Profiling maize embryonic leaf development and discovering new genes using high-resolution spatial long-read isoform sequencing.

Profiling transcriptome isoforms in their spatial context is instrumental for deciphering plant embryogenesis. By combining high-throughput full-length isoform sequencing and spatial transcriptomics (spatial MAS-IsoSeq) in maize embryogenesis, we identified 285,639 isoforms, 72.87% of which were previously uncharacterized. Gene models based on these full-length isoforms increased short-read exon mapping by 5.52%. Furthermore, spatial transcription expression detection improved by up to 97.45% in an extreme example. Using these isoforms, we constructed a new gene-model database (MaizeV5_IsoAnn) by integrating 5,228 novel genes and 1,674 genes with 5'- and/or 3'-flanking region extensions into the current maize reference gene models. Leveraging MaizeV5_IsoAnn, we reanalysed embryonic leaf cell transcriptomes to construct a refined time-ordered regulatory network and integrated it into multi-omics analyses with chromatin accessibility dynamics profiling, providing new insights into maize embryonic leaf development. Moreover, we propose LBD26 as an essential transcription factor in maize embryonic vein development. This study underscores the power of spatial MAS-IsoSeq to construct gene-model databases and elucidate developmental processes and mechanisms.