Curacin A from the marine cyanobacterium Lyngbya majuscula has unusual chemical groups compared to most polyketide natural products: a cyclopropane ring and a terminal alkene.  The polyketide synthase (PKS) for curacin A biosynthesis is also unusual: the loading module includes an apparent N-acetyltransferase, and the cyclopropanation module is only subtly different from a vinyl-chloride-producing module in the jamaicamide PKS from another strain of L. majuscula.  Structural, biochemical and analytical experiments have provided an explanation for some of these unusual features of curacin A and its PKS.  The loading module catalytic domain is a member of the GNAT superfamily (GCN5-related N-acetyltransferases) with unprecedented dual activities of decarboxylation and S-acetyl transfer.  Homologs of the curacin A GNAT loading module in several PKSs have conserved His and Thr side chains that are critical for decarboxylation but not for acetyl transfer.  Further along the curacin A gene cluster are a highly conserved HCS (hydroxymethyl-glutaryl-CoA synthase – like) gene cassette and coding sequences for an Fe-dependent halogenase (Hal) and a PKS enoyl reductase (ER).  These highly similar enzymes form cyclopropane in the curacin A PKS and vinyl chloride in the jamaicamide PKS.  The HCS cassette encodes a ketosynthase, an acyl carrier protein (ACP), an HCS, a dehydratase (ECH1), and a decarboxylase (ECH2) domain fused to the ER.  Subtle differences in the ECH2 and ER catalytic domains account for the strikingly different cyclopropane and vinyl chloride products.