Identification of Small Molecules Selectively Targeting Leukemic Stem Cells within Their Stromal Niche

Abstract 413

Current therapies for acute myeloid leukemia (AML) are highly toxic, yet the relapse rate remains high. New therapies are needed to improve cure rates while decreasing toxicity. Because therapies may be affected by the tumor niche, we aimed to test new compounds on leukemic stem cells (LSCs) within their stromal microenvironment. A niche-based high throughput screen identified candidate small molecules potentially toxic to MLL-AF9 murine leukemic stem cells (LSCs) while sparing normal hematopoietic stem cells (HSCs) and bone marrow stroma (Hartwell et al, Blood 118, Abs 760, 2011.) Three such compounds, including a selective serotonin receptor antagonist highly specific for the 5-HT1B receptor, SB-216641, and two antihelminthics, parbendazole and methiazole, were found to be effective and selected for studies on human leukemias.

We first examined SB-216641, studying the effects of this compound on 7 human primary AML samples. We began by assessing the compound's effect on LSCs using the week 5 cobblestone area forming cell (CAFC) assay, a standard in vitro stem cell assay. CD34+ cells were isolated with immunomagnetic beads. The leukemic cells were pulse treated for 18 hours and washed prior to placement on MS-5 murine stroma. We performed serial drug dilutions using the CAFC assay with the human primary samples as well as with HSCs derived from cord blood. All human leukemic samples formed cobblestone areas in the control setting (46-200 CAFCs/10⁶ cells plated). IC50 for the human primary leukemia CAFCs was 630 nm, and at 10 μM all LSCs were killed while normal human HSCs had 100% survival. A combination of the AML cell line HL60 transduced with GFP-luciferase and normal cord blood CD34+ cells (1:200) were then pre-incubated overnight with SB-216641 at 5 and 10 μM and injected into Nod Scid IL2R-gamma null (NSG) mice. The control mice had leukemic engraftment by luciferase imaging and flow cytometry and the mice that received treated cells had no leukemic engraftment but normal multilineage engraftment of cord blood. Primary patient AML samples were also pre-incubated overnight with SB-216641 at 10 μM and injected into NSG mice. As shown by flow cytometry, control mice engrafted with leukemia and mice that received pre-treated cells had no engraftment following exposure to SB-216641. Finally, an in vivo study was completed on NSG mice injected intraperitoneally with 20 mg/kg/day beginning on day 1 or day 8 after inoculation with HL60 (500 cells). The mice were imaged at 2 and 3 week time points and both treatment groups had significantly less leukemia on imaging than the control group with minimal toxicity noted. Another specific 5-HT1B receptor antagonist, SB-224289, was found to have similar activity to SB-216641 against leukemic cells and to spare HSCs in preliminary studies.

Similar CAFC studies with serial dilutions on primary AML samples were performed on the two anti-helminthic agents. IC50 for parbendazole was 1.25 μM and for methiazole 5 μM. As shown by luciferase imaging and flow cytometry, when injected with combined HL60 and cord blood pre-incubated overnight at 5 and 10 μM with each compound as described above, the control mice engrafted with leukemia and the mice that received treated cells had no leukemic engraftment but normal multilineage engraftment of cord blood. NSG mice were then injected with primary AML pretreated overnight with parbendazole at 10 μM. As shown by flow cytometry, control mice engrafted with leukemia and mice that received pre-treated cells had significantly lower engraftment following exposure to parbendazole (p = 0.01).

Two new avenues of leukemia therapy were discovered warranting further investigation. SB-216641, an agent with a completely novel receptor target in leukemia therapy, has shown both in vitro success in human leukemia as well as preliminary success in vivo with minimal toxicity. We aim to move forward with this agent while also testing parbendazole in vivo, as this compound is already known to have good pharmacokinetics and minimal toxicity in animals. The high toxicity to LSCs and sparing of normal HSCs give both these agents an attractive profile for future clinical trials.