The combination of FLT3 and SYK kinase inhibitors is toxic to leukaemia cells with CBL mutations

Abstract Mutations in the E3 ubiquitin ligase CBL, found in several myeloid neoplasms, lead to decreased ubiquitin ligase activity. In murine systems, these mutations are associated with cytokine‐independent proliferation, thought to result from the activation of hematopoietic growth receptors, including FLT3 and KIT. Using cell lines and primary patient cells, we compared the activity of a panel of FLT3 inhibitors currently being used or tested in AML patients and also evaluated the effects of inhibition of the non‐receptor tyrosine kinase, SYK. We show that FLT3 inhibitors ranging from promiscuous to highly targeted are potent inhibitors of growth of leukaemia cells expressing mutant CBL in vitro, and we demonstrate in vivo efficacy of midostaurin using mouse models of mutant CBL. Potentiation of effects of targeted FLT3 inhibition by SYK inhibition has been demonstrated in models of mutant FLT3‐positive AML and AML characterized by hyperactivated SYK. Here, we show that targeted SYK inhibition similarly enhances the effects of midostaurin and other FLT3 inhibitors against mutant CBL‐positive leukaemia. Taken together, our results support the notion that mutant CBL‐expressing myeloid leukaemias are highly sensitive to available FLT3 inhibitors and that this effect can be significantly augmented by optimum inhibition of SYK kinase.

mutations that confer a growth advantage to leukaemic cells, such as mutations in receptor tyrosine kinases (RTKs) (FLT3 and KIT, for example), non-receptor tyrosine kinases (non-RTKs) (JAK2 and ABL) and downstream signalling molecules such as RAS. Loss of function or deletion of genes that normally turn off growth signalling may also be oncogenic. One example is the E3 ubiquitin ligase, CBL, which is mutated in approximately 10% of myeloid neoplasms, with a low occurrence of 1.1% in AML/MDS and a higher occurrences of 16% in AMLs carrying inv(16)/t(16;16) chromosomal abnormalities (CBL splicing mutations) or 13%-15% in patients with juvenile myelomonocytic leukaemia (JMML) or chronic myelomonocytic leukaemia (CMML). [2][3][4][5][6] CBL mutations cluster in exons 8 and 9, with approximately 85% occurring as point mutations and 15% presenting as small deletions in the linker or RING finger domain of CBL. 2,7,8 These CBL mutations are generally thought to decrease E3 ubiquitin ligase activity.
Interestingly, CBL is known to have as targets several non-RTKs, such as SYK, and RTKs including FLT3, KIT and FMS, as well as the oncogenic variants of those RTKs. 9,10 In myeloid neoplasms, mutations in CBL are thought to decrease turnover of these RTKs and promote growth and viability signalling. In these diseases, CBL, functionally and genetically, acts like a tumour suppressor. 8 This hypothesis is supported by a number of studies demonstrating spontaneous activation of RTK signalling pathways in cells with CBL mutations and also limited studies where multitargeted tyrosine kinase inhibitors (TKI's) such as midostaurin or SU11248 2,8,9,11 can reverse factor independence or factor-hyperresponsiveness of leukaemia cells. 12 As the direct targeting of CBL is a challenge, finding upstream/ downstream vulnerabilities related to mutated CBL is important to develop a successful treatment approach. Previous work has established that the effects of mutated CBL and mutated FLT3 on intracellular signalling, particularly transforming hyperstimulation of signalling molecules downstream of FLT3, are similar, 12 and we thus sought to comprehensively evaluate clinical-grade inhibitors of FLT3 for efficacy against mutant CBL AML. In this study, we have compared the activity of multiple TKIs that have a varied spectrum of activity against FLT3 or KIT and which are available for clinical use or testing, including the N-indolocarbazole midostaurin (PKC412; Rydapt; Novartis Pharma AG), 13 19 and sorafenib (Nexavar; co-developed and co-marketed by Bayer and Onyx Pharmaceuticals). 20 In our cell line model, co-expression of wt FLT3, but not c-KIT, in Ba/F3 cells with mutant CBL was found to be necessary and sufficient to achieve growth factor-independent growth 12 and we therefore tested the effects of the midostaurin and other FLT3 inhibitors on these cells.
For mutant FLT3-positive AML patients to achieve maximum clinical benefit, it has been imperative that midostaurin be administered in combination with other anti-cancer agents. We and others have previously shown that inhibition of the non-receptor cytoplasmic tyrosine kinase (non-RTK) SYK, established to play a critical role in AML transformation, potentiates the anti-leukaemic activity of targeted FLT3 inhibition in models of FLT3-ITD AML and AML characterized by hyperactivated SYK. 21,22 Given the potential of SYK as a putative therapeutic target in AML and the demonstrated ability of additional SYK suppression to augment the anti-leukaemic effects of FLT3 inhibition in mutant FLT3-positive leukaemia, we were interested in exploring the combination of FLT3 and SYK suppression in the context of mutant CBL-positive leukaemia, which is characterized by aberrant FLT3 signalling.
Here, using these cell lines and primary mutant CBL-positive patient cells, as well as murine xenografts, we performed a side-by-side comparison of highly targeted and broad-spectrum FLT3 inhibitors

| Cell proliferation studies
Details are provided in the Appendix S1.

| Immunoblotting and immunoprecipitation
Protein lysate preparation, immunoblotting and immunoprecipitation were carried out as has been previously described. 13

| Antibodies
Antibodies purchased from Cell Signaling Technology were used at a dilution of 1:1000 and include beta-tubulin (rabbit polyclonal,

| Drug combination studies
Details about drug combination studies, which employed the method of Chou and Talalay, 23 are provided in the Appendix S1.

| Non-invasive in vivo bioluminescence study
All animal studies were performed according to protocols approved by the Dana-Farber Cancer Institute's Institutional Animal Care and Use Committee.
Bioluminescence imaging was carried out as previously described. 24 Briefly, Ba/F3.FLT3(wt).CBL.Ins (SK366)-luc+ cells suspended in PBS were implanted intravenously (1.5 × 10 6 cells/mouse) in the female NCr nude mice (7 weeks of age; Taconic, NY). Animals were randomized 3 days post-implantation using total flux values (sum of prone and supine bioluminescence values) into vehicle control and midostaurin, 100 mg/kg once daily by oral gavage for 21 days (n = 9-11/group). Bioluminescence imaging was performed once weekly after treatment initiation, and bodyweights were measured twice weekly.
Midostaurin was formulated as a pre-concentrate/microemulsion with 5% drug powder, 34% Vit E TPGS, 42.5% PEG400, 8.5% corn oil and 10% ethanol. The pre-concentrate was then dissolved in purified water at a 24:76 ratio on the day of treatment. Stocks of midostaurin were purchased from LC Laboratories and MedChemExpress.
For all in vivo studies, P < .05 was considered to be statistically significant. The data had similar variance and met the assumptions of the tests carried out. For in vivo studies investigating the single-agent effects of midostaurin, the Mann-Whitney test (two-tailed) was carried out to assess differences in leukaemia burden between vehicle and drug-treated mice and the Gehan-Breslow-Wilcoxon test was carried out for survival curve comparisons.

| RNA sequencing analysis
Total RNA was isolated from midostaurin-and DMSO-treated cells using TRIzol (Ambion by Life Technologies) followed by an RNeasy cleanup step (RNAeasy kit; Qiagen). Library preparation, sequencing of RNA (RNAseq), and changes in gene expression were performed at the Molecular Biology Core Facility (DFCI). Reads were aligned against the mouse genome; P-values were calculated from raw counts, and false discovery rate (FDR) values were calculated using the method of Benjamini and Hochberg. 25 Gene set enrichment analysis (GSEA) 26

| RE SULTS
Given the demonstrated effectiveness of FLT3 inhibitors against leukaemia characterized by mutated CBL and consequent increased expression of wt FLT3, we were interested in exploring this phenomenon more thoroughly across a panel of widely investigated FLT3 inhibitors in late-stage clinical development or that have been FDA approved for mutant FLT3-positive AML. We tested and compared a range of concentrations of sorafenib, gilteritinib, midostaurin, crenolanib and quizartinib against growth factor-independent Ba/F3 cells engineered to co-express human wt FLT3 and mutated human CBL (ΔY371, Y371H or Ins(SK366)), as well as growth factor-dependent parental Ba/F3 cells, Ba/F3 cells engineered to express wt human FLT3 or Ba/F3 cells engineered to co-express both wt human FLT3 and wt human CBL. As shown in Figure 1A The murine mutant CBL and wt FLT3-expressing cell lines provide a relatively simplistic and clean system with which we were able to directly measure effects of FLT3 inhibition, alone and combined with SYK inhibition. We were interested in seeing whether or not these effects were able to be generalized to human AML cell lines and primary AML cells. As expected, midostaurin notably most potently inhibits human AML cell lines, such as MV4-11 and MOLM14, which express FLT3-ITD, as compared to other human AML cell lines that express wt FLT3 and that are driven by other oncogenes ( Figure 2A). MOLM14 is unique in that it co-expresses FLT3-ITD and mutant CBL (monoallelic CBL deletion transcript identified), 2 however, due to the co-expression of both oncogenes, growth-suppressive effects due to inhibition of oncogenic FLT3 cannot be separated from any drug effects that may be associated with mutant CBL expression in these cells.
We next investigated the effects of our panel of kinase inhibitors on primary AML and CMML cells, including those expressing mutant CBL, as well as those expressing wt FLT3/wt CBL, and mutant FLT3/ wt CBL. Normal PBMCs acquired from a donor were tested as a control. The wt FLT3/wt CBL and mutant FLT3/wt CBL primary cells generally showed more sensitivity to the inhibitors (at 1000 nmol/L) than the mutant CBL-expressing AML samples ( Figure 2B,D). One exception was a mutant CBL-positive CMML sample, which showed a loss of around 50% viability with only 100 nmol/L crenolanib ( Figure 2C). However, it is important to note that gilteritinib, midostaurin and crenolanib killed over 50% of mutant CBL-expressing cells, compared with around 70%-80% of wt FLT3/wt CBL-expressing AML or mutant FLT3/wt CBL-expressing AML. In contrast, less than 20% of normal PBMCs were killed by these drugs, suggesting that drug effects were selective for transformed primary patient cells. Table S1 shows a number of genetic mutations that were identified in the mutant CBL-positive and wt CBL-expressing samples that may influence the sensitivity or resiliency of these samples to the effects of the kinase inhibitors.
As CBL ubiquitin ligase activity has been implicated in the negative regulation of both SYK and FLT3, 9,10 and inactivation of CBL via mutation would be expected to impair this regulation, we were interested in measuring the expression levels and activity of the two proteins in mutant CBL-expressing cells to gain a better understanding of the sensitivity of these cells to FLT3 inhibitors. Specifically, we com- The increases in levels/activity of FLT3 in mutant CBL-expressing cells may be due to failure of mutated CBL to function properly as an E3 ubiquitin ligase 9,10 and may also explain the higher susceptibility of these cells to FLT3 inhibitor treatment. Although differences in transfection efficiency cannot be completely ruled out as contributing to these differences, the observed elevated wt FLT3 expression in mutant CBL-expressing cells is consistent with and supportive of previously published studies focused on effects of mutant CBL in leukaemia. 2,8,9,12 We next investigated the ability of midostaurin to inhibit the pro-  Primary mutant CBL CMML1 (<5% blasts; CMML is divided into two classifications based on cell counts in the blood and bone marrow. CMML-1 is characterized by blasts that make up less than 5% of white cells in blood and less than 10% of cells in bone marrow. CMML-2 is characterized by blasts that make up 5%-20% of white cells in blood or 10%-20% of cells in bone marrow): CBL mutations: I383M, C384Y. Primary wt FLT3 AML2 (95% blasts); Primary FLT3-ITD AML3 (99% blasts); Primary wt FLT3 AML5 (90% blasts). Assay conditions: Primary samples were cultured in RPMI or DMEM supplemented with 10% FBS and 1% pen/strep, with no added cytokines. Cell viability prior to seeding was 98%, as samples were Ficoll-Paque Plus purified prior to testing in the assay. After 72 h, AML patient cell death is maximum 10%-12% in the culture system In addition to looking at downstream effectors, we also investigated effects of midostaurin or quizartinib combined with PRT062607 against SYK and FLT3 directly. A combination effect was observed between midostaurin and PRT062607 against phospho-SYK, presumably due to the fact that SYK is a target of both midostaurin and PRT062607 ( Figure 5C). In contrast, there was no combination effect observed between quizartinib and PRT062607, which may be due to the fact that SYK, while a target of PRT062607, is not a target of quizartinib ( Figure S7F). Neither midostau-rin+PRT062607 nor quizartinib+PRT062607 showed combination effects against phospho-FLT3, which may be due to the fact that although FLT3 is a target of midostaurin and quizartinib, FLT3 is not a target of PRT062607 (Fig 5C and Fig S7F). Drug concentrations used for these combination studies were derived from dose-response studies carried out with each inhibitor (Figures S8 and S9).  Figure 5D).   Clinical studies with midostaurin have shown that its effects as a single agent are at best partial and transient, and for patients to achieve maximum clinical benefit midostaurin must be administered in combination with other anti-cancer drugs. 14,38 Highly activated SYK has been observed to be enriched in FLT3-ITD-positive patient AML, and its cooperation with oncogenic FLT3 leads to activation of MYC transcriptional programmes. 21 Consistent with this, in the present study we observed higher SYK activity in mutant CBL-expressing cells relative to control cells. Mutant CBL-expressing cells were also noted to more robustly express FLT3 and showed increased phosphorylation of FLT3, relative to control cells. As we have previously shown that the inhibitory effect of midostaurin can be further enhanced by additional SYK inhibition against mutant FLT3-positive AML, 22 we were particularly interested in investigating the synergizing potential of FLT3 and SYK inhibition in the context of mutant CBL-positive leukaemia, given our observations.

| D ISCUSS I ON
We have previously shown that midostaurin, a multitargeted kinase inhibitor, is an inhibitor of both SYK and FLT3; we accomplished this using Ba/F3 cells expressing mutant SYK fusion genes and FLT3-ITD. 22 In addition, we have found midostaurin to be unique among the FLT3 inhibitors tested here in terms of its ability to inhibit SYK as well as FLT3 (unpublished results). Here, we expand on our previous findings to show that, because of the similarities between FLT3 signalling in mutant FLT3-and mutant CBL-expressing AML, the inhibitory effects of FLT3 inhibitors can similarly be potentiated by SYK inhibition in both AML subtypes. Cooperativity and transactivation of FLT3-ITD and SYK are transforming, and this causes the two proteins, acting jointly, to be viable targets for combination therapy. 21 In similar fashion, it is possible that hyperactivated wt FLT3 mimics FLT3-ITD, and it is also plausible that overstimulated FLT3 acts in concert with activated SYK, and this may account for observed synergy between FLT3 inhibition and SYK inhibition.
Specifically, we found that additional SYK inhibition, both that of  These results yield novel insights for possible clinical investigation of targeted therapy for mutant CBL-positive AML and suggest that therapies beneficial for mutant FLT3-positive leukaemia may be similarly beneficial for mutant CBL-positive leukaemia due to the similarities between the two leukaemia subtypes.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.