Actinomycetes, mainly Streptomyces, are the most important group of bacteria as producers of bioactive compounds such as antibiotics or antitumor compounds. However, production yields by native strains are usually rather low and development programs for increasing yields are generally required. Thus, pharmaceutical companies have carried out large programs of random mutagenesis and selection to improve production yields of target compounds. Last, other strategies have been explored that involve the use of genetic engineering technologies.1 In this sense, higher production yields have been achieved by expressing antibiotic biosynthesis gene clusters in optimized heterologous production strains. Thus, industrial producer strains have been successfully used as hosts to express biosynthesis gene clusters from other organisms.2, 3 Also, some Streptomyces strains have been engineered (by deleting antibiotic gene clusters from their chromosomes), to be used as hosts for expressing heterologous antibiotic gene clusters improving production yields.4 This approach requires a previous knowledge of the biosynthesis gene cluster, including their limits, and to be able to clone the entire gene cluster in a suitable vector. This latter could be a handicap especially when biosynthesis genes are not clustered in the chromosome as it occurs with the elloramycin or with some phenazine gene clusters,5, 6 or for large gene clusters. Although in the past years several strategies to rescue large gene clusters and vectors suitable to harbor large pieces of DNA have been developed,7, 8 manipulation of large pieces of DNA is still a difficult task and requires appropriate skills.8 An alternative approach to increase production yields of bioactive compounds is to apply metabolic engineering strategies to native producer strains.1 Thus, higher production yields have been achieved by overexpressing or inactivating activator and repressor genes, respectively.9 To apply this strategy, target regulatory genes should be first identified, which can be hard to accomplish if the cluster is under a complex regulatory network. Also, specific mutations in RNA polymerase and in ribosomal protein-coding genes have been used to enhance antibiotic production yields.10 An alternative approach is to increase the intracellular pool of precursor metabolites that are used in the biosynthesis pathway of the target antibiotic.1 In this respect, overexpressing genes coding for enzymes that catalyze specific steps in the biosynthesis of these precursor metabolites could be a suitable strategy.

Many bioactive drugs belong either to the polyketide (PK) or the hybrid polyketide-nonribosomal peptide (PK-NRP) families of compounds. PKs are synthesized through the condensation of acyl-CoA units such as malonyl-CoA, which is derived from acetyl-CoA in a reaction catalyzed by an acetyl-CoA carboxylase (ACC). In addition, many bioactive compounds are glycosylated by one or several deoxysugars mainly derived from glucose-1-phosphate. The biosynthesis of glucose-1-phosphate involves the enzyme phosphoglucomutase (Pgm) that interconverts glucose-6-phosphate and glucose-1-phosphate. We have previously shown that the heterologous overexpression of the ACC genes (ovmGIH) from the oviedomycin gene cluster11 and the Pgm gene from Streptomyces coelicolor12 results in an increase in the intracellular pool of malonyl-CoA and glucose-1-phosphate, respectively, and also increases the production of mithramycin in S. argillaceus.13 Here we report the construction of conjugative plasmids for overexpressing the ACC, Pgm or both, and the use of these plasmids to easily improve production yields of different bioactive compounds produced by Streptomyces strains.

Plasmids pPGM, pGIH and pMIX13 contain the phosphoglucomutase pgm, the ACC ovmGIH or both sets of genes, respectively, cloned under the control of the erythromycin resistance promoter in the multicopy plasmid pIAGO.14 To facilitate expression of these plasmids in different Streptomyces strains, overcoming difficulties of protoplast transformation, they were converted into conjugative plasmids by inserting the oriT in the above-mentioned constructs and also in pIAGO, generating pPGMT, pGIHT, pMIXT and pIAGOT, respectively (Figure 1).

Figure 1
figure 1

Conjugative plasmids generated in this work. bla, β-lactamase gene; tsr, thiostrepton resistance gene; pgm, phosphoglucomutase gene; ovmG, ovmI and ovmH, acetyl-CoA carboxylase genes; PermE, erythromycin resistance gene promoter.

These plasmids were tested as tools to increase production yields of malonyl- and/or glucose-derived compounds, by introducing them into different Streptomyces producer strains. We selected native Streptomyces strains known to produce (i) PKs derived from malonyl-CoA (actinorhodin producer S. coelicolor M145; tetracenomycin C producer S. glaucescens Tü49); (ii) glycosylated PKs derived from malonyl-CoA and glucose-1-phosphate (aclacinomycin A producer S. galilaeus ATCC 13615; chromomycin A3 producer S. griseus subsp. griseus ATCC 13273; nogalamycin producer S. nogalater NRRL 3035; elloramycin producer S. olivaceus Tü2353; steffimycin producer S. steffisburgensis NRRL 3193); (iii) hybrid PK-NRP (collismycin A producer S. sp. CS40);15 and (iv) glycosylated hybrid PK-NRP (streptolydigin producer S. lydicus NRRL 2433). In addition, the recombinant strain S. lividans 16F4 containing cosmid cos16F4 that directs the biosynthesis of the PK 8-demethyl-tetracenomycin C was also used.16

pGIHT was introduced into strains that are not known to produce glycosylated compounds (that is, S. coelicolor, S. glaucescens, S. lividans 16F4 and S. sp CS40), whereas pPGMT, pGIHT and pMIXT were used to conjugate the rest of strains (S. galilaeus, S. griseus subsp. griseus, S. nogalater, S. olivaceus, S. steffisburgensis and S. lydicus). pIAGOT was used as a control. Five thiostrepton-resistant colonies (50 μg ml−1 final concentration) were selected from each conjugation event, and were used to determine production yields of the corresponding compounds in comparison with control strains (IAGO strains). Antibiotic production was analyzed in 3 ml of R5A solid medium,17 using 25-compartment squared Petri plates inoculated with 20 μl of a spore suspension (106 spores per ml). Cultures from 8 days old at 30 °C were extracted twice with a solution of ethyl acetate and 1% formic acid, and organic extracts were evaporated under vacuum and suspended in 200 μl (400 μl in the case of actinorhodin and streptolydigin containing extracts) of DMSO:methanol (50:50). Samples (2 μl) were analyzed by reversed-phase chromatography in an Acquity UPLC equipment with a BEH C18 column (1.7 μm, 2.1 × 100 mm, Waters, Taunton, MA, USA), using as mobile phase acetonitrile and 0.1% trifluoroacetic acid in water. Elution was carried out with 10% acetonitrile for 1 min, followed by a linear gradient (10 to 80% acetonitrile) for 7 min, at a flow rate of 0.5 ml min−1. HPLC-MS analysis was carried as described.18 Compounds were identified by their absorption and mass spectra, and chromatographic characteristics. Concentration of compounds produced by the different strains was estimated by calculating peak areas of the corresponding compound. Four (six in the case of S. lydicus) independent cultures of five biological samples were run and analyzed for each experiment.

Figure 2 shows increments in production yields for each recombinant strain as percentages in relation to the control strain. As can be observed, expression of pGIHT resulted in an increase in the production yields of all nonglycosylated PK (actinorhodin, 8-demethyl-tetracenomycin C and tetracenomycin C) and hybrid PK-NRP (collismycin) compounds: 943% for actinorhodin, 100% for 8-demethyl-tetracenomycin C, 39% for tetracenomycin C and 20% for collismycin (Figure 2a). In the case of actinorhodin, a similar increment has been reported by overexpressing its own ACC.12 Interestingly, although 8-demethyl-tetracenomycin C and tetracenomycin C only differs by the presence of an additional methyl group in the latter, increments in their productions were quite different. This could reflect intrinsic differences between strains, such as the presence/absence of other biosynthesis gene clusters that could compete for the use of malonyl-CoA. Also, it can not be excluded a higher requirement for S-adenosyl-methionine in tetracenomycin C biosynthesis to explain this lower production.

Figure 2
figure 2

Production yields in R5A solid medium of engineered Streptomyces strains producing antibiotic compounds derived from (a) malonyl-CoA and (b) malonyl-CoA and glucose-1-phosphate. Productions are shown as percentages in relation to the control strain (IAGOT) that was set as 100. GIHT, PGMT and MIXT are strains expressing the acetyl-CoA carboxylase (ovmGIH), the phosphoglucomutase (pgm) and both set of genes, respectively. ACL, aclacinomycin A; ACT, actinorhodin; COL, collismycin A; CHR, chromomycin A3; DTC, 8-demethyl-tetracenomycin C; ELL, elloramycin; NOG, nogalamycin; STL, streptolydigin; STF, steffimycin; TCM, tetracenomycin C.

The lowest increase achieved was for collismycin (20%). This could be explained because the biosynthesis of collismycin requires cysteine, lysine and leucine as precursor metabolites, in addition to malonyl-CoA15 (Figure 3). By favoring the biosynthesis of malonyl-CoA that of leucine and/or lysine could be harmed as that would decrease the intracellular concentration of acetyl-CoA and also the carbon flux through the tricarboxylic acids cycle, lowering the formation of lysine from oxaloacetate (Figure 3).

Figure 3
figure 3

Overview of the metabolic origin of precursor metabolites for the biosynthesis of bioactive compounds used in this study. Engineered biosynthesis steps by overexpression of acetyl-CoA carboxylase (ACC) and phosphoglucomutase (PGM) are indicated by thick arrows. Biosynthesis precursors are highlighted by gray squares. TCA, tricarboxylic acids cycle.

All plasmids were also expressed in strains producing compounds that use malonyl-CoA and glucose-1-phosphate as precursor metabolites. In general, the best results were obtained using plasmid pMIXT that co-expresses Pgm- and ACC-coding genes, followed by pGIHT and pPGMT (Figure 2b). Production of chromomycin A3 was most significantly improved (470%), whereas production of elloramycin, nogalamycin, aclacinomycin A and steffimycin was increased by 38%, 42%, 51% and 62%, respectively. Chromomycin A3 is structurally similar to antitumor mithramycin from which only differs in some of the sugars that are attached to the PK aglycon.19 Although both biosynthesis pathways use the same precursors, the increments achieved in chromomycin A3 production were higher than those observed for mithramycin in S. argillaceus in the same conditions.13 As mentioned above, this could also reflect differences in the number and expression of biosynthesis gene clusters in both strains. A striking result was observed in relation to the glycosylated hybrid PK-NRP streptolydigin. In this case, overexpression of the Pgm-coding gene (pPGM) in S. lydicus improved streptolydigin production ~506%, whereas by expressing the ACC-coding genes alone or together with pgm the increments were 74% and 90%, respectively. These results suggest that the intracellular concentration of glucose-1-phosphate, the biosynthetic precursor for the l-rhodinose moiety in the final compound, could be an important bottleneck for the biosynthesis of streptolydigin in S. lydicus. Surprisingly, the overexpression of ACC together with Pgm decreased streptolydigin production. A possible explanation could be that by overexpressing ACC the concentration of acetyl-CoA and the flux of acetyl-CoA through the tricarboxylic acid cycle would decrease. This would diminish formation of propionyl-CoA, and also of 2-oxoglutarate and consequently that of 3-methylaspartate, which have been reported to be precursor metabolites in the biosynthesis of streptolydigin in S. lydicus20, 21 (Figure 3).

To summarize, different approaches have been reported to improve the production of secondary metabolites. We report here the construction of three conjugative plasmids to express the Pgm, ACC or both, which can be used as metabolic engineering tools to improve production yields of compounds derived from malonyl-CoA and/ or glucose-1-phosphate, in natural and recombinant Streptomyces strains. Using these plasmids, we were able to improve production yields (between 20 and 943%) of different PK and hybrid PK-NRP compounds. These plasmids could tentatively be used with highly producer industrial strains to further improve their production yields.