Antileukemia Effects of Notch-Mediated Inhibition of Oncogenic PLK1 in B-Cell Acute Lymphoblastic Leukemia

B-cell acute lymphoblastic leukemia (B-ALL) is the most common single cancer in children, and patients of any age with relapsed B-ALL have poor outcomes (1–3). Treatment strategies and outcomes for B-ALL are partially based on the presence of high-risk genomic aberrations (2, 4). These include presence of the BCR-ABL oncogene, or the so-called Philadelphia chromosome (Ph+; refs. 1, 2). Another high-risk subgroup in both pediatric and adult populations lacks BCR-ABL, but has gene expression profiles similar to Ph+ cases, and is thus referred to as “Ph-like” (5–7). Currently, standard treatment for B-ALL invariably includes cytotoxic chemotherapies that have significant long-term side effects (8). Therefore, treatment strategies that minimize the need for chemotherapy or utilize other targeted antileukemia agents are highly desirable. Thus, the need for effective and targeted therapeutic approaches is critical for both adult and pediatric patients with B-ALL. Despite recent advances in chimeric antigen receptor T cells, CD19-CD3 bi-specifics, and CD22 antibody–drug conjugates, there remains a significant unmet medical need (9–11). We sought to use our knowledge of the apoptotic effects of Notch signaling on B-ALL cells (12, 13) to find a novel therapeutic approach that could be rapidly tested in the clinic.

Normal development of lymphoid cells is dependent on Notch signaling (14). When Notch receptors (Notch1–4) on the surface of a cell are bound by ligand, 2 successive proteolytic cleavages occur, resulting in the release of intracellular Notch (ICN1-4), which subsequently translocates to the nucleus (15). Once in the nucleus, ICN interacts with a number of cofactors to influence transcription of a variety of downstream target genes, including HES1, HEY1, and DTX1 (16). Constitutive activation of Notch signaling leads to T-ALL (17, 18). Conversely, potent activation of Notch signaling exerts proapoptotic effects in B-ALL cells (12, 13).

Inhibitors of the Notch pathway have demonstrated preclinical success and are in clinical trials for malignancies with overactive Notch signaling, such as breast, colorectal, gliomas, and T-cell malignancies (19). In contrast, options for translatable Notch agonists are extremely limited, due largely in part to lack of specificity. Therefore, we aimed to identify clinically actionable targets downstream of Notch that could mimic its proapoptotic effects in B-ALL. In this study, we identified Polo-like kinase 1 (PLK1) as a B-ALL–specific pathway that contributes to the tumor suppressive effect of Notch activation.

PLK1 is a serine/threonine kinase and negative regulator of p53 that mediates mitotic entry, spindle formation, and chromosome segregation (20, 21). Its expression is elevated in solid tumors arising from several anatomic locations, including bladder, melanoma, colorectal, esophageal, and lung (22). In a variety of malignancies, PLK1 knockdown stabilizes p53, resulting in apoptosis (23). In this study, we describe a mechanism by which Notch activation downregulates PLK1, allowing for p53-mediated cell death. Using a clinically relevant PLK1 inhibitor, we demonstrate that inhibition of PLK1 in cell lines and patient-derived xenograft (PDX) models of B-ALL mimics Notch activation, resulting in cell-cycle arrest and apoptosis. Thus, our work supports the use of PLK1 inhibitors in B-ALL.

Pertinent cell line details are summarized in Table 1. All cells were maintained in RPMI1640 medium (GIBCO) containing 10% heat-inactivated FCS (Hyclone) and 1 mmol/L HEPES, 1 mmol/L glutamine, 1 mmol/L sodium pyruvate, and 1 × nonessential amino acids (all, GIBCO). Cell samples from patients with T-ALL or B-ALL were acquired from the Leukemia Tissue Bank at The University of Texas MD Anderson Cancer Center, with approval from the MD Anderson Institutional Review Board.

HS5 cells were transduced with GFP-expressing MSCV-based retroviral plasmid MigR1 (as control) or full-length human DLL1, sorted for 100% GFP positivity and plated. B-ALL cell lines, T-ALL CCRF-CEM cells, and secondarily engrafted B-ALL cells from patients were counted and plated (100,000 cells/well) onto the GFP or DLL1-expressing HS5 feeder system. Alternatively, Notch was activated by the plate-bound DLL1 ligand in protein-G–coated 6-well plates. Plates were treated with 1% BSA (Sigma-Aldrich), washed with PBS, and incubated with 1 μg of human ligands fused to IgG Fc domains, that is, DLL1-Fc (ALX-201-765-0025; Enzo Life Sciences), DLL3-Fc, DLL4-Fc (Adipogen), Jagged-1Fc or -2Fc (1277-JG-050 and 1726-JG-050; R&D Systems) or control IgG-Fc (Jackson Laboratory) in 0.1% BSA independently, with rocking, overnight at 4°C.

Human cDNA for Notch target gene HES1 was cloned into the MigR1 retroviral vector (12, 24). PLK1 short-hairpin (sh) RNA from a TRIPZ-Inducible Transfection Starter Kit (RHS4741-EG5347; Dharmacon) was cotransfected with second-generation packaging vectors psPAX2 and pMD2 (ratio 1:1:0.5, respectively), using jetPEI transfection reagents according to the manufacturer's protocol, for 72 hours into 293T cells (provided by Dr. Faye Johnson, MD Anderson). Cells were transduced using the following method: Cells (0.1–2 × 10⁶) were plated with 250 to 500 μL of a viral supernatant and 4 to 8 μg/mL of polybrene (Sigma-Aldrich). After centrifugation at 1,000 × g for 90 minutes, cells were incubated at 37°C in 5% CO₂ for 3 to 6 hours before addition of fresh complete culture medium. Upon recovery, cells transduced with doxycycline-inducible PLK1 shRNA or nontargeted control were selected against puromycin. To induce PLK1 knockdown, B-ALL cells were exposed to doxycycline (2 μg/mL) for 2 days. Doxycycline-induced red fluorescent protein expression was confirmed by flow cytometry analysis.

Oligonucleotide small-interfering RNA (siRNA) targeting human PLK1 (sc36277) or CHFR (sc37567) and siFITC-nontargeting control scramble (sc36869) were purchased from Santa Cruz Biotechnology. Each siRNA (100 nm) was transfected into cells using Lipofectamine 2000 (Invitrogen) along with control siFITC transfection reagent according to the manufacturer's protocol. After 12 hours of siRNA transfection, FITC-positive leukemic cells were sorted for coculture.

Total RNA was isolated from the cells and reverse-transcribed using the RNeasy Mini Kit (Qiagen). Prepared RNA was primed with random hexamers to synthesize cDNA using AMV reverse-transcriptase (Amersham) according to the manufacturer's instructions. qRT-PCR was performed using human-specific Taqman probes for HES1 (Hs00172878_m1), Deltex1 (Hs00614837), HEY1 (Hs03005884), PLK1 (Hs00153444_m1), and GAPDH (4333764F); (Applied Biosystems) per the manufacturer's instructions. CHFR cDNA was amplified with the following primers (forward/reverse, 5′-3′): CAGCAGTCCAGGATTACGTGTG and AGCAGTCAGGACGGGATGTTAC (25). The qPCR reactions were performed on an iCycler (Bio-Rad). All values were normalized to the expression of the control GAPDH housekeeping gene using the 2^ΔΔC_t method.

Cocultured cells were collected and whole-cell lysates were prepared according to established methods utilizing Triton X-100 or RIPA lysis buffer and 1 mmol/L PMSF supplemented with protease inhibitor cocktail (Calbiochem) and phosphatase inhibitor cocktails I and II (Sigma-Aldrich). BCA Protein Assay Kit (Thermo Fisher Scientific) was used to determine protein concentrations. HRP-linked secondary antibodies against rabbit, mouse, and goat IgGs were purchased from Amersham-GE Life Sciences. Fifty micrograms of protein were loaded for Western analyses as described previously (26).

For immunoprecipitation of PLK1, CHFR, MDM2, pMDM2, and p53, cells were subjected to lysis with 400 μL of HEPES lysis buffer (pH 7.5) containing 50 mmol/L HEPES, 150 mmol/L NaCl, 1 mmol/L EGTA, 10 mmol/L Na pyrophosphate, 10 mmol/L NaF, 10% glycerol, 1% Triton X-100, and 1.5 mmol/L MgCl₂ at 4°C for 30 minutes. Whole-cell lysates were subjected to centrifugation at 15,000 × g for 20 minutes at 4°C and diluted in RIPA immunoprecipitation buffer (500 μL) containing 50 mm Tris-HCl (pH 7.5), 150 mm NaCl, 0.5% Nonidet P-40, 0.1% SDS, and 0.1% sodium deoxycholate with 1 × protease inhibitor mixture (Roche Applied Science) and were precleared with 30 μL of a 50% slurry of protein A-Sepharose (Amersham). Supernatants were incubated with normal mouse serum and appropriate antibodies at 4°C for 3 hours. Immunocomplexes were incubated with protein G–coated magnetic beads (53014; Active Motif) for 45 minutes at 4°C. Immunoprecipitation and wash steps were performed by pelleting the magnetic beads on the side of the tube using a magnetic bar for the removal of supernatant. The precipitates were washed 3 times with the lysis buffer at 4°C, resuspended in 30 μL of the SDS sample buffer, and heated at 100°C for 5 minutes. Proteins were then resolved by 10% SDS-PAGE and transferred onto Immobilon-P membranes for Western blot analysis using the specified antibodies.

Antibodies β-actin (A2066; Sigma-Aldrich), cleaved Notch2 (C651.6DbHN; Developmental Studies Hybridoma Bank, University of Iowa, Iowa City, IA), HES1 (ab71559; Abcam; TA500014; Origene), Notch-1, -3, and -4 (ab52627, ab23426, and ab184742; Abcam), Jagged1 (sc11376; Santa Cruz Biotechnology), DLL1 (sc73899; Santa Cruz Biotechnology), p53 (DO1; Santa Cruz Biotechnology), FITC anti-p53 antibody (645803; BioLegend), PLK1 (ab17057; Abcam), CHFR (H00055743-M01; Abnova), PAR (4335-MC-100-AC; Trevigen), MDM2 (OP115; Millipore), pMDM2(ser²⁶⁰) (orb129684; Biorbyt LLC; Bioworld Technology, Inc.), phycoerythrin-labeled donkey anti-rabbit IgG antibody (406421; BioLegend), and anti-ubiquitin (U5379; Sigma-Aldrich) were used. PARP1 inhibitor 3ABA (300 μmol/L; Sigma-Aldrich; ref. 27), DMSO, Notch inhibitor N-[N-(3,5-difluorophenacetyl)-l-alanyl]-S-phenylglycine t-butyl ester (DAPT; Sigma-Aldrich), and PLK1 inhibitors BI6727 (volasertib; ref. 28) and BI2536 (29) (Selleck Chemicals) were used.

B-ALL patient cells were injected into the tail vein of 5- to 6-week-old NSG-SGM3 mice (500,000 per mouse) after sublethal total body irradiation (800 rads). Peripheral blood was drawn from the tail vein at relevant time points and monitored for engraftment via flow cytometry for human CD45 and CD19 expression. When circulating leukemic cells were observed (3 weeks from initial injection), engrafted mice were treated twice weekly for 2 weeks with DMSO or PLK1 inhibitor BI2536 or BI6727. Peripheral blood leukemia burden was monitored beginning with the first treatment, weekly for 4 weeks. At the end of the 4-week monitoring period, mice were euthanized and peripheral blood, femurs, and tibias harvested. Animals were maintained and all animal experiments were performed under Institutional Animal Care and Use Committee with prior approval from the MD Anderson in compliance with the US Department of Health and Human Services guidelines.

Cells were stained with antibodies against human PLK1, total p53, and pMDM2 were quantified by flow cytometry (30). For intracellular staining, cells were fixed and permeabilized with BD Cytofix/Cytoperm (554714), stained with antibody for 40 minutes at room temperature with gentle shacking, and run on a FACSCalibur. Annexin V staining was performed using standard methods (550475; BD Biosciences). FlowJo software (FlowJo LLC) was used for analysis.

Metal-tagged antibodies were provided by the MD Anderson Flow Cytometry Core, and samples were stained as described previously (31, 32), and run on a DVS Sciences CyTOF mass cytometer (Fluidigm). Samples were analyzed using Spanning Tree Progression of Density Normalized Events (SPADE) algorithm as described elsewhere (33).

mRNA expression levels in samples from 206 children with ALL (34 T-ALL, 172 B-ALL) were measured on Affymetrix U133A arrays. To identify significant differences, we used linear models for microarray analysis (LIMMA), the empirical Bayes t test implemented in Bioconductor, and the Benjamini–Hochberg method of false discovery rate estimation. Data were deposited at the Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo/) under accession no. GSE26366.

Differences between treatment groups were analyzed by the Student t test or multiple measures ANOVA to assess statistical significance. All in vitro experiments were performed in triplicate unless otherwise noted.

Previous work from our laboratory (12, 13) demonstrated that Notch signaling results in growth arrest and apoptosis in B-cell malignancies. To better understand the expression of Notch receptors on B-ALL, we evaluated surface expression of Notch1 to Notch4 in patient samples with various Ph status by flow cytometry. Expression of Notch2 was consistently highest, whereas the other receptors were present at lower and variable levels (Fig. 1A). Importantly, Notch receptors were present in both Ph-negative and Ph-like B-ALL samples (Fig. 1A).

Given the presence of all Notch receptors and the variety in receptor distribution, we next sought to determine whether cell death would be preferentially induced by individual Notch ligands. Plate-bound DLL1-Fc, Jagged1-Fc, or DLL4-Fc were all able to decrease viability in patient with B-ALL samples (Fig. 1B). In contrast, culture with DLL3-Fc or Jagged2-Fc had limited impact on cell viability. Notably, no ligand induced B-ALL cell growth (Fig. 1B). Although several Notch ligands could induce B-ALL cell death, we selected the most potent ligand, DLL1, as our model ligand. Annexin V positivity was observed in a dose-dependent manner when patient samples were cultured with increasing concentrations of DLL1-Fc (Supplementary Fig. S1A). Similarly, B-ALL cell lines and patient samples cocultured with human stromal cells overexpressing DLL1 (HS5-DLL1) exhibited decreased growth (Fig. 1C) and increased apoptosis (Fig. 1D) compared with those cultured with an HS5-GFP control. In line with previous work, this effect was not seen in a T-ALL cell line (Fig. 1C and D). In support of active Notch signaling, B-ALL cells cultured on HS5-DLL1 demonstrated a consistent increase in Notch downstream target genes HES1, HEY1, and DTX1 mRNA (Fig. 1E) and in HES1 protein (Fig. 1F) when compared with control cocultures. The addition of a DLL1-blocking antibody (xDLL1) to coculture mitigated the induction of HES1 protein expression (Fig. 1F), suggesting that the additional HES1 expression was due to the additional DLL1 ligand on these HS5 cells (HS5-DLL1). Induction of HES1 expression was seen across B-ALL cell lines and patient samples (Supplementary Fig. S1B). Together, these results illustrate that ligand-mediated Notch activation is capable of suppressing growth and survival of B-ALL.

Previous work showed that B-ALL cell death was induced by expression of the intracellular domain of any of the 4 Notch receptors (12). Here, we demonstrate that at least 3 recombinant Notch ligands induced significant B-ALL cell death (Fig. 1B). These observations suggest that B-ALL cells apoptose via a common Notch-mediated mechanism, which is potentially shared to different degrees by many Notch ligand–receptor pairs. Given the challenge of translating Notch-activating therapeutics, we sought to target the downstream mechanism of Notch activation in B-ALL cells. PLK1, a putatively oncogenic kinase, is stabilized by a negative regulator of Notch in melanoma (34). Hematologic malignancies, including ALL, reportedly express high levels of PLK1 (35, 36). We thus chose to examine whether the inverse was true—namely, if Notch activation was a negative regulator of PLK1 expression in B-ALL cells, ultimately leading to apoptosis.

Flow cytometry demonstrated that our B-ALL cell lines and patient samples did express PLK1 (Fig. 2A and C). Although expression was observed across B-ALL samples of various Ph-status, PLK1 was highest in Ph-like B-ALLs (Fig. 2B and D). Upon coculture with HS5-DLL1 cells, B-ALL cell lines and patient samples demonstrated marked reduction in the expression of PLK1 protein in the context of activation of Notch signaling, supported by the appearance of cleaved intracellular Notch2 (ICN2) and the common Notch target gene HES1 (Fig. 2E and G). Notably, Ph-like samples had the greatest reduction in PLK1 in response to DLL1-Notch activation (Fig. 2D). Expression of the intracellular domain of any of the 4 Notch receptors was capable of decreasing PLK1 expression (Fig. 2F), although ICN2 and ICN1 elicited the greatest reductions, and ICN4 only had a marginal decrease. These results are consistent with prior work demonstrating that expression of ICN1-4 could induce B-ALL cell death (12), and the activating potencies of Notch receptors, that is, ICN1 = ICN2 > ICN3 > ICN4.

Importantly, overexpression of HES1, the most commonly reported Notch downstream target gene, was sufficient to decrease PLK1 protein expression in B-ALL cell lines, but not in T-ALL cell lines (Fig. 2G), suggesting that the Notch/HES-mediated downregulation of PLK1 occurs in a B-ALL–specific manner.

Given these observations that Notch activation downregulates PLK1 expression, we sought to determine whether direct downregulation of PLK1 expression was sufficient to inhibit growth and survival in patient B-ALL cells. B-ALL with shPLK1 exhibited a significant decrease in viability, but this effect was not seen in T-ALL (Fig. 3A and B). This reduction in cell viability with partial PLK1 downregulation suggests the importance of PLK1 in the survival of B-ALL cells.

In other malignancies, PLK1 is proposed to exert oncogenic functions by suppressing the proapoptotic function of p53 (20, 37). In contrast, p53 can repress transcription of PLK1 in response to DNA damage (38). To explore this inverse relationship between PLK1 and p53, we evaluated p53 expression in B-ALL samples following PLK1 knockdown. p53 expression was significantly increased in shPLK1 cells compared with control, and this effect was not seen in T-ALL cells (Fig. 3C). Together, this demonstrates that PLK1 is required for survival in B-ALL, and may exert its effects in part through suppression of p53.

PLK1 has been shown to modulate p53 stability and function via phosphorylation of MDM2 at Ser260, resulting in p53 degradation (21). To evaluate whether these PLK1–p53 relationships are maintained in B-ALL, we tested the effect of activated Notch signaling on pMDM2^Ser260 and p53 levels in B-ALL cell lines and patient with B-ALL samples. Notch ligand-mediated activation dramatically decreased pMDM2^Ser260 without altering total MDM2 levels, and markedly increased p53 protein levels (Fig. 4A). Importantly, inhibition of Notch receptor cleavage with DAPT, a γ-secretase inhibitor that acts as a nonspecific inhibitor of Notch activation, could abrogate the majority of the p53 accumulation (Fig. 4B) and attenuate HES1 induction in the context of Notch ligand exposure (Fig. 4C), supporting the hypothesis that this mechanism is driven by canonical Notch signaling.

Adding further to this mechanism, coimmunoprecipitation assays in a B-ALL patient sample demonstrated loss of endogenous MDM2–p53 interaction following Notch activation (Fig. 4D). Likewise, flow cytometry revealed a decrease in pMDM2^Ser260 and a corresponding increase in p53 staining intensity in these patient with B-ALL samples (Fig. 4E and F). Together, these results demonstrate that Notch-mediated PLK1 downregulation leads to a significant reduction of pMDM2^Ser260 and subsequent p53 stabilization. This regulation of the MDM2:p53 interaction through PLK1-mediated phosphorylation of MDM2 at Ser260 is a novel mechanism in B-ALL and provides insight into Notch-mediated effects in B-ALL while also highlighting PLK1 as a potential therapeutic target in B-ALL.

To further understand the mechanism of Notch/HES-mediated downregulation of PLK1, we examined PLK1 mRNA expression following exposure to Notch ligand and found no significant differences (Fig. 5A), suggesting that Notch-mediated regulation of PLK1 protein levels occurs posttranslationally. Because PLK1 is known to be ubiquitinated and subsequently degraded by the proteasome (39, 40), we tested whether PLK1 protein was ubiquitinated following Notch activation. Following Notch activation, PLK1 was ubiquitinated in B-ALL cells, with ubiquitination seen at 24 and 48 hours (Fig. 5B and C). PLK1 was ubiquitinated in the same cells following HES1 overexpression (Fig. 5D), confirming the role for Notch signaling in the ubiquitination of PLK1.

Because the E3 ligase CHFR has been reported to target PLK1 for ubiquitination (40), we sought to investigate whether CHFR was important for PLK1 ubiquitination in B-ALL. Indeed, knockdown of CHFR impaired HES1-mediated loss of PLK1 expression (Fig. 5E), linking HES1 to CHFR and PLK1. We then confirmed an interaction between PLK1 and CHFR via immunoprecipitation of endogenous proteins in B-ALL cell lines (Fig. 5F). To determine whether CHFR was necessary for the Notch-mediated growth inhibition in B-ALL, we partially knocked down CHFR via siRNA and observed partial rescue of Notch ligand-mediated reductions in B-ALL cell counts, but not in a T-ALL cell line (Fig. 5G), suggesting an important role for CHFR in this mechanism. In line with the notion that this cell death is due, at least in part, to p53 stabilization, we sought to understand the role of CHFR in p53 stabilization in B-ALL. Indeed, CHFR knockdown prevented the loss of pMDM2^Ser260 and the stabilization of p53 in patient with B-ALL cells (Fig. 5H).

Our previous work demonstrated that HES1 activates PARP1, resulting in global poly-ADP-ribosylation (PARylation) in B-ALL (13). Following exposure to DLL1, CHFR was found to be PARylated in B-ALL cells (Fig. 5I). Addition of a PARP1 inhibitor (3ABA) rescued the majority of Notch ligand-mediated growth inhibition in B-ALL samples (Fig. 5J), underscoring the importance of CHFR and PARP activity in this mechanism of Notch-mediated growth inhibition.

The experiments described above reveal a novel mechanism of Notch/HES-mediated PARP/CHFR-mediated PLK1 loss and subsequent dysregulation of MDM2/p53, leading to growth arrest and apoptosis in B-ALL cells. However, as Notch ligands/agonists are a challenge to translate into a clinical therapeutic, the remainder of the manuscript focuses on the use of PLK1 inhibitors as a clinically available method to mimic Notch-mediated apoptosis in B-ALL.

The PLK1 inhibitor BI2536 dramatically decreased cell growth of B-ALL cell lines and patient samples, but had little effect on a T-ALL cell line (Fig. 6A). Consistent with the Notch/HES-mediated mechanism revealed above, exposure to BI2536 resulted in a decrease in pMDM2^Ser260 and subsequent p53 stabilization in B-ALL, but not T-ALL (Fig. 6B and C). These results were further corroborated by Western blot analysis (Fig. 6D). These data confirm the importance of this PLK1/MDM2/p53 mechanism in B-ALL and demonstrate the therapeutic potential of PLK1 inhibition in this disease.

To evaluate whether PLK1 inhibition reduces leukemia burden in vivo, we treated a Ph-like B-ALL PDX mouse model with the PLK1 inhibitors BI2536 or BI6727 (volasertib). Strikingly, mice that received either PLK1 inhibitor had leukemic burden reductions of 70% in blood (Fig. 6E), 90% in bone marrow, and 95% in spleen compared with mice receiving vehicle (Fig. 6F). Analysis of bone marrow cells isolated from mice receiving BI2636 or BI6727 revealed a decrease in pMDM2^Ser260 with concomitant p53 stabilization compared with vehicle-treated mice (Fig. 6G). In addition to increased p53, mice that received BI6727 also had increased expression of the proapoptotic protein BAX in leukemia blasts, suggesting a further downstream mechanism of apoptosis (S2a-f). These data support the existence of a PLK1–MDM2–p53 mechanism in B-ALL and demonstrate the potential therapeutic efficacy of direct PLK1 inhibition in B-ALL.

Although PLK1 inhibition appears to have marked antileukemic effects in B-ALL, single-agent therapies are only rarely utilized clinically. Thus, we evaluated whether addition of a PLK1 inhibitor would confer any additional benefit to standardly used chemotherapeutic agents. When combined with dexamethasone or vincristine, a low dose of BI2536 (5-fold lower than previous in vitro experiments) significantly reduced cell viability compared with any of the 3 agents alone (Fig. 6H). These suggest that the antileukemia effects of dexamethasone and vincristine can be potentiated by PLK1 inhibition.

Our data suggest an important role for Notch-mediated PLK1 inhibition, occurring via PARylation of CHFR, hypophosphorylation of MDM2^Ser260, stabilization of p53, and upregulation of proapoptotic proteins such as BAX. To evaluate whether this mechanism could be reasonably generalized across a larger subset of B-ALL, we analyzed data from 172 bone marrow samples from B-ALL cases (representing all major biologic subtypes) and 34 T-ALL cases compared with 9 leukemia-free donors. Supplementary Figure S3 displays a heat map of the relative mRNA expression of the genes involved in this mechanism. Of the Notch receptors, transcript levels of NOTCH1 and NOTCH2 were highest, consistent with their relatively abundant surface expression (Fig. 1A). HES1 levels were elevated in T-ALL, but were uniformly low in B-ALL, consistent with the constitutive Notch activation seen in many T-ALL cases, with low HES1 protein (Fig. 1E and F) and with our prior findings of a lack of constitutive Notch signaling in B-ALL (12, 13). CHFR and PLK1 were expressed at similar levels in B-ALL and T-ALL, although these transcript expression array data do not reflect the posttranslational regulation described here. Likewise, TP53 levels appear similar between B-ALL and T-ALL. Importantly, both PARP1 and MDM2 transcripts were significantly higher in B-ALL than T-ALL, potentially revealing why B-ALL is sensitive to the PLK1–MDM2–p53 mechanism described here.

This study demonstrates that Notch ligand-mediated Notch activation induces B-ALL–specific growth inhibition and apoptosis, in part through PLK1 reduction. Our data demonstrate that PLK1 is ubiquitinated by CHFR, which is PARylated by PARP1 in response to Notch activation via HES1 expression. The decrease in PLK1 results in MDM2 hypophosphorylation, p53 stabilization, and upregulation of proapoptotic proteins (BAX).

It appears that multiple Notch ligands and receptors may contribute to this mechanism. DLL1 was used as the model ligand in these studies, although multiple Notch ligands were able to induce apoptosis (Fig. 1B), suggesting that this mechanism is not specific to a single ligand. Although patient samples had variable distribution of Notch receptors (Fig. 1A; Supplementary Fig. S3), each sample tested exhibited PLK1 downregulation and ultimately cell death in response to Notch activation. Likewise, expression of ICN1-4 or HES1 phenocopied Notch-ligand–mediated PLK1 downregulation (Fig. 2F and G), suggesting that any Notch receptor can induce this mechanism through HES gene upregulation. Finally, DAPT abrogated the effects of Notch activation on PLK1 downregulation in B-ALL (Fig. 4C and D) suggesting that this mechanism relies on cleavage of the Notch receptors and canonical Notch signaling. Together, these observations allude to a general Notch-HES pathway effect, that is not exclusive to one ligand or receptor. Deciphering relative contributions of each Notch receptor or ligand in different patient samples may be helpful in further understanding the physiologic effects of Notch signaling in B-ALL. However, by therapeutically targeting the downstream mechanism via direct PLK inhibition, the contribution of Notch pathway becomes less relevant for the clinic.

In this study, PLK1 inhibitor dramatically potentiated the effects of both vincristine and dexamethasone (Fig. 6H). These may suggest that addition of a PLK1-targeted agent may reduce the exposure to chemotherapy necessary to achieve clinical response. To this end, PLK1 inhibitors, including volasertib (BI6727), have shown acceptable toxicity profiles and have had modest clinical response in combination with other agents for advanced solid tumors and acute myeloid leukemia (41–43). Although PLK1 inhibition has not been used clinically in B-ALL, accumulating evidence supports a role for PLK1 in B-cell oncogenesis. In B-cell lymphoma, overexpression of PLK1 has been described as an independent prognostic factor (44). A recent publication has demonstrated a pivotal role for PLK1 in double-hit diffuse large B-cell lymphoma, and proposed combination therapy of PLK1 inhibition and venetoclax, a BCL2 inhibitor, in this disease (45). This aligns nicely with our current study, in which PLK1 inhibition upregulated p53 and the proapoptotic BAX protein, ultimately resulting in B-ALL growth arrest and apoptosis. Future studies using these combinations in B-ALL would certainly be of use given the widespread use and clinical efficacy of venetoclax (46, 47).

In this study, we utilized samples with wild-type p53. The majority of pediatric B-ALL presents with wild-type p53, although relapse is associated with a higher incidence of TP53 mutations (48). In adult B-ALL, p53 is mutated in ∼15% of newly diagnosed cases (49–51). More frequently, other mechanisms of p53 inactivation—such as deregulation of MDM2 or ARF—occur (52, 53). This is also evident in our analysis of patient samples, where MDM2 levels were elevated (Supplementary Fig. S3). Hypophosphorylation of MDM2 by Notch-HES1–mediated PLK1 reduction appears to act as a means to reactivate wild-type p53, which results in upregulation of proapoptotic proteins, such as BAX (Supplementary Fig. S2). Future studies evaluating whether this mechanism remains intact when p53 is mutated in B-ALL would be worth exploring.

Previously, we showed that HES1 interacts with and activates PARP1 in B-ALL, resulting in nuclear translocation of apoptosis inducing factor (AIF), caspase cleavage, and apoptosis of B-ALL cells (13). In this study, we expand on these findings by further relating PARP1 to PARylation and activation of CHFR, which is required for the ubiquitination and subsequent degradation of PLK1 in B-ALL. These mechanisms are likely to work in concert with one another, which may explain why knockdown of CHFR only partially rescues DLL1-mediated reductions in B-ALL cell counts (Fig. 5G). Therefore, a therapeutic means to selectively activate Notch in B-ALL cells may ultimately be more beneficial than PLK1 inhibition in isolation. The antileukemic activity of Notch signaling in B-ALL resides in striking contrast to the oncogenic Notch activation seen in various malignancies, including T-ALL (54, 55). This reasonably causes hesitation regarding the notion of developing Notch-activating therapeutics to treat B-ALL. Should such a venture be evaluated, ensuring specificity to B-lineage cells via conjugation of Notch ligand to an antibody raised against an antigen such as CD19, CD20, or CD22 may be useful.

Notch-activation induced B-ALL apoptosis appears to be a phenomenon observed across B-ALL subtypes (Fig. 2). Intriguingly, this effect is particularly pronounced in Ph-like samples, representing a high-risk phenotype (Fig. 2D). Likewise, Ph-like cell lines and patient samples had the highest PLK1 expression (Fig. 2C), which may brand this subtype as particularly amenable to PLK1 inhibition. These compelling findings require further investigation.

In summary, this study demonstrated a role for active Notch-HES signaling in downregulating PLK1 in a PARP1/CHFR-dependent manner. This led to hypophosphorylation of MDM2, stabilization of p53, and upregulation of the proapoptotic protein BAX, ultimately resulting in B-ALL cell death. PLK1 inhibitors had marked proapoptotic effects in vitro and in PDX models of B-ALL. Together, these data strongly suggest that Notch activators or PLK1 inhibitors are viable therapeutic approaches in B-ALL that warrant clinical investigation.

M. Konopleva has ownership interest (including stock, patents, etc.) in Reata Pharmaceutical, and is a consultant/advisory board member for Abbvie, Genentech, F. Hoffman LaRoche, Cellectis, and Stemline Therapeutics. C.G. Mullighan reports of receiving Loxo Oncoloy, Pfizer, and Abbvie; has received speakers bureau honoraria from Amgen, Pfizer, and Illumina; and is a consultant/advisory board member for St. Justine CHU. P.A. Zweidler-McKay is a Senior Medical Director at and has ownership interest (including stock, patents, etc.) in ImmunoGen. No potential conflicts of interest were disclosed by the other authors.

Conception and design: S. Kannan, M.J.L. Aitken, S.M. Herbrich, J. Chandra, P.A. Zweidler-McKay

Development of methodology: S. Kannan, L.S. Golfman, J.K. Burks, P.A. Zweidler-McKay

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Kannan, D.H. Mak, J.K. Burks, M. Konopleva, C.G. Mullighan, P.A. Zweidler-McKay

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S. Kannan, M.J.L. Aitken, S.M. Herbrich, M.G. Hall, J.K. Burks, G. Song, C.G. Mullighan, J. Chandra, P.A. Zweidler-McKay

Writing, review, and/or revision of the manuscript: S. Kannan, M.J.L. Aitken, S.M. Herbrich, J.K. Burks, C.G. Mullighan, J. Chandra, P.A. Zweidler-McKay

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S. Kannan, S.M. Herbrich

Study supervision: S. Kannan, J. Chandra, P.A. Zweidler-McKay

Other (assisted with blood and tissue sampling and flow analysis.): L.S. Golfman

Other (secured funding for the study.): J. Chandra

This work was supported by the NCI (R01CA138816, to P.A. Zweidler-McKay) and by a research grant from Alex's Lemonade Stand (ALSF ID# 4497, to J. Chandra). Additional funding support was provided by Cancer Prevention & Research Institute of Texas (CPRIT: RP150006, to M. Konopleva) and by a Laboratory Incentive Fund from the Division of Pediatrics (to S. Kannan), and additional resources from Richard Gorlick, MD. Thanks to Dr. Michael Roth (Division of Pediatrics) and the Department of Scientific Publications (MD Anderson Cancer Center) for reviewing the manuscript. The authors are grateful to Drs. Michelle Barton, Dean of the University of Texas Graduate School of Biomedical Sciences, and Faye M. Johnson, Associate Professor of Thoracic/Head and Neck Medical Oncology at MD Anderson Cancer Center (Houston, TX), for helpful discussion about the manuscript and providing MDM2 and p53 reagents. The authors also are thankful to the Cellular Imaging Core, Department of Leukemia, for their valuable support in analyzing the CyTOF samples with the support of the NIH/NCI through the MD Anderson Cancer Center Support Grant (CCSG) under award number P30CA016672.

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.