The retinoid receptors RARA and RXRA are both important transcription factors that influence hematopoietic cell growth and differentiation. RARA and RXRA are both ligand-dependent transcription factors. The natural ligand for RARA is thought to be all-trans retinoic acid (ATRA), the natural ligand for RXRA is unclear, although 9-cis retinoic acid is an active ligand in vitro. Aldehyde dehydrogenase (ALDH) metabolism is the rate-limiting step in ATRA synthesis, and ALDH activity is associated with stem cell self-renewal in hematopoietic stem cell (HSCs) and in cancer stem cells. It is unknown whether these two functions are related.

In order to measure the presence and regulation of natural retinoids in vivo, we developed a UAS-GFP reporter mouse. We found this system to be highly sensitive and specific. We transplanted UAS-GFP mouse bone marrow cells with virus expressing either Gal4-RARA-IRES-mCherry (Gal4-RARA-IC) or Gal4-RXRA-IRES-mCherry (Gal4-RXRA-IC). Because PML-RARA is proposed to act as a dominant-negative, and ATRA induces significant differentiation of myeloid cells, we were surprised to observe no GFP expression in mice transplanted with Gal4-RARA-IC. Instead, we observed GFP in mice transplanted with Gal4-RXRA-IC, consistent with natural RXRA ligands, but not RARA ligands, in bone marrow cells. When we treated mice with either ATRA or bexarotene, we observed that most hematopoietic cell types can respond to an active retinoid, with the exception of Kit+Lin-Sca+ HSCs, which had a significantly attenuated response. Ex vivo, we found that the P450 inhibitors liarozole and talarozole augmented response to retinoids in Kit+ hematopoietic stem/progenitor cells, suggesting that HSCs have high rates of retinoid degradation via active P450 pathways, and thus maintain a retinoid deplete environment. We further tested whether hematopoietic cells might respond to hematopoietic stress through retinoid receptor signaling. We observed a significant increase in the number of GFP+ cells when Gal4-RXRA-IC transplanted mice were treated with either 5FU or GCSF, but no response in Gal4-RARA-IC transplanted mice. As a control, we repeated the studies using a Gal4-RXRA vector with the AF2 domain deleted. The RXRAdeltaAF2 mutation can bind to ligand, but does not respond to it, although it still can be activated through a heterodimeric partner. We observed no GFP induction by 5FU or GCSF with the RXRAdeltaAF2 mutation, suggesting that the GFP response is to natural RXRA ligands, and not to alternative signaling through a heterodimeric partner (e.g. Lxra or Pparg).

These data suggest that HSCs maintain low levels of natural retinoids, and that the mechanism of stem-cell associated ALDH is therefore not through ATRA production and RARA activation. In addition, bone marrow cells are exposed to natural RXRA ligands, but not RARA ligands, under homeostatic conditions, and this increases during response to 5FU and GCSF, suggesting that ligand-dependent RXRA activation may play a critical role in hematopoietic response to 5FU and GCSF.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.