Senior Scholar Award in Global Infectious Disease
Jon Clardy, Ph.D.
Harvard Medical School

New Antibiotics from Environmental DNA

Cultured soil microbes provide many of our most important drugs including the antibiotics erythromycin and vancomycin, the immunosuppressive drugs FK506 and rapamycin, and the anticancer agents mitomycin C and actinomycin D, among many others. Culturing soil microbes is still a useful way to discover new medicinal agents, but rediscovery rates are discouragingly high. Over 99% of antibiotic ‘hits’ rediscover previously known antibiotics. DNA-based probes of soil have shown that only a tiny, unrepresentative minority (~0.1%) of the microbes can be cultured. This research project proposes methods for accessing the potentially useful molecules produced by the 99.9% of the microbes that cannot be cultured.

The instructions to produce a molecule such as erythromycin are on a continuous stretch of microbial DNA along with instructions for resistance (auto-immunity) and regulation. DNA extracted directly from soil, the collective DNA of the billions of unnamed microbes in every pinch of soil, can be isolated in the laboratory. This environmental DNA (eDNA) can then be chopped into conveniently sized pieces, and some of these pieces will contain the instructions to make a new antibiotic. The eDNA fragments can be put into organisms that are easily cultured in the laboratory, and these cultured organisms can then be screened for the production of useful molecules. There are several problems with this scenario, especially problems with finding appropriate host organisms for statement and problems with screening.

The problem of getting DNA from soil is largely solved although improvements in scale would be very useful. The eDNA can be introduced into laboratory organisms in several formats, and initially we’ve elected to use cosmid libraries. The relatively small size of the inserted cosmid DNA is more than compensated by the multiple copy number. At the moment we’re using E. coli for heterologous statement in the laboratory, and future plans call for developing other hosts. Screening a cosmid library of several million, potentially several billion, members is difficult enough, but the low hit rate—the overwhelming majority of library members will contain absolutely nothing of interest—makes screening the biggest bottleneck. Screening will also be complicated by low statement rates due to the presence of unusual promoter sequences or missing pathway intermediates in the laboratory organisms. We’ve developed a double antibiotic selection method that allows screening to be done on the selection plates, and using this scheme, a several million-member library can be screened for antibiotic activity in a matter of weeks. Several small molecule antibiotic producing clones were found in the first attempt.

A major advantage of the heterologous statement of eDNA for the discovery of new antibiotics is the tight linkage between the antibiotic with its biosynthetic gene cluster. Shortly after the discovery of the first antibiotic producing clone, the single novel open reading frame responsible for its production was found using trasnposon mutagenesis. Locating one ORF can lead to many others as probes, both virtual and real, to scan DNA data bases and DNA libraries will lead to other ORFs responsible for the biosynthesis of small molecules.