Plant & Animal Genomes XIII Conference

January 15-19, 2005
Town & Country Convention Center
San Diego, CA

Loline And Ergot Alkaloid Synthesis Genes In Grass Endophytes

Christopher L. Schardl¹ , Robert B. Grossman² , Daniel G. Panaccione³

¹  Department of Plant Pathology, University of Kentucky, 201F PSB, 1405 Veterans Dr., Lexington, KY, 40546-0312, USA

²  Department of Chemistry, University of Kentucky, Lexington, KY, 40506-0055, USA

³  Division of Plant & Soil Sciences, West Virginia University, 401 Brooks Hall, Morgantown, WV, 26506-6058, USA

Fungi of genera Neotyphodium and Epichloë are seed-transmissible endophytes symbiotic with cool-season grasses, and often help protect host plants from biotic insults and abiotic stresses. Four classes of endophyte alkaloids kill, sicken or deter herbivores: the ergot alkaloids and indole-diterpenes are active against mammals, and the lolines (1-aminopyrrolizidines) and peramine act against insects. Also, ergot alkaloids produced by Claviceps spp. are used medicinally. Recently, genes for two ergot alkaloid synthesis enzymes have been characterized: dmaW, which encodes a tryptophan prenyl transferase, and two genes for the heterodimeric lysergyl peptide synthetase. In Claviceps spp. these are clustered with several genes predicted to encode enzymes for oxidation and cyclization steps. Interestingly, although there is no apparent relationship of DmaW to other known prenyl transferases, homologues of dmaW are present in other fungal gene clusters. Therefore, DmaW may be an archetype of a novel class of prenyl transferases in natural product biosynthesis. Loline alkaloid gene clusters have also been identified in Neotyphodium and Epichloë spp. Several genes in these clusters are related to amino acid metabolism genes. LolC probably evolved from homocysteine synthase, a methionine pathway enzyme that catalyzes the condensation of homoserine with H₂S. However, feeding studies indicate that proline and homoserine - but not methionine - are precursors of the core loline structure. Characterizations of dmaW and lolC illustrate both the power and current limitations of bioinformatic tools for predicting gene function.