GATA Factor-Dependent Positive-Feedback Circuit Controls Acute Myeloid Leukemia Cell Proliferation: Mechanisms and Targets for Therapeutic Intervention

Recent studies have established a relationship between GATA-2, a master regulator of hematopoiesis, and acute myeloid leukemia (AML). Loss-of-function mutations and elevated GATA2 expression are implicated in AML. While inhibitory GATA2 mutations occur in patients with primary immunodeficiency, which progresses to AML, GATA2 can be overexpressed in AML patients with poor prognosis. Many questions remain unanswered regarding mechanisms by which GATA2 deregulation results in AML. We reported that the Ras-p38α pathway induces multi-site GATA-2 phosphorylation in proerythroblasts, and Ser192 is required for the phosphorylation. Because RAS is mutated in 20% of AML patients, and elevated GATA2 expression correlates with poor prognosis, we tested whether this pathway functions in AML. GATA-2 was phosphorylated in the steady-state in AML cell lines, and oncogenic Ras(G12V) expression induced GATA-2 hyperphosphorylation in a p38/ERK-dependent manner. GATA-2 contains a sequence that conforms to the MAPK docking site termed "DEF motif". Mutation of S192 or the DEF motif abrogated Ras(G12V)-mediated GATA-2 hyperphosphorylation. We analyzed function using a Mouse Aortic Endothelial (MAE) cell line, in which GATA-2 and Ras(G12V) synergistically induce expression of specific GATA-2 target genes, e.g. Hdc. The DEF mutation abrogated GATA-2 hyperphosphorylation by Ras(G12V) and attenuated GATA-2 activity to induce Hdc expression in the presence of Ras(G12V) in MAE cells (50% of control, p < 0.05). GATA-2 hyperphosphorylation was also induced by expression of constitutively-active (ca) p38α and caMEK-1. caMEK-1-induced hyperphosphorylation was suppressed by the p38 inhibitor, SB203580 (p < 0.05) while the MEK inhibitor U0126 did not influence cap38α-induced hyperphosphorylation. The phosphatase inhibitor okadaic acid induced GATA-2 hyperphosphorylation, and p38 was more important than ERK in this context. These data support a model in which a hierarchical network of kinases controls GATA-2 phosphorylation and function in AML. Phospho-proteomics is being used to establish interconnectivity between network components and temporal aspects of the multi-step signaling mechanism.

ChIP-Seq analysis in Kasumi-3 AML cells revealed GATA-2 occupancy at IL1B and CXCL2. IL1B and CXCR2, encoding the CXCL2 receptor, are implicated in AML. GATA-2 downregulation decreased expression of these genes (IL1B: 81.5% decrease, p < 0.001, CXCL2: 73% decrease, p < 0.001). Inhibition of ERK or p38 attenuated GATA-2 phosphorylation, decreased GATA-2 occupancy at GATA2, IL1B, and CXCL2, and lowered mRNA expression of these genes. To determine the consequences of the reduced occupancy, FAIRE was used to quantitate accessibility at the occupancy sites. Under conditions of reduced GATA-2 occupancy, accessibility decreased 50-65% at the GATA2 -77 enhancer and GATA-2 occupancy sites of IL1B and CXCL2 (p < 0.05). IL-1β and CXCL2 stimulated GATA-2 phosphorylation via p38 and ERK activation in Kasumi-1 and primary AML cells and increased GATA-2 chromatin occupancy and target gene expression. These results reveal a Ras-p38/ERK-GATA-2-IL-1β/CXCL2 axis that constitutes a positive-feedback circuit. We tested whether this circuit impacts Kasumi-1 cell proliferation. GATA2 downregulation reduced the proliferative rate (p < 0.001), which was partially rescued with recombinant CXCL2 (p < 0.01). To determine whether the signal-dependent mechanism can be extrapolated to human AML patients, we analyzed AML datasets from TCGA and GEO. GATA2 and CXCL2 expression was elevated in AML (p < 0.01). GATA2, IL1B, and CXCL2 mRNA levels correlated in M5 AML (GATA2 vs. IL1B: p=0.0015, GATA2 vs. CXCL2: p=0.0046, IL1B vs. CXCL2: p=0.000001). Among the patients with normal karyotype AML, high GATA2 and CXCL2 expression correlated with significantly reduced survival (p=0.003).

In aggregate, our results establish a Ras-p38/ERK-GATA-2-IL-1β/CXCL2 axis in AML. It is attractive to propose that the GATA-2-dependent positive-feedback circuit is deployed upon ectopic elevation of GATA-2 activity in certain AML contexts and may harbor potential therapeutic targets. Given the profound capacity of this mechanism to alter GATA-2 activity, it is crucial to consider the GATA-2 activation state, rather than mRNA or total GATA-2 protein levels in clinical scenarios.