TIM-3 and Its Ligand, Galectin-9 Constitute a Pan-Myeloid Autocrine Loop to Develop and Maintain Malignant Stem Cells in Most Human Myeloid Leukemia

Acute myeloid leukemia (AML) originates from self-renewing leukemic stem cells (LSCs), an ultimate therapeutic target for AML. We have reported that the T-cell immunoglobulin mucin-3 (TIM-3) is expressed on the surface of LSCs in most types of AML, and that TIM-3-targeted therapy could eradicate AML LSCs, based on in vivo xenograft experiments using patients’ AML samples (Kikushige et al, Cell Stem Cell, 2010). Since only the TIM-3+ but not the TIM-3- fraction of human AML cells can reconstitute human AML in immunodeficient mice, we hypothesized that the TIM-3 plays indispensable function to maintain AML LSCs. Galectin-9 is a ligand for TIM-3, and ligation of TIM-3 by galectin-9 has been shown to phosphorylate tyrosine residues of TIM-3 and activate Src family kinases through its SH2 domain in T cells and monocytes.

We found that serum galectin-9 concentration was significantly elevated in AML patients (431.1+58.6pg/ml, n=13) but not in the healthy individuals (25.2+6.8pg/ml, n=7) or patients with B cell malignancies (36.1+15.5pg/ml, n=10). Primitive CD34+ AML cells had abundant galectin-9 protein in their cytoplasms. We then transplanted human CD34⁺ primitive AML cells into irradiated immunodeficient mice. Strikingly, only in mice reconstituted with human AML, but not in those with normal cord blood or human ALL cells had elevated serum levels of human galectin-9: Serum galectin-9 levels were 234.7+69.0pg/ml (n=8) in mice reconstituted with primary human AML cells, whereas 4.64+4.64pg/ml (n=12) in mice with normal human hematopoiesis. These results collectively suggested that AML cells secreted galectin-9 in vivo in an autocrine manner. We then performed transcriptome analyses of primary CD34⁺TIM-3⁺ AML cells after galectin-9 ligation. Results were further evaluated by a pathway enrichment analysis, which demonstrated that both NF-κB and β-catenin pathways were activated. In fact, galectin-9 ligation of TIM-3⁺ human AML cells induced activation of the NF-κB pathway via ERK and AKT phosphorylation. Activation of ERK and AKT pathways are known to inhibit the GSK3β activity, and to promote the nucleus translocation of β-catenin in several cancers. To demonstrate significant nucleus translocation of β-catenin of primary AML cells, we established a quantitation system by utilizing the Array Scan VTI system. By this system, we could formally prove that TIM-3 ligation by galectin-9 significantly promoted the nucleus translocation of β-catenin in primary AML cells.

We also have extensively analyzed TIM-3 and galectin-9 interaction in cells from patients with myeloid malignancies including myelodysplastic syndromes, myeloproliferative neoplasms and chronic myelogeneous leukemia. Strikingly, in all cases, frequencies of CD34⁺CD38^-TIM-3⁺ cells dramatically increased along with disease progression from early/chronic phase to overt leukemias. Furthermore, serum levels of galectin-9 were also dramatically elevated after leukemic transformation. Significant nucleus translocation of β-catenin by galectin-9 ligation was also found in these diseases after leukemic transformation. A recent study has shown that NF-κB and β-catenin pathways could play a cooperative role in conferring the cancer-stem-cell properties to non stem cells in intestinal cancer model. These data collectively suggest that TIM-3 and galectin-9 constitutes a pan-myeloid autocrine loop to develop malignant stem cells in the vast majority of human myeloid malignancies. Thus, signaling molecules downstream of TIM-3 and galectin-9 ligation, as well as surface TIM-3 itself might be good candidates for cancer stem cell-target therapy common to most myeloid malignancies.