Quantitative phosphoproteomics uncovers synergy between DNA-PK and FLT3 inhibitors in acute myeloid leukaemia

Acute myeloid leukaemia (AML) is the most common and aggressive form of acute leukaemia in adults. The most common driver mutations in AML are activating mutations in the FMS-like tyrosine kinase 3 (FLT3) gene, occurring in approximately one third of cases [1]. Internal tandem duplication (FLT3-ITD) mutations in the FLT3 juxtamembrane domain are the most common (~25%), and are associated with genomic instability, and intermediateadverse prognosis [2]. Point mutations in the FLT3 activation loop, most commonly at amino acid D835, occur in ~8% of AML patients and their prognostic effect remains to be defined. The landscape of treatment options for AML is rapidly changing, however there remains limited durable treatment options for molecular subtypes such as mutant-FLT3 AML. To identify novel therapeutic targets, we undertook quantitative phosphoproteomic profiling of primary AML blasts (Supplementary Methods) [3]. Across seven patient samples (3 × wildtype-FLT3, 4 × mutant-FLT3; Tables S1 and S2), 4345 unique phosphosites were identified from 1994 proteins, with expected ratios of serine, threonine, and tyrosine sites [pS:pT:pY 88.9%:10.6%:0.6%]. The top pathways phosphorylated in AML blasts were growth and survival signalling pathways (ERK/MAPK, AMPK signalling, Fig. S1) and DNA damage repair signalling pathways (ATM signalling, DNA double strand break (DSB) repair, Fig. S1). Kinase substrate enrichment analysis (KSEA) revealed activation of the cell cycle and apoptosis regulator, Casein Kinase 2 (CK2-A1), in all blast samples (Figs. 1a and S2). Increased CK2-A1 expression has been demonstrated in a range of haematological cancers, including AML; driving cell proliferation, survival, and drug resistance [4]. Further These authors contributed equally: Nicole M. Verrills, Matthew D. Dun


To the Editor:
Acute myeloid leukaemia (AML) is the most common and aggressive form of acute leukaemia in adults. The most common driver mutations in AML are activating mutations in the FMS-like tyrosine kinase 3 (FLT3) gene, occurring in approximately one third of cases [1]. Internal tandem duplication (FLT3-ITD) mutations in the FLT3 juxtamembrane domain are the most common (~25%), and are associated with genomic instability, and intermediateadverse prognosis [2]. Point mutations in the FLT3 activation loop, most commonly at amino acid D835, occur iñ 8% of AML patients and their prognostic effect remains to be defined.
The landscape of treatment options for AML is rapidly changing, however there remains limited durable treatment options for molecular subtypes such as mutant-FLT3 AML. To identify novel therapeutic targets, we undertook quantitative phosphoproteomic profiling of primary AML blasts (Supplementary Methods) [3]. Across seven patient samples (3 × wildtype-FLT3, 4 × mutant-FLT3; Tables S1 and S2), 4345 unique phosphosites were identified from 1994 proteins, with expected ratios of serine, threonine, and tyrosine sites [pS:pT:pY 88.9%:10.6%:0.6%]. The top pathways phosphorylated in AML blasts were growth and survival signalling pathways (ERK/MAPK, AMPK signalling, Fig.  S1) and DNA damage repair signalling pathways (ATM signalling, DNA double strand break (DSB) repair, Fig. S1). Kinase substrate enrichment analysis (KSEA) revealed activation of the cell cycle and apoptosis regulator, Casein Kinase 2 (CK2-A1), in all blast samples (Figs. 1a and S2). Increased CK2-A1 expression has been demonstrated in a range of haematological cancers, including AML; driving cell proliferation, survival, and drug resistance [4]. Further validating our approach, Glycogen Synthase Kinase 3β (GSK3β), a downstream regulator of mutant-FLT3 proliferative signalling [5], and Cyclin-Dependent Kinase 5 (CDK5), an AML drug target [6], showed activation in mutant-FLT3 samples (Fig. 1a). The serine/threonine protein kinase KIS, a regulator of proliferation in leukaemia cells [7]; and the DSB repair protein kinases DNAdependent Protein Kinase (DNA-PK) and Ataxia Telangiectasia-Mutated (ATM), were also activated in the majority of samples (Figs. 1a and S2). In mutant-FLT3 samples compared to wildtype-FLT3, 143 peptides displayed a significant twofold increase and 90 peptides displayed a significant twofold decrease (Fig. S3). Pathway enrichment analysis revealed increased phosphorylation of proteins involved in the error-prone DNA-PK-dependent Non Homologous End Joining (NHEJ) pathway in mutant-FLT3 AML patient samples, compared to wildtype-FLT3 patients (Figs. 1b, S1B, and S4). This included core NHEJ proteins DNA-PKcs (PRKDC), XRCC5, XRCC4, and 53BP1 (Figs. 1b and S4), suggesting NHEJ pathway activation in mutant-FLT3 samples. In support of this, mutant-FLT3 patient samples displayed increased phosphorylation of the DNA-PK activating autophosphorylation site, S2612 (p = 0.047, Fig. 1b), analogous to previous results in FLT3-, NRAS-, and BRAF-mutant AML [8]. In addition, phosphorylation of Base Excision Repair (BER) pathway µ µ µ µ µ Fig. 1 Quantitative phosphoproteomic profiling of human AML blasts identifies phosphorylation of DNA repair, and growth and survival signalling pathways. The phosphoproteome of seven human AML blast samples was quantified by iTRAQ mass spectrometry. a Kinase activity profile of AML blast samples was determined by individual kinase substrate enrichment analysis (KSEA), performed on each sample separately using the mass spectrometry mediannormalised data. Colour scale indicates PHOXTRACK kinase enrichment score, with a positive value predictive of kinase activation and a negative value predictive of inhibited kinase activity. b DNA repair pathways displayed altered phosphorylation in mutant-FLT3 AML blasts, with increased phosphorylation of proteins (yellow) within the Non Homologous End Joining pathway and decreased phosphorylation (blue) within the Base Excision Repair pathway analysed using 2-way ANOVA p = 0.0397 and p = 0.0436, respectively. Phosphosites with a greater than 2-fold change in abundance are shown. Darker shading indicates individual phosphorylation sites that are statistically significantly different between wildtype and mutant-FLT3 AML samples, as analysed using one-tailed test *p < 0.05. c DNA-PKcs phosphorylation levels were assessed by targeted mass spectrometry in cell lines treated for 1 h with DNA-PK inhibitors (NU7441, M3814), FLT3 inhibitors (sorafenib, midostaurin, AC220), or their combination, as indicated. *p < 0.05, **p < 0.01, n = 3. proteins were decreased in mutant-FLT3 compared to wildtype-FLT3 AML patients (Figs. 1b and S4), which has not been previously reported.
We next tested a panel of FLT3 inhibitors in combination with DNA-PK inhibitors. Mutant-FLT3 cell lines were sensitive to the selective type II FLT3 inhibitor AC220, combined with NU7441, which effected a synergistic reduction in cell growth in the mutant-FLT3 cell lines  Table S4). Interestingly, DNA-PK inhibitor combinations with the type II FLT3 inhibitor sorafenib, were more potent than combinations with midostaurin in FLT3-ITD cells, possibly due to dual FLT3 and MAPK inhibition (Figs. 2a and S5, and Table S4). As expected, FLT3-D835 mutant lines were resistant to the type II inhibitor sorafenib, and no synergy was observed ( Fig. 2a and Table S4), suggesting that synergy is dependent on FLT3 inhibition. In comparison, DNA-PK inhibitor combinations with DNAdamaging AML chemotherapeutics, cytarabine and daunorubicin, inhibited proliferation in all lines, irrespective of FLT3 status (Figs. 2a and S6, and Table S4).
To investigate the clinical relevance of DNA-PK and FLT3 inhibitor combinations, drug sensitivity was assessed by Annexin V flow cytometry in human AML blasts ex vivo. NU7441 combined with sorafenib was synergistic in both wildtype and mutant-FLT3 blasts (Figs. 2b and S8A). In contrast, mutant-FLT3 blasts were more sensitive than wildtype-FLT3 blasts to the more potent DNA-PK inhibitor M3814 combined with sorafenib, with the combination effecting synergy (p = 0.041, Figs. 2b and S8B).
The potentiation of DNA-PK and FLT3 inhibitors in mutant-FLT3 cells was further shown in an in vivo preclinical AML model. FLT3-ITD MV4-11 cells, stably expressing luciferase, were engrafted in wildtype-DNA-PK Nod-Rag-Gamma (NRG) mice (Supplementary Methods). Once mean leukaemia burden reached 2% in the peripheral blood, mice were randomised to receive vehicle, M3814 (150 mg/kg), sorafenib (5 mg/kg), or M3814 combined with sorafenib (150 mg/kg M3814 + 5 mg/kg sorafenib). As a monotherapy, M3814 had no effect on leukaemia burden (Fig. 2c). However, sorafenib monotherapy and M3814 combined with sorafenib reduced the proportion of leukaemia cells in the peripheral blood, with a deeper and more sustained response achieved with the combination (Fig. 2c). Bioluminescence measurements also demonstrated a deeper remission of systemic leukaemia burden in the combination group (Fig. S9). Consequently, M3814 combined with sorafenib led to a significant survival benefit; with a median survival of 81 days, compared to 67.5 for mice receiving sorafenib alone (p = 0.0039), 43 days for mice receiving M3814 alone (p = 0.0029), and 42 days for mice receiving vehicle (p = 0.0018, Fig. 2c).
Herein, the increased phosphorylation of DNA-PKdependent NHEJ proteins, and decreased phosphorylation of BER pathway proteins, identified by phosphoproteomics µ µ µ µ µ µ µ µ in mutant-FLT3 AML blasts (Fig. 1b) together uncovers a potential mechanistic explanation for the mutation signatures reported in mutant-FLT3 AML. FLT3-ITD AML displays a C > A and T > G transversion signature [10] consistent with a lack of repair of oxidative DNA damage lesions, which are substrates for BER. Mutant-FLT3 cells have increased error-prone DNA DSBR [11,12], which has not been previously linked with activation of DNA-PKdependent NHEJ. Mutant-FLT3 AML patients display a high rate of cytogenetic evolution [13], and a high frequency of rare structural chromosomal variations at relapse [14], consistent with the mutation signature of over-active DNA-PK-dependent NHEJ [15].
Collectively, we have shown the utility of quantitative phosphoproteomic profiling of AML blasts for identifying activated pathways and guiding the rational selection of drug target combinations. As DNA-PK displayed activation in the majority of AML blast samples, the use of multikinase inhibitors in combination with DNA-PK inhibitors may provide therapeutic benefit in a range of AML subtypes, not limited to mutant-FLT3 AML. Although the mechanism linking FLT3 with DNA-PK activation remains to be determined, our studies demonstrate that DNA-PK is an attractive therapeutic target in AML. Our preclinical results support the clinical evaluation of DNA-PK inhibitors in combination with FLT3 inhibitors as a novel therapeutic strategy for mutant-FLT3 AML.
Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons. org/licenses/by/4.0/. Fig. 2 DNA-PK and FLT3 inhibitors are synergistic and potent in mutant-FLT3 AML. a Proliferation and apoptosis assessed in Ba/F3 and AML cell lines. Upper: Cell survival assessed by resazurin assay, after 72 h treatment with DNA-PK inhibitors (2 µM NU7441, 4 µM M3814, 0.5 µM CC115) and FLT3 inhibitors (0.6 nM AC220, 6 nM sorafenib, 20 nM midostaurin (AML lines), 5 nM midostaurin (Ba/F3 lines)). +, Synergistic drug combinations, determined by the method of Chou Talalay (Table S4). nd, not determined. Lower: Apoptosis was assessed by Annexin V + flow cytometry, after 48 h treatment with the indicated inhibitors. +, Synergistic drug interaction, calculated using the fractional product method of Webb. n = 3 + SEM. b Viability of wildtype-FLT3 and mutant-FLT3 AML patient blast samples after 24 h treatment with the indicated DNA-PK (NU7441, M3814) and FLT3 (sorafenib) inhibitors (Table S5). Cell survival was determined by Annexin V and PI negativity. *, p < 0.05 between wildtype-FLT3 and mutant-FLT3 groups. +, Synergistic drug interaction, calculated using the fractional product method of Webb using the group mean. n = 3 + SEM. c NRG mice were transplanted with FLT3-ITD MV4-11 cells. Once engraftment reached 2% in the peripheral blood, mice were randomised and treated 5 days/week for 4 weeks. Treatment with vehicle (n = 5), 150 mg/kg M3814 (n = 3), 5 mg/kg sorafenib (n = 4), or 150 mg/kg M3814 + 5 mg/kg sorafenib (n = 5) commenced 5 weeks post inoculation of MV4-11 cells. Left, Leukaemia burden in the peripheral blood was measured by flow cytometric analysis of the levels of human CD45 (hCD45) positive cells as a percentage of total human and mouse CD45+ cells. *p < 0.05. Right, Kaplan Meier survival analysis revealed a significant survival advantage in mice treated with M3814 combined with sorafenib. Log-rank test: vehicle vs M3814 + sorafenib, p = 0.0018; M3814 vs M3814 + sorafenib, p = 0.0029, sorafenib vs M3814 + sorafenib, p = 0.0039.