Insight into the structure–function relationship of the nonheme iron halogenases involved in the biosynthesis of 4-chlorothreonine – Thr3 from Streptomyces sp. OH-5093 and SyrB2 from Pseudomonas syringae pv. syringae B301DR

Authors


Correspondence

I. Grgurina, Department of Biochemical Sciences ‘A. Rossi Fanelli’, Sapienza University of Rome, P. le A. Moro 5, 00185 Rome, Italy

Fax: +39 06 4991 7566

Tel: +39 06 4991 7571

E-mail: ingeborg.grgurina@uniroma1.it

Abstract

Molecular cloning of the biosynthetic gene cluster involved in the production of free 4-chlorothreonine in Streptomyces sp. OH-5093 showed the presence of six ORFs: thr1, thr2, thr3, orf1, orf2 and thr4. According to bioinformatic analysis, thr1, thr2, thr3 and thr4 encode a free-standing adenylation domain, a carrier protein, an Fe(II) nonheme α-ketoglutarate-dependent halogenase and a thioesterase, respectively, indicating the role of these genes in the activation and halogenation of threonine and the release of 4-chlorothreonine in a pathway closely reflecting the formation of this amino acid in the biosynthesis of the lipodepsipeptide syringomycin from Pseudomonas syringae pv. syringae B301DR. Orf1 and orf2 show sequence similarity with alanyl/threonyl-tRNA synthetases editing domains and drug metabolite transporters, respectively. We show that thr3 can replace the halogenase gene syrB2 in the biosynthesis of syringomycin, by functional complementation of the mutant P. s. pv. syringae strain BR135A1 inactivated in syrB2. We also provide an insight into the structure–function relationship of halogenases Thr3 and SyrB2 using homology modelling and site-directed mutagenesis.

Database

Nucleotide sequence data have been deposited in the ebi.ac.uk/EMBL/GenBank databases under accession numbers: thr1 (gi:378781338|CCF23454); thr2 (gi:378781340|CCF23456); thr3 (gi:378781341|CCF23457); orf1 (gi:378781342|CCF23458); orf2 (gi:378781343|CCF23459); thr4 (gi:378781344|CCF23460)

Abbreviations
ACP

acyl carrier protein

DMT

drug metabolite transporter

NBY

nutrient broth yeast

NRPS

nonribosomal peptide synthesis

TAIL-PCR

thermal asymmetric interlaced PCR

YMG

yeast malt glucose

Introduction

4-Chlorothreonine is found in nature both as a component of secondary metabolites and as a free amino acid. It is a conserved C-terminal residue of nonribosomally synthesized phytotoxic and antifungal lipodepsinonapeptides produced by many strains of the plant colonizing bacterium belonging to Pseudomonas spp.: syringomycin [1, 2], syringotoxin [3], syringostatin [2], pseudomycin [4] and cormycin [5], as well as of two structural analogues of the antitumor compound actinomycin from Streptomyces fradiae [6]. The presence of chlorine contributes to the antifungal and cytotoxic activities of lipodepsinonapeptides and actinomycin, respectively [6, 7]. As a free amino acid, 4-chlorothreonine has been isolated from Streptomyces sp. OH-5093 and reported to inhibit the growth of radish, sorghum and also Candida albicans [8]. This amino acid was also envisaged as a possible antitumor agent as a result of its ability to inactivate serine hydroxymethyltransferase as a mechanism-based inhibitor [9].

The mechanism of formation of 4-chlorothreonine in Pseudomonas syringae pv. syringae B301DR was clarified in biosynthetic studies of syringomycin. Radiolabelling experiments indicated threonine as a precursor of its chlorinated analogue [10]. Structural and enzymatic studies on SyrB2, a component of the syringomycin synthetase multienzyme system [11], demonstrated the chlorination of the threonine methyl group [12, 13] and led to the discovery of a novel class of halogenases. Indeed, the direct insertion of halogen atom on unactivated carbon centres has subsequently been shown in the formation of several secondary metabolites and also of biosynthetic intermediates in nonribosomal and mixed nonribosomal/polyketide pathways [14, 15]. The substrates of halogenases are either amino acids and their derivatives tethered to thiolation domains, as in the biosynthesis of syringomycin [12, 13], coronatine [16], kuntzerizide [17, 18] and 4,4′-dichloroaminobutyrate [19], or polyketide precursors bound to acyl carrier proteins (ACP) as in the biosynthesis of curacin [20] and jamaicamide [21].

The enzymes catalyzing the halogenation of unactivated carbon centres belong to the superfamily of Fe(II) nonheme α-ketoglutarate-dependent enzymes that are involved in a number of oxidative reactions [14, 15]. However, by contrast to the oxygenases, where iron is coordinated by two histidine residues and a carboxylate, in the halogenating enzymes, the carboxylate ligand in the iron coordination sphere is replaced by chloride. In the resting state, the iron in the active site is also bound to α-ketoglutarate and a water molecule [14]. The initial part of the catalytic cycle is common to both classes of Fe(II) nonheme enzymes. Following the entry of the substrate into the active site, the water molecule is removed, allowing the binding of the dioxygen to the iron centre. The subsequent formation of the high valent ferryl(IV)-oxo species by cleavage of the dioxygen bond is simultaneous with oxidative decarboxylation of α-ketoglutarate and the formation of succinate. The ferryl(IV)-oxo intermediate abstracts hydrogen atom from the substrate forming a substrate radical and an Fe(III)-OH species. In hydroxylases, the cycle is completed by the rebound of the hydroxyl radical from the resultant X-Fe(III)-OH complex to the substrate radical, whereas, in halogenases, the radical rebound involves the coordinated halogen. These enzymes are active only on substrates tethered to the phosphopantetheine arm of a carrier protein and the interaction with the carrier protein was shown to influence the reactivity of the halogenase SyrB2 by triggering the chloroferryl formation [22].

To evaluate whether an Fe(II) nonheme α-ketoglutarate-dependent halogenase was involved in the formation of free 4-chlorothreonine in Streptomyces sp. OH-5093, we used PCR primers homologous to the conserved sequences in the iron nonheme halogenases to amplify a fragment from genomic DNA. This was then used to identify the biosynthetic cluster of 4-chlorothreonine in Streptomyces sp. OH-5093. The cluster contains the genes thr1, thr2, thr3 and thr4 that encode a free-standing adenylation domain, a carrier protein, a Fe(II) nonheme halogenase and a thioesterase, respectively. It also contains orf1, which shows sequence similarity with Ala/Thr-tRNA synthetases editing domains and orf2 belonging to the superfamily of drug metabolite transporter (DMT) proteins. Moreover, we provide insight into the structure–function relationship of the halogenases SyrB2 and Thr3 using homology modelling, site-directed mutagenesis and functional complementation.

Results

Molecular cloning of the gene cluster involved in the biosynthesis of 4-chlorothreonine

To capture the 4-chlorothreonine biosynthetic gene cluster, a PCR-based approach was chosen. A pair of degenerate primers (Table 1) deduced from highly conserved regions of Fe(II) nonheme halogenases shown in the alignment (Fig. S1) was used to obtain a gene fragment homologous to syrB2. This fragment was then labelled and used as a probe in Southern blot analysis of Streptomyces sp. OH-5093 genomic DNA. Streptomyces sp. OH-5093 total genomic DNA was digested with several restriction enzymes. The best results were obtained with BamHI, which produced a band of approximately 5.5 kb. A partial Streptomyces sp. OH-5093 genomic library was constructed by cloning BamHI-derived DNA fragments in pBluescript and the library was screened with the same probe used for the Southern blot analysis. The sequencing of the 5.5-kb DNA genomic fragment identified six ORFs: thr1, thr2, thr3, orf1, orf2 and thr4.

Table 1. Sequences of oligonucleotides used in the present study. Wobble positions: M(AC); R(AG); W(AT); S(GC); Y(CT); N(AGCT)
Oligonucleotide nameSequence
N65-D71 ACCTACGACCGSCACCTSGAY
G194Y200 GTAGTCGTASCCGWASAMSCC
GCAD2 GTSGRCWGRSMCGSAT
GCAD9 TCRGSYGWCNGSNSTG
GCAD11 TGSWGNGCNRSWCG
GCAD12 GASYGWCSRGWGNSTC
SP1-R TCCGCGCCGGTGCTCAGCGCACC
SP2-R GAAGGAGAACAGGGCGGGAGCAGGCCTCG
SP3-R TCCGCGCCGGTGCTCAGCGCACC
thr3-F GGATCCATGAGCACCAGTGAG
thr3flag-F GGATCCATGGACTACAAGGATGACGATGACAAGATGAGCACAGTGAG
thr3-R AAGCTTTCAGAAGCGCCGCGGAGT
syrB2-F GGATCCATGAGCAAAAAATTCGC
syrB2-R AAGCTTTCAGACCGCTTCGAATT
mEsyrB2-F GCAGCCTTAAGATGAGCAAAAAATTCG
mEsyrB2-R TCAGTAAGCTTGCGCGAGGCATTTGTGTAG
N123A-F CCAGGCCGACACCTTCGCCGCTGCCTCCGG
N123A-R CGGAGGCAGCGGCGAAGGTGTCGGCCTG
Y107A-F ACCGAGTTCTTTCCCAAAGCTCCGGGCGATGAAGGCAC
Y107A-R GTGCCTTCATCGCCCGGAGCTTTGGGAAAGAACTCG
E102A-F CTCTGCTGGCGTACCGCGTTCTTTCCCAAATATCC
E102A-R CCGGATATTTGGGAAAGAACGCGGTACGCCAGCAGAG
F121A-F CCAGGCCGACACCGCCGCCAATGCCTCCGG
F121A-R CGGAGGCATTGGCGGCGGTGTCGGCCTG
F195A-F GTGCGTCGCGGTGCCTTCGGCTATGACTATCGC-
F195A-R GATAGTCATAGCCGAAGGCACCGCGACGCAC
F196A-F GATAGTCATAGCCAGCGAAACCGCGACGCAC
F196A-R GTGCGTCGCGGTTTCGCTGGCTATGACTATCGC
R40A-F ACCTGGAAGCGCACCGCGCTGCGCCTGCTCGAC
R40A-R GTCGAGCAGGCGCAGCGCGGTGCGCTTCCAG
Y178A-F AAGCGCATGACCGCCGAGCCCGATG
Y178A-R CATCGGGCTCGGCGGTCATGCGC
N126A-F GGCGGCGACGTTCGCCGCCGCCAGCAAGAAGGC
N126A-R GGCGGCGACGTTCGCCGCCGCCAGCAAGAAGGCCC
Y110A-R GGAGTTCTTCCCCAAGGCCCCGGGTGACGAGGG
Y110A-F CCTCGTCACCCGGGGCCTTGGGGAAGAACTC
E105A-F CTGGCGCACGGCGTTCTTCCC
E105A-R GAAGAACGCCGTGCGCCAG
F124A-F CACCAGGCGGCGACGGCCGCCAACGCCAGC
F124A-R GCTGGCGTTGGCGGCCGTCGCCGCCTGG
F201A-F GCGCGGCTTCGCCGGCTACGACTACCGTCAGC
F201A-R CTGACGGTAGTCGTAGCCGGCGAAGCCGCG
R44A-F AACACCATCGCGCGCCAGCTG
R44A-R CAGCTGGCGCGCGATGGTG
Y184A-F ATGGACGCCGACCCCAACGCC
Y184A-R GGCGTTGGGGTCGGCGTCC

The inspection of the sequence of thr1 showed the presence of several core motifs characteristic of adenylation domains [23] and, in particular, the residues determining the threonine nonribosomal code [24], indicating that Thr1 is a threonine-activating free-standing adenylation domain. However, the sequence of thr1 was incomplete at the 5′-region, lacking A1 and A2 core motifs. To obtain the full sequence of this gene, a thermal asymmetric interlaced PCR (TAIL-PCR), adapted to amplify genes with high GC content, was performed [25]. This method is an efficient tool for the recovery of DNA fragments adjacent to known sequences. It utilizes three nested specific primers in successive reactions, together with a shorter arbitrary degenerate (GCAD) primer to improve the relative amplification efficiencies of specific nucleotide sequences. The primer sequences are shown in Table 1 and the PCR conditions are provided in Table 2. Two PCR fragments of 750 and 850 bp, amplified by using GCAD12 and GCAD2, were cloned into the pGEMT vector. Sequence analysis demonstrated that both contained the 5′-thr1-lacking sequence, together with a promoter region. Thr1 encodes a protein containing 501 or 529 amino acids residues. Indeed, the starting codon could be either GTG (288–290) or ATG (372–374) located 5 bp downstream of the putative ribosome-binding sites: GGAG and AGGGC, respectively. The gene thr2 (270 bp) encodes a free-standing phosphopantetheine-binding carrier protein (89 amino acids) where the sequence DNFLDLGGHS, containing the consensus motif characteristic of thiolation domains, can be observed [23]. The gene thr3 (957 bp) that was targeted in our probing strategy, encodes an Fe(II) nonheme α-ketoglutarate-dependent halogenase (318 amino acids) sharing 65% amino acid identity with the halogenase SyrB2 [11]. The gene thr4 (765 bp), positioned at the 3′-end of the cluster, is assumed to start with a GTG codon that is preceded by a ribosome-binding site (GGAGG) located 7 bp upstream. It encodes a protein (254 amino acids) belonging to the family of thioesterases, as indicated by the sequence analysis and, in particular, by the presence of the conserved motif GxSxG (GHSMG) [23]. It is reasonable to assume that these ORFs indicate the biosynthetic sequence of adenylation, thiolation and halogenation of threonine followed by the release of 4-chlorothreonine, a pathway similar to that operating in the biosynthesis of syringomycin where the two-step activation and the halogenation of threonine are catalyzed by a didomain adenylation–thiolation module SyrB1(A-T) and the halogenase SyrB2, respectively [12, 13]. The cluster contains, located between thr3 and thr4, two additional ORFs: orf1 (657 bp) and orf2 (1008 bp). Orf1 is predicted to encode an Ala/Thr-tRNA synthetase free-standing editing domain (218 amino acids). The alignment with the sequences of most similar proteins whose 3D structures have been solved shows the presence of the conserved motifs HXXXH and CXXXH characteristic of alanyl- and threonyl-tRNA editing domains [26, 27] (Fig. S2). Orf2 encodes a protein (335 amino acids) homologous to DMT proteins [28]. The scheme of the thr cluster is shown in Fig. 1 and the most similar sequences are reported in Table 3.

Figure 1.

Scheme of the biosynthetic cluster for the production of 4-chlorothreonine in Streptomyces sp. OH-5093.

Table 2. Settings for GC TAIL-PCR. GCADX represents one of the GCAD primers reported in Table 1
StepsCyclesThermal programsMaterials
Primary194 °C, 5 min 
594 °C, 30 s; 70 °C, 30 s; 72 °C, 2 min5 μm GCADX primer
194 °C, 30 s; 30 °C, 3 min; ramping to 72 °C, 0.3 °C·s−1; 72 °C, 2 min

0.15 μm SP1-R primer

0.2 mm each dNTPs

1594 °C, 30 s; 70 °C, 30 s; 72 °C, 2 min 
 94 °C, 30 s; 70 °C, 30 s; 72 °C, 2 min 
 94 °C, 30 s; 50 °C, 1 min; 72 °C, 2 min 
Secondary194 °C, 5 minSame as the primary PCR reaction except for SP2-R primer (0.2 μm)
1594 °C, 30 s; 70 °C, 30 s; 72 °C, 2 min
 94 °C, 30 s; 70 °C, 30 s; 72 °C, 2 minTemplate (50-fold diluted primary PCR products)
 94 °C, 30 s; 50 °C, 1 min; 72 °C, 2 min
172 °C, 7 min 
Tertiary The same as the secondary thermal programSame as the primary PCR reaction except for SP3-R primer (0.2 μm)
Template (50-fold diluted secondary PCR products)
Table 3. Predicted functions of proteins present in 4Cl-threonine biosynthetic gene cluster in Streptomyces sp. OH-5093
Predicted polypeptideSize (amino acids)Predicted functionOrigin of the most similar proteinsIdentity/similarity (%)Accession number
Thr1529Peptide synthetaseStreptomyces hygroscopicus ATCC 53653 (ZP_05514100.1)60/71 CCF23454
Thr289Thiolation domainStreptomyces hygroscopicus ATCC 53653 (ZP_05514099)52/71 CCF23456
Thr3318Chlorinating enzymeStreptomyces hygroscopicus ATCC 53653 (ZP_05514098)73/84 CCF23457
Orf1218Threonyl-alanyl tRNA synthetase related proteinProvidencia rettgeri (ZP_06125639.1)41/61 CCF23458
Orf2335DMTFrankia alni ACN14a (YP_716816)81/91 CCF23459
Thr4254Thioesterase domainStreptomyces hygroscopicus ATCC (ZP_05514097)57/68 CCF23460

In trans functional complementation assays of P. s. pv. syringae strain BR135A1 inactivated in the gene syrB2 with gene thr3

Previous work from our laboratory (M. R. Fullone & I. Grgurina, unpublished results) indicated that a P. s. pv. syringae BR135A1 strain, selectively inactivated in the syrB2 and recA genes and impaired in syringomycin production, could be complemented in trans by the gene syrB2. The construction of the syrB2 knockout strain and the functional complementation experiments are described in the present study. To assay the capability of the gene thr3 to substitute the gene syrB2 in the biosynthesis of syringomycin, the gene thr3 was cloned into the plasmid pKm12 [29] and the obtained construct was electroporated into P. s. pv. syringae BR135A1. Both halogenase genes, thr3 and syrB2, can restore the production of syringomycin, as monitored by the antifungal activity test (Fig. 2). The availability of a simple biological assay for the production of syringomycin was exploited for structure–function studies of these halogenases using site-directed mutagenesis and in silico studies, as described below.

Figure 2.

Antifungal activity assay against R. pilimanae for syringomycin production by three strains of P. s. pv. syringae: (1) inactivated in the gene syrB2: BR135A1; (2) complemented with the gene syrB2: BR135A1pKm12syrB2; and (3) complemented with the gene thr3: BR135A1pKm12thr3.

In silico studies of the interaction of the halogenases SyrB2 and Thr3 with SyrB1-T

To identify the amino acid residues relevant for the interaction of the two threonine halogenases SyrB2 and Thr3 with the trimodular substrate, namely threonine tethered to the phosphopantetheine arm of the thiolation domain, in silico models of the proteins involved were built and docking experiments were carried out. The homology model of Thr3 was constructed using the crystal structure of SyrB2 in complex with Fe(II), chloride and α-ketoglutarate [13] as a template. As expected, the 3D structures of the two halogenases, which exhibit 65% identity and share the same substrate, are very similar and the active sites are almost identical. The thiolation domain SyrB1-T was then modelled using as a template the 3D structure of the H state of the TycC3 thiolation domain [30], which shows the highest similarity with the SyrB1-T domain (28% sequence identity). The docking simulations between SyrB2 and the substrate allowed the identifiation of several residues, located in different regions of the halogenase, which are probably important for the reaction. The relevance of some of these residues and the corresponding residues in Thr3 was then checked, as described below.

In the amino acid-binding pocket, four residues important for the stabilization of threonine in the active site were identified: E102(SyrB2)/E105(Thr3), N123(SyrB2)/N126(Thr3), F121SyrB2/F124(Thr3) and F195(SyrB2)/F200(Thr3). Their interactions with the substrate amino acid are shown in Fig. 3. The glutamic acid occupying positions 102 and 105 in SyrB2 and Thr3, respectively, could stabilize the threonine residue by electrostatic interaction between the carboxyl group of the glutamate side chain and the amino group of the substrate amino acid. The asparagine residue N123(SyrB2)/N126(Thr3) could stabilize the hydroxyl group of threonine displacing a water molecule which, according to the crystal structure analysis of SyrB2, forms a hydrogen bond with the asparagine side chain [13]. The methyl group of threonine could be stabilized by hydrophobic interactions with phenylalanine residues F195(SyrB2)/F200(Thr3) and F121SyrB2/F124(Thr3), which are also involved in the interaction with chloride in the crystal structure of SyrB2 [13]. Among the residues lining the phosphopantetheine channel, F196(SyrB2), which had been envisaged as limiting the access to the tunnel [13] (corresponding to F201 in Thr3), is shown. The docking between SyrB2 crystal structure and SyrB1-T homology model indicated several short peptides located on the surface of the halogenase at a distance < 4 Å from the thiolation domain. The residues comprised between 36–46, 53–65 and 178–184 in SyrB2 (corresponding to residues 40–50, 57–68 and 184–190 in Thr3) are in contact with the region predicted as a loop connecting helices I and II in the T domain, containing the active site residue Ser45 with the attached 4′-phosphopantetheine, whereas 103–107 in SyrB2 (106–110 in Thr3) and 240–246 in SyrB2 (246–251 in Thr3) are in contact with regions predicted as helix II and loop III. Figure 4 shows the contacts between R40 and Y107 in SyrB2 (corresponding to R44 and Y110 in Thr3) and the SyrB1 thiolation domain, which are the first two residues that we mutated at the beginning of our work on the putative protein–protein interaction regions.

Figure 3.

Docking simulation of the interaction between the SyrB2 active site and threonine bound to the phosphopantetheine arm. Nitrogen atoms are shown in blue, oxygen in red, sulfur in orange, and carbon in white. An analogous image is obtained when the homology model of Thr3 is used. The possible interaction of F195 (F200 in Thr3) and the methyl group of threonine is shown. In the background, residues R40 (R44 in Thr3) and Y107 (Y110 in Thr3) are indicated, which are in contact with the thiolation domain.

Figure 4.

View of the docking between the halogenase SyrB2 or Thr3 and the SyrB1-T thiolation domain, showing the contacts with the residues R40(SyrB2)/R44(Thr3) and Y107(SyrB2)/Y110(Thr3).

Mutational analysis of the interaction of the halogenases SyrB2 and Thr3 with threonine tethered to the phosphopantetheine arm of the thiolation domain

The effect on chlorinating activity of the mutations in Ala of selected residues of iron nonheme halogenases SyrB2 from P. s. pv. syringae B301DR and Thr3 from Streptomyces sp. OH-5093 was evaluated by a functional complementation assay, which allows crude data to be obtained rapidly with respect to the relevance of structural modifications of a protein on its biosynthetic activity in vivo. The P. s. pv. syringae strain BR135A1, inactivated in the gene syrB2, was transformed with constructs carrying the genes syrB2, thr3 or their mutant forms. SyrB2 and Thr3 fused to maltose-binding protein and to FLAG octapeptide, respectively, were produced by cloning and expressing the halogenases in the plamids pMEKm12 [29] and pKm12FLAG [31], as described in the Materals and methods. Western blot analysis indicated that all the mutated genes were expressed as soluble proteins at equal levels (data not shown). The transformed P. s. pv. syringae BR135A1 strains were grown on nutrient broth yeast (NBY) medium and the production of syringomycin was tested on isopropyl thio-β-d-galactoside-containing potato dextrose agar plates against the fungus Rhodotorula pilimanae. The results of the functional complementation experiments of the SyrB2 and Thr3 mutant forms compared to the respective wild-type forms are shown in Fig. 5.

Figure 5.

Effects of the mutations in halogenases SyrB2 and Thr3 on the antifungal activity of the P. s. pv. syringae strain BR135A1, complemented with the mutant forms of the halogenases. Data are shown as the mean ± SE of three independent experiments. □, SyrB2 and its mutants; ■, Thr3 and its mutants.

The effects of the mutations showed the same trend in both halogenases. In the amino acid binding pocket, the substitution of E102(SyrB2)/E105(Thr3), F121SyrB2/F124(Thr3) and F195(SyrB2)/F200(Thr3) with alanine abolished the production of syringomycin, as shown by a lack of antifungal activity. The mutation of the asparagine residue in the active site caused a strong decrease in antifungal activity. The residual activity was approximately 26–30% for N123A SyrB2 and N126A Thr3 mutants. Two residues situated at the entrance of the phosphopantetheine channel were mutated into Ala: F196(SyrB2)/F201(Thr3) and Y178(SyrB2)/Y184(Thr3). The mutants of the former were inactive in both halogenases, whereas some residual activity (23–26%) was maintained in Y178A and Y184A mutants. We also report the results of two Ala mutations carried out at the beginning of our studies on the possible recognition determinants of the interaction between the halogenase and the thiolation domain, Y107A/(SyrB2)/Y110A(Thr3) and R40A(SyrB2)/R44A(Thr3), which maintained a residual activity of 75% and 26%, respectively.

Discussion

The molecular cloning and sequence analysis of the biosynthetic cluster of free 4-chlorothreonine in Streptomyces sp. OH-5093 indicates a pathway very similar to that operating in the formation of this amino acid in the biosynthesis of syringomycin in P. s. pv. syringae B301DR. In both systems, the phosphopantetheine-tethered threonine residue is chlorinated by an α-ketoglutarate-dependent Fe(II) nonheme halogenase. The differences regard the activation of threonine and the release of the product. In Streptomyces, the first step is carried out by self-standing adenylation (Thr1) and thiolation domains (Thr2) and, in Pseudomonas, by an adenylation-thiolation didomain module SyrB1(A-T) [11, 13]. In syringomycin biosynthesis, 4-chlorothreonine is then inserted into the megasynthetase SyrE with the participation of the aminoacyltransferase SyrC [32], whereas, in Streptomyces sp. OH-5093, it is probably released from Thr2 by the putative thioesterase Thr4.

The occurrence of analogous pathways in the formation of compounds that, in Streptomyces sp., are released as free metabolites and, in Pseudomonas sp., are incorporated into more complex compounds has been observed previously. For example, the plant associated bacterium Streptomyces scabies 87–22 contains a gene cluster involved in the biosynthesis of a metabolite similar to coronafacic acid [33], a component of the phytotoxin coronatine produced by many plant colonizing P. syringae pathovars [34]. Both coronatine and the coronofacic acid-like metabolite are involved in plant–microbe interactions [33, 34]. It would be interesting to determine whether 4-chlorothreonine plays a role in plant–microbe interactions analogous to that shown for syringomycin [35].

In the 4-chlorothreonine cluster, besides the thr genes, two additional genes, orf1 and orf2, located between thr3 and thr4, were found. Orf2, which shows similarity with the superfamily DMT [28], could be involved in the export of the amino acid. Orf1 shows homology with a family of Ala/Thr-tRNA synthetases editing domains [26, 27]. The presence of this gene in a nonribosomal peptide synthesis (NRPS)-like cluster could represent a new example of an evolutionary link between ribosomal and nonribosomal mechanisms, which came to light from a number of studies on the participation of aminoacyl-tRNA synthetases and their homologues in processes not directly related to protein synthesis [36]. For example, several aminoacyl-tRNA synthetases were shown to catalyze the aminoacylation of thiols [37]. Homologues of aminoacyl-tRNA synthetases deprived of the tRNA-binding domain that catalyze the ATP-dependent amino acid activation and the transfer of the aminoacyl adenylate to the phosphopantetheine arm of a carrier protein have also been identified [38]. More recently, an unusual mechanism of peptide bond formation was discovered, namely transfer of the alanyl residue from Ala-tRNA to the peptidyl intermediate tethered to a thiolation domain in the NRPS assembly line of pacidamycin biosynthesis, as catalyzed by the aminoacyltransferase PacB [39]. Investigations of the possible role, if any, of orf1 in the production of 4-chlorothreonine are underway in our laboratory.

Considering the high degree of similarity between the Fe(II) nonheme halogenases from Streptomyces sp. OH-5093 and P. s. pv. syringae B301DR, we attempted the functional complementation of the P. s. pv. syringae strain BR135A1 inactivated in the gene syrB2 by transforming it with a plasmid carrying the Streptomyces halogenase gene thr3. The observed restoration of syringomycin biosynthesis represents the first example of productive interaction in vivo between an Fe(II) nonheme α-ketoglutarate halogenase and a carrier protein originating from different organisms. Previously, studies conducted in vitro showed that the contact between the Pseudomonas halogenase SyrB2 and the noncognate thiolation domain CytC2 from Streptomyces loaded with threonine could promote chloroferryl formation [22]. On the other hand, in the biosynthesis of coronamic acid, selectivity of the halogenase CmaB towards different thiolation domains within the same biosynthetic pathway was observed: it recognizes the free-standing thiolation domain CmaD but not the thiolation domain fused to the adenylation domain CmaA [16]. Similarly, curacin halogenase can distinguish between two components of the multi-enzyme system of curacin biosynthesis, CurA ACPI and CurB ACP assayed as substrate donors, as shown in the studies on molecular interactions between these acyl-carrier proteins and the halogenase [40].

Taking advantage of the availability of a simple functional complementation assay, we evaluated the effects of selected mutations on the biosynthetic activity of SyrB2 and Thr3. The choice of the residues to be mutated in each region of contact between a halogenase and its trimodular substrate was guided by studies performed in silico. The homology model of the thiolation domain SyrB1-T was docked with either SyrB2 crystal structure or the homology model of Thr3. Substitution with alanine of the amino acid residues potentially relevant for molecular recognition either abolished or caused a decrease in activity, affecting both SyrB2 and Thr3 to the same extent, at least within the limits of the sensitivity of the functional complementation assay. According to the docking simulation, the carboxyl group of glutamic acid E102(SyrB2)/E105(Thr3) could form an electrostatic interaction with the amino group of threonine. An analogous interaction was also envisaged by Borowski et al. [41]. Previously it was proposed, on the basis of density functional theory calculations, that this residue could participate in the hydrogen-bond relay also involving R254, which would protonate the Fe(III)-oxo catalytic intermediate, making it inaccessible for the radical rebound step and thus favouring the chloride rebound [42]. Because this Glu residue is conserved in most Fe(II) nonheme α-ketoglutarate-dependent halogenases acting on T-tethered amino acids (Fig. S1), it could be hypothesized that the interaction with the amino group might be a common trait of the substrate-binding interactions in the active sites of these enzymes. This hypothesis gains support from a comparison of our model with the 3D structure of curacin halogenase whose substrate, (S)-HMG-ACP, does not contain an amino group [20] and where the position analogous to Glu in SyrB2 is occupied by Ala (Fig. S3). Our docking simulation indicates that the hydroxyl group of threonine could be stabilized by the interaction with N123(SyrB2)/N126(Thr3) replacing the water molecule which, in the crystal structure of SyrB2, is bound to N123 and chloride [13]. The substitution of N123(SyrB2)/N126(Thr3) with Ala causes a significant decrease in chlorinating activity.

The mutations of both phenylalanine residues, F121(SyrB2)/F124(Thr3) and F195(SyrB2)/F200(Thr3), which could stabilize the methyl group of threonine, abolished the production of syringomycin, indicating the importance of hydrophobic environment in the active site pocket. However, because F121 belongs to the group of residues located in the pocket hosting the chloride ligand in the SyrB2 active site [13], the inactivation of the halogenase in the F121A mutant could be a result of the lack of the stabilization of either chloride or the threonine methyl group, or both. Taken together, the results of the mutations in the threonine binding pocket highlight the importance of the residues lining the halogenase active site for the stabilization of the substrate. The orientation of the substrate in the active site was shown to be crucial for amino acid chlorination [43]. The mutations of two residues located at the entrance of the phosphopantetheine channel, F196(SyrB2)/F201/(Thr3), which controls access to the channel [13], and Y178(SyrB2)/Y184(Thr3), caused inactivation and a strong decrease in activity, respectively. Y178(SyrB2)/Y184(Thr3) belongs to a group of residues forming surface patches, which, according to the simulation, are in contact with the carrier protein. Indeed, the docking indicated several discontinuous binding sites, made up of short peptide fragments that are not adjacent in the sequence but are in spatial proximity as a result of folding. The amino acids in the regions 36–46(SyrB2)/40–50(Thr3), 53–65(SyrB2)/57–68(Thr3), 178–184(SyrB2)/184–190(Thr3) are in contact with the region connecting helices I and II of the T domain, whereas 103–107 in SyrB2 (106–110 in Thr3) and 240–246 in SyrB2 (246–251 in Thr3) are in contact with helix II and loop III. As can be seen from the alignment (Fig. S1), these regions are conserved, in particular the region 103–107(SyrB2)/106–110(Thr3). The decrease in chlorinating activity in Y107 in SyrB2 (Y110 in Thr3) and R40 in SyrB2 (R44 in Thr3) mutants indicates their relevance for the halogenation reaction; it is not surprising that the exchange of only one amino acid does not abolish the activity because several other residues appear to be involved in the interaction. Potential rearrangements occurring upon contact with the carrier protein should also be taken into consideration. The efficient in trans functional complementation of the P. s. pv. syringae strain inactivated in syrB2 with thr3 indicates a productive interaction of the halogenase Thr3 with SyrB1-T and points to possible structural similarities between this heterologous thiolation domain and the putative cognate carrier protein, Thr2. The sequence alignment of these two proteins, which share 27% identity, is shown in Fig. S4 where the conserved residues are highlighted. Further studies aim to identify the structural elements in the T domains that are important for the recognition by the halogenase. Investigations on the interactions of carrier proteins with their protein partners in NRPS assembly lines showed that the specificity can be determined by a few residues. Moreover, conformational plasticity and dynamic changes, which, in turn, can be influenced by the interacting proteins, play an important role [44].

In conclusion, the molecular cloning of 4-chlorothreonine biosynthetic cluster in Streptomyces sp. OH-5093, which reflects the part of the syringomycin gene cluster involved in the production of 4-chlorothreonine in P. s. pv. syringae B301DR, provides a new contribution to studies on similar and differing features of biosynthetic pathways involved in the production of related secondary metabolites in pseudomonads and streptomycetes [33, 45, 46]. Moreover, using an in silico modelling approach, site-directed mutagenesis and functional complementation, we obtained data on the relevance of several selected residues, located in different portions of the halogenases, with respect to the efficiency of the reaction. Analogous results obtained for two halogenases substantiate the validity of the structure–function relationship. The presence of halogen atoms can influence significantly the biological properties of a compound [14, 15]. Among a number of examples, syringomycin [7], clorobiocin [47] and bahlimycin [48] are worthy of note. Knowledge about the structural determinants of molecular recognition between halogenases and their substrates is important in the context of their biotechnological application in the production of new halogenated metabolites by combinatorial biosynthesis.

Materials and methods

Strains and culture conditions

Streptomyces sp. OH-5093 strain was cultured in yeast malt glucose (YMG) broth medium at 30 °C on a rotary shaker at 200 r.p.m. Escherichia coli strain DH10B used for plasmid propagation, were grown in LB liquid medium at 37 °C. Kanamycin (50 μg·mL−1) was used for the selection of E. coli and P. s. pv. syringae transformants. R. pilimanae was cultured in yeast malt broth at 25 °C as described previously [7]. P. s. pv. syringae strain BR135A1 was grown at 25 °C on NBY medium in the presence of piperacillin (50 μg·mL−1). P. s. pv. syringae strain BR135A1 transformed with plasmids carrying the genes syrB2 or thr3 was grown in the presence of both piperacillin (50 μg·mL−1) and kanamycin (50 μg·mL−1).

Genomic DNA isolation

Genomic DNA was isolated by modification of the procedure reported in Pootoolal et al. [49]. In total, 10 mL of Streptomyces sp. OH-5093 cultures in YMG medium (4 g·L−1 yeast extract, 10 g·L−1 malt extract, 4 g·L−1 glucose) were grown at 30 °C for 24 h. Cells were centrifuged at 3500 g for 10 min, washed in 2 mL of SET buffer (20 mm Tris, pH 7.5, 75 mm NaCl, 25 mM EDTA) and lysed in 0.5 mL of the same buffer supplemented with lysozime (2 mg·mL−1 final concentration) at 25 °C for 10 min. 1% Sarcosyl and 100 mg·mL−1 RNAase were then added and the sample was incubated at 65 °C for 10 min. After this step, 600 μg·mL−1 of proteinase K was added and the sample was incubated at 65 °C for 1 h. NaCl was added to 1 m final concentration and the solution was mixed thoroughly and cooled to 37 °C. The preparation was chloroform-extracted and the genomic DNA was precipitated with isopropanol, washed with 70% (v/v) ethanol and dissolved in water.

Genomic DNA digestion and Southern blot

Genomic DNA was digested with BamHI, HindIII or PstI, separated on a 0.8% agarose gel, transferred by capillarity to a Nytran-N filter (Schleicher and Schuell, Dassel, Germany) and probed with a 500-bp DNA fragment corresponding to the a fragment homologous to a portion of syrB2 amplified by PCR with oligonucleotides N65-D71 and G194-Y200 (Table 1). These primers were designed on the basis of alignment analysis of SyrB2 from P. s. pv. syringae B301DR (gi:5748808|AAD50521), chlorinating enzyme from P. s. pv. syringae 642 (gi:302189256|ZP_07265929.1), CmaB from P. syringae (gi:2673891|AAC46036.1), unnamed protein product from Pseudomonas putida KT2440 (gi:26990488|NP_745913.1), BarB1 from Lyngbya majuscula (gi:23452292|AAN32975.1) and BarB2 from Lyngbya majuscula (gi:23452293|AAN32976). PCR with this primer pair was performed, and the amplified fragment (approximately 400 bp) was purified in agarose gel, ligated into the cloning vector pGEM-T (Promega, Madison, WI, USA) and sequenced. Sequence analysis and its predicted amino acids showed that the fragment had 73% identity to the syrB2 gene. The probe was labelled by random priming in the presence of 50 μCi of [α-32P]dCTP and purified by chromatography on a Sephadex G-50 column. Filters were hybridized in 2 mm sodium phosphate buffer (pH 7.2) containing 50% formamide, 5× SSPE (150 mm NaCl, 10 mm sodium phosphate buffer, pH 7, 0.1 mM EDTA), 5× Denhart's solution, 0.5% SDS and 100 μg·mL−1 herring sperm DNA at 42 °C for 18 h. The final washing conditions were 5× SSPE and 0.1% SDS at 37 °C for 15 min.

Construction and screening of the plasmid genomic library

Streptomyces sp. OH-5093 genomic DNA was digested with BamHI. Restriction fragments from selected size areas (5–6 kb) were purified from agarose gels and ligated into pBluescript II KS digested with BamHI. Competent E. coli DH5α cells were transformed to generate pBluescript-based restriction fragment size-libraries. The colonies were screened with the same probe and the same conditions used for Southern blot. Positive clones were sequenced on both strands by automated fluorescent DNA sequencing at MWG Biotech (Ebersberg, Germany).

TAIL-PCR analysis

To obtain the full-length adenylation gene (thr1), a modified TAIL-PCR, named GC TAIL-PCR was used [25]. Three successive PCR reactions were performed using the same degenerated GCAD primer and three nested specific primers (SP1-R–SP3-R). The sequence of primers (GCAD2, GCAD9, GCAD11 and GCAD12; SP1-R, SP2-R and SP3-R) used are reported in Table 1. In the primary PCR reaction, 10 ng of genomic DNA was used as a template. For the secondary and tertiary reaction, 50-fold diluted PCR products from the primary or secondary PCR reaction were used as templates, respectively. The thermal cycling condition are summarized in Table 2. All PCR products were electrophoresed in 1% agarose gels; the secondary and tertiary products showing expected sizes were chosen, purified and cloned into pGEM-T vector for sequencing.

Construction of P. s. pv. syringae BR135 A1 transposon mutant (ΔsyrB2-recA)

To mutate the syrB2 gene, Tn3HoHo1 mutagenesis of plasmid pYM1 [50] was performed as described previously [51] and marker exchange mutagenesis was conducted by transferring a pYM1 derivative (pYM1-135) containing a syrB2::Tn3HoHo1 insert to B301D-R by triparental mating to obtain the syrB2 mutant BR135 (Y. Y. Mo and D. C. Gross, unpublished data). To ensure the stable maintenance of overexpression constructs in strain BR135, a recA mutation was introduced by marker exchange mutagenesis into the mutant of P. s. pv. syringae to generate a recA mutant. The plasmid pEMH9 [52] carries a recA gene mutated by insertion of an Ω fragment containing the spectinomycin and streptomycin resistance genes from P. s. pv. syringae strain B728a. A 10-kb BamHI fragment containing the mutated recA gene was isolated from pEMH9, polished with T4 DNA polymerase and then cloned into pBR325 [53] at the polished EcoRI site to generate pSL66. The mutated recA gene in plasmid pSL66 was introduced into the genomes of strain BR135 by marker exchange mutagenesis. Colonies resistant to spectinomycin and sensitive to tetracycline were selected as potential recA mutants. The recA mutants were confirmed by testing for sensitivity to UV light [52]. Cell suspensions (2 × 108 CFU·mL−1) of the candidate colonies were streaked on NBY agar [54]. Half of the agar plates were covered with thick glass, exposed to UV (254 nm) for 1 s, and then incubated at 25 °C overnight. Parental strains of recA mutants (B301D and BR135) and BR132A1 [55] were included as negative and positive controls, respectively.

Cloning of the gene syrB2

To express syrB2 into P. s. pv. syringae BR135A1 as an N-terminal free protein, pKm12 plasmid was used [29]. The syrB2 gene amplified by PCR from the p601D-1-R using the specific primers syrB2F and syrB2R (Table 1) was directly cloned into pGEM-T vector in accordance with the manufacturer's instructions (Promega). The resulting construct was digested with BamHI and HindIII and subcloned into the pKm12 vector. To express SyrB2 as a maltose-binding fusion protein, a 0.8-kb fragment carrying the syrB2 gene amplified by PCR with the primers mEsyrB2F and mEsyrB2R (Table 1) was digested with EcoR1 and HindIII and cloned into pMEKm12 [31]. The obtained plasmids, pMEKm12syrB2 and pKm12syrB2, were sequenced and electroporated into P. s. pv. syringae BR135A1.

Cloning of the gene thr3

Two pairs of PCR primers thr3F/thr3R and thr3flagF/thr3R (Table 1) were used to amplify the thr3 gene from pBluescript carrying the 5.5-kb genomic DNA fragment containing the 4-chlorothreonine gene cluster. The thr3flagF was designed to introduce the nucleotides coding the FLAG peptide into the amplified fragments. Amplified products were subcloned into pGEMT vector. The resulting plasmids were digested with BamHI and HindIII and the fragments obtained were cloned into pKm12 plasmid [29] previously digested with the same enzymes. The obtained plasmids pKm12thr3 and pKm12FLAGthr3 were electroporated into P. s. pv. syringae BR135A1.

Site-directed mutagenesis of syrB2 and thr3

Site-directed mutagenesis was performed using the QuickChange kit (Stratagene, La Jolla, CA, USA). Puc18 vector carrying syrB2 gene was used as template for mutagenesis of N123, Y107, E102, F121, R40, F196, F195A and Y178A. Mutated plasmids were digested with EcoRI and HindIII and the fragments obtained were subcloned into pMEKm12. pGEMT vector carrying thr3 gene was used as a template for mutagenesis of N126, Y110, E105, F124, R44, F201, F200 and Y184. Mutated plasmids were digested with BamHI and HindIII and the fragments obtained were subcloned into pKm12. The mutagenic primers are reported in Table 1. All the constructs obtained were sequenced and electroporated into P. s. pv. syringae BR135A1.

Bioassay for syringomycin production

Briefly, P. s. pv. syringae strains were grown overnight in 2 mL of NBY liquid medium: BR135A1 in the presence of 50 μg·mL−1 of piperacillin and BR135A1 transformed with plasmids carrying the genes syrB2 or thr3 and the relative mutated forms in the presence of piperacillin 50 μg·mL−1 and kanamycin 100 μg·mL−1. Bacterial cells were harvested by centrifugation and washed once with sterile distilled water. The cells were resuspended in sterile distilled water to a concentration of approximately 2 × 108 CFU·mL−1, and 5-μL aliquots of bacterial suspension were spotted onto potato dextrose agar plates containing 100 μg·mL−1 of piperacillin and kanamycin in the presence of isopropyl thio-β-d-galactoside (5 mm). The plates were incubated for 4 days at 25 °C and the colonies oversprayed with a suspension of R. pilimanae, which is sensitive to syringomycin but not syringopeptin [56]. After 48 h, the P. s. pv. syringae mutant strains were compared for formation of zones of antifungal growth inhibition to R. pilimanae. The effects of the mutations on the ability of the halogenases to restore the production of syringomycin in the syrB2 knockout strain were evaluated by comparing the diameters of the inhibition zones produced in the functional complementation assay carried out with the wild-type and mutant forms of the halogenases.

Computational analysis

Alignment of sequence data was performed using the multialign server (http://multalin.toulouse.inra.fr/multalin/) (multiple sequence alignment with hierarchical clustering) [57]. Predictions of secondary structure and intrinsically disordered regions were carried out as described previously [58]. A monomeric model of Thr3 was constructed using modeller-7 [59], using the crystal structure of the nonheme iron halogenase SyrB2 from P. s. pv. syringae (Protein Data Bank code: 2FCV) as a structural template [13]. 2-oxoglutaric acid, Fe(II) and chloride ions were also inserted in the model as heteroatoms. Ten different models were built and evaluated using several criteria: the model displaying the lowest objective function [60] was taken as the most representative and analyzed with prosaii [61] and verify_3d [62] to monitor its structural consistency. The initial alignment was then subjected to minor changes in an attempt to increase the low score regions, and resubmitted each time to a new modeller session. The final overall verify_3d and prosaii plots showed a structure of good quality. The same protocol was applied to model the phosphopantetheine-binding T domain of SyrB1, on the basis of the solution structure of the TycC3(PCP) from Brevibacillus brevis (Protein Data Bank code: 2GDX) [30]. Sequence alignments and computation of sequence similarity were performed by means of the pymod [63]. The cluspro server was used to model the complex between the crystal structure of SyrB2 and the T domain of SyrB1 [64]. The phosphopantetheine moiety was then docked by means of autodock, version 4.0 (http://autodock.scripps.edu/), keeping all parameters at their default values [65].

Acknowledgements

This work was supported by funds from the Italian Ministero dell'Istruzione, dell'Università e della Ricerca (MIUR) and by the Consorzio Interuniversitario per le Applicazioni di Supercalcolo per Università e Ricerca (CASPUR, Roma, Italy) (std12-008).

Ancillary