Hybrid transcriptome sequencing uncovers widespread shifts in transcript usage between mid-lactation and dry-off in...

Lactation and mammary involution and remodeling are complex biological processes that require the coordinate expression of thousands of genes. Differential expression analyses comparing lactating and dry goats have revealed extensive changes in the expression of protein-coding and non-coding RNAs in the mammary gland. Here, we hypothesize that lactation and mammary involution/remodeling may also involve changes in the abundance of transcripts differing in exon composition and functional properties. To test this hypothesis, we analyzed the mammary transcriptomes of 5 lactating and 4 dry goats by using a hybrid approach based on the integration of data from short-read Illumina and long-read Nanopore sequencing. After data filtering, we detected 21,598 transcripts derived from 12,300 genes (≈1.7 transcripts per locus) in the goat mammary gland. Among them, we found 14,092 annotated isoforms, 6,291 novel isoforms, and 1,215 novel loci. Around 39.3% of expressed genes generated multiple transcript variants. Overall, the goat mammary transcriptome showed exon skipping as the most frequent splicing event, followed by alternative use of initial exons, intron retention, and variations in 3' and 5' splicing sites. We also observed that a limited number of isoforms accounted for a very substantial fraction of the total expression output of the lactating mammary gland, with the top 10 and top 50 genes representing approximately 66% and 82% of total expression, respectively. This transcriptomic specialization is driven primarily by genes encoding caseins CSN1S1, CSN2, and CSN3, and major whey proteins such as progestagen associated endometrial protein (PAEP), α-lactalbumin (LALBA), and lactophorin (GLYCAM1), which are essential milk nutrients. Finally, differential transcript usage (DTU) analysis comparing lactating and dry goats revealed 443 isoform switches affecting 355 unique genes and 563 transcripts. Besides, 266 and 297 transcripts were upregulated and downregulated in lactating goats, respectively, and 413 DTU, affecting 248 genes, were predicted to have functional consequences. Among these functional consequences, the most important ones were protein domain gain, non-reference domain isoform gain, nonsense-mediated insensitivity and coding transcripts. Several of the genes showing DTU have important roles in lactation, being of particular relevance those encoding epidermal growth factor receptor (EGFR), glycerol-3-phosphate acyltransferase, mitochondrial (GPAM), hydroxysteroid 11-β dehydrogenase 1 (HSD11B1), insulin receptor substrate 1 (IRS1), nuclear receptor subfamily 3 group C member 1 (NR3C1), and phosphoinositide-3-kinase regulatory subunit 1 (PIK3R1). Moreover, we also detected DTU for several genes integrated in the mitogen-activated protein kinase and Rho GTPase pathways. In summary, we provide a comprehensive catalog of RNA isoforms expressed in the goat mammary gland and demonstrate that mammary transcript splicing patterns differ substantially between lactating and dry goats.