TIM-3 signaling hijacks the canonical Wnt/β-catenin pathway to maintain cancer stemness in acute myeloid leukemia

Abstract

The activation of β-catenin plays critical roles in normal stem cell function, and, when aberrantly activated, the maintenance and enhancement of cancer stemness in many solid cancers. Aberrant β-catenin activation is also observed in acute myeloid leukemia (AML), and crucially contributes to self-renewal and propagation of leukemic stem cells (LSCs) regardless of mutations in contrast with such solid tumors. In this study, we showed that the AML-specific autocrine loop comprised of T-cell immunoglobulin mucin-3 (TIM-3) and its ligand, galectin-9 (Gal-9), drives the canonical Wnt pathway to stimulate self-renewal and propagation of LSCs, independent of Wnt ligands. Gal-9 ligation activates the cytoplasmic Src homology 2 domain of TIM-3 to recruit hematopoietic cell kinase (HCK), a Src family kinase highly expressed in LSCs but not in HSCs, and HCK phosphorylates p120-catenin to promote formation of the LDL receptor-related protein 6 (LRP6) signalosome, hijacking the canonical Wnt pathway. This TIM-3/HCK/p120-catenin axis is principally active in immature LSCs compared with TIM-3-expressed differentiated AML blasts and exhausted T cells. These data suggest that human AML LSCs constitutively activates β-catenin via autocrine TIM-3/HCK/p120-catenin signaling, and that molecules related to this signaling axis should be critical targets for selective eradication of LSCs without impairing normal HSCs.

Graphical abstract

Figure 1.

Identification of the canonical Wnt pathway as major downstream signaling of TIM-3/Gal-9 autocrine loop. (A) Fluorescence-activated cell sorting (FACS) analysis of surface TIM-3 expression (left) and intracellular Gal-9 expression (right) in KASUMI-3 cells (B) Percentage of BM chimerism in NSG mice with xenografts with TIM-3 KD KASUMI-3 cells (n = 6) and scrambled control cells (n = 6) (left panel), and human Gal-9 concentration in serum of mice with xenografts (right panel). ∗∗P < .01 vs scrambled control. These data were obtained at 13 weeks after xenotransplantation. (C-D) Enrichment plots for gene sets significantly enriched in scrambled control cells compared with KASUMI-3 cells transfected with shHAVCR2-1 and shHAVCR2-2. (C) HSC- and LSC-related genes. (D) Canonical Wnt pathway–related genes (BIOCARTA). (E) Quantification of nucleus translocation of β-catenin evaluated by ArrayScan system. Left panels show the representative images for localization of β-catenin in scrambled control and TIM-3 KD KASUMI-3 cells. Scale bar represents 10 μm. Right panel shows the quantification of β-catenin (green) translocation to the nucleus calculated from the fluorescence intensity and area overlapped with nucleus (blue), which were analyzed in TIM-3 KD KASUMI-3 cells and scrambled control cells. Data are presented as mean ± SEM, ∗∗P < .01 vs scrambled control.

Figure 2.

Gal-9 ligation to TIM-3 induces LRP6 signalosome formation and subsequent β-catenin accumulation in AML. (A) WB analysis of β-catenin accumulation induced by stimulation with Gal-9 in the presence of inhibitors in KASUMI-3 cells. KASUMI-3 cells were stimulated with Gal-9 in the presence of dimethyl sulfoxide, 10 μM of U0126 (MEK1/2 inhibitor), 10 μM of LY294002 (PI3K inhibitor), or 200 ng/mL of DKK-1, and their total cell lysates subjected to WB analysis. (B) Summarized data of Gal-9 stimulation–induced changes in the accumulation of β-catenin from 3 independent experiments. (C) Extent of β-catenin translocation to the nucleus evaluated using the ArrayScan system compared with that observed in nonstimulated controls. Data in panels B and C are presented as mean ± SEM, ∗P < .05. (D) Immunoblotting analysis of total lysates (left) and IP lysates (right) of Gal-9–stimulated primary TIM-3⁺ AML samples. Cells were lysed and subjected to IP with an anti-LRP6 antibody or normal mouse IgG. A representative result of TIM-3⁺ primary AML cells out of 3 independent cases (AML#4, #9, and #13) is shown. (E) WB analysis of Gal-9 stimulation–induced phosphorylation of LRP6 at Thr1479 in mock-transfected and TIM-3-OX THP-1 cells. (F) WB analysis of cell lysates from TIM-3⁺ AML cells (left) and TIM-3⁻ AML cells (right). TIM-3⁺ AML cells were stimulated with Gal-9 in the absence or presence of DKK-1 (200 ng/mL). Representative results out of 4 independent TIM-3⁺ AML cells (AML#1, #2, #4, and #16) and 2 TIM-3⁻ AML cells (AML#10 and #11) are shown. n.s., no significant difference.

Figure 3.

HCK is a critical signal transducing molecule involved in TIM-3–induced canonical Wnt pathway activation. (A) WB analysis of Gal-9–stimulated KASUMI-6 cells in the presence of SFK inhibitors: AMGT, PP1, PP2, and A-419259 (potent HCK inhibitor). Inhibitable members of SFKs at each concentration of inhibitors are listed. Because KASUMI-6 cells showed obvious response to SFKs inhibitors, we presented the representative data of KASUMI-6 cells. (B) WB analysis of cell lysates from Gal-9–stimulated TIM-3⁺ primary samples in the presence of 0, 1, and 10 nM of A-419259. Cells were preincubated with A-419259 for 2 hours before stimulation with Gal-9. Representative results out of 4 independent TIM-3⁺ AML cases (AML#6, #9, #14, and #17) are shown. (C) Images of KASUMI-3 cells captured from representative fields obtained by ArrayScan analysis. These cells were stimulated with or without Gal-9 in the presence or absence of A-419259 for 20 hours. Scale bar represents 10 μm. (D) Extent of β-catenin translocation to the nucleus evaluated using the ArrayScan system compared with nonstimulated control. Data are presented as mean ± SEM, ∗P < .05 and ∗∗P < .01. (E) Immunoblotting analysis with HCK, pSFK, and TIM-3 antibodies of total cell lysates (left) and immunoprecipitated lysates (right). Cells were stimulated with Gal-9 for 0, 5, and 10 minutes and subsequently lysed and subjected to IP with an HCK antibody or normal rabbit IgG. Representative results (AML#13) out of 3 independent experiments are shown here.

Figure 4.

p120-catenin plays a crucial role for bridging TIM-3/Gal-9 signaling to the canonical Wnt pathway in AML cells. (A) Cellular growth of KASUMI-3 cells transfected with scrambled control and shRNA-mediated CTNND1 KD vectors. Two kinds of shRNA (shCTNND1-1: target sites in the 3' untranslated region, and shCTNND1-2: target sites in the coding sequence) were used. Data are presented as mean ± SEM, ∗P < .05 and ∗∗P < .01 vs scrambled control. (B) WB analysis of the canonical Wnt pathway–related protein using cell lysates from KASUMI-3 cells transfected with scrambled control and 2 kinds of shRNA-targeting CTNND1. (C) WB analysis of the phosphorylation status of p120-catenin at Tyr228 induced by stimulation with Gal-9 (AML#7). (D) Quantification of the levels of phosphorylated p120-catenin at Tyr228 by Gal-9 stimulation in KASUMI-3 cells. Results of 4 independent experiments (AML#1, #7, #9, and #16) are summarized. Data are presented as mean ± SEM, ∗P < .05 vs nonstimulated control. (E) WB analysis of the inhibitory effect of A-419259 (0, 1, or 10 nM) on Gal-9 stimulation–induced phosphorylation of p120-catenin using a primary AML sample (AML#9). (F) Quantification of the levels of phosphorylated p120-catenin induced by Gal-9 stimulation in the presence of each concentration of A-419259 (0, 1, and 10 nM). Data are presented as mean ± SEM, ∗P < .05 and ∗∗P < .01. Results from 4 independent AML experiments (AML#6, #9, #14, and #17) are summarized. (G) Immunoblotting analysis with total p120-catenin and phosphorylated p120-catenin (Tyr228) antibodies of total cell lysates (left), and immunoprecipitated lysates (right). Cells were stimulated with Gal-9 for 0, 5, and 10 minutes, and subsequently lysed and subjected to IP with an anti-HCK antibody or normal rabbit IgG. Representative results out of 3 independent experiments are shown here.

Figure 5.

TIM-3 signaling induces β-catenin accumulation in accordance with p120-catenin expression level in AML. (A) FACS analysis of the expression of TIM-3 on CD34⁺CD38^– (middle; LSCs-enriched fraction) and CD34⁺CD38⁺ (right; main blasts–enriched fraction) AML cells from the CD45^dimSSC^low fraction. Representative FACS plots of 4 independent AML samples (AML#7, #8, #13, and #14) are shown. (B) RT-qPCR analysis of the expression of CTNND1 mRNA in LSCs compared with blasts from AML samples of 3 independent AML cases. Results were normalized to the mRNA expression of GAPDH. Data are presented as mean ± SEM, ∗∗P < .01 and ∗P < .05. (C) WB analysis of cell lysate from Gal-9–stimulated CD34⁺CD38^– AML cells (LSCs) and CD34⁺CD38⁺ cells (blasts) (AML#14). Of note, LSCs with higher p120-catenin expression exhibited profound β-catenin accumulation.

Figure 6.

The expression of p120-catenin results in a striking difference in the signal transduction of TIM-3 between AML and exhausted T cells. (A) WB analysis of Gal-9–stimulated TIM-3-OX THP-1 and PD-1⁺TIM-3⁺ exhausted T cells. Of note, Gal-9 ligation to TIM-3 induced LRP6 phosphorylation in TIM-3-OX THP-1 cells but not in the exhausted T cells devoid of p120-catenin expression. (B) RT-qPCR analysis of CTNND1 mRNA in FACS-purified AML cells and the exhausted (PD-1⁺TIM-3⁺) T cells from 2 patients with AML (AML#12 and #15). Results were normalized to the mRNA expression of GAPDH. Data are presented as mean ± SEM, ∗P < .05. (C) Intracellular FACS analysis of p120-catenin expression in TIM-3⁺ AML cells and PD-1⁺TIM-3⁺ exhausted T cells in the identical patient with AML. A representative result out of 3 AML samples (AML#8, #12, and #15) is shown. (D) FACS analysis of surface TIM-3 expression in Ki-JK cells. TIM-3 expression of KASUMI-3 and KASUMI-6 are shown as positive controls. (E) The results of WB analysis using mock-transfected (left panels) and p120-catenin-OX Ki-JK cells (right panels) stimulated with Gal-9 are shown.

Figure 7.

Schematic summary of a novel molecular mechanisms of the hijacking of the canonical Wnt pathway by TIM-3 signaling in AML LSCs. The schema shows how TIM-3 signaling induces constitutively canonical Wnt pathway activation and aberrant accumulation of β-catenin in AML-LSCs. As shown in the left panel, β-catenin is constantly destroyed by the β-catenin degradation complex in the absence of canonical Wnt pathway activation. As shown in the right panel, in AML, a TIM-3/Gal-9 autocrine loop constitutively recruits and activates HCK, leading to the induction of p120-catenin phosphorylation. The activated p120-catenin initiates LRP6 signalosome formation to inhibit the function of the β-catenin degradation complex, leading to the aberrant accumulation of β-catenin in AML cells. Through the use of LSCs-specific molecules such as TIM-3, HCK, and p120-catenin, the AML LSCs hijack the canonical Wnt pathway.