Properties of lanK-Based Regulatory Circuit Involved in Landomycin Biosynthesis in Streptomyces cyanogenus S136

Abstract

LanK is TetR-like regulatory protein recently shown to regulate the export and glycosylation of landomycins in Streptomyces cyanogenus S136. Here, several properties of the lanK-mediatcd regulation were deciphered. LanK seems to function as oligomer as evident from experiments in vitro. In vivo, it is able to recognize various landomycins with altered aglycon structure and the minimal concentration of landomycin A sensed by LanK lies in low nanomolar range. Coexpression studies showed that the positive regulatory gene lanI upregulates lanK-dependent lan genes once the negative LanK-regulation is cancelled. Gene lanK can be useful for the construction of biosensor strains for sensitive and specific identification of producers of landomycin-like molecules with long glycosidic chains.

The TetR family of transcriptional repressors is one of the largest groups of regulators governing various aspects of prokaryotic physiology [1]. Although all TetR-like proteins share a similar mode of action, each of them is uniquely tailored to sense and respond to the presence of specific small molecule ligands in a cellular environment. Currently, reliable in silico prediction of functions of regulatory proteins is a major challenge, emphasizing the need for experimental verification. Genes encoding putative TetR-like proteins are especially abundant in actinomycete genomes, and are particularly often found within gene clusters encoding biosyntheses of secondary metabolites. Despite this fact, there is little known on the significance and exact role of TetR-like regulators in antibiotic production.

Recently we have characterized the tetR-like gene lanK involved in the biosynthesis of landomycin A in Streptomyces cyanogenus S136 [2]. Our long-term interest in landomycin biosynthesis stems from the unusual structural features and unique spectrum of bioactivities displayed by the members of landomycin family [3–5]. The potency of landomycins depends on the length of their carbohydrate chain; namely, landomycin A possessing the longest glycoside chain (6 deoxysugars) is the most active antiproliferative agent [6]. The biosynthesis of landomycins with long glycosidic chains is coordinated with their export from the producer cells by the LanK repressor, which interacts with landomycin A, its penta- and triglycosylated intermediates and triggers the expression of exporter gene lanJ [2]. There is keen interest in the generation of landomycins with longer glycoside chain/modified aglycon [7], and we suggest that lanK-based reporter strains can be exploited to discover such novel compounds. The results of our studies reported here further elucidate the function of LanK in context with landomycin biosynthesis and support the idea that the lanK gene can be a useful biotechnological tool.

MATERIALS AND METHODS

Strains and Compounds Used in the Work

Streptomyces albus J1074 was a gift from Prof. J. Salas (University of Oviedo, Spain). Vector pIJ6902 (kindly provided by Prof. S. Cohen; [8]) was used to construct lanI expression plasmid pMO19. Plasmid pMO11 was described previously [2]. E. coli BW25113 (pKD20) is a gift from Prof. J. Beckwith (Harvard Medical School, United States); it was used to carry out RedET-mediated gene replacement as described in following paragraph. Solid oatmeal and soy-mannitol media [9, 10] were used for plating of E. coli—Streptomyces matings and for maintenance of Streptomyces strains. LB agar supplemented with 150 mcg/ml kanamycin was used to plate the freshly harvested spore suspensions of pMO11⁺ S. albus reporter strain and its derivatives. Standard genetic techniques for E. coli and Streptomyces and for DNA manipulations were applied as described [10, 11]. All plasmids were introduced in Streptomyces strains via conjugation as described [10]. Landomycin resistance was analyzed via antibiotic disc diffusion as described by Ostash et al. (2007) [12].

Landomycins E, G and prejadomycin C triglycoside were purified from S. globisporus strains Smy622, GT4.1(lndGT4) and ΔlndE(urdGT2), respectively [9, 13, 14]. Urdamycins A and B were purified from S. fradiae Tü2717. Saquayamycin Z and galtamycin were isolated from Micromonospora sp. Tü6368 [15]. Landomycins A, B, M were purified from S. cyanogenus strains S136 (LaA, LaB) and OJΔGT3 (LaM; [3]). Simocyclinone D8 was obtained from S. antibioticus Tü6040. We purified these compounds according to procedures described in the papers mentioned above. The chromatographic mobility and m/z value of each molecule coincided with the published ones. All compounds were at least 90% pure. Crude extract containing roughly 80% of cosmomycin D was kindly provided by Prof. G. Padilla (University of Sao-Paulo, Brazil) and used without further purification. Doxorubicin, aclacinomycin and nogalamycin were obtained from Sigma. All compounds were tested via TLC prior to the bioassays to rule out the presence of degradation products. Structures of the molecules are shown on Fig. 1.

Fig. 1.

Structural formulae of the molecules used in this study. Eight-angle star marks those metabolites that interact with LanK protein.

In vitro Experiments

DNA of lanI was amplified from S. cyanogenus S136 chromosome using primers lanINdeIup (5'-AAACATATGGGTCAGTTTTCGACGG-3') and lanIMfeIrp (5'-AAACAATTGTCACTGGTTAC-CGAGCCG-3'). The PCR product was cloned into NdeI-EcoRI sites of pIJ6902 to give pMO15. The construct was sequenced to verify the cloning of lanI. Hygromycin resistance gene hyg was amplified from plasmid pHYG1 [10] with primers P1Am-Hyg-up (5'-GTGCAATACGAATGGCGAAAAGCCGAGCT-CATCGGTCAGCCCGTAGAGATTG GCGATCCC-3') and P2Am-Hyg-rp (5'-TCATGAGCTCAGCCAATC-GACTGGCGAGCGGCATCGCATCAGGCGCC-GGGGGCGGTGTC-3'). This amplicon was introduced into E. coli BW25113 (pKD20, pMO15) through electroporation and RedET-mediated replacement of aac(3)IV with hyg was achieved according to standard procedures [16]. The final construct carrying tipAp-lanI fusion and hyg gene instead of aac(3)IV has been marked as pMO19.

A 190 bp DNA fragment containing lanKJp, conditions of LanK purification and the DNA binding assays were as described [2], except that we varied the amount of LanK (0, 70, 100, 200, 300, 600 and 700 ng of the protein per reaction mixture – see Fig. 2.

Fig. 2.

Electrophoretic mobility shift assay of the stoichiometry of LanK binding to lanKJp DNA fragment. (a) Changes in distribution of shifted lanKJp in response to different amounts of LanK. Lane 1, free lanKJp; lane 2, lanKJp + 70 ng of LanK; lane 3, lanKJp + 100 ng of LanK; lane 4, lanKJp + 200 ng of LanK; lane 5, lanKJp + 500 ng of LanK; lane 6, lanKJp + 700 ng of LanK. (b) Pattern of the shifted lanKJp bands in presence of 120 ng of LanK and scheme, explaining the molecular nature of the 5 DNA bands seen on the gel (to the right). Lane 1, purified LanK; lane 2, lanKJp + LanK.

RESULTS

LanK Is Likely to Bind DNA as an Oligomer

In previous in vitro assays 22 kDa LanK repressor was shown to shift the mobility of rather big DNA fragments (more than 400 kDa; [2]). It is possible therefore that LanK acts as a homooligomer, as it was demonstrated for many members of TetR family [1]. Indeed, during FPLC-assisted purification LanK eluted from the Sephadex column as a single 44 kDa fraction (data not shown). We performed additional DNA shift assays where all parameters but the LanK amounts were constant (Fig. 2a). At certain LanK concentrations four shifted DNA bands were detected, an indicative of binding of four LanK molecules to the studied DNA fragment (lanKJp) under saturation conditions. Taking into account the results of FPLC analysis of LanK, we propose that LanK acts as a homodimer and it has two binding sites within lanKJp. Probably, because either LanK binds DNA most efficiently as a dimer or there are just more ways to produce the DNA fragment bound to two LanK molecules (see scheme on Fig. 2b), the shifted band corresponding to binding of two LanK subunits to lanKJp was always the most prominent on the gels (Figs. 2a and 2b).

Improved Reporter Strain for Detection of Landomycins

The regulatory elements involved in antibiotic production/resistance can be used to develop the reporter strains for identification of novel molecules of certain chemical or functional type [17–19]. We demonstrated previously that a S. lividans strain carrying plasmid pMO11 with lanK, lanKJp and kanamycin resistance gene (neo) fused together can be used to detect certain landomycins. However, this strain has several shortcomings – long incubation time (6 days) and, consequently, high background of spontaneous kanamycin resistant clones [2]. We suggested that inefficient penetration of landomycins into S. lividans cells is the main reason for the aforementioned problems, which can be ameliorated by switching to a more convenient strain. Therefore, using antibiotic disc diffusion assay we checked the L’viv university collection of streptomycete strains in order to find one with increased sensitivity to landomycin A. This way we selected strain S. albus J1074 and used it as a host for reporter plasmid pMO11. Under the described conditions (see Methods) pMO11⁺ S. albus strain sporulated abundantly in 4 days around disc with landomycin E and no spontaneous Km^r colonies were observed (Fig. 3).

Fig. 3.

Activity of lanK-based reporter system in S. albus. Halo of growth of Km^r colonies was observed around the disc containing 25 mcg of landomycin E (bottom right), while no growth was seen around the discs impregnated with 25 mcg of landomycin D, urdamycin B and prejadomycin C-triglycoside (from bottom left, clockwise).

Interplay between lanK and lanI Regulatory Genes

Gene lanI encodes a positive pathway-specific regulator of landomycin A biosynthetic genes, particularly those which are targets for LanK negative regulation [20]. The significance of positive regulation exerted by LanI upon lanK-dependent genes remained ambiguous, since the absence of lanJ transcription in lanI-deficient S. cyanogenus mutant (ΔlanI7) could be explained either by absence of landomycins A and B directly responsible for abolishment of LanK repressing function, or by strict dependence of lanJ expression on LanI. The latter speculation seems to be questionable since promoter lanKJp works in S. lividans and S. albus strains apparently without lanI assistance. Therefore, we set out to assess the role of lanI in regulation of lanKJp activity via coexpression studies. For this purpose actinophage φC31-based plasmid pMO19 has been constructed (see Materials and Methods) in which coding region of lanI has been placed under control of inducible tipA promoter (tipAp). We verified pMO19 via its introduction into lanI deficient S. cyanogenus strain ΔlanI7 [20] and induction of lanI expression with thiostrepton (15 ng/ml). The pMO19⁺ ΔlanI7 strain overproduced landomycin A, showing that pMO19 carries functionally proficient lanI gene. Notably, pMO19⁺ ΔlanI7 strain produced landomycin A in absence of an inducer (albeit to a lesser extent than in case of induction; data not shown), pointing to “leaky” expression of lanI from tipAp.

We constructed pMO19⁺ pMO11⁺ S. albus strain and tested the efficiency of its response to landomycins A, E and D. The halo of growth of Km^r pMO19⁺ pMO11⁺ S. albus colonies around landomycin A and E discs on media supplemented with 10 ng/ml thiostrepton was observed after two days, while colonies of the control strain (pMO11⁺ S. albus plus empty vector) have been detected after 4 days. The number of Km^r pMO19⁺ pMO11⁺ S. albus colonies scattered over the plate also increased significantly, probably reflecting the fact that LanI competes with LanK for binding to lanKJp and thus may trigger neo expression. This experiment supports the idea that lanI contributes significantly to lanJ expression once LanK repressor is inactivated. In absence of lanI lanKJp possesses basal transcriptional activity strong enough to drive the expression of reporter gene neo under heterologous conditions.

Sensitivity and Specificity of the Reporter Strain towards the Ligand Molecules

We decided to determine the lowest landomycin A concentration capable of eliciting the growth of Km^r pMO11⁺ S. albus clones. The halo of growth was detected around the disc containing 4 ng of landomycin A, whereas 0.5 ng of it did not induce the growth of the reporter strain. The latter also sensed landomycin E accumulated in the agar plugs cut off the 7-day lawn of wild type S. globisporus 1912 strain grown on Bennett plate. Landomycin E production by 1912 strain on Bennett medium is very poor and can’t be recognized as a change in medium/colony pigmentation or accumulation of antibiotic activity. Therefore, pMO11⁺ S. albus appears to be useful for identification of producers of certain landomycins.

LanK is able to recognize quite different landomycins as ligands [2], however, motifs of their structure critical for interaction with LanK remained largely unknown. We approached this issue by testing different anthracycline and angucycline molecules (Fig. 1) as the potential inducers of kanamycin resistance phenotype in pMO11⁺ S. albus strain. All landomycins carrying at least trisaccharide glycoside chain, even with modified aglycon structures (e.g. landomycins G and M) induced the halo of growth of Km^r colonies around respective discs. Anthracyclines (aclacinomycin, nogalamycin, doxorubicin, cosmomycin D, galtamycin) and angucyclines carrying C-glycosidically attached carbohydrate chains (urdamycins A and B, simocyclinone D8, saquayamycin B, prejadomycin C-triglycoside) were tested in a wide range of concentrations (10–50–500–10³–10⁴–30⁴ ng). Under no conditions did they induce the growth of the reporter strain. We also tested crude extracts from S. fradiae Tü2717 (urdamycin producer), Micromonospora sp. Tü6368 (saquayamycin Z and galtamycin producer) and S. globisporus ΔlndE(urdGT2) (prejadomycin C-glycosides producer) in hope to detect some intermediates capable of inducing the reporter strain. However, none of the extracts activated the expression of the reporter gene (data not shown).

DISCUSSION

SARP-like pathway-specific transcriptional activators are essential for angucycline production [4], however there is limited information on other regulatory genes within angucycline biosynthetic gene clusters. We recently showed that, besides transcriptional activator gene lanI, repressor gene lanK governs late steps of landomycins production in S. cyanogenus S136 [2]. Here we further investigated the properties of LanK repressor and show that it is likely to function as a dimer, as all studied to date TetR-like regulators do [1]. Our work provides first evidence that both regulatory genes control the expression from lanKJp promoter; namely, LanK represses transcription, while LanI accelerates it once lanKJp is relieved from repression. Although our study supports the idea that LanI promotes lanJ expression, it remains inconclusive whether LanI is important to trigger the lanK expression. Work to investigate this is underway.

Development of biosensor systems for metabolite discovery is a very attractive research direction, which may significantly improve the screening of certain molecules in terms of specificity, sensitivity, speed and cost-effectiveness. Our initial studies show that the lanK gene may be useful for the identification of producers of various landomycins. The LanK protein senses these metabolites in a highly specific manner in complex mixtures and at low concentrations that are hardly detectable by current chemical techniques. Landomycins differing in both their glycosidic chains and their aglycon hydroxylation pattern are efficiently recognized by the described biosensor strain, showing that the lanK-mediated discovery of novel landomycins is a feasible task.

While both prejadomycin C-triglycoside and landomycin E have the same carbohydrate moiety (Fig. 1), only the latter is recognized by LanK (Fig. 3). The reason for that, obviously, lies in two major structural differences between these molecules. Firstly, in landomycins deoxysugar moiety is attached O-glycosidically (while C-glycosidically in the prejadomycins). Secondly, their aglycon structures are quite different. Either of these differences (or both of them) may perturb the interaction of prejadomycin C-triglycoside with LanK. A similar line of reasoning is true for the other molecules that turned out to be “non-ligands”. Due to the limited structural diversity of the molecules being tested, the minimal structural motif responsible for their interaction with LanK can’t be inferred from this study. Nevertheless, our study shows that both glycoside and polyketide portions of the landomycins are essential for binding to the repressor, although limited variations in the landomycin glycan and aglycon halves do not abolish their interaction with LanK.