BMC Plant Biology Науки о растениях
Deciphering the gene regulatory network underlying storage protein and lipid accumulation during walnut embryo development via machine learning and a co‑regulation strategy
Аннотация
Background Walnuts ( Juglans regia L.) are a rich source of proteins and lipids, the accumulation of which during embryo development is governed by complex gene regulatory networks (GRNs). However, the transcriptional regulation underlying protein and lipid biosynthesis during walnut embryo development remains poorly understood. Results In this study, we collected 90 transcriptomes datasets spanning various embryonic developmental stages from six Juglans regia cultivars and one Juglans sigillata cultivar. Utilizing two machine learning algorithms, GENIE3 and GRNBoost2, we reconstructed the first comprehensive GRNs for walnut embryo development. Based on these dual GRNs, we de novo predicted the co-regulation of four CRU1 genes, which encode glutelin, the most abundant storage protein in walnuts, by the FUS3 transcription factor. Furthermore, the GENIE3- and GRNBoost2-derived GRNs identified potential co-regulators for 1,115 and 1,127 co-expressed paralogous gene pairs (CEPs), respectively. We employed a rigorous co-regulation strategy evaluating the correspondence between co-regulator binding sites and constrained regions derived from pairwise promoter alignments of these CEPs. Through this analytical framework, we identified the cross-species conserved regulation of nine CEP groups by the WRI1 transcription factor, and further validated two previously overlooked lipid biosynthesis-related target genes, CUE1 and PLE2 . Additionally, we de novo predicted and confirmed that the transcription factor NF-YA1-1 modulates embryonic fatty acid composition by activating a KASII paralogous gene pair. Conclusions Collectively, these data provide a foundation for unraveling the regulatory mechanisms of protein and lipid accumulation in walnut embryos. Moreover, this proposed co-regulation strategy offers a highly transferable approach for extracting evidence of regulation from GRNs in other species.