Female fertility is anchored to a biological countdown. Mammalian females are born with a finite reserve of oocytes, and both the quantity and the quality of that reserve erode with time, a process that underlies age-related subfertility, poorer outcomes in assisted reproduction and disorders such as premature ovarian insufficiency. Oocytes do not age alone: they are enveloped by granulosa cells, somatic nurses that exchange metabolites and signaling molecules with the egg through gap junctions and coordinate its growth. When either partner in this duo falters, the whole follicle suffers. Yet most of what scientists know about the aging ovarian transcriptome comes from short-read RNA sequencing, which measures gene-level expression by tallying millions of tiny fragments. That approach has a blind spot. A single gene can yield many isoforms through alternative splicing, alternative promoters and alternative polyadenylation, and those isoforms may encode different proteins or carry distinct regulatory elements. Gene-level counts collapse this diversity, so a gene can appear unchanged with age even as it quietly swaps one transcript version for another.

To see past that blind spot, the team turned to Oxford Nanopore Technologies sequencing, in which RNA molecules, or their cDNA copies, are threaded through protein nanopores and read end to end. Where Illumina short reads chop RNA into pieces of a few hundred bases and reconstruct transcripts computationally, a strategy that quantifies genes well but struggles to assemble full-length isoforms from complex loci, nanopore reads can span an entire messenger RNA, capturing the complete splice pattern along with the transcription start and end sites that define each isoform. The researchers applied this platform to granulosa cells and germinal vesicle oocytes, the immature eggs arrested before ovulation, isolated from young mice aged 6 to 8 weeks and aged mice at 10 months of the outbred ICR strain. Illumina short-read data were generated in parallel as orthogonal support for quantification. On top of the sequencing, the team layered transcript annotation, differential expression analysis, alternative polyadenylation analysis and weighted gene co-expression network analysis, adapting analytical frameworks first developed to chart isoform diversity in the developing human brain.

The annotation effort alone was revealing. The pipeline classified 130,730 high-confidence transcripts, including over 100,000 putative novel isoforms, molecules that either extend known gene models or splice exons together in combinations absent from reference databases. Many of these newly described transcripts fall into the categories of novel-in-catalog and novel-not-in-catalog, meaning they reshuffle known gene structures or invent entirely new ones, while others represent incomplete splice matches that refine existing annotations. Strikingly, the authors report that several novel isoforms of disease-associated genes are undetectable at the gene level, and that many genes relevant to reproductive medicine show cell-type-specific isoform usage, deploying different transcript versions in granulosa cells than in oocytes. In other words, the ovary’s transcriptomic repertoire is far richer than standard annotations suggest, and a meaningful slice of that richness had been hidden inside gene-level averages that conventional sequencing was never designed to resolve.

Aging, the study found, bends this repertoire in a consistent direction: toward isoforms with lower predicted coding potential. The researchers scored each transcript with the Coding-Potential Assessment Tool, an algorithm that distinguishes protein-coding RNAs from non-coding ones based on features such as open reading frame length and nucleotide composition. Across both cell types, aged ovaries showed a shift in isoform usage favoring transcripts predicted to be less protein-coding than their counterparts in young tissue. Exploratory enrichment analysis then hinted at what those low-coding-potential transcripts might be doing, linking them to biological processes that include protein synthesis and chromosome segregation, two processes on which oocyte quality depends exquisitely. The authors are careful to frame this as exploratory and prediction-based rather than proof that aged ovaries abandon full-length protein production at scale. Still, the observation raises an intriguing possibility: that part of reproductive aging plays out not in how much RNA the ovary makes, but in what kind of RNA it chooses to make.

The study also caught aging in the act of trimming transcripts from their tails. Through alternative polyadenylation, a gene can route its mRNA to different stop signals, producing versions with longer or shorter 3′ untranslated regions, the stretches that do not code for protein but act as docking sites for microRNAs and other regulators of mRNA stability, localization and translation. Using the percentage of distal polyadenylation site usage index, a standard metric for such shifts, the team identified age-associated 3’UTR shortening in the ovary. Shortened 3’UTRs typically shed microRNA binding sites, which can stabilize messages and change how much protein they yield; in oocytes, where maternal RNA stockpiles must be precisely remodeled to support fertilization and early embryonic development, that kind of regulation is anything but cosmetic. The finding suggests that aging may retune post-transcriptional gene control in the ovary, a layer of regulation that conventional gene-expression studies cannot see.

The most consequential numbers came from the isoform-switching analysis. The team identified 795 significant switching events across the two cell types, cases in which young and aged samples favor different isoforms of the same gene, measured as a change in each transcript’s share of the gene’s total output. These were not silent swaps. Many of the switches were associated with predicted changes to open reading frames, meaning the protein a gene produces could differ structurally between young and old ovaries, and with the potential loss of protein domains, the functional building blocks that give proteins their binding and catalytic abilities. A granulosa cell in a 10-month-old mouse may therefore be running a different molecular toolkit than its 8-week-old counterpart, even when gene-level expression looks identical. Because granulosa cells orchestrate ovulation, steroid production and oocyte maturation, isoform-level rewiring in these cells offers a plausible mechanistic thread connecting aging to declining egg quality.

To map the aging transcriptome’s higher-order structure, the researchers applied weighted gene co-expression network analysis, which groups transcripts into modules whose members rise and fall together and flags the highly connected hub transcripts within each module. The analysis highlighted modules tied to aging and to cell-type identity, and among the hubs sat TALONT000180938, a transcript derived from Esr1, the gene encoding estrogen receptor alpha. That choice of hub gene is provocative: Esr1 has long been linked to ovarian function and ovarian disease, sitting at the center of the estrogen signaling that drives follicle growth and granulosa cell proliferation. That an Esr1 isoform emerges as a hub in these networks suggests estrogen signaling itself may be remodeled at the isoform level with age, a hypothesis the authors present as a candidate for follow-up rather than a settled mechanism. Many other disease-associated genes likewise exhibited cell-type-specific isoform usage that would have been invisible to gene-level studies.

The findings come with appropriate guardrails. The work was done in mice, and while 10-month ICR mice reproduce key features of reproductive aging, mice are not women, and the timing of ovarian decline differs between species. The coding-potential results are computational predictions, the enrichment findings are labeled exploratory, and the network analyses describe correlations rather than causes. The authors note that laboratory validation, for instance with reverse transcription PCR and rapid amplification of cDNA ends, techniques that pin down exact splice junctions and transcript boundaries, will be needed to confirm individual candidates. Even so, the resource itself stands out: a full-length, isoform-resolved transcriptome of the mammalian ovary across age, built cell type by cell type. It hands reproductive biologists a dense map of where to look next, from the 795 switching events to the shortened 3’UTRs to the disease genes whose novel isoforms had previously escaped detection.

The broader stakes reach toward the fertility clinic. Ovarian reserve is currently gauged with indirect markers such as hormone levels and antral follicle counts, and no routine test measures the transcript-level state of the aging egg environment. If age-associated isoform remodeling is confirmed in human ovaries, transcript signatures like the ones catalogued here could eventually inform how clinicians assess oocyte quality, predict responses to ovarian stimulation or investigate premature ovarian insufficiency. For now, the study’s message is more fundamental: the transcriptome is not a flat list of genes but a population of molecules that ages in its own right. In the aging ovary, that population tilts, truncates and switches, and with long-read sequencing, researchers finally have the resolution to watch it happen, one full-length RNA molecule at a time.