Derbinski et al. 1 reported the expression of self-molecules in the mouse thymus. Their work follows earlier reports showing that self-molecules with tissue-restricted expression (or “peripheral” proteins) are also produced in the thymus through “ectopic” or

“promiscuous” transcription and translation2–5 and that such expression contributes to shaping a self-tolerant T cell repertoire6. Derbinski et al. used RT-PCR to detect peripheral protein transcripts in thymic cell populations purified by fluorescence-activated cell sorting after staining with markers specific for medullar or cortical thymic epithelial cells (mTECs and cTECs, respectively) and for bone marrow–derived APCs such as macrophages and DCs. Essentially all the molecules studied—including several autoantigens involved in autoimmune diseases such as type I diabetes and multiple sclerosis—were expressed by mTECs and not by DCs or macrophages.

These findings contrast with earlier reports3, 5, which showed that thymic cells expressing type I diabetes autoantigens (proinsulin and insulin, GAD and IA-2 and other pancreatic hormones) had surface markers that are typical of DCs and macrophages in human and mouse thymi, respectively. In our experiments, immunohistochemistry and double-immunofluorescence techniques were used to stain frozen tissue sections5. This approach prevents the detection of artifacts potentially associated with tissue processing and cell manipulation and allows the examination of cells in their native state. Unlike Derbinski et al. 1, but like Throsby et al. 3, we were unable to colocalize proinsulin, GAD and IA-2 with cytokeratin (a marker of thymic epithelial cells) 5. We were also unable to colocalize proinsulin with AIRE5, which is expressed in thymic epithelial cells and a subset of DCs7. If our findings were based on