RAR-Alpha Targeting Compounds Overcome Bone Marrow (BM) Stromal Protection of AML By CYP26

All-trans retinoic acid (ATRA) causes terminal differentiation and apoptosis of non-APL AML cells in vitro, but has not proven clinically effective. We have recently shown that BM stroma expressing CYP26 inactivates ATRA and protects leukemia cells from differentiation (Su M et al 2015). ATRA signals through retinoic acid receptors (RARs) α, β and γ and induces not only terminal differentiation of AML, but also upregulation of CYP26, potentially forming an even more protective niche for leukemia. In an effort to decouple these two biological effects and improve differentiation therapy for AML, we investigated the relative contribution of RARα and RARγ to differentiation and CYP26 upregulation.

ATRA (pan-RAR agonist), AM80 (strong RARα agonist, weak RARγ), and IRX5183 (RARα-specific agonist; previously known as NRX 195183, Io Therapeutics, Inc.) all induced significant differentiation of NB4 APL cells (i.e., upregulation of myeloid differentiation antigen CD11b and decreased clonogenic activity) at 0.1 μM. In contrast, the RARγ-specific agonist CD437 (0.1 μM) produced no evidence of differentiation. To determine the impact of RARα and RARγ signaling on stromal CYP26B1 levels, we treated murine OP9 BM stroma cells in low serum conditions with the aforementioned agents. ATRA induced a 40.5±30.1 fold upregulation of CYP26B1 at 24h (p=0.02 compared to control) and 24.7±20.4 fold upregulation at 48 hours (p=0.03). CD437 showed progressive upregulation of CYP26B1 (14.4±5.2 fold at 24h and 27.7±8.8 fold at 48h, p<0.01 for both). In contrast, the RARα-specific IRX5183 had only modest effects on CYP26B1 levels (2.9±0.9 fold at 24h and 2.7±0.2 fold at 48h, p<0.01 for both). AM80 showed the highest up-regulation of CYP26B1 levels both at 24h and 48h of treatment (29.1±7.9 and 77±24.4, p<0.01 for both). Since RARγ activation appeared to be dispensable for differentiation, we tested if the two RARα active agents, AM80 and IRX5183, could bypass stromal-mediated protection against ATRA-induced differentiation, as both of these compounds have also been reported to be resistant to CYP26 mediated degradation. Several AML cell lines were cultured in the presence or absence of OP9 BM stroma and treated with ATRA, AM80 or IRX5183 at concentrations of 0.01 to 0.1 μM. We have previously shown that ATRA can induce differentiation of most AML cell lines in stroma-free conditions, but it is inactive in the presence of BM stroma; further, its effect can be restored by inhibition of stromal CYP26 (Su M et al. 2015). Whereas stroma blocked upregulation of CD11b and inhibition of clonogenicity of NB4 cells by ATRA, both AM80 and IRX5183 showed similar activity in the presence or absence of stroma (e.g. AM80 resulted in a fold change CD11b from control of 9.8±5.3 off stroma vs 11.8±8.4 on stroma, p=NS). Stroma also blocked ATRA's activity against NPM1-mutated OCI-AML3 cells (clonogenic recovery from control 6.4±3.3% vs 61.4±12.2% on stroma, p=0.002), but AM80 and IRX5183 were active both in the presence and absence of stroma (e.g. IRX5183 resulted in clonogenic recovery from control of 7.3±2.5% vs 11.2±2.8% on stroma, p=NS). Similar effects were observed using Kasumi-1 core binding factor AML cells.

In conclusion, we found that RARα and γ have distinct effects on niche vs AML cells. Stimulation of RARγ does not induce differentiation of AML cells, but leads to the upregulation of stromal CYP26B1, and thus enhanced resistance to ATRA. In contrast, CYP26 resistant, RARα active synthetic retinoids AM80 and IRX5183 are able to differentiate and eliminate AML cells even in the presence of BM stroma regardless of induction of CYP26 expression. We are currently exploring a phase I/II clinical trial using IRX5183 in non-APL AML.