Curcumin, genistein, epigallocatechin gallate (EGCG), resveratrol and decitabine were purchased from Target Molecule Corp. Afuresertib (GSK2110183) and SC79 were purchased from Selleck Chemicals. Antibodies against phosphorylated (p)-P70S6 kinase (P70S6K; T421/S424; cat. no. AP0540), p-AKT1(S473; cat. no. AP0140), total AKT1 (cat. no. A11016), poly(ADP-ribose) polymerase (PARP; cat. no. A11010), ACTB (cat. no. AF0198) and caspase 3 (cat. no. A2156) were obtained from ABclonal Biotech Co., Ltd. Antibodies against p-RAF-1 (S301; cat. no. AF0047), p-proline-rich Akt substrate, 40 kDa (PRAS40; T246; cat. no. AF2387), p-p27/Kip1 (T198; cat. no. AF3325), p-eukaryotic translation initiation factor 4E-binding protein 1 (4E-BP1; T36; cat. no. AF3431), β-tubulin (cat. no. AF7011) and Ki67 (cat. no. AF0198) were obtained from Affinity Biosciences. PE-conjugated (clone HI30; cat. no. 560975; BD Bioscience) and unconjugated mouse anti-human CD45 antibody (clone HI30; cat. no. 555480; BD Bioscience) were used for flow cytometry and immunohistochemistry (IHC), respectively. A FITC TUNEL cell apoptosis detection kit was purchased from Wuhan Servicebio Technology Co., Ltd.

AML cell lines (HL-60, ML-2, MOLM-13, OCI-AML3, OCI-AML5 and U937) were obtained from the American Type Culture collection, and were cultured according to the manufacturer's instructions. All cell lines were mycoplasma-free and were authenticated by Yubo Biological Technology Co., Ltd. using short tandem repeat analysis.

Cells were cultured in a 96-well plate until the cell confluence reached ~70%, and then cells were treated with different concentrations (0, 5, 10, 20, 40 and 80 µM) of curcumin, genistein, EGCG, resveratrol or decitabine. After 48 h, cell viability was determined using a MTT assay as described previously (15). Based on the results of the MTT assay, the half maximal inhibitory concentration (IC₅₀) of each chemical was calculated.

As reviewed by Kouhpeikar et al (16), in vitro examination of the efficacy of curcumin against AML cells was conducted using 10–50 µM curcumin to treat cells for 24–48 h. In the present study, cells were treated with 25 µM curcumin for 24 h. After treatment, cell cycle and apoptosis were analyzed using a PI staining kit [Hangzhou Multi Sciences (Lianke) Biotech Co., Ltd.] and an Annexin V-FITC/PI staining kit (Invitrogen; Thermo Fisher Scientific, Inc.), respectively, according to the manufacturer's instructions. After staining, cells were analyzed using a flow cytometer (CytoFlex; Beckman Coulter, Inc.).

A human phosphorylation pathway profiling array (cat. no. AAH-PPP-1-4) was purchased from RayBiotech, Inc., which can detect 55 phosphorylated proteins in five signaling pathways: MAPK, AKT, JAK/STAT, NF-κB and TGF-β. ML-2 cells were cultured in a 10-cm dish until cells reached 90% confluence, and then cells were treated with or without curcumin (25 µM) for 6 h. After treatment, the cells were harvested and lysed using the cell lysis buffer with a protease inhibitor cocktail and a phosphatase inhibitor cocktail. Phosphorylation array analysis was performed according to the manufacturer's protocol. The array was sequentially incubated with the sample and horseradish peroxidase-conjugated antibodies (provided within the kit), and then scanned with ImageQuant LAS4000 Scanner (Cytiva). In total, two biological replicates were performed, and the average expression levels were compared between the treatment and control samples.

Cells were cultured in a 12-well plate until the cell confluence reached ~90%, and then cells were treated with the indicated chemicals. Cell lysates were prepared using the cell lysis buffer with a protease inhibitor cocktail and a phosphatase inhibitor cocktail (Cell Signaling Technology). After quantification of the protein concentration, cell lysates containing equal amounts of total protein were denatured and separated on 10–12% SDS-PAGE. Following separation, the proteins were blotted onto PVDF membranes and blocked. After sequentially incubated with primary antibodies and appropriate secondary antibodies, the membranes were exposed to Pierce ECL Western Blotting Substrate (Thermo Fisher Scientific, Inc.) and were imaged using a gel imaging system (Tanon 4600SF; Tanon Science and Technology Co., Ltd.).

A total of 20 male NOD/SCID mice (age, 5–6 weeks; average weight, 23 mg) were purchased from Hunan Slaccas Jingda Laboratory Animal Co. Ltd., and housed in groups of 5 per cage with water and food ad libitum, in a specific-pathogen-free room with filtered air and controlled light/dark cycle (12/12 h), temperature (24±2°C) and relative humidity (45–65%). All mice were pretreated with an intraperitoneal injection of 20 mg/kg busulfan (APExBIO Technology LLC) 24 h before inoculation, and were then injected intravenously with 1×10⁶ ML2 cells. At 15 days after inoculation, the mice were randomly divided into four groups (5 mice per group), and then treated with vehicle, curcumin (2 mg/mouse), afuresertib (1 mg/mouse) or curcumin (2 mg/mouse) + afuresertib (1 mg/mouse) via oral gavage every other day for 16 days. Curcumin and afuresertib were dissolved with 5% DMSO + 10% PEG300 + 5% Tween-80. The humane endpoints were defined by body weight loss of 20%. All mice were euthanized by asphyxiation (CO₂ displacement rate was ~20% vol/min) 4 days after the last treatment, and the death was verified by respiratory arrest and cardiac arrest for >10 min. The experiments were performed from to July 10 to August 12. The spleens were fixed in 10% formalin and processed for hematoxylin and eosin (H&E) staining, immunohistochemistry (IHC) analysis and TUNEL assay, as described previously (17,18). Bone marrow (obtained from tibias and femurs) was crushed in PBS and created into single cell suspensions for flow cytometry analysis.

RStudio (https://rstudio.com) was used for statistical analysis. ANOVA and Tukey's post hoc test were performed to evaluate the significance of difference between samples, adjust P<0.05 was considered as the level of significance.

The four phytochemicals (curcumin, EGCG, genistein and resveratrol) have been reported to function as epigenetic modulating agents in cancer (19), while the DNA methyltransferase inhibitor decitabine is an FDA-approved chemical for the treatment for AML. Thus, the present study compared the cytotoxicity of these four phytochemicals with decitabine in six AML cell lines (HL-60, ML-2, MOLM-13, OCI-AML3, OCI-AML5 and U937). Cell viabilities at 48 h after exposure to various concentrations of drugs were determined using MTT assays, and IC₅₀ values were calculated. Compared with decitabine, curcumin had a similar or lower IC₅₀ in the AML cell lines (Table I). Moreover, curcumin had the strongest cytotoxic activity against AML cells (except for OCI-AML3) among the four phytochemicals, and ML-2 cells were the most sensitive to curcumin. Therefore, curcumin was selected for further study of its function and mechanism in AML.

It has been reported that curcumin can affect protein phosphorylation (12–14). Thus, the influence of curcumin treatment on protein phosphorylation was examined using a human phosphorylation pathway profiling array, which can detect 55 phosphorylated proteins in five signaling pathways: MAPK, AKT, JAK/STAT, NF-κB and TGF-β. The results demonstrated that the phosphorylation levels of 14 proteins were downregulated (fold change ≤0.83), while those of four proteins were upregulated (fold change ≥1.2) after treatment with curcumin for 6 h (Fig. 1A; Table SI). Functional annotation analysis based on the Kyoto Encyclopedia of Genes and Genomes (KEGG) database identified that the five most enriched pathway were the ‘PD-L1/L1 pathway’, ‘AML’, ‘ErbB signaling pathway’, ‘EGFR tyrosine kinase inhibitor (TKI) resistance’ and ‘B cell receptor signaling pathway’ (Fig. 1B; Table SII).

To understand the association between these 18 proteins, a protein-protein interaction (PPI) network was constructed using STRING software (https://string-db.org). In a PPI network, the nodes (proteins) with multiple edges (interactions) are defined as hubs, and hubs are more essential for the global network structure compared with non-hubs (20). In the present network (Fig. 1C), the AKT node had the highest number of edges, indicating that AKT was the key target of curcumin. Moreover, AKT was involved in the top 20 pathways affected by curcumin (Table SII).

To confirm the result of the phosphorylation array analysis, western blotting was conducted to detect the influence of curcumin (CCM) on the phosphorylation status of AKT and its interacting proteins in ML-2 cells. Curcumin suppressed the phosphorylation of AKT1, PRAS40, 4E-BP1, P70S6K, RAF-1 and p27 in a dose-dependent manner (Fig. 1D).

To determine whether the cytotoxicity of curcumin is dependent on AKT activity, AML cells were treated with curcumin (CCM), an AKT activator (SC-79) or an AKT inhibitor (afuresertib) alone (AFU) or in combination (CCM+SC79/CCM+AFU). Based on the results of the MTT assay (Fig. 2A), SC-79 reversed the antitumor effects of curcumin, while afuresertib augmented its cytotoxicity on both OCI-AML5 and ML-2 cells. Flow cytometry results demonstrated that treatment with curcumin or the AKT inhibitor afuresertib led to cell cycle arrest in the G₁ phase, while treatment with the AKT activator SC-79 promoted cell division (Fig. 2B and C). Moreover, AKT activation by SC-79 rescued the curcumin-induced cell cycle arrest, while AKT inhibition by afuresertib enhanced this effect (Fig. 2B and C). These results suggested that curcumin suppressed AML cell arrest in the G₁ phase by inactivating AKT.

The influence of curcumin, SC-79 and afuresertib on cyclin D1 and p21, which are positive and negative regulators of the cell cycle progression from the G₁ to S phase (21,22), respectively, was then examined. Treatment of the AML cell lines with curcumin or afuresertib suppressed the expression of cyclin D1, but increased the expression of p21 (Fig. 2D and E). However, SC-79 produced the opposite results. Thus, the effects of curcumin were enhanced by afuresertib but attenuated by SC-79.

Annexin V and PI labeling followed by flow cytometry were used to detect apoptotic cells. The results demonstrated that both curcumin and afuresertib promoted apoptosis, while SC-79 suppressed apoptosis. Moreover, curcumin-induced apoptosis was stimulated by afuresertib, but diminished by SC-79 (Fig. 3A and B).

To identify the proteins involved in curcumin-induced apoptosis, the expression of three apoptosis-related proteins, including Bcl-2, caspase-3 and PARP, were examined. The results indicated that curcumin treatment decreased the antiapoptotic Bcl-2 protein expression but increased the cleavage of caspase-3 (C-Casp3) and PARP (C-PARP) (Fig. 3C and D). Furthermore, the influence of curcumin on these three proteins could be enhanced by afuresertib, but was abrogated by SC-79.

The above results suggested that curcumin promoted AML cell arrest and apoptosis by inactivating AKT. However, IC₅₀ values of curcumin were very weakly correlated with the levels of phosphorylated AKT in AML cell lines (Fig. S1). Thus suggested that curcumin also exerted antitumor roles via other pathways, besides the AKT pathway.

Next, the in vivo efficacy of curcumin and afuresertib for the treatment of AML was evaluated. NOD/SCID mice were intravenously injected with 1×10⁶ ML-2 cells. Drug treatment began 15 days after injection and continued every other day for 16 days. After treatment, peripheral blood mononuclear cells (PBMCs) and bone marrow mononuclear cells (BMMCs) were isolated and evaluated for human hematopoietic (hCD45) chimerism via flow cytometry (Fig. 4). Compared with the control group (VEH), the mice treated with curcumin (CCM) or afuresertib (AFU) either alone or in combination (CCM+AFU) had fewer human CD45⁺ cells in the bone marrow and peripheral blood. Moreover, combination drug therapy was more effective than single drug therapy in reducing the chimerism of hCD45 (Fig. 4). These results indicated that curcumin and afuresertib synergistically suppressed the engraftment of AML cells.

The mice treated with curcumin and afuresertib either alone or in combination had a smaller and lighter spleen compared with the control mice (Fig. 5A and B). Thus, it was suggested that treatment with curcumin or afuresertib could decrease splenomegaly in AML mice. IHC using an anti-hCD45 antibody demonstrated that, compared with control mice, the mice treated with curcumin or afuresertib had decreased dissemination of AML cells in the spleen, and the combinational use of curcumin and afuresertib was more effective compared with the use of a single drug (Fig. 5C).

Subsequently, Ki-67 staining and TUNEL assay were conducted to evaluate cell proliferation and apoptosis, respectively. The results (Fig. 5C) demonstrated that treatment with curcumin or afuresertib significantly increased apoptosis but decreased AKT phosphorylation and the cell proliferation rate in spleen, while treatment with both drugs had stronger effects compared with treatment with a single drug. These findings suggested that treatment with curcumin or afuresertib suppressed the engraftment, proliferation and survival of AML cells, and that combination therapy had increased efficacy compared with monotherapy.