The X-RARA fusion proteins have reduced sensitivity to all-trans-retinoic acid (ATRA), and have been proposed to act by decreasing retinoid-dependent transcription required for myeloid maturation. 1-3 We therefore sought to determine whether maturing myeloid cells are exposed to natural retinoid ligands in vivo. Surprisingly, we detected a paucity of natural retinoids capable of transactivating RARA-dependent transcription in adult mouse bone marrow cells, and the trace activity we observed tended to be in erythroidlineage cells, rather than in myeloid-progenitors. This suggests that the resistance to retinoid-mediated transactivation likely has a limited role in X-RARA-dependent leukemogenesis because natural retinoids are largely absent during normal myeloid maturation. The presence and distribution of natural retinoids have not been studied in adult hematopoiesis. To detect retinoids capable of transactivating RARA, we developed a UAS-GFP reporter mouse. UAS promoter sequences are recognized by the yeast Gal4 transcription factor, and are not activated by mammalian proteins (schema in Figure 1a). When the modular Gal4-DNA-binding domain is fused to the RARA ligand-binding domain, and expressed in UAS-GFP bone marrow cells, the reporter specifically detects intracellular retinoids that bind and transactivate RARA. This approach improves specificity of retinoid detection compared with alternative approaches, such as using the retinoid response element from the RARB promoter, which may respond nonspecifically to Rara/Rxra and Rarg/Rxra heterodimers, and Rxra/Rxra homodimers. The mouse embryonic stem cell clone used to generate the UAS-GFP mice was selected through a series of functional assays, to determine responsiveness and background of the randomly integrated transgene (Supplementary Figure 1). We observed only trace background GFP expression in UAS-GFP mice in the absence of a Gal4 fusion protein, with the exception of a small population of GFPdim lymphocytes in the peripheral blood and spleen (Supplementary Figure 2).

The UAS-GFP reporter was sensitive and specific to retinoids ex vivo. When Kit+ bone marrow cells were transduced with retrovirus expressing Gal4-RARA-IRES-mCherry (Gal4-RARA-IC) or Gal4-RARG-IRES-mCherry (Gal4-RARG-IC), we observed a dosedependent response to ex vivo ATRA, with subnanomolar sensitivity (Figure 1a, Gal4-RARA-IC: EC50 0.36±0.14 nM; Gal4-RARG-IC: EC50 0.16±0.1 nM). This corresponds with the in vitro-measured Kd using radiolabeled ATRA (Kd= 0.2 and 0.2 nM, respectively). 4 Synthetic, receptor-specific ligands induced receptor-specific GFP expression (Figure 1b, RARA-specific agonist: BMS753; RARG-specific agonist: BMS961). RARG binds to co-repressors with lower affinity than RARA. 5 Consistent with this, Gal4-RARG-IC was modestly more sensitive to ATRA than Gal4-RARA-IC (Figure 1a), and Gal4-RARG-IC induced a small increase in the background GFP expression in the absence of exogenous ligand (Figure 1b column 5). In order to determine whether bone marrow cells are exposed to natural retinoids in vivo, and whether this correlates with specific stages of hematopoietic differentiation, we transduced UAS-GFP Kit+ cells with Gal4-RARA-IC or Gal4-RARG-IC, and transplanted these cells into lethally irradiated recipient mice (the RARG vector was included to improve sensitivity of natural retinoid detection and to validate the RARA findings). The recipient mice were then maintained on standard chow (PicoLab 20, Labdiet, St. Louis, MO, USA, which contains 15 IU/g vitamin A) and then analyzed after complete engraftment …