Although inhibitors of the kinases CHK1, ATR, and WEE1 are undergoing clinical testing, it remains unclear how these three classes of agents kill susceptible cells and whether they utilize the same cytotoxic mechanism. Here we observed that CHK1 inhibition induces apoptosis in a subset of acute leukemia cell lines in vitro, including TP53-null acute myeloid leukemia (AML) and BCR/ABL–positive acute lymphoid leukemia (ALL), and inhibits leukemic colony formation in clinical AML samples ex vivo. In further studies, downregulation or inhibition of CHK1 triggered signaling in sensitive human acute leukemia cell lines that involved CDK2 activation followed by AP1-dependent TNF transactivation, TNFα production, and engagement of a TNFR1- and BID-dependent apoptotic pathway. AML lines that were intrinsically resistant to CHK1 inhibition exhibited high CHK1 expression and were sensitized by CHK1 downregulation. Signaling through this same CDK2–AP1–TNF cytotoxic pathway was also initiated by ATR or WEE1 inhibitors in vitro and during CHK1 inhibitor treatment of AML xenografts in vivo. Collectively, these observations not only identify new contributors to the antileukemic cell action of CHK1, ATR, and WEE1 inhibitors, but also delineate a previously undescribed pathway leading from aberrant CDK2 activation to death ligand–induced killing that can potentially be exploited for acute leukemia treatment.

Significance:

This study demonstrates that replication checkpoint inhibitors can kill AML cells through a pathway involving AP1-mediated TNF gene activation and subsequent TP53-independent, TNFα-induced apoptosis, which can potentially be exploited clinically.

Inhibitors of the kinases CHK1, ATR, and WEE1 have undergone extensive preclinical and clinical testing. The ability of these agents to disrupt the S and G2–M checkpoints, destabilize stalled replication forks, and enhance the cytotoxicity of antimetabolites, TOP1 poisons, and platinating agents provides a rationale for combinations that include these inhibitors (1–4).

In addition to modulating the cytotoxicity of agents that induce replication stress, loss or inhibition of CHK1 has been observed to kill certain neoplastic cell lines in the absence of exogenous DNA damage (e.g., 5, 6–10). WEE1 inhibitors (WEE1i) and ATR inhibitors (ATRi) also kill neoplastic cells as single agents (11–14). Despite extensive preclinical and clinical studies, incomplete understanding of the action of these agents has impeded clinical development. For example, the second-generation checkpoint kinase inhibitor prexasertib exhibits single-agent activity in ovarian cancer (15), but it has been difficult to identify the subset of tumors most likely to respond (16). Better understanding of the cytotoxic mechanism of these agents could potentially inform this issue.

Many anticancer drugs, including targeted agents and conventional chemotherapies, modulate BCL2 family member expression or activity to trigger the mitochondrial apoptotic pathway (17, 18). For example, DNA-damaging agents upregulate the TP53 transcriptional targets PUMA and NOXA to trigger activation of BAX and BAK (19). In contrast, only a few anticancer treatments have been shown to trigger the death receptor pathway. In particular, the CD95/FAS receptor plays an important role in fluoropyrimidine-induced colon cancer cell death (20), and the death ligand TRAIL contributes to tretinoin-induced killing of promyelocytic leukemia cells (21, 22). In both cases, killing is initiated by binding of the death ligand to cell surface receptors, which leads to adaptor protein-mediated CASP8 activation and subsequent proteolytic activation of the proapoptotic BCL2 family member BID (23, 24).

The third major death ligand, tumor necrosis factor α (TNFα), was initially identified as a cytokine that induces necrosis but has been subsequently recognized as an important contributor to inflammation and tumorigenesis (25). To our knowledge, TNFα production by tumor cells has not been previously implicated in the apoptotic response to targeted antineoplastic agents. Moreover, CDK2 activation has not previously been linked to death receptor pathway activation.

Here we conducted experiments designed to (i) assess the impact of CHK1 inhibitors in acute leukemia, particularly AML, as single agents; (ii) investigate signaling pathways leading to the death of sensitive leukemia cells; and (iii) determine whether ATR and WEE1 inhibitors kill through a similar mechanism. Results of these studies show that inhibitors of CHK1, ATR, and WEE1 can all trigger CDK2-dependent TNFα upregulation that contributes to the death of susceptible cells, thereby providing the broad outline of a previously unrecognized cytotoxic mechanism for AML.

Tissue culture

All cell lines were grown under 5% CO2 at 37°C. Cell lines were obtained as follows: HL-60, U937, HEL 92.1.7, and THP.1 from ATCC; ML-2, Molm13, and SET2 from Raoul Tibes (New York University, New York, NY); ML-1, K562, Jurkat, and Z181 from Michael Kastan (Duke University, Durham, NC), Robert Abraham (Pfizer), Paul Leibson (Mayo Clinic Rochester, Rochester, MN) and Joya Chandra (M.D. Anderson Cancer Center, Houston, TX), respectively. All lines were authenticated by short tandem repeat profiling in the Mayo Clinic Cytogenetics Core, verified to be mycoplasma free, maintained at <1 × 106 cells/mL in RPMI1640 containing 20% (SET2) or 10% (other lines) heat-inactivated FBS, and passaged <3 months before use.

Transfections

For transient knockdown experiments, cells growing in antibiotic-free medium were suspended in 380 μL RPMI1640 medium containing 10% FBS and 20–40 μmol/L siRNA (see Supplementary Table S1). To diminish transfection-associated killing, 12.5 mmol/L HEPES (pH 7.4) was added during transfection of THP.1 cells. After incubation for 5 minutes, cells were subjected to electroporation using a BTX830 square wave electroporator (BTX) delivering a single 10-ms pulse at 280 V. After a 15-minute incubation, cells were transferred to 25 mL antibiotic-free medium for 24–48 hours and then treated, harvested for immunoblotting, or assayed for Annexin V binding. Protection from drug-induced apoptosis afforded by each siRNA was calculated according to the formula % protection = 100 [1 – (apoptosisdrug + siRNA – apoptosisdiluent + siRNA)/ (apoptosisdrug + control siRNA – apoptosisdiluent + control siRNA)], where apoptosisdrug is the percentage of apoptotic cells after transfection with the indicated siRNA followed by subsequent drug treatment and apoptosisdiluent is the percentage of apoptotic cells after transfection with the indicated siRNA and subsequent diluent treatment.

To generate BID−/− cells, oligonucleotides (5′-CGCAGAGAGCTGGACGCACT-3′) guiding to human BID 591–610 (accession number: NM_197966.2) were synthesized, annealed, and cloned into the BsmBI site of lentiCRISPR-v2 plasmid (Addgene). BID targeting virus and empty vector were packaged in lentiviral particles by transfecting HEK293T cells with the packaging vector psPAX3, envelope vector pMD2.G, and lentiCRISPR-v2-BID 591–610 or empty vector using Lipofectamine 2000 (Thermo Fisher Scientific). Two days after viral transduction, U937 cells were selected with 3 μg/mL puromycin for a week, cloned by limiting dilution in 96-well plates, and assayed for gene interruption by immunoblotting.

MTS assay

Aliquots containing 2 × 104 cells in 120 μL of their usual growth medium were treated with varying prexasertib concentrations. After incubation for 6 days, plates were treated with MTS and phenazine methosulfate as instructed by the manufacturer and incubated for 2–6 hours to obtain an absorbance of 0.5–1.0 at 490 nm in control samples.

Normal and leukemic cell drug sensitivity testing ex vivo

Studies in human samples were performed in accordance with the Declaration of Helsinki. Patients with newly diagnosed AML underwent bone marrow aspiration prior to therapy. Marrows from hip arthroplasty specimens served as normal controls. Under the aegis of Institutional Review Board–approved protocols, all subjects provided informed written consent for research use of their specimens.

For colony-forming assays, mononuclear cells were isolated on ficoll-Hypaque step gradients (density 1.077 gm/mL), washed with RPMI1640 medium, resuspended at 1.5 × 106 cells/mL, and plated at 600,000 cells/plate in MethoCult methylcellulose containing the indicated concentrations of prexasertib, MK-8776, or cytarabine. Normal colonies (sum of CFU-G, CFU-GM, and BFU-E) or leukemic colonies were counted after a 10- to 14-day incubation (26).

To assess the short-term impact of treatments on normal stem and progenitor cells, freshly isolated cells were resuspended at 1 × 106 cells/mL in Iscove modified Dulbecco's medium containing 20% (v:v) FBS, 2 mmol/L glutamine, and diluent (0.1% DMSO), 10 nmol/L prexasertib, or 100 nmol/L venetoclax. After a 24-hour incubation, cells were sedimented and subjected to multi-parameter flow cytometry as described previously (27). After collection of 800,000 events/sample, data were analyzed using FlowJo software and the number of surviving progenitor (CD34+/CD45dim/CD38+) and hematopoietic stem cells (CD34+/CD45dim/CD38/CD90+/CD45RA) were compared in diluent- versus drug-treated samples.

Animal studies

Under protocols approved by the Mayo Clinic Institutional Animal Care and Use Committee, xenograft studies were performed according to the NIH Guide for the Care and Use of Laboratory Animals. In brief, 4- to 5-week old nude mice (HSD:Athymic nude-Foxn1nu, Envigo) were implanted subcutaneously with radiofrequency identification chips along the neck and 100 μL of a 1:1 slurry containing Matrigel (BD Biosciences) and 5 × 106 washed, log-phase ML-1, or U937 cells in the right flank. When the neoplastic implants reached an average volume of 100 mm3, mice were randomized to receive diluent [(2-hydroxypropyl) β-cyclodextrin, 45% w/v solution in water] or prexasertib (10 mg/kg or 15 mg/kg as indicated in the figure legends) at 12-hour intervals on days 1–3, 8–10, 15–17, and 22–24 (9). Alternatively, 4- to 5-week-old female NSGS mice were inoculated intravenously with 5 × 106 marrow cells from founder mice with the Ph+ ALL patient-derived xenograft (PDX) and randomized to treatment with diluent or prexasertib 15 mg/kg on the schedule described above beginning on day 40 after inoculation. Animals were sacrificed when implants exceeded 1,000 mm3 or body conditioning score dropped below 6.

Quantitative RT-PCR

qRT-PCR was performed in triplicate using 100 ng RNA and TaqMan One-Step RT-PCR Master Mix (Applied Biosystems) per the manufacturer's instructions using human TNFα and GAPDH probes (Supplementary Table S1). PCR was performed on a ABI Prism 7900HT Real Time System using a program of 48°C for 30 minutes, 95°C for 10 minutes, then 40 cycles of 95°C for 15 seconds, and 60°C for 1 minute. Data analysis was performed using the following equations: ΔCt = Ct (sample)-Ct(endogenous control); ΔΔCt = ΔCt(sample)-ΔCt(untreated); and fold change = 2–ΔΔCt.

ELISA

The concentration of human TNFα protein secreted into the culture medium was assayed by ELISA (kit #HSTA00E, R&D Systems).

Statistical analysis

Prexasertib dose–response curves were performed in cell lines at least three times independently. Error bars in all experiments represent mean ± SD of three independent experiments. *, **, and *** indicate P < 0.05, P < 0.01, and P < 0.001, respectively, using unpaired t tests. Survival curves were plotted by the Kaplan–Meier method and analyzed using Cox proportional hazard models.

Additional methods

Additional methods, including those for cell-cycle analysis, apoptosis assays, immunoblotting, immunoprecipitation, chromatin immunoprecipitation, and RNA-sequencing (RNA-seq), have been reported previously (28, 29) and are described in the Supplementary Methods. The RNA-sequencing data have been deposited into the GEO database as GSE131912 and are available at https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE131912. Sources of materials are provided in Supplementary Table S1.

Effect of CHK1 knockdown or inhibition on AML cell viability

To investigate the antileukemic effects of CHK1i monotherapy, 10 myeloid leukemia lines with various genetic abnormalities (Fig. 1A, inset) were assayed for sensitivity to the checkpoint kinase inhibitor prexasertib. After a 6-day exposure, MTS assays indicated IC50 values from 1–22 nmol/L (Fig. 1A), a clinically achievable concentration range (30). These observations prompted us to further examine the impact of CHK1 perturbation on AML cell-cycle progression and survival using prexasertib, the highly selective CHK1i MK-8776 (31), and CHK1 siRNA.

Figure 1.

Proapoptotic effects of CHK1 inhibition or knockdown. A, Human AML lines with the indicated genomic features were treated for 6 days with prexasertib and assayed for MTS reduction. U937 (B–E) and THP.1 cells (C, D, and F) were treated for 24 hours with prexasertib ± the caspase inhibitor Q-VD-OPh (5 μmol/L) and analyzed for DNA fragmentation (B and C), Annexin V binding (D), or cleavage of caspase substrates (E and F). Similar results were observed in ML-1 cells (Supplementary Fig. S1). B, Inset, immunoblot of CHK1 Ser296 autophosphorylation, a marker of CHK1 activity, after a 6-hour treatment with prexasertib and 30 nmol/L bortezomib (to prevent proteasome-mediated CHK1 degradation). G, after siRNA transfection, THP.1 cells were incubated for 48 hours and harvested for Annexin V binding (bar graph) or immunoblotting. Error bars in A, C, D, and G, mean ± SD of three independent experiments. *, P < 0.05; **, P < 0.01; ***, P < 0.001, respectively, relative to control siRNA or treatment with prexasertib alone. H, Freshly isolated clinical AML isolates (black or green lines) were plated in methylcellulose with prexasertib (0, 3, 6, and 10 nmol/L or 0, 3, 6, and 12 nmol/L) and assayed for leukemic colonies (CFU-L). Green lines, samples with TP53 mutations. Black lines, samples without known TP53 mutations (Supplementary Table S3). Orange lines, myeloid colony formation by cells from normal marrows. Numbers correspond to sample numbers in Supplementary Table S3. I, Effect of diluent versus 10 nmol/L prexasertib (Prex) on relative survival of normal myeloid stem and progenitor cells (CD34+/CD45dim/CD38+) or hematopoietic stem cells (HSCs, CD34+/CD38/CD90+/CD45RA) at 24 hours as assessed by 10-color flow cytometry (27). Venetoclax (100 nmol/L) was a positive control for toxicity in stem and progenitor cells. Circles in I, results with individual samples. Error bars, mean ± SD of assays in 4–5 separate samples.

Figure 1.

Proapoptotic effects of CHK1 inhibition or knockdown. A, Human AML lines with the indicated genomic features were treated for 6 days with prexasertib and assayed for MTS reduction. U937 (B–E) and THP.1 cells (C, D, and F) were treated for 24 hours with prexasertib ± the caspase inhibitor Q-VD-OPh (5 μmol/L) and analyzed for DNA fragmentation (B and C), Annexin V binding (D), or cleavage of caspase substrates (E and F). Similar results were observed in ML-1 cells (Supplementary Fig. S1). B, Inset, immunoblot of CHK1 Ser296 autophosphorylation, a marker of CHK1 activity, after a 6-hour treatment with prexasertib and 30 nmol/L bortezomib (to prevent proteasome-mediated CHK1 degradation). G, after siRNA transfection, THP.1 cells were incubated for 48 hours and harvested for Annexin V binding (bar graph) or immunoblotting. Error bars in A, C, D, and G, mean ± SD of three independent experiments. *, P < 0.05; **, P < 0.01; ***, P < 0.001, respectively, relative to control siRNA or treatment with prexasertib alone. H, Freshly isolated clinical AML isolates (black or green lines) were plated in methylcellulose with prexasertib (0, 3, 6, and 10 nmol/L or 0, 3, 6, and 12 nmol/L) and assayed for leukemic colonies (CFU-L). Green lines, samples with TP53 mutations. Black lines, samples without known TP53 mutations (Supplementary Table S3). Orange lines, myeloid colony formation by cells from normal marrows. Numbers correspond to sample numbers in Supplementary Table S3. I, Effect of diluent versus 10 nmol/L prexasertib (Prex) on relative survival of normal myeloid stem and progenitor cells (CD34+/CD45dim/CD38+) or hematopoietic stem cells (HSCs, CD34+/CD38/CD90+/CD45RA) at 24 hours as assessed by 10-color flow cytometry (27). Venetoclax (100 nmol/L) was a positive control for toxicity in stem and progenitor cells. Circles in I, results with individual samples. Error bars, mean ± SD of assays in 4–5 separate samples.

Close modal

Analysis of three acute myeloid leukemia (AML) cell lines with low to medium IC50s (ML-1, THP.1, and U937) revealed that prexasertib induced DNA fragmentation (Fig. 1B and C; Supplementary Fig. S1), phosphatidylserine exposure (Fig. 1D), and cleavage of caspase substrates (Fig. 1E and F)–three hallmarks of apoptosis–within 24 hours. These apoptotic changes accompanied induction of replication stress, as indicated by phosphorylation of replication protein A (RPA; Fig. 1E and F). Addition of the broad-spectrum caspase inhibitor Q-VD-OPh did not affect the replication stress (Fig. 1E and F) but did diminish phosphatidylserine externalization (Fig. 1D), caspase-induced proteolysis (Fig. 1E and F), and DNA fragmentation (Supplementary Fig. S1B and S1D), providing further evidence for caspase-mediated killing. Similar effects were observed with MK-8776 (Supplementary Fig. S1B and S1D). Importantly, CHK1 knockdown also induced replication stress accompanied by caspase activation and phosphatidylserine externalization (Fig. 1G; Supplementary Fig. S2). Collectively, these observations indicate that loss of CHK1 activity or protein is deleterious to multiple AML lines.

To place these results in the context of previous studies of CHK1i-induced sensitization to antimetabolites (32, 33), we also treated U937 cells with various CHK1is in the absence and presence of cytarabine (Supplementary Fig. S3). These studies revealed that multiple CHK1is induce apoptosis as single agents at concentrations only 2- to 3-fold higher than those that sensitize cells to cytarabine (Supplementary Fig. S3, red symbols).

Consistent with these results, analysis of primary AML specimens using colony-forming assays indicated that CHK1 inhibition not only sensitized leukemic colony-forming cells to cytarabine, but also markedly diminished colony formation as monotherapy in approximately 50% of primary AML specimens (Supplementary Fig. S4; Supplementary Table S2). More detailed analysis demonstrated that colony formation was inhibited by >90% at 10–12 nmol/L prexasertib in 7 of 13 clinical AML isolates ex vivo (Fig. 1H; Supplementary Table S3). Notably, 2 of 3 TP53-mutant AMLs were among the sensitive isolates (Fig. 1H, green lines). In contrast, prexasertib had much less impact on normal myeloid stem and progenitor cells at the same concentrations (Fig. 1H, orange lines; Fig. 1I, orange bars).

CHK1is induce AML cell apoptosis through a CDC25A-CDK2–dependent mechanism

Although CHK1 inhibition can induce mitotic catastrophe in certain cell types (34), AML cells treated with prexasertib did not show evidence of mitotic arrest as assessed by flow cytometry, morphologic analysis, or staining for the mitotic marker phospho-Ser28 Histone H3 (Supplementary Fig. S5). Instead, prexasertib induced accumulation of AML lines in S phase (Supplementary Fig. S5A, S5B, S5D, and S5E), prompting us to focus on S-phase events as initiators of cytotoxicity.

Further experiments showed that CHK1 inhibition was accompanied by increased levels of the CDK2 activator CDC25A (Supplementary Fig. S6A), in agreement with earlier studies. Importantly, experiments performed in U937 cells because of their facile transfection with siRNA showed that knockdown of CDC25A (Fig. 2A; Supplementary Fig. S6B), CDK2 or its binding partner CCNA2 diminished CHK1i-induced apoptosis (Fig. 2B). Likewise, CDK2 inhibition diminished CHK1i-induced apoptosis in AML lines (e.g., Fig. 2C; Supplementary Fig. S6C and S6D) and blunted the effects of prexasertib in clinical AML isolates ex vivo (Fig. 2D), suggesting that CDK2 contributes to CHK1i-induced killing.

Figure 2.

Contribution of CDC25A and CDK2 to CHK1i killing. A and B, Twenty-four hours after siRNA transfection, U937 cells were treated for 24 hours with the indicated CHK1i and analyzed by Annexin V staining. Insets, immunoblots showing knockdown. C, U937 cells were treated for 24 hours with 10 nmol/L prexasertib (Prex) in the absence or presence of 350 nmol/L CDK2i and analyzed for Annexin V staining. D, AML isolates were plated in methylcellulose in the presence of 10–12 nmol/L prexasertib ± CDK2i and examined for leukemic colonies. Values are normalized to diluent treated control samples. Error bars in A–C, mean ± SD from three independent experiments. *, P < 0.05; **, P < 0.01.

Figure 2.

Contribution of CDC25A and CDK2 to CHK1i killing. A and B, Twenty-four hours after siRNA transfection, U937 cells were treated for 24 hours with the indicated CHK1i and analyzed by Annexin V staining. Insets, immunoblots showing knockdown. C, U937 cells were treated for 24 hours with 10 nmol/L prexasertib (Prex) in the absence or presence of 350 nmol/L CDK2i and analyzed for Annexin V staining. D, AML isolates were plated in methylcellulose in the presence of 10–12 nmol/L prexasertib ± CDK2i and examined for leukemic colonies. Values are normalized to diluent treated control samples. Error bars in A–C, mean ± SD from three independent experiments. *, P < 0.05; **, P < 0.01.

Close modal

Contribution of TNFα to CHK1i-induced killing of susceptible cells

Subsequent experiments focused on identifying the CHK1i-induced apoptotic pathway and tying that pathway to CDK2 activation. An RNA-seq analysis comparing diluent- and prexasertib-treated U937 cells revealed that 384 transcripts were highly differentially expressed after prexasertib treatment, including 67 transcripts associated with cell death (labeled in Fig. 3A). Among the upregulated transcripts were those encoding the death ligand TNFα and the BH3-only protein PMAIP1/NOXA (labeled red). Gene-set enrichment analysis of all significantly upregulated genes indicated that TNFα signaling, inflammatory signaling, and apoptotic signaling were among the five most significantly altered pathways after prexasertib treatment (Fig. 3B).

Figure 3.

TNFα contribution to CHK1i-induced apoptosis. A, Heat-map of RNA-seq experiment comparing biological replicates of U937 cells treated for 24 hours with diluent (Dil) or 10 nmol/L prexasertib (Prex) or 10 nmol/L prexasertib + CDK2i (P + C). Of 384 transcripts that were highly differentially expressed, defined as |log2 fold change|>2 and P value < 0.01, in response to prexasertib, 67 (labeled) were implicated in cell death or apoptosis. Red labels, transcripts that figured prominently in subsequent analysis. B, Significantly upregulated differentially expressed genes, defined as [|log2 fold change|>1 and FDR < 0.01], were queried against the hallmark gene sets in the Molecular Signatures Database v7.0 (MSigDB) as described in the Supplementary Methods. Natural log of FDR (q value) given for the top ten gene sets is shown on the y-axis, with the relative proportion of genes in the overlap (genes in overlap/genes in gene set) shown on the x-axis. TNFα signaling via NFκB is the most significantly upregulated pathway by FDR (q value) as well as by the relative proportion of genes in the overlap. C, Cells treated with diluent, 10 nmol/L prexasertib (Prex), or 1 μmol/L MK-8776 for 24 hours were stained with Annexin V. D and E, Cells treated for 24 hours with prexasertib or MK-8776 in the presence of Q-VD-OPh (to prevent apoptosis-associated mRNA degradation – 28) were assayed for TNFα mRNA (D) and TNFα protein release into the medium (E). F,TNF luciferase reporter construct (top) containing 1,100 bp of upstream sequence (the canonical TNF promoter – 40) was transfected into U937 cells along with pTK Renilla transfection control. After 24 hours, cells were treated with diluent or CHK1i for 24 hours, then assayed for firefly and Renilla luciferase. The ratio was then normalized to diluent-treated controls. G, After siRNA transfection, THP.1 cells were incubated for 48 hours in the presence of Q-VD-OPh and assayed for TNFα mRNA. H–J, Twenty-four hours after siRNA transfection, U937 cells were treated for 24 hours with diluent or CHK1i and analyzed for Annexin V binding (H and I) or relative TNFα mRNA (J). Insets in G, I, and J, immunoblots to assess knockdowns. K, U937 cells treated for 24 hours with recombinant human (rh) TNFα were stained with Annexin V (bar graph) or subjected to immunoblotting (inset). Error bars, mean ± SD from three independent experiments. *, P < 0.05; **, P < 0.01.

Figure 3.

TNFα contribution to CHK1i-induced apoptosis. A, Heat-map of RNA-seq experiment comparing biological replicates of U937 cells treated for 24 hours with diluent (Dil) or 10 nmol/L prexasertib (Prex) or 10 nmol/L prexasertib + CDK2i (P + C). Of 384 transcripts that were highly differentially expressed, defined as |log2 fold change|>2 and P value < 0.01, in response to prexasertib, 67 (labeled) were implicated in cell death or apoptosis. Red labels, transcripts that figured prominently in subsequent analysis. B, Significantly upregulated differentially expressed genes, defined as [|log2 fold change|>1 and FDR < 0.01], were queried against the hallmark gene sets in the Molecular Signatures Database v7.0 (MSigDB) as described in the Supplementary Methods. Natural log of FDR (q value) given for the top ten gene sets is shown on the y-axis, with the relative proportion of genes in the overlap (genes in overlap/genes in gene set) shown on the x-axis. TNFα signaling via NFκB is the most significantly upregulated pathway by FDR (q value) as well as by the relative proportion of genes in the overlap. C, Cells treated with diluent, 10 nmol/L prexasertib (Prex), or 1 μmol/L MK-8776 for 24 hours were stained with Annexin V. D and E, Cells treated for 24 hours with prexasertib or MK-8776 in the presence of Q-VD-OPh (to prevent apoptosis-associated mRNA degradation – 28) were assayed for TNFα mRNA (D) and TNFα protein release into the medium (E). F,TNF luciferase reporter construct (top) containing 1,100 bp of upstream sequence (the canonical TNF promoter – 40) was transfected into U937 cells along with pTK Renilla transfection control. After 24 hours, cells were treated with diluent or CHK1i for 24 hours, then assayed for firefly and Renilla luciferase. The ratio was then normalized to diluent-treated controls. G, After siRNA transfection, THP.1 cells were incubated for 48 hours in the presence of Q-VD-OPh and assayed for TNFα mRNA. H–J, Twenty-four hours after siRNA transfection, U937 cells were treated for 24 hours with diluent or CHK1i and analyzed for Annexin V binding (H and I) or relative TNFα mRNA (J). Insets in G, I, and J, immunoblots to assess knockdowns. K, U937 cells treated for 24 hours with recombinant human (rh) TNFα were stained with Annexin V (bar graph) or subjected to immunoblotting (inset). Error bars, mean ± SD from three independent experiments. *, P < 0.05; **, P < 0.01.

Close modal

Consistent with these results, CHK1i-induced apoptosis (Fig. 3C) was accompanied by 10- to 50-fold upregulation of TNFα mRNA and increased TNFα secretion (Fig. 3D and E), which reflected TNF promoter activation (Fig. 3F). CHK1 knockdown likewise induced TNFα (Fig. 3G). In contrast, the death ligand TRAIL was not upregulated to the same extent at either the mRNA (Supplementary Fig. S7A) or protein level (Supplementary Fig. S7B). Importantly, downregulation of TNFα (Fig. 3H) or its receptor TNFR1 (Fig. 3I) diminished CHK1i-induced apoptosis, placing TNFα and TNFR1 in the CHK1i killing pathway. Moreover, TNFα upregulation was diminished by the same treatments that decreased CHK1i-induced apoptosis, including CDK2 or CCNA2 downregulation (Fig. 3J) and CDK2 inhibition (Supplementary Fig. S7C and S7D), placing TNFα between CDK2 activation and cell death. Finally, TNFα as a single agent also induced apoptosis in these cells (Fig. 3K). Collectively, these observations suggest that activated CDK2 signals through TNFα and TNFR1 to induce AML cell apoptosis.

Prexasertib upregulated not only TNFα, but also the PMAIP1 transcript encoding the proapoptotic BCL2 family member NOXA (Fig. 3A), a BH3-only protein previously implicated in triggering the intrinsic apoptotic pathway after treatment of neoplastic cells with the proteasome inhibitor bortezomib, the NEDD8-activating enzyme inhibitor pevonedistat or the PARP inhibitor olaparib (35–37). Accordingly, we also examined the impact of CHK1is on NOXA and other BCL2 family members. Despite changes in several pro- and antiapoptotic BCL2 family members at the mRNA level (Supplementary Fig. S8A), prexasertib and MK-8776 failed to consistently alter BCL2 family member expression at the protein level (Fig. 4A). Instead, CHK1 inhibition (Fig. 4A) or knockdown (Fig. 4B) induced cleavage of BID, a BH3-only protein that amplifies the proapoptotic signal after death ligand–mediated caspase 8 activation (38, 39). This cleavage was associated with localization of the BID C-terminal (active) fragment (23, 24) to mitochondria (Fig. 4C) and activation of the mitochondrial permeabilizers BAX and BAK (Supplementary Fig. S8B). Consistent with a critical role for BID in CHK1i-induced killing, BID knockdown protected almost as effectively as BAX and BAK knockdown (Fig. 4D), whereas BIM, PUMA, or NOXA knockdown had more limited effects. CRISPR/Cas9-mediated BID knockout likewise reduced prexasertib-induced apoptosis (Fig. 4E). CASP8, which activates BID during death receptor signaling, was also cleaved to its signature active fragments during CHK1i treatment (Fig. 4B and F; Supplementary Fig. S8C). Importantly, prexasertib-induced cleavages of both CASP8 and BID were diminished by CDK2 inhibition (Fig. 4F) and by knockdown of CDK2, CCNA2, and TNFα (Fig. 4G; Supplementary Fig. S8C), placing proteolytic activation of CASP8 and BID downstream of CDK2 activation and TNFα production during prexasertib-induced killing.

Figure 4.

Contribution of BID to CHK1i-induced apoptosis. A, After cells were treated with diluent, 10 nmol/L prexasertib (Prex) or 1 μmol/L MK-8776 for 24 hours, whole-cell lysates were blotted for BCL2 family members and α-tubulin was used as loading control. B, Forty-eight hours after siRNA transfection, THP.1 whole-cell lysates were prepared for immunoblotting. C, After U937 cells were treated with diluent or 10 nmol/L prexasertib for 24 hours, cell lysates, cytosol, or mitochondria were blotted for the indicated BCL2 family members. D, Twenty-four hours after siRNA transfection, cells were treated with diluent or 10 nmol/L prexasertib for 24 hours and stained with Annexin V. Blots above bar graph show knockdown at 48 hours. E, Pooled U937 cells transduced with empty vector (EV) or two separate BID knockout clones were treated with diluent, 10 nmol/L prexasertib (24 hours), or 200 nmol/L cytarabine (48 hours) and assayed for DNA fragmentation. Baseline apoptosis was subtracted to correct for differences in apoptosis in diluent-treated cells incubated for 24 versus 48 hours. F, After U937 cells were treated for 24 hours with 10 nmol/L prexasertib ± CDK2i, whole-cell lysates were subjected to immunoblotting. G, Beginning 24 hours after transfection with control or TNFα siRNA, cells were treated with diluent or 10 nmol/L prexasertib for 24 hours and subjected to immunoblotting. Error bars in D and E, mean ± SD from three independent experiments. *, P < 0.05; **, P < 0.01.

Figure 4.

Contribution of BID to CHK1i-induced apoptosis. A, After cells were treated with diluent, 10 nmol/L prexasertib (Prex) or 1 μmol/L MK-8776 for 24 hours, whole-cell lysates were blotted for BCL2 family members and α-tubulin was used as loading control. B, Forty-eight hours after siRNA transfection, THP.1 whole-cell lysates were prepared for immunoblotting. C, After U937 cells were treated with diluent or 10 nmol/L prexasertib for 24 hours, cell lysates, cytosol, or mitochondria were blotted for the indicated BCL2 family members. D, Twenty-four hours after siRNA transfection, cells were treated with diluent or 10 nmol/L prexasertib for 24 hours and stained with Annexin V. Blots above bar graph show knockdown at 48 hours. E, Pooled U937 cells transduced with empty vector (EV) or two separate BID knockout clones were treated with diluent, 10 nmol/L prexasertib (24 hours), or 200 nmol/L cytarabine (48 hours) and assayed for DNA fragmentation. Baseline apoptosis was subtracted to correct for differences in apoptosis in diluent-treated cells incubated for 24 versus 48 hours. F, After U937 cells were treated for 24 hours with 10 nmol/L prexasertib ± CDK2i, whole-cell lysates were subjected to immunoblotting. G, Beginning 24 hours after transfection with control or TNFα siRNA, cells were treated with diluent or 10 nmol/L prexasertib for 24 hours and subjected to immunoblotting. Error bars in D and E, mean ± SD from three independent experiments. *, P < 0.05; **, P < 0.01.

Close modal

CHK1 expression as a determinant of myeloid leukemia sensitivity

Additional studies examined KG1a, K562, and the widely studied Jurkat ALL line, all of which were less sensitive to prexasertib (Fig. 1A; Supplementary Fig. S9). In these lines, CHK1 inhibition failed to induce TNFα (Supplementary Fig. S9A and S9B), BID cleavage (Supplementary Fig. S9C), and phosphatidylserine externalization (Supplementary Fig. S9D), further strengthening the association between TNFα production and cell killing. In subsequent experiments, a number of previously reported determinants of CHK1i sensitivity were examined (Supplementary Fig. S10A), including levels of the nucleases MRE11 and MUS81 as well as basal levels of phospho-H2AX and phospho-RPA as markers of preexisting replication stress. None of these putative predictors of CHK1i sensitivity correlated with response of these cell lines to prexasertib. Instead, the resistant lines had higher CHK1 expression (Supplementary Fig. S10A) and, upon CHK1 knockdown, were sensitized to CHK1i-induced apoptosis (Supplementary Fig. S10B and S10C), suggesting that CHK1 expression is an important factor in determining the response of acute leukemia to these agents.

Contribution of the CDK2–TNFα–BID pathway to killing by ATR and WEE1 inhibitors

Because CDK2 activity is also modulated by inhibitors of ATR and WEE1, we next examined cytotoxic effects of the ATRi berzosertib (VX-970) and the WEE1i adavosertib (AZD1775). ATRi-induced killing (Supplementary Fig. S11A) was accompanied by TNFα upregulation at the mRNA and protein levels (Supplementary Fig. S11B and S11C) as well as cleavage of CASP8 and BID (Supplementary Fig. S11D). WEE1i-induced apoptosis (Fig. 5A) was likewise accompanied by TNFα upregulation (Fig. 5B and C) and activating cleavages of CASP8 and BID (Fig. 5D, top). These ATRi- and WEE1i-induced changes were diminished by CDK2 inhibition (Fig. 5A–D; Supplementary Fig. S11A-S11D) as well as downregulation of CDK2 or CCNA2 (Fig. 5E). Moreover, knockdown of TNFα or TNFR1 (Fig. 5F and G; Supplementary Fig. S11E) diminished the cytotoxicity of the ATRi and WEE1i, supporting an important role for the CDK2–TNFα–BID pathway in killing by these agents as well.

Figure 5.

CDK2/TNFα/BID pathway contributes to WEE1i-induced apoptosis. A–D, U937 cells treated for 24 hours with adavosertib (AZD1775, 0.5 - 1 μmol/L) ± CDK2i were assayed for Annexin V binding (A), TNFα mRNA (B), TNFα release (C), and cleavage of caspase substrates (D). Q-VD-OPh was included in B and C. E–G, Tweny-four hours after siRNA transfection, U937 cells were treated for 24 hours with diluent or adavosertib 0.5 μmol//L (E and G) or the indicated concentrations (F) and analyzed for Annexin V binding. Insets in E, F, and G, immunoblots or qRT-PCR to assess knockdown. Error bars in A–C and E–G, mean ± SD from three independent experiments. *, P < 0.05; **, P < 0.01.

Figure 5.

CDK2/TNFα/BID pathway contributes to WEE1i-induced apoptosis. A–D, U937 cells treated for 24 hours with adavosertib (AZD1775, 0.5 - 1 μmol/L) ± CDK2i were assayed for Annexin V binding (A), TNFα mRNA (B), TNFα release (C), and cleavage of caspase substrates (D). Q-VD-OPh was included in B and C. E–G, Tweny-four hours after siRNA transfection, U937 cells were treated for 24 hours with diluent or adavosertib 0.5 μmol//L (E and G) or the indicated concentrations (F) and analyzed for Annexin V binding. Insets in E, F, and G, immunoblots or qRT-PCR to assess knockdown. Error bars in A–C and E–G, mean ± SD from three independent experiments. *, P < 0.05; **, P < 0.01.

Close modal

Drug-induced activation of TNFα transcription factors

To gain insight into the mechanism of TNFα upregulation, we further examined the RNA-seq data from U937 cells (Fig. 3A) and a similar experiment in ML-1 cells. Of 384 mRNAs that were highly altered in prexasertib-treated U937 cells, 187 were also highly altered in prexasertib-treated ML-1 cells (Supplementary Fig. S12A). Upstream regulators of those altered transcripts with predicted activation or inhibition in the dataset (|z-score| > 2) were computed for each cell line through the use of Ingenuity Pathway Analysis (https://www.qiagen.com/us/products/discovery-and-translational-research/next-generation-sequencing/informatics-and-data/interpretation-content-databases/ingenuity-pathway-analysis/). Because the TNF promoter is also activated by CHK1i (Fig. 3F), further analysis focused on changes upstream of TNF promoter activation. This analysis pointed to twelve transcription factors (TF) that are known to regulate TNF directly or indirectly (40) and could potentially contribute to the additional mRNA changes (Supplementary Fig. S12B–S12D).

When the possible role of these TFs was assessed, siRNA-induced downregulation of EGR1, NFκB1, or RELA had no impact on CHK1i-induced killing (Supplementary Fig. S12E). A different view emerged upon analysis of JUN, a TF identified as an upregulated gene in the RNA-seq analysis (Fig. 3A), and its binding partner FOS. CHK1i treatment upregulated both of these TFs in U937 cells at the mRNA and protein levels (Fig. 6A and B; Supplementary Fig. S12F and S12G). These increases reflected, in part, increased JUN and FOS phosphorylation at sites known to result in stabilization and activation (Fig. 6B). Similar changes in JUN/FOS phosphorylation and expression also occurred in CHK1i-treated ML-1 and THP.1 cells but not in the resistant lines (Fig. 6B). Phosphorylation and upregulation of JUN and FOS were also observed after CHK1 knockdown (Fig. 6C), although the degree of change was slightly lower than observed after prexasertib or MK-8776 treatment of the same cells (Supplementary Fig. S12H), possibly reflecting less complete inhibition of the pathway by knockdown versus chemical inhibition. JUN and FOS were likewise phosphorylated and upregulated after ATR or WEE1 inhibition (Supplementary Fig. S12I). Moreover, CDK2 inhibition diminished CHK1i-induced JUN and FOS upregulation (Fig. 6D), paralleling the impact of CDK2 inhibition on CHK1i-induced TNF promoter activation, TNFα expression, and killing (Figs. 2C, 3F, and 5B and C).

Figure 6.

Role of AP-1 in CDK2-mediated TNF transactivation. A, After U937 cells were treated with 10 nmol/L prexasertib (Prex) and 5 μmol/L Q-VD-OPh ± CDK2i for 24 hours, JUN and FOS mRNA were measured by qRT-PCR. B, Twenty-four hours after treatment with 10 nmol/L prexasertib or 1 μmol/L MK-8776 with 5 μmol/L Q-VD-OPh, whole-cell lysates were harvested for immunoblotting with α-tubulin as the loading control. C, After siRNA transfection, THP.1 cells were incubated with 5 μmol/L Q-VD-OPh for 48 hours and harvested for immunoblotting. D, After U937 cells were treated for 24 hours with 10 nmol/L prexasertib ± CDK2i in the presence of Q-VD-OPh, whole-cell lysates were harvested. E and F, After chromatin immunoprecipitation with the indicated antibody, samples were subjected to PCR with primers that amplify a 160-bp fragment encompassing the AP1-binding site in the TNF promoter, followed by agarose gel electrophoresis (E) or qPCR with primers that span this site (F). Control IgG served as a negative control. G, Twenty-four hours after siRNA transfection, U937 cells were treated for 24 hours with 10 nmol/L prexasertib or 1 μmol/L MK-8776 and analyzed for TNFα mRNA (top), TNFα concentration in the supernatant (middle), and Annexin V binding (bottom). Treatments in the top and middle panels included Q-VD-OPh. Inset in G, immunoblots showing knockdown efficiency. Error bars in A, F and G, mean ± SD from three independent experiments. H, Summary of steps involved in killing of AML cell lines by CHK1, ATR, or WEE1 inhibitors. Manipulations used to interrogate the pathway in this study are indicated in gray. *, P < 0.05; **, P < 0.01.

Figure 6.

Role of AP-1 in CDK2-mediated TNF transactivation. A, After U937 cells were treated with 10 nmol/L prexasertib (Prex) and 5 μmol/L Q-VD-OPh ± CDK2i for 24 hours, JUN and FOS mRNA were measured by qRT-PCR. B, Twenty-four hours after treatment with 10 nmol/L prexasertib or 1 μmol/L MK-8776 with 5 μmol/L Q-VD-OPh, whole-cell lysates were harvested for immunoblotting with α-tubulin as the loading control. C, After siRNA transfection, THP.1 cells were incubated with 5 μmol/L Q-VD-OPh for 48 hours and harvested for immunoblotting. D, After U937 cells were treated for 24 hours with 10 nmol/L prexasertib ± CDK2i in the presence of Q-VD-OPh, whole-cell lysates were harvested. E and F, After chromatin immunoprecipitation with the indicated antibody, samples were subjected to PCR with primers that amplify a 160-bp fragment encompassing the AP1-binding site in the TNF promoter, followed by agarose gel electrophoresis (E) or qPCR with primers that span this site (F). Control IgG served as a negative control. G, Twenty-four hours after siRNA transfection, U937 cells were treated for 24 hours with 10 nmol/L prexasertib or 1 μmol/L MK-8776 and analyzed for TNFα mRNA (top), TNFα concentration in the supernatant (middle), and Annexin V binding (bottom). Treatments in the top and middle panels included Q-VD-OPh. Inset in G, immunoblots showing knockdown efficiency. Error bars in A, F and G, mean ± SD from three independent experiments. H, Summary of steps involved in killing of AML cell lines by CHK1, ATR, or WEE1 inhibitors. Manipulations used to interrogate the pathway in this study are indicated in gray. *, P < 0.05; **, P < 0.01.

Close modal

If JUN and FOS contribute to CHK1i-induced TNF transactivation, then (i) these TFs should bind the TNF promoter after CHK1 inhibition and (ii) JUN or FOS downregulation should diminish CHK1i-induced TNFα induction and killing. Consistent with these predictions, chromatin immunoprecipitation demonstrated CHK1i-induced binding of JUN and FOS to the TNF promoter (Fig. 6E and F). Moreover, knockdown of either JUN or FOS diminished CHK1i-induced TNFα upregulation and killing (Fig. 6G). Time-course experiments also showed that JUN and FOS phosphorylation and upregulation preceded TNFα upregulation, which in turn preceded activating cleavages of caspase 8 and BID (Supplementary Fig. S13). Experiments designed to assess whether TNFα signaling also contributed to prexasertib action upstream of CHK1 inhibition (41) failed to show a correlation between baseline TNFα mRNA and prexasertib sensitivity (Supplementary Fig. S14A) or an impact of TNFα downregulation on CDC25A induction, CDK2 phosphorylation or JUN phosphorylation (Supplementary Fig. S14B). Collectively, these results support the signaling pathway outlined in Fig. 6H, where inhibition of CHK1, ATR, or WEE1 leads sequentially to CDK2 activation, JNK activation, FOS and JUN upregulation, FOS- and JUN-mediated TNF gene transactivation, and TNFα-mediated apoptosis.

Prexasertib-induced antileukemic effects in vivo

To rule out the possibility that the CHK1i-induced events summarized in Fig. 6H and Supplementary Fig. S13 only occur in vitro, mice bearing AML xenografts were treated with prexasertib on a previously described schedule (9, 10). Within 12–24 hours, prexasertib sequentially induced all of the changes in the xenografts that were observed in vitro, including increased expression and phosphorylation of JUN and FOS, increased TNFα mRNA, and cleavage of CASP8 and BID (Fig. 7A and B; Supplementary Fig. S15A and S15B). Moreover, prexasertib induced antiproliferative effects in a dose-dependent manner, with inhibition of xenograft growth (Supplementary Fig. S15C) and enhanced host survival (Supplementary Fig. S15D) at 10 mg/kg and marked xenograft shrinkage accompanied by lack of regrowth for up to 90 days in some engrafted animals at 15 mg/kg (Fig. 7C and D). Accordingly, events observed in vitro were also readily observed in vivo.

Figure 7.

Effect of prexasertib on acute leukemia xenografts. A and B, Mice bearing U937 flank xenografts were treated with prexasertib 10 mg/kg every 12 hours and harvested for immunoblotting at the indicated time (A) or qRT-PCR for TNFα mRNA at 12 and 24 hours (B). Our previous studies indicated that mRNA destruction in cells undergoing apoptosis in the absence of caspase inhibition (as in this in vivo experiment) can result in underestimation of the change in mRNA encoding a proapoptotic protein (28). C and D, Xenograft volumes of surviving mice (C) and survival (D) of mice randomized to diluent versus prexasertib 15 mg/kg twice daily for 3 days/week × 4 weeks, followed by observation (9). Inset in C, weights of mice in the two cohorts. E–G,BCR/ABL–positive Z181 ALL cells (42) were treated with 10 nmol/L prexasertib ± CDK2i for 24 hours in the presence (E) or absence (F and G) of Q-VD-OPh and assayed for TNFα mRNA (E), caspase-mediated cleavage of BID and PARP1 (arrows, F), and Annexin V binding (G). H, Forty days after a Ph+ ALL PDX was expanded into 8 NSGS mice, animals were randomized to diluent or prexasertib 15 mg/kg twice daily for 3 days/week × 4 weeks, followed by observation. Error bars in B, E, and G, mean ± SD from three independent experiments. Error bars in C, mean ± SD of values from surviving mice. *, P < 0.05; **, P < 0.01.

Figure 7.

Effect of prexasertib on acute leukemia xenografts. A and B, Mice bearing U937 flank xenografts were treated with prexasertib 10 mg/kg every 12 hours and harvested for immunoblotting at the indicated time (A) or qRT-PCR for TNFα mRNA at 12 and 24 hours (B). Our previous studies indicated that mRNA destruction in cells undergoing apoptosis in the absence of caspase inhibition (as in this in vivo experiment) can result in underestimation of the change in mRNA encoding a proapoptotic protein (28). C and D, Xenograft volumes of surviving mice (C) and survival (D) of mice randomized to diluent versus prexasertib 15 mg/kg twice daily for 3 days/week × 4 weeks, followed by observation (9). Inset in C, weights of mice in the two cohorts. E–G,BCR/ABL–positive Z181 ALL cells (42) were treated with 10 nmol/L prexasertib ± CDK2i for 24 hours in the presence (E) or absence (F and G) of Q-VD-OPh and assayed for TNFα mRNA (E), caspase-mediated cleavage of BID and PARP1 (arrows, F), and Annexin V binding (G). H, Forty days after a Ph+ ALL PDX was expanded into 8 NSGS mice, animals were randomized to diluent or prexasertib 15 mg/kg twice daily for 3 days/week × 4 weeks, followed by observation. Error bars in B, E, and G, mean ± SD from three independent experiments. Error bars in C, mean ± SD of values from surviving mice. *, P < 0.05; **, P < 0.01.

Close modal

To assess whether these effects were unique to AML cells, we also examined the impact of prexasertib on BCR/ABL+ acute lymphocytic leukemia (ALL). Initial studies demonstrated that prexasertib induces BID cleavage and apoptosis equally in murine Baf3 cells transformed with wild-type BCR/ABL or the TKI-resistant E225K and T315I variants (Supplementary Fig. S16A and S16B). Further studies showed that prexasertib also induces TNFα upregulation, BID cleavage, and apoptosis in a CDK2-dependent manner (Fig. 7E–G) in the BCR/ABL+ human ALL line Z181 (42), extending these results to human BCR/ABL+ ALL. Consistent with these results, prexasertib prolonged survival in a BCR/ABL+ ALL patient–derived xenograft as well (Fig. 7H).

The current studies, which utilized a series of AML cell lines as a model system to investigate the cytotoxic effects of CHK1, ATR, and WEE1 inhibitors, demonstrate that signaling through a CDK2–AP1–TNFα pathway (Fig. 6H) contributes to killing by all three classes of agents. The fact that this pathway differs from the typical DNA damage–induced pathway has potential implications for the development of these agents.

The contribution of the CDK2–AP1–TNFα pathway to the demise of susceptible leukemia cells is supported by several observations. Time-course studies demonstrate the sequential upregulation of CDC25A, CDK2 activity, JUN, FOS, and TNFα that precedes caspase 8 activation, BID cleavage, and cell death (Supplementary Fig. S13). Moreover, knockdown of critical participants, including CDK2, CCNA2, JUN, FOS, TNFα, TNFR1, and BID, inhibits subsequent drug-induced steps (summarized in Fig. 6H). Importantly, the initial steps in this pathway are distinct from the intrinsic apoptotic pathway triggered by many kinase inhibitors, BH3 mimetics, and other targeted agents (18, 43).

Although TNF upregulation has previously been widely observed after DNA damage, concomitant upregulation of BH3-only proteins such as PUMA, NOXA, and BIK has previously been thought to trigger the subsequent apoptosis (19, 44). TNFα has more recently been implicated in natural killer cell–mediated cytotoxicity (45), tuberculosis-induced programmed necrosis in macrophages (46), and BH3 mimetic–induced necroptosis (47). Here, we extend these findings by identifying TNFα production by the neoplastic cells themselves as the initiating event for induction of leukemia cell apoptosis by inhibitors of CHK1, ATR, and WEE1.

Although the TNF gene can be regulated by a number of transcription factors, several observations indicate that the TNFα induction described here is mediated by JUN and FOS. First, these TFs are upregulated by CHK1 knockdown and by treatment with CHK1, ATR, and WEE1 inhibitors, reflecting transcriptional activation as well as JNK-mediated phosphorylation during replication stress in vitro and in vivo. Second, JUN and FOS bind the TNF promoter in a CHK1i-inducible fashion. Finally, knockdown of either JUN or FOS diminishes TNFα upregulation as well as apoptosis induction by the replication checkpoint inhibitors.

The current demonstration that these checkpoint kinase inhibitors can kill acute leukemia cells through AP1-mediated TNF induction has potentially important implications. AMLs that harbor TP53 mutations respond poorly to conventional cytarabine + anthracycline therapy (48, 49). In contrast, the CDK2–AP1–TNFα pathway described here appears to be TP53 independent, as indicated by killing of AML lines that are TP53 mutant (e.g., THP.1) or null (U937) as well as TP53 wild-type (ML-1). Moreover, several TP53-mutant clinical AML specimens are among the most sensitive to CHK1is ex vivo. These results provide new mechanistic insight into the previous observation that killing by ATR or CHK1 loss is at least as effective in TP53-null cells (50).

The starting point for the current study was our observation that two- or three-fold increases in concentration changed the behavior of CHK1 inhibitors from agents that enhance cytarabine-induced killing to agents that induce cytotoxicity in AML cell lines and clinical samples as monotherapy (Supplementary Fig. S3). Although the concentrations of MK-8776 needed to induce the single-agent effects cannot be maintained for 24 hours in the clinical setting, the 10–12 nmol/L prexasertib concentrations needed for single-agent effects are far below the 50 nmol/L trough concentration observed when prexasertib is administered on a multi-day schedule in solid tumor patients (30). The single-agent activity observed in xenografts (Fig. 7; Supplementary Fig. S15) also suggests that active prexasertib concentrations can be achieved in vivo.

Previous studies of CHK1, ATR, and WEE1 inhibitors have largely focused on combining these agents with drugs that induce replication stress. In acute leukemia, this approach has led to combinations of CHK1 inhibitors with cytarabine (ClinicalTrials.gov identifier NCT01870596) or cytarabine + fludarabine (NCT02649764). The current results not only provide new insight into these drug combination studies by showing that CHK1is activate BID, which would complement the cytarabine-induced upregulation of other BH3-only family members, but also suggest that CHK1is might also have activity in combination with agents that facilitate TNFα-induced apoptosis.

This study extends current understanding of the cytotoxic effects of CHK1, ATR, and WEE1 inhibitors but also has some limitations. First, the cell line panel utilized was chosen, in part, for relatively uniform growth conditions and high transfection efficiency to permit mechanistic studies. Accordingly, this panel might not reflect the full spectrum of AML genotypes. In particular, KMT2A fusions are found in a number of the lines studied (ML-1, ML-2, Molm13, MV-4–11, and THP.1), whereas other genetic abnormalities are not represented. Importantly, however, U937 cells lack these KMT2A fusions but are still sensitive. Second, only a limited number of clinical AML samples were examined. Accordingly, further studies using genetically diverse AML lines and clinical samples are required to better determine how widespread the involvement of the CDK2–AP1–TNFα pathway is and to more fully define the range of genotypes that are susceptible to killing by achievable concentrations of CHK1, ATR, or WEE1 inhibitors. Finally, this study indicates that TNFα or TNFR1 siRNAs diminish prexasertib-induced apoptosis somewhat less than CDK2 siRNAs (mean inhibition 50% vs. 70%; Fig. 3H and I vs. Fig. 2B). Protection provided by BID knockout (Fig. 4F) is more substantial but nonetheless incomplete. These observations raise the possibility that another pathway, possibly downstream of CDK2, might also contribute to leukemia cell killing. Further studies are required to identify the second CHK1i-activated killing pathway in AML cells and its relative role in killing of other cell types.

The observation that a similar apoptotic pathway is activated by CHK1, ATR, and WEE1 inhibitors might also help guide the future development of these agents more broadly. Recent studies have focused on differences in the ways that replication checkpoint inhibitors contribute to replication stress and the possibility that preexisting replication stress might determine response to these agents. The present demonstration that a CDK2–AP1–TNFα pathway participates in killing by all of these agents does not negate these prior studies, but instead identifies common steps such as JNK activation, TNF induction, and death receptor pathway engagement where changes could potentially modulate sensitivity to multiple agents. As clinical studies of replication checkpoint inhibitors progress, particularly as monotherapy, it will be important to determine whether changes at these steps contribute to resistance and assess the possibility of cross-resistance between these classes of agents.

B.D. Koh reports other support from Gilead Sciences outside the submitted work. J.A. Webster reports personal fees from Pfizer and Amgen outside the submitted work. K.W. Pratz reports grants and personal fees from Abbvie and Agios, grants from Astellas and Millenium, personal fees from Celgene, Jazz, Boston Biomedical, and personal fees from Daiichi Sankyo outside the submitted work. L.M. Karnitz reports grants from NIH R01CA190473 during the conduct of the study. S.H. Kaufmann reports grants from NCI and grants from Eli Lilly during the conduct of the study. No disclosures were reported by the other authors.

H. Ding: Conceptualization, investigation, writing–original draft, writing–review and editing. N.D. Vincelette: Conceptualization, investigation, writing–original draft, writing–review and editing. C.D. McGehee: Conceptualization, formal analysis, writing–original draft, writing–review and editing. M.A. Kohorst: Investigation, writing–review and editing. B.D. Koh: Investigation, writing–review and editing. A. Venkatachalam: Investigation, writing–review and editing. X.W. Meng: Investigation, writing–review and editing. P.A. Schneider: Investigation, writing–review and editing. K.S. Flatten: Investigation, writing–review and editing. K.L. Peterson: Investigation, writing–review and editing. C. Correia: Data curation, formal analysis, writing–review and editing. S.-H. Lee: Investigation, writing–review and editing. M. Patnaik: Resources, writing–review and editing. J.A. Webster: Resources, writing–review and editing. G. Ghiaur: Resources, writing–review and editing. B.D. Smith: Resources, writing–review and editing. J.E. Karp: Resources, writing–review and editing. K.W. Pratz: Resources, writing–review and editing. H. Li: Conceptualization, data curation, formal analysis, supervision, writing–review and editing. L.M. Karnitz: Conceptualization, funding acquisition, writing–review and editing. S.H. Kaufmann: Conceptualization, resources, supervision, funding acquisition, investigation, writing–original draft, project administration, writing–review and editing.

This work was supported by grants from the NIH (R01 CA172503 to S.H. Kaufmann; R01 CA190473 to L.M. Karnitz), T32 GM125633 (to B.D. Koh, M.A. Kohorst), F30 CA213737 (to C.D. McGehee), a fellowship from the Mayo Foundation for Education and Research (to N.D. Vincelette), and a grant from Eli Lilly, Inc. (to S.H. Kaufmann). The authors also gratefully acknowledge encouragement of Richard Beckmann and Aimee Bence Lin; gifts of reagents or cell lines from David Toft, David Huang, Michael Kastan, Robert Abraham, Joya Chandra, Zeev Estrov, and Raoul Tibes; advice regarding ChIP assays from Luciana Almada and Martin Fernandez-Zapico; and helpful comments from the three anonymous reviewers.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Dai
Y
,
Grant
S
. 
New insights into checkpoint kinase 1 in the DNA damage response signaling network
.
Clin Cancer Res
2010
;
16
:
376
83
.
2.
Karnitz
LM
,
Zou
L
. 
Molecular pathways: targeting ATR in cancer therapy
.
Clin Cancer Res
2015
;
21
:
4780
5
.
3.
Brandsma
I
,
Fleuren
EDG
,
Williamson
CT
,
Lord
CJ
. 
Directing the use of DDR kinase inhibitors in cancer treatment
.
Expert Opin Investig Drugs
2017
;
26
:
1341
55
.
4.
Ubhi
T
,
Brown
GW
. 
Exploiting DNA replication stress for cancer treatment
.
Cancer Res
2019
;
79
:
1730
9
.
5.
Luo
Y
,
Rockow-Magnone
SK
,
Kroeger
PE
,
Frost
L
,
Chen
Z
,
Han
EK
, et al
Blocking Chk1 expression induces apoptosis and abrogates the G2 checkpoint mechanism
.
Neoplasia
2001
;
3
:
411
9
.
6.
Sakurikar
N
,
Thompson
R
,
Montano
R
,
Eastman
A
. 
A subset of cancer cell lines is acutely sensitive to the Chk1 inhibitor MK-8776 as monotherapy due to CDK2 activation in S phase
.
Oncotarget
2016
;
7
:
1380
94
.
7.
Sen
T
,
Tong
P
,
Stewart
CA
,
Cristea
S
,
Valliani
A
,
Shames
DS
, et al
CHK1 inhibition in small-cell lung cancer produces single-agent activity in biomarker-defined disease subsets and combination activity with cisplatin or olaparib
.
Cancer Res
2017
;
77
:
3870
84
.
8.
Lee
HJ
,
Cao
Y
,
Pham
V
,
Blackwood
E
,
Wilson
C
,
Evangelista
M
, et al
Ras-MEK signaling mediates a critical Chk1-dependent DNA damage response in cancer cells
.
Mol Cancer Ther
2017
;
16
:
694
704
.
9.
Lowery
CD
,
VanWye
AB
,
Dowless
M
,
Blosser
W
,
Falcon
BL
,
Stewart
J
, et al
The checkpoint kinase 1 inhibitor prexasertib induces regression of preclinical models of human neuroblastoma
.
Clin Cancer Res
2017
;
23
:
4354
63
.
10.
Lowery
CD
,
Dowless
M
,
Renschler
M
,
Blosser
W
,
VanWye
AB
,
Stephens
JR
, et al
Broad spectrum activity of the checkpoint kinase 1 inhibitor Prexasertib as a single agent or chemopotentiator across a range of preclinical pediatric tumor models
.
Clin Cancer Res
2019
;
25
:
2278
89
.
11.
Guertin
AD
,
Li
J
,
Liu
Y
,
Hurd
MS
,
Schuller
AG
,
Long
B
, et al
Preclinical evaluation of the WEE1 inhibitor MK-1775 as single-agent anticancer therapy
.
Mol Cancer Ther
2013
;
12
:
1442
52
.
12.
Wright
G
,
Golubeva
V
,
Remsing Rix
LL
,
Berndt
N
,
Luo
Y
,
Ward
GA
, et al
Dual targeting of WEE1 and PLK1 by AZD1775 elicits single agent cellular anticancer activity
.
ACS Chem Biol
2017
;
12
:
1883
92
.
13.
Kurmasheva
RT
,
Kurmashev
D
,
Reynolds
CP
,
Kang
M
,
Wu
J
,
Houghton
PJ
, et al
Initial testing (stage 1) of M6620 (formerly VX-970), a novel ATR inhibitor, alone and combined with cisplatin and melphalan, by the pediatric preclinical testing program
.
Pediatr Blood Cancer
2018
;
65
.
DOI: 10.1002/pbc.26825.
14.
Young
LA
,
O'Connor
LO
,
de Renty
C
,
Veldman-Jones
MH
,
Dorval
T
,
Wilson
Z
, et al
Differential activity of ATR and WEE1 inhibitors in a highly sensitive subpopulation of DLBCL linked to replication stress
.
Cancer Res
2019
;
79
:
3762
75
.
15.
Parmar
K
,
Kochupurakkal
BS
,
Lazaro
JB
,
Wang
ZC
,
Palakurthi
S
,
Kirschmeier
PT
, et al
The CHK1 inhibitor prexasertib exhibits monotherapy activity in high-grade serous ovarian cancer models and sensitizes to PARP inhibition
.
Clin Cancer Res
2019
;
25
:
6127
40
.
16.
Lee
JM
,
Nair
J
,
Zimmer
A
,
Lipkowitz
S
,
Annunziata
CM
,
Merino
MJ
, et al
Prexasertib, a cell cycle checkpoint kinase 1 and 2 inhibitor, in BRCA wild-type recurrent high-grade serous ovarian cancer: a first-in-class proof-of-concept phase 2 study
.
Lancet Oncol
2018
;
19
:
207
15
.
17.
Kaufmann
SH
,
Earnshaw
WC
. 
Induction of apoptosis by cancer chemotherapy
.
Exp Cell Res
2000
;
256
:
42
9
.
18.
Cragg
MS
,
Harris
C
,
Strasser
A
,
Scott
CL
. 
Unleashing the power of inhibitors of oncogenic kinases through BH3 mimetics
.
Nat Rev Cancer
2009
;
9
:
321
6
.
19.
Villunger
A
,
Michalak
EM
,
Coultas
L
,
Mullauer
F
,
Bock
G
,
Ausserlechner
MJ
, et al
p53- and drug-induced apoptotic responses mediated by BH3-only proteins puma and noxa
.
Science
2003
;
302
:
1036
8
.
20.
Petak
I
,
Tillman
DM
,
Harwood
FG
,
Mihalik
R
,
Houghton
JA
. 
Fas-dependent and -independent mechanisms of cell death following DNA damage in human colon carcinoma cells
.
Cancer Res
2000
;
60
:
2643
50
.
21.
Altucci
L
,
Rossin
A
,
Raffelsberger
W
,
Reitmair
A
,
Chomienne
C
,
Gronemeyer
H
. 
Retinoic acid-induced apoptosis in leukemia cells is mediated by paracrine action of tumor-selective death ligand TRAIL
.
Nat Med
2001
;
7
:
680
6
.
22.
Sakoe
Y
,
Sakoe
K
,
Kirito
K
,
Ozawa
K
,
Komatsu
N
. 
FOXO3A as a key molecule for all-trans retinoic acid-induced granulocytic differentiation and apoptosis in acute promyelocytic leukemia
.
Blood
2010
;
115
:
3787
95
.
23.
Taylor
RC
,
Cullen
SP
,
Martin
SJ
. 
Apoptosis: controlled demolition at the cellular level
.
Nat Rev Mol Cell Biol
2008
;
9
:
231
41
.
24.
Strasser
A
,
Cory
S
,
Adams
JM
. 
Deciphering the rules of programmed cell death to improve therapy of cancer and other diseases
.
EMBO J
2011
;
30
:
3667
83
.
25.
Sethi
G
,
Sung
B
,
Aggarwal
BB
. 
TNF: a master switch for inflammation to cancer
.
Front Biosci
2008
;
13
:
5094
107
.
26.
Eaves
C
,
Lambie
K
.
Atlas of human hematopoietic colonies
.
Vancouver, British Columbia, Canada
:
StemCell Technologies Inc.
; 
1995
.
27.
Knorr
KL
,
Finn
LE
,
Smith
BD
,
Hess
AD
,
Foran
JM
,
Karp
JE
, et al
Assessment of drug sensitivity in hematopoietic stem and progenitor cells from acute myelogenous leukemia and myelodysplastic syndrome ex vivo
.
Stem Cells Transl Med
2017
;
6
:
840
50
.
28.
Yun
S
,
Vincelette
ND
,
Knorr
KLB
,
Schneider
PA
,
Peterson
KL
,
Flatten
KS
, et al
Dual mTOR inhibitiors induce bim-dependent apoptosis in acute lymphoblastic leukmeia via inhibition of eIF4E/eIF4G and activation of NF-kB/Egr-1
.
Blood
2016
;
127
:
2711
22
.
29.
Ding
H
,
Peterson
KL
,
Correia
C
,
Koh
B
,
Schneider
PA
,
Nowakowski
GS
, et al
Histone deacetylase inhibitors interrupt HSP90*RASGRP1 and HSP90*CRAF interactions to upregulate BIM and circumvent drug resistance in lymphoma cells
.
Leukemia
2017
;
31
:
1593
602
.
30.
Hong
D
,
Infante
J
,
Janku
F
,
Jones
S
,
Nguyen
LM
,
Burris
H
, et al
Phase I study of LY2606368, a checkpoint kinase 1 inhibitor, in patients with advanced cancer
.
J Clin Oncol
2016
;
34
:
1764
71
.
31.
Guzi
TJ
,
Paruch
K
,
Dwyer
MP
,
Labroli
M
,
Shanahan
F
,
Davis
N
, et al
Targeting the replication checkpoint using SCH 900776, a potent and functionally selective CHK1 inhibitor identified via high content screening
.
Mol Cancer Ther
2011
;
10
:
591
602
.
32.
Shi
Z
,
Azuma
A
,
Sampath
D
,
Li
YX
,
Huang
P
,
Plunkett
W
. 
S-Phase arrest by nucleoside analogues and abrogation of survival without cell cycle progression by 7-hydroxystaurosporine
.
Cancer Res
2001
;
61
:
1065
72
.
33.
Schenk
EL
,
Koh
BD
,
Flatten
KS
,
Peterson
KL
,
Parry
D
,
Hess
AD
, et al
Effects of selective checkpoint kinase 1 inhibition on cytarabine cytotoxicity in acute myelogenous leukemia cells in vitro
.
Clin Cancer Res
2012
;
18
:
5364
73
.
34.
Ma
CX
,
Janetka
JW
,
Piwnica-Worms
H
. 
Death by releasing the breaks: CHK1 inhibitors as cancer therapeutics
.
Trends Mol Med
2011
;
17
:
88
96
.
35.
Smith
AJ
,
Dai
H
,
Correia
C
,
Lee
S-H
,
Takahashi
R
,
Kaufmann
SH
. 
Noxa/Bcl-2 interactions contribute to bortezomib resistance in human lymphoid cells
.
J Biol Chem
2011
;
286
:
17682
92
.
36.
Knorr
KLB
,
Schneider
PA
,
Meng
XW
,
Dai
H
,
Smith
BD
,
Hess
AD
, et al
MLN4924 induces Noxa upregulation in acute myelogenous leukemia and synergizes with Bcl-2 inhibitors
.
Cell Death Differ
2015
;
22
:
2133
42
.
37.
Sun
C
,
Fang
Y
,
Yin
J
,
Chen
J
,
Ju
Z
,
Zhang
D
, et al
Rational combination therapy with PARP and MEK inhibitors capitalizes on therapeutic liabilities in RAS mutant cancers
.
Sci Transl Med
2017
;
9
:
eaal5148
.
38.
Luo
X
,
Budihardjo
I
,
Zou
H
,
Slaughter
C
,
Wang
X
. 
Bid, a Bcl2 interacting protein, mediates cytochrome C release from mitochondria in response to activation of cell surface death receptors
.
Cell
1998
;
94
:
481
90
.
39.
Li
H
,
Zhu
H
,
Xu
CJ
,
Yuan
J
. 
Cleavage of BID by caspase 8 mediates the mitochondrial damage in the fas pathway of apoptosis
.
Cell
1998
;
94
:
491
501
.
40.
Falvo
JV
,
Tsytsykova
AV
,
Goldfeld
AE
. 
Transcriptional control of the TNF gene
.
Curr Dir Autoimmun
2010
;
11
:
27
60
.
41.
Blosser
WD
,
Dempsey
JA
,
McNulty
AM
,
Rao
X
,
Ebert
PJ
,
Lowery
CD
, et al
A pan-cancer transcriptome analysis identifies replication fork and innate immunity genes as modifiers of response to the CHK1 inhibitor prexasertib
.
Oncotarget
2020
;
11
:
216
36
.
42.
Chandra
J
,
Tracy
J
,
Loegering
D
,
Flatten
K
,
Sverstovsek
S
,
Beran
M
, et al
Adaphostin-induced oxidative stress overcomes BCR/ABL mutation-dependent and -independent imatinib resistance
.
Blood
2006
;
107
:
2501
6
.
43.
Delbridge
AR
,
Grabow
S
,
Strasser
A
,
Vaux
DL
. 
Thirty years of BCL-2: translating cell death discoveries into novel cancer therapies
.
Nat Rev Cancer
2016
;
16
:
99
109
.
44.
Vousden
KH
,
Lu
X
. 
Live or let die: the cell's response to p53
.
Nat Rev Cancer
2002
;
2
:
594
604
.
45.
Ruscetti
M
,
Leibold
J
,
Bott
MJ
,
Fennell
M
,
Kulick
A
,
Salgado
NR
, et al
NK cell-mediated cytotoxicity contributes to tumor control by a cytostatic drug combination
.
Science
2018
;
362
:
1416
22
.
46.
Roca
FJ
,
Whitworth
LJ
,
Redmond
S
,
Jones
AA
,
Ramakrishnan
L
. 
TNF induces pathogenic programmed macrophage necrosis in tuberculosis through a mitochondrial-lysosomal-endoplasmic reticulum circuit
.
Cell
2019
;
178
:
1344
61
.
47.
Giampazolias
E
,
Zunino
B
,
Dhayade
S
,
Bock
F
,
Cloix
C
,
Cao
K
, et al
Mitochondrial permeabilization engages NF-kappaB-dependent anti-tumour activity under caspase deficiency
.
Nat Cell Biol
2017
;
19
:
1116
29
.
48.
Bejar
R
,
Stevenson
K
,
Abdel-Wahab
O
,
Galili
N
,
Nilsson
B
,
Garcia-Manero
G
, et al
Clinical effect of point mutations in myelodysplastic syndromes
.
N Engl J Med
2011
;
364
:
2496
506
.
49.
Rampal
R
,
Ahn
J
,
Abdel-Wahab
O
,
Nahas
M
,
Wang
K
,
Lipson
D
, et al
Genomic and functional analysis of leukemic transformation of myeloproliferative neoplasms
.
Proc Natl Acad Sci U S A
2014
;
111
:
E5401
10
.
50.
Reinhardt
HC
,
Aslanian
AS
,
Lees
JA
,
Yaffe
MB
. 
p53-deficient cells rely on ATM- and ATR-mediated checkpoint signaling through the p38MAPK/MK2 pathway for survival after DNA damage
.
Cancer Cell
2007
;
11
:
175
89
.