Targeting SLFN11-regulated pathways restores chemotherapy sensitivity in AML

Abstract

Chemoresistance represents an ongoing challenge in treating patients with acute myeloid leukemia (AML), and a better understanding of the resistance mechanisms can lead to the development of novel AML therapies. Here, we demonstrated that low expression of the DNA damage response gene Schlafen 11 (SLFN11) correlates with poor overall survival and worse prognosis in patients with AML. Moreover, we showed that SLFN11 plays an essential role in regulating chemotherapy sensitivity in AML. AML cells with suppressed levels of SLFN11 do not undergo apoptosis in response to cytarabine because of aberrant activation of the Ataxia telangiectasia and Rad3-related protein (ATR)/Checkpoint kinase 1 (Chk1) pathway, allowing for DNA damage repair, whereas sensitivity to cytarabine can be restored by inhibiting the ATR pathway. Importantly, SLFN11 knockout AML cells retain sensitivity to hypomethylating agents and the B-cell lymphoma 2 (BCL-2) inhibitor venetoclax. Altogether, these results reveal SLFN11 as an important regulator and predictor of chemotherapy sensitivity in AML and suggest that targeting pathways suppressed by SLFN11 may offer potential combination therapies to enhance and optimize chemotherapy responses in AML.

Graphical abstract

Figure 1.

High SLFN11 expression correlates with increased OS in patients with AML. (A) Kaplan-Meier survival curves for patients with AML expressing high (above the median, n = 80) vs low (below the median, n = 81) SLFN11 expression. Data were extracted from the TCGA AML data set using the UCSC Xena browser. (B) Kaplan-Meier survival curves for patients with AML presenting (left) high (n = 65) vs low (n = 129) methylation levels at TSS 1500 of SLFN11 and (right) high (n = 130) vs low (n = 64) methylation levels at TSS 200 of SLFN11. Raw methylation data used in the MethSurv web tool were extracted from the TCGA AML data set available on the Broad Institute website. (C) Heat map showing the methylation levels of various CpG sites on SLFN11 in patients with AML. Data were extracted from the TCGA AML data set available on the Broad Institute website and graphed using the MethSurv web tool. CpG sites shown in panel B are highlighted in yellow. (A-B) Statistical analyses were performed using the log-rank test, and P values are shown. HR, hazard ratio; UTR, untranslated region.

Figure 2.

High SLFN11 expression is associated with better prognostic groups in AML. (A) Relative SLFN11 messenger RNA expression in normal CD34⁺ cells isolated from healthy donors (n = 12) and in patients with favorable-risk (n = 117) or adverse risk (n = 164) AML based on the ELN 2017 risk criteria. Data were extracted from the Beat AML data set, available through Vizome. (B-D) Relative SLFN11 mRNA expression in patients with (B) newly diagnosed (n = 428) and relapsed (n = 23) AML, (C) de novo (n = 223) and therapy-related AML (tAML) or secondary AML (sAML) (n = 228), and (D) NPM1-mutated (mut) (n = 108) or NPM1 wild-type (WT; n = 340) AML. Data were extracted from the Beat AML data set, and accessed through Vizome. (E) Relative SLFN11 mRNA expression in TP53-WT (n = 117) and TP53-mutated (n = 27) AML. Data were extracted from the OSHU AML cohort using the cBioPortal. (F) Log₂ of SLFN11 mRNA expression in patients with AML with inv(16)/t(16;16) (n = 47), t(11q23)/mixed lineage leukemia (MLL) (n = 42), or a complex karyotype (n = 49). Data were extracted from the TCGA AML data set available through BloodSpot. (G) Relative SLFN11 mRNA expression in normal CD34⁺ cells (n = 32) and patients with (n = 7) or without (n = 444) biallelic CEBPA mutations. Data were extracted from the Beat AML data set available through Vizome. (A,F,G) Statistical analysis was performed using the Kruskal-Wallis test followed by Dunn multiple comparison adjustment. (B-E) Statistical analysis was performed using the 2-tailed Mann-Whitney test. (A-G) The median is represented by the black line for each AML subgroup. ∗P < .05; ∗∗∗P < .001; ∗∗∗∗P < .0001. NS, not significant; RPKM, reads per kilobase million.

Figure 3.

Low SLFN11 expression correlates with lower sensitivity to AraC. (A) Correlation analysis between AraC activity and SLFN11 gene methylation in cancer cell lines from the NCI-60 panel. Data were extracted from CellMiner Cross-Database (CDB) (n = 60). (B) Correlation analysis between AraC activity and Log₂ of SLFN11 mRNA expression in cancer cell lines from the NCI-60 panel. Data were extracted from CellMiner CDB (n = 59). (A-B) Statistical analysis was performed using simple linear regression, and P values are shown for deviation of the line slope from 0. (C) U937 cells were treated with the vehicle control (dimethyl sulfoxide [DMSO]) or 1 μM AZA for 24 hours. Protein lysates from U937 cells were resolved by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE), followed by immunoblotting analysis with the indicated antibodies. GAPDH is shown as a loading control. (D) Clonogenic capability of primary leukemic blasts isolated from patients with AML expressing either high levels of SLFN11 (SLFN11-high, n = 3 for 8 ng/ml AraC and n = 4 for 0.08 and 0.8 ng/ml AraC) or low levels of SLFN11 (SLFN11-low, n = 3) treated with either vehicle control (VC; water) or increasing concentrations of AraC, as indicated (left panel). Data are expressed as the percentage of colony formation over VC-treated cells for each patient (Ctrl) and are shown as mean ± standard error of the mean (SEM). Statistical analyses were performed using Mixed-effects analysis followed by Tukey multiple comparisons test and P values are shown for the highest AraC dose for each group. There was a significant interaction (P = .043), that is, the relationship between concentration and percentage of colony formation was significantly different between patients with AML that were SLFN11-low vs SLFN11-high. (D) Relative SLFN11 mRNA expression, normalized to GAPDH, was assessed by qRT-PCR analysis and is shown for the primary leukemic blasts isolated from patients with AML and used in the clonogenic assays (SLFN11-low, n = 3 and SLFN11-high, n = 4; right panel).

Figure 4.

SLFN11 KO AML cells are resistant to AraC but not to HMAs and VEN. (A-B) Protein lysates from U937-Cas9 and U937-SLFN11 KO cells (pooled and single clones, as indicated) were resolved by SDS-PAGE, followed by immunoblotting with the indicated antibodies. GAPDH is shown as a loading control. (C) U937-Cas9 and U937-SLFN11 KO cells (pooled clones) were seeded in 96-well plates and treated with either VC (water) or increasing concentrations of AraC for 24 hours. (D) U937-Cas9 and U937-SLFN11 KO cells (single clones) were seeded in 96-well plates and treated with either VC (DMSO) or increasing concentrations of AZA for 24 hours. (E) Correlation analysis between AZA activity and Log₂ of SLFN11 mRNA expression in cancer cell lines from the NCI-60 panel. Data extracted from CellMiner CDB (n = 59). Statistical analysis was performed using simple linear regression, and the P value is shown for the deviation of the line slope from 0. (F) U937-Cas9 and U937-SLFN11 KO cells (pooled clones) were seeded in 96-well plates and treated with either VC (DMSO) or increasing concentrations of DAC for 48 hours. (G) U937-Cas9 and U937-SLFN11 KO cells (single clones) were seeded in 96-well plates and treated with either VC (DMSO) or increasing concentrations of VEN for 24 hours. (H) U937-Cas9 and U937-SLFN11 KO cells (single clones) were seeded in 96-well plates and treated with either VC (DMSO) or increasing concentrations of AZA and VEN for 24 hours. (I) U937-Cas9 control cells and U937-SLFN11 KO cells (single clones) were seeded in 96-well plates and treated with either VC (DMSO) or increasing concentrations of AraC and/or AZA for 24 hours. (C,D,F-I) Cell viability was assessed using the WST-1 reagent. Data are expressed as a percentage of cell viability of VC-treated cells. Means ± SEM of 4 independent experiments in panel C or 3 independent experiments in panels D,F-I are shown. Statistical analyses were performed using a 2-way analysis of variance (ANOVA) followed by the Sidak multiple comparison test. In panel C, there is a significant interaction between cell type and concentration (P < .0001) in AraC-treated cells (the effect of AraC concentration on cell viability was significantly different between U937-Cas9 and U937-SLFN11 KO cells). ∗P < .05; ∗∗P < .01; ∗∗∗P < .001; ∗∗∗∗P < .0001. NS, not significant.

Figure 5.

Effects of knocking out SLFN11 in AML cells and in response to AraC treatment in vivo. (A) Schematic illustration of the AML xenograft mouse model and therapeutic regimen. Briefly, mice were injected SQ with U937-Cas9 or U937-SLFN11 KO cells (single clones). On day 8, mice from each tumor genotypic group were randomized by tumor volume and body weight into 2 treatment groups: VC (PBS) or AraC and daily treatments by intraperitoneal (IP) injection were given on days 9 to 18. Tumor volumes were measured 3 times per week until the study end point. Two independent in vivo studies were performed. (B-C) Tumor volumes (mean ± SEM) are shown for (B) PBS-treated or (C) AraC-treated mice implanted with either U937-Cas9 or U937-SLFN11 KO cells until the first tumor for either cohort reached 2000 mm³ (n = 8 for the PBS-treated groups and U937-Cas9 AraC-treated group and n = 9 for the U937-SLFN11 KO AraC-treated group). Mixed effects regression models were used to compare tumor growth between the groups. Tumor volume was the outcome variable, and log(volume + 1) transformation was used to stabilize the variance and satisfy the normality assumption. Day, group, and their interaction were included as fixed effects were fitted, and the within animal correlation between repeated tumor measurements over time was accounted for using a first-order autoregressive covariance structure (AR(1)). P value from the day × group interaction is reported. (D) Kaplan-Meier survival curves for the 4 treatment/genotypic groups (data compiled from the 2 in vivo studies, n = 12 for the PBS-treated and U937-Cas9 AraC-treated groups and n = 13 for the U937-SLFN11 KO AraC-treated group). Survival curves were compared using the log-rank test, and P values were adjusted for 4 pairwise comparisons (denoted in the figure) using the method of Holm-Sidak. ∗P < .05; ∗∗P < .01; ∗∗∗P < .001. NS, not significant; S.Q, subcutaneous.

Figure 6.

SLFN11 KO AML cells present enhanced activation of the ATR/CHK1 pathway and decreased cell death in response to AraC treatment. (A-B) U937-Cas9 and U937-SLFN11 KO cells (single clones) were treated with VC (water) or AraC (500 ng/mL) for 6 hours and stained with propidium iodide (PI), followed by flow cytometry analysis. (A) Representative flow cytometry histograms showing U937-Cas9 and U937-SLFN11 KO cell populations in the sub-G1, G1, S, and G2-M phases of the cell cycle before (untreated [UT]) and after AraC treatment. (B) The bar graph of flow cytometry data shows the cell cycle distribution for each cell type and treatment condition. The mean ± standard deviation of 3 independent experiments are shown. For each phase of the cell cycle (sub-G1, G1, S, and G2/M), statistical analysis was performed using 2-way ANOVA followed by Sidak multiple comparisons test (∗∗∗P < .001; ∗∗∗∗P < .0001). (C) U937-Cas9 and U937-SLFN11 KO cells (pooled clones) were treated with either VC (water) or AraC (500 ng/mL) for the indicated lengths of time (hours). The cells were then lysed, and protein lysates were resolved by SDS-PAGE, followed by immunoblotting analyses using the indicated antibodies. GAPDH is shown as a loading control. Immunoblots are representative of 3 independent experiments. (D) U937-Cas9 and U937-SLFN11 KO cells (pooled clones) were seeded into 96-well plates and treated with either VC (DMSO) or increasing concentrations of the ATRi VE822 for 24 hours. Cell viability was assessed using the WST-1 assay. Data are expressed as a percentage of cell viability of VC-treated cells. Means ± SEM of 3 independent experiments are shown. (E) U937-Cas9 (left panel) and U937-SLFN11 KO cells (right panel; pooled clones) were seeded in 96-well plates and treated with either VC (DMSO), 0.3 μM VE822, or increasing concentrations of AraC alone or in combination with 0.3 μM VE822 for 24 hours. Cell viability was assessed by WST-1 assay. Data are expressed as a percentage of cell viability of VC-treated cells (for AraC alone) or 0.3 μM VE822-treated cells (for AraC + 0.3 μM VE822). Means ± SEM of 3 independent experiments are shown. Statistical analysis was performed using 2-way ANOVA followed by the Sidak multiple comparison test. ∗∗P < .01; ∗∗∗P < .001; ∗∗∗∗P < .0001. (F) U937-SLFN11 KO cells (pooled clones) were treated with VC (DMSO), AraC (500 ng/mL), and/or VE822 (0.3 μM) for 1, 4, or 8 hours, as indicated. Cells were lysed and protein lysates were resolved by SDS-PAGE, followed by immunoblotting analyses using the indicated antibodies. GAPDH is shown as a loading control. Immunoblots are representative of 3 independent experiments. ns, not significant.