NATURE BIOTECHNOLOGY VOLUME 24 NUMBER 12 DECEMBER 2006 1541New antibiotics from bacterialnatural productsJon Clardy, Michael A Fischbach & Christopher T WalshFor the past five decades, the need for new antibiotics has been met largely by semisynthetic tailoring of natural product scaffolds discovered in the middle of the 20th century. More recently, however, advances in technology have sparked a resurgence in the discovery of natural product antibiotics from bacterial sources. In particular, efforts have refocused on finding new antibiotics from old sources (for example, streptomycetes) and new sources (for example, other actinomycetes, cyanobacteria and uncultured bacteria). This has resulted in several newly discovered antibiotics with unique scaffolds and/or novel mechanisms of action, with the potential to form a basis for new antibiotic classes addressing bacterial targets that are currently underexploited.Because the use of an antibiotic inevitably selects for resistant microbes, there is a continuing and cyclical need for new antibiotics. The clock begins to tick down on an antibiotic’s useful lifetime before clinically significant resistance emerges1, impelling the need for new drugs to combat the current generation of resistant pathogens. In the 20th cen-tury, two parallel lines of discovery yielded success in the antibiotic arena: isolation of natural products with antibiotic activity and prepa-ration of synthetic antibiotics. The golden age of antibiotic natural product discovery spanned only the 1940s to 1950s. In accord with Sir James Black’s famous observation that “the most fruitful basis for the discovery of a new drug is to start with an old drug2,” new antibiot-ics from the subsequent decades have come from iterative, semisyn-thetic tailoring of the natural scaffolds to create successive generationsof antibiotics (for example, erythromycin → clarithromycin → telithro-mycin)3.This review examines the prospects for renewed discovery of natu-ral antibiotics—especially those with new molecular templates and/or mechanisms of action—from two perspectives. First, we consider esti-mates of how many new molecules are still to be found and discuss strategies for finding them, with an emphasis on antibiotics from the uncultured microbial majority. Second, we use examples of recently dis-covered (or rediscovered) molecular classes and newly tailored scaffolds to discuss novel strategies for engaging old and new bacterial targets. Our emphasis on naturally occurring antibiotics has two motivations: first, the well-established ability to discover useful antibiotics from natural sources suggests that continued efforts are likely to be fruitful; and second, as antibiotics often reach their targets by transport rather than diffusion, antibiotic candidates benefit from having structural features rarely found in the synthetic libraries of ‘drug-like’ molecules used in most high-throughput screening facilities.How many natural antibiotics remain to be discovered?Over the past two decades, pharmaceutical companies deemphasized microbial screening programs for many reasons, although they fre-quently cited declining productivity and diminishing discoveries of novel molecules3. Work by two research groups4–6 allow at least a semiquantitative analysis of the chances for continued discovery from microbial sources.In 2001, Watve et al.4 estimated that from the first report of streptothricin in 1942 and streptomycin a year later, the order Actinomycetales had yielded ~3,000 known antibiotics (90% of those from Streptomyces, an Actinomycetales genus). On the basis of past experience, these authors proposed that if streptomycetes (exclusively) were screened as widely as they had been in 1995, 15–20 antibiotics would be discovered each year for the next 50 years. Over the subsequent five decades, these ~1,000 new molecules would yield 20–40 new antibiotics for human clinical use, assuming that the his-torical trend of one marketed antibiotic for 25–50 novel molecules remains the same. Apparently, this projected discovery rate was too low to be economically viable because several large pharmaceutical companies closed their antibiotic discovery programs shortly after its publication.More recent (and complementary) analyses by Baltz5,6 have esti-mated the frequency of antibiotic production by actinomycetes. In his 2005 paper, Baltz5 noted a 1958 report from H.B. Woodruff andL.E. McDaniel at Merck (Whitehouse Station, NJ, USA) indicating that 104 strains would include 2,500 antibiotic producers. Of these 2,500, 2,250 would make streptothricin, 125, streptomycin and 40, tetracy-cline, with frequencies of 2 × 10−1, 1 × 10−2 and 4 × 10−3, respectively. In 1976, Arai7 extended the Woodruff and McDaniel analysis to include vancomycin and erythromycin, and found frequencies of 1.5 × 10−5 and 5 × 10−6, respectively4. Baltz5 also points out that daptomycin (Table 1 and Fig. 1) was found once in 107 actinomycete cultures. Thus, the frequency of discovery for antibiotics from the most productive bac-teria, the actinomycetes, ranges 106-fold in the ~107 strains screened historically. In addition, Baltz5 estimates that less than one part in 1012 of the earth’s soil surface has been screened for actinomycetes.Department of Biological Chemistry & Molecular Pharmacology, Harvard Medical School, Boston, Massachusetts 02115, USA. Correspondence should be addressed to C.T.W. ([email protected]).Published online 11 December 2006; doi:10.1038/nbt1266REVIEW© 2006 Nature Publishing Group http://www.nature.com/naturebiotechnology1542 VOLUME 24 NUMBER 12 DECEMBER 2006 NATURE BIOTECHNOLOGYAlthough only 1–3% of all streptomycete antibiotics have been dis-covered, to find the remaining 97–99% will require a combination of high-throughput screening by modern technologies (108–109 strains per year), selection against the most common antibiotics, methods to enrich rare and slow-growing actinomycetes, a prodigious microbial collecting and culturing effort, and combinatorial biosynthesis in streptomycetes. As one example, Baltz6 notes that screening strains of Escherichia coli K12—engineered to harbor 15 antibiotic-resistance genes to exclude the most common antibiotics produced by actino-mycetes—can enhance the signal-to-noise ratio for new molecules with novel modes of action.Accessing greater bacterial diversity and optimizingnatural