. Fractions exhibiting quercetinase activity were pooled, washed with 50 mM Tris/HCl buffer, pH 8, and concentrated by ultrafiltration. After addition of 10% (v/v) glycerol, the concentrate was stored at −80°C. To determine the purity of the enzyme, it was subjected to SDS-PAGE (12.5% acrylamide gel according to Laemmli (1970)). Polyacrylamide gels were stained with Coomassie Brilliant Blue G-350 (0.15% Coomassie Brilliant Blue G-350 (w/w), 0.25% CuSO4·5H2O (w/w), 45% methanol (v/v), 10% (v/v) acetic acid), and destained in an aqueous solution of 30% (v/v) methanol and 10% (v/v) acetic acid.

Gel filtration of QueDHis6 for estimation of its native molecular mass was performed on a Bio-Prep SE-1000/17 column (BioRad Laboratories) in 50 mM Tris/HCl buffer, pH 7.5 containing 150 mM NaCl. Gel Filtration Standard marker from BioRad was used for calibration of the column.

Nucleotide sequence accession numbers

The sequences of the 4.5 kb KpnI/BamHI fragment that includes queD and of the 16S rDNA of Streptomyces sp. FLA have been deposited in the EMBL Nucleotide Sequence Database under accession numbers AM234612 and AM234613, respectively.

Results

Quercetin-degrading Streptomycetes, and characterization of Streptomyces sp. strain FLA

Out of 18 Streptomyces spp. tested besides strain FLA, the strains S. alboniger DSM 40043, S. eurythermus DSM 40014, S. flaveolus Tü 55, and S. tendae Tü 901 were able to decompose the flavonol quercetin as indicated by formation of colorless zones on quercetin agar. Comparison of the 16S rDNA gene sequence of strain FLA revealed significant homology to S. eurythermus T DSM 40014 (99.46% identity within 1,482 nucleotides), S. nogalater T DSM 40546 (99.18% identity within 1,469 nucleotides), and S. tendae T DSM 40101 (98.38% identity within 1,483 nucleotides). Streptomyces sp. FLA is able to utilize quercetin as sole carbon source; however, growth is very poor. After approximately 5 weeks of growth, the measured gain of cell dry weight, compared with cells incubated without any carbon source, was only 1.8 ± 0.3 mg. Such poor increase in biomass was not due to limitation by other components of the mineral salts medium, since strain FLA showed abundant growth in the same medium containing 1% glucose.

Quercetinase of Streptomyces sp. FLA

After induction of Streptomyces sp. strain FLA with quercetin, specific quercetinase activities between 0.17 and 0.19 U mg−1 were measured in crude extracts (soluble fraction). The four-step purification procedure resulted in a 590-fold enrichment of quercetinase from crude extract supernatant in 27% yield. About 0.2 mg protein with a specific activity of 98 U mg−1 were obtained from 38 g of wet biomass of strain FLA. Enzyme-catalyzed quercetinase conversion resulted in release of carbon monoxide, as shown by detection of CO-hemoglobin. When quercetin was omitted from the growth medium, quercetinase activity was not detected in crude extracts, indicating that synthesis of the enzyme is induced by the flavonol. The presence of glucose as a second carbon source in the medium besides quercetin did not significantly influence quercetinase activity in cell extract supernatants, suggesting that there is no carbon catabolite control of quercetin metabolism by glucose.

Cloning of the quercetinase gene queD of Streptomyces sp. FLA

A 2.9 kb KpnI fragment of total DNA of Streptomyces sp. FLA specifically hybridized with probe Q1. DNA fragments sized between 2.5 and 3.5 kb were ligated into KpnI-digested pUC18, and colony hybridization of the corresponding library in E. coli DH5α using the same probe resulted in the identification of a clone harbouring pUC18KQ (Table 1). Since sequence analysis of the 2.9 kb insert revealed that it only contained part of the presumed quercetinase gene, the obtained sequence was used to synthesize specific probe Q2, which was used to screen a BamHI library, yielding an E. coli clone that contained the complete quercetinase gene within a 3.4 kb fragment (pUC18BQ). It is interesting to note that the Q2 probe also specifically hybridized to distinct fragments of genomic DNA of the S. alboniger, S. eurythermus, S. flaveolus, and S. tendae strains that convert quercetin, whereas no hybridization signals were observed with DNA of Bacillus subtilis, even when attenuating the stringency of hybridization (data not shown).

Sequence analysis of queD and flanking regions

From the E. coli clones harbouring pUC18KQ and pUC18BQ (Table 1), a total DNA sequence of 4.5 kb was obtained, which was predicted to comprise three full and two truncated ORFs (Fig. 1). The hypothetical protein fragment encoded by truncated ORF1 is related to corresponding C-terminal segments of a broad range of proteases, e.g., to tripeptidylaminopeptidase Tap (acc no AAA92338) from Streptomyces lividans 66 (36% identity). The similarity includes a conserved GxSxG “nucleophile elbow” motif whose serine is potentially involved in the catalytic action of this protease (Krieger et al. 1994; Butler et al. 1995) and of related α/β-hydrolases, suggesting that ORF1 encodes a serine hydrolase.

Fig. 1
figure 1

DNA region (4.5 kb) of Streptomyces sp. FLA containing the gene coding for quercetinase (queD). Putative ORFs are indicated by open arrows. For a description, see text

Full size image

Putative ORF2 is located 1,378 nucleotides upstream of queD and is deduced to code for a protein of 130 aa with a calculated molecular mass of 14,373 Da. The ORF2 protein shows close similarity to endoribonuclease L-PSP of Nitrosospira multiformis ATCC 25196 (acc no ABB75555, 129 aa, 69% identity). The amino acid segment spanning positions 18–126 is related to the conserved domain of the L-PSP (rat liver perchloric acid-soluble protein) endoribonucleases (Pfam signature PF01042). Members of this family are thought to inhibit protein synthesis by cleavage of mRNA (Morishita et al. 1999) and are presumed to be involved in the regulation of purine biosynthesis (Rappu et al. 1999).

ORF3 starts 61 nucleotides downstream of the queD stop codon, and a putative ribosome binding site (GGAG) was identified eight nucleotides upstream of the alternative start codon GTG. The 30,200 Da hypothetical protein (281 aa) exhibits 42% identity to amidohydrolase two from Arthrobacter sp. FB24 (acc no ZP_00414293). A conserved domain search revealed that the region spanning amino acids 9–195 is related to the amidohydrolase two protein family (Pfam signature PF04909).

Potential ORF4, with a possible Shine Dalgarno sequence (GAGG) 6 nucleotides upstream of its start codon, starts 89 nucleotides downstream of ORF3. The putative protein fragment (146 aa) resembles the N-terminal segment of dienelactone hydrolase (323 aa) from Novosphingobium aromaticivorans DSM 12444 with 41% identity, and the N-terminal segments of several esterases, e.g., carboxylesterase EST2 (310 aa) from Alicyclobacillus acidocaldarius (acc no 1EVQ_A, Manco et al. 1999), showing 35% identity within the aligned region. The amino acid segment comprising positions 60–139 of the ORF4 protein aligns with the N-terminal part of a carboxylesterase domain (Pfam signature PF00135). Carboxylesterases and dienelactone hydrolases are members of the α/β-hydrolase fold clan (Nardini and Dijkstra 1999).

FramePlot analysis of the 983 bp DNA sequence between putative ORF2 and queD showed that two regions exhibit a relatively high G + C content, suggesting two short open reading frames, starting 900 and 563 bp upstream of queD, which may code for hypothetical proteins of 144 and 188 aa. However, comparison of the deduced hypothetical proteins with the NCBI database did not reveal any significant matches, and it is not clear whether these ORFs indeed represent functional genes.

The quercetinase gene queD consists of 558 bp, encoding a protein of 186 amino acids with a calculated molecular mass of 21,065 Da. As described for other genes of Streptomycetes (Wright and Bibb 1992), queD exhibits a high overall G + C content (67.6%), with pronounced preference of G and C at the third codon position (93.6% G + C). Six nucleotides upstream of the initiation codon ATG, a possible ribosome binding site was identified (GGAGG). Analysis of the deduced amino acid sequence with SignalP and TatP did not predict a signal peptide, indicating that Streptomyces quercetinase in contrast to the fungal enzymes (Oka et al. 1971; Hund et al. 1999; Fusetti et al. 2002) is a cytoplasmic protein, as already obvious by its preparation from cell extracts of strain FLA as a soluble protein. Streptomyces QueD is most closely related to quercetinase of Bacillus subtilis (QueD, or YxaG, acc no P42106), with 35.9 and 29.0% identity of Streptomyces QueD to the C- and N-terminal domain of Bacillus QueD, respectively. Secondary structure predictions suggested that the order of α-helices and β-strands of QueD from strain FLA is similar to that of the C-terminal cupin domain of Bacillus quercetinase, indicating that the Streptomyces protein also exhibits the typical cupin β-barrel fold. The N- and C-terminal domains of quercetinase from Aspergillus japonicus (acc no Q7SIC2) exhibit 27.9 and 23.4% identity, respectively, to QueD from strain FLA. In a multiple comparison performed with ClustalW, the Streptomyces protein aligns with the C-terminal cupin domains of both Bacillus and Aspergillus quercetinase (Fig. 2). The alignment and secondary structure predictions show that, in contrast to the quercetinases of B. subtilis and A. japonicus which are both bicupins, Streptomyces QueD belongs to the monocupins.

Fig. 2
figure 2

Alignment of quercetinases from Streptomyces sp. FLA, Bacillus subtilis (acc no P42106) and Aspergillus japonicus (acc no Q7SIC2), compiled with the ClustalW program. Amino acids identical in two and three enzymes are highlighted in gray and black, respectively. Each catalytically active cupin domain comprises three conserved histidine residues and a glutamate (marked by asterisks), which act as ligands of the active site metal

Full size image

The sequences 63 GEVIPAHSHADTYEVFYITQG 83 and 104 GDFGFVPKNCVHAYRM119 of QueD from strain FLA match the consensus motifs G(X)5 HXH(X)3,4 E(X)6 G and G(X)5 PXG(X)2 H(X)3N of the cupin superfamily (Dunwell et al. 2000, 2001, 2004). In the enzymatic members of this superfamily, the two histidine residues and the glutamate in motif 1 as well as histidine in motif 2 act as ligands for an active-site divalent metal ion. Conservation of these residues suggests that QueD contains a catalytically relevant metal cofactor as well.

Heterologous expression of queD in Streptomyces lividans TK23

To confirm that the DNA fragment harbouring the queD gene confers quercetinase activity, the 3.4 kb BamHI fragment (Fig. 1) was ligated into the Streptomyces vector pIJ702, resulting in pIJ702HM. Streptomyces lividans strain TK23, which does not exhibit quercetinase activity, was used as a host for the recombinant plasmid. Transformants formed zones of decolorisation on quercetin agar plates (Fig. 3), indicating formation of functional quercetinase in S. lividans. Crude extract of S. lividans TK23 pIJ702HM exhibited a specific quercetinase activity of 5.13 U mg−1, which is about 26-fold higher than measured in extracts of Streptomyces sp. FLA. Such increase of specific activity in the recombinant strain is consistent with the approximate copy number of the cloning vector.

Fig. 3
figure 3

Streptomyces lividans strain TK23 transformed with pIJ702 (a) and pIJ702HM (b) on mineral medium containing 2 mM quercetin and 0.5% glucose as carbon sources. Quercetinase activity is indicated by formation of colorless zones on the yellow quercetin agar

Full size image

Overexpression of queD in E. coli, and properties of recombinant QueD His6

The gene queD of Streptomyces sp. FLA was overexpressed in E. coli BL21 (DE3) pLysS pET23aqueD, and the recombinant QueDHis6 protein (calculated molecular mass of 23,041 Da) was purified by Ni2+ chelate affinity chromatography to near electrophoretic homogeneity (Fig. 4). About 30 mg of QueDHis6 were obtained from 4.8 g of cells (wet biomass). The enrichment factor from crude extract supernatant was approximately 21, reflecting the high content of recombinant protein in the cell extract (Fig. 4). Gel filtration of the purified protein indicated a molecular mass of approximately 63 kDa, suggesting that QueDHis6 is a multimeric protein in its native state. Recombinant QueDHis6 catalyzed the cleavage of quercetin with a specific activity of 3.96 U mg−1 (at pH 8) with concomitant release of CO, indicating that the gene product of queD is sufficient for quercetinase activity. From a V max of 10.0 ± 0.77 μM min−1, an apparent k cat of 1.45 ± 0.11 s−1 was calculated for recombinant QueDHis6. This k cat value is slightly higher than that of (likewise His-tagged) B. subtilis quercetinase, as isolated from recombinant E. coli cells grown in LB (0.8 s−1; Bowater et al. 2004). The apparent K m value of QueDHis6 for quercetin, determined in the standard assay (pH 8) in air-saturated buffer, was 14.1 ± 0.7 μM. This value is higher than the K m values reported for the quercetinases from A. flavus and A. niger of 5.2 μM (Oka et al. 1971) and 6.6 μM (Hund et al. 1999), respectively. B. subtilis quercetinase prepared from LB-grown recombinant E. coli also showed a low K m of 3.8 μM (Bowater et al. 2004).

Fig. 4
figure 4

SDS-PAGE of recombinant QueDHis6. Lane 1 Molecular mass standards; lane 2 crude extract supernatant of E. coli BL21 (DE3) pLysS pET23aqueD, obtained from cells harvested 6 h after IPTG induction; lane 3 recombinant QueDHis6 after purification by Ni2+ chelate affinity chromatography

Full size image

Several flavonoids were tested for their decomposition by recombinant Streptomyces quercetinase (Table 2). Besides quercetin, the flavonols kaempferol, myricetin, galangin, fisetin, and morin were converted by the enzyme. Kaempferol, which compared to quercetin lacks the 3′-OH group at the B-ring, was cleaved with relatively high rate, indicating that the substituent at this position of the benzene ring is not of major importance for catalysis. On the other hand, a hydroxy substituent at position 2′ appears to drastically impair the catalytic activity of QueDHis6, since morin, which has a 2′,4′-dihydroxy substitution pattern at the B-ring, was converted with a relative rate of only 1.7%. This may be due to steric effects, and/or the particularly low pK a (3.46) of one of its dissociating OH-groups (Jovanovic et al. 1994). In contrast, an additional hydroxy substituent at position C5′ of quercetin had a comparatively minor effect on catalysis by QueDHis6, as indicated by its relative activity of about 50% towards myricetin (Table 2). Enzyme-catalyzed release of carbon monoxide was detected from all flavonols (3-hydroxy-flavones) tested, suggesting that they all undergo 2,4-dioxygenolytic ring cleavage (Table 2). An apparent relative activity of QueDHis6 towards taxifolin of about 0.1% was measured by monitoring dioxygen consumption, but carbon monoxide formation was not observed. Since this flavonoid differs from the physiological substrate quercetin only by saturation of C2 and C3 of the O-heterocyclic ring, it is obvious that the C2–C3 double bond of the flavonols, which is conjugated with the 4-oxo group involved in electron delocalization, plays an important role in catalysis. Enzyme-catalyzed dioxygen consumption and CO formation were not observed when QueDHis6 was incubated with the flavone luteolin, which apart from the missing hydroxyl group at C3 of the C-ring is identical to quercetin, demonstrating that the flavonol structure is indispensable for the dioxygenolytic ring cleavage reaction by QueDHis6.

Table 2 Basic flavone structure and activity of recombinant QueDHis6 towards flavonoids
Full size table

Discussion

In this study, the gene queD coding for CO-forming quercetinase was identified in Streptomyces sp. strain FLA, a soil isolate related to Streptomyces eurythermus T. As indicated by hybridization of a queD-specific probe with genomic DNA of quercetin degrading S. alboniger, S. eurythermus, S. flaveolus, and S. tendae strains, the gene coding for quercetinase is fairly conserved among these Streptomycetes. Sequence analysis of DNA regions flanking queD suggests that the quercetinase gene in Streptomyces sp. strain FLA is not part of a catabolic gene cluster involved in the degradation of aromatic compounds. However, we can not exclude the possibility that the gene product of ORF4, which shows similarities to carboxylesterases, is an esterase catalyzing cleavage of 2-protocatechuoylphloroglucinol carboxylic acid, the product of the quercetinase reaction.

Bicupin proteins of the pirin family (pfam acc no PF02678), which bind Fe2+ (Pang et al. 2004), are highly conserved among both eukaryotes and prokaryotes, but their function is poorly understood. They have been proposed to be involved in co-regulation of transcription and DNA replication; however, recently it has been shown that human pirin and its E. coli homologue Yhhw possess quercetinase activity. Since quercetin interferes with various cellular pathways in both eukaryotes and prokaryotes, it has been discussed that the quercetinase activity of pirins might be important to counteract its inhibitory effects (Adams and Jia 2005). In enteric bacteria like E. coli as well as in soil bacteria like Streptomyces sp. FLA, which are very likely to encounter plant polyphenols in their natural habitat, quercetinase activity might indeed be most important for detoxification, since quercetin exhibits antimicrobial activities (for review, see Cushnie and Lamb 2005). It was reported to cause an increase in the permeability of the inner bacterial membrane and a dissipation of the membrane potential (Mirzoeva et al. 1997), to inhibit bacterial DNA gyrase B at its ATP binding site, and to induce DNA cleavage at the DNA-gyrase complex (Plaper et al. 2003). Bacterial quercetinases thus may play a primary role in detoxification, but additionally in some organisms, such as Aspergillus spp., these proteins may have evolved to potent catabolic enzymes. However, the observed marginal growth of Streptomyces sp. FLA on quercetin suggests that in this strain, quercetin metabolism is not optimized for using this compound as source of carbon and energy.

Amino acid sequences and crystal structures are available for A. japonicus and B. subtilis quercetinase. In the latter protein, the N- and C-terminal cupin domains both comprise a set of three strictly conserved histidines and a glutamate residue, each set coordinating a divalent metal ion, and both domains are catalytically active (Bowater et al. 2004; Gopal et al. 2005). In contrast to Bacillus quercetinase, the three potential ligands to a metal are missing in the C-terminal cupin domain of the A. japonicus enzyme (Fig. 2). A. japonicus quercetinase thus possesses only one active site in its N-terminal cupin domain, where H66, H68, H112, and a water molecule or E73 coordinate a Cu2+ ion (Fusetti et al. 2002). Dunwell et al. (2004) reported that the majority of bicupin dioxygenases have only one active domain, whilst the other is a non-functional remnant.

Since pirins as well as Bacillus and Aspergillus quercetinases are bicupins, the enzyme of Streptomyces sp. FLA is the only monocupin quercetinase described as yet, indicating that the interface between the two domains of the bicupin quercetinases is not essential for catalytic activity. Bicupins have been suggested to have evolved by gene duplication and then fusion from ancestral single cupin progenitors, or in some cases by fusion of two different monocupin precursor genes (Dunwell et al. 2000, 2001, 2004). Based on this hypothesis, the Streptomyces enzyme might represent an “ancient” bacterial quercetinase.

Purified recombinant QueDHis6 exhibited a specific activity of about 4 U mg−1 and an apparent k cat of only 1.45 s−1 towards quercetin. Such poor specific activity compared to wild-type quercetinase from strain FLA (98 U mg−1) might be due to the cultivation of the E. coli clone in LB medium. Bacillus quercetinase also showed a low k cat of 0.8 s−1 when purified from recombinant E. coli grown in LB (Bowater et al. 2004); under these conditions, the Bacillus enzyme was synthesized as an iron protein (Barney et al. 2004; Bowater et al. 2004). Notably, the divalent metal ions present in the growth medium during gene expression significantly influence the metal content and catalytic activity of recombinant Bacillus quercetinase. Addition of Mn2+ and Co2+ to minimal medium (M9) resulted in highest quercetinase activity in cell extracts, and purified Mn-QueD and Co-QueD showed k cat values of 25 and 6.7 s−1, respectively (Schaab et al. 2006). The authors suggested that the low k cat of quercetinase purified from LB-grown clones is due to the 20- to 30-fold lower intracellular concentration of manganese compared to iron in LB-grown E. coli (Gabbianelli et al. 1995), which controls Fe incorporation into the active site. Concerning Streptomyces quercetinase, it is conceivable that the LB medium likewise does not provide sufficient amounts of the metal required for optimal activity.

QueDHis6 of Streptomyces sp. FLA was found to catalyze cleavage of the flavonols kaempferol, myricetin, galangin, fisetin, and morin. These compounds are also converted by quercetinase of A. flavus, and the relative activities of the two enzymes show similar tendencies, apart from the activity towards kaempferol, which in case of the fungal enzyme is 2.5-fold higher than towards quercetin (Oka et al. 1972). The failure of recombinant QueDHis6 to decompose the flavone luteolin (“3-desoxyquercetin”) and to catalyze CO release from the 2,3-dihydroflavonol taxifolin suggests that the flavonol scaffold is essential for the enzyme-catalyzed reaction. Such observation is consistent with the reaction mechanisms proposed for Cu2+-quercetinase of A. japonicus (Steiner et al. 2002) and the Mn2+-enzyme of B. subtilis (Schaab et al. 2006). Despite important mechanistic differences, both reactions involve quercetin binding to the metal centre via its O3 atom, and single-electron oxidation of the flavonol ring, in which the generated radical is stabilized by the delocalised pi-electron system.

The functionally highly variable cupin superfamily comprises several types of dioxygenases (Dunwell et al. 2000, 2001, 2004). Within the monocupin dioxygenases, the predominant subclass is that of the 2-oxoglutarate- and Fe2+-dependent dioxygenases. Further monocupin dioxygenases are 3-hydroxyanthranilate 3,4-dioxygenase (EC 1.13.11.6), cysteine dioxygenase (EC 1.13.11.20), and acireductone dioxygenase (ARD) from Klebsiella pneumoniae (gi:75475369; PDB 1M4O). A most interesting feature of the latter protein is its ability to catalyze different reactions with Ni2+ or Fe2+ bound to its active site. Fe-ARD catalyzes a reaction that is part of the ubiquitous methionine salvage pathway, converting 1,2-dihydroxy-3-keto-5-(methylthio)pentene to formate and 2-keto-4-(methylthio)butanoate, the α-ketoacid precursor of methionine. In contrast, Ni-ARD yields methylthiopropanoate, CO and formate from the same substrate, i.e., it catalyzes the dioxygenolytic cleavage of two carbon–carbon bonds with concomitant release of CO, a reaction analogous to that of quercetinase. However, the in vivo significance of this reaction is unknown (Dai et al. 1999, 2001; Pochapsky et al. 2002). Ni-ARD-type activity was observed when the apoprotein was reconstituted with Mn2+, Co2+, and Ni2+, whereas Fe-ARD activity was conferred by Fe2+ and Mg2+ (Pochapsky et al. 2002). ARD proteins constitute an own enzyme family (Pfam acc no PF03079); however, Klebsiella ARD and Streptomyces QueD share 21.7% identity of their amino acid sequences.

The cupin fold is well adapted for dioxygenases, since its signature motifs provide His and Glu ligands for binding a divalent metal ion that can be used for the activation of molecular oxygen, the organic substrate, or both. Among dioxygenases, iron is the most commonly used transition metal. Copper, which became biologically available later than iron, is found in a much smaller range of oxygenases (Sariaslani 1989; Harayama et al. 1992). Based on structural as well as mechanistic considerations, quercetinase of A. japonicus has been suggested to have evolved from a primordial Fe-containing enzyme (Fusetti et al. 2002; Steiner et al. 2002). The Bacillus enzyme, on the other hand, may be a member of the manganese dioxygenase family, and recent kinetic and spectroscopic evidence suggests that the catalytic mechanisms of Aspergillus and Bacillus quercetinases are different (Schaab et al. 2006). Identification of the metal cofactor of Streptomyces quercetinase may contribute further insight into the evolutionary relationship of bacterial and fungal quercetinases and into the versatility of these enzymes in terms of their metal cofactor and catalytic mechanism.