Because of tolerance mechanisms, it has been hard to identify the T cell receptors (TCRs) of high-avidity T cells against self (for example, tumor) antigens. TCRs that are specific for foreign human antigens from the nontolerant T cell repertoire can be identified in mice. Moreover, if mice are constructed to express the human TCR repertoire, they can be used to analyze the unskewed repertoire against human self antigens. Here we generated transgenic mice with the entire human TCRalphabeta gene loci (1.1 and 0.7 Mb), whose T cells express a diverse human TCR repertoire that compensates for mouse TCR deficiency. A human major histocompatibility class I transgene increases the generation of CD8+ T cells with human compared to mouse TCRs. Functional CD8+ T cells against several human tumor antigens were induced, and those against the Melan-A melanoma antigen used similar TCRs to those that have been detected in T cell clones from individuals with autoimmune vitiligo or melanoma. These mice will allow researchers to identify pathogenic and therapeutic human TCRs.

Identifying T-cell receptors (TCRs) that bind tumor-associated antigens (TAAs) with optimal affinity is a key bottleneck in the development of adoptive T-cell therapy of cancer. TAAs are unmutated self proteins, and T cells bearing high-affinity TCRs specific for such antigens are commonly deleted in the thymus. To identify optimal-affinity TCRs, we generated antigen-negative humanized mice with a diverse human TCR repertoire restricted to the human leukocyte antigen (HLA) A*02:01 (ref. 3). These mice were immunized with human TAAs, for which they are not tolerant, allowing induction of CD8⁺ T cells with optimal-affinity TCRs. We isolate TCRs specific for the cancer/testis (CT) antigen MAGE-A1 (ref. 4) and show that two of them have an anti-tumor effect in vivo. By comparison, human-derived TCRs have lower affinity and do not mediate substantial therapeutic effects. We also identify optimal-affinity TCRs specific for the CT antigen NY-ESO. Our humanized mouse model provides a useful tool for the generation of optimal-affinity TCRs for T-cell therapy.

Cortical (cTEC) and medullary (mTEC) thymic epithelial cells establish key microenvironments for T-cell differentiation and arise from thymic epithelial cell progenitors (TEP). However, the nature of TEPs and the mechanism controlling their stemness in the postnatal thymus remain poorly defined. Using TEC clonogenic assays as a surrogate to survey TEP activity, we found that a fraction of cTECs generates specialized clonal-derived colonies, which contain cells with sustained colony-forming capacity (ClonoTECs). These ClonoTECs are EpCAM+MHCII-Foxn1lo cells that lack traits of mature cTECs or mTECs but co-express stem-cell markers, including CD24 and Sca-1. Supportive of their progenitor identity, ClonoTECs reintegrate within native thymic microenvironments and generate cTECs or mTECs in vivo. Strikingly, the frequency of cTECs with the potential to generate ClonoTECs wanes between the postnatal and young adult immunocompetent thymus, but it is sustained in alymphoid Rag2-/-Il2rg-/-counterparts. Conversely, transplantation of wild-type bone marrow hematopoietic progenitors into Rag2-/-Il2rg-/-mice and consequent restoration of thymocyte-mediated TEC differentiation diminishes the frequency of colony-forming units within cTECs. Our findings provide evidence that the cortical epithelium contains a reservoir of epithelial progenitors whose abundance is dynamically controlled by continual interactions with developing thymocytes across lifespan.

Axons have generally been represented as straight cylinders. It is not at all uncommon for anatomists to take single cross-sections of an axonal bundle, and from the axonal diameter compute expected conduction velocities. This assumes that each cross-section represents a slice through a perfect cylinder. We have examined the three-dimensional geometry of 98 central and peripheral unmyelinated axons, using computer-assisted serial electron microscopy. These reconstructions reveal that virtually all unmyelinated axons have highly irregular axial shapes consisting of periodic varicosities. The varicosities were, without exception, filled with membranous organelles frequently including mitochondria, and have obligatory volumes similar to that described in other neurites. The mitochondria make contact with microtubules, while the other membraneous organelles were frequently found free floating in the cytoplasm. We conclude that unmyelinated axons are fundamentally varicose structures created by the presence of organelles, and that an axon's calibre is dynamic in both space and time. These irregular axonal geometries raise serious doubts about standard two dimensional morphometric analysis and suggest that electrical properties may be more heterogeneous than expected from single section data. These results also suggest that the total number of microtubules contained in an axon, rather than its single section diameter, may prove to be a more accurate predictor of properties such as conduction velocity. Finally, these results offer an explanation for a number of pathological changes that have been described in unmyelinated axons.