Long-read transcriptome assembly reveals vast isoform diversity in the placenta associated with metabolic and endocrine function

The placenta plays a critical role in fetal development and mediates maternal metabolic effects on offspring health outcomes. Despite its importance, the placenta remains understudied in large-scale genomic initiatives, with most analyses focusing on gene-level expression based on annotations from adult tissue references that obscure isoform diversity particularly vital to understanding function in developmental tissues. Here, we employed largest-in-class long-read RNA sequencing (N = 72) to generate a comprehensive placental transcript reference, yielding 37,661 high-confidence isoforms across 12,302 genes. Our assembly revealed 14,985 novel isoforms of known genes and 2,759 transcripts representing previously unannotated genes, with extensive diversity in genes involved in obesity, placental lactogen production, and growth control. We demonstrate this approach has two immediate practical advantages: First, the improved reference reduced inferential uncertainty in isoform quantification by approximately 37%, reduced the frequency of low-confidence annotations and increased the yield (read depth) of high-confidence annotations. Second, in analyses of gestational diabetes mellitus (GDM) with short-read RNA-seq datasets from two independent, multi-ancestry birth cohorts (N = 344) as an exemplar, we were able to identify novel isoforms of chorionic somatomammotropin hormone 1 ( CSH1 ) mediating the effect of maternal hyperglycemia on birth weight that was not apparent in conventional gene-level analyses. These findings demonstrate that isoform-level annotation of placental transcriptomics provides greater sensitivity and specificity than gene-level approaches. This enhanced precision may be essential for understanding placental function in developmental programming of metabolism and reveal ancestry- and context-specific variation in placental function and responses to environmental exposures.