Synergy and contingency as driving forces for the evolution of multiple secondary metabolite production by Streptomyces species

Since the discovery of actinomycin in Selman Waksman's laboratory at Rutgers University in 1940, followed in 1943 by streptomycin, the first really effective drug to treat tuberculosis, the actinomycetes have been famous as producers of antibiotics and other “secondary metabolites” with biological activity. During the Golden Age of antibiotic discovery, in the 1950s and 1960s, such well-known antibacterial drugs as tetracycline, erythromycin, and kanamycin, antifungal agents like candicidin and nystatin, and anticancer drugs such as adriamycin were discovered through the efforts of academic and industrial researchers. After 1970 the rate of discovery of useful compounds declined progressively, although several important agents nevertheless came to light, including the antihelmintic avermectin, the immunosuppressants rapamycin and tacrolimus (FK506), and the natural herbicide bialaphos. In fact, the total number of known biologically active molecules continued to grow steadily after the end of the Golden Age. By the mid to late 1990s, thousands of antibiotics and compounds with other biological activities had been described. Estimates of the numbers vary. For example, Demain and Fang (1) gave the total number of antibiotics produced by bacteria and fungi as 5,000, whereas Berdy (2) had more than twice that number. Nevertheless there is agreement that the actinomycetes are responsible for more than two-thirds of the total. Within the actinomycetes, members of the genus Streptomyces account for 70–80% of secondary metabolites, with smaller contributions from genera such as Saccharopolyspora, Amycolatopsis, Micromonospora, and Actinoplanes. What are secondary metabolites, what is their evolutionary significance, and why should the actinomycetes, those soil-dwelling, sporulating members of the high G + C branch of the Gram-positive bacteria, be such prolific producers of them?

There is no pithy one-line definition of the term secondary metabolite, but it nevertheless remains an indispensable epithet in discussions about microbial (and plant) biochemistry and ecology. It embraces the ideas that such compounds are characteristic of narrow taxonomic groups of organisms, such as strains within species, and have diverse, unusual, and often complex chemical structures. They are nonessential for growth of the producing organism, at least under the conditions studied, and are indeed typically made after the phase of most active vegetative growth when the producer is entering a dormant or reproductive stage (1). Their range of biological activities is wide, including the inhibition or killing of other microorganisms (the narrow definition of an antibiotic), but also toxic effects against multicellular organisms like invertebrates and plants. Then there are hormone-like roles in microbial differentiation, and roles in metal transport, a function that blurs the distinction between primary and secondary metabolism. More problematic are the many compounds that either have (so far) no demonstrated biological activity or an activity that is hard to relate to any competitive advantage to the producer, such as a specific effect on the vertebrate immune system exerted by a compound made by a saprophytic soil inhabitant.

The last category was in particular responsible for a widespread view that secondary metabolites were either neutral in evolutionary terms or significant merely as waste products or to keep metabolism ticking over. The implausibility of such general explanations was admirably discussed by Williams et al. (3), who emphasized two powerful arguments for the adaptive significance of secondary metabolites: the complex genetic determination of their biosynthesis, and the exquisite adaptation of many classes of compounds (six were described in detail) to interact with their targets. The former point has been further reinforced by innumerable genetic studies of the biosynthesis of natural products since 1989. Take, for example, erythromycin (4). The producer, Saccharopolyspora erythrea, devotes some 60 kb of its DNA to making this macrolide from propionate units by an amazing assembly-line process involving no fewer than 28 active sites arranged along three giant proteins, followed by hydroxylation and glycosylation involving 18 further proteins, and then has to protect its ribosomes from the highly specific toxicity of the antibiotic by an equally specific methylation of a site on the ribosomal RNA. This is not merely a means of dealing with any excess propionate the cell may produce, nor an idling of the metabolic machinery!

Firn and Jones (5) have made an important recent contribution to the debate. They point out that, whereas some secondary metabolites are indeed highly potent at the concentrations produced in nature and, in the case of antibiotics, against target organisms that actually interact with the producer in the wild, many have only moderate activity. They proposed a unifying model in which it is acknowledged that high-affinity, reversible, noncovalent interactions between a ligand and a protein only occur when the ligand has exactly the right molecular configuration to interact with the complex 3D structure of the protein (what they call “biomolecular activity”) and that this is a rare property. On the other hand, many more molecules have biological activity when tested against a whole organism containing thousands of potential targets that might be inhibited relatively inefficiently. There will therefore be a selective advantage in the evolution of traits that optimize the production and retention of chemical diversity at minimal cost, to allow for the gradual emergence of true biomolecular activity. As long as we accept a selective advantage for the ability to produce secondary metabolites, then, it is legitimate to ask why the streptomycetes and their relatives are preeminent in this ability. The answer almost certainly reflects both the habitat of the organisms and their lifestyle.

The classical habitat of Streptomyces species is as free-living saprophytes in terrestrial soils. Although this is doubtless correct, there is now abundant evidence that some species colonize the rhizosphere of plant roots and even plant tissues; in some cases antibiotic production by the streptomycete may protect the host plant against potential pathogens; the symbiont in turn acquires nutrients from the plant (6, 7). There is now good evidence also for the growth of actinomycetes in marine soils (8).

The soil is a proverbially complex environment in which innumerable stresses (chemical, physical, and biological) occur in a temporally and spatially variable manner. Moreover, streptomycetes are nonmotile, so stresses cannot be avoided but have to be met. The need to combat stress was the explanation invoked by Bentley et al. (9) for the enormous numbers of genes that would encode regulators, transport proteins, and nutritional enzymes identified in the Streptomyces coelicolor genome sequence. A striking feature of this 8.7-megabase (Mb) genome is its notional division into a “core” region of ≈4.9 Mb and left and right “arms” of ≈1.5 and 2.3 Mb, respectively. Classes of genes that would encode unconditionally essential functions such as the machinery of DNA replication, transcription, and translation were found in the core, whereas examples of conditionally adaptive functions, such as the ability to grow on complex carbohydrates like cellulose, chitin, and xylan, occurred predominately in the arms. The genome sequence also revealed ≈23 clusters of genes, representing ≈4.5% of the genome, that were predicted to encode biosynthetic enzymes for a wide range of secondary metabolites (Table 1), only half a dozen of which had previously been identified. Interestingly in the present context, many of the clusters reside in the arms or close to their boundary with the core of the genome, as pointed out by Piepersberg (10) in a review that emphasizes the roles of secondary metabolites in chemical communication, the topic of this Sackler symposium. An even larger number of secondary metabolic gene clusters was found in the recently sequenced Streptomyces avermitilis genome (30 clusters covering ≈6% of the genome), again many of them in the arm regions (11). Thus, assuming a selective advantage for secondary metabolite production, such an advantage for many of the compounds is likely to be conditional, or sporadic.

As pointed out by Chater and Merrick (12), Streptomyces antibiotics are typically produced in small amounts at the transition phase in colonial development when the growth of the vegetative mycelium is slowing as a result of nutrient exhaustion and the aerial mycelium is about to develop at the expense of nutrients released by breakdown of the vegetative hyphae (13). Such antibiotics are proposed to defend the food source when other soil microorganisms threaten it. The hypothesis is strengthened by the finding of large numbers of antibiotics in other groups of differentiating microbes such as filamentous fungi and myxobacteria. The concept has been extended to the possibility that competitors might be attracted to the amino acids, sugars, and other small molecules arising from the degraded vegetative mycelium and be killed and recycled by the developing Streptomyces colony, a concept dubbed “fatal attraction” in relation to Myxococcus by Shi and Zusman (14).