Tracking Down Biotransformation to the Genetic Level: Identification of a Highly Flexible Glycosyltransferase from Saccharothrix espanaensis

Abstract

Saccharothrix espanaensis is a member of the order Actinomycetales. The genome of the strain has been sequenced recently, revealing 106 glycosyltransferase genes. In this paper, we report the detection of a glycosyltransferase from Saccharothrix espanaensis which is able to rhamnosylate different phenolic compounds targeting different positions of the molecules. The gene encoding the flexible glycosyltransferase is not located close to a natural product biosynthetic gene cluster. Therefore, the native function of this enzyme might be not the biosynthesis of a secondary metabolite but the glycosylation of internal and external natural products as part of a defense mechanism.

INTRODUCTION

Biotransformation is the process of modification of organic compounds into recoverable products by chemical reactions catalyzed by cellular enzymes. Common chemical modifications occurring during biotransformation include oxidation, reduction, hydrolysis, hydroxylation, isomerization, and glycosylation (1–3). Detection of the enzymes catalyzing the biotransformation reactions is of great interest since the use of these enzymes as a tool for the synthesis of modified drug leads allows researchers to economically generate the desired products.

Many natural products are decorated with carbohydrates essential for the interaction with their biological targets. Particularly, deoxyhexoses are important carbohydrate components which mediate biological activity of the glycoconjugates. They often play a key role as recognition elements in the mechanism of action of respective bioactive compounds (4). Adding or exchanging sugars can dramatically improve the pharmacological properties of a compound and change the molecular mechanism of its mode of action (5–7). As a result of the potential to alter these properties, there is intense interest in efficient methods to glycosylate small molecules and thus develop improved therapeutic agents. As chemical synthesis of glycosylated compounds is challenging and expensive (8, 9), the use of glycosyltransferases to perform glycosylation is an effective and economical alternative. Several in vitro and in vivo strategies have been employed to glycosylate a given compound. However, these approaches are limited by the high substrate specificity of glycosyltransferases (10, 11). Until now, only a few enzymes have been known to glycosylate different acceptor substrates (12). Consequently, new glycosyltransferases with low substrate specificity are urgently needed.

Strains of the genus Saccharothrix, which is a member of the order Actinomycetales, are known for their ability to glycosylate natural products (13). Recently, the complete genome of Saccharothrix espanaensis has been sequenced. Altogether, 213 genes involved in carbohydrate transport and metabolism have been detected, among them 106 glycosyltransferase genes (14). In this study, we describe the exploration of the metabolic potential of S. espanaensis in glycosylating small molecules in vivo as well as the identification and characterization of a highly flexible glycosyltransferase.

MATERIALS AND METHODS

Strains and media.

Saccharothrix espanaensis DSM 44229, Streptomyces albus J1074, and their mutants were grown on mannitol soy flour agar plates or tryptone soy agar plates (15). Liquid growth for strain maintenance was performed in tryptone soy broth (TSB), and feeding experiments were conducted in NL111 medium (16), both in double-baffled flasks at 28°C and 180 rpm on a rotary incubator. Apramycin, spectinomycin, and thiostrepton were added to the medium to final concentrations of 100 μg/ml, 100 μg/ml, and 50 μg/ml, respectively. The Escherichia coli strain Turbo was used for cloning, the strain ET12567(pUZ8002) was used to drive the conjugative transfer of nonmethylated DNA from E. coli to the actinomycete recipient, and the strain BL21 Codon Plus RP(pL1SL2) was used for recombinant protein synthesis by applying standard protocols (17–21).

Feeding experiments.

One hundred milliliters of NL111 medium was inoculated with 1 ml of a 24-h-old seed culture of the respective actinomycete strain grown in TSB. After 24 h on the rotary incubator, the culture was supplemented with 20 μl of the test compound dispersed in dimethyl sulfoxide (DMSO) (25 mg/ml). This feeding procedure was repeated four times every 12 h. After 7 days of cultivation, the pH of the culture was adjusted to 7, the culture was centrifuged, the supernatant was extracted with ethyl acetate, and the received crude extract was analyzed for biotransformation products by high-pressure liquid chromatography–mass spectrometry (HPLC-MS).

Origin of utilized test compounds.

The reactants which were used in feeding experiments (Table 1) were supplied by Sigma-Aldrich and Roth. Novobiocic acid was synthesized as already described (22). Two chemical compounds, 1-amino-4-chloranthraquinone (23) and 1,2,5,8-tetrahydroxyanthraquinone (24), were synthesized as described previously.

Table 1.

Biotransformation products generated by adding different substrates to S. espanaensis liquid cultures^a

Product type and substrate	Biotransformation product(s)	Analytical method(s)
Anthraquinone
Alizarin (1,2-dihydroxyanthraquinone)	alipro1 (2-O-α-l-rhamnosylalizarin)	HPLC-MS, NMR
	alic1 (1,2-di-O-α-l-rhamnosylalizarin)	HPLC-MS, NMR
	alipro2 (2,3-di-O-α-l-rhamnosylanthragallol)	HPLC-MS, NMR
Emodin (6-methyl-1,3,8-trihydroxyanthraquinone)	emopro1 (8-O-α-l-rhamnosylemodin)	HPLC-MS, NMR
	emopro2 (1-O-α-l-rhamnosylemodin)	HPLC-MS, NMR
	emopro3 (6,8-di-O-α-l-rhamnosylemodin)	HPLC-MS, NMR
1-Amino-4-chloranthraquinone	ant14pro1 (1-amino-4-chloro-2-hydroxy-2-O-α-l-rhamnosylanthraquinone)	HPLC-MS, NMR
Quinizarin (1,4-dihydroxyanthraquinone)	1 rhamnosylated derivative	HPLC-MS
Anthrarufin (1,5-dihydroxyanthraquinone)	1 rhamnosylated derivative	HPLC-MS
Dantron (1,8-dihydroxyanthraquinone)	1 rhamnosylated derivative	HPLC-MS
Chrysophanol (1,8-dihydroxy-3-methylanthraquinone)	3 rhamnosylated derivatives	HPLC-MS
Aloe emodin (1,8-dihydroxy-3-hydroxymethylanthraquinone)	1 rhamnosylated derivative	HPLC-MS
Quinalizarin (1,2,5,8-tetrahydroxyanthraquinone)	7 rhamnosylated derivatives	HPLC-MS
Other
Quercetin	quercpro1 (3-O-α-l-rhamnosylquercetin)	HPLC-MS, NMR
Novobiocic acid	novpro1 (11-hydroxynovobiocic acid)	HPLC-MS, NMR
	novpro3 (10-hydroxynovobiocic acid)	HPLC-MS, NMR
	novpro5 (4′-O-α-l-rhamnosylnovobiocic acid)	HPLC-MS, NMR

^aIf no NMR data were available, the suggested structures for the biotransformation products were based on their mass and UV spectra.

Genes or proteins from S. espanaensis.

The locus tag prefix “BN6_” for genes or proteins from S. espanaensis assigned by the ENA is replaced by “ses” or “Ses” throughout, respectively.

Plasmid construction.

The plasmids used in this work are displayed in Table 2. The primer sequences and the protocol for the generation of the plasmids are described in the supplemental material. The plasmids were verified by restriction endonuclease mapping and sequencing.

Table 2.

Strains and plasmids used in this study

Strain or plasmid	Descriptiona	Source or reference(s)
Strains
S. espanaensis	WT strain, biotransformation host	14, 29
Δses44520 mutant	WT with deletion of ses44520 gene	This study
Δses45900 mutant	WT with deletion of ses45900 gene	This study
Δses47640 mutant	WT with deletion of ses47640 gene	This study
Δses49200 mutant	WT with deletion of ses49200 gene	This study
Δses60310 mutant	WT with deletion of ses60310 gene	This study
Δses60310 mutant carrying pSET-ses60310	Deletion mutant of ses60310 complemented with ses60310 gene	This study
S. albus	WT strain, heterologous host	30
Rham	WT expressing the biosynthetic genes for dTDP-rhamnose	This study
Rham(pUWL-A-ses60310)	S. albus Rham expressing ses60310 gene	This study
Rham(pUWL-A-ses60350)	S. albus Rham expressing ses60350 gene	This study
E. coli
Turbo	General cloning host	NEB
ET12567(pUZ8002)	Strain for intergeneric conjugation	31, 32
BL21 Codon Plus RP(pL1SL2)	Heterologous expression host, coexpressing Streptomyces chaperonin genes	17, 20
Plasmids
pKC1132	Replicative vector in E. coli, nonreplicative in actinomycetes; aac(3)IV	33
pKGLP2	Replicative vector in E. coli, nonreplicative in actinomycetes; hph, gusA gene	34
pKCXY02	pKC1132 derivative with gusA gene	This study
pLERE	Cloning vector containing amp, two loxLE sites, and two loxRE sites	25
pIJ778	Cloning vector with amp, aadA, and oriT	35
pLERE-spec-oriT	Cloning vector with amp, aadA, and oriT flanked by two loxLE sites and two loxRE sites	This study
pKCΔses44520	Vector for deletion of ses44520, based on pKC1132	This study
pKCΔses45900	Vector for deletion of ses45900, based on pKC1132	This study
pKCΔses47640	Vector for deletion of ses47640, based on pKC1132	This study
pKCΔses49200	Vector for deletion of ses49200, based on pKC1132	This study
pKCΔses60310	Vector for deletion of ses60310, based on pKCXY02	This study
pSET-1cerm	Integrative vector for actinomycetes; oriT, int, attP (ΦC31), amp, aac(3)IV, and ermE promoter (pUWL201)	36
pSET-ses60310	pSET-1cerm derivative with ses60310 gene under ermE promoter (pUWL201)	This study
pTOS	Vector for the marker-free integration of genes into the genome of actinomycetes; oriT, aac(3)IV, int, and attP (VWB) flanked by rox sites	25
pRham	Replicative vector for actinomycetes; amp, tsr, and oleS, oleE, oleL, and oleU under ermE* promoter	37
pTOS-Rham	pTOS derivative with oleS, oleE, oleL, and oleU under ermE* promoter	This study
pUWL-Dre	Replicative vector for actinomycetes; oriT, amp, tsr, and dre gene under ermE promoter (pUWL201)	25
pUWL201	Replicative vector for actinomycetes; pIJ101 replicon, amp, tsr, and ermE promoter	38
pUWL-A [pUWL-oriT-aac(3)IV]	Replicative vector for actinomycetes; oriT, amp, aac(3)IV, ermE (pUWL201)	39
pUWL-A-ses60310	pUWL-A derivative with ses60310 under ermE promoter	This study
pUWL-A-ses60350	pUWL-A derivative with ses60350 under ermE promoter	This study
pET28a(+)	Expression vector with aphII and N-terminal hexahistidine affinity tag	Novagen
pET28a-ses60310-N-his-Tev	pET28a(+) derivative for expression of ses60310 with cleavage site for TEV protease	This study

^aaac(3)IV, apramycin resistance-conferring gene; aadA, spectinomycin resistance-conferring gene; amp, ampicillin resistance-conferring gene; aphII, kanamycin resistance-conferring gene; attP, attachment site on plasmid for phage integration; dre, gene encoding Dre recombinase; ermE, constitutive promoter in streptomycetes; ermE*, upregulated variant of ermE promoter; gusA, gene encoding β-glucuronidase in actinomycetes; int, phage integrase gene; loxLE and loxRE, recognition sites for Cre recombinase containing mutations within the inverted repeats; oleE, dTDP-glucose 4,6-dehydratase gene; oleL, dTDP-4-keto-6-deoxyglucose 3,5-epimerase gene; oleS, dTDP-d-glucose synthase; oleU, dTDP-4-ketohexulose reductase; oriT, origin of transfer; rox, recognition site for Dre recombinase; ses44520, ses45900, ses47640, ses49200, ses60310, and ses60350, genes coding for glycosyltransferases from S. espanaensis belonging to CAZy family 1; tsr, thiostrepton resistance-conferring gene; WT, wild type; TEV, tobacco etch virus.

Inactivation of glycosyltransferase genes in the genome of S. espanaensis.

The glycosyltransferase genes were inactivated by the insertion of a spectinomycin resistance cassette via homologous recombination. Therefore, two homologous regions of approximately 2.5 kb, including a few base pairs of either the beginning or the end of the gene, were amplified and cloned into the suicide vector pKC1132 or pKCXY02. Between the start and the end of the gene, a spectinomycin resistance cassette was cloned. S. espanaensis was conjugated with E. coli 12567(pUZ8002) containing the resulting plasmids. Overlaying the conjugation with apramycin selected for single crossover mutants. These mutants were passaged in liquid medium without antibiotics. After 10 to 12 passages, single colonies were picked on both apramycin- and spectinomycin-containing agar plates. The requested double crossover mutants grew on spectinomycin-containing but not on apramycin-containing agar plates. For the construct pKCΔses60310, the double crossover mutants were identified by blue/white selection as well. Therefore, the single colonies were overlaid with X-Gluc (5-bromo-4-chloro-3-indolyl-β-d-glucuronic acid). The successful gene inactivation was proven by PCR. Feeding experiments were performed to investigate the biotransformation activities of the resulting mutants.

Complementation of S. espanaensis Δses60310 strain with an intact copy of ses60310.

The mutant S. espanaensis Δses60310 strain was conjugated with E. coli 12567(pUZ8002) containing the plasmid pSET-ses60310. The resulting apramycin-resistant mutant was used for feeding experiments.

Development and application of a heterologous test system.

For integration of the dTDP-l-rhamnose biosynthetic genes oleS, oleE, oleL, and oleU into the genome of S. albus, the plasmid pTOS-Rham was cloned. This plasmid was conjugated into S. albus followed by the conjugation of a second plasmid, pUWL-Dre, expressing the dre recombinase gene. The Dre recombinase removed the integrase gene and the apramycin resistance-conferring gene of pTOS-Rham from the genome as already described (25). The resulting mutant S. albus Rham carried the genes responsible for the biosynthesis of dTDP-l-rhamnose marker-free in the genome. Glycosyltransferase genes were cloned into the replicative vector pUWL-A and conjugated into S. albus Rham. The resulting mutants were used for feeding experiments.

Purification of biotransformation products.

For structure elucidation, the biotransformation products were isolated from the crude extracts. Therefore, the extracts were fractionated using Oasis HLB columns (Waters), applying a methanol-water gradient and preparative thin-layer chromatography (silica gel 60 F₂₅₄; Macherey-Nagel) with dichloromethane-methanol-acetic acid (9%:1%:0.1%) or acetonitrile-water-acetic acid (85%:10%:0.1%) as the developing solvent or flash chromatography on LiChroprep RP-18 columns (Merck) with a methanol-water gradient as well. As a final step, the compounds were purified by gel filtration on a Sephadex LH20 column (GE Healthcare) with methanol or by crystallization from the methanol-water mixture. The compounds alipro1, novpro1, novpro3, and novpro5 were purified using preparative HPLC as well.

Sample analysis by HPLC-MS.

HPLC-MS analysis was performed on an Agilent 1100 series LC/MS system with electrospray ionization (ESI) and detection in the positive and negative modes. The LC system was equipped with a Zorbax Eclipse XDB-C8 column (5-μm particle size; 150 mm by 4.6 mm; Agilent) and a Zorbax XDB-C8 precolumn (5-μm particle size; 12.5 mm by 4.6 mm; Agilent), maintained at room temperature. Detection wavelengths of the diode array were 254/360 nm, 480/800 nm, 360/580 nm, 430/600 nm, and 550/700 nm. The gradient consisted of solvent A (acetic acid [0.5%, vol/vol] in acetonitrile) and solvent B (acetic acid [0.5%, vol/vol]). Starting with 80%, B linear gradients (67% B after 9 min, 50% B after 16 min, 30% B after 20 min, and 5% after 24 min) were followed by a hold at 5% B for 6 min and a hold at 80% B for 5 min. The solvent flow was 0.7 ml/min.

Structure elucidation.

Nuclear magnetic resonance (NMR) spectra of alipro2, alic1, alipro4, and aliproN were measured on a Varian VNMR-S 600-MHz spectrometer equipped with 3-mm triple resonance inverse and 3-mm dual broadband probes. Pulse sequences were taken from the Varian pulse sequence library. These spectra are recorded in 150 μl DMSO-d₆ at T = 35°C. NMR spectra of alipro1, emopro1, emopro2, emopro3, ant14pro1, quercpro1, novpro1, novpro3, and novpro5 were measured on a Bruker Avance DRX 400 spectrometer equipped with a 5-mm broadband probe and were recorded in 550 μl DMSO-d₆ at room temperature (around 22°C). Pulse sequences were taken from the Bruker pulse sequence library. The nuclear Overhauser effect spectroscopy (NOESY) spectrum of alipro1 was recorded on a JEOL ECX400 spectrometer, equipped with a 5-mm broadband probe, and was recorded in 600 μl DMSO-d₆ at 40°C. Solvent signals were used as an internal standard (DMSO-d₆: δ_H = 2.5, δ_C = 39.5 ppm). For alipro1, emopro1, emopro2, emopro3, ant14pro1, quercpro1, novpro1, novpro3, and novpro5, high-resolution electron spray ionization mass spectra (HR-ESI-MS) were measured on an LTQ Orbitrap (Thermo Scientific). The absolute configuration of the sugar moiety of quercpro1 was confirmed by comparison of circular dichroic data of 3-O-α-l-rhamnosylquercetine (Roth). The molar circular dichroisms were recorded with 0.26 mM methanolic solutions in cuvettes with a light pass length of 5 mm on a J-810 spectropolarimeter (Jasco) with the following parameters: sensitivity, standard; scanning speed, 100 nm/min; data pitch, 1 nm; bandwidth, 2 nm; response, 2 s; accumulation, 3.

Protein synthesis, purification, and in vitro activity tests. (i) Protein synthesis.

Plasmid pET28a-ses60310-N-his-Tev was cloned for the heterologous expression of ses60310. Expression was accomplished in E. coli BL21 Codon Plus RP(pL1SL2) (17, 20). Optimal production conditions were achieved for cells grown in Luria-Bertani medium at 28°C with shaking at 180 rpm in baffled Erlenmeyer flasks. Overexpression of ses60310 was finally induced by addition of 0.1 mM IPTG (isopropyl-β-d-thiogalactopyranoside) at an optical density at 600 nm (OD₆₀₀) of 0.5 to 0.7. Cells were subsequently cultivated at 20°C for 20 h, before they were harvested by a centrifugation step at 5,000 rpm for 10 min, and the pellet was stored at −20°C.

(ii) Purification of Ses60310.

The cell pellet was resuspended in binding buffer (40 mM Tris-HCl, pH 8.0, 100 mM NaCl, 5 mM MgCl₂, 1 mM TCEP [tris(2-carboxyethyl)phosphine], 10% [mass/vol] glycerol), and subsequent cell disruption was carried out with a French pressure cell press. Afterwards, cell debris was removed by centrifugation (10,000 rpm, 1 h, 4°C), and the protein Ses60310 was purified from the cytosolic fraction by immobilized-metal affinity chromatography (IMAC) with nickel-nitrilotriacetic acid (Ni-NTA) as follows. The supernatant was loaded onto a 5-ml His Trap FF column (GE Healthcare), preequilibrated with binding buffer. Following extensive washing with 7% elution buffer (40 mM Tris-HCl, pH 8.0, 100 mM NaCl, 5 mM MgCl₂, 1 mM TCEP, 10% [mass/vol] glycerol, 500 mM imidazole) to discard impurities, the protein was eluted with 15% elution buffer. Fractions containing Ses60310 were pooled and concentrated to a final volume of 1 ml by centrifugation in a Vivaspin concentrator (Sartorius Stedim) with a molecular mass cutoff size of 10 kDa. As a last chromatographic step, the protein sample was loaded onto a size exclusion HiLoad 16/60 Superdex-200 column (GE Healthcare), preequilibrated with buffer A. Fractions containing Ses60310 were pooled, concentrated, and immediately used for in vitro assays after determination of the protein concentration by UV absorption at 280 nm.

(iii) In vitro activity tests.

Recombinant Ses60310 was deployed to catalyze the reaction of alizarin and dTDP-l-rhamnose to alipro1 and dTDP, respectively. Standard analytical enzyme assays (100 μl) were performed at room temperature for 1 h. One incubation mixture contained Ses60310 (5 μM), alizarin (0.2 mM), Tris buffer (pH 8.1; 0.025 mM), bovine serum albumin (BSA; 0.5 mg/ml), MgCl₂ (2.5 mM), dTDP-l-rhamnose (0.2 mM), and TCEP (1 mM). The pH of the Tris buffer was diversified for the determination of the pH which leads to the largest amount of alipro1. To identify the optimal pH and incubation temperature for the production of alipro1, the reaction was stopped by the addition of 200 μl ethyl acetate followed by 10 μl of 1,4-dihydroxyanthraquinone (0.2 mM in methanol) as an internal standard. The organic phase was subsequently removed and dried. The residue was dissolved in methanol (200 μl), and the amount of synthesized alipro1 was measured by HPLC-MS. The enzyme kinetic parameters K_m and V_max were determined using a photometric assay as already described (26). Phosphoenol pyruvate (0.7 mM; Sigma-Aldrich), NADH (0.15 mM; Roth), pyruvate kinase (3.6 units; Sigma-Aldrich), and lactate dehydrogenase (3.6 units; Sigma-Aldrich) were added to the standard incubation mixture. The amount of synthesized alipro1 was determined indirectly by the continuous photometric measurement of the decrease of NADH at 340 nm (Ultrospec 2100 pro UV/visible spectrophotometer with SWIFT II software).

RESULTS

Whole-cell biotransformation.

A set of compounds, including the anthraquinones alizarin and emodin, the flavonoid quercetin, and the aminocoumarin novobiocic acid were dispensed to S. espanaensis under whole-cell biotransformation conditions. The crude extracts of the fed cultures analyzed by HPLC-MS exhibited new species. Due to similar UV spectra, the detected compounds are supposed to be derivatives of the fed substrates. Additionally, the reduced retention time suggests that these derivatives are more hydrophilic than are the reactants. Consequently, the increase in their molecular mass indicates the presence of a hydrophilic substituent. NMR spectroscopy data confirmed that the biotransformation products are α-l-rhamnosylated derivatives of the dispensed compounds (see the supplemental material).

The biotransformation profile of the feeding experiment with alizarin is depicted in Fig. 1. Structure elucidation revealed that S. espanaensis biotransforms alizarin to alipro1, alic1, and alipro2, the 2-O-α-l-rhamnosylated, the 1,2-O-α-l-dirhamnosylated, and the 3-hydroxylated-2,3-O-α-l-dirhamnosylated derivative, respectively. Table 1 shows a set of anthraquinones and other structural classes, of both biological and synthetic origins. All compounds were successfully converted into their deoxysugar glycosides. Reactants without the OH-group were biotransformed into rhamnosylated products as well, involving additional hydroxylation of the substrate.

Fig 1.

(A) HPLC graph (430 nm) of biotransformation products generated by adding alizarin to an S. espanaensis culture. (B) Chemical structures and UV spectra of alizarin and analyzed biotransformation products.

Glycosyltransferase genes from the genome of S. espanaensis.

In 2012, the complete genome of S. espanaensis was sequenced. In accordance with the general eggNOG functional categories, 213 of the 8,427 protein-coding sequences are connected to carbohydrate transport and metabolism (14). Altogether, 1.26% of the protein-coding sequences, 106 genes, code for glycosyltransferases. The gene products of 86 glycosyltransferase genes are expected to be involved in common cellular processes like cell wall synthesis or DNA formation. However, S. espanaensis possesses glycosyltransferase genes whose gene products participate in secondary metabolism as well. Enzymes responsible for the glycosylation of secondary metabolites and external small molecules usually exhibit protein sequence homology to the glycosyltransferases classified in CAZy family 1 (27, 28). In the genome of S. espanaensis, the gene products of 20 glycosyltransferase genes show homology to glycosyltransferases of this family (Ses11290, Ses44490, Ses44520, Ses45900, Ses47640, Ses49200, Ses54000, Ses58720, Ses58730, Ses58740, Ses58750, Ses58760, Ses58770, Ses58780, Ses58790, Ses58800, Ses58810, Ses60310, Ses60350, and Ses68280). Ten of these 20 glycosyltransferase genes (ses58720 to ses58810) are located in the saccharomicin biosynthetic gene cluster (14), while all other genes are distributed throughout the genome. The promiscuous glycosyltransferase(s) responsible for rhamnosylation of different substrates was expected to be part of CAZy family 1 as well. Figure 2 shows a phylogenetic tree of the 20 glycosyltransferases from S. espanaensis belonging to CAZy family 1 and several other enzymes either known to transfer rhamnose or showing substrate flexibility.

Fig 2.

Phylogenetic tree of the 20 glycosyltransferases from S. espanaensis belonging to CAZy family 1 and known glycosyltransferases from other actinomycetes. These selected glycosyltransferases represent enzymes which were proven to transfer rhamnose or to be flexible concerning the acceptor molecule. Alignment of the amino acid sequences of glycosyltransferases was performed by using the Clustal X 2.0 program (53). The phylogenetic tree was then constructed with Clustal X 2.0 and Tree View software (54) using the neighbor-joining method. The percentage of replicate trees in which the associated glycosyltransferases clustered together in the bootstrap test (1,000 replicates) is shown next to the nodes. The branch length indicates the evolutionary distance between different enzymes. The accession numbers for each protein are as follows: AknN, AAF73450; AraGT, ABL09968; CalG1, AAM70336; DesVII, AAC68677; ElmGT, CAC16413; EryBV, AAB84072; MycB, BAC57037; OleD, ABA42119; OleG2, CAA05642; OleI, ABA42118; RebG, BAC15749; Ses11290, YP_007035328; Ses44490, YP_007038603; Ses44520, YP_007038606; Ses45900, YP_007038742; Ses47640, YP_007038902; Ses49200, YP_007039063; Ses54000, YP_007039532; Ses58720, YP_007040002; Ses58730, YP_007040003; Ses58740, YP_007040004; Ses58750, YP_007040005; Ses58760, YP_007040006; Ses58770, YP_007040007; Ses58780, YP_007040008; Ses58790, YP_007040009; Ses58800, YP_007040010; Ses58810, YP_007040011; Ses60310, YP_007040160; Ses60350, YP_007040164; Ses68280, YP_007040938; StaG, CAD58668; UrdGT2, AAF00209.

Gene deletion, complementation, and heterologous expression experiments.

In an effort to delineate the promiscuous rhamnosyltransferase(s), gene deletion and gene expression experiments have been performed. Gene inactivation experiments were carried out with genes encoding glycosyltransferases belonging to CAZy family 1 which are located beyond the saccharomicin gene cluster. Mutants with deletions in ses44520, ses45900, ses47640, and ses49200 were still able to rhamnosylate alizarin, while S. espanaensis with a mutation in ses60310 was not able to perform the glycosylation reaction. Figure 3A, panel II, shows that the S. espanaensis Δses60310 inactivation mutant was not able to glycosylate alizarin. Nevertheless, biotransformation products were observed for the feeding of this mutant. alipro3 and alipro4, the 3-hydroxylated and the 2-methoxylated and 3-hydroxylated derivatives of alizarin, respectively, were detected. Additionally, the S. espanaensis Δses60310 strain also did not convert emodin, quercetin, and novobiocic acid to rhamnosylated biotransformation products, indicating that Ses60310 is the glycosyltransferase responsible for promiscuous glycosylation in S. espanaensis (data not shown). This is also supported by the results for the feeding of the S. espanaensis Δses60310 strain carrying pSET-ses60310 with alizarin. As shown in Fig. 3A, panel III, the mutant is able to glycosylate alizarin. The complementation of the S. espanaensis Δses60310 deletion mutant with an intact copy of the gene ses60310 restored the biotransformation activity, leading to a biotransformation profile of the S. espanaensis Δses60310 strain carrying pSET-ses60310 comparable to the profile of the wild type.

Fig 3.

(A) HPLC graph (430 nm) of the crude extracts of cultures of S. espanaensis wild type (I), S. espanaensis Δses60310 strain (II), and S. espanaensis Δses60310 strain carrying pSET-ses60310 (III) fed with alizarin. (B) Chemical structure of the biotransformation products alipro3 and alipro4 and the UV spectrum of alizarin overlaid with those of alipro3 and alipro4, respectively. The structure of alipro3 is based on HPLC-MS data; the structure of alipro4 is based on HPLC-MS and NMR data.

For a final proof, ses60310 was expressed in a heterologous test system. Therefore, the genes oleS, oleE, oleL, and oleU, responsible for the formation of dTDP-l-rhamnose and controlled by the ermE* promoter, were integrated into the genome of S. albus using a marker-free integration system (25). The overexpression of ses60310 in the resulting strain S. albus Rham led to the mutant S. albus Rham(pUWL-A-ses60310). As shown in Fig. 4, alizarin was converted to glycosylated products by this strain, indicating as well that Ses60310 is responsible for the glycosylation reactions. Besides the rhamnosylated biotransformation product alipro1, another new compound, named aliproN, was detected. Structure elucidation by NMR analysis demonstrated that this compound is a 2-O-l-olivosylated derivative of alizarin. When emodin, quercetin, and novobiocic acid were fed to the strain, rhamnosylated and olivosylated derivatives were detected as well, indicating that Ses60310 is a highly flexible enzyme (data not shown).

Fig 4.

(A) HPLC graph (430 nm) of the crude extracts of cultures of S. albus wild type (I), S. albus Rham (II), S. albus Rham(pUWL-A-ses60350) (III), and S. albus Rham(pUWL-A-ses60310) (IV) fed with alizarin. The MS of the peak at 13.7 min (*) refers to a biotransformation product which is rhamnosylated and olivosylated. (B) Chemical structure of the biotransformation product aliproN and its UV spectrum overlaid with that of alizarin. The structure of aliproN is based on HPLC-MS and NMR data.

Sequence of ses60310-surrounding genes.

The sequence of a 10-kb DNA fragment containing ses60310 revealed that the gene is surrounded by a regulatory gene encoding an ArsR regulator, by genes involved in gluconate metabolism, by a glycosyltransferase gene, by a peptidase gene, and by a sugar-specific regulatory gene encoding a protein belonging to the TrmB family (Fig. 5).

Fig 5.

Environment of the flexible glycosyltransferase gene ses60310 in the genome of S. espanaensis.

The presence of putative dNDP-l-rhamnose biosynthetic genes in the genome of S. espanaensis was indicated by sequence comparison with known rhamnose biosynthetic gene products. These genes are distributed throughout the genome, and none of them is located close to ses60310.

Purification and in vitro characterization of Ses60310.

The gene ses60310 was heterologously expressed in an E. coli strain coexpressing Streptomyces chaperonins (17, 20). Thereby, soluble protein was produced that was purified to homogeneity. The protein Ses60310 was subsequently deployed in an in vitro assay. The enzyme catalyzed the reaction of alizarin and dTDP-l-rhamnose to alipro1 and dTDP, respectively. At pH 8.8 and a 37°C incubation temperature, the production of alipro1 reached its peak value. The enzymatic parameters of this reaction were approximated to K_m(alizarin) of 31 μM and V_{max(alizarin)} of 0.0015 s⁻¹.

DISCUSSION

The bioactivity of glycosylated natural products synthesized by actinomycetes can often be attributed to their saccharide moiety. Recently, glycodiversification studies, with the overall goal of generating novel compounds with altered activities or improved properties, have attracted a lot of attention.

Glycosyltransferases, which catalyze the attachment of a sugar moiety to an aglycon, are key enzymes for glycodiversification. Some natural product glycosyltransferases are sufficiently promiscuous for use in altering the glycosylation patterns, but most of them show strict substrate specificity, which is a limiting factor in natural product diversification. Examples for flexible glycosyltransferases are listed in Table 3. Most flexible glycosyltransferases show flexibility toward the sugar donor, but only a few glycosyltransferases, including OleD and OleI, exhibit remarkable aglycon promiscuity. In order to find glycosyltransferases with broad substrate specificity, we became interested in Saccharothrix espanaensis. We observed that phenolic compounds fed to the strain were converted to modified compounds with high conversion rates. Structure elucidation revealed mono- and dirhamnosylated derivatives. The acceptor spectrum included a diverse range of “drug-like” structures, such as anthraquinones, a flavonoid, and an aminocoumarin derivative.

Table 3.

Flexible glycosyltransferases from actinomycetes

Glycosyltransferase	Description	Reference(s)
AknN	l-2-Desoxyfucosyltransferase: some sugar flexibility, some aglycon flexibility	40
AraGT	l-Rhamnosyltransferase: broad sugar flexibility	41
AveB1	l-Oleandrosyltransferase: broad sugar flexibility	42, 43
CalG1	l-Rhamnosyltransferase: broad sugar flexibility	42, 43
CalG3	4,6-Dideoxy-4-hydrxylamino-α-d-glucosyltransferase: broad sugar flexibility	44
DesVII	d-Desosaminyltransferase: some sugar flexibility, broad macrolide flexibility	42, 43
ElmGT	l-Rhamnosyltransferase: broad sugar flexibility	45
EryBV	l-Mycarosyltransferase: broad sugar flexibility	46
GtfA	4-epi-Vancosaminyltransferase: some sugar flexibility	11
GtfC	4-epi-Vancosaminyltransferase: broad sugar flexibility, some aglycon flexibility	11
GtfD	l-Vancosaminyltransferase: broad sugar flexibility, some aglycon flexibility	42, 43
GtfE	d-Glycosyltransferase: broad sugar flexibility, some aglycon flexibility	42, 43
OleD	d-Glucosyltransferase: broad sugar flexibility, broad aglycon flexibility	47
OleG2	l-Oleandrosyltransferase: some sugar flexibility, some aglycon flexibility	48
OleI	d-Glucosyltransferase: some sugar flexibility, broad aglycon flexibility	47
RebG	d-Glucosyltransferase: some sugar flexibility, some aglycon flexibility	49
Ses60310	l-Rhamnosyltransferase: some sugar flexibility, broad aglycon flexibility	This study
StaG	l-Ristosaminyltransferase: broad sugar flexibility	50
UrdGT2	d-Olivosyltransferase: broad sugar flexibility, some aglycon flexibility	51
VinC	d-Vicenisaminyltransferase: broad sugar flexibility, some aglycon flexibility	52

The desired glycosyltransferase gene(s) was expected to be similar to flexible glycosyltransferases involved in natural product glycosylation or to enzymes which transfer rhamnose to a phenolic compound. In the genome of S. espanaensis, 20 suitable glycosyltransferase genes were identified. Among these candidates, Ses60310 was detected to be responsible for the rhamnosylation of the compounds used in our studies.

The glycosyltransferase Ses60310 uses different nucleotide-activated sugars, accepts different aglycons, and has a remarkable regioflexibility. The gene ses60310 is not part of one of the 26 secondary metabolite gene clusters in the genome of S. espanaensis (14). Therefore, it is reasonable that Ses60310 is not involved in natural product biosynthesis but might be involved in a defense mechanism protecting the strain by glycosylating natural products entering the cell, comparable to OleD and OleI. Hence, this kind of xenobiotic-glycosylating enzyme seems to be much more suitable for glycodiversification than are biosynthetic enzymes. Since we proved that Ses60310 has some sugar flexibility and accepts a variety of phenolic compounds as the substrates, we identified a valuable enzyme for further studies.