A Streptomyces coelicolor host for the heterologous expression of Type III polyketide synthase genes

Background Recent advances in genome sequencing, combined with bioinformatic analysis, has led to the identification of numerous novel natural product gene clusters, particularly in actinomycetes of terrestrial and marine origin. Many of these gene clusters encode uncharacterised Type III polyketide synthases. To facilitate the study of these genes and their potentially novel products, we set out to construct an actinomycete expression host specifically designed for the heterologous expression of Type III PKS genes and their gene clusters. Results A derivative of Streptomyces coelicolor A3(2) designed for the expression of Type III polyketide synthase (PKS) genes was constructed from the previously engineered expression strain S. coelicolor M1152 [Δact Δred Δcpk Δcda rpoB(C1298T)] by removal of all three of the endogenous Type III PKS genes (gcs,srsA,rppA) by PCR targeting. The resulting septuple deletion mutant, M1317, proved to be an effective surrogate host for the expression of actinobacterial Type III PKS genes: expression of the reintroduced gcs gene from S. coelicolor and of the heterologous rppA gene from Streptomyces venezuelae under the control of the constitutive ermE* promoter resulted in copious production of germicidin and flaviolin, respectively. Conclusions The newly constructed expression host S. coelicolor M1317 should be particularly useful for the discovery and analysis of new Type III polyketide metabolites.


Background
Type III polyketides are generally small molecules generated by Type III polyketide synthases (PKSs). These multifunctional ketosynthases typically utilise free CoA thioesters as substrates without the involvement of an acyl carrier protein (ACP), and frequently catalyse specific cyclization reactions found in aromatic and pyrone polyketides [1,2]. The recent explosion in genome sequencing, combined with bioinformatic analysis, has led to the identification of numerous uncharacterized Type III PKS genes in plants, fungi and bacteria, including actinomycetes [1,[3][4][5][6]. Bacterial Type III PKSs are attractive to study for a number of reasons: they are widely distributed, make a variety of structurally different products and, compared to Type I and Type II PKSs, relatively simple to manipulate genetically [7]. Some Type III PKSs are solely responsible for product formation. For example, germicidin synthase (Gcs) and the 1,3,6,8-tetrahydroxynaphthalene (THN) (RppA) synthases, commonly found in Streptomyces spp., are the only enzymes required to produce germicidin and THN, respectively [5,8,9]. THN can oxidise spontaneously to yield the red-brown pigment flaviolin, which is involved in bacterial melanin production, although a cytochrome P450 monooxygenase was reported to be involved in the conversion of THN to flaviolin in Streptomyces antibioticus [6,10,11]. Some Type III PKS genes are organized in small operons, such as srsABC of Streptomyces griseus that produces alkylresorcinols [12]. Such operons may also reside in complex gene clusters where they provide a specific precursor for the biosynthesis of more elaborate specialized metabolites; for example, the dpgABCD operon that synthesizes dihydroxyphenylacetic acid which is used as a precursor for vancomycin, balhimycin, teicoplanin and kendomycin biosynthesis in Amycolatopsis orientalis [13], A. balhimycina [14], Actinoplanes teichomyceticus [15] and Streptomyces violaceoruber [16], respectively.
Within the actinomycetes, Streptomyces species such as Streptomyces lividans TK24, Streptomyces coelicolor, Streptomyces avermitilis and Streptomyces albus J1074 [17,18] have been used as surrogate hosts for the heterologous expression of gene clusters encoding specialized metabolites. S. avermitilis, the industrial producer of avermectin [19,20], and S. coelicolor, for many years the model species for the study of the genetics and biology of actinomycetes [21], have both been manipulated genetically to enhance the level of expression of heterologous gene clusters [22][23][24][25], and derivatives of each have been used to produce a range of different heterologous specialized metabolites [23,24,[26][27][28][29][30][31][32]. In this work, we have further engineered S. coelicolor M1152 [24], a derivative of strain M145 from which four of the known antibiotic biosynthetic gene clusters had been deleted and which contains a mutation (C1298T) in rpoB that increases the level of secondary metabolite production, by removing all of its native Type III PKS genes. Deletion of the resident Type III PKS genes could potentially increase precursor pool levels and prevent undesirable interference with biosynthesis encoded by heterologous Type III PKS genes.

Type III PKS genes in S. coelicolor
Analysis of the genome sequence of S. coelicolor M145 revealed three endogenous Type III PKS genes: sco1206, sco7221 and sco7671 [33]. sco1206 encodes RppA which synthesizes 1,3,6,8-tetrahydroxynaphthalene (THN), the intermediate involved in bacterial melanin biosynthesis, by the condensation of five malonyl CoA molecules [8]. sco1206 lies upstream of sco1207 and sco1208 (Fig. 1), which are homologues of the characterized cytochrome P450 genes mel [6] and momA [10] in S. griseus and S. antibioticus, respectively; in each of these species, the single P450 gene lies adjacent to a rppA homologue. S. coelicolor thus differs in possessing two tandemly arranged P450 genes 3′ of the Type III PKS gene, with all three genes apparently translationally coupled. MomA was predicted to play a role in the oxidation of THN to form flaviolin [10], while Mel was thought to couple molecules of THN to generate hexahydroxyperylenequinone melanin [6]. Since S. coelicolor is not known to produce any of these three molecules, this gene cluster may not be expressed under commonly used laboratory growth conditions. sco7221 was shown to play a role in germicidin biosynthesis [5,34] and its product thus designated as germicidin synthase (Gcs). Analysis of Sco7671 revealed closest homology to a family of mycobacterial Type III PKSs [34]. sco7671, sco7669 and sco7670 are likely to form an operon ( Fig. 1) organized in a similar fashion to srsABC of S. griseus (known to produce phenolic lipids, the alkylresorcinols) [12]. Although the function, substrate specificity and the product profile of sco7671 have been investigated in vitro [34], the final product of the gene cluster has not been identified. To generate a clean host for the heterologous expression of Type III PKS gene clusters, we set out to delete sco7221 (gcs), sco7669-7670-7671 (the srs operon) and sco1206-1207-1208 (the rppA operon) from S. coelicolor M1152 [24].

Deletion of the three native Type III PKS genes and operons of S. coelicolor M1152
All three Type III PKS genes and operons were deleted from S. coelicolor M1152 by PCR targeting [35,36] ( Fig. 2; see reference 35 for a detailed description of PCR targeting). For two of the mutations (Δsco7221 and Δsco7669-7670-7671), in-frame deletions were generated in a two-step process; in the first step, the entire gene or operon in the corresponding cosmid was replaced with an apramycin (Apr) resistance cassette in Escherichia coli and the mutant allele then introduced into S. coelicolor M1152 to create an Apr R -marked mutant; in the second step, the Apr R cassette in the cosmid was eliminated in E. coli using FLP-recombinase to leave an 81 bp scar sequence which was then introduced into the Apr Rmarked mutant, replacing the Apr R cassette and creating an in-frame deletion mutant. Deletion of gcs (sco7221; 1122 bp) yielded M1314, which was then used for the in-frame deletion of the srs operon (sco7669-7670-7671; 2786 bp) to yield the double Type III PKS deletion mutant, M1316. Finally, the rppA operon (sco1206-1207-1208; 2859 bp) was removed from M1316 by replacement with the Apr R cassette to yield the triple mutant M1317 (Fig. 2). Confirmation of the mutations in M1314, M1316 and M1317 was achieved by PCR amplification using primers flanking the deleted regions of the chromosome followed by DNA sequencing (data not shown). With the previous removal of four antibiotic biosynthetic gene clusters (to generate M1152), the resulting septuple mutant M1317 has lost approximately 178 kb of chromosomal DNA compared to the parental strain M145.
All three deletion mutants grew and sporulated in a manner that was indistinguishable from M1152 when cultures were grown on SFM agar (data not shown).

Validation of S. coelicolor M1317 as a host for the expression of Type III PKS genes
The well-characterized germicidin synthase gene of S. coelicolor was chosen initially to validate the utility of M1317 as a host for the expression of Type III PKS genes. gcs (sco7221), which is capable of producing five congeners [5], was inserted downstream of the constitutive ermE* promoter in two bifunctional multi-copy vectors, pIJ86 and pIJ12477 (a derivative of pIJ86 conferring kanamycin resistance), to give pIJ86 + sco7221 and pIJ12477 + sco7221, respectively. pIJ86 + sco7221 and pIJ86 were then transferred by conjugation into M1152 to yield M1152/pIJ86 + sco7221 and M1152/pIJ86, respectively, while pIJ12477 + sco7221 and pIJ12477 were transferred similarly to M1317 to obtain M1317/ pIJ12477 + sco7221 and M1317/pIJ12477, respectively. Germicidin production was assessed by HPLC analysis of 5 days culture supernatants of all four strains (Fig. 3a), and the masses of the peaks corresponding to each germicidin congener [5] confirmed by mass spectrometry. While no production was observed, as expected, in M1317/pIJ12477, peaks corresponding to most of the germicidin congeners were detected in culture supernatants from the other three strains, with total production levels in M1317/pIJ12477 + sco7221 and M1152/ pIJ86 + sco7221 10.7-and 7.8-fold higher than in M1152/ pIJ86, presumably at least partly attributable to the multi-copy nature of the vectors; production in M1317/ pIJ12477 + sco7221 was 1.4-fold higher than in M1152/ pIJ86 + sco7221, potentially reflecting increased precursor supply in the septuple deletion mutant. In each case, germicidin A and germicidin B7 were the major products ( Fig. 3; Table 1).
To evaluate the use of M1317 as a host for the heterologous expression of Type III PKS genes, the rppA homologue of S. venezuelae (sven5367) was cloned in pIJ12477 to yield pIJ12477 + sven5367 which was introduced into M1152 and M1317 to produce M1152/ pIJ12477 + sven5367 and M1317/pIJ12477 + sven5367, respectively. Both strains produced a red-brownish pigment on IPM agar and in GYM liquid medium that was not observed in strains carrying the vectors alone (M1152/pIJ86 and M1317/pIJ12477) (Fig. 4). RppA initially generates 1,3,6,8-tetrahydroxynapthalene (THN) which is rapidly converted to flaviolin [6,8]. Since we were unable to detect THN by chromatography, we instead quantified flaviolin production. The mass of the HPLC peak corresponding to flaviolin was confirmed by mass spectrometry [37] (Fig. 5a). Flaviolin production was readily detected in both M1152/pIJ12477 + sven5367 and M1317/pIJ12477 + sven5367, while none was detected in M1152/pIJ86 and M1317/pIJ12477 ( Fig. 5b; Table 1), thus establishing the utility of M1317 as a host for the heterologous expression of Type III PKS genes. Production in M1317/pIJ12477 + sven5367 was 1.3-fold higher than in M1152/pIJ12477 + sven5367, again potentially reflecting increased precursor supply in the septuple deletion mutant. While some bacterial Type III PKSs, like plant chalcone and stilbene synthases [7], are sufficient alone for final product formation (e.g. the germicidin synthase of S. coelicolor [5] and the THN synthases of several Streptomyces species [8,38,39]), some require a cytochrome P450 encoded by a neighbouring gene to convert the intermediate they produce into the final product, e.g. the THN synthase of S. antibioticus [10]. Interestingly, while rppA of S. venezuelae lies adjacent to a gene encoding a cytochrome P450 that may be required for flaviolin biosynthesis in that strain, expression of rppA alone was all that was required for heterologous production of flaviolin in S. coelicolor. It is possible that in this strain THN is either oxidized spontaneously to flaviolin or oxidized by one of the remaining 18 P450 monooxygenases encoded by the S. coelicolor genome [40].
Our inability to detect flaviolin production in M1152/ pIJ86 is consistent with previous work that also suggested that sco1206 is not expressed in S. coelicolor under standard laboratory conditions [8]. However, when the rppA operon of S. coelicolor (sco1206-1207-1208) expressed from the ermE* promoter was introduced into M1317 to yield M1317/pIJ12477 + sco1206-1207-1208, redbrown pigment production occurred in both liquid-and agar-grown cultures. Flaviolin production in M1317/ pIJ12477 + sco1206-1207-1208 was confirmed by HPLC analysis and mass spectrometry (data not shown).

Conclusions
We have successfully engineered the expression host S. coelicolor M1152 by removing all three of the endogenous Type III PKS gene and operons, and validated the resulting septuple deletion M1317 as a host for the ectopic and heterologous expression of Type III PKS genes. We believe that S. coelicolor M1317 will prove to be a useful strain for the future analysis of Type III PKS genes and their products from a wide range of actinomycetes.  (Table 1) indicated by bars

Bacterial strains, plasmids, cosmids and culture conditions
Bacterial strains, plasmids and cosmids used and generated in this study are listed in Table 2. Escherichia coli strains were grown and manipulated following standard methods [35,36,41]. Media for the growth of   (Table 1) indicated by the bars S. coelicolor strains were: instant potato mash (IPM) agar [42], SFM agar [43], tryptic soy broth (TSB) and glucoseyeast extract-malt extract (GYM) medium [44]. Liquid cultures of Streptomyces strains were performed in 50 ml of medium in 250 ml flasks containing stainless steel springs at 30 °C with shaking at 250 rpm. The following antibiotics were used when required: 50 µg/ml apramycin (Apr), 50 µg/ml kanamycin (Kan), 50 µg/ml hygromycin (Hyg) and 25 µg/ml nalidixic acid (Nal). An overnight culture of each Streptomyces strain was used to inoculate 50 ml of medium to give an initial OD 600 nm = 0.1.

DNA manipulations for E. coli and Streptomyces
Extraction of DNA from E. coli was carried out using standard methods [36] or by using a QIAprep Spin Miniprep Kit (QIAGEN) and following the manufacturer's instructions. Genomic DNA was isolated from Streptomyces strains using the modified Kirby Mix procedure [43]. PCR amplifications for confirmation of engineered cosmids, mutant strains and expression constructs were performed using Taq DNA polymerase (QIAGEN). Expand High Fidelity PCR System (Roche) was used to amplify disruption cassettes [35] and Phusion ® High-Fidelity DNA Polymerase (NEB) was used to amplify PCR products for gene expression.

Deletion of the Type III PKS genes of S. coelicolor
The three native Type III PKS genes and operons of S. coelicolor (sco7221, sco7669-7670-7671 and sco1206-1207-1208) were deleted from strain M1152 [24] by PCR targeting [35,41]. Three pairs of flanking primers, SCO7221FSpe and SCO7221RNhe, SCO7671FSpe and SCO7669RNhe, and SCO1206FSpe and SCO1208RNhe (Table 3), were designed to amplify a 1458 bp fragment carrying an apramycin resistance (Apr R ) oriT cassette from pIJ773 [35]. Each of the Apr R cassettes were flanked by 39 nt sequences matching the ends of the gene or operon to be deleted, and the targeted regions replaced with the Apr R cassette by homologous recombination between these 39 nt sequences. Cosmids St2H12, St4C2 and 2StG58, containing sco7221, sco7669-7670-7671 and sco1206-1207-1208, respectively, were transferred into E. coli BW25113 [47]/pIJ790 [35] and the resulting strains were then transformed with the corresponding replacement cassette. The mutagenized cosmids pIJ13000, pIJ13001 and pIJ13002, respectively, were verified by restriction enzyme digestion and PCR amplification with primers SCO7221TF and SCO7221TR, SCO7671TF and SCO7669TR, and SCO1206TF and SCO1208TR, respectively (Table 3). pIJ13000 (Δsco7221::(oriT-aac(3)IV)) was then introduced into S. coelicolor M1152 by conjugation with selection for apramycin resistance; subsequent screening for loss of kanamycin resistance (encoded by the SuperCos 1 vector) but retention of apramycin resistance was used to isolate the required Δsco7221::(oriTaac(3)IV) replacement mutant. In the meantime, the Apr R cassettes were removed from pIJ13000, pIJ13001 and pIJ13002 in E. coli BT340 [46] using the FLP-recombinase, in each case leaving a 81 bp scar sequence in place of the cassette. The resulting non-transmissible cosmids, pIJ13003, pIJ13005 and pIJ13007, respectively, were confirmed by restriction enzyme digestion and PCR SVEN5367RHindIII AAAAAAAGCTTTCAGGCGACGGACGTGCGGG amplification using the same primer sets that were used for verification of the Apr R cassettes (Table 3). To allow transfer of these cosmids from E. coli to Streptomyces via conjugation, a bla-oriT-hyg-bla cassette was amplified from pIJ10701 [35] and used to target the bla gene in the SuperCos1 backbone of pIJ13003, pIJ3005 and pIJ13007, yielding pIJ13004, pIJ13006 and pIJ13008, respectively. Intergeneric conjugation between the S. coelicolor Δsco7221::(oriT-aac(3)IV) replacement mutant and E. coli ET12567/pUZ8002 [47,50] containing pIJ13004 was performed as previously described [35,41]. Exconjugants were selected based on hygromycin resistance and the desired Δsco7221 in-frame deletion mutant M1314 obtained by screening for Apr, Kan and Hyg sensitive segregants and confirmed by sequencing of the PCR product obtained from genomic DNA using primers SCO7221TF and SCO7221TR. Essentially the same procedure was used to generate the double in-frame deletion mutant M1316 (Δsco7221, Δsco7669-7671) from M1314 using pIJ13001 and pIJ13006 (and confirmed by sequencing the PCR amplicon obtained from genomic DNA using PCR primers SCO7671TF and SCO7669TR). Similarly the triple deletion mutant M1317 (Δsco7221, Δsco7669-7671, Δsco1206-1208::aac(3)IV) was derived from M1316 using pIJ13002. Despite repeated attempts, it was not possible to remove the Apr R cassette from M1317 using pIJ13008. Therefore, M1317 was confirmed by sequencing of the PCR product obtained from genomic DNA using PCR primers SCO1206TF and SCO1208TR. In each case the PCR fragment was cloned into pGEM ® -T Easy and sequenced using M13 universal primers to verify that the expected deletions had been made. The three mutants were grown on SFM agar until sporulation. Spores were harvested and stored in 20 % glycerol at −20 °C.

Analysis of germicidin and flaviolin production
Cultures were grown for 5 days in 50 ml of GYM liquid medium, 5 ml of each culture supernatant lyophilized, the dried samples dissolved in 500 µl water and filtered through a VectaSpin Micro Polysulphone 0.2 µm column (Whatman). The samples were injected onto a Phenomenex Gemini-NX3u C18 110A, 150 × 4.6 mm column fitted to an Agilent 1100 HPLC system and analyzed using a linear methanol:water gradient in 0.1 % (w/v) formic acid as follows: 0 min, 10 % methanol; 1 min, 25 % methanol; 21 min, 100 % methanol; 25.50 min, 10 % methanol; 33 min, 10 % methanol, at a flow rate of 0.8 ml/min with UV absorbance monitoring at 280 nm. Fractions of interest were collected and analyzed by mass spectrometry (Thermo Scientific ™ ). Germicidins were analyzed using positive ion mode mass spectrometry whereas flaviolin was analyzed using negative ion mode. Each experiment was carried out in duplicate. The mycelium from each 50 ml culture was harvested by centrifugation, dried and the dry weight determined. Relative levels of germicidin and flaviolin production were calculated by multiplying the average peak area (mAU) obtained from a 5 ml sample by ten and dividing by the average dry weight obtained from a 50 ml culture.