Transcriptional remodeling shapes therapeutic vulnerability to necroptosis in acute lymphoblastic leukemia

Abstract

Insufficient eradication of cancer cells and survival of drug tolerant clones are major relapse driving forces. Underlying molecular mechanisms comprise activated prosurvival and antiapoptotic signaling, leading to insufficient apoptosis and drug resistance. The identification of programmed cell death pathways alternative to apoptosis opens up possibilities to antagonize apoptosis escape routes. We have earlier shown that acute lymphoblastic leukemia (ALL) harbors a distinct propensity to undergo cell death by receptor-interacting protein kinase 1 (RIPK1)-dependent necroptosis, activated by small-molecule second mitochondria-derived activators of caspase (SMAC) mimetics. Despite demonstrated safety and tolerability of SMAC mimetics in clinical trials, their efficacy as single agent seems still limited, highlighting the need for combinatorial treatments. Here, we investigate so far unexplored regulatory mechanisms of necroptosis and identify targets for interference to augment the necroptotic antileukemia response. Ex vivo drug response profiling in a model of the bone marrow microenvironment reveals powerful synergy of necroptosis induction with histone deacetylase (HDAC) inhibition. Subsequent transcriptome analysis and functional in vivo CRISPR screening identify gene regulatory circuitries through the master transcription regulators specificity protein 1 (SP1), p300, and HDAC2 to drive necroptosis. Although deletion of SP1 or p300 confers resistance to necroptosis, loss of HDAC2 sensitizes cells to RIPK1-dependent cell death by SMAC mimetics. Consequently, our data inform strong in vivo antileukemic activity of combinatorial necroptosis induction and HDAC inhibition in patient-derived human leukemia models. Thus, transcriptional dependency of necroptosis activation is a key regulatory mechanism that identifies novel targets for interference, pointing out a strategy to exploit alternative nonapoptotic cell death pathways to eradicate resistant disease.

Graphical abstract

Figure 1.

Apoptosis modulators and HDACis potentiate SM response, showing synergistic potential in ALL PDXs. (A) Schematic representation of experimental set up. (B) DRP from 6 different PDXs, screened for synergy between the SM birinapant and 24 anticancer compounds. Each dot represents a sample. Mean, and standard deviation are also shown. Samples included BCP-ALL (PID0083, PID0874, PID0852, PID0859, and PID0858) and T-ALL (PID0884). Within each PDX, n = 1. (C) Heat map depicting mean synergy scores (ZIP, Loewe, HSA, and Bliss) for combinations of SM + apoptosis modulators and SM + HDACis, indicated for each PDX. White represents synergy scores <0; pink represents synergy scores between 0 and +5; red represents synergy scores above +5. HSA, highest single agent; IC₅₀, 50% inhibitory concentration; MSC, mesenchymal stem cell.

Figure 2.

SM synergizes with HDACi across ALL PDXs in a RIPK1-dependent fashion. (A) Heat map depicting mean synergy scores (ZIP, Loewe, HSA, and Bliss) obtained from matrix synergy screens performed in 18 BCP-ALL and T-ALL, including relapse and diagnosis samples. White represents synergy scores <0; pink represents synergy scores between 0 and +5; red represents synergy scores above +5 (left). The Wilcoxon matched-pairs signed rank test revealing significant differences in mean ZIP scores between SM + Mo combination and the other combinations. Lighter dots, BCP-ALL; darker dots, T-ALL (right). (B) Individual sensitivities (log₁₀IC₅₀) toward SM or Mo and mean ZIP scores value for the SM + Mo combination, calculated for 47 BCP-ALL PDXs, grouped by subtype. (C) Immunoblot showing RIPK1 depletion after 6-hour treatment with LD4172 in 5 PDXs. GAPDH is used as a loading control. Dot plot depicting normalized logarithmic SM AUC (normalized AUC) upon SM treatment alone vs pretreatment with LD4172. (D) The Wilcoxon matched-pairs signed rank test indicating the significance of the variation in ZIP scores between SM + Mo combination with and without LD4172-mediated RIPK1 degradation, in 42 BCP-ALL PDXs. ns, P > .05; ∗∗∗P ≤ .001; ∗∗∗∗P ≤ .0001. Replicates within each PDX are indicated on respective underlying data table (supplemental Data 1). SM, birinapant. AUC, area under the curve; En, entinostat; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; ns, not significant; Pa, panobinostat; UNT, untreated.

Figure 3.

Transcriptome analysis indicates early events during necroptosis, suggesting transcriptional dependency. (A) Validation of RIPK1^−/− PDX: immunoblot showing efficient knockout of RIPK1, whereas GAPDH is used as a loading control (left), and viability curves indicate response to the SM birinapant in RIPK1^+/+ vs RIPK1^−/− PDX (right). (B) Schematic representation of experimental setup. (C) Functional enrichment analysis (displayed: Kyoto Encyclopedia of Genes and Genomes [KEGG] pathways) upon 5 hours of SM treatment in PID0117 RIPK1^−/− cells (log₂FC > 1). (D) Functional enrichment analysis (displayed: KEGG pathways) for the RIPK1-specific signature (log₂FC > 0.58). To obtain this signature, log₂FC values resulting from SM treatment vs untreated conditions were additionally compared between the 2 isogenic PDXs. Overrepresentation of genes belonging to specific pathways (fold enrichment) is shown on the x-axis; dot sizes represent the number of genes belonging to the specific pathway; FDR is indicated using different colors (green, lower FDR; the enrichment is less likely to be caused by chance). (E) Volcano plot showing differential gene expression (upregulation/downregulation) of genes in RIPK1^+/+ cells upon 5 hours of SM treatment. Highlighted are genes upregulated with DNA-binding transcription activity (GO:MF, red) and downregulated with DNA modification activity (GO:MF, blue) in RIPK1^+/+ cells upon SM treatment. Horizontal dotted line, −log₁₀ (.05); vertical dotted lines, log₂FC (±1). (F) Representation of differentially regulated genes in RIPK1^+/+ cells (x-axis) in relation to RIPK1 signature (y-axis) after 5 hours of SM treatment. Colored, labeled dots: genes differentially regulated by |log₂FC| >0.58 and P value (adjusted) < .05. (G) TRRUST analysis of differentially regulated genes for RIPK1^+/+ cells (log₂FC > 1), highlighting the regulators of genes of interest upon 5 hours SM treatment. (H) TRRUST analysis of differentially regulated genes for RIPK1 signature (log₂FC > 0.58). RNA sequencing was conducted in triplicates. CTRL, control; FC, fold cange; FDR, false discovery rate; UNTR, untreated.

Figure 4.

sgRNA CRISPR screen highlights key regulators of SM response, including SP1 and acetylation-modulating enzymes. (A) Schematic representation of experimental setup. (B) Volcano plot displaying significantly enriched or depleted sgRNAs (cutoffs: |robust rank aggregation [score]| > 0.495; FDR < 0.05; left). sgRNA rank plot depicting enriched (red) or dropout (blue) sgRNAs for the genes shown with the volcano plot plus intergenic controls (right). Each line represents 1 sgRNA; total sgRNAs per gene are 12 (4 sgRNAs from the Brunello library, 3 replicates). (C) Transcription frequency in TPM of sgRNAs for genes of interest. RIPK1 is included as a positive control. One dot represents 1 replicate. TPM, transcript per million.

Figure 5.

Depletion of SP1, p300, or HDAC2 alters sensitivity toward SM in different PDXs. (A) Immunoblot showing depletion of p300 upon treatment for 6 hours with various concentrations of dCBP1 or DMSO in 2 PDXs. GAPDH is used as a loading control (left). Dot plot showing normalized AUC upon SM treatment in 10 SM-sensitive BCP-ALL with or without pretreatment with dCBP1 for 6 hours (right). The Wilcoxon matched-pairs signed rank test revealing significant differences in normalized AUC. PDXs included PID0036, PID0556, PID0872, PID0083, PID0788, PID0874, PID0117, PID0852, PID0859, and PID0858. (B) Box plot showing SM normalized AUC from 7 BCP-ALL PDXs treated with indicated p300 inhibitors (A-485, 0.1-10 μM; INB, 0.01-1 μM; concentration ranges were selected based on different potencies of the compounds). The Wilcoxon matched-pairs signed rank test detecting differences in the normalized AUC upon SM treatment after pretreatment with different p300 inhibitors compared with SM alone (gray∗) or compared with increasing concentrations of the same inhibitors (black∗). Median is displayed for each condition. PDXs included PID0556, PID0083, PID0874, PID0117, PID0852, PID0859, and PID0858. (C) Bar plot depicting the ratio of RPF657⁺ population (construct-containing population) calculated on human CD19⁺ (hCD19⁺) population in spleens of mice treated with SM over spleens of mice treated with vehicle. HDAC2, population containing LC.RFP657.HDAC2 construct; shuttle, population containing LC.RFP657.shuttle construct (used as control); SP1, population containing LC.RFP657.SP1 construct. No difference, y = 1; enrichment, y > 1; depletion, y < 1. Dot plot depicts hCD19⁺ cells percent over the total lymphocyte population (hCD19⁺ + mCD45⁺) in mice’s spleens at the end of treatment. Multiple comparison 2-way analysis of variance (ANOVA) analysis showing significant difference in ALL population in SM-treated HDAC2-depleted PID0117 spleen compared with the shuttle counterpart. ns, P > .05; ∗P ≤ .05; ∗∗P ≤ .01 ∗∗∗∗P ≤ .0001.

Figure 6.

SM and HDACi cooperate to delay leukemia progression in vivo. (A) Schematic representation of experimental setup. (B) Representative examples of leukemia progression dynamics during treatment of 4 PDXs. Shaded in blue is the treatment window. (C) Time elapse (in days) to reach ALL progression of 20% in peripheral blood, calculated between treated mouse and the respective control (vehicle) for each PDX (vehicle – treatment). A log-rank (Mantel-Cox) test revealing significant differences in progression for mice treated with the single agents vs combinatorial treatment. (D) ALL-progression difference, indicating the difference in leukemia percent (in peripheral blood) calculated between control mouse and treated mouse (vehicle – treatment) at the end point of control mice, relative to each PDX. Same treatment conditions are plotted together. The Wilcoxon matched-pairs signed rank test revealing significant differences in progression for mice treated with the single agents vs combinatorial treatment. Median is depicted. Leukemia percent is calculated based on hCD19⁺ cells over the total lymphocyte population (hCD19⁺ + mCD45⁺). Gray, vehicle; red, birinapant; green, Mo; blue, combination. SM, birinapant 15 mg/kg; Mo, 25 mg/kg; SM + Mo, birinapant 15 mg/kg + Mo 25 mg/kg. Eight PDXs are included. ∗P ≤ .05; ∗∗P ≤ .01; ∗∗∗∗P ≤ .0001. (E) Graphical representation of the proposed mechanism of transcriptional regulation of SM-induced cell death in ALL.