Abstract
Cyclodipeptide synthases (CDPSs) use as substrates two amino acids activated as aminoacyl-tRNAs to synthesize cyclodipeptides in secondary metabolites biosynthetic pathways. Since the first description of a CDPS in 2002, the number of putative CDPSs in databases has increased exponentially, reaching around 800 in June 2017. They are likely to be involved in numerous biosynthetic pathways but the diversity of their products is still under-explored. Here, we describe the activity of 32 new CDPSs, bringing the number of experimentally characterized CDPSs to about 100. We detect 16 new cyclodipeptides, one of which containing an arginine which has never been observed previously. This brings to 75 the number of cyclodipeptides formed by CDPSs out of the possible 210 natural ones. We also identify several consensus sequences related to the synthesis of a specific cyclodipeptide, improving the predictive model of CDPS specificity. The improved prediction method enables to propose the main product synthesized for about 80% of the CDPS sequences available in databases and opens the way for the deciphering of CDPS-dependent pathways. Analysis of phylum distribution and predicted activity for all CDPSs identified in databases shows that the experimentally characterized set is representative of the whole family. Our work also demonstrates that some cyclodipeptides, precursors of diketopiperazines with interesting pharmacological properties and previously described as being synthesized by fungal non-ribosomal peptide synthetases, can also be produced by CDPSs in bacteria.