Organoid transplants rebuild uterine lining and restore fertility

What the work claims

This is a primary in vivo study in mice, still a preprint. Mopure, Kim, Ang, Davis, Spencer, McKinley and Kelleher built a genetic ablation of the uterine epithelium, transplanted cultured endometrial epithelial organoids (EEOs) into the denuded lumen, and scored the readout that ultimately matters: live pups. Three claims follow. First, organoids engraft and rebuild both epithelial compartments of the uterus after near-total epithelial loss, and this restores fertility that spontaneous repair did not. Second, when the transplanted organoids are stripped of their glandular lineage before grafting, the surviving luminal-lineage cells convert to functional glandular epithelium in vivo, a conversion the authors show does not happen in culture. Third, the same luminal-to-glandular trajectory operates during the endogenous repair of selectively ablated glands, meaning the plasticity is a property of the tissue, not an artifact of transplantation.¹

How it works

The engineering is the strength of the paper. Crossing an epithelium-specific Ltf-iCre driver with a diphtheria-toxin receptor line lets a single intraperitoneal toxin dose (25 micrograms per kilogram) kill about 94 percent of luminal and glandular epithelium while sparing stroma; flow and histology quantification of the ablation in three mice per group gave P = 0.0025. The denuded surface re-epithelializes within ten days from surviving cells, yet females remained infertile across a three-month breeding trial (P = 0.0020, four mice per group). Morphological repair, in other words, is not functional repair. Donor EEOs came from a dual-reporter line labeling the glandular lineage with GFP and everything else with RFP; at passage 3 the cultures ran roughly 80 percent RFP and 20 percent GFP (P = 0.0006, three donors). One million cells in matrix were injected into the uterine lumen a day after ablation.¹

Thirty days later, donor cells formed both compartments: RFP-positive luminal epithelium expressing the luminal marker CALB1, and GFP-positive, FOXA2-positive glands. Mated recipients implanted only in the transplanted horn by gestational day 12 ultrasound, carried pregnancies to term, and produced live pups (one-way ANOVA across groups, P less than 0.0001, five mice per group). The decisive experiment depletes glandular cells from the donor organoids in vitro with diphtheria toxin before grafting, leaving a pure luminal-lineage graft. These cells still built FOXA2-positive glands in vivo, switching on the glandular reporter as they did so. A parallel in situ experiment ablated only the glands (5 nanograms of toxin delivered into one horn) and watched repair over ten days: CALB1-positive buds invaginate from the luminal surface by day 3, begin expressing the glandular marker by day 6, and mature into CALB1-negative, FOXA2-positive glands by day 10. Fertility after this endogenous regeneration was indistinguishable from controls (P = 0.7854, five mice per group), confirming the regenerated glands work.¹

The authors' interpretation is that the adult uterine epithelium carries facultative progenitor behavior, reactivating a latent developmental program in which glands bud from the luminal epithelium, and that stromal signals are the likely instructors of the conversion.¹

Where a skeptic should push

The load-bearing assumption is that luminal-lineage cells convert, rather than a rare unlabeled bipotent progenitor expanding preferentially. The authors flag this honestly: the Cxcl15-based glandular labeling could miss rare non-glandular recombination, and bipotent cells at the luminal-glandular junction have been proposed by others. The cleanest support for true conversion is the temporal sequence in situ, but formal clonal proof is absent. Second, generality: everything is mouse, on a C57BL/6J background, with small groups (three to five animals for the headline statistics). Third, and most important for translation, the injury model preserves a healthy underlying stroma. Real clinical targets such as intrauterine adhesions (Asherman's syndrome) or a fibrotic postpartum endometrium present an injured niche, and this study cannot tell us whether an injured niche still instructs luminal-to-glandular conversion or instead fails the graft. The work is also epithelial-focused: stromal, vascular and immune contributions to regeneration are explicitly out of scope.¹

What engraftment plasticity means for organoid therapies

For the organoid field, this paper is a working demonstration of organoids as transplanted therapeutic products, and its non-obvious lesson cuts against the current instinct in cell therapy. The manufacturing reflex is to purify the exact cell type you want and reject lots that drift from identity. Here, purity bought nothing: organoids stripped of the glandular lineage rebuilt fully functional glands, because the host niche reprogrammed what arrived. If that holds beyond this model, the release criteria for organoid cell therapies should weigh engraftment competence and responsiveness to niche signals at least as heavily as starting identity, and potency assays measured only in a dish will misjudge products whose fate is decided after transplant.¹

There is also a benchmark lesson hidden in the controls. Spontaneous re-epithelialization after 94 percent ablation restored a closed luminal surface within ten days but left animals infertile; only gland reconstruction, by graft or by endogenous conversion, restored function. Any regenerative-medicine program that scores morphology or resurfacing as its endpoint is measuring the thing that failed here. Fertility, implantation and decidualization are the readouts that separated success from failure, and organoid transplant platforms are unusually good at hosting such functional readouts because the graft is genetically marked and its lineage contribution can be tracked.¹

The threat is the mirror image of the opportunity. A graft whose identity is set by the host will take the host's instructions wherever it lands, including places you did not intend: ectopic endometrial-like tissue is itself a disease model (endometriosis is ectopic endometrium), so engrafted, plasticity-prone epithelium carries a real risk of establishing lesions at leakage sites, and the paper's own design shows how little starting identity constrains the outcome. The mouse-to-human gap is equally unpriced: human endometrial organoids exist and engraft in other models, but whether the human luminal-to-glandular conversion operates on the same rules, in a perimenopausal or fibrotic niche, is exactly what this study cannot say. A genuine threat to translation, not a footnote.¹

The bottom line

Established in mice: near-total epithelial ablation renders the uterus infertile despite luminal resurfacing; transplanted endometrial organoids rebuild luminal and glandular epithelia and restore fertility to term; luminal-lineage grafts convert to functional glands in vivo; and the same conversion runs endogenous repair after selective gland ablation. Open: whether true lineage conversion (rather than a rare bipotent progenitor) explains the result, what the stromal instructive signals are, and whether any of it holds in human tissue or in injured stroma. What would settle it: clonal lineage tracing of the converting cells, defined-signal perturbations in the engraftment niche, and a replication in a fibrosis model where the stroma is the variable.¹