The target specificity of the transcription factor AIRE, which regulates a subset of gene expression in medullary thymic epithelial cells (mTECs), has so far been unclear. Reporting in Nature, Fang et al. use two independent approaches — involving a deep-learning algorithm and analysis of natural variation in regulatory sequences — to show that AIRE targets Z-DNA, and to a lesser extent NFE2-binding motifs, associated with double-strand breaks (DSBs) and promoter poising.

To confirm the contribution of (CA)n repeats and NFE2-binding motifs to the cis-regulation of AIRE-induced gene expression, Fang et al. looked at the effects of natural genetic variation between non-obese diabetic (NOD; a model for autoimmune diabetes) and B6 mouse strains on the expression of AIRE-induced genes in mTECs. In F1 hybrids, the team identified genetic variants in (CA)n-containing motifs and NFE2L2-binding motifs that are associated with allelic differences in chromatin accessibility and transcription of nearby AIRE-induced genes. For example, the NOD allele of the AIRE-induced gene Marcksl1 had a longer (CA)n motif than the B6 allele, and expression of Marcksl1 was greater from the NOD allele than the B6 allele in F1 hybrids.