AMPK activation induces immunogenic cell death in AML

Abstract

Survival of patients with acute myeloid leukemia (AML) can be improved by allogeneic hematopoietic stem cell transplantation (allo-HSCT) because of the antileukemic activity of T and natural killer cells from the donor. However, the use of allo-HSCT is limited by donor availability, recipient age, and potential severe side effects. Similarly, the efficacy of immunotherapies directing autologous T cells against tumor cells, including T-cell recruiting antibodies, chimeric antigen receptor T-cell therapy, and immune checkpoint inhibitors are limited in AML because of multiple mechanisms of leukemia immune escape. This has prompted a search for novel immunostimulatory approaches. Here, we show that activation of adenosine 5'-monophosphate-activated protein kinase (AMPK), a master regulator of cellular energy balance, by the small molecule GSK621 induces calreticulin (CALR) membrane exposure in murine and human AML cells. When CALR is exposed on the cell surface, it serves as a damage-associated molecular pattern that stimulates immune responses. We found that GSK621-treated murine leukemia cells promote the activation and maturation of bone marrow-derived dendritic cells. Moreover, vaccination with GSK621-treated leukemia cells had a protective effect in syngeneic immunocompetent recipients bearing transplanted AMLs. This effect was lost in recipients depleted of CD4/CD8 T cells. Together, these results demonstrate that AMPK activation by GSK621 elicits traits of immunogenic cell death and promotes a robust immune response against leukemia. Pharmacologic AMPK activation thus represents a new potential target for improving the activity of immunotherapy in AML.

Graphical abstract

Figure 1.

GSK621 induces preapoptotic surface CALR exposure in murine and human cells. (A) C1498 cells were treated at the indicated concentrations with GSK621 for 48 hours, and the percentage apoptotic cells was measured by flow cytometry using annexin V/propidium iodide (PI) staining (early apoptotic cells: annexin V–positive/PI⁻; late apoptotic: annexin V–positive/PI⁺). Ordinary two-way analysis of variance (ANOVA) with Dunnett multiple comparisons test. (B) Flow cytometry of C1498 cells used as controls for surface CALR staining. CALR KO or constitutive cell surface CALR were stained with a CALR–AF647 antibody or isotype control. Graphs represent mean fluorescence intensity (MFI) ± standard deviation (SD). (C) Fold change in MFI relative to vehicle of surface CALR staining on live, unpermeabilized cells by flow cytometry (MFI ± standard error of the mean) after 6 hours of treatment with the indicated drugs (GSK621, MTO, or AraC); n = 3 biological replicates. Ordinary two-way ANOVA with Dunnett multiple comparisons test. (D) Schematic of the procedure to generate MLL-AF9 transformed AML cells from leukemia-bearing mice. After multiple replatings in methylcellulose supplemented with IL-6 (10 ng/mL), stem cell factor (SCF, 10 ng/mL), and IL-3 (6 ng/mL), colonies were enriched in granulocyte colony-forming units (CFU-G; picture). Colonies were then transferred in liquid media and grown in suspension. (E) MLL-AF9 cells were treated at various concentrations with GSK621 for 48 hours and percentages of apoptotic cells were measured as described earlier. Ordinary two-way ANOVA with Šidák multiple comparisons test. (F) Representative dot plots of surface CALR staining in unpermeabilized MLL-AF9 cells treated for 24 hours with GSK621 30 μM or dimethyl sulfoxide (DMSO). (G) MFI of surface CALR staining measured on live, unpermeabilized MLL-AF9 cells; n = 3 replicates. Paired t test. (H) Percent of CALR⁺ PI⁻ blasts of 3 samples from patients with AML after treatment with GSK621 (20 μM), MTO (0.5 nM), or DMSO. Human AML samples 1 and 2 were cryopreserved, thawed, and treated with GSK621. Sample 3 was received fresh and treated immediately. Ordinary one-way ANOVA with Dunnett multiple comparisons test. (I) Representative dot plots for sample 3.

Figure 2.

Calreticulin surface exposure is AMPK dependent in AML cells. (A) Western blot showing knockdown efficiency of AMPKα in MOLM14 and OCI-AML2 cells transduced with a nontargeting control sgRNA guide (NTG) or PRKAA1 sgRNA. (B) Representative dot plots of MOLM14 and OCI-AML2 surface CALR. Percentages represent the fraction of CALR⁺ among 4′,6-diamidino-2-phenylindole (DAPI)-negative cells. (C) Control (NTG) or AMPK KO MOLM14 and OCI-AML2 cells were treated with GSK621 (30 μM) or vehicle for 24 hours, and surface exposure of CALR was measured by flow cytometry. Results are MFI of CALR on live (DAPI negative) cells. Baseline fluorescence intensity of the isotype control stained or unstained DMSO treated cells are shown on the same graphs. Ordinary one-way ANOVA with Šidák multiple comparisons test. (D) MOLM14 AMPK KO or control cells (MOLM14 NTG) were treated for 24 hours with DMSO, GSK621 (30 μM), idarubicin (IDA; 25 nM), venetoclax (VEN; 500 nM), or AraC (2 μM), and surface CALR was measured by flow cytometry. The MFI of 3 independent replicates is shown. Ordinary two-way ANOVA with Dunnett multiple comparisons test for comparison to DMSO within each cell line group (NTG and AMPK KO), and ordinary two-way ANOVA with Tukey multiple comparisons test to compare GSK621 treatment in the NTG vs AMPK KO cell line. (E) Western blot showing the expression of a V5-tagged truncated form of AMPKα1 lacking its regulatory region (AMPK 1-312; selected in blasticidin; predicted size: 37.2 kDa). (F) MFI of surface CALR on AMPK KO MOLM14 cells with or without AMPK (1-312) rescue after 24 hours treatment with DMSO, GSK621, VEN, or AraC. Ordinary two-way ANOVA with Šidák multiple comparisons test. (G) MOLM14 cells lacking endogenous PERK expression were generated by CRISPR/Cas9 editing of EIF2AK3 at an intron-exon junction, compared with NTG cells. Cells were rescued with doxycycline-inducible expression of wild-type (WT) PERK bearing a Myc 9E10 tag. Western blot showing the indicated proteins in PERK knockdown and rescue cells, with and without treatment with GSK621 (30 μM). (H) Surface exposure of CALR was measured by flow cytometry before and after induction of PERK rescue, with or without GSK621 treatment. Ordinary two-way ANOVA with Šidák multiple comparisons test.

Figure 3.

AML cells pretreated with GSK621 induce murine BMDC maturation in vitro. (A) Experimental schema for coculture experiments. BMDCs were generated over 7 days from fresh bone marrow cells harvested from female C57BL/6 mice. BMDC maturation was measured by flow cytometry after overnight coculture with tumor cells pretreated for 24 hours before mixing. (B) Representative dot plots of the percent activated DCs after overnight coculture with cells pretreated with DMSO, AraC (10 μM), GSK621 (30 μM), or MTO (100 nM), with a DC–to–tumor cell ratio of 1:2. Controls are BMDCs treated overnight with LPS (0.1 μg/mL) or vehicle (PBS). (C) Graphs representing the level of maturation of BMDCs as measured by percentage of MHCII⁺CD80⁺ or MHCII⁺CD86⁺ cells after overnight coculture with C1498 cells pretreated as indicated, with 2 ratios of BMDC to tumor cells; n = 3 replicates (mean ± SD). BMDCs alone were treated with LPS as a positive control to induce activation markers. Ordinary one-way ANOVA with Dunnett multiple comparisons test for multiple comparison to DMSO within each ratio of cocultures, and a t test to compare BMDC cultured without tumor cells, with or without LPS. (D) Coculture experiments using murine AML cells driven by MLL-AF9. Cells were pretreated for 24 hours with GSK621 or vehicle, then mixed overnight with BMDCs in a 1:8 ratio (1 BMDC for 8 AML cells). Percent of live coculture-matured DCs as defined by DAPI-negative, CD11⁺ cells expressing MHCII and CD80, or MHCII and CD86, measured by flow cytometry. Ordinary one-way ANOVA with Šidák multiple comparisons test.

Figure 4.

AML cells pretreated with GSK621 induce a vaccination effect in vivo. (A) Schema of the vaccination assay performed with C1498 cells or MLL-AF9 cells in immunocompetent C57BL/6 mice. (B) Representative interferon gamma enzyme-linked immunospot (ELISpot) assay using single-cell suspensions from the spleens of 1 animal of each arm of the experiment activated by culture with the parental AML cells. Splenocytes were cocultured with a syngeneic C57BL/6 melanoma cell line (B16F10) or the ovalbumin antigen alone (OVA) as negative controls, or anti-CD3 as a nonspecific positive control. Each column represents ELISpots from splenocytes of 1 spleen seeded in triplicates. (C) Kaplan-Meier curves showing the overall survival in each group (5 mice per group) vaccinated with MLL-AF9 cells pretreated as indicated and fixed before injection, then challenged with live MLL-AF9 cells. Log-rank test. (D) Tumor growth on the challenge site of individual animals in panel C. Of note, none of the mice vaccinated with GSK621-treated cells developed tumors (red line on x-axis). (E) Kaplan-Meier curves showing the overall survival in each group (5 mice per group) vaccinated with MLL-AF9 cells pretreated with GSK621 before injection, then challenged with live MLL-AF9 cells. One group was depleted of CD4/CD8 T cells with intraperitoneal injections of neutralizing antibodies starting 2 days before the challenge (dashed line). The mice of the other group received a control IgG (solid line). Log-rank test. (F) Tumor growth on the challenge site of individual animals in panel E.