Phylogenetic analysis of polyketide synthase I domains from soil metagenomic libraries allows selection of promising clones

Abstract

The metagenomic approach provides direct access to diverse unexplored genomes, especially from uncultivated bacteria in a given environment. This diversity can conceal many new biosynthetic pathways. Type I polyketide synthases (PKSI) are modular enzymes involved in the biosynthesis of many natural products of industrial interest. Among the PKSI domains, the ketosynthase domain (KS) was used to screen a large soil metagenomic library containing more than 100,000 clones to detect those containing PKS genes. Over 60,000 clones were screened, and 139 clones containing KS domains were detected. A 700-bp fragment of the KS domain was sequenced for 40 of 139 randomly chosen clones. None of the 40 protein sequences were identical to those found in public databases, and nucleic sequences were not redundant. Phylogenetic analyses were performed on the protein sequences of three metagenomic clones to select the clones which one can predict to produce new compounds. Two PKS-positive clones do not belong to any of the 23 published PKSI included in the analysis, encouraging further analyses on these two clones identified by the selection process.

FIG. 1.

Phylogenetic analysis of KS domains. The reconstruction was computed for 207 protein sequences on 413 sites by the ML method (BIONJ, JTT matrix) and the distance method (NJ, PAM matrix) with 500 bootstrap replicates. The group named hybrid NRPS/PKS concerns KS domains preceded by an NRPS. KS^Q domains and KS domains from loading modules with the organization ACP-KS-AT-AT-ACP are clustered and are called the KS^Q/2AT group. As the tree topologies obtained with both methods are similar, bootstrap values are indicated as ML/NJ. The strongly clustered KS domains of a PKSI are given with the name of the produced polyketide and their respective bootstrap values. The number of amino acid substitutions is proportional to the length of the scale. Sequences obtained from the metagenomic library are given in boldface type. The entire tree is available upon request.

FIG. 2.

Phylogenetic analyses of AT domains. The mtaBbis domain corresponds to the second AT domain in the unusual organization ACP-KS-AT-AT-ACP. The reconstruction was computed for 210 protein sequences on 169 sites by the ML method (BIONJ, JTT matrix) and the distance method (NJ, PAM matrix) with 500 bootstrap replicates. As the tree topologies obtained with both methods are similar, bootstrap values are indicated as ML/NJ. The AT domains located outside the two clusters for malonyl and methylmalonyl remain unsolved by phylogenetic analysis. The number of amino acid substitutions is proportional to the length of the scale except for mcyD, which has its branch length written on the dashed line. Sequences obtained from the metagenomic library are given in boldface type. The entire tree is available upon request.