Caracterização in silico de enzimas-chave da biossíntese de triacilgliceróis em dendezeiro, Elaeis guineensis, por anotação funcional e modelagem estrutural

Abstract

Oil palm (Elaeis guineensis Jacq.) is one of the main oilseed crops cultivated worldwide, and the oil extracted from its fruits has high economic and industrial relevance. Understanding the molecular mechanisms that control fatty acid biosynthesis and triacylglycerol (TAG) assembly in this species is essential to support breeding strategies aimed at increasing productivity and improving oil quality. In this study, the oil palm proteome was functionally annotated to in silico characterize central lipogenic enzymes in E. guineensis, integrating functional annotation, metabolic pathway analysis and structural prediction. The species proteome was obtained from NCBI and annotated with KofamScan using Hidden Markov Model (HMM) profiles specific for KEGG Orthology (KO). Only hits with HMM Score above the KO-specific threshold (as recommended by KOfamScan) and E- value < 1×10⁻¹⁰ were considered high-confidence. From the list of valid KOs, metabolic pathways were reconstructed in KEGG Mapper, revealing strong coverage of fatty acid biosynthesis and glycerolipid metabolism. Based on this annotation, five key enzymes were selected: acetyl-CoA carboxylase (ACC), 3-oxoacyl-ACP synthase I (KAS I), stearoyl-ACP desaturase (SAD/FAD2-like), glycerol-3- phosphate acyltransferase (GPAT) and diacylglycerol acyltransferase (DGAT). These proteins were subjected to physicochemical and structural characterization with ProtParam, SignalP, TMHMM, DeepLoc2 and CDDSearch, which confirmed the presence of conserved catalytic domains consistent with the function assigned to each KO, while post-translational regulation was evaluated with NetPhos 3.1 . The analysis of conserved motifs with MEME Suite identified signatures shared between plastidial and endoplasmic reticulum enzymes, suggesting common structural and regulatory elements along the lipid pathway. Finally, three-dimensional modeling of the five proteins with AlphaFold/ColabFold, followed by visualization in PyMOL, allowed mapping of domains, motifs and phosphorylation sites at the structural level. Taken together, the results support these five enzymes as in silico markers of lipogenesis in E. guineensis.