Dissecting Signaling Network Responses to PAK4 Allosteric Modulators (PAMs) in Cell Subsets within Primary Human Acute Myeloid Leukemia Samples

Background: Acute myeloid leukemia (AML) is the most common acute leukemia in adults and has a poor outcome with limited treatment options in patients with relapsed or resistant disease. Therapy resistance in AML is likely related to the inadequacy of therapy within leukemia cell subsets, including leukemia stem cells (LSCs). The p21-activated kinase (PAK) family of proteins was shown to be overexpressed in cancer cells and to play a key role in proliferation, survival, and maintenance of cellular structure. The series of orally bioavailable PAK4 allosteric modulators (PAM) have previously been shown to have activity in hematological cancer cell lines, including those derived from acute myeloid leukemia (AML) (Senapedis et al. Blood124, 2208-2208). Understanding how therapies target cellular subsets within primary patient samples could aid drug development by revealing any subset specific drug effects. In this project, we studied the effects of p21-activated kinase 4 (PAK4) modulation in AML samples. PAK4 modulation has been shown to have significant effects on many intracellular signaling pathways, including PI3K/AKT, MAPK/ERK and WNT/β-catenin pathways (Senapedis et al. Blood124, 2208-2208). It is unknown whether PAMs will have similar activity in primary leukemia cells. Likewise, it is currently unclear to what extent PAMs will differentially impact primary cell subsets including leukemia stem cells and non-malignant cell subsets that may be critical to recovery of bone marrow functions. We have previously shown that the single cell biology platform of flow cytometry is well-suited for dissecting clinically relevant signaling network mechanisms in primary human AML (Irish et al. Cell, 118(2):217-28).

Methods: Flow cytometry was used to dissect the impact of an orally bioavailable PAM in AML cell lines and primary patient tissue. Cell lines chosen for this study included NRAS mutant KG-1 and Kasumi-1, which carry t(8;21) and express the AML1:ETO fusion protein. Primary AML biopsies were acquired from bone marrow or blood prior to any treatment and patients were identified and consented for this study according to a local Institutional Review Board-approved protocol. AML tissue samples were viably cryopreserved and then assayed ex vivo. Established protocols were used for phospho-specific flow cytometry, fluorescent cell barcoding, and data analysis in Cytobank (Irish et al. Cell, 118(2):217-28, Doxie and Irish, Curr Top Microbiol Immunol. 377:1-21).

Results: Differential effects of PAK4 inhibition were observed between cell lines and among cell subsets from AML patient bone marrow. In leukemia cell lines and patient samples, p-ERK and p-S6 showed marked inhibition via PAM, though degree of inhibition varied. In AML patient samples, PAMs blocked signaling responses in p-ERK specifically in AML blasts, but spared normal CD45^hi mononuclear cells (0.88 vs. 0.29-fold reduction (arcsinh scale) in p-ERK at 10 nM). Within the AML blast population, CD34⁺ CD38^- and CD34⁺ CD38⁺ AML subsets showed similar PAM dose response via p-ERK.

Conclusions: Single cell analysis effectively distinguishes effects of PAK4 inhibition via a series of allosteric modulators of PAK4 (PAMs) on leukemia and non-leukemia subsets in the same sample. PAM reduced immediate p-ERK and p-S6 levels in primary leukemia and cell lines. Notably, inhibition in various subsets within human AML was successfully measured by phospho-flow cytometry. Signaling changes in p-ERK were minimal within non-leukemic mature CD45⁺ mononuclear cells found in primary patient biopsies. Analysis of CD34⁺ CD38^- cells indicates that PAMs could have activity within leukemia stem cells, and, at least, effect the AML progenitors. These findings support further investigation into the mechanism of action and treatment potential of PAMs in AML.