iPSC Model Grows Human Thymus Cells That Train T Cells In Vitro

The thymus is the immune system's training academy, and its epithelial cells — thymic epithelial cells (TECs) — are the instructors. TECs guide developing T cells through selection, tolerance, and maturation. Yet how these specialized cells themselves arise from a common progenitor during human embryogenesis has remained deeply unclear, partly because human fetal thymic tissue is nearly impossible to obtain and study. This research from Kyoto University's Center for iPS Cell Research addresses that gap by building a complete, reproducible in vitro model of human thymus organogenesis using induced pluripotent stem cells (iPSCs).

The protocol begins by directing iPSCs through definitive endoderm and anterior foregut endoderm — well-established intermediary stages — before applying a narrow, precisely titrated dose of retinoic acid (RA) from day 7 to day 18 of culture. This RA window proved critical: it upregulated HOXA3, a transcription factor specifying the positional identity of the third pharyngeal pouch (3rd PP), the embryonic structure from which the thymus originates. FGF8 further enhanced expression of pharyngeal endoderm markers TBX1 and PAX9 and boosted FOXN1 (p<0.05 for each). Notably, by day 28, FOXN1 expression reached approximately 50% of the level seen in purified primary TECs from pediatric donors — a benchmark not previously achieved in fully in vitro systems. WNT activation at high doses abolished FOXN1 expression entirely, consistent with mouse data showing thymic disruption under elevated WNT signaling.

A key innovation was the use of FOXN1-mCherry reporter iPSC lines to track and isolate FOXN1-expressing cells in real time. After FOXN1+ thymic epithelial progenitor (TEP)-like cells emerged, the protocol shifted to self-directed differentiation — deliberately avoiding exogenous patterning signals to allow the cells to follow their own developmental logic. This yielded a highly heterogeneous population that, when profiled by single-cell RNA sequencing and benchmarked against published primary human fetal and postnatal TEC datasets, closely matched cortical TECs (cTECs), medullary TEC low (mTEClow), medullary TEC high (mTEChigh), and mimetic mTEC subpopulations — the full spectrum of mature TEC lineages. Pseudotime trajectory analyses allowed inference of developmental pathways from progenitors to each mature subtype.

The most stringent functional test came from co-culturing the induced TECs (iTECs) with hematopoietic progenitor-derived thymocytes. This co-culture system successfully generated naïve CD4+ and CD8+ T cells with diverse T cell receptor (TCR) repertoires, demonstrating that the iTECs are functionally competent to support positive selection. Strikingly, upon thymocyte co-culture, AIRE+ cells and post-AIRE mimetic subpopulations — which are responsible for central tolerance and tissue antigen presentation — emerged within the iTEC populations, suggesting that thymocyte-TEC crosstalk drives further TEC maturation in vitro just as it does in vivo.

The system was validated across three independent iPSC lines (201B7, 409B2, 1383D6), demonstrating reproducibility across genetic backgrounds. The authors acknowledge important caveats: the 2D culture system lacks the three-dimensional architecture of the native thymus, mesenchymal and vascular components present in vivo are absent, and the mimetic populations produced in vitro may not fully replicate their in vivo counterparts in terms of tissue-specific antigen repertoire. Nonetheless, this platform represents a transformative advance for studying congenital thymic disorders (such as DiGeorge syndrome), immune aging and thymic involution, and for developing cell-based regenerative therapies aimed at rebuilding immune competence.