Phylogenetic relationships and characterization of UDP-glycosyltransferases (UGTs) in dicotyledonous plants

HUGO MARCELO ATENCIO; NICOLÁS STOCCHI; AGUSTIN AMALFITANO; FERNANDO VILLARREAL; ARJEN TEN HAVE

Mendoza

Congreso; X Congreso Argentino de Bioinformática y Biología Computacional.; 2019

A2B2C

BACKGROUND The protein superfamily UDP-Glycosyl Transferase (UGT) is highly divergent. UGTs transfer sugar moieties from a variety of UDP-activated sugars to a variety of acceptor molecules. With respect to both acceptor and donor, promiscuity has been demonstrated and has likely resulted in erroneous function assignation. Furthermore, since UGTs are important in secondary metabolism they have evolved towards a complex superfamily. This might explain why many UGT phylogenies do not correspond with functional classification. As such, current classification is based on sequence identity, were plant UGTs fall into the classes 71 to 100. We are interested in plant UGTs involved in flavonoid and anthocyanin synthesis and need to identify the paralogs involved. In the initial sequence mining we identifiied many partial sequences as well as many pseudogenes. The presence of these erroneous sequences in datasets results in erroneous MSAs and therewith erroneous phylogenies. Identification of such ?bad? sequences from a complex dataset is difficult and in general subjective and labor consuming.OBJECTIVETo reconstruct and characterize a high fidelity phylogeny for structure-function prediction of plant dicotyledoneous UGTs by applying an automated, objective sequence scrutiny.RESULT To identify UGT homologues we applied HMMER to 17 complete proteomes of dicotyledonous plants and the contro-set of SwissProt. SEQrutinator was used for the objective sequence scrutiny. SEQrutinator detects and removes short and non-homologous sequences; sequences that instigate large contiguous gaps; or sequences with large gaps in an MSA. A total of 2339 sequences were accepted as verified UGT homologues. This dataset was clustered using HMMERCTTER, to group sequences based on a phylogeny from high quality data, to get cluster-specific HMMER profiles detecting 14 clusters of UGTs, without the presence of orphan sequences, with 100% precision and recall (P&R). Meta-analysis of properly annotated sequences showed the clustering presents functional clustering. well. Finally, using the SDPFox software, 33 CDPs (Cluster Determining Positions) were detected that, upon verification by Mutual Information Network analysis, will be used for structure-function prediction..CONCLUSIONSA high fidelity phylogenetic tree of UGTs was obtained and will contribute to a more precise functional annotation based on studies of structure-function prediction.