To further explore possible avenues for accessing microbial biodiversity for drug discovery from natural products, we constructed and screened a 5,000-clone "shotgun" environmental DNA library by using an Escherichia coli-Streptomyces lividans shuttle cosmid vector and DNA inserts from microbes derived directly (without cultivation) from soil. The library was analyzed by several means to assess diversity, genetic content, and expression of heterologous genes in both expression hosts. We found that the phylogenetic content of the DNA library was extremely diverse, representing mostly microorganisms that have not been described previously. The library was screened by PCR for sequences similar to parts of type I polyketide synthase genes and tested for the expression of new molecules by screening of live colonies and cell extracts. The results revealed new polyketide synthase genes in at least eight clones. In addition, at least five additional clones were confirmed by high-pressure liquid chromatography analysis and/or biological activity to produce heterologous molecules. These data reinforce the idea that exploiting previously unknown or uncultivated microorganisms for the discovery of novel natural products has potential value and, most importantly, suggest a strategy for developing this technology into a realistic and effective drug discovery tool.

Bacteria belonging to the order Actinomycetales produce most microbial metabolites thus far described, several of which have found applications in medicine and agriculture. However, most strains were discovered by their ability to produce a given molecule and are, therefore, poorly characterized physiologically and genetically. Thus, methodologies for genetic manipulation of actinomycetes are not available and efficient tools have been developed for just a few strains. This constitutes a serious limitation to applying molecular genetics approaches to strain development and structural manipulation of microbial metabolites. To overcome this hurdle, we have developed bacterial artificial chromosomes (BAC) that can be shuttled among Escherichia coli, where they replicate autonomously, and a suitable Streptomyces host, where they integrate site-specifically into the chromosome. The existence of gene clusters and of genetically amenable host strains, such as Streptomyces coelicolor or Streptomyces lividans, makes this a sensible approach. We report here that 100 kb segments of actinomycete DNA can be cloned into these vectors and introduced into genetically accessible S. lividans, where they are stably maintained in integrated form in its chromosome.

Using a cell-free protein-synthesis system, we have established that the elongation factor (EF) Tu (EF-Tu) of the actinomycete Planobispora rosea, the producer of the thiazolyl peptide GE2270, a specific EF-Tu inhibitor, is highly resistant to its own antibiotic, while it is completely inhibited by kirromycin, which is another inhibitor of this factor. P. rosea was found to possess a single tuf gene, located between fus and rpsJ, encoding other components of the protein-synthesis machinery. The P. rosea tuf gene was expressed as a translational fusion to malE in Escherichia coli, and the resulting EF-Tu with an N-terminal Gly-Met extension was able to promote poly(U)-directed poly(Phe) synthesis in cell-free systems. This activity was not affected by GE2270, and the recombinant protein was incapable of binding the antibiotic, indicating that the P. rosea EF-Tu is intrinsically resistant to this inhibitor. Inspection of the translated tuf sequence revealed a number of amino acid substitutions in highly conserved positions. These residues, which are likely to be involved in conferring GE2270 resistance, map in EF-Tu domain II, as do the only two known mutations conferring resistance to this class of thiazolyl peptides in Bacillus subtilis.

No abstract

The antibiotic kirromycin (Kr) inhibits bacterial protein synthesis by binding to elongation factor Tu (EF-Tu). Streptomyces cinnamoneus and Nocadia lactamdurans, producers of antibiotics of the Kr class, are known to possess an EF-Tu resistant to Kr. Both micro-organisms appear to possess a single tuf gene and we have characterized the one from 5. cinnamoneus, which belongs to the tuff family. To assess the molecular determinants of Kr resistance, the S. cinnamoneus tuf gene was expressed in Escherichia coli as a translational fusion to ma/€, which enabled the recovery by affinity chromatography of the recombinant protein uncontaminated by the host factor. The recombinant EF-T u was able to catalyse polyU-directed polyphe synthesis in two heterologous cell-free systems, even as an uncleaved fusion. When tested for antibiotic sensitivity it behaved like the natural S. cinnamoneus protein, showing equivalent resistance to Kr but sensitivity to the antibiotic GE2270, indicating that all determinants for Kr resistance are intrinsic to the EF-Tu sequence. Multiple sequence analysis of EF-Tu proteins, together with knowledge of mutations conferring Kr resistance, allowed the identification of key residues as likely candidates for the natural Kr resistance of the S. cinnamoneus EF-Tu. One of these, Tht.378, was mutated to the consensus Ala and the resulting mutant protein was sensitive to Kr. Interestingly, it retained some activity (30% of the control) even at high Kr concentrations.